Neuromelanin organelles are specialized autolysosomes that accumulate undegraded proteins and lipids in aging human brain and are likely involved in Parkinson’s disease

Neuromelanin organelles are specialized autolysosomes that accumulate undegraded proteins and... www.nature.com/npjparkd ARTICLE OPEN Neuromelanin organelles are specialized autolysosomes that accumulate undegraded proteins and lipids in aging human brain and are likely involved in Parkinson’s disease 1 1,2 1 1 1 1 1 Fabio A. Zucca , Renzo Vanna , Francesca A. Cupaioli , Chiara Bellei , Antonella De Palma , Dario Di Silvestre , Pierluigi Mauri , 3 3 4 5,6,7 1,5 Sara Grassi , Alessandro Prinetti , Luigi Casella , David Sulzer and Luigi Zecca During aging, neuronal organelles filled with neuromelanin (a dark-brown pigment) and lipid bodies accumulate in the brain, particularly in the substantia nigra, a region targeted in Parkinson’s disease. We have investigated protein and lipid systems involved in the formation of these organelles and in the synthesis of the neuromelanin of human substantia nigra. Membrane and matrix proteins characteristic of lysosomes were found in neuromelanin-containing organelles at a lower number than in typical lysosomes, indicating a reduced enzymatic activity and likely impaired capacity for lysosomal and autophagosomal fusion. The presence of proteins involved in lipid transport may explain the accumulation of lipid bodies in the organelle and the lipid component in neuromelanin structure. The major lipids observed in lipid bodies of the organelle are dolichols with lower amounts of other lipids. Proteins of aggregation and degradation pathways were present, suggesting a role for accumulation by this organelle when the ubiquitin-proteasome system is inadequate. The presence of proteins associated with aging and storage diseases may reflect impaired autophagic degradation or impaired function of lysosomal enzymes. The identification of typical autophagy proteins and double membranes demonstrates the organelle’s autophagic nature and indicates that it has engulfed neuromelanin precursors from the cytosol. Based on these data, it appears that the neuromelanin-containing organelle has a very slow turnover during the life of a neuron and represents an intracellular compartment of final destination for numerous molecules not degraded by other systems. npj Parkinson’s Disease (2018) 4:17 ; doi:10.1038/s41531-018-0050-8 INTRODUCTION to play a dual role, both toxic and protective, that is determined 13,14 by the cellular context and conditions. The synthesis of NM is Electron microscopy studies of neurons of numerous brain regions neuroprotective since it removes from the cytosol the reactive/ have demonstrated that organelles containing neuromelanin (NM) toxic quinones that would otherwise induce neurotoxicity. NM exhibit abundant clear “lipid bodies” (sometimes referred to as further plays a protective role by chelating potentially toxic “lipid droplets” in the literature, although this term is widely used 3,15 metals, including Fe, Zn, Cu, Al, Cr, Mo, Pb, and Hg (refs. ), drugs for a different lipid storage organelle) and a dark electron-dense 16–18 1–3 and organic toxicants. However, NM can play a toxic role matrix. The number of these organelles and the concentration 3–6 when released by degenerating neurons of the SN during PD: of NM pigment increase linearly during aging. These organelles under these conditions, NM acutely discharges high amounts of are highly concentrated in dopamine (DA) neurons of the metals and organic chemicals accumulated over many years of substantia nigra (SN) (Fig. 1a,b,c) and norepinephrine neurons of 2,5,6 locus coeruleus, brain regions strongly targeted in Parkinson’s life. NM released by degenerating neurons in PD activates 7,8 microglia, producing reactive and pro-inflammatory molecules disease (PD). The pigments of these organelles are a family of compounds formed by a melanic, aliphatic, and protein compo- that induce further neuronal death and release of NM, thus establishing a vicious cycle of neuroinflammation and neurode- nents with variable ratios. NM pigment also accumulates large amounts of metals, further confirming that these organelles generation. The activation of microglia by NM can drive antigen presentation by SN and locus coeruleus catecholaminergic continuously accumulate in aging due to very slow turnover. The formation of NM appears to provide a protective neurons, a response that may play a crucial role in PD 3,10 process, but the amount of NM accumulated in neurons is pathogenesis. NM can also stimulate dendritic cells in vitro 8,11,12 related to their vulnerability in PD. Due to its biochemical inducing their maturation. properties, NM has long been suggested as a critical factor The structure of the melanic component is different in various 8 9,22 underlying neuronal vulnerability in PD. Indeed, NM is suggested types of NM pigments, and multiple features of NM structure, 1 2 3 Institute of Biomedical Technologies, National Research Council of Italy, Segrate, Milan, Italy; IRCCS Don Carlo Gnocchi ONLUS Foundation, Milan, Italy; Department of Medical 4 5 Biotechnology and Translational Medicine, University of Milan, Segrate, Milan, Italy; Department of Chemistry, University of Pavia, Pavia, Italy; Department of Psychiatry, Columbia University Medical Center, New York State Psychiatric Institute, New York, NY, USA; Department of Neurology, Columbia University Medical Center, New York, NY, USA and Department of Pharmacology, Columbia University Medical Center, New York, NY, USA Correspondence: Luigi Zecca (luigi.zecca@itb.cnr.it) These authors contributed equally: Fabio A. Zucca, Renzo Vanna. Received: 14 December 2017 Revised: 10 April 2018 Accepted: 17 April 2018 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. as well as protein and lipid composition of NM-containing isolation of ORG samples, the membranes can be broken with organelles, remain incompletely characterized. Indeed, the protein consequent mixing of intra- and extracellular proteins. The components of NM-containing organelle have been partially washing procedure of ORG isolation can moreover change the characterized, and these are consistent with its lysosomal protein content of organelles due to membrane damage. The 23,24 nature. However, the characterization of proteins belonging native proteins detected by analyzing ORG samples therefore to different portions of the NM-containing organelle and their likely underestimate the original protein content, while there may localization are yet to be clarified, although this knowledge is be an increased number of proteins due to contamination. fundamental to understanding the complex nature of these Thus, we also isolated TIS-NM for protein characterization by LC- organelles. Current data do not indicate the mechanisms of NM MS. During the isolation process of TIS-NM, membranes are accumulation, protein and lipid transport and accumulation within broken and additional proteins are likely aspecifically adsorbed by the organelle, or the role of these organelles inside neurons. It is NM pigment. TIS-NM was prepared following the procedure 3,9,28 further unknown whether the NM pigment is synthesized within reported by several previous studies. Here, proteinase K was these organelles or transported inside after synthesis elsewhere. employed during the isolation procedure of TIS-NM from SN tissue Here we investigate the proteins and lipids of NM-containing as it was necessary to remove non-specifically associated proteins. organelles from human SN and their accumulation inside these A preliminary study reported that TIS-NM isolated without organelles. The NM-containing organelle analyzed here is a proteinase K contained a higher percentage of extracellular and paradigmatic case in which an abundant accumulation of the nuclear proteins than TIS-NM isolated with proteinase K, indicating melanic pigment occurs; however, the presence of these a higher contamination by non-specific proteins during the organelles is observed throughout the entire brain as a result of isolation procedure. Indeed, the isolation of NM pigment physiological aging. without proteinase K generates macroaggregates difficult to The aim of this study is to perform an extended characterization purify. These macroaggregates could contain proteins originating of proteins and lipids of NM-containing organelles from human from cytosol, other organelles and tissue compartments which SN. To this end, we required highly purified preparations of NM- interact and bind to NM pigment and NM-conjugated proteins containing organelles. In order to control contaminations and to during isolation, that may form S–S bridges and other means of avoid the loss of some proteins or lipids, we prepared three types conjugation. In this case, this kind of sample would contain of NM-containing samples with different procedures and com- proteins not related to the NM-containing organelle. This pared the data obtained by liquid chromatography-mass spectro- procedure and its rationale are described in detail in previous 3,9 metry (LC-MS) determinations on these samples. To further studies (Methods). We elected to prepare the TIS-NM samples confirm the reliability of our results, independent determinations with proteinase K as this type of sample was used in several were also made by immunoelectron microscopy (IEM), western previous studies, and so necessary for comparison of the present 3,9,28 blotting (WB), and thin-layer chromatography (TLC) in addition to data with previous reports. LC-MS. In this study, protein profiles were analyzed (i) in three Finally, to overcome the above experimental limitations, we preparations derived from human SN, the organelles containing further analyzed proteins in ORG-NM samples, i.e., the NM NM pigment (ORG), NM pigment purified from SN tissues (TIS-NM), pigment isolated from ORG samples. This type of NM was purified and NM pigment isolated from organelles (ORG-NM) using LC-MS; by disrupting and eliminating the membranes and the soluble (ii) by WB of ORG samples; and (iii) by IEM of SN tissue slices. The portion from ORG samples, in order to study the protein lipid pathways were analyzed by TLC and LC-MS analyses of lipid components strictly associated with NM pigment, without molecules associated with lipid bodies and with NM pigment, and exposure to contaminants from other intraneuronal components. through characterization of transport proteins and related This experimental design was in our opinion the best approach enzymes by LC-MS, IEM, and WB analyses. to exclude contaminating proteins. In addition, this approach The combination of proteomics, lipidomics, imaging, and avoided loss of some proteins of the NM-containing organelle, a biochemical techniques in the study of NM-containing organelles central aspect of this study, beyond preventing contamination. is required for a detailed description of the molecular mechanisms We then compared the three sets of proteins observed in the associated with brain aging and neurodegeneration. Previous following preparations (Supplementary Fig. 1): two independent 23–27 studies have only partially addressed these issues. The samples of ORG, two independent samples of TIS-NM, and two identification of these pathways is crucial for elucidating the independent samples of ORG-NM (Methods). ORG, TIS-NM, and processes mediating neuronal survival and vulnerability during ORG-NM samples were analyzed by a total of 15 LC-MS analyses. aging and PD. As a result, 1020 proteins were identified and a group of 293 was selected as representative of the samples based on their amount (Table 1), as estimated from the number of spectral count (SpC), RESULTS defined as the sum of all peptides of a single protein observed in NM organelles proteins were identified by analyzing three types of the mass spectrum. The threshold for inclusion in the list of samples: isolated NM-containing organelles, NM purified from SN representative proteins was two or more SpC (i.e., peptides) for tissues, and NM purified from NM-containing organelles each protein as average value in at least one of the three types of The purpose of this study is to describe the proteins present inside samples. the NM-containing organelles, and then distinguish those The cellular distribution of the 293 representative proteins is covalently bound to NM pigment from those not attached to represented in Fig. 2 (for details refer to Supplementary Table 1; NM. The proteins bound to NM pigment may be more related to Supplementary Data 1), showing that 34 proteins, detected by the initial steps of NM synthesis, while those not bound to NM ∼60 % of the number of SpC in all samples (7916 of 13,157 overall pigment may play a role in the membrane, transport, and storage number of SpC for representative proteins), were lysosomal processes involved in NM-containing organelle formation. There proteins. We can speculate that if all the protein classes were are several experimental limitations in the analyses of proteins in equally represented in our samples, the 34 lysosomal proteins NM-containing organelles. In human post mortem brain, the among the 293 representative proteins (Supplementary Table 1) membranes of organelles undergo degradation, and their would be represented by ∼1527 SpC and not by 7916 SpC, as we constituents can be released from organelles to the cytosol, and observed. With this assumption, we estimate that the lysosomal conversely, the organelles can be contaminated by cytosolic class is >5-fold enriched in our samples. This is a striking evidence components. Furthermore, when processing brain tissues for of lysosomal protein overrepresentation. npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation 1234567890():,; Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Table 1. Summary of the proteomic analysis ORG (n = 2) TIS-NM (n = 2) ORG-NM (n = 2) Overall Number of proteins 563 407 220 1020 Number of SpC for all 1020 proteins 4229 7743 2023 13,995 Representative proteins (with SpC ≥ 2 as average value) 174 [164] 130 [116] 125 [101] 293 Number of SpC for the group of 293 representative proteins 3792 [3778] 7439 [7420] 1926 [1898] 13,157 [13,096] Summary of proteomic analyses of different types of samples (ORG, TIS-NM and ORG-NM), calculated from Supplementary Data 1. Each type of sample was prepared in duplicate and then analyzed by multiple LC-MS analyses. For details of subjects and preparation of samples for LC-MS analysis of proteins, see Methods. The first two rows show the number of proteins and their number of SpC for each kind of sample. Among the group of overall proteins detected in these samples (1020 by 13,995 SpC), representative proteins were selected on the basis of their SpC, defined as the sum of all peptides of a single protein observed in the mass spectrum. We have classified a protein as representative in one type of sample, if detected by SpC ≥ 2 as average value. In the third row, we report for each type of sample the following values: (i) in brackets, the number of proteins detected as representative (with SpC ≥ 2) uniquely in that type of sample; (ii) without brackets, the number of representative proteins in that type of sample plus those identified as non-representative (with SpC < 2) in that specific sample but listed as representative (with SpC ≥ 2) in at least one of the other type of samples (e.g., a protein that was detected as non-representative in TIS-NM but as representative in ORG or ORG-NM samples would be included in the count for TIS-NM). The fourth row of the table shows the number of SpC detected for the group of 293 representative proteins described in the third row, and in particular: (i) in brackets, the total number of SpC detected for representative proteins (with SpC ≥ 2) uniquely found in that type of sample (e.g., 3778 is the number of SpC for 164 proteins uniquely detected in ORG sample); (ii) without brackets, the total number of SpC of representative proteins plus the SpC of proteins identified as non-representative (with SpC < 2) in that specific sample but listed as representative (with SpC ≥ 2) in at least one of the other type of samples (e.g., 7439 is the number of SpC for 130 proteins detected in TIS-NM sample, including SpC of proteins uniquely found in TIS-NM plus those of proteins detected as non-representative in TIS-NM but as representative in ORG or ORG-NM samples). The “Overall” column represents the overall number of proteins and peptides detected in all analyzed samples (i.e., 293 is the number of all representative proteins detected in ORG, TIS-NM, and ORG-NM, considering only one time a protein present in two or more samples) An Euler diagram in Fig. 3 shows the distribution of the 293 membrane proteins including lysosome membrane protein 2 representative proteins among different samples, as described in (SCARB2), CD63 antigen, type 1 phosphatidylinositol 4,5-bispho- Table 1. The ORG sample contains the highest number of proteins sphate 4-phosphatase (only in the ORG sample) and some functional and shares 80 proteins with the other two samples. TIS-NM and subunits of the lysosomal V-type proton ATPase (in ORG sample one ORG-NM, representing two different preparations of NM pigment, subunit as representative while other two subunits in very low contain fewer proteins. TIS-NM shares 68 proteins with ORG and amounts and categorized as non-representative proteins), but not lysosome-associated membrane glycoprotein 1 (LAMP1) or ORG-NM samples, while ORG-NM shares 89 proteins with ORG and lysosome-associated membrane glycoprotein 2 (LAMP2). TIS-NM samples. A portion of representative proteins was uniquely Due to the difficulties in preserving ORG membranes during identified in each type of sample (94 proteins in ORG, 62 proteins their isolation and related problems in LC-MS detection of in TIS-NM, and 36 proteins in ORG-NM), but these were present in transmembrane proteins, we also performed WB and IEM very small quantities (<10 % of the overall number of SpC experiments to investigate lysosomal features of these organelles. detected for representative proteins); interestingly, 35 proteins We confirmed the presence of SCARB2 and V-type proton ATPase were commonly detected in all three samples, and these shared subunit B, brain isoform (ATP6V1B2), a subunit of the lysosomal V- proteins were present in high quantities (∼80 % of the overall type proton ATPase, in NM-containing organelles by WB on ORG number of SpC detected for representative proteins; see data and samples and by IEM in SN sections (Figs. 4 and 5; Supplementary discussion in the legends of Fig. 3 and Table 1, and details in Fig. 2, 3). Likewise, both LAMP1 and LAMP2, while undetected by following sections). LC-MS in ORG samples, were observed by WB and IEM (Figs. 4 and 5; Supplementary Fig. 2, 3). Proteins found in NM-containing organelles Some non-lysosomal proteins were also detected, most of The ORG samples were obtained directly from fresh SN tissue, which are classified as cytoskeletal and cytoplasmatic proteins. In without freezing and after soft homogenization/centrifugation addition to typical tubulin chains (mainly beta chains), we found procedures in order to preserve the original structure and tubulin polymerization-promoting protein, which is involved in composition of these organelles. Transmission electron micro- protein aggregation, inclusion bodies formation and neurodegen- scopy confirmed that the contents of the purified organelles were eration; we also detected heat shock protein HSP 90-alpha and mainly intact, with preserved lipid membranes and lipid bodies alpha-crystallin B chain, which has been reported as a component that were morphologically identical to those observed in slices of of Lewy bodies and has been characterized as a chaperone. SN tissues. Importantly, low magnification images demonstrated Additional cytoskeletal proteins included for example the the absence of other cellular contaminants (Fig. 1d). microtubule-associated protein tau and microtubule-associated A group of 164 proteins was identified with SpC ≥ 2 exclusively in protein 6, which play roles in microtubule stability and are 32,33 ORG sample (Table 1; Supplementary Data 1), while an additional implicated in neurodegenerative mechanisms. 34,35 ten proteins were identified with SpC < 2 in ORG but detected with As a potentially important clue in PD pathogenesis, we also SpC ≥ 2 in TIS-NM or ORG-NM samples. Then, we are considering detected alpha-synuclein (SNCA) exclusively in ORG samples. We that the number of representative proteins in ORG sample is 174 confirmed this result by WB experiments on ORG samples and IEM (Table 1;Fig. 3; Supplementary Table 1).These data revealed that experiments on SN tissue slices, observing SNCA signals mainly in ORG is characterized by a large group of lysosomal proteins (25 the NM pigment and rarely in lipid bodies (Figs. 4 and 5). proteins, ~43 % rel. # SpC), with far fewer proteins from other We also found major histocompatibility complex, class I (HLA) in organelles or cytosol (Fig. 2; Supplementary Table 1). In addition to ORG samples, consistent with our recent report of the first the set of soluble lysosomal proteins observed in all three types of identification of this protein in adult neurons. The identification samples (proteases, esterases, sulfatases, glycosidases, hydrolases, of HLA was obtained by matching experimental spectra to peptide and other lysosomal proteins), here we identified typical lysosomal sequences in specific databases for HLA (Supplementary Data 2). Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Fig. 1 Transmission electron microscopy images of NM-containing organelles in human SN tissue (a–c) and after the isolation procedure (d). a–c SN tissue of 89 y.o. healthy subject. Intraneuronal NM-containing organelles of the SN are membrane bounded (black arrowhead in a and b) and contain large amounts of NM pigment (black and electron dense), a protein matrix and lipid bodies (asterisk). Scale bar =1µm in a. Large lipid bodies (asterisk) are surrounded by a membrane as demonstrated in b (arrow), although the images do not distinguish between a bilayer and single layer membrane. Considering that brain samples used in this study were post mortem tissues, it is striking that there is often a double membrane around many of the organelles. At higher magnification, a double membrane delimiting NM-containing organelle is clearly visible (empty arrowhead in c). d NM-containing organelles isolated from the SN tissue of 89 y.o. healthy subject (the same subject of a–c). The purity and integrity of isolated NM-containing organelles is clearly demonstrated by transmission electron microscopy: low magnification d demonstrates that cellular and subcellular debris are completely absent. The outer limiting membrane is not apparent, but the constituents of the organelles appear intact with NM pigment, many lipid bodies (asterisks) and membranes of lipid bodies. Scale bar = 1 µm in d Fig. 2 Histogram of cellular distribution of the 293 representative proteins found in all analyzed samples shown as relative number of SpC vs. cellular compartments. For details of subjects and preparation of samples for LC-MS analysis of proteins, see Methods. Some proteins may have multiple cellular locations: for each protein the most typical and representative cellular location was assigned. The different types of samples are represented by different colors (ORG, TIS-NM, and ORG-NM) and gray bars refer to overall representative proteins considered as a single data set (indicated with “All Samples”). The “Rel. # SpC (%)” is the total number of SpC for a specific class of proteins (i.e., lysosomal) referred to the overall number of SpC of representative proteins in each sample: this value represents the relative abundance of a particular class of proteins in one sample (see also Supplementary Table 1). The term “Vesicles” refers to vesicle trafficking, including proteins involved in vesicular transport, fusion, etc. The category “Unknown cell location” consists of proteins for which a cellular location was still unclear, while the class “Uncharacterized proteins” comprises proteins for which, at the moment of data analyses, a complete characterization and/or role was missing npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. The presence of the antigen presenting protein HLA was confirmed with IEM on SN tissue, indicating a high accumulation of HLA on NM granules of the SN (Supplementary Fig. 4). Interestingly, some additional proteins identified only in ORG sample were dynein heavy chain 12, axonemal and some ras- related proteins (RAB8B, RAB14, and RAB33B as representative proteins, while RAB2A, RAB5C, and RAB8A were present in low amounts and categorized as non-representative), each of which are involved in intracellular vesicle trafficking. Proteins found in NM pigment isolated from SN tissue The TIS-NM samples were obtained from pooled SN tissues using the purification procedure adopted for previous chemical and 3,9,28 structural investigations on NM pigment. Despite the chemical treatments (i.e., high-salt solutions, sodium dodecyl sulfate, methanol, and hexane), repeated washings and proteinase K digestions, proteomic analysis resulted in the identification of 116 proteins with SpC ≥ 2 uniquely in TIS-NM (Table 1; Supple- mentary Data 1). However, an additional 14 proteins with SpC < 2 in TIS-NM were found with SpC ≥ 2 in ORG or ORG-NM samples: Fig. 3 Area-proportional Euler diagram of the 293 representative therefore, we estimate 130 to be the number of representative proteins (detected by SpC ≥ 2 as average value in at least one of the proteins in TIS-NM sample (Table 1; Fig. 3; Supplementary Table 1). three types of samples) identified in ORG, TIS-NM, or ORG-NM. For details of subjects and preparation of samples for LC-MS analysis of The most highly represented and abundant class of proteins was proteins, see Methods. The diagram was calculated using the again lysosomal, even more so than in ORG samples. We detected EulerAPE tool (Methods) and by using NCBI accession (GI 26 lysosomal proteins representing ~73 % of rel. # SpC in this number). Outside the diagram we report for each type of sample sample (Fig. 2; Supplementary Table 1). Among these proteins, the following values: (i) in brackets, the number of proteins detected high quantities of typical lysosomal enzymes were found, and as representative (with SpC ≥ 2) uniquely in that type of sample, as these were also identified in ORG and ORG-NM samples. However, reported in Table 1; (ii) without brackets, the number of some lysosomal proteins were detected exclusively in TIS-NM representative proteins plus those identified as non-representative samples (e.g., epididymal secretory protein E1, fatty acid synthase, (with SpC < 2) in that specific sample but listed as representative cathepsin L1, lysosomal alpha-mannosidase, and ribonuclease T2). (with SpC ≥ 2) in at least one of the other type of samples (e.g., a protein that was detected as non-representative in ORG sample but Only one lysosomal membrane protein, the transmembrane as representative in TIS-NM would be included in the count for protein 106B, was identified exclusively in the TIS-NM sample. ORG). Numbers in non-overlapping areas of circles report the Non-lysosomal proteins found in TIS-NM included ferritins, representative proteins found uniquely in that type of sample. The mostly ferritin light chain (FTL) in contrast to ferritin heavy chain overlapping areas correspond to proteins shared by two or three (FTH1), heat shock protein HSP 90-alpha, and glyceraldehyde-3- different types of samples: e.g., a protein detected in all samples but phosphate dehydrogenase. HLA was also detected in this sample as representative only in ORG would be included in the overlapping (Supplementary Data 2). By means of LC-MS analysis, we found area of 35 proteins shared between ORG, TIS-NM, and ORG-NM. exclusively in this sample the mature chain of ATP synthase F(0) Percentages in parentheses represent the ratio between the total complex subunit C3, mitochondrial (ATP5G3), the major storage number of SpC of proteins belonging to one area of the diagram and the overall number of SpC of representative proteins detected material accumulated in ceroid lipofuscinosis, as well as in all samples. The highest percentage value is located in the area cerebellin-2, the PD-associated protein DJ-1 and protein shared between all three types of samples. The detailed list of disulfide-isomerase A3. Additionally superoxide dismutase proteins is reported in Supplementary Data 1 [Cu–Zn] was detected in TIS-NM and also as a fragment in ORG sample. The ATP synthase F(0) complex subunit C1, mitochondrial (ATP5G1) was also detected by WB in ORG samples; its presence were identified in ORG-NM with SpC < 2 but detected with SpC ≥ 2 was also confirmed by IEM, indicating that is localized in the NM- in ORG or TIS-NM samples. Therefore, we estimate the number of containing organelles, where it is mainly bound to the NM representative proteins in ORG-NM sample to be 125 (Table 1; Fig. pigment. The antibody used in these experiments is expected to 3; Supplementary Table 1); the majority (89 proteins) were also recognize the three mature chains of proteins, namely ATP5G1, detected in ORG and TIS-NM samples (Fig. 3). Proteins identified ATP synthase F(0) complex subunit C2, mitochondrial (ATP5G2) and ATP5G3, which are identical and encoded by three different only in ORG-NM sample (36 proteins) were found in very low genes (Supplementary Fig. 2, 3). amounts (<1 % rel. # SpC) (Fig. 3), thus indicating that almost no contaminants affected the analysis of this type of sample. As observed in ORG and TIS-NM, lysosomal proteins were also Proteins found in NM pigment isolated from organelles prevalent in this sample (23 proteins; ~44 % rel. # SpC) (Fig. 2; In order to study the protein matrix associated with the NM Supplementary Table 1). Moreover, lysosomal membrane proteins pigment inside organelles and considering that aspecific proteins (i.e., SCARB2 and CD63 antigen) and other cytoskeletal and might interact with NM during the isolation of TIS-NM from SN cytoplasmatic proteins (i.e., alpha-crystallin B chain, tubulin tissues, we isolated NM pigment directly from ORG samples. The polymerization-promoting protein, microtubule-associated pro- amount of ORG-NM sample was lower than the other two types of tein tau, and low amounts of microtubule-associated protein 6) samples (ORG-NM < ORG < TIS-NM), as it was prepared from the found in ORG samples were observed in ORG-NM as well. Again, ORG sample; TIS-NM was isolated processing several SN pooled LAMP2 was not detected in this sample by LC-MS, while LAMP1 tissues, while ORG and ORG-NM samples were prepared from one was identified at very low levels and categorized as a non- or occasionally two SN tissues (Methods). This is clearly shown by the number of SpC detected in each sample (Table 1). A group of representative protein. These data suggest that thermal shock 101 proteins was detected with SpC ≥ 2 exclusively in ORG-NM treatments to purify NM pigment from organelles do not (Table 1; Supplementary Data 1), while an additional 24 proteins completely disrupt the organelle. It is possible that some Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. membrane portions are strongly connected to the contents of the experiments, which confirmed their localization inside the NM- organelle and prevent the leakage of some proteins. Indeed, containing organelles (Figs. 4 and 5; Supplementary Fig. 4, 5, 6). membranous, cytoskeletal and cytoplasmic proteins were still found in the analysis of ORG-NM sample. As observed in ORG and Relevant proteins not revealed by mass spectrometry but found in TIS-NM samples, HLA was also detected in this sample by other techniques (Supplementary Data 2). Additional IEM and WB studies were performed for some relevant Similarly, ORG-NM samples share several proteins with TIS-NM proteins not detected by LC-MS, probably due to instrumental (Fig. 3): e.g., the lysosomal prosaposin, heat shock protein HSP 90- limitations and low abundance. Due to the lysosomal features of alpha (here detected in low amounts, but present as representa- the NM-containing organelle and its proposed autophagic origin, tive also in ORG and TIS-NM), glyceraldehyde-3-phosphate we performed WB and IEM experiments to verify the presence of dehydrogenase, some peptides of ferritins (again mostly FTL if the autophagic marker microtubule-associated proteins 1A/1B compared to FTH1), and the PD-associated protein ubiquitin light chain 3B (MAP1LC3B), which was not detected by LC-MS like carboxyl-terminal hydrolase isozyme L1 (here as representative, 23,24 in previous studies. Here, IEM experiments on SN tissues while in TIS-NM it was in low amounts and categorized as a non- revealed MAP1LC3B signals mainly engulfed inside NM-containing representative protein). ORG-NM and TIS-NM are analogous organelles, on NM pigment, around lipid bodies and sometimes samples since they are both isolated NM pigment. However, the lining their membranes (Fig. 4). In order to confirm the presence of amount of ORG-NM sample is normally lower than that of TIS-NM, this crucial protein, IEM experiments were also repeated using a while the ORG sample is the intact organelle containing the NM different antibody, confirming the specific accumulation of pigment. Due to the identification by LC-MS of FTL and FTH1 in MAP1LC3B inside NM-containing organelles (Supplementary Fig. both ORG-NM and TIS-NM samples but not in ORG samples, IEM 4). In addition to IEM findings, WB analyses of ORG samples and WB experiments were performed. IEM confirmed the presence demonstrated the presence of MAP1LC3B, thus confirming the of FTL and to a lesser extent FTH1 in NM-containing organelles of autophagic nature of NM-containing organelle. Specifically, WB the SN tissue. The accumulation of this iron-storage protein was analyses demonstrated mainly the MAP1LC3B-I form, while the confirmed by WB analyses that detected low levels of FTL but MAP1LC3B-II form was undetectable in ORG samples (Fig. 5). failed to detect FTH1 in ORG samples (Supplementary Fig. 2, 3). Another important autophagy-related protein, the autophagic This may be due to chains of iron hydroxides that formed bridges adaptor sequestosome-1 (SQSTM1), was not detected by LC-MS 23,24 connecting NM and partially degraded ferritins, so that this here and in previous studies, but was found by IEM and WB protein was bound to NM and was not present as a separate analyses (Supplementary Fig. 5, 6). molecule. Moreover, due to the presence of some Ras-related proteins, we Examples of proteins identified as representative only in ORG- confirmed the presence of Ras-related protein Rab-5A (RAB5A), NM include phospholipid hydroperoxide glutathione peroxidase, which is of interest due to its involvement in retrograde axonal mitochondrial, and lysosomal acid phosphatase. endosomal transport, by IEM and WB (Supplementary Fig. 5, 6) but not by LC-MS. Proteins detected in all three types of samples A common group of proteins was identified in all types of Lipid bodies contain dolichols involved in NM synthesis and samples, corresponding to ~80 % of overall SpC detected for 293 typical membrane lipids representative proteins (Fig. 3; Table 2). Within NM-containing organelles, electron micrographs demon- Within the set of lysosomal proteins detected in all three types strated the presence of membrane-bound lipid bodies, generally of samples (17 entries in Table 2), were peptidases (tripeptidyl- ranging from 200 to 500 nm, sometimes reaching sizes as large as peptidase 1, gamma-glutamyl hydrolase and lysosomal Pro-X 1 µm, with some smaller lipid bodies (50–100 nm) entrapped in carboxypeptidase), proteases [cathepsin B, cathepsin Z and the NM regions of the organelle (Fig. 1a–c). Conventional electron cathepsin D (CTSD)], esterases (sialate O-acetylesterase and microscopy on post mortem SN tissues does not clearly reveal if palmitoyl-protein thioesterase 1), sulfatases (iduronate 2-sulfatase, these lipid bodies are surrounded by membranes formed of N-sulphoglucosamine sulphohydrolase, arylsulfatase B and N- normal bilayer or single layer. acetylgalactosamine-6-sulfatase), glycosidases (lysosomal alpha- LC-MS analyses of solvent extracts prepared from both TIS-NM glucosidase), lipid hydrolases (acid ceramidase), and other and ORG samples demonstrate that dolichols and dolichoic acids lysosomal proteins [mammalian ependymin-related protein 1 were the major lipid components (Fig. 6). LC-MS analyses of both and putative phospholipase B-like 2 (PLBD2)]. CTSD, a classical solvent extracts further revealed the presence of signals lysosomal marker, was detected in ORG samples by WB as heavy corresponding to different glycerophospholipids and sphingoli- chain mature form and was confirmed by IEM as present pids (Supplementary Table 2). Among sphingolipids, signals abundantly in the NM-containing organelles (Figs. 4 and 5). We attributable to sphingomyelin, neutral glycolipids (lactosylcera- confirmed by IEM and WB (Supplementary Fig. 2, 3) the presence mide), sulfatides, and gangliosides (mono-, di-, and tri-sialogan- 37,38 of the recently described PLBD2. In addition, we detected by gliosides) together with other lipid molecules such as free fatty LC-MS phospholipase D3, apolipoprotein D (APOD), cerebellin-1, acids were identified (Supplementary Table 2). Thus, the lipid transmembrane glycoprotein NMB (GPNMB), F-box only protein bodies of ORG and the lipid mixtures adsorbed to TIS-NM contain 11, heat shock protein HSP 90-alpha, and two ubiquitin-related a wide variety of membrane amphipathic lipids, encompassing proteins, namely polyubiquitin-C (UBC) and ubiquitin-60S riboso- both lipids typically enriched in neurons, such as gangliosides, and mal protein L40 (UBA52). HLA peptides associated with NM were lipids involved in oligodendrocyte function and myelin formation, 25,41 identified by LC-MS analyses in all samples isolated from human such as sphingomyelin and sulfatide. The profile of amphi- SN (Supplementary Data 2). In each type of sample we also found pathic lipids was not identical in TIS-NM and in ORG, in particular protein phosphatase 1 regulatory subunit 1B, which potently as signals attributable to sulfatides were far higher in TIS-NM inhibits protein phosphatase-1. Moreover, various tubulins (Supplementary Table 2), suggesting that some myelin lipids could (mainly beta chains) and other abundant proteins including glial be adsorbed to TIS-NM during the purification procedure. Indeed, fibrillary acidic protein, myelin basic protein, and creatine kinase B- the presence of amphipathic lipids in the lipid bodies from ORG type were detected in all three samples (for other proteins see and in lipid mixtures adsorbed to TIS-NM was confirmed by TLC Table 2; Supplementary Data 1). Due to the important roles of analysis (Fig. 7). In the solvent extracts of the TIS-NM and ORG APOD, GPNMB, ubiquitins, and HLA, we conducted IEM and WB samples, the amounts of dolichols and dolichoic acids were higher npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Table 2. Representative proteins commonly detected by LC-MS in all analyzed samples NCBI accession (GI UniProt accession Protein name Gene name Cellular location All samples number) number Total # Average # SpC SpC 193785841 P61981 14-3-3 protein gamma YWHAG Cytoplasm 13 2 21536286 P12277 Creatine kinase B-type CKB Cytoplasm 41 6 4503979 P14136 Glial fibrillary acidic protein GFAP Cytoplasm 124 14 21735492 Q9UD71 Protein phosphatase 1 regulatory PPP1R1B Cytoplasm 28 3 subunit 1B 32526901 Q8WYA0 Intraflagellar transport protein 81 IFT81 Cytoskeleton 17 2 homolog 105990539 P07196 Neurofilament light polypeptide NEFL Cytoskeleton 10 2 2119276 Q6LC01 Tubulin beta chain (fragment) N/A Cytoskeleton 126 11 4507729 Q13885 Tubulin beta-2A chain TUBB2A Cytoskeleton 22 3 21361322 P04350 Tubulin beta-4A chain TUBB4A Cytoskeleton 60 8 60729665 Q6Y288 Beta-1,3-glucosyltransferase B3GLCT Endoplasmic 17 2 reticulum 1575347 Q8IV08 Phospholipase D3 PLD3 Endoplasmic 266 19 reticulum 4502163 P05090 Apolipoprotein D APOD Extracellular 74 10 8247915 Q13510 Acid ceramidase ASAH1 Lysosome 225 17 825628 P15848 Arylsulfatase B ARSB Lysosome 53 4 4503139 P07858 Cathepsin B CTSB Lysosome 1058 63 3929733 Q6LAF9 Cathepsin B (fragment) N/A Lysosome 88 10 4503143 P07339 Cathepsin D CTSD Lysosome 343 24 22538442 Q9UBR2 Cathepsin Z CTSZ Lysosome 169 11 4503987 Q92820 Gamma-glutamyl hydrolase GGH Lysosome 1003 62 4557659 P22304 Iduronate 2-sulfatase IDS Lysosome 95 8 126590 P10253 Lysosomal alpha-glucosidase GAA Lysosome 37 3 4826940 P42785 Lysosomal Pro-X carboxypeptidase PRCP Lysosome 91 7 24475586 Q9UM22 Mammalian ependymin-related EPDR1 Lysosome 1489 88 protein 1 4503899 P34059 N-acetylgalactosamine-6-sulfatase GALNS Lysosome 26 2 4506919 P51688 N-sulphoglucosamine SGSH Lysosome 47 4 sulphohydrolase 4506031 P50897 Palmitoyl-protein thioesterase 1 PPT1 Lysosome 287 19 27734917 Q8NHP8 Putative phospholipase B-like 2 PLBD2 Lysosome 57 4 6808138 Q9HAT2 Sialate O-acetylesterase SIAE Lysosome 2043 124 2408232 O14773 Tripeptidyl-peptidase 1 TPP1 Lysosome 428 26 4505405 Q14956 Transmembrane glycoprotein NMB GPNMB Melanosome 49 11 68509930 P02686 Myelin basic protein MBP Membrane 1867 125 5912201 Q86XK2 F-box only protein 11 FBXO11 Nucleus 21 4 31542868 P13807 Glycogen [starch] synthase, muscle GYS1 Nucleus 6 1 4506645 P63173 60S ribosomal protein L38 RPL38 Ribosome 8 2 4757922 P23435 Cerebellin-1 CBLN1 Synapse 226 14 List of representative proteins commonly detected by LC-MS analyses in all samples. In the table for each protein, NCBI accession (GI number), UniProt accession number, protein and gene name, cellular location, and few details of LC-MS analyses corresponding to all samples are reported. The column “All Samples” refers to overall representative proteins commonly present in all three types of samples which are considered as a single data set. Note that some proteins may have multiple cellular locations: for each protein the most typical and representative cellular location was assigned. A protein eligible to be inserted in this list must be detected in all samples, and classified as representative (detected by SpC ≥ 2 as average value) in at least one of the three types of samples. This list describes the 35 proteins contained in the central overlapping area of Euler diagram in Fig. 3. For full details see Supplementary Data 1. Two ubiquitin-related proteins (UBC and UBA52) and heat shock protein HSP 90-alpha, although present in the three types of samples, were not included in the group of proteins detected in all three types of samples (in this list and in the central overlapping area of Euler diagram in Fig. 3, containing 35 proteins) because they were identified with different GI accession numbers in different samples. Note that the Euler diagram calculation was performed using NCBI accession (GI number) only. Nevertheless, the above mentioned ubiquitin-related proteins and heat shock protein HSP 90-alpha, considered as unique proteins, were detected in all three samples Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Fig. 4 IEM of SN from healthy aged subjects for selected proteins. For number of IEM experiments, see Methods. CTSD (73 y.o.; gold particles = 20 nm). LAMP2 (86 y.o.; gold particles = 20 nm). MAP1LC3B (69 y.o.; gold particles = 15 nm). SCARB2 (69 y.o.; gold particles = 15 nm). SNCA (63 y.o.; gold particles = 15 nm). UBA52 (63 y.o.; gold particles = 15 nm). Lipid bodies are indicated by asterisks. NM pigment of the NM- containing organelles appears as black and electron dense granular aggregates. Scale bar in each panel = 250 nm than other lipid molecules identified in both samples, as shown by likely indicates an inhibition of the lysosomal activity inside NM- TLC analysis. In the lipid extracts from both samples, bands co- containing organelles. The presence of mono- and polysialogan- migrating with sphingomyelin, phosphatidylcholine, lactosylcer- gliosides was confirmed by cholera toxin staining after sialidase amide, and phosphatidylethanolamine were identified. In addi- treatment of the aqueous phases obtained from TIS-NM, but not tion, bands corresponding to galactosylceramide and sulfatide from ORG, likely due to the paucity of the sample (Supplementary (some of the typical myelin lipids) were clearly visible in the lipid Fig. 7). We note that in TIS-NM the membranes and any extracts from TIS-NM, particularly sulfatides as also confirmed by component of organelles are removed after sodium dodecyl LC-MS, but not from ORG samples (Fig. 7). In the ORG sample, sulfate treatment, but lipids and especially dolichols are still there were comparable amounts of some sphingolipids (lacto- adsorbed into NM structure, as shown after solvent (methanol and sylceramide and sphingomyelin, a typical myelin lipid) and hexane) extraction of NM and analysis of these lipids extracts by glycerophospholipids (phosphatidylcholine and phosphatidy- LC-MS and TLC. The results indicate a selective affinity of NM lethanolamine) (Fig. 7): since in membranes the sphingolipid pigment for some lipids, particularly dolichols and dolichoic acids, content is usually lower than that of glycerophospholipids, the 3,9 as previously reported. relative increase of sphingolipids vs. glycerophospholipids in ORG npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. 3,42 LC-MS studies on lipids extracts from TIS-NM. This is probably a consequence of membrane disruption during isolation of TIS-NM, so that oxidized dolichols and dolichoic acids present in cytosol, mitochondria and other organelles are released and then adsorbed by NM pigment. However, the presence of dolichols and dolichoic acids was also confirmed by LC-MS in lipid extracts from ORG samples, thus confirming the specific accumulation of this particular class of lipids inside the NM-containing organelles (Fig. 6), mainly in their lipid bodies. Indeed, Fig. 1d shows an electron microscopy image of NM-containing organelles isolated from SN (ORG samples) with many lipid bodies. In addition, the distribution of dolichols and dolichoic acids chain lengths in lipids extracted from ORG samples was similar to that observed in TIS- NM samples: 14–22 isoprene units for dolichols, and 14–21 isoprene units for dolichoic acids. This suggests that artifacts were absent, as the same types of lipids were identified using two different isolation procedures. No known enzymes involved in dolichol metabolism were observed in these organelles, suggesting that oxidation of dolichols on the double bonds with formation of epoxides does not occur inside NM-containing organelles. The conclusion that dolichols are not synthesized within NM-containing organelles is further supported by the absence of two key enzymes required for dolichol synthesis, i.e., dehydrodolichyl diphosphate synthase Fig. 5 WB (for proteins detected by IEM in Fig. 4) performed on SN complex subunit DHDDS (DHDDS) and polyprenol reductase tissue lysates and on ORG samples. For number of WB analyses, see (SRD5A3) which were not detected by LC-MS, WB, or IEM Methods. CTSD (protein content ratio SN tissue lysate/ORG = 3.1). (Supplementary Fig. 5, 6). The band present in both SN tissue lysate (16 pooled tissues, from The presence of dolichols, dolichoic acids, and other lipids we 48 to 85 years of age) and ORG sample (isolated from one subject, describe in the NM-containing organelle has never been reported 81 y.o.) corresponds to the mature CTSD heavy chain which is highly in previous studies. In the past, dolichols, dolichoic acids, and enriched in ORG sample, considering that the total protein content in ORG was 3.1-fold lower than that of SN tissue lysate. LAMP2 other lipids were reported only in isolated NM pigment, a different (protein content ratio SN tissue lysate/ORG = 1.0). LAMP2 protein situation, as they can be adsorbed into NM during isolation and was lightly present in ORG sample (isolated from one subject, 83 y. originate from cytosol and other organelles that are broken during 28,42,43 o.), while in SN tissue lysate (13 pooled tissues, from 62 to 86 years of the isolation process. Thus, in the present study we provide age) this protein is largely expressed. The antibody used here the demonstration of the presence of these lipids in the lipid recognizes all three LAMP2 isoforms. MAP1LC3B (protein content bodies of intact NM-containing organelles. ratio SN tissue lysate/ORG = 1.8). The black arrowhead indicates the MAP1LC3B-I form, which was more prevalent in SN tissue lysate (five pooled tissues, from 73 to 85 years of age) than the MAP1LC3B-II DISCUSSION form (empty arrowhead indicating the phosphatidylethanolamine conjugated form). In ORG sample (isolated from one subject, 77 y.o.), Integrated methodology for NM samples preparations and the use the MAP1LC3B-I form was abundant while MAP1LC3B-II form was of different analytical methods undetectable. SCARB2 (protein content ratio SN tissue lysate/ORG = The study of protein and lipid pathways of NM-containing 2.3). Here we note an enrichment of SCARB2 in ORG sample organelles requires highly purified and well preserved organelles. (isolated from one subject, 77 y.o.) if compared to SN tissue lysate This is challenging because during their preparation, membrane (five pooled tissues, from 73 to 85 years of age), considering that the and soluble proteins can be lost and contamination of organelles total protein content in ORG was 2.3-fold lower than that of SN by proteins or lipids arising from other cellular compartments may tissue lysate. SNCA (protein content ratio SN tissue lysate/ORG = 3.4). The black arrowhead indicates the soluble-monomeric form of occur. Another factor is that human brain tissues used for SNCA which is clearly visible in SN tissue lysate (nine pooled tissues, preparation of organelles are post mortem and thus affected by from 67 to 85 years of age) while undetectable in ORG sample degradation. The preparation of organelles can itself provide a (isolated from one subject, 66 y.o.). Other bands at higher molecular source of changes in the distribution of proteins and lipids, as weight are present in SN tissue lysate, corresponding to fibrils and discussed in the first paragraph of Results. aggregates with possible modifications. In the ORG sample, two A previous report analyzed by LC-MS/MS the NM-containing main bands are clearly visible corresponding to some aggregated/ organelles isolated from SN tissues, but that isolation procedure modified forms of SNCA (at ~50 and ~58 kDa) which are present also in SN tissue lysate. UBA52 (protein content ratio SN tissue lysate/ was different than that used here and started from frozen ORG= 2.7). The black arrowhead indicates the free ubiquitin that is tissues. It is well known that freezing and thawing of post scarcely visible in SN tissue lysate (eight pooled tissues, from 62 to 89 mortem tissues can break the membranes, with leakage of years of age), but abundant in the ORG sample (isolated from one proteins that can diffuse among organelles as a possible source subject, 66 y.o.). The WB also reveals the presence of large number of of contamination. A later study proposed a new centrifugation immunoreactive high molecular weight bands corresponding to method for the combined isolation and enrichment of NM high amounts of poly-ubiquitinated proteins, both in SN tissue lysate granules (i.e., NM-containing organelles) and synaptosomes from and highly enriched in the ORG sample, although the total protein human SN for proteomic analysis, but again this method content in ORG was 2.7-fold lower than that of SN tissue lysate processed frozen tissues. More recently, the same group performed a proteomic analysis on NM-containing organelles The LC-MS results indicate that a high level of dolichols at obtained by laser capture microdissection from human SN frozen different molecular weights (with 14–22 isoprene units) and their slices. The clear advantage of this new methodology is in oxidized derivatives such as dolichoic acids (with 14–21 isoprene isolating NM-containing samples from very low quantities of units) were present in TIS-NM samples, consistent with previous tissues, but the laser capture microdissection lacks sufficient Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Fig. 6 LC-MS analysis of lipids isolated from TIS-NM and ORG samples. The TIS-NM sample here represented was isolated from a pool of seven subjects (from 71 to 85 years of age), while the ORG sample was isolated from two pooled subjects (74 and 89 y.o.). Mass spectra (averaged mass spectra, range 1200–1500 m/z) demonstrate the presence of dolichols species in both samples. We highlight the series of singly charged ions with different chain lengths, corresponding to dolichols with terminal hydroxyl group, their oxidized derivative dolichoic acids, and acetate adducts of dolichols species. Both spectra selectively show dolichol species with chain lengths ranging from 17 to 21 isoprene units, although few dolichol species with lower and higher number of isoprene units were found in lipid extracts from both samples (Results). Abbreviations used in the figure: Dol, dolichol; Dol-Ac, dolichol acetate; Dol-CA, dolichoic acid resolution to discern among different subcellular components that technology method, which is an excellent gel-free approach and are clearly present among NM-containing organelles of the tissue provides improved selectivity and resolution of peptide separa- 1,3,9,10 (Fig. 1a–c and see previous findings) and therefore can tion, with an increased number of identified proteins and better quantitative determinations in complex mixtures. In order to represent a source of contamination. With such a procedure, the collected samples inevitably would contain different types of improve the identification of protein and lipid pathways, and to debris deriving from other cellular components. distinguish proteins and lipids related to different components of Additionally, the LC-MS/MS employed here for proteomic the NM-containing organelle, we analyzed three different NM analysis provides a multidimensional protein identification preparations (ORG, TIS-NM, and ORG-NM): the NM-containing npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. It is noteworthy that the larger mass of our samples in any of the three types of preparations contained the same 35 proteins (corresponding to ∼80 % of the overall SpC of representative proteins detected in all samples, as shown in Fig. 3), thus demonstrating the high reproducibility and pertinence of the detected proteins in the NM-containing organelle. Euler diagrams (Supplementary Fig. 8) show the comparison of our proteomic data with those previously reported by Tribl et al. 23,24 and Plum et al. Considering all the proteins here identified (Supplementary Fig. 8a), there are 50 proteins found also within the 72 proteins (∼69 %) identified by Tribl and colleagues. Among these overlapping proteins, we observed that 36, 32, and 34 proteins belong to ORG, TIS-NM, and ORG-NM respectively, representing ∼6%, ∼8 %, and ∼15 % of all proteins detected in each of the three types of samples (Supplementary Data 1; Table 1). The observation that NM-containing organelles analyzed in the mentioned study had the highest similarity with our NM pigment isolated from organelles (ORG-NM) suggests that samples in that study featured broken membranes and contained mainly proteins strictly bound to NM pigment. In the present study, we isolated the NM pigment from its organelle by intentionally breaking membranes with a freeze/thaw procedure, while Tribl and colleagues isolated the NM-containing organelles from SN frozen tissues. In parallel, the comparison between the list of all the proteins we identified and the list of 1000 proteins reported by Plum and colleagues shows that 188 proteins (∼19 % of proteins identified by Plum’s group) are shared by the two studies (Supplementary Fig. 8a). A more detailed analysis of the overlapping proteins shows that 102, 99, and 71 proteins belong to ORG, TIS-NM, and ORG-NM respectively, representing the ∼18 %, ∼24 %, and ∼32 % Fig. 7 High performance TLC analysis of total lipid extracts obtained of all proteins detected in each of the three types of samples from TIS-NM and ORG samples. The TIS-NM sample here repre- (Supplementary Data 1; Table 1). Also in this case, the ORG-NM sented was isolated from a pool of four subjects (from 62 to 86 years sample shows the highest similarity with the samples analyzed by of age), while lipids of from three ORG samples (each isolated from Plum et al. (Supplementary Data 1). three different subjects, respectively 62, 61 and 77 y.o.) were pooled before loading onto the TLC plates. After separation, lipids were Among new proteins related to NM-containing organelles detected by spraying the TLC plate with anisaldehyde. In TIS-NM detected by LC-MS, here we report some lysosomal proteins that sample the intense spot at the solvent front likely corresponds to were not found in the two previous studies: e.g., iduronate 2- dolichols and dolichoic acids, as confirmed by LC-MS (Fig. 6). The sulfatase, carboxypeptidase Q, lysosomal acid phosphatase, content of sphingomyelin, galactosylceramide, sulfatides (typical lysosomal alpha-mannosidase, and ribonuclease T2. In addition, myelin lipids), and lactosylceramide is higher than phosphatidy- we reported many other proteins of noticeable interest among lethanolamine and phosphatidylcholine (glycerophospholipids). In those never identified before as associated with NM-containing the ORG samples, the main components are again dolichols and organelles: examples include HLA, GPNMB, transmembrane dolichoic acids at the solvent front. In this sample there are comparable amounts of sphingolipids (lactosylceramide and sphin- protein 106B, F-box only protein 11, protein phosphatase 1 gomyelin) and glycerophospholipids. The arrows at the margin of regulatory subunit 1B, intraflagellar transport protein 81 homolog, the image indicate the position of pure standard lipids co- etc. (Supplementary Data 1). Some of these proteins have been chromatographed with the samples, as described in Methods. independently confirmed by IEM and/or WB as discussed below. Abbreviations used in the figure: GalCer, galactosylceramide; GD1a, Finally, it should be noted that there are 43 proteins shared by GD1b, GM1, GT1b, gangliosides GD1a, GD1b, GM1, GT1b; GlcCer, 23,24 our study and other two studies, considering all the proteins glucosylceramide; LacCer, lactosylceramide; PC, phosphatidylcho- we detected (Supplementary Fig. 8a; Supplementary Data 1). If we line; PE, phosphatidylethanolamine; SM, sphingomyelin; ST, evaluate the most enriched proteins detected in all the studies by sulfatides using different samples and methodologies, we should overlap our representative proteins with the 166 significantly over- organelles isolated from SN, NM pigment isolated from SN tissues, represented proteins reported by Plum’s group and with those and NM pigment isolated from NM-containing organelles. The detected by Tribl et al. The result of this evaluation is a small localization of some proteins in NM-containing organelles and group of 18 proteins (Supplementary Fig. 8b), 16 of which are other cellular organelles was also confirmed by IEM in intact SN lysosomal proteins. If we exclude these 16 typical lysosomal tissue slices. proteins, there are two particularly interesting proteins detected in Notably, in the present study, fresh (rather than frozen and all three studies. These two proteins, both primarily assigned to thawed) SN tissue was examined and the isolation procedure endoplasmic reticulum, are phospholipase D3 and protein enabled us to obtain highly purified NM-containing organelles disulfide-isomerase A3, which seem to be closely related to NM- with well preserved membranes and lipid bodies, as demon- containing organelles and are briefly discussed below. strated in electron micrographs (Fig. 1c). The only contaminant rarely observed in ORG preparations were a few red blood cells, The NM-containing organelle is an autophagic lysosome with 47,48 and so their associated previously identified proteins were particular catabolic features subtracted from the proteomic data sets of ORG samples Protein analyses of the three different types of NM-derived (Methods). samples (ORG, TIS-NM, and ORG-NM) revealed that lysosomal Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. proteins are the major class of proteins in the NM-containing principally involved in glycoproteins, glycosaminoglycans and 59,60 organelle, representing ~60 % of overall representative peptides glycosphingolipids degradation pathways in lysosomes, were identified in all samples (Fig. 2; Supplementary Table 1). detected and in very low amounts (overall 49 SpC, <1 % of Comparison of our data with recent lists of defined human lysosomal SpC). Concerning the shortage of enzymes related to 38,49,50 lysosomal proteins indicates good overlap but also some catabolic pathways of lipids in our samples, an exception is acid differences in the distribution of lysosomal enzymatic classes. In ceramidase and its specific prosaposin (which undergoes proteo- particular, the Euler diagrams (Supplementary Fig. 9) and lytic cleavage to form saposins), both involved in the last step of additional table (Supplementary Data 3) show the comparison sphingolipids degradation pathway in lysosomes. Acid cerami- between our data and the study by Sleat and colleagues, which dase and prosaposin were found in NM-containing organelles by is the most detailed human brain proteomic study performed by relative high amount of peptides (overall 225 SpC and 70 SpC, detecting only the soluble resident lysosomal proteins using respectively). mannose 6-phosphate (Man-6-P) as an univocal lysosomal marker. Thus, it appears that NM-containing organelles possess a low Among all the proteins here identified (Supplementary Fig. 9a), we representation of typical components of phospholipids and found 26 of the 48 (∼54 %) confirmed lysosomal soluble proteins sphingolipids degradation pathways. This could indicate that of human brain; if we consider our representative proteins only, NM-containing organelles lose the ability to conduct specific we found 22 of 48 (∼46 %) confirmed lysosomal proteins enzymatic pathways as it accumulates in neurons, and could be (Supplementary Fig. 9b). related to the particular lipid storage content of NM-containing 3,42 It thus appears that peptidases and a majority of esterases are organelles, consisting mainly of dolichols, for which catabolic overrepresented in our samples, while lipases and glycosylases are pathways are still unclear and unrelated to phospholipid/ underrepresented (Supplementary Data 3). In detail, 10 of 14 sphingolipid pathways. On the other hand, our findings are peptidases (E.C. 3.4.-) belonging to human brain lysosomes were consistent with the presence of undegraded glycerophospholipids detected and in large amounts in our samples (with overall 3243 and sphingolipids in the lipid extract from the lipid bodies. SpC, corresponding to ∼41 % of lysosomal SpC), suggesting again an overrepresented complement of protein degradation pathways Lysosomal membrane proteins are less represented in NM- in NM-containing organelles (see also first paragraph of Results). containing organelles than conventional lysosomes The identification of CTSD in its heavy chain (mature form) as Among lysosomal membrane proteins, we detected by different abundant in all samples, revealed by LC-MS data and confirmed techniques SCARB2, LAMP1, LAMP2, CD63 antigen, type 1 by WB and high IEM gold signals, is notable considering its role in phosphatidylinositol 4,5-bisphosphate 4-phosphatase, and some limiting lysosomal storage diseases and in inhibiting SNCA V-type proton ATPase subunits. In particular, SCARB2 was shown aggregation. Similarly, 6 of 11 lysosomal esterases (E.C. 3.1.-) by LC-MS, IEM and WB to be the most abundant lysosomal identified in human brain lysosomes were detected by overall membrane protein in NM-containing organelles. One function of 2523 SpC, corresponding to ∼32 % of lysosomal SpC (Supplemen- SCARB2 may be to transport β-glucocerebrosidase into the tary Data 3). Among these proteins, sialate O-acetylesterase, a key lysosome, although as above β-glucocerebrosidase was not enzyme involved in sialic acid catabolism, was the lysosomal identified in this study by LC-MS. SCARB2 also belongs to the esterase we detected by the highest number of peptides (i.e., scavenger receptor class B family involved in the transport of high overall 2043 SpC). density and low density lipoproteins, cholesterol esters, phospho- In contrast, typical lysosomal human brain enzymes mainly 62,63 lipids and oxidized phospholipids, and if overexpressed, involved in lipids, phospholipids, and sphingolipids catabolism, induces endosomes/lysosomes enlargement and cholesterol including lysosomal acid lipase, group XV phospholipase A2, and accumulation in the enlarged compartments. The abundance sphingomyelin phosphodiesterase were not detected. We of dolichols into the NM-containing organelles could be related to observed only elevated quantities of PLBD2 and phospholipase the presence of high amounts of SCARB2 in NM-containing D3, two poorly characterized proteins of unknown functions, that organelles. This protein is present both on the organelle were identified both as potential proteins of human brain membrane and its lumen, especially in lipid bodies and sometimes lysosomes and in previous studies on NM-containing orga- 23,24 in NM pigment, and is apparently accumulated in the NM- nelles. PLBD2, also confirmed and localized by IEM signals, is a 37,38 containing organelle during aging (see below). new putative lipase with uncertain enzymatic activity, with the Other lysosomal membrane components were identified to a exception of a homolog protein in amoeba that cleaves acyl lesser extent, including LAMP1, LAMP2, CD63 antigen, type 1 chains of some phospholipids (phosphatidylinositol, phosphatidy- phosphatidylinositol 4,5-bisphosphate 4-phosphatase, and some lethanolamine, and phosphatidylcholine). Due to the accumula- subunits of V-type proton ATPase. Considering the probable tion of dolichols in NM-containing organelles, and considering partial loss of membranes during isolation procedures and/or due that dolichols may be transported to lysosomes as dolichyl esters to problems in detecting membrane proteins by LC-MS, IEM and and then hydrolyzed by an unknown dolichyl esterase, we WB analyses were performed, which demonstrated the presence attempted to test by molecular docking if dolichyl esters are of LAMP1, LAMP2, and ATP6V1B2 subunits, mainly in the luminal possible substrate of PLBD2, but obtained no fit (not shown). portion of the NM-containing organelles. The antibody used for Similarly, despite its classification as an esterase, neither a definite LAMP2 detection by IEM and WB recognizes all three LAMP2 enzymatic activity nor specific substrates have been clearly 55,56 isoforms, and so we cannot distinguish between LAMP2A (the reported for phospholipase D3. Nevertheless, phospholipase 55,56 chaperone-mediated autophagy receptor), LAMP2B (probably D3 is abundantly expressed in brain and neural tissues, is involved in macroautophagy), and LAMP2C. Autophagosomes correlated with the modulation of cellular resistance to oxidative possess LAMP2B and LAMP2C isoforms, which have uncertain stress and has been recently indicated as a key factor in the functions distinct from chaperone-mediated autophagy. How- pathological processes of Alzheimer’s disease. The finding of ever, LAMP2 appears to have a low presence with abnormal large amounts of both of these two recently discovered enzymes location in the NM-containing organelle, which could be in all samples indicates a need for further investigation into their consistent with the age-related decreased level of this protein, roles in NM-containing organelles and lysosomes, in particular on particularly for LAMP2A in lysosomes, as well as a general metabolism and storage of lipids. Another group of less represented enzymes are glycosylases: decline of autophagic-lysosomal function that occurs in normal indeed, only 5 of at least 14 enzymes classified as glycosylases and aging. These observations suggest that NM-containing organelles reported by Sleat and colleagues (Supplementary Data 3), are likely derived from macroautophagic organelles (see next npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. paragraph), rather than lysosomes specialized for chaperone- oxidative stress. APOD is consistently upregulated and highly mediated autophagy. expressed during normal aging, in overall SN tissue of PD 75 72 The identification by LC-MS of only low levels of V-type proton patients and other neurodegenerative diseases, where oxida- ATPase functional subunits, responsible for acidification of tive stress and lipid abnormalities are implicated. As mentioned, lysosomes, may indicate a decreased acidification and diminished SCARB2 accumulates in the NM-containing organelle as also lysosomal catabolism. As vacuole fusion requires an electroche- reported for lysosomal inclusion bodies. While SCARB2 is a mical membrane potential created by the V-type proton ATPase, protein of the lysosomal membrane, it was also found on the NM-containing organelles may have a low capacity for fusion with luminal side of the NM-containing organelle particularly in lipid lysosomes or autophagosomes. IEM experiments revealed that bodies and sometimes in NM pigment. Although some of these ATP6V1B2 subunit is not evident on the membrane of NM- proteins may be enriched in the NM-containing organelle due to containing organelles (in contrast to lysosomes) but is sparsely their particular role or because they are normally overexpressed located in NM pigment and lipid bodies, confirming a likely during aging, these proteins may be also accumulated due to functional deficiency of V-type proton ATPase in the NM- impaired degradation in NM-containing organelles. containing organelles. We note that ATP5G1/2/3, that we found in NM-containing organelles, is the primary marker of broad range of neuronal ceroid lipofuscinoses (i.e., CLN2, -3, -4, -5, -6, -7, -8, -9, CLCN7). The NM-containing organelle originates from macroautophagy ATP5G1/2/3 accumulates in autophagic vacuoles and lysosomes of and accumulates MAP1LC3B neurons where autophagy or some lysosomal enzymes are Little is known about the origin of the NM-containing organelle. 76,77 blocked, as in lysosomal storage disorders. Intriguingly, this Experiments in cultured neurons showed that an excess of protein also accumulates inside autophagic vacuoles in normal cytosolic DA induces NM synthesis and NM-containing organelles aged mice, consistent with a decline of autophagic-lysosomal formation, and the induced pigment was chemically identical to function during normal aging. Similar to the age-dependent 2,10 human NM as assessed by electron paramagnetic resonance. In 78,79 accumulation of dolichols in brains of the elderly, which was addition, the identification of a double membrane around these greatly increased in patients with neuronal ceroid lipofuscino- induced organelles of NM and surrounding NM-containing 80,81 sis, ATP5G1/2/3 accumulates in lipofuscin-like lipopigments 2,10 organelles of the human SN, as previously reported and here inside normal neurons during aging, a process amplified in confirmed (Fig. 1c), together with presence of many lysosomal neuronal ceroid lipofuscinosis and other lysosomal disorders. hydrolases, suggests that the NM-containing organelle is a Interestingly, these previously reported observations are consis- pigmented autophagic vacuole. tent with the LC-MS detection of this protein specifically in TIS-NM We have shown in previous studies electron microscopy images samples, which can be associated to the early stages of NM- of NM-containing organelles of human brain that have a clearly containing organelle formation. Additionally, saposins, which were different ultrastructural appearance than lipofuscin and lyso- detected in our samples in the form of prosaposin, have been 2,3,9,10 somes. Another paper described the species-specific ultra- identified as the main storage material (particularly saposins A and structure of neuronal lipofuscin by electron microscopy, thus D) in two types of neuronal ceroid lipofuscinosis, CLN1 and showing morphological features of lipofuscins that are quite CLN10. different from that of NM-containing organelles. Other proteins accumulated inside NM-containing organelles Here we report the presence of the macroautophagy marker are cerebellin-1 and cerebellin-2, which have unknown functions. MAP1LC3B inside NM-containing organelles, confirming its Cerebellin-1 (which we found in all analyzed samples) is autophagic nature. Using IEM on SN tissues, we found MAP1LC3B preferentially expressed in cerebellar synapses, where it is localized around lipid bodies, lining membranes and remarkably required for synapse integrity and plasticity, but is also present on NM pigment within organelles. This evidence confirms that at variable concentrations elsewhere in the brain, and a study NM-containing organelles may derive by formation of MAP1LC3B- reports its presence in the endo-lysosomal compartments of positive autophagosomes that engulf forming NM and related neurons. Cerebellin-1 is more highly expressed in SN (A9) proteins and lipids components present in the cytosol, and neurons than ventral tegmental area (A10) neurons of mice, successively fuse with lysosomes, forming NM-containing auto- suggesting a mechanism for its accumulation similar to that of NM lysosomes that become NM-containing organelles (Fig. 8). in dopaminergic neurons of SN, and is also highly expressed in The level of MAP1LC3B-II is thought to be a reliable indicator of mucopolysaccharidosis type IIIB mouse brains. autophagosome formation. However, in our ORG samples, the Previous studies have suggested the storage features of the MAP1LC3B-II form was undetectable by WB compared to NM-containing organelle: (i) indigestible NM pigment increases in MAP1LC3B-I, either because it was delivered at low levels or concentration in neurons from very early life over the entire life because it was normally degraded by lysosomal enzymes. It is 3–6 3,9,42,43 span; (ii) NM pigment binds high quantities of dolichols; likely that MAP1LC3B-II form could be degraded by the lysosomal and (iii) NM pigment accumulates large amount of endogenous hydrolases we found, as normally occurs after the fusion of our 3,15 and environmental metals. We now report the identification of special autophagosomes with lysosomes (Fig. 8), and therefore storage material in these aged organelles, suggesting that this transient supply of lysosomal enzymes would degrade some mechanisms for degrading protein and lipid components are substrates. Indeed, after autophagosome fusion with lysosomes to impaired, despite the presence of several enzymes inside NM- form autolysosomes, intra-autolysosomal MAP1LC3B-II is normally containing organelles. This evidence, together with low levels of degraded by lysosomal hydrolases and is difficult to detect. some catabolic lysosomal enzymes and shortage of lysosomal membrane proteins (as discussed in the previous sections on The NM-containing organelle is an aged and impaired lysosomal- lysosomal proteins), including V-type proton pump ATPases related organelle that accumulates proteins, indigestible NM, and involved in acidification, suggests that the NM-containing dolichols organelle is an aged impaired lysosomal-related organelle unable In addition to acid ceramidase, other proteins (as previously to completely digest its content. described) including tripeptidyl-peptidase 1 and APOD were This organelle, due to its particular content of undegradable NM present in high amounts in the samples and have been also pigment, proteins, lipids and metals (see below), could be also a reported as major components of lipofuscin-like lysosomal source of oxidative stress (Fig. 8). Therefore, the presence of inclusion bodies. APOD is involved in binding and transport of proteins involved in protection against oxidative stress are lipids, in their protection from oxidation and consequent notable: i.e., phospholipid hydroperoxide glutathione peroxidase, Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. b c β-structured Alpha-crystallin B MAP1LC3B, Proteasome proteins chain, ubiquitin, etc. SQSTM1 3+ a Phagophore Fe Polymerization 3+ Fe formation DA-quinones 3+ Fe Oxidation Macroautophagy 3+ Fe 3+ DA-protein Fe GPNMB, tubulin adduct polymerization- Iron-melanin-protein promoting protein, etc. complex 3+ Fe 3+ DA Fe Lipids Vesicles or other accumulation autophagic vacuoles Autophagosome NM-containing organelle Aging Fusion Lysosome Proteases, NM RAB5A, tubulin Proteins glycosylases, lipases, accumulation polymerization- Autolysosome accumulation proton pumps, etc. promoting protein, etc. h f Fig. 8 Hypothesized scheme summarizing NM-containing organelle formation in human SN. a, b In the cytosol of SN neurons, DA can be oxidized to semiquinones and quinones via iron catalysis, and these highly reactive compounds can react with aggregated and β-structured proteins that accumulate in the cytosol. c The oxidative polymerization of quinones initiates with the formation of the melanin-protein complex which can also bind high levels of metals, especially iron. During this step drugs and toxicants can also bind to the melanin-protein complex. Proteins damaged by misfolding and DA-adducts formation may be recognized and bound by ubiquitins (green) and alpha- crystallin B chain, in the attempt to degrade damaged and/or misfolded proteins in the proteasome pathway. It may be that ubiquitinated- NM-derived products are too large and damaged to be degraded by the proteasome system. d, e The resulting undegradable material accumulates in the cytosol. GPNMB and tubulin polymerization-promoting protein could be engaged at this step, since these proteins are 105,110 involved in the formation of aggresome-like structure and degradation of cellular debris. The accumulated undegradable material is then taken up into autophagic vacuoles by the phagophore, an isolation double membrane which engulfs bulk material for macroautophagy. This is confirmed by the presence of some typical macroautophagic markers such as MAP1LC3B (red) and SQSTM1 (orange), and by the presence of a double membrane surrounding the NM-containing organelle as displayed by electron microscopy (Fig. 1c).f These autophagic vacuoles fuse with lysosomes, shown in the scheme with numerous enzymes (black), different membrane proteins (blue) and the proton pumps (violet), to become autolysosomes containing the enzymes, proteins and lipids of lysosomes. After fusion with lysosomes, the undegraded and NM-derived material contained in the autophagic vacuoles can interact with other lipids and other proteins carried by lysosomes. A decreased lysosomal enzyme efficiency and reduced fusion capacity could occur also as a consequence of aging, oxidative stress and NM accumulation. g These organelles can fuse with other vesicles or with other autophagic vacuoles containing NM precursors or with old NM-containing organelles, etc. This fusion could lead to the accumulation in the lumen of the organelle of membranous portions which would otherwise be degraded, while dolichols in particular are not chemically decomposed and accumulate with other undegraded lipids leading to the formation of lipid bodies. Dolichols (or dolichyl esters) may be transported into the organelle by vesicle transport and membrane fusion. Fusion of organelles could be mediated by RAB5A, tubulin polymerization-promoting protein, microtubule-associated protein tau and other related proteins. This aged organelle, due to its particular content of undegradable NM pigment together with damaged and oxidized proteins, lipids and metals, is a reservoir of buffered toxins (red stars). Under conditions of cellular damage these toxins together with NM pigment could be released and induce neuroinflammation and neurodegeneration. h The final NM-containing organelle is the result of a complex and continuous process occurring during aging, that leads to the accumulation of undegradable material in specialized pigmented “autophagic lysosomes”. Figure modified from previously published papers, by permission of Springer and by permission of Elsevier protein DJ-1, superoxide dismutase [Cu–Zn], and APOD. Here, we of neurons not containing NM or glia. Oligodendrocytes showed briefly highlight two examples. Phospholipid hydroperoxide the highest positive reactions for both FTL and FTH1, while in glutathione peroxidase, which contributes to redox balance in neurons enriched with NM-containing organelles the staining for 5,6 cells, protects lipid membranes from oxidation and was previously FTH1 and FTL was generally undetectable with this technique. observed to co-localize with NM pigment of dopaminergic SN Due to technical limitations of peroxidase immunohistochemistry, neurons. Indeed, this hydroperoxidase was significantly reduced in low quantities of these iron storage proteins could not be overall parkinsonian SN compared to controls, but was increased revealed. By using techniques with high sensitivity (LC-MS and 86 89 in the surviving SN neurons. Protein DJ-1 has been identified as IEM), we now confirm the presence of FTL previously reported an atypical peroxidase that scavenges hydrogen peroxide derived and, to a lesser extent, of FTH1 in NM-containing organelles of the toxicity and mutations in its gene are associated with early-onset SN. In conclusion there are low levels of ferritins inside NM- PD. containing organelles, while the abundant NM pigment is the 3,5,6,14 We investigated FTH1 and FTL, proteins involved in iron major iron storage complex of pigmented neurons. homeostasis/storage, in NM-containing organelles. Using immu- The presence of HLA in NM-containing organelles and its nohistochemistry, we previously reported that FTH1 and FTL accumulation in NM pigment may have important consequences content in NM-containing neurons of SN is much lower than that for preferential vulnerability of catecholaminergic pigmented npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. neurons in SN and locus coeruleus of PD patients. HLA expression Tubulin polymerization-promoting protein, a protein involved in 105,106 is higher in SN and locus coeruleus pigmented neurons than other the maintenance of microtubule network stability, plays also brain neurons, and SN dopaminergic neurons in culture can a role in SNCA aggregation and co-localizes with aggregated SNCA in Lewy bodies inclusions in a group of α-synucleinopa- express HLA that can bind peptides from endogenous or thies. We found tubulin polymerization-promoting protein in exogenous proteins and present them on neuronal membrane + 20,90 ORG and ORG-NM samples, suggesting its possible involvement in leading to targeting by CD8 lymphocytes and death. NM synthesis and NM-containing organelle formation due to its reported role in initiating the formation of cellular aggresome-like Proteins involved in aggregation, degradation pathways, and structures and inclusion bodies. potential precursors of NM synthesis The presence of GPNMB in all isolated samples, as also The presence of abundant UBC and UBA52 in the NM pigment confirmed by WB and IEM, is remarkable. The melanosomal inside the NM-containing organelles, as demonstrated by LC-MS 107,108) protein GPNMB (refs. is found in melanosomes of MNT-1 and IEM, suggests that ubiquitinated proteins likely participate in 100 109 cells by proteomics analysis, participates in melanogenesis, early steps of NM synthesis in the cytosol. In addition, WB data and in control of macroautophagy and bulk degradation in the demonstrate that high levels of ubiquitinated proteins with high cytosol, as a recruiter for MAP1LC3B. GPNMB could play a role molecular weight are detected in the NM-containing organelle. in formation of NM autophagic vacuoles and fusion to produce Ubiquitination can direct proteins, particularly when partially the final NM-containing organelle. Notably, its corresponding unfolded or damaged, to either proteasome or lysosome; if gene is a new PD risk loci reported in a wide association meta- proteasome is unable to degrade all ubiquitinated proteins, analysis. This suggests that mutations of this gene could macroautophagy could provide an important compensatory encode a modified GPNMB protein unable to participate in the 91–93 mechanism. macroautophagic process producing the NM-containing orga- Ubiquitination probably occurs during NM formation attempt- nelle, leaving the neuron exposed to toxic species. A recent gene ing to degrade proteins damaged by DA-modification. Proteins expression study on 6-hydroxydopamine animal models of PD may be modified by DA-quinones, as we have previously detected revealed that GPNMB (as well as other genes belonging to by chemical degradation of NM pigment isolated from SN (TIS- regeneration-associated genes) was highly upregulated in SN NM) high amounts of cysteinyl adducts with DA and with 3,4- early after the lesion. This upregulation could be a response dihydroxyphenylalanine (DOPA), in addition to DA and DOPA. associated with axodegenerative process of SN neurons after This suggests that quinones of DA and DOPA are trapped by lesions, and could also exhibit axoprotective or regenerative cysteine residues of proteins (although histidine residues can properties. similarly react with quinones), producing DA- and DOPA-modified proteins during the early steps of NM biosynthesis. It may be that The lipid bodies of NM-containing organelles contain mainly ubiquitinated-, NM-derived products are too large to be degraded dolichols by the proteasome and therefore are removed by macroauto- The lipid bodies in NM-containing organelles contain mostly phagy and finally stored in the NM-containing organelle (Fig. 8). dolichols and dolichoic acids, although we did not find proteins We further detected SQSTM1, an important partner of MAP1LC3B, for dolichol synthesis or transport, including two important required for the degradation of ubiquitinated aggregates by enzymes for the last steps of synthesis of dolichols (DHDDS and autophagy. Interestingly, SQSTM1 accumulates in ubiquitin-rich SRD5A3). No dolichol-specific transporter has been yet inclusion bodies in neurodegenerative protein aggregation characterized, and only a single study to our knowledge reported 95,96 diseases. The formation of protein inclusion bodies enriched that dolichols intermediates in the human blood are normally in SQSTM1 and ubiquitins is a typical response to stress conditions transported by low-density lipoproteins and may be involved in including amino acid starvation, oxidative stress, and inhibition of their accumulation in lysosomes during normal aging and 94,97,98 lysosomes and autophagy. 114 lysosomal diseases. However, we could not detect the receptor Alpha-crystallin B chain and heat shock protein HSP 90-alpha for low-density lipoproteins as representative in our samples. were found in our samples and could play a role similar to that of Dolichols may exist as free forms with free terminal hydroxyl SQSTM1 and ubiquitins during NM-containing organelle forma- group, or in phosphorylated forms (dolichol phosphate is required tion. Heat shock protein HSP 90-alpha is a chaperone that for glycoprotein biosynthesis), or esterified with fatty acid, which is promotes protein folding and might rescue damaged proteins, 113,115 the dominant form of dolichols in animal tissues. It is and has been detected also in melanosomes. Alpha-crystallin B possible that esters of dolichols with fatty acids (dolichols are chain is a small heat-shock protein that can function as a usually transported to the lysosomes in the esterified form) are molecular chaperone and prevents fibrillization of proteins, transferred into NM-containing organelles, where they could be 31,101 particularly SNCA: it is sometimes present with ubiquitinated slowly hydrolyzed to free dolichols by the abundant hydrolase proteins and SNCA in Lewy bodies, and recently was found activities (esterases) we detected. The intracellular accumulation markedly upregulated in the SN of PD patients. The finding of of dolichyl esters may occur because a majority of lipids are this protein in both the ORG and ORG-NM, but not the TIS-NM degraded by cytosolic, lysosomal, and mitochondrial enzymes, sample, suggests that the protein might not be strictly bound to while dolichols are poorly degraded by unknown pathways and 116,117 NM pigment and is present as a component of the protein matrix unidentified specific catabolic enzymes. It may be that of the NM-containing organelle. dolichols (or dolichyl ester) can be transported into the NM- Protein disulfide-isomerase A3, a protein found only in TIS-NM containing organelle by vesicle transport and membrane fusion sample and then strictly bound to NM pigment, was also reported from other organelles (other NM-containing organelles, lysosomes, by previous proteomic studies as non-lysosomal protein of NM- etc.). Indeed, the largest lipid bodies present into the NM- 23,24 containing organelles. This enzyme catalyzes disulfide bond containing organelles are mainly located on the external portion formation, reduction, or isomerization and belongs to a family of organelle and are membrane bound (Fig. 1a–c), although we responsible for quality-control system to ensure the correct cannot distinguish between a normal bilayer or a single layer; the folding of proteins. Interestingly, a member of this family of smallest lipid bodies are observed both embedded into NM proteins was shown to co-localize with SNCA in brainstem and pigment and in the external portion of the organelle. cortical Lewy bodies of subjects with neurodegenerative Although the precise role of dolichols is not established, they 104 118 diseases. may be important for membrane trafficking, and stimulating Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. fusion of lipid vesicles. It is possible that dolichol-containing high number of cysteine and histidine groups, which readily react membranes continuously fuse during NM-containing organelle with quinones. formation and maturation (NM-containing autophagosomes with It appears that enzymatic (if any) oxidation of DA to form melanic oligomers occurs in the cytosol, likely via iron catalysis or new lysosomes, with old NM-containing organelles, etc.), and by some of the oxidative enzymes (Fig. 8). Evidence for enzymatic consequently, dolichols may accumulate in NM-containing orga- control of the oxidative process leading to NM pigment formation nelles that are unable to efficiently degrade them (Fig. 8). Thus, has been discussed but has not been demonstrated dolichols in the organelle could derive from incompletely 13,14,127,128 directly. Interestingly, we found two oxidoreductase degraded membranes of other organelles. enzymes in TIS-NM sample by LC-MS. The first is the 3 beta- The phospholipids and sphingolipids we found in NM- hydroxysteroid dehydrogenase type 7 (found exclusively in TIS- containing organelles likely derive from the organelle’s own NM sample), which is normally localized in the endoplasmic membranes and incomplete degradation of membranes of reticulum. The other enzyme is the superoxide dismutase [Cu–Zn], autophagic vacuoles that fuse with NM-containing organelles, as normally found in the cytoplasm, which was found in TIS-NM with well as by vesicle transport into NM-containing organelles. SpC higher than that of the previously mentioned enzyme, and Sulfatides are synthesized and accumulated predominantly in was also found with slightly higher SpC as fragment in the ORG oligodendrocytes and are a major component in myelin sheaths, sample. This suggests that superoxide dismutase [Cu–Zn] could be although low amounts of sulfatides have been detected in 120 an additional candidate enzyme involved in the oxidation of DA neurons and astrocytes. The sulfatides accumulated in NM- and related catecholamines during the early phases of NM containing organelles (at very low level in ORG samples compared synthesis. In addition, the existence of an auto-oxidation process to purified TIS-NM, see Results) could derive from disruption of of DA and related catecholamines must be also considered in the oligodendrocytes, dopaminergic and non-dopaminergic axons biosynthesis of NM pigment, a process that to date is only partially and enter the NM-containing organelles through endocytosis, understood. The presence of iron(III) promotes the oxidation of DA vesicle transport and fusion. NM-containing organelles are also into highly reactive quinones that can form NM pigment. characterized by the presence of some typical neuronal sphingo- Indeed, NM synthesis may be driven by an excess of cytosolic lipids, for example gangliosides (as revealed by LC-MS), which catecholamines not accumulated in synaptic vesicles, as sug- likely do not undergo complete degradation due to the loss of 10 gested previously. Moreover, the oxidation of DA to semiqui- activity of glycosyl hydrolases involved in their catabolism, nones and quinones can generate aminochrome and 5,6- consistent with the notion that the NM-containing organelles indolequinone, which can induce toxicity if these species are are a form of aged and impaired lysosomal-related organelle. not taken into NM synthesis. These reactive compounds are reported to cause dysfunction in mitochondria and protein NM synthesis begins in cytosol and intermediate products are degradation, and to promote aggregation of SNCA to toxic transported into organelles protofibrils and produce oxidative damage. The melanic component derived from the oxidative polymer- We did not find significant amounts of typical enzymes and ization of DA and/or other related catecholamines likely binds to proteins involved in melanogenesis (i.e., tyrosinase, dopachrome 100,121,122 β-structured proteins, since X-ray scattering studies performed on tautomerase, melanocyte protein PMEL17, etc.), with the isolated NM pigment showed a diffraction pattern of about 4.7 Å, sole exception of the melanosomal protein GPNMB, a glycoprotein typical of cross-β sheet structured protein aggregates. Since with high homology to the structural melanocyte protein 108,122 GPNMB and SNCA (both capable of forming insoluble fibrils with PMEL17. In addition to the role suggested above for GPNMB cross β-structure, and WB on ORG samples indicate the presence in autophagy to form the NM-containing organelle, this protein of complex aggregates) were found in different parts of the NM- could produce amyloid fibrils assembling into sheets on which containing organelle, including NM itself, we propose that NM DA-quinones would start NM synthesis (Fig. 8). Indeed, it was formation may start with the accumulation of β-structured protein shown that in the absence of glycosylation, the NTR-PKD domain aggregates in the cytosol, resulting in the formation of peptide/ region of GPNMB retains the intrinsic capacity to form amyloid in protein “seeds”, either protofibrils or even fibrils, which would cell cultures. react with excess of DA and/or quinones of DA metabolites, Likewise, we did not find DA transporters, with the sole followed by polymerization to form melanin-protein conjugates. exception of the synaptic vesicular amine transporter which was The protein-DA modifications and subsequent polymerization was detected at very low levels only in ORG-NM sample and 126,132 recently demonstrated in the synthesis of NM models. The categorized as non-representative. However, it was shown in the oxidation of DA and the resulting complex between melanin and past that human DA neurons of SN with high levels of NM many β-structured proteins, which could include fibrillar forms of pigment have low levels of synaptic vesicular amine transporter 133–135 SNCA, GPNMB, and other proteins with cross β-structure, is expression and vice versa. Similarly, the overexpression of promoted by reactive iron which is abundant in cytosol of SN synaptic vesicular amine transporter in neuronal cell cultures, 5,14 neurons. Iron(III) could accumulate within the conjugates treated with L-DOPA which is rapidly converted to DA, inhibited during this process. Indeed, NM pigment in the organelle binds 3,5,14 the synthesis of NM by lowering the cytosolic concentration of iron(III) in two distinct iron-binding sites with different affinity. DA. It is noteworthy that the protein phosphatase 1 regulatory This ternary complex of iron-melanin-protein formed in the subunit 1B, that we found in all samples, is involved in L-DOPA cytosol can be accumulated into autophagic vacuoles and induced dyskinesia in PD. The L-DOPA increases cytosolic DA and transported to the lysosome. Here, the proteases would cleave protein phosphatase 1 regulatory subunit 1B phosphoryla- most protein chains, generating a complex with higher ratio of 124,125 3,9 tion. This protein could be involved in the first steps of melanin:protein in the final NM pigment. This lower polarity NM synthesis by reacting with high cytosolic DA and related melanin:protein complex will more easily react with dolichols, and catecholamines or their derived products (semiquinones and to a lesser extent with other lipids, released by lipid bodies present quinones), due to the presence of cysteines and histidines in its in the organelle to form the insoluble NM pigment which is then structure. Another protein that may react with quinones during continuously accumulated inside membrane-bound organelles. the early steps of NM synthesis could be the cysteine-rich protein This reaction of the complex iron-melanin-protein with dolichols 2, which was detected in ORG-NM samples. More broadly, likely occurs through iron mediated radical oxidation at the candidate proteins that could be incorporated into NM pigment carbon atom adjacent to the double bond in the isoprenic unit of during the early phases of biosynthesis are those containing a dolichols. npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. We found that dolichols are covalently bound to the melanic component of NM pigment. The protein-melanin conjugate thus 3,9,42 component within the NM structure. The isolated NM formed traps iron and other metals, and is then accumulated into pigment is quite insoluble both in water and organic solvents, autophagic vacuoles and carried to lysosomes. Here, the proteases although NM pigment contains a relatively low molecular weight likely cleave most of protein/peptide chains of the protein- component that is slightly soluble in dimethyl sulfoxide with melanin conjugate which react with dolichols to form the final essentially the same composition of the insoluble component. NM. The insoluble portion seems to contain polymers of larger size, more dolichols and less saturated lipids than its dimethyl METHODS sulfoxide-soluble counterpart. The soluble portion of NM pigment likely consists of oligomeric precursors of polymeric NM that have Antibodies molecular weights between 1.4 and 52 kDa. The presence of large For WB and IEM experiments, the following primary antibodies against amounts of oligomeric and polymeric structures further confirms different proteins were used: APOD, mouse monoclonal (Abcam, Cam- that NM synthesis and accumulation are continuous processes bridge, UK); ATP5G1, mouse monoclonal against 18–137 sequence occurring in neuronal organelles (Fig. 8). (Abcam), and this antibody recognizes mature chains of ATP5G1, ATP5G2 and ATP5G3, which are identical in their sequence; ATP6V1B2, mouse monoclonal (Santa Cruz Biotechnologies Inc., Santa Cruz, CA, USA); CTSD, CONCLUSIONS goat polyclonal against the C-term (Santa Cruz Biotechnologies); DHDDS, rabbit polyclonal (Atlas Antibodies AB, Stockholm, Sweden); FTH1, goat The comparison and integration of data from different methods of polyclonal (Abcam); FTL, rabbit polyclonal (Abcam); GPNMB, goat preparation and analysis provide a highly reliable and compre- polyclonal against 23–486 sequence (R&D Systems, Minneapolis, MN, hensive description of protein and lipid pathways of NM- USA); HLA, mouse monoclonal (Santa Cruz Biotechnologies); LAMP1, containing organelle of the human SN. We found that the NM- mouse monoclonal (Developmental Studies Hybridoma Bank, Iowa City, IA, containing organelle possesses particular membrane and soluble USA); LAMP2, mouse monoclonal recognizing all LAMP2 isoforms proteins typical of lysosomes, while other characteristic lysosomal (Developmental Studies Hybridoma Bank); MAP1LC3B, rabbit polyclonal proteins are missing or scarcely expressed. Typical lysosomal (Cell Signaling Technology, Danvers, MA, USA) and rabbit polyclonal enzymes abundant in NM-containing organelles are peptidases, (Abgent, Inc., San Diego, CA, USA); PLBD2, rabbit polyclonal against sulfatases and esterases. However, the organelle has lower levels 448–565 sequence at C-term (Atlas Antibodies AB); RAB5A, rabbit of lipases and glycosylases than typical lysosomes, and so exhibits polyclonal (Abgent, Inc.); SCARB2, mouse monoclonal (Santa Cruz limited degradation pathways for some molecules. The reduced Biotechnologies); SNCA, rabbit polyclonal (EMD Millipore, Temecula, CA, lysosomal activity and fusion capacity could be a consequence of USA); SQSTM1, mouse monoclonal (Abcam); SRD5A3 rabbit polyclonal inadequate localization of some lysosomal membrane proteins, (Atlas Antibodies AB); UBA52, rabbit polyclonal against ubiquitin (DakoCy- tomation, Glostrup, Denmark). We have indicated detailed immunogenic including V-type proton ATPase, which normally acidifies sequence only for proteins that undergo relevant molecule processing. lysosomes. The LAMP2 is at low levels and inadequately located in NM- containing organelles, suggesting that lysosomes specialized for Brain tissues chaperone-mediated autophagy do not form NM-containing This study was approved by the Institutional Review Board of the Institute organelles. In contrast, the presence of double membrane of Biomedical Technologies—National Research Council of Italy (Segrate, surrounding many of the pigmented organelles and the finding Milan, Italy) and was carried out in agreement with the Policy of National of proteins such as MAP1LC3B and SQSTM1 demonstrate the Research Council of Italy. Written informed consents for using brain macroautophagic nature of NM-containing organelles, which samples for research purposes were obtained from closest relatives and derive from autophagosomes that engulf NM precursors, lipids are stored at the Section of Legal Medicine and Insurances, Department of and proteins from cytosol. Biomedical Sciences for Health, University of Milan, Milan, Italy. Final approval was given by the pathologist performing the autopsy. All tissues In lipid bodies of NM-containing organelles, the major samples were analyzed anonymously. components accumulated are dolichols and dolichoic acids, likely In this work, SN samples were obtained during autopsies of male and transported by SCARB2 and APOD. The high accumulation of female subjects who died at different ages without evidence of dolichols in NM-containing organelles may also be mediated by neuropsychiatric and neurodegenerative disorders. The healthy subjects membrane fusion from other organelles and vesicle transport of included in this study at pathological examination did not show any dolichyl esters, and their subsequent hydrolysis to dolichols. macroscopic alteration of neurological and vascular type. Histological The NM-containing organelle accumulates undegradable NM examination (on formalin-fixed and paraffin-embedded tissue sections) pigment, dolichols, lipids, proteins, and metals over the entire revealed no Lewy bodies and other pathological markers. lifespan. Several proteins have been detected in spite of the All samples analyzed and included in this study were obtained from presence of different proteases and other degradative enzymes, as healthy elderly subjects, and the age ranges were the following: (i) ORG degradation processes including those dependent on organelle samples for LC-MS and WB analyses of proteins, LC-MS and TLC analyses of acidification are inhibited or inefficient. Alpha-crystallin B chain lipids (subjects ranging from 61 to 93 years of age); (ii) TIS-NM samples for and heat shock protein HSP 90-alpha were found as well as LC-MS analyses of proteins, LC-MS and TLC analyses of lipids (isolated from pooled SN of subjects ranging from 48 to 92 years of age); (iii) ORG-NM tubulin polymerization-promoting protein, GPNMB, ubiquitins samples for LC-MS analyses of proteins (subjects ranging from 60 to 82 (along with several other proteins), likely consistent with their years of age); (iv) SN tissue lysates for WB analyses of proteins (subjects involvement in macroautophagy and bulk degradation in the ranging from 48 to 89 years of age); and (v) SN tissue slices for IEM process of NM synthesis. (subjects ranging from 63 to 91 years of age). For each type of preparation Among proteins accumulated inside NM-containing organelles, intended for a specific determination, we have used tissues with similar (as a striking presence is that of HLA, that could increase the much as possible) post mortem intervals to use the most reproducible vulnerability of NM-containing neurons, since this protein can conditions. It is noteworthy that all the analyses described in this study present antigens on cell membranes so that neurons would be were performed on sets of samples coming from subjects with overlapping targeted by T-lymphocytes. age ranges. We hypothesize that the NM synthesis starts in the cytosol with Indeed, we were studying brain aging and wanted to investigate elderly accumulation of aggregated and β-structured proteins, which may subjects, and in this age range the amount of NM accumulated in SN was include SNCA, GPNMB, and other proteins, that bind oxidized DA sufficient for our determinations. Furthermore, we employed samples from or DA-derived compounds to produce adducts which undergo subjects in this age range since it was difficult to collect brain samples in a further oxidation and polymerization to form the melanic narrower age range. Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Isolation of NM-containing organelles, NM purified from SN 300 SB-C18, 0.3 i.d. × 5 mm, 5 µm, 300 Å; Agilent Technologies, Santa Clara, CA, USA) for concentration and desalting prior to final separation by tissues, and NM purified from NM-containing organelles reversed phase C18 column (Biobasic-18, 0.180 i.d. × 100 mm, 5 µm, 300 Å; The isolation of intact ORG samples from SN tissue of single subjects, or Thermo Fisher Scientific, Inc.). Peptides were eluted using an acetonitrile rarely from SN of two subjects depending on the availability of brain gradient (eluent A, 0.1 % formic acid in water; eluent B, 0.1 % formic acid in tissues, was performed immediately after SN dissection, in order to avoid acetonitrile): the gradient profile was 5 % eluent B for 3 min, 5–40 % B in freezing and thawing steps of tissues that otherwise would have altered 50 min, 40–80 % B in 10 min, 80–95 % B in 5 min, 95 % B in 10 min. the integrity of ORG samples. The isolation procedure of ORG samples was The flow rate was 100 μl/min, which was split to achieve a final flux of 2 slightly modified and adapted for LC-MS analyses from our previously μl/min. Then, eluting peptides were electrosprayed directly into a hybrid published protocol. TM ion trap-Orbitrap mass spectrometer (LTQ Orbitrap XL ETD; Thermo TIS-NM samples were isolated from pooled SN tissues and prepared as 3,9,42 Fisher Scientific, Inc.), equipped with a nanospray ion source. The spray previously reported. capillary voltage was set at 1.5 kV, and the ion transfer capillary ORG-NM samples, consisting of the NM pigment fraction isolated from temperature was maintained at 220 °C. For each step of peptide elution NM-containing organelles, were prepared from ORG samples and from C18 column, full mass spectra were recorded in the positive ion mode described as follows. Intact ORG samples suspended in their isolation 3 over a 400–1600 m/z range, with a resolving power of 60,000 (full width at buffer were processed with three freeze/thaw cycles (−78 °C/+37 °C) in half-maximum) and a scan rate of 2 spectra/s. This step was followed by order to disrupt organelles membranes. The sample was then centrifuged four low-resolution MS/MS events that were sequentially generated in a (17,500 × g, 15 min, 4 °C) to obtain a pellet containing the NM pigment. data-dependent manner on the top four ions selected from the full MS This pellet was subsequently washed twice in the same buffer, centrifuged spectrum, using dynamic exclusion for the MS/MS analysis. In particular, as described above, and then treated for LC-MS analyses of proteins. the MS/MS scans were acquired by setting a normalized collision energy of 35 % on the precursor ion and, when a peptide ion was analyzed twice, Samples preparation for liquid chromatography-mass applying an exclusion duration of 0.5 min. Mass spectrometer scan spectrometry analyses of proteins functions and high performance liquid chromatography solvent gradients were controlled by the Xcalibur data system version 1.4 (Thermo Fisher Samples analyzed by LC-MS for proteomic analyses were isolated from the Scientific, Inc.). following subjects: (i) two ORG samples isolated from two different subjects (respectively 86 and 69 y.o.); (ii) two TIS-NM samples isolated from pooled SN tissues, the first one from a pool of five subjects (from 72 to 86 Identification of proteins detected by liquid chromatography-mass years of age) and the second one from pooled twelve subjects (from 70 to spectrometry analyses 92 years of age); and (iii) two ORG-NM samples, the first one isolated from Protein identification was carried out by matching experimental spectra to one subject (82 y.o.) and the second one from two pooled subjects (60 and peptide sequences using the SEQUEST database search algorithm, 69 y.o.). contained in BioWorks version 3.3.1 SP1 (University of Washington, TIS-NM samples (dried NM pigments) were hydrated in bi-distilled water licensed to Thermo Fisher Scientific, Inc.) using SEQUEST PC Cluster. under gentle shaking for 3 days in order to obtain a homogeneous For peptide matching, an updated non-redundant human protein suspension in water for tryptic digestion. The resulting TIS-NM suspension sequence database of 276,790 entries, downloaded in January 2009 from was then concentrated (Concentrator 5301; Eppendorf, Hamburg, Ger- the National Center for Biotechnology Information (NCBI) website (http:// many). ORG samples were concentrated in their isolation buffer while ORG- www.ncbi.nlm.nih.gov), was used. In addition, to make a thorough NM was directly treated with RapiGest SF (Waters, Milford, MA, USA) as 23,24,38 comparison of our data with data obtained from other studies and detailed below. to avoid possible bias and false results due to the use of not-aligned In order to dissolve membrane proteins and break NM pigment in all protein codes, the GI accession numbers of identified proteins were samples, a solution of surfactant RapiGest in 100 mM (pH 7.9) ammonium updated to those downloaded in April 2017 from the UniProt repository bicarbonate was added to ORG, TIS-NM and ORG-NM samples according to (http://www.uniprot.org/). Therefore, for these comparisons, we have the manufacturer’s protocol. A final concentration of 0.25 % (v/v) of considered only one time the few proteins we identified in our samples RapiGest was adjusted with 100 mM ammonium bicarbonate (pH 7.9). by two different GI accession numbers but with same UniProt accession Mixtures were then heated at 100 °C for 5 min, cooled to room number. temperature and digested overnight at 37 °C by adding sequencing grade Known abundant contaminating proteins such as keratins, trypsin and modified trypsin (Promega, Madison, WI, USA) at an enzyme/substrate ratio typical abundant proteins of red blood cells (rarely observed as probable of 1:50 (w/w). An additional aliquot of 0.5 µg trypsin was added in the contaminants in ORG samples) were removed prior to the final data morning and digestion was then prolonged for 4 h. The addition of 0.5 % 47,48 analysis, referring to published proteomic data sets of red blood cells. trifluoroacetic acid stopped the enzymatic reaction and subsequent HLA peptides were identified using the updated non-redundant human incubation at 37 °C for 45 min completed the acidic hydrolysis of RapiGest. HLA isoform (HLA gene) database downloaded from NCBI. The water insoluble degradation products were removed by centrifuging This analysis enabled the identification of peptide sequences and at 13,000 × g for 10 min and supernatants containing digested proteins related proteins. Since the confidence of protein identification depends on were desalted using PepClean C-18 spin columns (Pierce Biotechnology, the stringency of the identification of the peptide sequence and peptide Inc., Rockford, IL, USA), concentrated and finally suspended in 20 µl of 0.1 matching, particularly when using data from a single peptide, a high % (v/v) formic acid. stringency was guaranteed by using the following method. The peptide mass search tolerance was set to 1.00 Da; the precursor ion tolerance was Liquid chromatography-mass spectrometry analyses of proteins set to 50 ppm and the intensity threshold was set to 100. Moreover, all Three different types of samples (ORG, TIS-NM, ORG-NM) were prepared searches were performed with no enzyme and in order to assign a final from SN tissues (Supplementary Fig. 1): two samples for ORG, two samples score to the proteins, the SEQUEST output data were filtered by setting the −3 for TIS-NM, and two samples for ORG-NM as described in the previous peptide probability to 1 × 10 , the minimum correlation score values paragraph. Each of the two ORG samples was injected three times for LC- (Xcorr) was chosen greater than 1.5, 2.0, 2.5, and 3.0 for single-, double-, MS analysis, each of the two TIS-NM samples was injected three times, triple-, and quadruple-charged ions respectively, and the consensus score while one ORG-NM sample was injected two times and the other one was higher than 10. False-positive peptides ratio, calculated through reverse injected only one time due to its scarce amount. database, was less than 5 %. For decoy searches a reversed version of the The multidimensional protein identification technology (MudPIT) target human protein database was generated using the reverse database 46,136 system is a 2DC-MS/MS platform, composed of a two dimensional function in Bioworks 3.3.1 software (Thermo Fisher Scientific, Inc.). The micro-high performance liquid chromatography system (Surveyor HPLC; number of peptides, notably the SpC, detected by MS/MS was utilized as Thermo Fisher Scientific, Inc., San Jose, CA, USA) coupled online to a mass indicator of relative protein abundance. The group of representative spectrometer, using ProteomeX-2 configuration (Thermo Fisher Scientific, proteins was selected considering proteins detected by SpC ≥ 2 as average Inc.). Briefly, digested peptide mixtures were loaded onto a capillary strong value in at least one of the three types of samples. cation exchange column (Biobasic-SCX column, 0.32 i.d. × 100 mm, 5 µm; The area-proportional Euler diagrams were calculated using the Thermo Fisher Scientific, Inc.) and eluted stepwise with ammonium eulerAPE drawing tool (software at http://www.eulerdiagrams.org/ chloride injections of increasing molarity (5, 10, 15, 20, 30, 40, 80, 120, 400, eulerAPE/). Analyses of proteins for Euler diagram calculation in ORG, 600, 700 mM). Fractions were captured in turn onto peptide traps (Zorbax TIS-NM, and ORG-NM, as shown in Fig. 3, were performed by using NCBI npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. accession (GI number). Euler diagram calculation in the comparison of our sample from pooled four subjects (from 62 to 86 years of age); (ii) three list of proteins with those previously reported for NM-containing ORG samples, each of which isolated from one subject (respectively 62, 61, 23,24 38 organelles of human SN and for human brain lysosomes were and 77 y.o.). performed by using UniProt accession number. Total lipids extracts for TLC analyses were prepared from TIS-NM 3,9,42 Individual cellular location was assigned to each protein according to samples as previously described by using methanol and hexane. the GOA database (http://geneontology.org/), the UniProt database (http:// Additionally, these TIS-NM lipids extracts dried under nitrogen stream were www.uniprot.org/) and authors’ PubMed search. It should be noted that then resuspended in chloroform-methanol-water (2:1:0.1, v/v/v) and subjected to a two-phase Folch’s partitioning, resulting in the separation some proteins may have multiple cellular location: in these cases the most of an aqueous phase containing gangliosides and an organic phase typical and representative cellular location was manually assigned. We containing all other lipids, including glycerophospholipids, neutral glyco- were unable to assign the cellular location for some proteins which were sphingolipids and sphingomyelin. therefore classified as “unknown cell location” and additional proteins The ORG samples in few µl of isolation buffer were snap frozen and then without a complete characterization at the moment of data analysis were lyophilized. Lipids were extracted with chloroform-methanol-water (2:1:0.1, classified as “uncharacterized proteins”. v/v/v). Then, total lipid extracts were subjected to a two-phase Folch’s The names of genes and proteins described in the main text and supplementary files were retrieved from updated UniProt and UniParc partitioning. Following the two-phase partitioning, organic phases deriving from total database (April 2017), and we adopt as the protein name symbol that of the approved gene symbol. lipids extracted from both TIS-NM and ORG samples were analyzed by TLC, loading equivalent amounts of lipids. Lipids were separated by mono- dimensional high performance TLC silica gel using chloroform-methanol- Samples preparation for liquid chromatography-mass 0.2 % aqueous calcium chloride (60:35:8, v/v/v) as a solvent system. After spectrometry analysis of lipids separation, lipids were detected by spraying the TLC plates with Samples analyzed by LC-MS for identification of lipids were isolated from anisaldehyde, a reagent for the general detection of lipids. Identification the following subjects: (i) two TIS-NM samples isolated from pooled SN of lipids after separation and chemical detection was assessed by co- tissues, the first one isolated from a pool of seven subjects (from 71 to 85 migration with lipid standards (Avanti Polar Lipids, Inc., Alabaster, AL, USA; years of age) and the second one from pooled four subjects (from 82 to 85 Sigma-Aldrich Co., St. Louis, MO, USA; some standard were synthesized in years of age); (ii) three ORG samples, two of which isolated from two laboratories of the Department of Medical Biotechnology and Translational different subjects (respectively 81 and 89 y.o.) and the third one from two Medicine, University of Milan, Segrate, Italy). pooled subjects (74 and 89 y.o.). Cholera toxin staining of lipid present in the aqueous phases was The extraction of total lipids from TIS-NM samples was performed by performed as described hereafter. Briefly, after chromatographic running 3,9,42 using methanol and hexane as previously reported. The organic using chloroform-methanol-0.2 % aqueous calcium chloride (50:42:11, v/v/ fractions (methanol and then hexane) derived from NM pigments were v) as a solvent system, the TLC plate was well dried and fixed with a combined and dried under nitrogen flow. polyisobutylmethacrylate solution prepared dissolving 1.3 g of polyisobu- The total lipids were extracted from ORG samples with methanol and tylmethacrylate in 10 ml chloroform and diluting 8 ml of this solution with hexane, as for TIS-NM samples. The ORG samples in few µl of isolation 42 ml of hexane. The TLC plate was immersed three times in this solution buffer were resuspended in about 0.5 ml of methanol and then centrifuged and then allowed to dry for 1 h. The dried TLC was soaked for 30 min in (1000 × g, 30 min, 20 °C). The supernatant was collected and the remaining 0.1 M Tris-hydrochloride (pH 8.0), 0.14 M sodium chloride containing 1 % pellet was resuspended in about 0.5 ml of hexane, and subsequently bovine serum albumin. The TLC was then incubated with Clostridium centrifuged as described above. Again, organic fractions (methanol and perfringens sialidase (0.12 U/ml in 0.05 M acetate buffer pH 5.4 and 4 mM then hexane) extracted from ORG samples were finally combined and calcium chloride) overnight at room temperature. Afterward, the TLC was dried under nitrogen flow. incubated with biotin-conjugated cholera toxin subunit B (10 μg/ml; Sigma-Aldrich Co.) in phosphate-buffered saline containing 1 % bovine serum albumin for 1 h, and subsequently for 1 h with horseradish Liquid chromatography-mass spectrometry analysis of lipids peroxidase-conjugated streptavidin (2 μg/ml; Sigma-Aldrich Co.) in the For LC-MS analysis, the dried lipids extracted from TIS-NM and ORG same solution. After several washings with phosphate-buffered saline, the samples were dissolved in 100 µl chloroform and diluted 1:1 (v/v) in eluent TLC plate was developed for 5 min with o-Phenylenediamine dihydrochlor- A, as described below. Typically, 5 µl of sample were injected on a C8 ide substrate (Sigma-Aldrich Co.), 1 tablet in 50 ml citrate-phosphate buffer reversed phase column (BioBasic C8, 100 × 0.18 mm, 5 µm, 300 Å; Thermo (pH 5.0) and 20 μl hydrogen peroxide. Fisher Scientific, Inc.), using a micro liquid chromatography system (Surveyor HPLC; Thermo Fisher Scientific, Inc.) coupled to a mass TM Electron microscopy spectrometer with Orbitrap mass analyzer (Exactive Plus; Thermo Fisher Transmission electron microscopy for morphological evaluation was Scientific, Inc.), equipped with a nanospray ion source. Liquid chromato- performed on SN tissue slices and isolated ORG samples prepared graphy was operated at a flow rate of 100 μl/min, split in order to achieve a according to our previously published method. final flux of 2 μl/min, with a gradient as follows: 100 % eluent A was held IEM experiments were carried out on SN tissue blocks which were fixed isocratically for 10 min, then linearly increased to 100 % B over 30 min and for 2 h at 4 °C in a mixture of 4 % paraformaldehyde and 0.25 % held at 100 % B for 12 min. Eluent A consisted of methanol-acetonitrile- glutaraldehyde in cacodylate buffer (0.12 M, pH 7.4). Tissue samples were aqueous 1 mM ammonium acetate (60:20:20, v/v/v). Eluent B consisted of extensively washed with cacodylate buffer, dehydrated in a graded isopropanol-acetonitrile (90:10, v/v). ethanol series and then embedded in LRW resin. Ultra thin sections Nanospray was achieved using a coated fused silica emitter (360 µm o. (80 nm) were prepared using a ultramicrotome (Leica Ultracut; Leica d./50 µm i.d. 730 µm tip i.d.; New Objective, Inc., Woburn, MA, USA) held to Microsystems GmbH, Wien, Austria), collected over nickel grids and 1.5 kV. The heated capillary was held at 220 °C. Full mass spectra were incubated for 90 min at room temperature with primary antibodies diluted recorded in negative and positive ion mode over a 400–2000 m/z range. in phosphate-buffered saline (pH 7.4). The concentrations of primary The resolution was set to 100,000 with 2 microscans/s, restricting the antibodies used for IEM experiments were the following: APOD (1:50), Orbitrap loading time to a maximum of 50 ms with a target value of 5E5 ATP5G1 (1:100), ATP6V1B2 (1:100), CTSD (1:100), DHDDS (1:200), FTH1 ions (ultimate mass accuracy mode). (1:250), FTL (1:300), GPNMB (1:100), HLA (1:200), LAMP1 (1:20), LAMP2 Manual interpretation of mass spectra was performed to evaluate the (1:20), MAP1LC3B (1:50 for antibody from Cell Signaling Technology; 1:25 major lipid species present in the analyzed lipid extracts. for antibody from Abgent Inc.), PLBD2 (1:100), RAB5A (1:100), SCARB2 (1:50), SNCA (1:200), SQSTM1 (1:30), SRD5A3 (1:25), and UBA52 (1:250). The Lipids extraction and partitioning for thin-layer chromatography grids were then washed with phosphate-buffered saline and incubated for analyses of lipids 60 min with gold-conjugated secondary antibodies (British Biocell Inter- Samples analyzed by TLC for identification and semiquantitative assess- national, Cardiff, UK) diluted 1:100 in phosphate-buffered saline. After ment of lipids were isolated from the following subjects: (i) four TIS-NM extensive washing, the grids were post fixed with 1 % glutaraldehyde in samples isolated from pooled SN tissues, the first three samples isolated cacodylate buffer. Samples were then stained with saturated uranyl acetate from a pool of five subjects each (respectively from 48 to 85 years of age, in water for 5 min, washed and then stained with 3 mM lead citrate for from 67 to 85 years of age, from 73 to 85 years of age) and the fourth 5 min. Finally, the sections were photographed using a transmission Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. electron microscope LEO 912AB (Carl Zeiss AG, Oberkochen, Germany) was confirmed by re-incubating the membranes with the chemilumines- equipped with a Proscan CCD camera (ProScan, Lagerlechfeld, Germany) cent substrate. controlled by EsivisionPro software (Soft Imaging System, Münster, The WB analyses in this study are qualitative rather than quantitative. Germany). Electron microscopy was performed at the Advanced Light For the electrophoresis separation of proteins in the samples, we kept an and Electron Microscopy BioImaging Center—San Raffaele Scientific empty lane between lanes containing SN tissue lysates and those Institute. containing ORG samples in order to avoid contamination leading to false Each protein was investigated by IEM in at least one to three samples positive results. For the preparation of figures, we divided the two lanes of deriving from different subjects, depending on SN tissue availability. Data interest by a thin white line between juxtaposed lanes. It should be noted of subjects analyzed by IEM are reported in the legends of figures chosen that: (i) ORG samples were loaded at high volumes, due to their low total as representative of all electron microscopy experiments. Each IEM protein concentration with respect to SN tissue lysates, and because the experiment was routinely checked for negative controls as described concentration procedure of ORG sample produced an insoluble and below: (i) the reaction was performed on SN tissue slices without either untreatable residue from which proteins could be no longer recovered primary or secondary antibody to confirm the absence of gold particles in (high salt composition of the isolation buffer, particular nature of the NM the examined sections; (ii) in the presence of primary and secondary pigment contained in ORG samples, etc.); (ii) the granular composition of antibodies, the absence of gold particles was verified in the resin devoid of the NM pigment present in ORG samples could interfere with the initial tissue; and (iii) the most relevant control we did was to check the absence electrophoresis run in the stacking gel, thus leading to lateral diffusion of of signal in cellular/tissue components where it was not expected. Only proteins; and (iii) the contamination of proteins from SN tissue lysate after passing these three negative controls, we evaluated the signal in NM- (largely rich of proteins) to ORG sample was less probable but could not be containing organelles which we judged as specific even if represented by excluded and lead as well to false positive results in ORG samples. The total few gold particles. amounts of proteins loaded into each lane of the gel were then chosen depending on the availability of the sample (particularly ORG sample), and Western blotting analyses of tissue lysates and NM-containing on the content of each selected protein in both SN tissue lysates and ORG organelles samples. The ratios of protein content between SN tissue lysates and ORG SN tissues were homogenized and sonicated with five volumes of lysis samples for each protein analyzed by WB are reported in the figure buffer as described above. In case of soluble proteins, the lysis buffer was legends. Note that for ATP6V1B2, GPNMB, and RAB5A proteins, the WB composed of 50 mM Tris-hydrochloride (pH 7.5), 150 mM sodium chloride, analyses were performed on SN tissue lysates and ORG samples on 0.1 % sodium dodecyl sulfate, 1 % Triton® X-100, and 1 % protease different days, due to the lack of either SN tissues or ORG samples, inhibitor cocktail (Sigma-Aldrich Co.). For membrane-bound proteins, the depending on human brain tissues availability, and it was sometimes lysis buffer was composed of phosphate-buffered saline (pH 7.4), 0.1 % impossible to run all types of samples at the same time. Relative lanes sodium dodecyl sulfate, 1 % Triton® X-100, and 1 % protease inhibitor where then juxtaposed as previously explained. The use of WB on SN tissue cocktail. After short sonication cycles in ice, SN homogenates were lysates was chiefly employed to assess the specificity of each antibody for centrifuged at 17,500 × g, 4 °C (30 min for soluble proteins, 10 min for the target protein as essential information for IEM experiments, and to membrane-bound proteins) and the supernatants were collected. From verify its presence or absence in the ORG samples (n ≥ 3 for each protein). supernatants, proteins were precipitated by acid treatment to remove components of buffers that could influence protein quantitation. Then, Data of subjects from which SN tissue lysates and ORG samples were Lowry method was performed on the protein pellets to measure the total prepared are reported in the legends of figures chosen as representative of protein concentration in both SN tissue lysates and ORG samples. all WB experiments performed on different samples. For electrophoresis, samples were heated in reducing sample buffer and then loaded onto 0.75 mm thick polyacrylamide mini gels (different pore Data availability sizes, depending on the desired separation of proteins) and separated by one-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophor- All relevant data are within the paper and its Supporting Information files. esis (Mini-PROTEAN® 3 Cell; Bio-Rad Laboratories, Inc., Hercules, CA, USA). However, raw data supporting the results reported in this article are Prestained protein markers (New England BioLabs, Inc., Ipswich, MA, USA) available upon request. were used as size standards in the electrophoresis. Proteins were then over-night transferred (Mini Trans-Blot® Electrophoretic Transfer Cell; Bio- Rad Laboratories, Inc.) to 0.45 µm supported nitrocellulose membranes ACKNOWLEDGEMENTS (Hybond®-C Extra; Amersham Biosciences, Little Chalfont, UK). Ponceau-S This work was supported by Italian Ministry of Education, University, and Research staining of membranes was performed to check the efficiency of proteins (MIUR)—National Research Programme (PNR)—National Research Council of Italy transfer. (CNR) Flagship “InterOmics” Project (PB.P05), by MIUR—PNR—CNR Aging program For immunoblotting, membranes were blocked for 2 h at room 2012–2014, by MIUR—Medical Research in Italy (MERIT) Project RBNE08ZZN7, by temperature with fatty acid/globulin-free bovine serum albumin (Sigma- Lombardy Region—CNR 2013–2015 Framework Agreement MbMM Project (18089/ Aldrich Co.) or skimmed milk, with concentrations ranging from 3 to 8 %, RCC) and by MIUR—Research Projects of National Interest (PRIN) 2015 Prot. depending on the properties of investigated protein, antibody specificity, 2015T778JW. D.S. also thanks the National Parkinson’s, Parkinson’s Disease and the etc., in phosphate-buffered saline (pH 7.4) with 0.1 % Tween® 20. Then JPB Foundations (Miami, FL and New York, NY, USA) for support. D.S. is a NARSAD membranes were incubated with primary antibodies diluted in blocking solutions, while temperature and time of incubation varied according to Brain and Behavior Distinguished Investigator. The funders had no role in study antibodies specificity. The concentrations of primary antibodies used for design, data collection and analysis, decision to publish, or preparation of the WB experiments were the following: APOD (1:300), ATP5G1 (1:100), manuscript. The authors thank Dr. Maria Carla Panzeri (Advanced Light and Electron ATP6V1B2 (1:400), CTSD (1:500), DHDDS (1:1000), FTH1 (1:3500), FTL Microscopy BioImaging Center—San Raffaele Scientific Institute) for expert and (1:2000), GPNMB (1:1500), LAMP1 (1:50), LAMP2 (1:200), MAP1LC3B (1:600), invaluable assistance in electron microscopy imaging. Authors gratefully acknowl- PLBD2 (1:300), RAB5A (1:1000), SCARB2 (1:600), SNCA (1:1500), SQSTM1 edge Dr. Simona Prioni (Department of Medical Biotechnology and Translational (1:300), SRD5A3 (1:100), and UBA52 (1:200). After washing, membranes Medicine, University of Milan, Segrate, Italy) for her skillful assistance in TCL analyses were incubated for 90 min at room temperature with horseradish of lipids. Authors are also thankful to Dr. Luciano Milanesi and Dr. Pasqualina D’Ursi peroxidase-labeled secondary antibodies with a concentration range for molecular docking of PLBD2 protein, to Dr. Eleonora Franceschi for her excellent 1:2000–1:5000 in block solution (Jackson ImmunoResearch, Inc., West technical support in LC-MS analysis of lipids, and to Dr. Alice Valmadre for her Grove, PA, USA; EMD Millipore). After extensive washing, the immunor- assistance during some WB analyses; additionally, authors thank Ms. Loredana eactive species were visualized by SuperSignal® West Pico chemilumines- Ansalone and Ms. Mariagiovanna Torti for their assistance in the preparation of cent substrate as described in the manufacturer’s protocol (Pierce financial reports of the above mentioned grants (all these collaborators are affiliated Biotechnology, Inc.). The films were exposed for few seconds up to with Institute of Biomedical Technologies—National Research Council of Italy). L.Z. 15 min. As a control for aspecific binding, the same procedure was also thanks the Grigioni Foundation for Parkinson’s Disease (Milan, Italy). Finally, all executed without primary antibody. When multiple blotting of membranes the authors gratefully thank the Section of Legal Medicine and Insurances, was necessary, as in case of loading control with CTSD antibody, the TM Restore PLUS Western Blot Stripping Buffer (Pierce Biotechnology, Inc.) Department of Biomedical Sciences for Health at University of Milan for providing was used according to manufacturer’s protocol, and successful stripping brain tissue samples. npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. AUTHOR CONTRIBUTIONS 21. Oberländer, U. et al. Neuromelanin is an immune stimulator for dendritic cells in vitro. BMC Neurosci. 12, 116 (2011). Conceived and designed the experiments: F.A.Z., R.V., P.M., L.Z. Performed the 22. Double, K. L. et al. Structural characteristics of human substantia nigra neuro- experiments: F.A.Z., R.V., F.A.C., C.B., A.D.P., S.G. Analyzed the data: F.A.Z., R.V., A.D.P., melanin and synthetic dopamine melanins. J. Neurochem. 75,2583–2589 D.D.S., P.M., A.P. Contributed reagents/materials/analysis tools: P.M., A.P., L.Z. Wrote (2000). the paper: F.A.Z., R.V., A.P., L.C., D.S., L.Z. 23. Tribl, F. et al. “Subcellular proteomics” of neuromelanin granules isolated from the human brain. Mol. Cell. Proteom. 4, 945–957 (2005). 24. Plum, S. et al. Proteomic characterization of neuromelanin granules isolated ADDITIONAL INFORMATION from human substantia nigra by laser-microdissection. Sci. Rep. 6, 37139 (2016). Supplementary information accompanies the paper on the npj Parkinson’s Disease 25. Han, X. Multi-dimensional mass spectrometry-based shotgun lipidomics and the website (https://doi.org/10.1038/s41531-018-0050-8). altered lipids at the mild cognitive impairment stage of Alzheimer’s disease. Biochim. Biophys. Acta 1801, 774–783 (2010). Competing interests: The authors declare no competing interests. 26. Licker, V. et al. Proteomic analysis of human substantia nigra identifies novel candidates involved in Parkinson’s disease pathogenesis. Proteomics 14, Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims 784–794 (2014). in published maps and institutional affiliations. 27. Plum, S. et al. Proteomics in neurodegenerative diseases: Methods for obtaining a closer look at the neuronal proteome. Proteom. Clin. Appl. 9, 848–871 (2015). 28. Zecca, L. et al. Interaction of human substantia nigra neuromelanin with lipids REFERENCES and peptides. J. Neurochem. 74, 1758–1765 (2000). 29. Engelen, M. et al. Proteomic profile of neuromelanin and neuromelanin- 1. Duffy, P. & Tennyson, V. M. Phase and electron microscopic observations of containing organelles. Abstract at “5th Congress of the Portuguese Proteomics Lewy bodies and melanin granules in the substantia nigra and locus coeruleus Network—ProCura 1st International Congress on Analytical Proteomics—ICAP”, in Parkinson’s disease. J. Neuropathol. Exp. Neurol. 24, 398–414 (1965). Sep 30th–Oct 3rd, (Caparica, Portugal, 2009). 2. Sulzer, D. et al. Neuronal pigmented autophagic vacuoles: lipofuscin, neuro- 30. Lindersson, E. et al. p25alpha Stimulates alpha-synuclein aggregation and is co- melanin, and ceroid as macroautophagic responses during aging and disease. J. localized with aggregated alpha-synuclein in alpha-synucleinopathies. J. Biol. Neurochem. 106,24–36 (2008). Chem. 280, 5703–5715 (2005). 3. Zecca, L. et al. New melanic pigments in the human brain that accumulate in 31. Rekas, A. et al. Interaction of the molecular chaperone alphaB-crystallin with aging and block environmental toxic metals. Proc. Natl Acad. Sci. USA 105, alpha-synuclein: effects on amyloid fibril formation and chaperone activity. J. 17567–17572 (2008). Mol. Biol. 340, 1167–1183 (2004). 4. Fedorow, H. et al. Evidence for specific phases in the development of human 32. Letournel, F., Bocquet, A., Dubas, F., Barthelaix, A. & Eyer, J. Stable tubule only neuromelanin. Neurobiol. Aging 27, 506–512 (2006). polypeptides (STOP) proteins co-aggregate with spheroid neurofilaments in 5. Zecca, L. et al. The role of iron and copper molecules in the neuronal vulner- amyotrophic lateral sclerosis. J. Neuropathol. Exp. Neurol. 62, 1211–1219 (2003). ability of locus coeruleus and substantia nigra during aging. Proc. Natl Acad. Sci. 33. Morris, M., Maeda, S., Vossel, K. & Mucke, L. The many faces of tau. Neuron 70, USA 101, 9843–9848 (2004). 410–426 (2011). 6. Zucca, F. A. et al. Neuromelanin and iron in human locus coeruleus and sub- 34. Halliday, G. M. et al. Alpha-synuclein redistributes to neuromelanin lipid in the stantia nigra during aging: consequences for neuronal vulnerability. J. Neural substantia nigra early in Parkinson’s disease. Brain 128, 2654–2664 (2005). Transm. (Vienna) 113, 757–767 (2006). 35. Rochet, J. C. et al. Interactions among alpha-synuclein, dopamine, and bio- 7. German, D. C. et al. Disease-specific patterns of locus coeruleus cell loss. Ann. membranes: some clues for understanding neurodegeneration in Parkinson’s Neurol. 32, 667–676 (1992). disease. J. Mol. Neurosci. 23,23–34 (2004). 8. Hirsch, E., Graybiel, A. M. & Agid, Y. A. Melanized dopaminergic neurons are 36. Jalanko, A. & Braulke, T. Neuronal ceroid lipofuscinoses. Biochim. Biophys. Acta differentially susceptible to degeneration in Parkinson’s disease. Nature 334, 1793, 697–709 (2009). 345–348 (1988). 37. Jensen, A. G. et al. Biochemical characterization and lysosomal localization of 9. Engelen, M. et al. Neuromelanins of human brain have soluble and insoluble the mannose-6-phosphate proteinp76 (hypothetical protein LOC196463). Bio- components with dolichols attached to the melanic structure. PLoS ONE 7, chem. J. 402, 449–458 (2007). e48490 (2012). 38. Sleat, D. E., Zheng, H., Qian, M. & Lobel, P. Identification of sites of mannose 6- 10. Sulzer, D. et al. Neuromelanin biosynthesis is driven by excess cytosolic cate- phosphorylation on lysosomal proteins. Mol. Cell. Proteom. 5, 686–701 (2006). cholamines not accumulated by synaptic vesicles. Proc. Natl Acad. Sci. USA 97, 39. Hemmings, H. C. Jr, Greengard, P., Tung, H. Y. & Cohen, P. DARPP-32, a 11869–11874 (2000). dopamine-regulated neuronal phosphoprotein, is a potent inhibitor of protein 11. Kastner, A. et al. Is the vulnerability of neurons in the substantia nigra of patients phosphatase-1. Nature 310, 503–505 (1984). with Parkinson’s disease related to their neuromelanin content? J. Neurochem. 40. Deinhardt, K. et al. Rab5 and Rab7 control endocytic sorting along the axonal 59, 1080–1089 (1992). retrograde transport pathway. Neuron 52, 293–305 (2006). 12. Liang, C. L., Nelson, O., Yazdani, U., Pasbakhsh, P. & German, D. C. Inverse 41. Boggs, J. M. Role of galactosylceramide and sulfatide in oligodendrocytes and relationship between the contents of neuromelanin pigment and the vesicular CNS myelin: formation of a glycosynapse. Adv. Neurobiol. 9, 263–291 (2014). monoamine transporter-2: human midbrain dopamine neurons. J. Comp. Neurol. 42. Ward, W. C. et al. Identification and quantification of dolichol and dolichoic acid 473,97–106 (2004). in neuromelanin from substantia nigra of the human brain. J. Lipid Res. 48, 13. Zucca, F. A. et al. Neuromelanin of the human substantia nigra: an update. 1457–1462 (2007). Neurotox. Res. 25,13–23 (2014). 43. Fedorow, H. et al. Dolichol is the major lipid component of human substantia 14. Zucca, F. A. et al. Interactions of iron, dopamine and neuromelanin pathways in nigra neuromelanin. J. Neurochem. 92, 990–995 (2005). brain aging and Parkinson’s disease. Prog. Neurobiol. 155,96–119 (2017). 44. Stingl, C., Söderquist, M., Karlsson, O., Borén, M. & Luider, T. M. Uncovering 15. Bohic, S. et al. Intracellular chemical imaging of the developmental phases of effects of ex vivo protease activity during proteomics and peptidomics sample human neuromelanin using synchrotron X-ray microspectroscopy. Anal. Chem. extraction in rat brain tissue by oxygen-18 labeling. J. Proteome Res. 13, 80, 9557–9566 (2008). 2807–2817 (2014). 16. Karlsson, O., Berg, C., Brittebo, E. B. & Lindquist, N. G. Retention of the cyano- 45. Plum, S. et al. Combined enrichment of neuromelanin granules and synapto- bacterial neurotoxin beta-N-methylamino-l-alanine in melanin and somes from human substantia nigra pars compacta tissue for proteomic ana- neuromelanin-containing cells--a possible link between Parkinson-dementia lysis. J. Proteom. 94, 202–206 (2013). complex and pigmentary retinopathy. Pigment Cell Melanoma Res. 22, 120–130 46. Di Silvestre, D., Brambilla, F. & Mauri, P. L. Multidimensional protein identification (2009). technology for direct-tissue proteomics of heart. Methods Mol. Biol. 1005,25–38 17. Karlsson, O. & Lindquist, N. G. Melanin affinity and its possible role in neuro- (2013). degeneration. J. Neural Transm. (Vienna) 120, 1623–1630 (2013). 47. Pasini, E. M. et al. In-depth analysis of the membrane and cytosolic proteome of 18. Karlsson, O. & Lindquist, N. G. Melanin and neuromelanin binding of drugs and red blood cells. Blood 108, 791–801 (2006). chemicals: toxicological implications. Arch. Toxicol. 90, 1883–1891 (2016). 48. Roux-Dalvai, F. et al. Extensive analysis of the cytoplasmic proteome of human 19. Zhang, W. et al. Neuromelanin activates microglia and induces degeneration of erythrocytes using the peptide ligand library technology and advanced mass dopaminergic neurons: implications for progression of Parkinson’s disease. spectrometry. Mol. Cell. Proteom. 7, 2254–2269 (2008). Neurotox. Res. 19,63–72 (2011). 49. Lübke, T., Lobel, P. & Sleat, D. E. Proteomics of the lysosome. Biochim. Biophys. 20. Cebrián, C. et al. MHC-I expression renders catecholaminergic neurons suscep- Acta 1793, 625–635 (2009). tible to T-cell-mediated degeneration. Nat. Commun. 5, 3633 (2014). Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. 50. Schröder, B. A., Wrocklage, C., Hasilik, A. & Saftig, P. The proteome of lysosomes. 79. Pullarkat, R. K. & Reha, H. Accumulation of dolichols in brains of elderly. J. Biol. Proteomics 10, 4053–4076 (2010). Chem. 257, 5991–5993 (1982). 51. Koike, M. et al. Cathepsin D deficiency induces lysosomal storage with ceroid 80. Hall, N. A. & Patrick, A. D. Dolichol and phosphorylated dolichol content of lipofuscin in mouse CNS neurons. J. Neurosci. 20, 6898–6906 (2000). tissues in ceroid-lipofuscinosis. J. Inherit. Metab. Dis. 8, 178–183 (1985). 52. Qiao, L. et al. Lysosomal enzyme cathepsin D protects against alpha-synuclein 81. Ng Ying Kin, N. M., Palo, J., Haltia, M. & Wolfe, L. S. High levels of brain dolichols aggregation and toxicity. Mol. Brain 1, 17 (2008). in neuronal ceroid-lipofuscinosis and senescence. J. Neurochem. 40, 1465–1473 53. Morgan, C. P., Insall, R., Haynes, L. & Cockcroft, S. Identification of phospholipase (1983). B from Dictyostelium discoideum reveals a new lipase family present in mam- 82. Elleder, M., Sokolová, J. & Hrebícek, M. Follow-up study of subunit c of mito- mals, flies and nematodes, but not yeast. Biochem. J. 382, 441–449 (2004). chondrial ATP synthase (SCMAS) in Batten disease and in unrelated lysosomal 54. Tollbom, O., Chojnacki, T. & Dallner, G. Hydrolysis of dolichyl esters by rat liver disorders. Acta Neuropathol. 93, 379–390 (1997). lysosomes. J. Biol. Chem. 264, 9836–9841 (1989). 83. Wei, P., Smeyne, R. J., Bao, D., Parris, J. & Morgan, J. I. Mapping of Cbln1-like 55. Munck, A., Böhm, C., Seibel, N. M., Hashemol Hosseini, Z. & Hampe, W. Hu-K4 is a immunoreactivity in adult and developing mouse brain and its localization to ubiquitously expressed type 2 transmembrane protein associated with the the endolysosomal compartment of neurons. Eur. J. Neurosci. 26, 2962–2978 endoplasmic reticulum. FEBS J. 272, 1718–1726 (2005). (2007). 56. Pedersen, K. M., Finsen, B., Celis, J. E. & Jensen, N. A. Expression of a novel 84. Chung, C. Y. et al. Cell type-specific gene expression of midbrain dopaminergic murine phospholipase D homolog coincides with late neuronal development in neurons reveals molecules involved in their vulnerability and protection. Hum. the forebrain. J. Biol. Chem. 273, 31494–31504 (1998). Mol. Genet. 14, 1709–1725 (2005). 57. Nagaoka-Yasuda, R., Matsuo, N., Perkins, B., Limbaeck-Stokin, K. & Mayford, M. 85. Villani, G. R. et al. Cytokines, neurotrophins, and oxidative stress in brain disease An RNAi-based genetic screen for oxidative stress resistance reveals retinol from mucopolysaccharidosis IIIB. J. Neurosci. Res. 85, 612–622 (2007). saturase as a mediator of stress resistance. Free Radic. Biol. Med. 43, 781–788 86. Bellinger, F. P. et al. Glutathione peroxidase 4 is associated with neuromelanin in (2007). substantia nigra and dystrophic axons in putamen of Parkinson’s brain. Mol. 58. Satoh, J. et al. PLD3 is accumulated on neuritic plaques in Alzheimer’s disease Neurodegener. 6, 8 (2011). brains. Alzheimers Res. Ther. 6, 70 (2014). 87. Andres-Mateos, E. et al. DJ-1 gene deletion reveals that DJ-1 is an atypical 59. Kolter, T. & Sandhoff, K. Lysosomal degradation of membrane lipids. FEBS Lett. peroxiredoxin-like peroxidase. Proc. Natl Acad. Sci. USA 104, 14807–14812 584, 1700–1712 (2010). (2007). 60. Schulze, H., Kolter, T. & Sandhoff, K. Principles of lysosomal membrane degra- 88. Bonifati, V. et al. Mutations in the DJ-1 gene associated with autosomal recessive dation: cellular topology and biochemistry of lysosomal lipid degradation. Bio- early-onset parkinsonism. Science 299, 256–259 (2003). chim. Biophys. Acta 1793, 674–683 (2009). 89. Tribl, F. et al. Identification of L-ferritin in neuromelanin granules of the human 61. Reczek, D. et al. LIMP-2 is a receptor for lysosomal mannose-6-phosphate- substantia nigra: a targeted proteomics approach. Mol. Cell. Proteom. 8, independent targeting of beta-glucocerebrosidase. Cell 131, 770–783 (2007). 1832–1838 (2009). 62. Gao, D. et al. Structural basis for the recognition of oxidized phospholipids in 90. Sulzer, D. et al. T cells from patients with Parkinson’s disease recognize α- oxidized low density lipoproteins by class B scavenger receptors CD36 and SR- synuclein peptides. Nature 546, 656–661 (2017). BI. J. Biol. Chem. 285, 4447–4454 (2010). 91. Hara, T. et al. Suppression of basal autophagy in neural cells causes neurode- 63. Eckhardt, E. R. et al. High density lipoprotein endocytosis by scavenger receptor generative disease in mice. Nature 441, 885–889 (2006). SR-BII is clathrin-dependent and requires a carboxyl-terminal dileucine motif. J. 92. Mizushima, N., Ohsumi, Y. & Yoshimori, T. Autophagosome formation in mam- Biol. Chem. 281, 4348–4353 (2006). malian cells. Cell Struct. Funct. 27, 421–429 (2002). 64. Kuronita, T. et al. The NH(2)-terminal transmembrane and lumenal domains of 93. Mizushima, N., Levine, B., Cuervo, A. M. & Klionsky, D. J. Autophagy fights disease LGP85 are needed for the formation of enlarged endosomes/lysosomes. Traffic through cellular self-digestion. Nature 451, 1069–1075 (2008). 6, 895–906 (2005). 94. Pankiv, S. et al. p62/SQSTM1 binds directly to Atg8/LC3 to facilitate degradation 65. Cuervo, A. M. & Dice, J. F. A receptor for the selective uptake and degradation of of ubiquitinated protein aggregates by autophagy. J. Biol. Chem. 282, proteins by lysosomes. Science 273, 501–503 (1996). 24131–24145 (2007). 66. Cuervo, A. M. Autophagy: many paths to the same end. Mol. Cell. Biochem. 263, 95. Kuusisto, E., Salminen, A. & Alafuzoff, I. Ubiquitin-binding protein p62 is present 55–72 (2004). in neuronal and glial inclusions in human tauopathies and synucleinopathies. 67. Cuervo, A. M. & Dice, J. F. Age-related decline in chaperone-mediated autop- Neuroreport 12, 2085–2090 (2001). hagy. J. Biol. Chem. 275, 31505–31513 (2000). 96. Zatloukal, K. et al. p62 Is a common component of cytoplasmic inclusions in 68. Boellaard, J. W., Schlote, W. & Hofer, W. Species-specific ultrastructure of neu- protein aggregation diseases. Am. J. Pathol. 160, 255–263 (2002). ronal lipofuscin in hippocampus and neocortex of subhuman mammals and 97. Komatsu, M. et al. Homeostatic levels of p62 control cytoplasmic inclusion body humans. Ultrastruct. Pathol. 28, 341–351 (2004). formation in autophagy-deficient mice. Cell 131, 1149–1163 (2007). 69. Kabeya, Y. et al. LC3, a mammalian homologue of yeast Apg8p, is localized in 98. Szeto, J. et al. ALIS are stress-induced protein storage compartments for sub- autophagosome membranes after processing. EMBO J. 19, 5720–5728 (2000). strates of the proteasome and autophagy. Autophagy 2, 189–199 (2006). 70. Tanida, I., Minematsu-Ikeguchi, N., Ueno, T. & Kominami, E. Lysosomal turnover, 99. Picard, D. Heat-shock protein 90, a chaperone for folding and regulation. Cell. but not a cellular level, of endogenous LC3 is a marker for autophagy. Autop- Mol. Life Sci. 59, 1640–1648 (2002). hagy 1,84–91 (2005). 100. Chi, A. et al. Proteomic and bioinformatic characterization of the biogenesis and 71. Schröder, B., Elsässer, H. P., Schmidt, B. & Hasilik, A. Characterisation of function of melanosomes. J. Proteome Res. 5, 3135–3144 (2006). lipofuscin-like lysosomal inclusion bodies from human placenta. FEBS Lett. 581, 101. Horwitz, J. Alpha-crystallin can function as a molecular chaperone. Proc. Natl 102–108 (2007). Acad. Sci. USA 89, 10449–10453 (1992). 72. Muffat, J. & Walker, D. W. Apolipoprotein D: an overview of its role in aging and 102. Lowe, J. et al. alpha B crystallin expression in non-lenticular tissues and selective age-related diseases. Cell Cycle 9, 269–273 (2010). presence in ubiquitinated inclusion bodies in human disease. J. Pathol. 166, 73. Ganfornina, M. D. et al. Apolipoprotein D is involved in the mechanisms reg- 61–68 (1992). ulating protection from oxidative stress. Aging Cell 7, 506–515 (2008). 103. Liu, Y. et al. Upregulation of alphaB-crystallin expression in the substantia nigra 74. de Magalhães, J. P., Curado, J. & Church, G. M. Meta-analysis of age-related gene of patients with Parkinson’s disease. Neurobiol. Aging 36, 1686–1691 (2015). expression profiles identifies common signatures of aging. Bioinformatics 25, 104. Conn, K. J. et al. Identification of the protein disulfide isomerase family member 875–881 (2009). PDIp in experimental Parkinson’s disease and Lewy body pathology. Brain Res. 75. Ordoñez, C. et al. Apolipoprotein D expression in substantia nigra of Parkinson 1022, 164–172 (2004). disease. Histol. Histopathol. 21, 361–366 (2006). 105. Lehotzky, A. et al. Dynamic targeting of microtubules by TPPP/p25 affects cell 76. Cao, Y. et al. Autophagy is disrupted in a knock-in mouse model of juvenile survival. J. Cell Sci. 117, 6249–6259 (2004). neuronal ceroid lipofuscinosis. J. Biol. Chem. 281, 20483–20493 (2006). 106. Hlavanda, E. et al. Brain-specific p25 protein binds to tubulin and microtubules 77. Ryazantsev, S., Yu, W. H., Zhao, H. Z., Neufeld, E. F. & Ohmi, K. Lysosomal and induces aberrant microtubule assemblies at substoichiometric concentra- accumulation of SCMAS (subunit c of mitochondrial ATP synthase) in neurons of tions. Biochemistry 41, 8657–8664 (2002). the mouse model of mucopolysaccharidosis III B. Mol. Genet. Metab. 90, 393–401 107. Basrur, V. et al. Proteomic analysis of early melanosomes: identification of novel (2007). melanosomal proteins. J. Proteome Res. 2,69–79 (2003). 78. Andersson, M., Appelkvist, E. L., Kristensson, K. & Dallner, G. Distribution of 108. Hoashi, T. et al. Glycoprotein nonmetastatic melanoma protein b, a melanocytic dolichol and dolichyl phosphate in human brain. J. Neurochem. 49, 685–691 cell marker, is a melanosome-specific and proteolytically released protein. FASEB (1987). J. 24, 1616–1629 (2010). npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. 109. Zhang, P. et al. Silencing of GPNMB by siRNA inhibits the formation of mela- 126. Ferrari, E et al. Synthesis, Structure characterization, and evaluation in microglia nosomes in melanocytes in a MITF-independent fashion. PLoS ONE 7, e42955 cultures of neuromelanin analogues suitable for modeling Parkinson’s disease. (2012). ACS Chem. Neurosci. 8 501–512 (2017). 110. Li, B. et al. The melanoma-associated transmembrane glycoprotein Gpnmb 127. Ikemoto, K. et al. Does tyrosinase exist in neuromelanin-pigmented neurons in controls trafficking of cellular debris for degradation and is essential for tissue the human substantia nigra? Neurosci. Lett. 253, 198–200 (1998). repair. FASEB J. 24, 4767–4781 (2010). 128. Tribl, F., Arzberger, T., Riederer, P. & Gerlach, M. Tyrosinase is not detected in 111. International Parkinson’s Disease Genomics Consortium (IPDGC); human catecholaminergic neurons by immunohistochemistry and western blot Wellcome Trust Case Control Consortium 2 (WTCCC2). A two-stage meta-ana- analysis. J. Neural Transm. Suppl. 72,51–55 (2007). lysis identifies several new loci for Parkinson’s disease. PLoS Genet. 7, e1002142 129. Segura-Aguilar, J. et al. Protective and toxic roles of dopamine in Parkinson’s (2011). disease. J. Neurochem. 129, 898–915 (2014). 112. Kanaan, N. M. et al. The longitudinal transcriptomic response of the substantia 130. Makin, O. S. & Serpell, L. C. Structures for amyloid fibrils. FEBS J. 272, 5950–5961 nigra to intrastriatal 6-hydroxydopamine reveals significant upregulation of (2005). regeneration-associated genes. PLoS ONE 10, e0127768 (2015). 131. Follmer, C. et al. Oligomerization and membrane-binding properties of covalent 113. Cantagrel, V. & Lefeber, D. J. From glycosylation disorders to dolichol bio- adducts formed by the interaction of α-synuclein with the toxic dopamine synthesis defects: a new class of metabolic diseases. J. Inherit. Metab. Dis. 34, metabolite 3,4-dihydroxyphenylacetaldehyde (DOPAL). J. Biol. Chem. 290, 859–867 (2011). 27660–27679 (2015). 114. Rip, J. W., Blais, M. M. & Jiang, L. W. Low-density lipoprotein as a transporter of 132. Ferrari, E. et al. Synthesis and structural characterization of soluble neurome- dolichol intermediates in the mammalian circulation. Biochem. J. 297, 321–325 lanin analogs provides important clues to its biosynthesis. J. Biol. Inorg. Chem. (1994). 18,81–93 (2013). 115. Tollbom, O. & Dallner, G. Dolichol and dolichyl phosphate in human tissues. Br. J. 133. Conway, K. A., Rochet, J. C., Bieganski, R. M. & Lansbury, P. T. Jr. Kinetic stabili- Exp. Pathol. 67, 757–764 (1986). zation of the alpha-synuclein protofibril by a dopamine-alpha-synuclein adduct. 116. Chojnacki, T. & Dallner, G. The biological role of dolichol. Biochem. J. 251,1–9 Science 294, 1346–1349 (2001). (1988). 134. Der-Sarkissian, A., Jao, C. C., Chen, J. & Langen, R. Structural organization of 117. Van Houte, H. A., Van Veldhoven, P. P., Mannaerts, G. P., Baes, M. I. & Declercq, P. alpha-synuclein fibrils studied by site-directed spin labeling. J. Biol. Chem. 278, E. Metabolism of dolichol, dolichoic acid and nordolichoic acid in cultured cells. 37530–37535 (2003). Biochim. Biophys. Acta 1347,93–100 (1997). 135. Fernández, C. O. et al. NMR of alpha-synuclein-polyamine complexes elucidates the 118. Valtersson, C. et al. The influence of dolichol, dolichol esters, and dolichyl mechanism and kinetics of induced aggregation. EMBO J. 23, 2039–2046 (2004). phosphate on phospholipid polymorphism and fluidity in model membranes. J. 136. Comunian, C. et al. A comparative MudPIT analysis identifies different expres- Biol. Chem. 260, 2742–2751 (1985). sion profiles in heart compartments. Proteomics 11, 2320–2328 (2011). 119. van Duijn, G. et al. Dolichyl phosphate induces non-bilayer structures, vesicle 137. Sadygov, R. G. et al. Code developments to improve the efficiency of automated fusion and transbilayer movement of lipids: a model membrane study. Biochim. MS/MS spectra interpretation. J. Proteome Res. 1, 211–215 (2002). Biophys. Acta 861, 211–223 (1986). 138. Micallef, L. & Rodgers, P. eulerAPE: drawing area-proportional 3-Venn diagrams 120. Eckhardt, M. The role and metabolism of sulfatide in the nervous system. Mol. using ellipses. PLoS ONE 9, e101717 (2014). Neurobiol. 37,93–103 (2008). 121. Raposo, G. & Marks, M. S. Melanosomes--dark organelles enlighten endosomal membrane transport. Nat. Rev. Mol. Cell Biol. 8, 786–797 (2007). Open Access This article is licensed under a Creative Commons 122. Watt, B., van Niel, G., Raposo, G. & Marks, M. S. PMEL: a pigment cell-specific Attribution 4.0 International License, which permits use, sharing, model for functional amyloid formation. Pigment Cell Melanoma Res. 26, adaptation, distribution and reproduction in any medium or format, as long as you give 300–315 (2013). appropriate credit to the original author(s) and the source, provide a link to the Creative 123. Theos, A. C. et al. The PKD domain distinguishes the trafficking and amyloido- Commons license, and indicate if changes were made. The images or other third party genic properties of the pigment cell protein PMEL and its homologue GPNMB. material in this article are included in the article’s Creative Commons license, unless Pigment Cell Melanoma Res. 26, 470–486 (2013). indicated otherwise in a credit line to the material. If material is not included in the 124. Santini, E. et al. Critical involvement of cAMP/DARPP-32 and extracellular signal- article’s Creative Commons license and your intended use is not permitted by statutory regulated protein kinase signaling in L-DOPA-induced dyskinesia. J. Neurosci. 27, regulation or exceeds the permitted use, you will need to obtain permission directly 6995–7005 (2007). from the copyright holder. To view a copy of this license, visit http://creativecommons. 125. Santini, E. et al. Dopamine- and cAMP-regulated phosphoprotein of 32-kDa org/licenses/by/4.0/. (DARPP-32)-dependent activation of extracellular signal-regulated kinase (ERK) and mammalian target of rapamycin complex 1 (mTORC1) signaling in experi- © The Author(s) 2018 mental parkinsonism. J. Biol. Chem. 287, 27806–27812 (2012). Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png npj Parkinson's Disease Springer Journals

Neuromelanin organelles are specialized autolysosomes that accumulate undegraded proteins and lipids in aging human brain and are likely involved in Parkinson’s disease

Free
23 pages

Loading next page...
 
/lp/springer_journal/neuromelanin-organelles-are-specialized-autolysosomes-that-accumulate-zS42lwU5mI
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Biomedicine; Biomedicine, general; Neurosciences; Neurology
eISSN
2373-8057
D.O.I.
10.1038/s41531-018-0050-8
Publisher site
See Article on Publisher Site

Abstract

www.nature.com/npjparkd ARTICLE OPEN Neuromelanin organelles are specialized autolysosomes that accumulate undegraded proteins and lipids in aging human brain and are likely involved in Parkinson’s disease 1 1,2 1 1 1 1 1 Fabio A. Zucca , Renzo Vanna , Francesca A. Cupaioli , Chiara Bellei , Antonella De Palma , Dario Di Silvestre , Pierluigi Mauri , 3 3 4 5,6,7 1,5 Sara Grassi , Alessandro Prinetti , Luigi Casella , David Sulzer and Luigi Zecca During aging, neuronal organelles filled with neuromelanin (a dark-brown pigment) and lipid bodies accumulate in the brain, particularly in the substantia nigra, a region targeted in Parkinson’s disease. We have investigated protein and lipid systems involved in the formation of these organelles and in the synthesis of the neuromelanin of human substantia nigra. Membrane and matrix proteins characteristic of lysosomes were found in neuromelanin-containing organelles at a lower number than in typical lysosomes, indicating a reduced enzymatic activity and likely impaired capacity for lysosomal and autophagosomal fusion. The presence of proteins involved in lipid transport may explain the accumulation of lipid bodies in the organelle and the lipid component in neuromelanin structure. The major lipids observed in lipid bodies of the organelle are dolichols with lower amounts of other lipids. Proteins of aggregation and degradation pathways were present, suggesting a role for accumulation by this organelle when the ubiquitin-proteasome system is inadequate. The presence of proteins associated with aging and storage diseases may reflect impaired autophagic degradation or impaired function of lysosomal enzymes. The identification of typical autophagy proteins and double membranes demonstrates the organelle’s autophagic nature and indicates that it has engulfed neuromelanin precursors from the cytosol. Based on these data, it appears that the neuromelanin-containing organelle has a very slow turnover during the life of a neuron and represents an intracellular compartment of final destination for numerous molecules not degraded by other systems. npj Parkinson’s Disease (2018) 4:17 ; doi:10.1038/s41531-018-0050-8 INTRODUCTION to play a dual role, both toxic and protective, that is determined 13,14 by the cellular context and conditions. The synthesis of NM is Electron microscopy studies of neurons of numerous brain regions neuroprotective since it removes from the cytosol the reactive/ have demonstrated that organelles containing neuromelanin (NM) toxic quinones that would otherwise induce neurotoxicity. NM exhibit abundant clear “lipid bodies” (sometimes referred to as further plays a protective role by chelating potentially toxic “lipid droplets” in the literature, although this term is widely used 3,15 metals, including Fe, Zn, Cu, Al, Cr, Mo, Pb, and Hg (refs. ), drugs for a different lipid storage organelle) and a dark electron-dense 16–18 1–3 and organic toxicants. However, NM can play a toxic role matrix. The number of these organelles and the concentration 3–6 when released by degenerating neurons of the SN during PD: of NM pigment increase linearly during aging. These organelles under these conditions, NM acutely discharges high amounts of are highly concentrated in dopamine (DA) neurons of the metals and organic chemicals accumulated over many years of substantia nigra (SN) (Fig. 1a,b,c) and norepinephrine neurons of 2,5,6 locus coeruleus, brain regions strongly targeted in Parkinson’s life. NM released by degenerating neurons in PD activates 7,8 microglia, producing reactive and pro-inflammatory molecules disease (PD). The pigments of these organelles are a family of compounds formed by a melanic, aliphatic, and protein compo- that induce further neuronal death and release of NM, thus establishing a vicious cycle of neuroinflammation and neurode- nents with variable ratios. NM pigment also accumulates large amounts of metals, further confirming that these organelles generation. The activation of microglia by NM can drive antigen presentation by SN and locus coeruleus catecholaminergic continuously accumulate in aging due to very slow turnover. The formation of NM appears to provide a protective neurons, a response that may play a crucial role in PD 3,10 process, but the amount of NM accumulated in neurons is pathogenesis. NM can also stimulate dendritic cells in vitro 8,11,12 related to their vulnerability in PD. Due to its biochemical inducing their maturation. properties, NM has long been suggested as a critical factor The structure of the melanic component is different in various 8 9,22 underlying neuronal vulnerability in PD. Indeed, NM is suggested types of NM pigments, and multiple features of NM structure, 1 2 3 Institute of Biomedical Technologies, National Research Council of Italy, Segrate, Milan, Italy; IRCCS Don Carlo Gnocchi ONLUS Foundation, Milan, Italy; Department of Medical 4 5 Biotechnology and Translational Medicine, University of Milan, Segrate, Milan, Italy; Department of Chemistry, University of Pavia, Pavia, Italy; Department of Psychiatry, Columbia University Medical Center, New York State Psychiatric Institute, New York, NY, USA; Department of Neurology, Columbia University Medical Center, New York, NY, USA and Department of Pharmacology, Columbia University Medical Center, New York, NY, USA Correspondence: Luigi Zecca (luigi.zecca@itb.cnr.it) These authors contributed equally: Fabio A. Zucca, Renzo Vanna. Received: 14 December 2017 Revised: 10 April 2018 Accepted: 17 April 2018 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. as well as protein and lipid composition of NM-containing isolation of ORG samples, the membranes can be broken with organelles, remain incompletely characterized. Indeed, the protein consequent mixing of intra- and extracellular proteins. The components of NM-containing organelle have been partially washing procedure of ORG isolation can moreover change the characterized, and these are consistent with its lysosomal protein content of organelles due to membrane damage. The 23,24 nature. However, the characterization of proteins belonging native proteins detected by analyzing ORG samples therefore to different portions of the NM-containing organelle and their likely underestimate the original protein content, while there may localization are yet to be clarified, although this knowledge is be an increased number of proteins due to contamination. fundamental to understanding the complex nature of these Thus, we also isolated TIS-NM for protein characterization by LC- organelles. Current data do not indicate the mechanisms of NM MS. During the isolation process of TIS-NM, membranes are accumulation, protein and lipid transport and accumulation within broken and additional proteins are likely aspecifically adsorbed by the organelle, or the role of these organelles inside neurons. It is NM pigment. TIS-NM was prepared following the procedure 3,9,28 further unknown whether the NM pigment is synthesized within reported by several previous studies. Here, proteinase K was these organelles or transported inside after synthesis elsewhere. employed during the isolation procedure of TIS-NM from SN tissue Here we investigate the proteins and lipids of NM-containing as it was necessary to remove non-specifically associated proteins. organelles from human SN and their accumulation inside these A preliminary study reported that TIS-NM isolated without organelles. The NM-containing organelle analyzed here is a proteinase K contained a higher percentage of extracellular and paradigmatic case in which an abundant accumulation of the nuclear proteins than TIS-NM isolated with proteinase K, indicating melanic pigment occurs; however, the presence of these a higher contamination by non-specific proteins during the organelles is observed throughout the entire brain as a result of isolation procedure. Indeed, the isolation of NM pigment physiological aging. without proteinase K generates macroaggregates difficult to The aim of this study is to perform an extended characterization purify. These macroaggregates could contain proteins originating of proteins and lipids of NM-containing organelles from human from cytosol, other organelles and tissue compartments which SN. To this end, we required highly purified preparations of NM- interact and bind to NM pigment and NM-conjugated proteins containing organelles. In order to control contaminations and to during isolation, that may form S–S bridges and other means of avoid the loss of some proteins or lipids, we prepared three types conjugation. In this case, this kind of sample would contain of NM-containing samples with different procedures and com- proteins not related to the NM-containing organelle. This pared the data obtained by liquid chromatography-mass spectro- procedure and its rationale are described in detail in previous 3,9 metry (LC-MS) determinations on these samples. To further studies (Methods). We elected to prepare the TIS-NM samples confirm the reliability of our results, independent determinations with proteinase K as this type of sample was used in several were also made by immunoelectron microscopy (IEM), western previous studies, and so necessary for comparison of the present 3,9,28 blotting (WB), and thin-layer chromatography (TLC) in addition to data with previous reports. LC-MS. In this study, protein profiles were analyzed (i) in three Finally, to overcome the above experimental limitations, we preparations derived from human SN, the organelles containing further analyzed proteins in ORG-NM samples, i.e., the NM NM pigment (ORG), NM pigment purified from SN tissues (TIS-NM), pigment isolated from ORG samples. This type of NM was purified and NM pigment isolated from organelles (ORG-NM) using LC-MS; by disrupting and eliminating the membranes and the soluble (ii) by WB of ORG samples; and (iii) by IEM of SN tissue slices. The portion from ORG samples, in order to study the protein lipid pathways were analyzed by TLC and LC-MS analyses of lipid components strictly associated with NM pigment, without molecules associated with lipid bodies and with NM pigment, and exposure to contaminants from other intraneuronal components. through characterization of transport proteins and related This experimental design was in our opinion the best approach enzymes by LC-MS, IEM, and WB analyses. to exclude contaminating proteins. In addition, this approach The combination of proteomics, lipidomics, imaging, and avoided loss of some proteins of the NM-containing organelle, a biochemical techniques in the study of NM-containing organelles central aspect of this study, beyond preventing contamination. is required for a detailed description of the molecular mechanisms We then compared the three sets of proteins observed in the associated with brain aging and neurodegeneration. Previous following preparations (Supplementary Fig. 1): two independent 23–27 studies have only partially addressed these issues. The samples of ORG, two independent samples of TIS-NM, and two identification of these pathways is crucial for elucidating the independent samples of ORG-NM (Methods). ORG, TIS-NM, and processes mediating neuronal survival and vulnerability during ORG-NM samples were analyzed by a total of 15 LC-MS analyses. aging and PD. As a result, 1020 proteins were identified and a group of 293 was selected as representative of the samples based on their amount (Table 1), as estimated from the number of spectral count (SpC), RESULTS defined as the sum of all peptides of a single protein observed in NM organelles proteins were identified by analyzing three types of the mass spectrum. The threshold for inclusion in the list of samples: isolated NM-containing organelles, NM purified from SN representative proteins was two or more SpC (i.e., peptides) for tissues, and NM purified from NM-containing organelles each protein as average value in at least one of the three types of The purpose of this study is to describe the proteins present inside samples. the NM-containing organelles, and then distinguish those The cellular distribution of the 293 representative proteins is covalently bound to NM pigment from those not attached to represented in Fig. 2 (for details refer to Supplementary Table 1; NM. The proteins bound to NM pigment may be more related to Supplementary Data 1), showing that 34 proteins, detected by the initial steps of NM synthesis, while those not bound to NM ∼60 % of the number of SpC in all samples (7916 of 13,157 overall pigment may play a role in the membrane, transport, and storage number of SpC for representative proteins), were lysosomal processes involved in NM-containing organelle formation. There proteins. We can speculate that if all the protein classes were are several experimental limitations in the analyses of proteins in equally represented in our samples, the 34 lysosomal proteins NM-containing organelles. In human post mortem brain, the among the 293 representative proteins (Supplementary Table 1) membranes of organelles undergo degradation, and their would be represented by ∼1527 SpC and not by 7916 SpC, as we constituents can be released from organelles to the cytosol, and observed. With this assumption, we estimate that the lysosomal conversely, the organelles can be contaminated by cytosolic class is >5-fold enriched in our samples. This is a striking evidence components. Furthermore, when processing brain tissues for of lysosomal protein overrepresentation. npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation 1234567890():,; Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Table 1. Summary of the proteomic analysis ORG (n = 2) TIS-NM (n = 2) ORG-NM (n = 2) Overall Number of proteins 563 407 220 1020 Number of SpC for all 1020 proteins 4229 7743 2023 13,995 Representative proteins (with SpC ≥ 2 as average value) 174 [164] 130 [116] 125 [101] 293 Number of SpC for the group of 293 representative proteins 3792 [3778] 7439 [7420] 1926 [1898] 13,157 [13,096] Summary of proteomic analyses of different types of samples (ORG, TIS-NM and ORG-NM), calculated from Supplementary Data 1. Each type of sample was prepared in duplicate and then analyzed by multiple LC-MS analyses. For details of subjects and preparation of samples for LC-MS analysis of proteins, see Methods. The first two rows show the number of proteins and their number of SpC for each kind of sample. Among the group of overall proteins detected in these samples (1020 by 13,995 SpC), representative proteins were selected on the basis of their SpC, defined as the sum of all peptides of a single protein observed in the mass spectrum. We have classified a protein as representative in one type of sample, if detected by SpC ≥ 2 as average value. In the third row, we report for each type of sample the following values: (i) in brackets, the number of proteins detected as representative (with SpC ≥ 2) uniquely in that type of sample; (ii) without brackets, the number of representative proteins in that type of sample plus those identified as non-representative (with SpC < 2) in that specific sample but listed as representative (with SpC ≥ 2) in at least one of the other type of samples (e.g., a protein that was detected as non-representative in TIS-NM but as representative in ORG or ORG-NM samples would be included in the count for TIS-NM). The fourth row of the table shows the number of SpC detected for the group of 293 representative proteins described in the third row, and in particular: (i) in brackets, the total number of SpC detected for representative proteins (with SpC ≥ 2) uniquely found in that type of sample (e.g., 3778 is the number of SpC for 164 proteins uniquely detected in ORG sample); (ii) without brackets, the total number of SpC of representative proteins plus the SpC of proteins identified as non-representative (with SpC < 2) in that specific sample but listed as representative (with SpC ≥ 2) in at least one of the other type of samples (e.g., 7439 is the number of SpC for 130 proteins detected in TIS-NM sample, including SpC of proteins uniquely found in TIS-NM plus those of proteins detected as non-representative in TIS-NM but as representative in ORG or ORG-NM samples). The “Overall” column represents the overall number of proteins and peptides detected in all analyzed samples (i.e., 293 is the number of all representative proteins detected in ORG, TIS-NM, and ORG-NM, considering only one time a protein present in two or more samples) An Euler diagram in Fig. 3 shows the distribution of the 293 membrane proteins including lysosome membrane protein 2 representative proteins among different samples, as described in (SCARB2), CD63 antigen, type 1 phosphatidylinositol 4,5-bispho- Table 1. The ORG sample contains the highest number of proteins sphate 4-phosphatase (only in the ORG sample) and some functional and shares 80 proteins with the other two samples. TIS-NM and subunits of the lysosomal V-type proton ATPase (in ORG sample one ORG-NM, representing two different preparations of NM pigment, subunit as representative while other two subunits in very low contain fewer proteins. TIS-NM shares 68 proteins with ORG and amounts and categorized as non-representative proteins), but not lysosome-associated membrane glycoprotein 1 (LAMP1) or ORG-NM samples, while ORG-NM shares 89 proteins with ORG and lysosome-associated membrane glycoprotein 2 (LAMP2). TIS-NM samples. A portion of representative proteins was uniquely Due to the difficulties in preserving ORG membranes during identified in each type of sample (94 proteins in ORG, 62 proteins their isolation and related problems in LC-MS detection of in TIS-NM, and 36 proteins in ORG-NM), but these were present in transmembrane proteins, we also performed WB and IEM very small quantities (<10 % of the overall number of SpC experiments to investigate lysosomal features of these organelles. detected for representative proteins); interestingly, 35 proteins We confirmed the presence of SCARB2 and V-type proton ATPase were commonly detected in all three samples, and these shared subunit B, brain isoform (ATP6V1B2), a subunit of the lysosomal V- proteins were present in high quantities (∼80 % of the overall type proton ATPase, in NM-containing organelles by WB on ORG number of SpC detected for representative proteins; see data and samples and by IEM in SN sections (Figs. 4 and 5; Supplementary discussion in the legends of Fig. 3 and Table 1, and details in Fig. 2, 3). Likewise, both LAMP1 and LAMP2, while undetected by following sections). LC-MS in ORG samples, were observed by WB and IEM (Figs. 4 and 5; Supplementary Fig. 2, 3). Proteins found in NM-containing organelles Some non-lysosomal proteins were also detected, most of The ORG samples were obtained directly from fresh SN tissue, which are classified as cytoskeletal and cytoplasmatic proteins. In without freezing and after soft homogenization/centrifugation addition to typical tubulin chains (mainly beta chains), we found procedures in order to preserve the original structure and tubulin polymerization-promoting protein, which is involved in composition of these organelles. Transmission electron micro- protein aggregation, inclusion bodies formation and neurodegen- scopy confirmed that the contents of the purified organelles were eration; we also detected heat shock protein HSP 90-alpha and mainly intact, with preserved lipid membranes and lipid bodies alpha-crystallin B chain, which has been reported as a component that were morphologically identical to those observed in slices of of Lewy bodies and has been characterized as a chaperone. SN tissues. Importantly, low magnification images demonstrated Additional cytoskeletal proteins included for example the the absence of other cellular contaminants (Fig. 1d). microtubule-associated protein tau and microtubule-associated A group of 164 proteins was identified with SpC ≥ 2 exclusively in protein 6, which play roles in microtubule stability and are 32,33 ORG sample (Table 1; Supplementary Data 1), while an additional implicated in neurodegenerative mechanisms. 34,35 ten proteins were identified with SpC < 2 in ORG but detected with As a potentially important clue in PD pathogenesis, we also SpC ≥ 2 in TIS-NM or ORG-NM samples. Then, we are considering detected alpha-synuclein (SNCA) exclusively in ORG samples. We that the number of representative proteins in ORG sample is 174 confirmed this result by WB experiments on ORG samples and IEM (Table 1;Fig. 3; Supplementary Table 1).These data revealed that experiments on SN tissue slices, observing SNCA signals mainly in ORG is characterized by a large group of lysosomal proteins (25 the NM pigment and rarely in lipid bodies (Figs. 4 and 5). proteins, ~43 % rel. # SpC), with far fewer proteins from other We also found major histocompatibility complex, class I (HLA) in organelles or cytosol (Fig. 2; Supplementary Table 1). In addition to ORG samples, consistent with our recent report of the first the set of soluble lysosomal proteins observed in all three types of identification of this protein in adult neurons. The identification samples (proteases, esterases, sulfatases, glycosidases, hydrolases, of HLA was obtained by matching experimental spectra to peptide and other lysosomal proteins), here we identified typical lysosomal sequences in specific databases for HLA (Supplementary Data 2). Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Fig. 1 Transmission electron microscopy images of NM-containing organelles in human SN tissue (a–c) and after the isolation procedure (d). a–c SN tissue of 89 y.o. healthy subject. Intraneuronal NM-containing organelles of the SN are membrane bounded (black arrowhead in a and b) and contain large amounts of NM pigment (black and electron dense), a protein matrix and lipid bodies (asterisk). Scale bar =1µm in a. Large lipid bodies (asterisk) are surrounded by a membrane as demonstrated in b (arrow), although the images do not distinguish between a bilayer and single layer membrane. Considering that brain samples used in this study were post mortem tissues, it is striking that there is often a double membrane around many of the organelles. At higher magnification, a double membrane delimiting NM-containing organelle is clearly visible (empty arrowhead in c). d NM-containing organelles isolated from the SN tissue of 89 y.o. healthy subject (the same subject of a–c). The purity and integrity of isolated NM-containing organelles is clearly demonstrated by transmission electron microscopy: low magnification d demonstrates that cellular and subcellular debris are completely absent. The outer limiting membrane is not apparent, but the constituents of the organelles appear intact with NM pigment, many lipid bodies (asterisks) and membranes of lipid bodies. Scale bar = 1 µm in d Fig. 2 Histogram of cellular distribution of the 293 representative proteins found in all analyzed samples shown as relative number of SpC vs. cellular compartments. For details of subjects and preparation of samples for LC-MS analysis of proteins, see Methods. Some proteins may have multiple cellular locations: for each protein the most typical and representative cellular location was assigned. The different types of samples are represented by different colors (ORG, TIS-NM, and ORG-NM) and gray bars refer to overall representative proteins considered as a single data set (indicated with “All Samples”). The “Rel. # SpC (%)” is the total number of SpC for a specific class of proteins (i.e., lysosomal) referred to the overall number of SpC of representative proteins in each sample: this value represents the relative abundance of a particular class of proteins in one sample (see also Supplementary Table 1). The term “Vesicles” refers to vesicle trafficking, including proteins involved in vesicular transport, fusion, etc. The category “Unknown cell location” consists of proteins for which a cellular location was still unclear, while the class “Uncharacterized proteins” comprises proteins for which, at the moment of data analyses, a complete characterization and/or role was missing npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. The presence of the antigen presenting protein HLA was confirmed with IEM on SN tissue, indicating a high accumulation of HLA on NM granules of the SN (Supplementary Fig. 4). Interestingly, some additional proteins identified only in ORG sample were dynein heavy chain 12, axonemal and some ras- related proteins (RAB8B, RAB14, and RAB33B as representative proteins, while RAB2A, RAB5C, and RAB8A were present in low amounts and categorized as non-representative), each of which are involved in intracellular vesicle trafficking. Proteins found in NM pigment isolated from SN tissue The TIS-NM samples were obtained from pooled SN tissues using the purification procedure adopted for previous chemical and 3,9,28 structural investigations on NM pigment. Despite the chemical treatments (i.e., high-salt solutions, sodium dodecyl sulfate, methanol, and hexane), repeated washings and proteinase K digestions, proteomic analysis resulted in the identification of 116 proteins with SpC ≥ 2 uniquely in TIS-NM (Table 1; Supple- mentary Data 1). However, an additional 14 proteins with SpC < 2 in TIS-NM were found with SpC ≥ 2 in ORG or ORG-NM samples: Fig. 3 Area-proportional Euler diagram of the 293 representative therefore, we estimate 130 to be the number of representative proteins (detected by SpC ≥ 2 as average value in at least one of the proteins in TIS-NM sample (Table 1; Fig. 3; Supplementary Table 1). three types of samples) identified in ORG, TIS-NM, or ORG-NM. For details of subjects and preparation of samples for LC-MS analysis of The most highly represented and abundant class of proteins was proteins, see Methods. The diagram was calculated using the again lysosomal, even more so than in ORG samples. We detected EulerAPE tool (Methods) and by using NCBI accession (GI 26 lysosomal proteins representing ~73 % of rel. # SpC in this number). Outside the diagram we report for each type of sample sample (Fig. 2; Supplementary Table 1). Among these proteins, the following values: (i) in brackets, the number of proteins detected high quantities of typical lysosomal enzymes were found, and as representative (with SpC ≥ 2) uniquely in that type of sample, as these were also identified in ORG and ORG-NM samples. However, reported in Table 1; (ii) without brackets, the number of some lysosomal proteins were detected exclusively in TIS-NM representative proteins plus those identified as non-representative samples (e.g., epididymal secretory protein E1, fatty acid synthase, (with SpC < 2) in that specific sample but listed as representative cathepsin L1, lysosomal alpha-mannosidase, and ribonuclease T2). (with SpC ≥ 2) in at least one of the other type of samples (e.g., a protein that was detected as non-representative in ORG sample but Only one lysosomal membrane protein, the transmembrane as representative in TIS-NM would be included in the count for protein 106B, was identified exclusively in the TIS-NM sample. ORG). Numbers in non-overlapping areas of circles report the Non-lysosomal proteins found in TIS-NM included ferritins, representative proteins found uniquely in that type of sample. The mostly ferritin light chain (FTL) in contrast to ferritin heavy chain overlapping areas correspond to proteins shared by two or three (FTH1), heat shock protein HSP 90-alpha, and glyceraldehyde-3- different types of samples: e.g., a protein detected in all samples but phosphate dehydrogenase. HLA was also detected in this sample as representative only in ORG would be included in the overlapping (Supplementary Data 2). By means of LC-MS analysis, we found area of 35 proteins shared between ORG, TIS-NM, and ORG-NM. exclusively in this sample the mature chain of ATP synthase F(0) Percentages in parentheses represent the ratio between the total complex subunit C3, mitochondrial (ATP5G3), the major storage number of SpC of proteins belonging to one area of the diagram and the overall number of SpC of representative proteins detected material accumulated in ceroid lipofuscinosis, as well as in all samples. The highest percentage value is located in the area cerebellin-2, the PD-associated protein DJ-1 and protein shared between all three types of samples. The detailed list of disulfide-isomerase A3. Additionally superoxide dismutase proteins is reported in Supplementary Data 1 [Cu–Zn] was detected in TIS-NM and also as a fragment in ORG sample. The ATP synthase F(0) complex subunit C1, mitochondrial (ATP5G1) was also detected by WB in ORG samples; its presence were identified in ORG-NM with SpC < 2 but detected with SpC ≥ 2 was also confirmed by IEM, indicating that is localized in the NM- in ORG or TIS-NM samples. Therefore, we estimate the number of containing organelles, where it is mainly bound to the NM representative proteins in ORG-NM sample to be 125 (Table 1; Fig. pigment. The antibody used in these experiments is expected to 3; Supplementary Table 1); the majority (89 proteins) were also recognize the three mature chains of proteins, namely ATP5G1, detected in ORG and TIS-NM samples (Fig. 3). Proteins identified ATP synthase F(0) complex subunit C2, mitochondrial (ATP5G2) and ATP5G3, which are identical and encoded by three different only in ORG-NM sample (36 proteins) were found in very low genes (Supplementary Fig. 2, 3). amounts (<1 % rel. # SpC) (Fig. 3), thus indicating that almost no contaminants affected the analysis of this type of sample. As observed in ORG and TIS-NM, lysosomal proteins were also Proteins found in NM pigment isolated from organelles prevalent in this sample (23 proteins; ~44 % rel. # SpC) (Fig. 2; In order to study the protein matrix associated with the NM Supplementary Table 1). Moreover, lysosomal membrane proteins pigment inside organelles and considering that aspecific proteins (i.e., SCARB2 and CD63 antigen) and other cytoskeletal and might interact with NM during the isolation of TIS-NM from SN cytoplasmatic proteins (i.e., alpha-crystallin B chain, tubulin tissues, we isolated NM pigment directly from ORG samples. The polymerization-promoting protein, microtubule-associated pro- amount of ORG-NM sample was lower than the other two types of tein tau, and low amounts of microtubule-associated protein 6) samples (ORG-NM < ORG < TIS-NM), as it was prepared from the found in ORG samples were observed in ORG-NM as well. Again, ORG sample; TIS-NM was isolated processing several SN pooled LAMP2 was not detected in this sample by LC-MS, while LAMP1 tissues, while ORG and ORG-NM samples were prepared from one was identified at very low levels and categorized as a non- or occasionally two SN tissues (Methods). This is clearly shown by the number of SpC detected in each sample (Table 1). A group of representative protein. These data suggest that thermal shock 101 proteins was detected with SpC ≥ 2 exclusively in ORG-NM treatments to purify NM pigment from organelles do not (Table 1; Supplementary Data 1), while an additional 24 proteins completely disrupt the organelle. It is possible that some Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. membrane portions are strongly connected to the contents of the experiments, which confirmed their localization inside the NM- organelle and prevent the leakage of some proteins. Indeed, containing organelles (Figs. 4 and 5; Supplementary Fig. 4, 5, 6). membranous, cytoskeletal and cytoplasmic proteins were still found in the analysis of ORG-NM sample. As observed in ORG and Relevant proteins not revealed by mass spectrometry but found in TIS-NM samples, HLA was also detected in this sample by other techniques (Supplementary Data 2). Additional IEM and WB studies were performed for some relevant Similarly, ORG-NM samples share several proteins with TIS-NM proteins not detected by LC-MS, probably due to instrumental (Fig. 3): e.g., the lysosomal prosaposin, heat shock protein HSP 90- limitations and low abundance. Due to the lysosomal features of alpha (here detected in low amounts, but present as representa- the NM-containing organelle and its proposed autophagic origin, tive also in ORG and TIS-NM), glyceraldehyde-3-phosphate we performed WB and IEM experiments to verify the presence of dehydrogenase, some peptides of ferritins (again mostly FTL if the autophagic marker microtubule-associated proteins 1A/1B compared to FTH1), and the PD-associated protein ubiquitin light chain 3B (MAP1LC3B), which was not detected by LC-MS like carboxyl-terminal hydrolase isozyme L1 (here as representative, 23,24 in previous studies. Here, IEM experiments on SN tissues while in TIS-NM it was in low amounts and categorized as a non- revealed MAP1LC3B signals mainly engulfed inside NM-containing representative protein). ORG-NM and TIS-NM are analogous organelles, on NM pigment, around lipid bodies and sometimes samples since they are both isolated NM pigment. However, the lining their membranes (Fig. 4). In order to confirm the presence of amount of ORG-NM sample is normally lower than that of TIS-NM, this crucial protein, IEM experiments were also repeated using a while the ORG sample is the intact organelle containing the NM different antibody, confirming the specific accumulation of pigment. Due to the identification by LC-MS of FTL and FTH1 in MAP1LC3B inside NM-containing organelles (Supplementary Fig. both ORG-NM and TIS-NM samples but not in ORG samples, IEM 4). In addition to IEM findings, WB analyses of ORG samples and WB experiments were performed. IEM confirmed the presence demonstrated the presence of MAP1LC3B, thus confirming the of FTL and to a lesser extent FTH1 in NM-containing organelles of autophagic nature of NM-containing organelle. Specifically, WB the SN tissue. The accumulation of this iron-storage protein was analyses demonstrated mainly the MAP1LC3B-I form, while the confirmed by WB analyses that detected low levels of FTL but MAP1LC3B-II form was undetectable in ORG samples (Fig. 5). failed to detect FTH1 in ORG samples (Supplementary Fig. 2, 3). Another important autophagy-related protein, the autophagic This may be due to chains of iron hydroxides that formed bridges adaptor sequestosome-1 (SQSTM1), was not detected by LC-MS 23,24 connecting NM and partially degraded ferritins, so that this here and in previous studies, but was found by IEM and WB protein was bound to NM and was not present as a separate analyses (Supplementary Fig. 5, 6). molecule. Moreover, due to the presence of some Ras-related proteins, we Examples of proteins identified as representative only in ORG- confirmed the presence of Ras-related protein Rab-5A (RAB5A), NM include phospholipid hydroperoxide glutathione peroxidase, which is of interest due to its involvement in retrograde axonal mitochondrial, and lysosomal acid phosphatase. endosomal transport, by IEM and WB (Supplementary Fig. 5, 6) but not by LC-MS. Proteins detected in all three types of samples A common group of proteins was identified in all types of Lipid bodies contain dolichols involved in NM synthesis and samples, corresponding to ~80 % of overall SpC detected for 293 typical membrane lipids representative proteins (Fig. 3; Table 2). Within NM-containing organelles, electron micrographs demon- Within the set of lysosomal proteins detected in all three types strated the presence of membrane-bound lipid bodies, generally of samples (17 entries in Table 2), were peptidases (tripeptidyl- ranging from 200 to 500 nm, sometimes reaching sizes as large as peptidase 1, gamma-glutamyl hydrolase and lysosomal Pro-X 1 µm, with some smaller lipid bodies (50–100 nm) entrapped in carboxypeptidase), proteases [cathepsin B, cathepsin Z and the NM regions of the organelle (Fig. 1a–c). Conventional electron cathepsin D (CTSD)], esterases (sialate O-acetylesterase and microscopy on post mortem SN tissues does not clearly reveal if palmitoyl-protein thioesterase 1), sulfatases (iduronate 2-sulfatase, these lipid bodies are surrounded by membranes formed of N-sulphoglucosamine sulphohydrolase, arylsulfatase B and N- normal bilayer or single layer. acetylgalactosamine-6-sulfatase), glycosidases (lysosomal alpha- LC-MS analyses of solvent extracts prepared from both TIS-NM glucosidase), lipid hydrolases (acid ceramidase), and other and ORG samples demonstrate that dolichols and dolichoic acids lysosomal proteins [mammalian ependymin-related protein 1 were the major lipid components (Fig. 6). LC-MS analyses of both and putative phospholipase B-like 2 (PLBD2)]. CTSD, a classical solvent extracts further revealed the presence of signals lysosomal marker, was detected in ORG samples by WB as heavy corresponding to different glycerophospholipids and sphingoli- chain mature form and was confirmed by IEM as present pids (Supplementary Table 2). Among sphingolipids, signals abundantly in the NM-containing organelles (Figs. 4 and 5). We attributable to sphingomyelin, neutral glycolipids (lactosylcera- confirmed by IEM and WB (Supplementary Fig. 2, 3) the presence mide), sulfatides, and gangliosides (mono-, di-, and tri-sialogan- 37,38 of the recently described PLBD2. In addition, we detected by gliosides) together with other lipid molecules such as free fatty LC-MS phospholipase D3, apolipoprotein D (APOD), cerebellin-1, acids were identified (Supplementary Table 2). Thus, the lipid transmembrane glycoprotein NMB (GPNMB), F-box only protein bodies of ORG and the lipid mixtures adsorbed to TIS-NM contain 11, heat shock protein HSP 90-alpha, and two ubiquitin-related a wide variety of membrane amphipathic lipids, encompassing proteins, namely polyubiquitin-C (UBC) and ubiquitin-60S riboso- both lipids typically enriched in neurons, such as gangliosides, and mal protein L40 (UBA52). HLA peptides associated with NM were lipids involved in oligodendrocyte function and myelin formation, 25,41 identified by LC-MS analyses in all samples isolated from human such as sphingomyelin and sulfatide. The profile of amphi- SN (Supplementary Data 2). In each type of sample we also found pathic lipids was not identical in TIS-NM and in ORG, in particular protein phosphatase 1 regulatory subunit 1B, which potently as signals attributable to sulfatides were far higher in TIS-NM inhibits protein phosphatase-1. Moreover, various tubulins (Supplementary Table 2), suggesting that some myelin lipids could (mainly beta chains) and other abundant proteins including glial be adsorbed to TIS-NM during the purification procedure. Indeed, fibrillary acidic protein, myelin basic protein, and creatine kinase B- the presence of amphipathic lipids in the lipid bodies from ORG type were detected in all three samples (for other proteins see and in lipid mixtures adsorbed to TIS-NM was confirmed by TLC Table 2; Supplementary Data 1). Due to the important roles of analysis (Fig. 7). In the solvent extracts of the TIS-NM and ORG APOD, GPNMB, ubiquitins, and HLA, we conducted IEM and WB samples, the amounts of dolichols and dolichoic acids were higher npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Table 2. Representative proteins commonly detected by LC-MS in all analyzed samples NCBI accession (GI UniProt accession Protein name Gene name Cellular location All samples number) number Total # Average # SpC SpC 193785841 P61981 14-3-3 protein gamma YWHAG Cytoplasm 13 2 21536286 P12277 Creatine kinase B-type CKB Cytoplasm 41 6 4503979 P14136 Glial fibrillary acidic protein GFAP Cytoplasm 124 14 21735492 Q9UD71 Protein phosphatase 1 regulatory PPP1R1B Cytoplasm 28 3 subunit 1B 32526901 Q8WYA0 Intraflagellar transport protein 81 IFT81 Cytoskeleton 17 2 homolog 105990539 P07196 Neurofilament light polypeptide NEFL Cytoskeleton 10 2 2119276 Q6LC01 Tubulin beta chain (fragment) N/A Cytoskeleton 126 11 4507729 Q13885 Tubulin beta-2A chain TUBB2A Cytoskeleton 22 3 21361322 P04350 Tubulin beta-4A chain TUBB4A Cytoskeleton 60 8 60729665 Q6Y288 Beta-1,3-glucosyltransferase B3GLCT Endoplasmic 17 2 reticulum 1575347 Q8IV08 Phospholipase D3 PLD3 Endoplasmic 266 19 reticulum 4502163 P05090 Apolipoprotein D APOD Extracellular 74 10 8247915 Q13510 Acid ceramidase ASAH1 Lysosome 225 17 825628 P15848 Arylsulfatase B ARSB Lysosome 53 4 4503139 P07858 Cathepsin B CTSB Lysosome 1058 63 3929733 Q6LAF9 Cathepsin B (fragment) N/A Lysosome 88 10 4503143 P07339 Cathepsin D CTSD Lysosome 343 24 22538442 Q9UBR2 Cathepsin Z CTSZ Lysosome 169 11 4503987 Q92820 Gamma-glutamyl hydrolase GGH Lysosome 1003 62 4557659 P22304 Iduronate 2-sulfatase IDS Lysosome 95 8 126590 P10253 Lysosomal alpha-glucosidase GAA Lysosome 37 3 4826940 P42785 Lysosomal Pro-X carboxypeptidase PRCP Lysosome 91 7 24475586 Q9UM22 Mammalian ependymin-related EPDR1 Lysosome 1489 88 protein 1 4503899 P34059 N-acetylgalactosamine-6-sulfatase GALNS Lysosome 26 2 4506919 P51688 N-sulphoglucosamine SGSH Lysosome 47 4 sulphohydrolase 4506031 P50897 Palmitoyl-protein thioesterase 1 PPT1 Lysosome 287 19 27734917 Q8NHP8 Putative phospholipase B-like 2 PLBD2 Lysosome 57 4 6808138 Q9HAT2 Sialate O-acetylesterase SIAE Lysosome 2043 124 2408232 O14773 Tripeptidyl-peptidase 1 TPP1 Lysosome 428 26 4505405 Q14956 Transmembrane glycoprotein NMB GPNMB Melanosome 49 11 68509930 P02686 Myelin basic protein MBP Membrane 1867 125 5912201 Q86XK2 F-box only protein 11 FBXO11 Nucleus 21 4 31542868 P13807 Glycogen [starch] synthase, muscle GYS1 Nucleus 6 1 4506645 P63173 60S ribosomal protein L38 RPL38 Ribosome 8 2 4757922 P23435 Cerebellin-1 CBLN1 Synapse 226 14 List of representative proteins commonly detected by LC-MS analyses in all samples. In the table for each protein, NCBI accession (GI number), UniProt accession number, protein and gene name, cellular location, and few details of LC-MS analyses corresponding to all samples are reported. The column “All Samples” refers to overall representative proteins commonly present in all three types of samples which are considered as a single data set. Note that some proteins may have multiple cellular locations: for each protein the most typical and representative cellular location was assigned. A protein eligible to be inserted in this list must be detected in all samples, and classified as representative (detected by SpC ≥ 2 as average value) in at least one of the three types of samples. This list describes the 35 proteins contained in the central overlapping area of Euler diagram in Fig. 3. For full details see Supplementary Data 1. Two ubiquitin-related proteins (UBC and UBA52) and heat shock protein HSP 90-alpha, although present in the three types of samples, were not included in the group of proteins detected in all three types of samples (in this list and in the central overlapping area of Euler diagram in Fig. 3, containing 35 proteins) because they were identified with different GI accession numbers in different samples. Note that the Euler diagram calculation was performed using NCBI accession (GI number) only. Nevertheless, the above mentioned ubiquitin-related proteins and heat shock protein HSP 90-alpha, considered as unique proteins, were detected in all three samples Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Fig. 4 IEM of SN from healthy aged subjects for selected proteins. For number of IEM experiments, see Methods. CTSD (73 y.o.; gold particles = 20 nm). LAMP2 (86 y.o.; gold particles = 20 nm). MAP1LC3B (69 y.o.; gold particles = 15 nm). SCARB2 (69 y.o.; gold particles = 15 nm). SNCA (63 y.o.; gold particles = 15 nm). UBA52 (63 y.o.; gold particles = 15 nm). Lipid bodies are indicated by asterisks. NM pigment of the NM- containing organelles appears as black and electron dense granular aggregates. Scale bar in each panel = 250 nm than other lipid molecules identified in both samples, as shown by likely indicates an inhibition of the lysosomal activity inside NM- TLC analysis. In the lipid extracts from both samples, bands co- containing organelles. The presence of mono- and polysialogan- migrating with sphingomyelin, phosphatidylcholine, lactosylcer- gliosides was confirmed by cholera toxin staining after sialidase amide, and phosphatidylethanolamine were identified. In addi- treatment of the aqueous phases obtained from TIS-NM, but not tion, bands corresponding to galactosylceramide and sulfatide from ORG, likely due to the paucity of the sample (Supplementary (some of the typical myelin lipids) were clearly visible in the lipid Fig. 7). We note that in TIS-NM the membranes and any extracts from TIS-NM, particularly sulfatides as also confirmed by component of organelles are removed after sodium dodecyl LC-MS, but not from ORG samples (Fig. 7). In the ORG sample, sulfate treatment, but lipids and especially dolichols are still there were comparable amounts of some sphingolipids (lacto- adsorbed into NM structure, as shown after solvent (methanol and sylceramide and sphingomyelin, a typical myelin lipid) and hexane) extraction of NM and analysis of these lipids extracts by glycerophospholipids (phosphatidylcholine and phosphatidy- LC-MS and TLC. The results indicate a selective affinity of NM lethanolamine) (Fig. 7): since in membranes the sphingolipid pigment for some lipids, particularly dolichols and dolichoic acids, content is usually lower than that of glycerophospholipids, the 3,9 as previously reported. relative increase of sphingolipids vs. glycerophospholipids in ORG npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. 3,42 LC-MS studies on lipids extracts from TIS-NM. This is probably a consequence of membrane disruption during isolation of TIS-NM, so that oxidized dolichols and dolichoic acids present in cytosol, mitochondria and other organelles are released and then adsorbed by NM pigment. However, the presence of dolichols and dolichoic acids was also confirmed by LC-MS in lipid extracts from ORG samples, thus confirming the specific accumulation of this particular class of lipids inside the NM-containing organelles (Fig. 6), mainly in their lipid bodies. Indeed, Fig. 1d shows an electron microscopy image of NM-containing organelles isolated from SN (ORG samples) with many lipid bodies. In addition, the distribution of dolichols and dolichoic acids chain lengths in lipids extracted from ORG samples was similar to that observed in TIS- NM samples: 14–22 isoprene units for dolichols, and 14–21 isoprene units for dolichoic acids. This suggests that artifacts were absent, as the same types of lipids were identified using two different isolation procedures. No known enzymes involved in dolichol metabolism were observed in these organelles, suggesting that oxidation of dolichols on the double bonds with formation of epoxides does not occur inside NM-containing organelles. The conclusion that dolichols are not synthesized within NM-containing organelles is further supported by the absence of two key enzymes required for dolichol synthesis, i.e., dehydrodolichyl diphosphate synthase Fig. 5 WB (for proteins detected by IEM in Fig. 4) performed on SN complex subunit DHDDS (DHDDS) and polyprenol reductase tissue lysates and on ORG samples. For number of WB analyses, see (SRD5A3) which were not detected by LC-MS, WB, or IEM Methods. CTSD (protein content ratio SN tissue lysate/ORG = 3.1). (Supplementary Fig. 5, 6). The band present in both SN tissue lysate (16 pooled tissues, from The presence of dolichols, dolichoic acids, and other lipids we 48 to 85 years of age) and ORG sample (isolated from one subject, describe in the NM-containing organelle has never been reported 81 y.o.) corresponds to the mature CTSD heavy chain which is highly in previous studies. In the past, dolichols, dolichoic acids, and enriched in ORG sample, considering that the total protein content in ORG was 3.1-fold lower than that of SN tissue lysate. LAMP2 other lipids were reported only in isolated NM pigment, a different (protein content ratio SN tissue lysate/ORG = 1.0). LAMP2 protein situation, as they can be adsorbed into NM during isolation and was lightly present in ORG sample (isolated from one subject, 83 y. originate from cytosol and other organelles that are broken during 28,42,43 o.), while in SN tissue lysate (13 pooled tissues, from 62 to 86 years of the isolation process. Thus, in the present study we provide age) this protein is largely expressed. The antibody used here the demonstration of the presence of these lipids in the lipid recognizes all three LAMP2 isoforms. MAP1LC3B (protein content bodies of intact NM-containing organelles. ratio SN tissue lysate/ORG = 1.8). The black arrowhead indicates the MAP1LC3B-I form, which was more prevalent in SN tissue lysate (five pooled tissues, from 73 to 85 years of age) than the MAP1LC3B-II DISCUSSION form (empty arrowhead indicating the phosphatidylethanolamine conjugated form). In ORG sample (isolated from one subject, 77 y.o.), Integrated methodology for NM samples preparations and the use the MAP1LC3B-I form was abundant while MAP1LC3B-II form was of different analytical methods undetectable. SCARB2 (protein content ratio SN tissue lysate/ORG = The study of protein and lipid pathways of NM-containing 2.3). Here we note an enrichment of SCARB2 in ORG sample organelles requires highly purified and well preserved organelles. (isolated from one subject, 77 y.o.) if compared to SN tissue lysate This is challenging because during their preparation, membrane (five pooled tissues, from 73 to 85 years of age), considering that the and soluble proteins can be lost and contamination of organelles total protein content in ORG was 2.3-fold lower than that of SN by proteins or lipids arising from other cellular compartments may tissue lysate. SNCA (protein content ratio SN tissue lysate/ORG = 3.4). The black arrowhead indicates the soluble-monomeric form of occur. Another factor is that human brain tissues used for SNCA which is clearly visible in SN tissue lysate (nine pooled tissues, preparation of organelles are post mortem and thus affected by from 67 to 85 years of age) while undetectable in ORG sample degradation. The preparation of organelles can itself provide a (isolated from one subject, 66 y.o.). Other bands at higher molecular source of changes in the distribution of proteins and lipids, as weight are present in SN tissue lysate, corresponding to fibrils and discussed in the first paragraph of Results. aggregates with possible modifications. In the ORG sample, two A previous report analyzed by LC-MS/MS the NM-containing main bands are clearly visible corresponding to some aggregated/ organelles isolated from SN tissues, but that isolation procedure modified forms of SNCA (at ~50 and ~58 kDa) which are present also in SN tissue lysate. UBA52 (protein content ratio SN tissue lysate/ was different than that used here and started from frozen ORG= 2.7). The black arrowhead indicates the free ubiquitin that is tissues. It is well known that freezing and thawing of post scarcely visible in SN tissue lysate (eight pooled tissues, from 62 to 89 mortem tissues can break the membranes, with leakage of years of age), but abundant in the ORG sample (isolated from one proteins that can diffuse among organelles as a possible source subject, 66 y.o.). The WB also reveals the presence of large number of of contamination. A later study proposed a new centrifugation immunoreactive high molecular weight bands corresponding to method for the combined isolation and enrichment of NM high amounts of poly-ubiquitinated proteins, both in SN tissue lysate granules (i.e., NM-containing organelles) and synaptosomes from and highly enriched in the ORG sample, although the total protein human SN for proteomic analysis, but again this method content in ORG was 2.7-fold lower than that of SN tissue lysate processed frozen tissues. More recently, the same group performed a proteomic analysis on NM-containing organelles The LC-MS results indicate that a high level of dolichols at obtained by laser capture microdissection from human SN frozen different molecular weights (with 14–22 isoprene units) and their slices. The clear advantage of this new methodology is in oxidized derivatives such as dolichoic acids (with 14–21 isoprene isolating NM-containing samples from very low quantities of units) were present in TIS-NM samples, consistent with previous tissues, but the laser capture microdissection lacks sufficient Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Fig. 6 LC-MS analysis of lipids isolated from TIS-NM and ORG samples. The TIS-NM sample here represented was isolated from a pool of seven subjects (from 71 to 85 years of age), while the ORG sample was isolated from two pooled subjects (74 and 89 y.o.). Mass spectra (averaged mass spectra, range 1200–1500 m/z) demonstrate the presence of dolichols species in both samples. We highlight the series of singly charged ions with different chain lengths, corresponding to dolichols with terminal hydroxyl group, their oxidized derivative dolichoic acids, and acetate adducts of dolichols species. Both spectra selectively show dolichol species with chain lengths ranging from 17 to 21 isoprene units, although few dolichol species with lower and higher number of isoprene units were found in lipid extracts from both samples (Results). Abbreviations used in the figure: Dol, dolichol; Dol-Ac, dolichol acetate; Dol-CA, dolichoic acid resolution to discern among different subcellular components that technology method, which is an excellent gel-free approach and are clearly present among NM-containing organelles of the tissue provides improved selectivity and resolution of peptide separa- 1,3,9,10 (Fig. 1a–c and see previous findings) and therefore can tion, with an increased number of identified proteins and better quantitative determinations in complex mixtures. In order to represent a source of contamination. With such a procedure, the collected samples inevitably would contain different types of improve the identification of protein and lipid pathways, and to debris deriving from other cellular components. distinguish proteins and lipids related to different components of Additionally, the LC-MS/MS employed here for proteomic the NM-containing organelle, we analyzed three different NM analysis provides a multidimensional protein identification preparations (ORG, TIS-NM, and ORG-NM): the NM-containing npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. It is noteworthy that the larger mass of our samples in any of the three types of preparations contained the same 35 proteins (corresponding to ∼80 % of the overall SpC of representative proteins detected in all samples, as shown in Fig. 3), thus demonstrating the high reproducibility and pertinence of the detected proteins in the NM-containing organelle. Euler diagrams (Supplementary Fig. 8) show the comparison of our proteomic data with those previously reported by Tribl et al. 23,24 and Plum et al. Considering all the proteins here identified (Supplementary Fig. 8a), there are 50 proteins found also within the 72 proteins (∼69 %) identified by Tribl and colleagues. Among these overlapping proteins, we observed that 36, 32, and 34 proteins belong to ORG, TIS-NM, and ORG-NM respectively, representing ∼6%, ∼8 %, and ∼15 % of all proteins detected in each of the three types of samples (Supplementary Data 1; Table 1). The observation that NM-containing organelles analyzed in the mentioned study had the highest similarity with our NM pigment isolated from organelles (ORG-NM) suggests that samples in that study featured broken membranes and contained mainly proteins strictly bound to NM pigment. In the present study, we isolated the NM pigment from its organelle by intentionally breaking membranes with a freeze/thaw procedure, while Tribl and colleagues isolated the NM-containing organelles from SN frozen tissues. In parallel, the comparison between the list of all the proteins we identified and the list of 1000 proteins reported by Plum and colleagues shows that 188 proteins (∼19 % of proteins identified by Plum’s group) are shared by the two studies (Supplementary Fig. 8a). A more detailed analysis of the overlapping proteins shows that 102, 99, and 71 proteins belong to ORG, TIS-NM, and ORG-NM respectively, representing the ∼18 %, ∼24 %, and ∼32 % Fig. 7 High performance TLC analysis of total lipid extracts obtained of all proteins detected in each of the three types of samples from TIS-NM and ORG samples. The TIS-NM sample here repre- (Supplementary Data 1; Table 1). Also in this case, the ORG-NM sented was isolated from a pool of four subjects (from 62 to 86 years sample shows the highest similarity with the samples analyzed by of age), while lipids of from three ORG samples (each isolated from Plum et al. (Supplementary Data 1). three different subjects, respectively 62, 61 and 77 y.o.) were pooled before loading onto the TLC plates. After separation, lipids were Among new proteins related to NM-containing organelles detected by spraying the TLC plate with anisaldehyde. In TIS-NM detected by LC-MS, here we report some lysosomal proteins that sample the intense spot at the solvent front likely corresponds to were not found in the two previous studies: e.g., iduronate 2- dolichols and dolichoic acids, as confirmed by LC-MS (Fig. 6). The sulfatase, carboxypeptidase Q, lysosomal acid phosphatase, content of sphingomyelin, galactosylceramide, sulfatides (typical lysosomal alpha-mannosidase, and ribonuclease T2. In addition, myelin lipids), and lactosylceramide is higher than phosphatidy- we reported many other proteins of noticeable interest among lethanolamine and phosphatidylcholine (glycerophospholipids). In those never identified before as associated with NM-containing the ORG samples, the main components are again dolichols and organelles: examples include HLA, GPNMB, transmembrane dolichoic acids at the solvent front. In this sample there are comparable amounts of sphingolipids (lactosylceramide and sphin- protein 106B, F-box only protein 11, protein phosphatase 1 gomyelin) and glycerophospholipids. The arrows at the margin of regulatory subunit 1B, intraflagellar transport protein 81 homolog, the image indicate the position of pure standard lipids co- etc. (Supplementary Data 1). Some of these proteins have been chromatographed with the samples, as described in Methods. independently confirmed by IEM and/or WB as discussed below. Abbreviations used in the figure: GalCer, galactosylceramide; GD1a, Finally, it should be noted that there are 43 proteins shared by GD1b, GM1, GT1b, gangliosides GD1a, GD1b, GM1, GT1b; GlcCer, 23,24 our study and other two studies, considering all the proteins glucosylceramide; LacCer, lactosylceramide; PC, phosphatidylcho- we detected (Supplementary Fig. 8a; Supplementary Data 1). If we line; PE, phosphatidylethanolamine; SM, sphingomyelin; ST, evaluate the most enriched proteins detected in all the studies by sulfatides using different samples and methodologies, we should overlap our representative proteins with the 166 significantly over- organelles isolated from SN, NM pigment isolated from SN tissues, represented proteins reported by Plum’s group and with those and NM pigment isolated from NM-containing organelles. The detected by Tribl et al. The result of this evaluation is a small localization of some proteins in NM-containing organelles and group of 18 proteins (Supplementary Fig. 8b), 16 of which are other cellular organelles was also confirmed by IEM in intact SN lysosomal proteins. If we exclude these 16 typical lysosomal tissue slices. proteins, there are two particularly interesting proteins detected in Notably, in the present study, fresh (rather than frozen and all three studies. These two proteins, both primarily assigned to thawed) SN tissue was examined and the isolation procedure endoplasmic reticulum, are phospholipase D3 and protein enabled us to obtain highly purified NM-containing organelles disulfide-isomerase A3, which seem to be closely related to NM- with well preserved membranes and lipid bodies, as demon- containing organelles and are briefly discussed below. strated in electron micrographs (Fig. 1c). The only contaminant rarely observed in ORG preparations were a few red blood cells, The NM-containing organelle is an autophagic lysosome with 47,48 and so their associated previously identified proteins were particular catabolic features subtracted from the proteomic data sets of ORG samples Protein analyses of the three different types of NM-derived (Methods). samples (ORG, TIS-NM, and ORG-NM) revealed that lysosomal Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. proteins are the major class of proteins in the NM-containing principally involved in glycoproteins, glycosaminoglycans and 59,60 organelle, representing ~60 % of overall representative peptides glycosphingolipids degradation pathways in lysosomes, were identified in all samples (Fig. 2; Supplementary Table 1). detected and in very low amounts (overall 49 SpC, <1 % of Comparison of our data with recent lists of defined human lysosomal SpC). Concerning the shortage of enzymes related to 38,49,50 lysosomal proteins indicates good overlap but also some catabolic pathways of lipids in our samples, an exception is acid differences in the distribution of lysosomal enzymatic classes. In ceramidase and its specific prosaposin (which undergoes proteo- particular, the Euler diagrams (Supplementary Fig. 9) and lytic cleavage to form saposins), both involved in the last step of additional table (Supplementary Data 3) show the comparison sphingolipids degradation pathway in lysosomes. Acid cerami- between our data and the study by Sleat and colleagues, which dase and prosaposin were found in NM-containing organelles by is the most detailed human brain proteomic study performed by relative high amount of peptides (overall 225 SpC and 70 SpC, detecting only the soluble resident lysosomal proteins using respectively). mannose 6-phosphate (Man-6-P) as an univocal lysosomal marker. Thus, it appears that NM-containing organelles possess a low Among all the proteins here identified (Supplementary Fig. 9a), we representation of typical components of phospholipids and found 26 of the 48 (∼54 %) confirmed lysosomal soluble proteins sphingolipids degradation pathways. This could indicate that of human brain; if we consider our representative proteins only, NM-containing organelles lose the ability to conduct specific we found 22 of 48 (∼46 %) confirmed lysosomal proteins enzymatic pathways as it accumulates in neurons, and could be (Supplementary Fig. 9b). related to the particular lipid storage content of NM-containing 3,42 It thus appears that peptidases and a majority of esterases are organelles, consisting mainly of dolichols, for which catabolic overrepresented in our samples, while lipases and glycosylases are pathways are still unclear and unrelated to phospholipid/ underrepresented (Supplementary Data 3). In detail, 10 of 14 sphingolipid pathways. On the other hand, our findings are peptidases (E.C. 3.4.-) belonging to human brain lysosomes were consistent with the presence of undegraded glycerophospholipids detected and in large amounts in our samples (with overall 3243 and sphingolipids in the lipid extract from the lipid bodies. SpC, corresponding to ∼41 % of lysosomal SpC), suggesting again an overrepresented complement of protein degradation pathways Lysosomal membrane proteins are less represented in NM- in NM-containing organelles (see also first paragraph of Results). containing organelles than conventional lysosomes The identification of CTSD in its heavy chain (mature form) as Among lysosomal membrane proteins, we detected by different abundant in all samples, revealed by LC-MS data and confirmed techniques SCARB2, LAMP1, LAMP2, CD63 antigen, type 1 by WB and high IEM gold signals, is notable considering its role in phosphatidylinositol 4,5-bisphosphate 4-phosphatase, and some limiting lysosomal storage diseases and in inhibiting SNCA V-type proton ATPase subunits. In particular, SCARB2 was shown aggregation. Similarly, 6 of 11 lysosomal esterases (E.C. 3.1.-) by LC-MS, IEM and WB to be the most abundant lysosomal identified in human brain lysosomes were detected by overall membrane protein in NM-containing organelles. One function of 2523 SpC, corresponding to ∼32 % of lysosomal SpC (Supplemen- SCARB2 may be to transport β-glucocerebrosidase into the tary Data 3). Among these proteins, sialate O-acetylesterase, a key lysosome, although as above β-glucocerebrosidase was not enzyme involved in sialic acid catabolism, was the lysosomal identified in this study by LC-MS. SCARB2 also belongs to the esterase we detected by the highest number of peptides (i.e., scavenger receptor class B family involved in the transport of high overall 2043 SpC). density and low density lipoproteins, cholesterol esters, phospho- In contrast, typical lysosomal human brain enzymes mainly 62,63 lipids and oxidized phospholipids, and if overexpressed, involved in lipids, phospholipids, and sphingolipids catabolism, induces endosomes/lysosomes enlargement and cholesterol including lysosomal acid lipase, group XV phospholipase A2, and accumulation in the enlarged compartments. The abundance sphingomyelin phosphodiesterase were not detected. We of dolichols into the NM-containing organelles could be related to observed only elevated quantities of PLBD2 and phospholipase the presence of high amounts of SCARB2 in NM-containing D3, two poorly characterized proteins of unknown functions, that organelles. This protein is present both on the organelle were identified both as potential proteins of human brain membrane and its lumen, especially in lipid bodies and sometimes lysosomes and in previous studies on NM-containing orga- 23,24 in NM pigment, and is apparently accumulated in the NM- nelles. PLBD2, also confirmed and localized by IEM signals, is a 37,38 containing organelle during aging (see below). new putative lipase with uncertain enzymatic activity, with the Other lysosomal membrane components were identified to a exception of a homolog protein in amoeba that cleaves acyl lesser extent, including LAMP1, LAMP2, CD63 antigen, type 1 chains of some phospholipids (phosphatidylinositol, phosphatidy- phosphatidylinositol 4,5-bisphosphate 4-phosphatase, and some lethanolamine, and phosphatidylcholine). Due to the accumula- subunits of V-type proton ATPase. Considering the probable tion of dolichols in NM-containing organelles, and considering partial loss of membranes during isolation procedures and/or due that dolichols may be transported to lysosomes as dolichyl esters to problems in detecting membrane proteins by LC-MS, IEM and and then hydrolyzed by an unknown dolichyl esterase, we WB analyses were performed, which demonstrated the presence attempted to test by molecular docking if dolichyl esters are of LAMP1, LAMP2, and ATP6V1B2 subunits, mainly in the luminal possible substrate of PLBD2, but obtained no fit (not shown). portion of the NM-containing organelles. The antibody used for Similarly, despite its classification as an esterase, neither a definite LAMP2 detection by IEM and WB recognizes all three LAMP2 enzymatic activity nor specific substrates have been clearly 55,56 isoforms, and so we cannot distinguish between LAMP2A (the reported for phospholipase D3. Nevertheless, phospholipase 55,56 chaperone-mediated autophagy receptor), LAMP2B (probably D3 is abundantly expressed in brain and neural tissues, is involved in macroautophagy), and LAMP2C. Autophagosomes correlated with the modulation of cellular resistance to oxidative possess LAMP2B and LAMP2C isoforms, which have uncertain stress and has been recently indicated as a key factor in the functions distinct from chaperone-mediated autophagy. How- pathological processes of Alzheimer’s disease. The finding of ever, LAMP2 appears to have a low presence with abnormal large amounts of both of these two recently discovered enzymes location in the NM-containing organelle, which could be in all samples indicates a need for further investigation into their consistent with the age-related decreased level of this protein, roles in NM-containing organelles and lysosomes, in particular on particularly for LAMP2A in lysosomes, as well as a general metabolism and storage of lipids. Another group of less represented enzymes are glycosylases: decline of autophagic-lysosomal function that occurs in normal indeed, only 5 of at least 14 enzymes classified as glycosylases and aging. These observations suggest that NM-containing organelles reported by Sleat and colleagues (Supplementary Data 3), are likely derived from macroautophagic organelles (see next npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. paragraph), rather than lysosomes specialized for chaperone- oxidative stress. APOD is consistently upregulated and highly mediated autophagy. expressed during normal aging, in overall SN tissue of PD 75 72 The identification by LC-MS of only low levels of V-type proton patients and other neurodegenerative diseases, where oxida- ATPase functional subunits, responsible for acidification of tive stress and lipid abnormalities are implicated. As mentioned, lysosomes, may indicate a decreased acidification and diminished SCARB2 accumulates in the NM-containing organelle as also lysosomal catabolism. As vacuole fusion requires an electroche- reported for lysosomal inclusion bodies. While SCARB2 is a mical membrane potential created by the V-type proton ATPase, protein of the lysosomal membrane, it was also found on the NM-containing organelles may have a low capacity for fusion with luminal side of the NM-containing organelle particularly in lipid lysosomes or autophagosomes. IEM experiments revealed that bodies and sometimes in NM pigment. Although some of these ATP6V1B2 subunit is not evident on the membrane of NM- proteins may be enriched in the NM-containing organelle due to containing organelles (in contrast to lysosomes) but is sparsely their particular role or because they are normally overexpressed located in NM pigment and lipid bodies, confirming a likely during aging, these proteins may be also accumulated due to functional deficiency of V-type proton ATPase in the NM- impaired degradation in NM-containing organelles. containing organelles. We note that ATP5G1/2/3, that we found in NM-containing organelles, is the primary marker of broad range of neuronal ceroid lipofuscinoses (i.e., CLN2, -3, -4, -5, -6, -7, -8, -9, CLCN7). The NM-containing organelle originates from macroautophagy ATP5G1/2/3 accumulates in autophagic vacuoles and lysosomes of and accumulates MAP1LC3B neurons where autophagy or some lysosomal enzymes are Little is known about the origin of the NM-containing organelle. 76,77 blocked, as in lysosomal storage disorders. Intriguingly, this Experiments in cultured neurons showed that an excess of protein also accumulates inside autophagic vacuoles in normal cytosolic DA induces NM synthesis and NM-containing organelles aged mice, consistent with a decline of autophagic-lysosomal formation, and the induced pigment was chemically identical to function during normal aging. Similar to the age-dependent 2,10 human NM as assessed by electron paramagnetic resonance. In 78,79 accumulation of dolichols in brains of the elderly, which was addition, the identification of a double membrane around these greatly increased in patients with neuronal ceroid lipofuscino- induced organelles of NM and surrounding NM-containing 80,81 sis, ATP5G1/2/3 accumulates in lipofuscin-like lipopigments 2,10 organelles of the human SN, as previously reported and here inside normal neurons during aging, a process amplified in confirmed (Fig. 1c), together with presence of many lysosomal neuronal ceroid lipofuscinosis and other lysosomal disorders. hydrolases, suggests that the NM-containing organelle is a Interestingly, these previously reported observations are consis- pigmented autophagic vacuole. tent with the LC-MS detection of this protein specifically in TIS-NM We have shown in previous studies electron microscopy images samples, which can be associated to the early stages of NM- of NM-containing organelles of human brain that have a clearly containing organelle formation. Additionally, saposins, which were different ultrastructural appearance than lipofuscin and lyso- detected in our samples in the form of prosaposin, have been 2,3,9,10 somes. Another paper described the species-specific ultra- identified as the main storage material (particularly saposins A and structure of neuronal lipofuscin by electron microscopy, thus D) in two types of neuronal ceroid lipofuscinosis, CLN1 and showing morphological features of lipofuscins that are quite CLN10. different from that of NM-containing organelles. Other proteins accumulated inside NM-containing organelles Here we report the presence of the macroautophagy marker are cerebellin-1 and cerebellin-2, which have unknown functions. MAP1LC3B inside NM-containing organelles, confirming its Cerebellin-1 (which we found in all analyzed samples) is autophagic nature. Using IEM on SN tissues, we found MAP1LC3B preferentially expressed in cerebellar synapses, where it is localized around lipid bodies, lining membranes and remarkably required for synapse integrity and plasticity, but is also present on NM pigment within organelles. This evidence confirms that at variable concentrations elsewhere in the brain, and a study NM-containing organelles may derive by formation of MAP1LC3B- reports its presence in the endo-lysosomal compartments of positive autophagosomes that engulf forming NM and related neurons. Cerebellin-1 is more highly expressed in SN (A9) proteins and lipids components present in the cytosol, and neurons than ventral tegmental area (A10) neurons of mice, successively fuse with lysosomes, forming NM-containing auto- suggesting a mechanism for its accumulation similar to that of NM lysosomes that become NM-containing organelles (Fig. 8). in dopaminergic neurons of SN, and is also highly expressed in The level of MAP1LC3B-II is thought to be a reliable indicator of mucopolysaccharidosis type IIIB mouse brains. autophagosome formation. However, in our ORG samples, the Previous studies have suggested the storage features of the MAP1LC3B-II form was undetectable by WB compared to NM-containing organelle: (i) indigestible NM pigment increases in MAP1LC3B-I, either because it was delivered at low levels or concentration in neurons from very early life over the entire life because it was normally degraded by lysosomal enzymes. It is 3–6 3,9,42,43 span; (ii) NM pigment binds high quantities of dolichols; likely that MAP1LC3B-II form could be degraded by the lysosomal and (iii) NM pigment accumulates large amount of endogenous hydrolases we found, as normally occurs after the fusion of our 3,15 and environmental metals. We now report the identification of special autophagosomes with lysosomes (Fig. 8), and therefore storage material in these aged organelles, suggesting that this transient supply of lysosomal enzymes would degrade some mechanisms for degrading protein and lipid components are substrates. Indeed, after autophagosome fusion with lysosomes to impaired, despite the presence of several enzymes inside NM- form autolysosomes, intra-autolysosomal MAP1LC3B-II is normally containing organelles. This evidence, together with low levels of degraded by lysosomal hydrolases and is difficult to detect. some catabolic lysosomal enzymes and shortage of lysosomal membrane proteins (as discussed in the previous sections on The NM-containing organelle is an aged and impaired lysosomal- lysosomal proteins), including V-type proton pump ATPases related organelle that accumulates proteins, indigestible NM, and involved in acidification, suggests that the NM-containing dolichols organelle is an aged impaired lysosomal-related organelle unable In addition to acid ceramidase, other proteins (as previously to completely digest its content. described) including tripeptidyl-peptidase 1 and APOD were This organelle, due to its particular content of undegradable NM present in high amounts in the samples and have been also pigment, proteins, lipids and metals (see below), could be also a reported as major components of lipofuscin-like lysosomal source of oxidative stress (Fig. 8). Therefore, the presence of inclusion bodies. APOD is involved in binding and transport of proteins involved in protection against oxidative stress are lipids, in their protection from oxidation and consequent notable: i.e., phospholipid hydroperoxide glutathione peroxidase, Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. b c β-structured Alpha-crystallin B MAP1LC3B, Proteasome proteins chain, ubiquitin, etc. SQSTM1 3+ a Phagophore Fe Polymerization 3+ Fe formation DA-quinones 3+ Fe Oxidation Macroautophagy 3+ Fe 3+ DA-protein Fe GPNMB, tubulin adduct polymerization- Iron-melanin-protein promoting protein, etc. complex 3+ Fe 3+ DA Fe Lipids Vesicles or other accumulation autophagic vacuoles Autophagosome NM-containing organelle Aging Fusion Lysosome Proteases, NM RAB5A, tubulin Proteins glycosylases, lipases, accumulation polymerization- Autolysosome accumulation proton pumps, etc. promoting protein, etc. h f Fig. 8 Hypothesized scheme summarizing NM-containing organelle formation in human SN. a, b In the cytosol of SN neurons, DA can be oxidized to semiquinones and quinones via iron catalysis, and these highly reactive compounds can react with aggregated and β-structured proteins that accumulate in the cytosol. c The oxidative polymerization of quinones initiates with the formation of the melanin-protein complex which can also bind high levels of metals, especially iron. During this step drugs and toxicants can also bind to the melanin-protein complex. Proteins damaged by misfolding and DA-adducts formation may be recognized and bound by ubiquitins (green) and alpha- crystallin B chain, in the attempt to degrade damaged and/or misfolded proteins in the proteasome pathway. It may be that ubiquitinated- NM-derived products are too large and damaged to be degraded by the proteasome system. d, e The resulting undegradable material accumulates in the cytosol. GPNMB and tubulin polymerization-promoting protein could be engaged at this step, since these proteins are 105,110 involved in the formation of aggresome-like structure and degradation of cellular debris. The accumulated undegradable material is then taken up into autophagic vacuoles by the phagophore, an isolation double membrane which engulfs bulk material for macroautophagy. This is confirmed by the presence of some typical macroautophagic markers such as MAP1LC3B (red) and SQSTM1 (orange), and by the presence of a double membrane surrounding the NM-containing organelle as displayed by electron microscopy (Fig. 1c).f These autophagic vacuoles fuse with lysosomes, shown in the scheme with numerous enzymes (black), different membrane proteins (blue) and the proton pumps (violet), to become autolysosomes containing the enzymes, proteins and lipids of lysosomes. After fusion with lysosomes, the undegraded and NM-derived material contained in the autophagic vacuoles can interact with other lipids and other proteins carried by lysosomes. A decreased lysosomal enzyme efficiency and reduced fusion capacity could occur also as a consequence of aging, oxidative stress and NM accumulation. g These organelles can fuse with other vesicles or with other autophagic vacuoles containing NM precursors or with old NM-containing organelles, etc. This fusion could lead to the accumulation in the lumen of the organelle of membranous portions which would otherwise be degraded, while dolichols in particular are not chemically decomposed and accumulate with other undegraded lipids leading to the formation of lipid bodies. Dolichols (or dolichyl esters) may be transported into the organelle by vesicle transport and membrane fusion. Fusion of organelles could be mediated by RAB5A, tubulin polymerization-promoting protein, microtubule-associated protein tau and other related proteins. This aged organelle, due to its particular content of undegradable NM pigment together with damaged and oxidized proteins, lipids and metals, is a reservoir of buffered toxins (red stars). Under conditions of cellular damage these toxins together with NM pigment could be released and induce neuroinflammation and neurodegeneration. h The final NM-containing organelle is the result of a complex and continuous process occurring during aging, that leads to the accumulation of undegradable material in specialized pigmented “autophagic lysosomes”. Figure modified from previously published papers, by permission of Springer and by permission of Elsevier protein DJ-1, superoxide dismutase [Cu–Zn], and APOD. Here, we of neurons not containing NM or glia. Oligodendrocytes showed briefly highlight two examples. Phospholipid hydroperoxide the highest positive reactions for both FTL and FTH1, while in glutathione peroxidase, which contributes to redox balance in neurons enriched with NM-containing organelles the staining for 5,6 cells, protects lipid membranes from oxidation and was previously FTH1 and FTL was generally undetectable with this technique. observed to co-localize with NM pigment of dopaminergic SN Due to technical limitations of peroxidase immunohistochemistry, neurons. Indeed, this hydroperoxidase was significantly reduced in low quantities of these iron storage proteins could not be overall parkinsonian SN compared to controls, but was increased revealed. By using techniques with high sensitivity (LC-MS and 86 89 in the surviving SN neurons. Protein DJ-1 has been identified as IEM), we now confirm the presence of FTL previously reported an atypical peroxidase that scavenges hydrogen peroxide derived and, to a lesser extent, of FTH1 in NM-containing organelles of the toxicity and mutations in its gene are associated with early-onset SN. In conclusion there are low levels of ferritins inside NM- PD. containing organelles, while the abundant NM pigment is the 3,5,6,14 We investigated FTH1 and FTL, proteins involved in iron major iron storage complex of pigmented neurons. homeostasis/storage, in NM-containing organelles. Using immu- The presence of HLA in NM-containing organelles and its nohistochemistry, we previously reported that FTH1 and FTL accumulation in NM pigment may have important consequences content in NM-containing neurons of SN is much lower than that for preferential vulnerability of catecholaminergic pigmented npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. neurons in SN and locus coeruleus of PD patients. HLA expression Tubulin polymerization-promoting protein, a protein involved in 105,106 is higher in SN and locus coeruleus pigmented neurons than other the maintenance of microtubule network stability, plays also brain neurons, and SN dopaminergic neurons in culture can a role in SNCA aggregation and co-localizes with aggregated SNCA in Lewy bodies inclusions in a group of α-synucleinopa- express HLA that can bind peptides from endogenous or thies. We found tubulin polymerization-promoting protein in exogenous proteins and present them on neuronal membrane + 20,90 ORG and ORG-NM samples, suggesting its possible involvement in leading to targeting by CD8 lymphocytes and death. NM synthesis and NM-containing organelle formation due to its reported role in initiating the formation of cellular aggresome-like Proteins involved in aggregation, degradation pathways, and structures and inclusion bodies. potential precursors of NM synthesis The presence of GPNMB in all isolated samples, as also The presence of abundant UBC and UBA52 in the NM pigment confirmed by WB and IEM, is remarkable. The melanosomal inside the NM-containing organelles, as demonstrated by LC-MS 107,108) protein GPNMB (refs. is found in melanosomes of MNT-1 and IEM, suggests that ubiquitinated proteins likely participate in 100 109 cells by proteomics analysis, participates in melanogenesis, early steps of NM synthesis in the cytosol. In addition, WB data and in control of macroautophagy and bulk degradation in the demonstrate that high levels of ubiquitinated proteins with high cytosol, as a recruiter for MAP1LC3B. GPNMB could play a role molecular weight are detected in the NM-containing organelle. in formation of NM autophagic vacuoles and fusion to produce Ubiquitination can direct proteins, particularly when partially the final NM-containing organelle. Notably, its corresponding unfolded or damaged, to either proteasome or lysosome; if gene is a new PD risk loci reported in a wide association meta- proteasome is unable to degrade all ubiquitinated proteins, analysis. This suggests that mutations of this gene could macroautophagy could provide an important compensatory encode a modified GPNMB protein unable to participate in the 91–93 mechanism. macroautophagic process producing the NM-containing orga- Ubiquitination probably occurs during NM formation attempt- nelle, leaving the neuron exposed to toxic species. A recent gene ing to degrade proteins damaged by DA-modification. Proteins expression study on 6-hydroxydopamine animal models of PD may be modified by DA-quinones, as we have previously detected revealed that GPNMB (as well as other genes belonging to by chemical degradation of NM pigment isolated from SN (TIS- regeneration-associated genes) was highly upregulated in SN NM) high amounts of cysteinyl adducts with DA and with 3,4- early after the lesion. This upregulation could be a response dihydroxyphenylalanine (DOPA), in addition to DA and DOPA. associated with axodegenerative process of SN neurons after This suggests that quinones of DA and DOPA are trapped by lesions, and could also exhibit axoprotective or regenerative cysteine residues of proteins (although histidine residues can properties. similarly react with quinones), producing DA- and DOPA-modified proteins during the early steps of NM biosynthesis. It may be that The lipid bodies of NM-containing organelles contain mainly ubiquitinated-, NM-derived products are too large to be degraded dolichols by the proteasome and therefore are removed by macroauto- The lipid bodies in NM-containing organelles contain mostly phagy and finally stored in the NM-containing organelle (Fig. 8). dolichols and dolichoic acids, although we did not find proteins We further detected SQSTM1, an important partner of MAP1LC3B, for dolichol synthesis or transport, including two important required for the degradation of ubiquitinated aggregates by enzymes for the last steps of synthesis of dolichols (DHDDS and autophagy. Interestingly, SQSTM1 accumulates in ubiquitin-rich SRD5A3). No dolichol-specific transporter has been yet inclusion bodies in neurodegenerative protein aggregation characterized, and only a single study to our knowledge reported 95,96 diseases. The formation of protein inclusion bodies enriched that dolichols intermediates in the human blood are normally in SQSTM1 and ubiquitins is a typical response to stress conditions transported by low-density lipoproteins and may be involved in including amino acid starvation, oxidative stress, and inhibition of their accumulation in lysosomes during normal aging and 94,97,98 lysosomes and autophagy. 114 lysosomal diseases. However, we could not detect the receptor Alpha-crystallin B chain and heat shock protein HSP 90-alpha for low-density lipoproteins as representative in our samples. were found in our samples and could play a role similar to that of Dolichols may exist as free forms with free terminal hydroxyl SQSTM1 and ubiquitins during NM-containing organelle forma- group, or in phosphorylated forms (dolichol phosphate is required tion. Heat shock protein HSP 90-alpha is a chaperone that for glycoprotein biosynthesis), or esterified with fatty acid, which is promotes protein folding and might rescue damaged proteins, 113,115 the dominant form of dolichols in animal tissues. It is and has been detected also in melanosomes. Alpha-crystallin B possible that esters of dolichols with fatty acids (dolichols are chain is a small heat-shock protein that can function as a usually transported to the lysosomes in the esterified form) are molecular chaperone and prevents fibrillization of proteins, transferred into NM-containing organelles, where they could be 31,101 particularly SNCA: it is sometimes present with ubiquitinated slowly hydrolyzed to free dolichols by the abundant hydrolase proteins and SNCA in Lewy bodies, and recently was found activities (esterases) we detected. The intracellular accumulation markedly upregulated in the SN of PD patients. The finding of of dolichyl esters may occur because a majority of lipids are this protein in both the ORG and ORG-NM, but not the TIS-NM degraded by cytosolic, lysosomal, and mitochondrial enzymes, sample, suggests that the protein might not be strictly bound to while dolichols are poorly degraded by unknown pathways and 116,117 NM pigment and is present as a component of the protein matrix unidentified specific catabolic enzymes. It may be that of the NM-containing organelle. dolichols (or dolichyl ester) can be transported into the NM- Protein disulfide-isomerase A3, a protein found only in TIS-NM containing organelle by vesicle transport and membrane fusion sample and then strictly bound to NM pigment, was also reported from other organelles (other NM-containing organelles, lysosomes, by previous proteomic studies as non-lysosomal protein of NM- etc.). Indeed, the largest lipid bodies present into the NM- 23,24 containing organelles. This enzyme catalyzes disulfide bond containing organelles are mainly located on the external portion formation, reduction, or isomerization and belongs to a family of organelle and are membrane bound (Fig. 1a–c), although we responsible for quality-control system to ensure the correct cannot distinguish between a normal bilayer or a single layer; the folding of proteins. Interestingly, a member of this family of smallest lipid bodies are observed both embedded into NM proteins was shown to co-localize with SNCA in brainstem and pigment and in the external portion of the organelle. cortical Lewy bodies of subjects with neurodegenerative Although the precise role of dolichols is not established, they 104 118 diseases. may be important for membrane trafficking, and stimulating Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. fusion of lipid vesicles. It is possible that dolichol-containing high number of cysteine and histidine groups, which readily react membranes continuously fuse during NM-containing organelle with quinones. formation and maturation (NM-containing autophagosomes with It appears that enzymatic (if any) oxidation of DA to form melanic oligomers occurs in the cytosol, likely via iron catalysis or new lysosomes, with old NM-containing organelles, etc.), and by some of the oxidative enzymes (Fig. 8). Evidence for enzymatic consequently, dolichols may accumulate in NM-containing orga- control of the oxidative process leading to NM pigment formation nelles that are unable to efficiently degrade them (Fig. 8). Thus, has been discussed but has not been demonstrated dolichols in the organelle could derive from incompletely 13,14,127,128 directly. Interestingly, we found two oxidoreductase degraded membranes of other organelles. enzymes in TIS-NM sample by LC-MS. The first is the 3 beta- The phospholipids and sphingolipids we found in NM- hydroxysteroid dehydrogenase type 7 (found exclusively in TIS- containing organelles likely derive from the organelle’s own NM sample), which is normally localized in the endoplasmic membranes and incomplete degradation of membranes of reticulum. The other enzyme is the superoxide dismutase [Cu–Zn], autophagic vacuoles that fuse with NM-containing organelles, as normally found in the cytoplasm, which was found in TIS-NM with well as by vesicle transport into NM-containing organelles. SpC higher than that of the previously mentioned enzyme, and Sulfatides are synthesized and accumulated predominantly in was also found with slightly higher SpC as fragment in the ORG oligodendrocytes and are a major component in myelin sheaths, sample. This suggests that superoxide dismutase [Cu–Zn] could be although low amounts of sulfatides have been detected in 120 an additional candidate enzyme involved in the oxidation of DA neurons and astrocytes. The sulfatides accumulated in NM- and related catecholamines during the early phases of NM containing organelles (at very low level in ORG samples compared synthesis. In addition, the existence of an auto-oxidation process to purified TIS-NM, see Results) could derive from disruption of of DA and related catecholamines must be also considered in the oligodendrocytes, dopaminergic and non-dopaminergic axons biosynthesis of NM pigment, a process that to date is only partially and enter the NM-containing organelles through endocytosis, understood. The presence of iron(III) promotes the oxidation of DA vesicle transport and fusion. NM-containing organelles are also into highly reactive quinones that can form NM pigment. characterized by the presence of some typical neuronal sphingo- Indeed, NM synthesis may be driven by an excess of cytosolic lipids, for example gangliosides (as revealed by LC-MS), which catecholamines not accumulated in synaptic vesicles, as sug- likely do not undergo complete degradation due to the loss of 10 gested previously. Moreover, the oxidation of DA to semiqui- activity of glycosyl hydrolases involved in their catabolism, nones and quinones can generate aminochrome and 5,6- consistent with the notion that the NM-containing organelles indolequinone, which can induce toxicity if these species are are a form of aged and impaired lysosomal-related organelle. not taken into NM synthesis. These reactive compounds are reported to cause dysfunction in mitochondria and protein NM synthesis begins in cytosol and intermediate products are degradation, and to promote aggregation of SNCA to toxic transported into organelles protofibrils and produce oxidative damage. The melanic component derived from the oxidative polymer- We did not find significant amounts of typical enzymes and ization of DA and/or other related catecholamines likely binds to proteins involved in melanogenesis (i.e., tyrosinase, dopachrome 100,121,122 β-structured proteins, since X-ray scattering studies performed on tautomerase, melanocyte protein PMEL17, etc.), with the isolated NM pigment showed a diffraction pattern of about 4.7 Å, sole exception of the melanosomal protein GPNMB, a glycoprotein typical of cross-β sheet structured protein aggregates. Since with high homology to the structural melanocyte protein 108,122 GPNMB and SNCA (both capable of forming insoluble fibrils with PMEL17. In addition to the role suggested above for GPNMB cross β-structure, and WB on ORG samples indicate the presence in autophagy to form the NM-containing organelle, this protein of complex aggregates) were found in different parts of the NM- could produce amyloid fibrils assembling into sheets on which containing organelle, including NM itself, we propose that NM DA-quinones would start NM synthesis (Fig. 8). Indeed, it was formation may start with the accumulation of β-structured protein shown that in the absence of glycosylation, the NTR-PKD domain aggregates in the cytosol, resulting in the formation of peptide/ region of GPNMB retains the intrinsic capacity to form amyloid in protein “seeds”, either protofibrils or even fibrils, which would cell cultures. react with excess of DA and/or quinones of DA metabolites, Likewise, we did not find DA transporters, with the sole followed by polymerization to form melanin-protein conjugates. exception of the synaptic vesicular amine transporter which was The protein-DA modifications and subsequent polymerization was detected at very low levels only in ORG-NM sample and 126,132 recently demonstrated in the synthesis of NM models. The categorized as non-representative. However, it was shown in the oxidation of DA and the resulting complex between melanin and past that human DA neurons of SN with high levels of NM many β-structured proteins, which could include fibrillar forms of pigment have low levels of synaptic vesicular amine transporter 133–135 SNCA, GPNMB, and other proteins with cross β-structure, is expression and vice versa. Similarly, the overexpression of promoted by reactive iron which is abundant in cytosol of SN synaptic vesicular amine transporter in neuronal cell cultures, 5,14 neurons. Iron(III) could accumulate within the conjugates treated with L-DOPA which is rapidly converted to DA, inhibited during this process. Indeed, NM pigment in the organelle binds 3,5,14 the synthesis of NM by lowering the cytosolic concentration of iron(III) in two distinct iron-binding sites with different affinity. DA. It is noteworthy that the protein phosphatase 1 regulatory This ternary complex of iron-melanin-protein formed in the subunit 1B, that we found in all samples, is involved in L-DOPA cytosol can be accumulated into autophagic vacuoles and induced dyskinesia in PD. The L-DOPA increases cytosolic DA and transported to the lysosome. Here, the proteases would cleave protein phosphatase 1 regulatory subunit 1B phosphoryla- most protein chains, generating a complex with higher ratio of 124,125 3,9 tion. This protein could be involved in the first steps of melanin:protein in the final NM pigment. This lower polarity NM synthesis by reacting with high cytosolic DA and related melanin:protein complex will more easily react with dolichols, and catecholamines or their derived products (semiquinones and to a lesser extent with other lipids, released by lipid bodies present quinones), due to the presence of cysteines and histidines in its in the organelle to form the insoluble NM pigment which is then structure. Another protein that may react with quinones during continuously accumulated inside membrane-bound organelles. the early steps of NM synthesis could be the cysteine-rich protein This reaction of the complex iron-melanin-protein with dolichols 2, which was detected in ORG-NM samples. More broadly, likely occurs through iron mediated radical oxidation at the candidate proteins that could be incorporated into NM pigment carbon atom adjacent to the double bond in the isoprenic unit of during the early phases of biosynthesis are those containing a dolichols. npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. We found that dolichols are covalently bound to the melanic component of NM pigment. The protein-melanin conjugate thus 3,9,42 component within the NM structure. The isolated NM formed traps iron and other metals, and is then accumulated into pigment is quite insoluble both in water and organic solvents, autophagic vacuoles and carried to lysosomes. Here, the proteases although NM pigment contains a relatively low molecular weight likely cleave most of protein/peptide chains of the protein- component that is slightly soluble in dimethyl sulfoxide with melanin conjugate which react with dolichols to form the final essentially the same composition of the insoluble component. NM. The insoluble portion seems to contain polymers of larger size, more dolichols and less saturated lipids than its dimethyl METHODS sulfoxide-soluble counterpart. The soluble portion of NM pigment likely consists of oligomeric precursors of polymeric NM that have Antibodies molecular weights between 1.4 and 52 kDa. The presence of large For WB and IEM experiments, the following primary antibodies against amounts of oligomeric and polymeric structures further confirms different proteins were used: APOD, mouse monoclonal (Abcam, Cam- that NM synthesis and accumulation are continuous processes bridge, UK); ATP5G1, mouse monoclonal against 18–137 sequence occurring in neuronal organelles (Fig. 8). (Abcam), and this antibody recognizes mature chains of ATP5G1, ATP5G2 and ATP5G3, which are identical in their sequence; ATP6V1B2, mouse monoclonal (Santa Cruz Biotechnologies Inc., Santa Cruz, CA, USA); CTSD, CONCLUSIONS goat polyclonal against the C-term (Santa Cruz Biotechnologies); DHDDS, rabbit polyclonal (Atlas Antibodies AB, Stockholm, Sweden); FTH1, goat The comparison and integration of data from different methods of polyclonal (Abcam); FTL, rabbit polyclonal (Abcam); GPNMB, goat preparation and analysis provide a highly reliable and compre- polyclonal against 23–486 sequence (R&D Systems, Minneapolis, MN, hensive description of protein and lipid pathways of NM- USA); HLA, mouse monoclonal (Santa Cruz Biotechnologies); LAMP1, containing organelle of the human SN. We found that the NM- mouse monoclonal (Developmental Studies Hybridoma Bank, Iowa City, IA, containing organelle possesses particular membrane and soluble USA); LAMP2, mouse monoclonal recognizing all LAMP2 isoforms proteins typical of lysosomes, while other characteristic lysosomal (Developmental Studies Hybridoma Bank); MAP1LC3B, rabbit polyclonal proteins are missing or scarcely expressed. Typical lysosomal (Cell Signaling Technology, Danvers, MA, USA) and rabbit polyclonal enzymes abundant in NM-containing organelles are peptidases, (Abgent, Inc., San Diego, CA, USA); PLBD2, rabbit polyclonal against sulfatases and esterases. However, the organelle has lower levels 448–565 sequence at C-term (Atlas Antibodies AB); RAB5A, rabbit of lipases and glycosylases than typical lysosomes, and so exhibits polyclonal (Abgent, Inc.); SCARB2, mouse monoclonal (Santa Cruz limited degradation pathways for some molecules. The reduced Biotechnologies); SNCA, rabbit polyclonal (EMD Millipore, Temecula, CA, lysosomal activity and fusion capacity could be a consequence of USA); SQSTM1, mouse monoclonal (Abcam); SRD5A3 rabbit polyclonal inadequate localization of some lysosomal membrane proteins, (Atlas Antibodies AB); UBA52, rabbit polyclonal against ubiquitin (DakoCy- tomation, Glostrup, Denmark). We have indicated detailed immunogenic including V-type proton ATPase, which normally acidifies sequence only for proteins that undergo relevant molecule processing. lysosomes. The LAMP2 is at low levels and inadequately located in NM- containing organelles, suggesting that lysosomes specialized for Brain tissues chaperone-mediated autophagy do not form NM-containing This study was approved by the Institutional Review Board of the Institute organelles. In contrast, the presence of double membrane of Biomedical Technologies—National Research Council of Italy (Segrate, surrounding many of the pigmented organelles and the finding Milan, Italy) and was carried out in agreement with the Policy of National of proteins such as MAP1LC3B and SQSTM1 demonstrate the Research Council of Italy. Written informed consents for using brain macroautophagic nature of NM-containing organelles, which samples for research purposes were obtained from closest relatives and derive from autophagosomes that engulf NM precursors, lipids are stored at the Section of Legal Medicine and Insurances, Department of and proteins from cytosol. Biomedical Sciences for Health, University of Milan, Milan, Italy. Final approval was given by the pathologist performing the autopsy. All tissues In lipid bodies of NM-containing organelles, the major samples were analyzed anonymously. components accumulated are dolichols and dolichoic acids, likely In this work, SN samples were obtained during autopsies of male and transported by SCARB2 and APOD. The high accumulation of female subjects who died at different ages without evidence of dolichols in NM-containing organelles may also be mediated by neuropsychiatric and neurodegenerative disorders. The healthy subjects membrane fusion from other organelles and vesicle transport of included in this study at pathological examination did not show any dolichyl esters, and their subsequent hydrolysis to dolichols. macroscopic alteration of neurological and vascular type. Histological The NM-containing organelle accumulates undegradable NM examination (on formalin-fixed and paraffin-embedded tissue sections) pigment, dolichols, lipids, proteins, and metals over the entire revealed no Lewy bodies and other pathological markers. lifespan. Several proteins have been detected in spite of the All samples analyzed and included in this study were obtained from presence of different proteases and other degradative enzymes, as healthy elderly subjects, and the age ranges were the following: (i) ORG degradation processes including those dependent on organelle samples for LC-MS and WB analyses of proteins, LC-MS and TLC analyses of acidification are inhibited or inefficient. Alpha-crystallin B chain lipids (subjects ranging from 61 to 93 years of age); (ii) TIS-NM samples for and heat shock protein HSP 90-alpha were found as well as LC-MS analyses of proteins, LC-MS and TLC analyses of lipids (isolated from pooled SN of subjects ranging from 48 to 92 years of age); (iii) ORG-NM tubulin polymerization-promoting protein, GPNMB, ubiquitins samples for LC-MS analyses of proteins (subjects ranging from 60 to 82 (along with several other proteins), likely consistent with their years of age); (iv) SN tissue lysates for WB analyses of proteins (subjects involvement in macroautophagy and bulk degradation in the ranging from 48 to 89 years of age); and (v) SN tissue slices for IEM process of NM synthesis. (subjects ranging from 63 to 91 years of age). For each type of preparation Among proteins accumulated inside NM-containing organelles, intended for a specific determination, we have used tissues with similar (as a striking presence is that of HLA, that could increase the much as possible) post mortem intervals to use the most reproducible vulnerability of NM-containing neurons, since this protein can conditions. It is noteworthy that all the analyses described in this study present antigens on cell membranes so that neurons would be were performed on sets of samples coming from subjects with overlapping targeted by T-lymphocytes. age ranges. We hypothesize that the NM synthesis starts in the cytosol with Indeed, we were studying brain aging and wanted to investigate elderly accumulation of aggregated and β-structured proteins, which may subjects, and in this age range the amount of NM accumulated in SN was include SNCA, GPNMB, and other proteins, that bind oxidized DA sufficient for our determinations. Furthermore, we employed samples from or DA-derived compounds to produce adducts which undergo subjects in this age range since it was difficult to collect brain samples in a further oxidation and polymerization to form the melanic narrower age range. Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. Isolation of NM-containing organelles, NM purified from SN 300 SB-C18, 0.3 i.d. × 5 mm, 5 µm, 300 Å; Agilent Technologies, Santa Clara, CA, USA) for concentration and desalting prior to final separation by tissues, and NM purified from NM-containing organelles reversed phase C18 column (Biobasic-18, 0.180 i.d. × 100 mm, 5 µm, 300 Å; The isolation of intact ORG samples from SN tissue of single subjects, or Thermo Fisher Scientific, Inc.). Peptides were eluted using an acetonitrile rarely from SN of two subjects depending on the availability of brain gradient (eluent A, 0.1 % formic acid in water; eluent B, 0.1 % formic acid in tissues, was performed immediately after SN dissection, in order to avoid acetonitrile): the gradient profile was 5 % eluent B for 3 min, 5–40 % B in freezing and thawing steps of tissues that otherwise would have altered 50 min, 40–80 % B in 10 min, 80–95 % B in 5 min, 95 % B in 10 min. the integrity of ORG samples. The isolation procedure of ORG samples was The flow rate was 100 μl/min, which was split to achieve a final flux of 2 slightly modified and adapted for LC-MS analyses from our previously μl/min. Then, eluting peptides were electrosprayed directly into a hybrid published protocol. TM ion trap-Orbitrap mass spectrometer (LTQ Orbitrap XL ETD; Thermo TIS-NM samples were isolated from pooled SN tissues and prepared as 3,9,42 Fisher Scientific, Inc.), equipped with a nanospray ion source. The spray previously reported. capillary voltage was set at 1.5 kV, and the ion transfer capillary ORG-NM samples, consisting of the NM pigment fraction isolated from temperature was maintained at 220 °C. For each step of peptide elution NM-containing organelles, were prepared from ORG samples and from C18 column, full mass spectra were recorded in the positive ion mode described as follows. Intact ORG samples suspended in their isolation 3 over a 400–1600 m/z range, with a resolving power of 60,000 (full width at buffer were processed with three freeze/thaw cycles (−78 °C/+37 °C) in half-maximum) and a scan rate of 2 spectra/s. This step was followed by order to disrupt organelles membranes. The sample was then centrifuged four low-resolution MS/MS events that were sequentially generated in a (17,500 × g, 15 min, 4 °C) to obtain a pellet containing the NM pigment. data-dependent manner on the top four ions selected from the full MS This pellet was subsequently washed twice in the same buffer, centrifuged spectrum, using dynamic exclusion for the MS/MS analysis. In particular, as described above, and then treated for LC-MS analyses of proteins. the MS/MS scans were acquired by setting a normalized collision energy of 35 % on the precursor ion and, when a peptide ion was analyzed twice, Samples preparation for liquid chromatography-mass applying an exclusion duration of 0.5 min. Mass spectrometer scan spectrometry analyses of proteins functions and high performance liquid chromatography solvent gradients were controlled by the Xcalibur data system version 1.4 (Thermo Fisher Samples analyzed by LC-MS for proteomic analyses were isolated from the Scientific, Inc.). following subjects: (i) two ORG samples isolated from two different subjects (respectively 86 and 69 y.o.); (ii) two TIS-NM samples isolated from pooled SN tissues, the first one from a pool of five subjects (from 72 to 86 Identification of proteins detected by liquid chromatography-mass years of age) and the second one from pooled twelve subjects (from 70 to spectrometry analyses 92 years of age); and (iii) two ORG-NM samples, the first one isolated from Protein identification was carried out by matching experimental spectra to one subject (82 y.o.) and the second one from two pooled subjects (60 and peptide sequences using the SEQUEST database search algorithm, 69 y.o.). contained in BioWorks version 3.3.1 SP1 (University of Washington, TIS-NM samples (dried NM pigments) were hydrated in bi-distilled water licensed to Thermo Fisher Scientific, Inc.) using SEQUEST PC Cluster. under gentle shaking for 3 days in order to obtain a homogeneous For peptide matching, an updated non-redundant human protein suspension in water for tryptic digestion. The resulting TIS-NM suspension sequence database of 276,790 entries, downloaded in January 2009 from was then concentrated (Concentrator 5301; Eppendorf, Hamburg, Ger- the National Center for Biotechnology Information (NCBI) website (http:// many). ORG samples were concentrated in their isolation buffer while ORG- www.ncbi.nlm.nih.gov), was used. In addition, to make a thorough NM was directly treated with RapiGest SF (Waters, Milford, MA, USA) as 23,24,38 comparison of our data with data obtained from other studies and detailed below. to avoid possible bias and false results due to the use of not-aligned In order to dissolve membrane proteins and break NM pigment in all protein codes, the GI accession numbers of identified proteins were samples, a solution of surfactant RapiGest in 100 mM (pH 7.9) ammonium updated to those downloaded in April 2017 from the UniProt repository bicarbonate was added to ORG, TIS-NM and ORG-NM samples according to (http://www.uniprot.org/). Therefore, for these comparisons, we have the manufacturer’s protocol. A final concentration of 0.25 % (v/v) of considered only one time the few proteins we identified in our samples RapiGest was adjusted with 100 mM ammonium bicarbonate (pH 7.9). by two different GI accession numbers but with same UniProt accession Mixtures were then heated at 100 °C for 5 min, cooled to room number. temperature and digested overnight at 37 °C by adding sequencing grade Known abundant contaminating proteins such as keratins, trypsin and modified trypsin (Promega, Madison, WI, USA) at an enzyme/substrate ratio typical abundant proteins of red blood cells (rarely observed as probable of 1:50 (w/w). An additional aliquot of 0.5 µg trypsin was added in the contaminants in ORG samples) were removed prior to the final data morning and digestion was then prolonged for 4 h. The addition of 0.5 % 47,48 analysis, referring to published proteomic data sets of red blood cells. trifluoroacetic acid stopped the enzymatic reaction and subsequent HLA peptides were identified using the updated non-redundant human incubation at 37 °C for 45 min completed the acidic hydrolysis of RapiGest. HLA isoform (HLA gene) database downloaded from NCBI. The water insoluble degradation products were removed by centrifuging This analysis enabled the identification of peptide sequences and at 13,000 × g for 10 min and supernatants containing digested proteins related proteins. Since the confidence of protein identification depends on were desalted using PepClean C-18 spin columns (Pierce Biotechnology, the stringency of the identification of the peptide sequence and peptide Inc., Rockford, IL, USA), concentrated and finally suspended in 20 µl of 0.1 matching, particularly when using data from a single peptide, a high % (v/v) formic acid. stringency was guaranteed by using the following method. The peptide mass search tolerance was set to 1.00 Da; the precursor ion tolerance was Liquid chromatography-mass spectrometry analyses of proteins set to 50 ppm and the intensity threshold was set to 100. Moreover, all Three different types of samples (ORG, TIS-NM, ORG-NM) were prepared searches were performed with no enzyme and in order to assign a final from SN tissues (Supplementary Fig. 1): two samples for ORG, two samples score to the proteins, the SEQUEST output data were filtered by setting the −3 for TIS-NM, and two samples for ORG-NM as described in the previous peptide probability to 1 × 10 , the minimum correlation score values paragraph. Each of the two ORG samples was injected three times for LC- (Xcorr) was chosen greater than 1.5, 2.0, 2.5, and 3.0 for single-, double-, MS analysis, each of the two TIS-NM samples was injected three times, triple-, and quadruple-charged ions respectively, and the consensus score while one ORG-NM sample was injected two times and the other one was higher than 10. False-positive peptides ratio, calculated through reverse injected only one time due to its scarce amount. database, was less than 5 %. For decoy searches a reversed version of the The multidimensional protein identification technology (MudPIT) target human protein database was generated using the reverse database 46,136 system is a 2DC-MS/MS platform, composed of a two dimensional function in Bioworks 3.3.1 software (Thermo Fisher Scientific, Inc.). The micro-high performance liquid chromatography system (Surveyor HPLC; number of peptides, notably the SpC, detected by MS/MS was utilized as Thermo Fisher Scientific, Inc., San Jose, CA, USA) coupled online to a mass indicator of relative protein abundance. The group of representative spectrometer, using ProteomeX-2 configuration (Thermo Fisher Scientific, proteins was selected considering proteins detected by SpC ≥ 2 as average Inc.). Briefly, digested peptide mixtures were loaded onto a capillary strong value in at least one of the three types of samples. cation exchange column (Biobasic-SCX column, 0.32 i.d. × 100 mm, 5 µm; The area-proportional Euler diagrams were calculated using the Thermo Fisher Scientific, Inc.) and eluted stepwise with ammonium eulerAPE drawing tool (software at http://www.eulerdiagrams.org/ chloride injections of increasing molarity (5, 10, 15, 20, 30, 40, 80, 120, 400, eulerAPE/). Analyses of proteins for Euler diagram calculation in ORG, 600, 700 mM). Fractions were captured in turn onto peptide traps (Zorbax TIS-NM, and ORG-NM, as shown in Fig. 3, were performed by using NCBI npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. accession (GI number). Euler diagram calculation in the comparison of our sample from pooled four subjects (from 62 to 86 years of age); (ii) three list of proteins with those previously reported for NM-containing ORG samples, each of which isolated from one subject (respectively 62, 61, 23,24 38 organelles of human SN and for human brain lysosomes were and 77 y.o.). performed by using UniProt accession number. Total lipids extracts for TLC analyses were prepared from TIS-NM 3,9,42 Individual cellular location was assigned to each protein according to samples as previously described by using methanol and hexane. the GOA database (http://geneontology.org/), the UniProt database (http:// Additionally, these TIS-NM lipids extracts dried under nitrogen stream were www.uniprot.org/) and authors’ PubMed search. It should be noted that then resuspended in chloroform-methanol-water (2:1:0.1, v/v/v) and subjected to a two-phase Folch’s partitioning, resulting in the separation some proteins may have multiple cellular location: in these cases the most of an aqueous phase containing gangliosides and an organic phase typical and representative cellular location was manually assigned. We containing all other lipids, including glycerophospholipids, neutral glyco- were unable to assign the cellular location for some proteins which were sphingolipids and sphingomyelin. therefore classified as “unknown cell location” and additional proteins The ORG samples in few µl of isolation buffer were snap frozen and then without a complete characterization at the moment of data analysis were lyophilized. Lipids were extracted with chloroform-methanol-water (2:1:0.1, classified as “uncharacterized proteins”. v/v/v). Then, total lipid extracts were subjected to a two-phase Folch’s The names of genes and proteins described in the main text and supplementary files were retrieved from updated UniProt and UniParc partitioning. Following the two-phase partitioning, organic phases deriving from total database (April 2017), and we adopt as the protein name symbol that of the approved gene symbol. lipids extracted from both TIS-NM and ORG samples were analyzed by TLC, loading equivalent amounts of lipids. Lipids were separated by mono- dimensional high performance TLC silica gel using chloroform-methanol- Samples preparation for liquid chromatography-mass 0.2 % aqueous calcium chloride (60:35:8, v/v/v) as a solvent system. After spectrometry analysis of lipids separation, lipids were detected by spraying the TLC plates with Samples analyzed by LC-MS for identification of lipids were isolated from anisaldehyde, a reagent for the general detection of lipids. Identification the following subjects: (i) two TIS-NM samples isolated from pooled SN of lipids after separation and chemical detection was assessed by co- tissues, the first one isolated from a pool of seven subjects (from 71 to 85 migration with lipid standards (Avanti Polar Lipids, Inc., Alabaster, AL, USA; years of age) and the second one from pooled four subjects (from 82 to 85 Sigma-Aldrich Co., St. Louis, MO, USA; some standard were synthesized in years of age); (ii) three ORG samples, two of which isolated from two laboratories of the Department of Medical Biotechnology and Translational different subjects (respectively 81 and 89 y.o.) and the third one from two Medicine, University of Milan, Segrate, Italy). pooled subjects (74 and 89 y.o.). Cholera toxin staining of lipid present in the aqueous phases was The extraction of total lipids from TIS-NM samples was performed by performed as described hereafter. Briefly, after chromatographic running 3,9,42 using methanol and hexane as previously reported. The organic using chloroform-methanol-0.2 % aqueous calcium chloride (50:42:11, v/v/ fractions (methanol and then hexane) derived from NM pigments were v) as a solvent system, the TLC plate was well dried and fixed with a combined and dried under nitrogen flow. polyisobutylmethacrylate solution prepared dissolving 1.3 g of polyisobu- The total lipids were extracted from ORG samples with methanol and tylmethacrylate in 10 ml chloroform and diluting 8 ml of this solution with hexane, as for TIS-NM samples. The ORG samples in few µl of isolation 42 ml of hexane. The TLC plate was immersed three times in this solution buffer were resuspended in about 0.5 ml of methanol and then centrifuged and then allowed to dry for 1 h. The dried TLC was soaked for 30 min in (1000 × g, 30 min, 20 °C). The supernatant was collected and the remaining 0.1 M Tris-hydrochloride (pH 8.0), 0.14 M sodium chloride containing 1 % pellet was resuspended in about 0.5 ml of hexane, and subsequently bovine serum albumin. The TLC was then incubated with Clostridium centrifuged as described above. Again, organic fractions (methanol and perfringens sialidase (0.12 U/ml in 0.05 M acetate buffer pH 5.4 and 4 mM then hexane) extracted from ORG samples were finally combined and calcium chloride) overnight at room temperature. Afterward, the TLC was dried under nitrogen flow. incubated with biotin-conjugated cholera toxin subunit B (10 μg/ml; Sigma-Aldrich Co.) in phosphate-buffered saline containing 1 % bovine serum albumin for 1 h, and subsequently for 1 h with horseradish Liquid chromatography-mass spectrometry analysis of lipids peroxidase-conjugated streptavidin (2 μg/ml; Sigma-Aldrich Co.) in the For LC-MS analysis, the dried lipids extracted from TIS-NM and ORG same solution. After several washings with phosphate-buffered saline, the samples were dissolved in 100 µl chloroform and diluted 1:1 (v/v) in eluent TLC plate was developed for 5 min with o-Phenylenediamine dihydrochlor- A, as described below. Typically, 5 µl of sample were injected on a C8 ide substrate (Sigma-Aldrich Co.), 1 tablet in 50 ml citrate-phosphate buffer reversed phase column (BioBasic C8, 100 × 0.18 mm, 5 µm, 300 Å; Thermo (pH 5.0) and 20 μl hydrogen peroxide. Fisher Scientific, Inc.), using a micro liquid chromatography system (Surveyor HPLC; Thermo Fisher Scientific, Inc.) coupled to a mass TM Electron microscopy spectrometer with Orbitrap mass analyzer (Exactive Plus; Thermo Fisher Transmission electron microscopy for morphological evaluation was Scientific, Inc.), equipped with a nanospray ion source. Liquid chromato- performed on SN tissue slices and isolated ORG samples prepared graphy was operated at a flow rate of 100 μl/min, split in order to achieve a according to our previously published method. final flux of 2 μl/min, with a gradient as follows: 100 % eluent A was held IEM experiments were carried out on SN tissue blocks which were fixed isocratically for 10 min, then linearly increased to 100 % B over 30 min and for 2 h at 4 °C in a mixture of 4 % paraformaldehyde and 0.25 % held at 100 % B for 12 min. Eluent A consisted of methanol-acetonitrile- glutaraldehyde in cacodylate buffer (0.12 M, pH 7.4). Tissue samples were aqueous 1 mM ammonium acetate (60:20:20, v/v/v). Eluent B consisted of extensively washed with cacodylate buffer, dehydrated in a graded isopropanol-acetonitrile (90:10, v/v). ethanol series and then embedded in LRW resin. Ultra thin sections Nanospray was achieved using a coated fused silica emitter (360 µm o. (80 nm) were prepared using a ultramicrotome (Leica Ultracut; Leica d./50 µm i.d. 730 µm tip i.d.; New Objective, Inc., Woburn, MA, USA) held to Microsystems GmbH, Wien, Austria), collected over nickel grids and 1.5 kV. The heated capillary was held at 220 °C. Full mass spectra were incubated for 90 min at room temperature with primary antibodies diluted recorded in negative and positive ion mode over a 400–2000 m/z range. in phosphate-buffered saline (pH 7.4). The concentrations of primary The resolution was set to 100,000 with 2 microscans/s, restricting the antibodies used for IEM experiments were the following: APOD (1:50), Orbitrap loading time to a maximum of 50 ms with a target value of 5E5 ATP5G1 (1:100), ATP6V1B2 (1:100), CTSD (1:100), DHDDS (1:200), FTH1 ions (ultimate mass accuracy mode). (1:250), FTL (1:300), GPNMB (1:100), HLA (1:200), LAMP1 (1:20), LAMP2 Manual interpretation of mass spectra was performed to evaluate the (1:20), MAP1LC3B (1:50 for antibody from Cell Signaling Technology; 1:25 major lipid species present in the analyzed lipid extracts. for antibody from Abgent Inc.), PLBD2 (1:100), RAB5A (1:100), SCARB2 (1:50), SNCA (1:200), SQSTM1 (1:30), SRD5A3 (1:25), and UBA52 (1:250). The Lipids extraction and partitioning for thin-layer chromatography grids were then washed with phosphate-buffered saline and incubated for analyses of lipids 60 min with gold-conjugated secondary antibodies (British Biocell Inter- Samples analyzed by TLC for identification and semiquantitative assess- national, Cardiff, UK) diluted 1:100 in phosphate-buffered saline. After ment of lipids were isolated from the following subjects: (i) four TIS-NM extensive washing, the grids were post fixed with 1 % glutaraldehyde in samples isolated from pooled SN tissues, the first three samples isolated cacodylate buffer. Samples were then stained with saturated uranyl acetate from a pool of five subjects each (respectively from 48 to 85 years of age, in water for 5 min, washed and then stained with 3 mM lead citrate for from 67 to 85 years of age, from 73 to 85 years of age) and the fourth 5 min. Finally, the sections were photographed using a transmission Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. electron microscope LEO 912AB (Carl Zeiss AG, Oberkochen, Germany) was confirmed by re-incubating the membranes with the chemilumines- equipped with a Proscan CCD camera (ProScan, Lagerlechfeld, Germany) cent substrate. controlled by EsivisionPro software (Soft Imaging System, Münster, The WB analyses in this study are qualitative rather than quantitative. Germany). Electron microscopy was performed at the Advanced Light For the electrophoresis separation of proteins in the samples, we kept an and Electron Microscopy BioImaging Center—San Raffaele Scientific empty lane between lanes containing SN tissue lysates and those Institute. containing ORG samples in order to avoid contamination leading to false Each protein was investigated by IEM in at least one to three samples positive results. For the preparation of figures, we divided the two lanes of deriving from different subjects, depending on SN tissue availability. Data interest by a thin white line between juxtaposed lanes. It should be noted of subjects analyzed by IEM are reported in the legends of figures chosen that: (i) ORG samples were loaded at high volumes, due to their low total as representative of all electron microscopy experiments. Each IEM protein concentration with respect to SN tissue lysates, and because the experiment was routinely checked for negative controls as described concentration procedure of ORG sample produced an insoluble and below: (i) the reaction was performed on SN tissue slices without either untreatable residue from which proteins could be no longer recovered primary or secondary antibody to confirm the absence of gold particles in (high salt composition of the isolation buffer, particular nature of the NM the examined sections; (ii) in the presence of primary and secondary pigment contained in ORG samples, etc.); (ii) the granular composition of antibodies, the absence of gold particles was verified in the resin devoid of the NM pigment present in ORG samples could interfere with the initial tissue; and (iii) the most relevant control we did was to check the absence electrophoresis run in the stacking gel, thus leading to lateral diffusion of of signal in cellular/tissue components where it was not expected. Only proteins; and (iii) the contamination of proteins from SN tissue lysate after passing these three negative controls, we evaluated the signal in NM- (largely rich of proteins) to ORG sample was less probable but could not be containing organelles which we judged as specific even if represented by excluded and lead as well to false positive results in ORG samples. The total few gold particles. amounts of proteins loaded into each lane of the gel were then chosen depending on the availability of the sample (particularly ORG sample), and Western blotting analyses of tissue lysates and NM-containing on the content of each selected protein in both SN tissue lysates and ORG organelles samples. The ratios of protein content between SN tissue lysates and ORG SN tissues were homogenized and sonicated with five volumes of lysis samples for each protein analyzed by WB are reported in the figure buffer as described above. In case of soluble proteins, the lysis buffer was legends. Note that for ATP6V1B2, GPNMB, and RAB5A proteins, the WB composed of 50 mM Tris-hydrochloride (pH 7.5), 150 mM sodium chloride, analyses were performed on SN tissue lysates and ORG samples on 0.1 % sodium dodecyl sulfate, 1 % Triton® X-100, and 1 % protease different days, due to the lack of either SN tissues or ORG samples, inhibitor cocktail (Sigma-Aldrich Co.). For membrane-bound proteins, the depending on human brain tissues availability, and it was sometimes lysis buffer was composed of phosphate-buffered saline (pH 7.4), 0.1 % impossible to run all types of samples at the same time. Relative lanes sodium dodecyl sulfate, 1 % Triton® X-100, and 1 % protease inhibitor where then juxtaposed as previously explained. The use of WB on SN tissue cocktail. After short sonication cycles in ice, SN homogenates were lysates was chiefly employed to assess the specificity of each antibody for centrifuged at 17,500 × g, 4 °C (30 min for soluble proteins, 10 min for the target protein as essential information for IEM experiments, and to membrane-bound proteins) and the supernatants were collected. From verify its presence or absence in the ORG samples (n ≥ 3 for each protein). supernatants, proteins were precipitated by acid treatment to remove components of buffers that could influence protein quantitation. Then, Data of subjects from which SN tissue lysates and ORG samples were Lowry method was performed on the protein pellets to measure the total prepared are reported in the legends of figures chosen as representative of protein concentration in both SN tissue lysates and ORG samples. all WB experiments performed on different samples. For electrophoresis, samples were heated in reducing sample buffer and then loaded onto 0.75 mm thick polyacrylamide mini gels (different pore Data availability sizes, depending on the desired separation of proteins) and separated by one-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophor- All relevant data are within the paper and its Supporting Information files. esis (Mini-PROTEAN® 3 Cell; Bio-Rad Laboratories, Inc., Hercules, CA, USA). However, raw data supporting the results reported in this article are Prestained protein markers (New England BioLabs, Inc., Ipswich, MA, USA) available upon request. were used as size standards in the electrophoresis. Proteins were then over-night transferred (Mini Trans-Blot® Electrophoretic Transfer Cell; Bio- Rad Laboratories, Inc.) to 0.45 µm supported nitrocellulose membranes ACKNOWLEDGEMENTS (Hybond®-C Extra; Amersham Biosciences, Little Chalfont, UK). Ponceau-S This work was supported by Italian Ministry of Education, University, and Research staining of membranes was performed to check the efficiency of proteins (MIUR)—National Research Programme (PNR)—National Research Council of Italy transfer. (CNR) Flagship “InterOmics” Project (PB.P05), by MIUR—PNR—CNR Aging program For immunoblotting, membranes were blocked for 2 h at room 2012–2014, by MIUR—Medical Research in Italy (MERIT) Project RBNE08ZZN7, by temperature with fatty acid/globulin-free bovine serum albumin (Sigma- Lombardy Region—CNR 2013–2015 Framework Agreement MbMM Project (18089/ Aldrich Co.) or skimmed milk, with concentrations ranging from 3 to 8 %, RCC) and by MIUR—Research Projects of National Interest (PRIN) 2015 Prot. depending on the properties of investigated protein, antibody specificity, 2015T778JW. D.S. also thanks the National Parkinson’s, Parkinson’s Disease and the etc., in phosphate-buffered saline (pH 7.4) with 0.1 % Tween® 20. Then JPB Foundations (Miami, FL and New York, NY, USA) for support. D.S. is a NARSAD membranes were incubated with primary antibodies diluted in blocking solutions, while temperature and time of incubation varied according to Brain and Behavior Distinguished Investigator. The funders had no role in study antibodies specificity. The concentrations of primary antibodies used for design, data collection and analysis, decision to publish, or preparation of the WB experiments were the following: APOD (1:300), ATP5G1 (1:100), manuscript. The authors thank Dr. Maria Carla Panzeri (Advanced Light and Electron ATP6V1B2 (1:400), CTSD (1:500), DHDDS (1:1000), FTH1 (1:3500), FTL Microscopy BioImaging Center—San Raffaele Scientific Institute) for expert and (1:2000), GPNMB (1:1500), LAMP1 (1:50), LAMP2 (1:200), MAP1LC3B (1:600), invaluable assistance in electron microscopy imaging. Authors gratefully acknowl- PLBD2 (1:300), RAB5A (1:1000), SCARB2 (1:600), SNCA (1:1500), SQSTM1 edge Dr. Simona Prioni (Department of Medical Biotechnology and Translational (1:300), SRD5A3 (1:100), and UBA52 (1:200). After washing, membranes Medicine, University of Milan, Segrate, Italy) for her skillful assistance in TCL analyses were incubated for 90 min at room temperature with horseradish of lipids. Authors are also thankful to Dr. Luciano Milanesi and Dr. Pasqualina D’Ursi peroxidase-labeled secondary antibodies with a concentration range for molecular docking of PLBD2 protein, to Dr. Eleonora Franceschi for her excellent 1:2000–1:5000 in block solution (Jackson ImmunoResearch, Inc., West technical support in LC-MS analysis of lipids, and to Dr. Alice Valmadre for her Grove, PA, USA; EMD Millipore). After extensive washing, the immunor- assistance during some WB analyses; additionally, authors thank Ms. Loredana eactive species were visualized by SuperSignal® West Pico chemilumines- Ansalone and Ms. Mariagiovanna Torti for their assistance in the preparation of cent substrate as described in the manufacturer’s protocol (Pierce financial reports of the above mentioned grants (all these collaborators are affiliated Biotechnology, Inc.). The films were exposed for few seconds up to with Institute of Biomedical Technologies—National Research Council of Italy). L.Z. 15 min. As a control for aspecific binding, the same procedure was also thanks the Grigioni Foundation for Parkinson’s Disease (Milan, Italy). Finally, all executed without primary antibody. When multiple blotting of membranes the authors gratefully thank the Section of Legal Medicine and Insurances, was necessary, as in case of loading control with CTSD antibody, the TM Restore PLUS Western Blot Stripping Buffer (Pierce Biotechnology, Inc.) Department of Biomedical Sciences for Health at University of Milan for providing was used according to manufacturer’s protocol, and successful stripping brain tissue samples. npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. AUTHOR CONTRIBUTIONS 21. Oberländer, U. et al. Neuromelanin is an immune stimulator for dendritic cells in vitro. BMC Neurosci. 12, 116 (2011). Conceived and designed the experiments: F.A.Z., R.V., P.M., L.Z. Performed the 22. Double, K. L. et al. Structural characteristics of human substantia nigra neuro- experiments: F.A.Z., R.V., F.A.C., C.B., A.D.P., S.G. Analyzed the data: F.A.Z., R.V., A.D.P., melanin and synthetic dopamine melanins. J. Neurochem. 75,2583–2589 D.D.S., P.M., A.P. Contributed reagents/materials/analysis tools: P.M., A.P., L.Z. Wrote (2000). the paper: F.A.Z., R.V., A.P., L.C., D.S., L.Z. 23. Tribl, F. et al. “Subcellular proteomics” of neuromelanin granules isolated from the human brain. Mol. Cell. Proteom. 4, 945–957 (2005). 24. Plum, S. et al. Proteomic characterization of neuromelanin granules isolated ADDITIONAL INFORMATION from human substantia nigra by laser-microdissection. Sci. Rep. 6, 37139 (2016). Supplementary information accompanies the paper on the npj Parkinson’s Disease 25. Han, X. Multi-dimensional mass spectrometry-based shotgun lipidomics and the website (https://doi.org/10.1038/s41531-018-0050-8). altered lipids at the mild cognitive impairment stage of Alzheimer’s disease. Biochim. Biophys. Acta 1801, 774–783 (2010). Competing interests: The authors declare no competing interests. 26. Licker, V. et al. Proteomic analysis of human substantia nigra identifies novel candidates involved in Parkinson’s disease pathogenesis. Proteomics 14, Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims 784–794 (2014). in published maps and institutional affiliations. 27. Plum, S. et al. Proteomics in neurodegenerative diseases: Methods for obtaining a closer look at the neuronal proteome. Proteom. Clin. Appl. 9, 848–871 (2015). 28. Zecca, L. et al. Interaction of human substantia nigra neuromelanin with lipids REFERENCES and peptides. J. Neurochem. 74, 1758–1765 (2000). 29. Engelen, M. et al. Proteomic profile of neuromelanin and neuromelanin- 1. Duffy, P. & Tennyson, V. M. Phase and electron microscopic observations of containing organelles. Abstract at “5th Congress of the Portuguese Proteomics Lewy bodies and melanin granules in the substantia nigra and locus coeruleus Network—ProCura 1st International Congress on Analytical Proteomics—ICAP”, in Parkinson’s disease. J. Neuropathol. Exp. Neurol. 24, 398–414 (1965). Sep 30th–Oct 3rd, (Caparica, Portugal, 2009). 2. Sulzer, D. et al. Neuronal pigmented autophagic vacuoles: lipofuscin, neuro- 30. Lindersson, E. et al. p25alpha Stimulates alpha-synuclein aggregation and is co- melanin, and ceroid as macroautophagic responses during aging and disease. J. localized with aggregated alpha-synuclein in alpha-synucleinopathies. J. Biol. Neurochem. 106,24–36 (2008). Chem. 280, 5703–5715 (2005). 3. Zecca, L. et al. New melanic pigments in the human brain that accumulate in 31. Rekas, A. et al. Interaction of the molecular chaperone alphaB-crystallin with aging and block environmental toxic metals. Proc. Natl Acad. Sci. USA 105, alpha-synuclein: effects on amyloid fibril formation and chaperone activity. J. 17567–17572 (2008). Mol. Biol. 340, 1167–1183 (2004). 4. Fedorow, H. et al. Evidence for specific phases in the development of human 32. Letournel, F., Bocquet, A., Dubas, F., Barthelaix, A. & Eyer, J. Stable tubule only neuromelanin. Neurobiol. Aging 27, 506–512 (2006). polypeptides (STOP) proteins co-aggregate with spheroid neurofilaments in 5. Zecca, L. et al. The role of iron and copper molecules in the neuronal vulner- amyotrophic lateral sclerosis. J. Neuropathol. Exp. Neurol. 62, 1211–1219 (2003). ability of locus coeruleus and substantia nigra during aging. Proc. Natl Acad. Sci. 33. Morris, M., Maeda, S., Vossel, K. & Mucke, L. The many faces of tau. Neuron 70, USA 101, 9843–9848 (2004). 410–426 (2011). 6. Zucca, F. A. et al. Neuromelanin and iron in human locus coeruleus and sub- 34. Halliday, G. M. et al. Alpha-synuclein redistributes to neuromelanin lipid in the stantia nigra during aging: consequences for neuronal vulnerability. J. Neural substantia nigra early in Parkinson’s disease. Brain 128, 2654–2664 (2005). Transm. (Vienna) 113, 757–767 (2006). 35. Rochet, J. C. et al. Interactions among alpha-synuclein, dopamine, and bio- 7. German, D. C. et al. Disease-specific patterns of locus coeruleus cell loss. Ann. membranes: some clues for understanding neurodegeneration in Parkinson’s Neurol. 32, 667–676 (1992). disease. J. Mol. Neurosci. 23,23–34 (2004). 8. Hirsch, E., Graybiel, A. M. & Agid, Y. A. Melanized dopaminergic neurons are 36. Jalanko, A. & Braulke, T. Neuronal ceroid lipofuscinoses. Biochim. Biophys. Acta differentially susceptible to degeneration in Parkinson’s disease. Nature 334, 1793, 697–709 (2009). 345–348 (1988). 37. Jensen, A. G. et al. Biochemical characterization and lysosomal localization of 9. Engelen, M. et al. Neuromelanins of human brain have soluble and insoluble the mannose-6-phosphate proteinp76 (hypothetical protein LOC196463). Bio- components with dolichols attached to the melanic structure. PLoS ONE 7, chem. J. 402, 449–458 (2007). e48490 (2012). 38. Sleat, D. E., Zheng, H., Qian, M. & Lobel, P. Identification of sites of mannose 6- 10. Sulzer, D. et al. Neuromelanin biosynthesis is driven by excess cytosolic cate- phosphorylation on lysosomal proteins. Mol. Cell. Proteom. 5, 686–701 (2006). cholamines not accumulated by synaptic vesicles. Proc. Natl Acad. Sci. USA 97, 39. Hemmings, H. C. Jr, Greengard, P., Tung, H. Y. & Cohen, P. DARPP-32, a 11869–11874 (2000). dopamine-regulated neuronal phosphoprotein, is a potent inhibitor of protein 11. Kastner, A. et al. Is the vulnerability of neurons in the substantia nigra of patients phosphatase-1. Nature 310, 503–505 (1984). with Parkinson’s disease related to their neuromelanin content? J. Neurochem. 40. Deinhardt, K. et al. Rab5 and Rab7 control endocytic sorting along the axonal 59, 1080–1089 (1992). retrograde transport pathway. Neuron 52, 293–305 (2006). 12. Liang, C. L., Nelson, O., Yazdani, U., Pasbakhsh, P. & German, D. C. Inverse 41. Boggs, J. M. Role of galactosylceramide and sulfatide in oligodendrocytes and relationship between the contents of neuromelanin pigment and the vesicular CNS myelin: formation of a glycosynapse. Adv. Neurobiol. 9, 263–291 (2014). monoamine transporter-2: human midbrain dopamine neurons. J. Comp. Neurol. 42. Ward, W. C. et al. Identification and quantification of dolichol and dolichoic acid 473,97–106 (2004). in neuromelanin from substantia nigra of the human brain. J. Lipid Res. 48, 13. Zucca, F. A. et al. Neuromelanin of the human substantia nigra: an update. 1457–1462 (2007). Neurotox. Res. 25,13–23 (2014). 43. Fedorow, H. et al. Dolichol is the major lipid component of human substantia 14. Zucca, F. A. et al. Interactions of iron, dopamine and neuromelanin pathways in nigra neuromelanin. J. Neurochem. 92, 990–995 (2005). brain aging and Parkinson’s disease. Prog. Neurobiol. 155,96–119 (2017). 44. Stingl, C., Söderquist, M., Karlsson, O., Borén, M. & Luider, T. M. Uncovering 15. Bohic, S. et al. Intracellular chemical imaging of the developmental phases of effects of ex vivo protease activity during proteomics and peptidomics sample human neuromelanin using synchrotron X-ray microspectroscopy. Anal. Chem. extraction in rat brain tissue by oxygen-18 labeling. J. Proteome Res. 13, 80, 9557–9566 (2008). 2807–2817 (2014). 16. Karlsson, O., Berg, C., Brittebo, E. B. & Lindquist, N. G. Retention of the cyano- 45. Plum, S. et al. Combined enrichment of neuromelanin granules and synapto- bacterial neurotoxin beta-N-methylamino-l-alanine in melanin and somes from human substantia nigra pars compacta tissue for proteomic ana- neuromelanin-containing cells--a possible link between Parkinson-dementia lysis. J. Proteom. 94, 202–206 (2013). complex and pigmentary retinopathy. Pigment Cell Melanoma Res. 22, 120–130 46. Di Silvestre, D., Brambilla, F. & Mauri, P. L. Multidimensional protein identification (2009). technology for direct-tissue proteomics of heart. Methods Mol. Biol. 1005,25–38 17. Karlsson, O. & Lindquist, N. G. Melanin affinity and its possible role in neuro- (2013). degeneration. J. Neural Transm. (Vienna) 120, 1623–1630 (2013). 47. Pasini, E. M. et al. In-depth analysis of the membrane and cytosolic proteome of 18. Karlsson, O. & Lindquist, N. G. Melanin and neuromelanin binding of drugs and red blood cells. Blood 108, 791–801 (2006). chemicals: toxicological implications. Arch. Toxicol. 90, 1883–1891 (2016). 48. Roux-Dalvai, F. et al. Extensive analysis of the cytoplasmic proteome of human 19. Zhang, W. et al. Neuromelanin activates microglia and induces degeneration of erythrocytes using the peptide ligand library technology and advanced mass dopaminergic neurons: implications for progression of Parkinson’s disease. spectrometry. Mol. Cell. Proteom. 7, 2254–2269 (2008). Neurotox. Res. 19,63–72 (2011). 49. Lübke, T., Lobel, P. & Sleat, D. E. Proteomics of the lysosome. Biochim. Biophys. 20. Cebrián, C. et al. MHC-I expression renders catecholaminergic neurons suscep- Acta 1793, 625–635 (2009). tible to T-cell-mediated degeneration. Nat. Commun. 5, 3633 (2014). Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17 Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. 50. Schröder, B. A., Wrocklage, C., Hasilik, A. & Saftig, P. The proteome of lysosomes. 79. Pullarkat, R. K. & Reha, H. Accumulation of dolichols in brains of elderly. J. Biol. Proteomics 10, 4053–4076 (2010). Chem. 257, 5991–5993 (1982). 51. Koike, M. et al. Cathepsin D deficiency induces lysosomal storage with ceroid 80. Hall, N. A. & Patrick, A. D. Dolichol and phosphorylated dolichol content of lipofuscin in mouse CNS neurons. J. Neurosci. 20, 6898–6906 (2000). tissues in ceroid-lipofuscinosis. J. Inherit. Metab. Dis. 8, 178–183 (1985). 52. Qiao, L. et al. Lysosomal enzyme cathepsin D protects against alpha-synuclein 81. Ng Ying Kin, N. M., Palo, J., Haltia, M. & Wolfe, L. S. High levels of brain dolichols aggregation and toxicity. Mol. Brain 1, 17 (2008). in neuronal ceroid-lipofuscinosis and senescence. J. Neurochem. 40, 1465–1473 53. Morgan, C. P., Insall, R., Haynes, L. & Cockcroft, S. Identification of phospholipase (1983). B from Dictyostelium discoideum reveals a new lipase family present in mam- 82. Elleder, M., Sokolová, J. & Hrebícek, M. Follow-up study of subunit c of mito- mals, flies and nematodes, but not yeast. Biochem. J. 382, 441–449 (2004). chondrial ATP synthase (SCMAS) in Batten disease and in unrelated lysosomal 54. Tollbom, O., Chojnacki, T. & Dallner, G. Hydrolysis of dolichyl esters by rat liver disorders. Acta Neuropathol. 93, 379–390 (1997). lysosomes. J. Biol. Chem. 264, 9836–9841 (1989). 83. Wei, P., Smeyne, R. J., Bao, D., Parris, J. & Morgan, J. I. Mapping of Cbln1-like 55. Munck, A., Böhm, C., Seibel, N. M., Hashemol Hosseini, Z. & Hampe, W. Hu-K4 is a immunoreactivity in adult and developing mouse brain and its localization to ubiquitously expressed type 2 transmembrane protein associated with the the endolysosomal compartment of neurons. Eur. J. Neurosci. 26, 2962–2978 endoplasmic reticulum. FEBS J. 272, 1718–1726 (2005). (2007). 56. Pedersen, K. M., Finsen, B., Celis, J. E. & Jensen, N. A. Expression of a novel 84. Chung, C. Y. et al. Cell type-specific gene expression of midbrain dopaminergic murine phospholipase D homolog coincides with late neuronal development in neurons reveals molecules involved in their vulnerability and protection. Hum. the forebrain. J. Biol. Chem. 273, 31494–31504 (1998). Mol. Genet. 14, 1709–1725 (2005). 57. Nagaoka-Yasuda, R., Matsuo, N., Perkins, B., Limbaeck-Stokin, K. & Mayford, M. 85. Villani, G. R. et al. Cytokines, neurotrophins, and oxidative stress in brain disease An RNAi-based genetic screen for oxidative stress resistance reveals retinol from mucopolysaccharidosis IIIB. J. Neurosci. Res. 85, 612–622 (2007). saturase as a mediator of stress resistance. Free Radic. Biol. Med. 43, 781–788 86. Bellinger, F. P. et al. Glutathione peroxidase 4 is associated with neuromelanin in (2007). substantia nigra and dystrophic axons in putamen of Parkinson’s brain. Mol. 58. Satoh, J. et al. PLD3 is accumulated on neuritic plaques in Alzheimer’s disease Neurodegener. 6, 8 (2011). brains. Alzheimers Res. Ther. 6, 70 (2014). 87. Andres-Mateos, E. et al. DJ-1 gene deletion reveals that DJ-1 is an atypical 59. Kolter, T. & Sandhoff, K. Lysosomal degradation of membrane lipids. FEBS Lett. peroxiredoxin-like peroxidase. Proc. Natl Acad. Sci. USA 104, 14807–14812 584, 1700–1712 (2010). (2007). 60. Schulze, H., Kolter, T. & Sandhoff, K. Principles of lysosomal membrane degra- 88. Bonifati, V. et al. Mutations in the DJ-1 gene associated with autosomal recessive dation: cellular topology and biochemistry of lysosomal lipid degradation. Bio- early-onset parkinsonism. Science 299, 256–259 (2003). chim. Biophys. Acta 1793, 674–683 (2009). 89. Tribl, F. et al. Identification of L-ferritin in neuromelanin granules of the human 61. Reczek, D. et al. LIMP-2 is a receptor for lysosomal mannose-6-phosphate- substantia nigra: a targeted proteomics approach. Mol. Cell. Proteom. 8, independent targeting of beta-glucocerebrosidase. Cell 131, 770–783 (2007). 1832–1838 (2009). 62. Gao, D. et al. Structural basis for the recognition of oxidized phospholipids in 90. Sulzer, D. et al. T cells from patients with Parkinson’s disease recognize α- oxidized low density lipoproteins by class B scavenger receptors CD36 and SR- synuclein peptides. Nature 546, 656–661 (2017). BI. J. Biol. Chem. 285, 4447–4454 (2010). 91. Hara, T. et al. Suppression of basal autophagy in neural cells causes neurode- 63. Eckhardt, E. R. et al. High density lipoprotein endocytosis by scavenger receptor generative disease in mice. Nature 441, 885–889 (2006). SR-BII is clathrin-dependent and requires a carboxyl-terminal dileucine motif. J. 92. Mizushima, N., Ohsumi, Y. & Yoshimori, T. Autophagosome formation in mam- Biol. Chem. 281, 4348–4353 (2006). malian cells. Cell Struct. Funct. 27, 421–429 (2002). 64. Kuronita, T. et al. The NH(2)-terminal transmembrane and lumenal domains of 93. Mizushima, N., Levine, B., Cuervo, A. M. & Klionsky, D. J. Autophagy fights disease LGP85 are needed for the formation of enlarged endosomes/lysosomes. Traffic through cellular self-digestion. Nature 451, 1069–1075 (2008). 6, 895–906 (2005). 94. Pankiv, S. et al. p62/SQSTM1 binds directly to Atg8/LC3 to facilitate degradation 65. Cuervo, A. M. & Dice, J. F. A receptor for the selective uptake and degradation of of ubiquitinated protein aggregates by autophagy. J. Biol. Chem. 282, proteins by lysosomes. Science 273, 501–503 (1996). 24131–24145 (2007). 66. Cuervo, A. M. Autophagy: many paths to the same end. Mol. Cell. Biochem. 263, 95. Kuusisto, E., Salminen, A. & Alafuzoff, I. Ubiquitin-binding protein p62 is present 55–72 (2004). in neuronal and glial inclusions in human tauopathies and synucleinopathies. 67. Cuervo, A. M. & Dice, J. F. Age-related decline in chaperone-mediated autop- Neuroreport 12, 2085–2090 (2001). hagy. J. Biol. Chem. 275, 31505–31513 (2000). 96. Zatloukal, K. et al. p62 Is a common component of cytoplasmic inclusions in 68. Boellaard, J. W., Schlote, W. & Hofer, W. Species-specific ultrastructure of neu- protein aggregation diseases. Am. J. Pathol. 160, 255–263 (2002). ronal lipofuscin in hippocampus and neocortex of subhuman mammals and 97. Komatsu, M. et al. Homeostatic levels of p62 control cytoplasmic inclusion body humans. Ultrastruct. Pathol. 28, 341–351 (2004). formation in autophagy-deficient mice. Cell 131, 1149–1163 (2007). 69. Kabeya, Y. et al. LC3, a mammalian homologue of yeast Apg8p, is localized in 98. Szeto, J. et al. ALIS are stress-induced protein storage compartments for sub- autophagosome membranes after processing. EMBO J. 19, 5720–5728 (2000). strates of the proteasome and autophagy. Autophagy 2, 189–199 (2006). 70. Tanida, I., Minematsu-Ikeguchi, N., Ueno, T. & Kominami, E. Lysosomal turnover, 99. Picard, D. Heat-shock protein 90, a chaperone for folding and regulation. Cell. but not a cellular level, of endogenous LC3 is a marker for autophagy. Autop- Mol. Life Sci. 59, 1640–1648 (2002). hagy 1,84–91 (2005). 100. Chi, A. et al. Proteomic and bioinformatic characterization of the biogenesis and 71. Schröder, B., Elsässer, H. P., Schmidt, B. & Hasilik, A. Characterisation of function of melanosomes. J. Proteome Res. 5, 3135–3144 (2006). lipofuscin-like lysosomal inclusion bodies from human placenta. FEBS Lett. 581, 101. Horwitz, J. Alpha-crystallin can function as a molecular chaperone. Proc. Natl 102–108 (2007). Acad. Sci. USA 89, 10449–10453 (1992). 72. Muffat, J. & Walker, D. W. Apolipoprotein D: an overview of its role in aging and 102. Lowe, J. et al. alpha B crystallin expression in non-lenticular tissues and selective age-related diseases. Cell Cycle 9, 269–273 (2010). presence in ubiquitinated inclusion bodies in human disease. J. Pathol. 166, 73. Ganfornina, M. D. et al. Apolipoprotein D is involved in the mechanisms reg- 61–68 (1992). ulating protection from oxidative stress. Aging Cell 7, 506–515 (2008). 103. Liu, Y. et al. Upregulation of alphaB-crystallin expression in the substantia nigra 74. de Magalhães, J. P., Curado, J. & Church, G. M. Meta-analysis of age-related gene of patients with Parkinson’s disease. Neurobiol. Aging 36, 1686–1691 (2015). expression profiles identifies common signatures of aging. Bioinformatics 25, 104. Conn, K. J. et al. Identification of the protein disulfide isomerase family member 875–881 (2009). PDIp in experimental Parkinson’s disease and Lewy body pathology. Brain Res. 75. Ordoñez, C. et al. Apolipoprotein D expression in substantia nigra of Parkinson 1022, 164–172 (2004). disease. Histol. Histopathol. 21, 361–366 (2006). 105. Lehotzky, A. et al. Dynamic targeting of microtubules by TPPP/p25 affects cell 76. Cao, Y. et al. Autophagy is disrupted in a knock-in mouse model of juvenile survival. J. Cell Sci. 117, 6249–6259 (2004). neuronal ceroid lipofuscinosis. J. Biol. Chem. 281, 20483–20493 (2006). 106. Hlavanda, E. et al. Brain-specific p25 protein binds to tubulin and microtubules 77. Ryazantsev, S., Yu, W. H., Zhao, H. Z., Neufeld, E. F. & Ohmi, K. Lysosomal and induces aberrant microtubule assemblies at substoichiometric concentra- accumulation of SCMAS (subunit c of mitochondrial ATP synthase) in neurons of tions. Biochemistry 41, 8657–8664 (2002). the mouse model of mucopolysaccharidosis III B. Mol. Genet. Metab. 90, 393–401 107. Basrur, V. et al. Proteomic analysis of early melanosomes: identification of novel (2007). melanosomal proteins. J. Proteome Res. 2,69–79 (2003). 78. Andersson, M., Appelkvist, E. L., Kristensson, K. & Dallner, G. Distribution of 108. Hoashi, T. et al. Glycoprotein nonmetastatic melanoma protein b, a melanocytic dolichol and dolichyl phosphate in human brain. J. Neurochem. 49, 685–691 cell marker, is a melanosome-specific and proteolytically released protein. FASEB (1987). J. 24, 1616–1629 (2010). npj Parkinson’s Disease (2018) 17 Published in partnership with the Parkinson’s Foundation Human neuromelanin organelles are impaired autolysosomes FA Zucca et al. 109. Zhang, P. et al. Silencing of GPNMB by siRNA inhibits the formation of mela- 126. Ferrari, E et al. Synthesis, Structure characterization, and evaluation in microglia nosomes in melanocytes in a MITF-independent fashion. PLoS ONE 7, e42955 cultures of neuromelanin analogues suitable for modeling Parkinson’s disease. (2012). ACS Chem. Neurosci. 8 501–512 (2017). 110. Li, B. et al. The melanoma-associated transmembrane glycoprotein Gpnmb 127. Ikemoto, K. et al. Does tyrosinase exist in neuromelanin-pigmented neurons in controls trafficking of cellular debris for degradation and is essential for tissue the human substantia nigra? Neurosci. Lett. 253, 198–200 (1998). repair. FASEB J. 24, 4767–4781 (2010). 128. Tribl, F., Arzberger, T., Riederer, P. & Gerlach, M. Tyrosinase is not detected in 111. International Parkinson’s Disease Genomics Consortium (IPDGC); human catecholaminergic neurons by immunohistochemistry and western blot Wellcome Trust Case Control Consortium 2 (WTCCC2). A two-stage meta-ana- analysis. J. Neural Transm. Suppl. 72,51–55 (2007). lysis identifies several new loci for Parkinson’s disease. PLoS Genet. 7, e1002142 129. Segura-Aguilar, J. et al. Protective and toxic roles of dopamine in Parkinson’s (2011). disease. J. Neurochem. 129, 898–915 (2014). 112. Kanaan, N. M. et al. The longitudinal transcriptomic response of the substantia 130. Makin, O. S. & Serpell, L. C. Structures for amyloid fibrils. FEBS J. 272, 5950–5961 nigra to intrastriatal 6-hydroxydopamine reveals significant upregulation of (2005). regeneration-associated genes. PLoS ONE 10, e0127768 (2015). 131. Follmer, C. et al. Oligomerization and membrane-binding properties of covalent 113. Cantagrel, V. & Lefeber, D. J. From glycosylation disorders to dolichol bio- adducts formed by the interaction of α-synuclein with the toxic dopamine synthesis defects: a new class of metabolic diseases. J. Inherit. Metab. Dis. 34, metabolite 3,4-dihydroxyphenylacetaldehyde (DOPAL). J. Biol. Chem. 290, 859–867 (2011). 27660–27679 (2015). 114. Rip, J. W., Blais, M. M. & Jiang, L. W. Low-density lipoprotein as a transporter of 132. Ferrari, E. et al. Synthesis and structural characterization of soluble neurome- dolichol intermediates in the mammalian circulation. Biochem. J. 297, 321–325 lanin analogs provides important clues to its biosynthesis. J. Biol. Inorg. Chem. (1994). 18,81–93 (2013). 115. Tollbom, O. & Dallner, G. Dolichol and dolichyl phosphate in human tissues. Br. J. 133. Conway, K. A., Rochet, J. C., Bieganski, R. M. & Lansbury, P. T. Jr. Kinetic stabili- Exp. Pathol. 67, 757–764 (1986). zation of the alpha-synuclein protofibril by a dopamine-alpha-synuclein adduct. 116. Chojnacki, T. & Dallner, G. The biological role of dolichol. Biochem. J. 251,1–9 Science 294, 1346–1349 (2001). (1988). 134. Der-Sarkissian, A., Jao, C. C., Chen, J. & Langen, R. Structural organization of 117. Van Houte, H. A., Van Veldhoven, P. P., Mannaerts, G. P., Baes, M. I. & Declercq, P. alpha-synuclein fibrils studied by site-directed spin labeling. J. Biol. Chem. 278, E. Metabolism of dolichol, dolichoic acid and nordolichoic acid in cultured cells. 37530–37535 (2003). Biochim. Biophys. Acta 1347,93–100 (1997). 135. Fernández, C. O. et al. NMR of alpha-synuclein-polyamine complexes elucidates the 118. Valtersson, C. et al. The influence of dolichol, dolichol esters, and dolichyl mechanism and kinetics of induced aggregation. EMBO J. 23, 2039–2046 (2004). phosphate on phospholipid polymorphism and fluidity in model membranes. J. 136. Comunian, C. et al. A comparative MudPIT analysis identifies different expres- Biol. Chem. 260, 2742–2751 (1985). sion profiles in heart compartments. Proteomics 11, 2320–2328 (2011). 119. van Duijn, G. et al. Dolichyl phosphate induces non-bilayer structures, vesicle 137. Sadygov, R. G. et al. Code developments to improve the efficiency of automated fusion and transbilayer movement of lipids: a model membrane study. Biochim. MS/MS spectra interpretation. J. Proteome Res. 1, 211–215 (2002). Biophys. Acta 861, 211–223 (1986). 138. Micallef, L. & Rodgers, P. eulerAPE: drawing area-proportional 3-Venn diagrams 120. Eckhardt, M. The role and metabolism of sulfatide in the nervous system. Mol. using ellipses. PLoS ONE 9, e101717 (2014). Neurobiol. 37,93–103 (2008). 121. Raposo, G. & Marks, M. S. Melanosomes--dark organelles enlighten endosomal membrane transport. Nat. Rev. Mol. Cell Biol. 8, 786–797 (2007). Open Access This article is licensed under a Creative Commons 122. Watt, B., van Niel, G., Raposo, G. & Marks, M. S. PMEL: a pigment cell-specific Attribution 4.0 International License, which permits use, sharing, model for functional amyloid formation. Pigment Cell Melanoma Res. 26, adaptation, distribution and reproduction in any medium or format, as long as you give 300–315 (2013). appropriate credit to the original author(s) and the source, provide a link to the Creative 123. Theos, A. C. et al. The PKD domain distinguishes the trafficking and amyloido- Commons license, and indicate if changes were made. The images or other third party genic properties of the pigment cell protein PMEL and its homologue GPNMB. material in this article are included in the article’s Creative Commons license, unless Pigment Cell Melanoma Res. 26, 470–486 (2013). indicated otherwise in a credit line to the material. If material is not included in the 124. Santini, E. et al. Critical involvement of cAMP/DARPP-32 and extracellular signal- article’s Creative Commons license and your intended use is not permitted by statutory regulated protein kinase signaling in L-DOPA-induced dyskinesia. J. Neurosci. 27, regulation or exceeds the permitted use, you will need to obtain permission directly 6995–7005 (2007). from the copyright holder. To view a copy of this license, visit http://creativecommons. 125. Santini, E. et al. Dopamine- and cAMP-regulated phosphoprotein of 32-kDa org/licenses/by/4.0/. (DARPP-32)-dependent activation of extracellular signal-regulated kinase (ERK) and mammalian target of rapamycin complex 1 (mTORC1) signaling in experi- © The Author(s) 2018 mental parkinsonism. J. Biol. Chem. 287, 27806–27812 (2012). Published in partnership with the Parkinson’s Foundation npj Parkinson’s Disease (2018) 17

Journal

npj Parkinson's DiseaseSpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off