Quiescin sulfhydryl oxidase 1 (QSOX1) glycosite mutation perturbs secretion but not Golgi localization

Quiescin sulfhydryl oxidase 1 (QSOX1) glycosite mutation perturbs secretion but not Golgi... Abstract Quiescin sulfhydryl oxidase 1 (QSOX1) catalyzes the formation of disulfide bonds in protein substrates. Unlike other enzymes with related activities, which are commonly found in the endoplasmic reticulum, QSOX1 is localized to the Golgi apparatus or secreted. QSOX1 is upregulated in quiescent fibroblast cells and secreted into the extracellular environment, where it contributes to extracellular matrix assembly. QSOX1 is also upregulated in adenocarcinomas, though the extent to which it is secreted in this context is currently unknown. To achieve a better understanding of factors that dictate QSOX1 localization and function, we aimed to determine how post-translational modifications affect QSOX1 trafficking and activity. We found a highly conserved N-linked glycosylation site to be required for QSOX1 secretion from fibroblasts and other cell types. Notably, QSOX1 lacking a glycan at this site arrives at the Golgi, suggesting that it passes endoplasmic reticulum quality control but is not further transported to the cell surface for secretion. The QSOX1 transmembrane segment is dispensable for Golgi localization and secretion, as fully luminal and transmembrane variants displayed the same trafficking behavior. This study provides a key example of the effect of glycosylation on Golgi exit and contributes to an understanding of late secretory sorting and quality control. enhanced aromatic sequon, N-linked glycosylation, O-linked glycosylation, subcellular localization, Tn antigen Introduction Successful protein folding and appropriate post-translational modifications (PTMs) allow trafficking along the secretory pathway, whereas defects in maturation result in protein degradation by quality control mechanisms. One major stage for quality assessment is at the exit from the endoplasmic reticulum (ER). Immature proteins or proteins not designated to progress along the secretory pathway are retrieved to or retained in the ER, and misfolded proteins are extracted from the ER for degradation. However, later decision points also exist in the secretory pathway (Briant et al. 2017). These points may be more sensitive to protein assembly state and PTMs than to folding per se. Furthermore, late secretory quality control is likely to occur by sorting and vesicular trafficking to the lysosome (Briant et al. 2017), rather than by direct retrotranslocation into the cytosol, though the latter has not been ruled out. Features by which proteins are recognized for quality control in the late secretory pathway are poorly understood, but a number of mechanisms are known to contribute to post-ER protein routing. The length and nature of transmembrane regions may influence whether proteins are retained in the Golgi or trafficked onwards (Munro 1995). Oligomerization of luminal domains (Chen et al. 2000) or the presence of certain cytosolic segments (Munro and Nichols 1999), can also promote Golgi localization. Other cytosolic motifs, typically short canonical amino acid sequences such as the acidic cluster/dileucine motif, contribute to cargo-selection for further transport out of the Golgi (reviewed in Bonifacino 2004; Kienzle and von Blume 2014). Cargo proteins lacking their own cytosolic selection motifs can also be recruited by interacting with transmembrane mediators that contain recognition motifs for the trafficking machinery (Chavez et al. 2007; Reczek et al. 2007; Canuel et al. 2008). Calcium is involved in certain Golgi cargo sorting events (von Blume et al. 2012). Lastly, glycan recognition appears to be an additional mechanism by which proteins can be recognized for trafficking out of the Golgi and directed to subsequent destinations. For example, N-linked glycosylation has been reported to be critical for apical vs. basolateral targeting of various proteins in polarized epithelial cells (Scheiffele et al. 1995; Gut et al. 1998; Martínez-Maza et al. 2001; Castillon et al. 2013). Other studies have shown a role for O-linked glycosylation in sorting and secretion (Yeaman et al. 1997; Zhang et al. 2010). Defective decisions regarding sorting and trafficking of cargo proteins play an important role in various diseases. Mutations within individual proteins that affect their trafficking lead to specific loss-of-function defects. For example, mutations in or near the transmembrane segment of the kidney Cl–/HCO3– anion exchanger result in retention of the protein in the Golgi and cause distal renal tubular acidosis (Cordat et al. 2006). Mutations in adaptors or other trafficking machinery components result in mislocalization of cargo proteins and produce diverse disease phenotypes (Dell’Angelica et al. 1999; Verkerk et al. 2009; Bauer et al. 2012). Proper intracellular trafficking and secretion is clearly essential for both cellular and organismal health, and insights into the mechanisms behind productive transport decisions in cell biology are often gained by the documentation and analysis of trafficking failures. Quiescin sulfhydryl oxidase 1 (QSOX1) (Sulfhydryl oxidase 1; UniProt O00391) is a catalyst of disulfide bond formation localized to the Golgi apparatus of many cell types (Ilani et al. 2013). Increased QSOX1 expression occurs in quiescent fibroblasts (Coppock et al. 1993, 2000), and the enzyme is then secreted (Coppock et al. 2000; Ilani et al. 2013). Secreted QSOX1 participates in assembly of the extracellular matrix (ECM) of cultured cells (Ilani et al. 2013). The presence of extracellular QSOX1 in serum (Israel et al. 2014; Zhang et al. 2016) and the correlation of extracellular QSOX1 and its degradation products with pancreatic cancer (Antwi et al. 2009; Pan et al. 2011) motivate a better understanding of QSOX1 trafficking. We therefore investigated QSOX1 glycan PTMs and how these influence trafficking and enzyme function. Results Identification of potential PTMs in QSOX1 QSOX1 is a four-domain protein consisting of two functional modules connected by a flexible linker (Alon et al. 2012) (Figure 1A). The carboxy-terminal of the two modules binds flavin adenine dinucleotide (FAD) (Thorpe et al. 2002). QSOX1 is naturally produced as two splice variants, one containing, and one lacking, a transmembrane segment downstream of the catalytic domains (Coppock et al. 1993; Thorpe et al. 2002; Rudolf et al. 2013). Both splice variants can be secreted, the longer following proteolytic cleavage to release the ectodomain from the transmembrane anchor (Ilani et al. 2013; Rudolf et al. 2013). We over-expressed in human embryonic kidney (HEK) cells a version of QSOX1 lacking both the transmembrane region and a poorly conserved segment of about 160 amino acids downstream of the catalytic modules. The enzyme was purified from cell culture supernatant and compared by mass spectrometry (MS) with a construct spanning the same region expressed in Escherichia coli (Figure 1B). An intact mass consistent with the amino acid sequence was readily obtained from the bacterial version. Specifically, a mass of 58,601.6 Da was measured, compared to a calculated value of 58,601.2 Da after considering the loss of 10 hydrogens due to formation of five disulfide bonds. In contrast, the QSOX1 version produced in mammalian cells was too heterogeneous in its PTMs to allow resolution of the charge states of the various sub-species (Figure 1B). Fig. 1. View largeDownload slide Potential post-translational modification sites in QSOX1. (A) The structure of QSOX1 is shown with domains colored and labeled. The flexible linker is shown as a dashed curve. The image is a composite of PDB IDs 3LLK and 3Q6O. Cysteine side chains are shown as spheres with yellow sulfur atoms, and asparagine side chains in N-X-T/S motifs are magenta (carbon) and blue (nitrogen). The bound flavin adenine dinucleotide (FAD) cofactor is orange. The two redox-active CXXC motifs are indicated. Glycosylation sites discussed in the text are labeled. (B) Comparison of intact mass analysis of QSOX1 produced in bacteria (top) or HEK cells (bottom). Extensive and heterogeneous post-translational modification of QSOX1 produced in mammalian cells led to a poorly resolved spectrum. (C) Sequence of QSOX1 without the transmembrane and transmembrane-proximal regions. The amino terminus after signal peptide cleavage is indicated by the bent arrow. Residues in peptides identified by MS in QSOX1 produced in bacteria but not observed in QSOX1 produced in mammalian cells are shown in red on a pink background. Residues observed in enzyme produced in both hosts are on a gray background. Residues observed only in the HEK version are on a gold background. Putative N-linked glycosylation sites are boxed, threonine residues in the linker are underlined, and a site of phosphorylation is circled. Fig. 1. View largeDownload slide Potential post-translational modification sites in QSOX1. (A) The structure of QSOX1 is shown with domains colored and labeled. The flexible linker is shown as a dashed curve. The image is a composite of PDB IDs 3LLK and 3Q6O. Cysteine side chains are shown as spheres with yellow sulfur atoms, and asparagine side chains in N-X-T/S motifs are magenta (carbon) and blue (nitrogen). The bound flavin adenine dinucleotide (FAD) cofactor is orange. The two redox-active CXXC motifs are indicated. Glycosylation sites discussed in the text are labeled. (B) Comparison of intact mass analysis of QSOX1 produced in bacteria (top) or HEK cells (bottom). Extensive and heterogeneous post-translational modification of QSOX1 produced in mammalian cells led to a poorly resolved spectrum. (C) Sequence of QSOX1 without the transmembrane and transmembrane-proximal regions. The amino terminus after signal peptide cleavage is indicated by the bent arrow. Residues in peptides identified by MS in QSOX1 produced in bacteria but not observed in QSOX1 produced in mammalian cells are shown in red on a pink background. Residues observed in enzyme produced in both hosts are on a gray background. Residues observed only in the HEK version are on a gold background. Putative N-linked glycosylation sites are boxed, threonine residues in the linker are underlined, and a site of phosphorylation is circled. To obtain further insight into possible sites of modification, the versions of QSOX1 produced in bacterial and mammalian cells were subjected to proteolysis and compared using liquid chromatography followed by tandem mass spectrometry (LC–MS-MS). When searching against unmodified amino acid sequences, peptide identifications missing only from mammalian cell-derived QSOX1 LC–MS/MS suggest sites of modification. Of the two potential N-linked glycosylation sites in this QSOX1 construct, one was detected in a peptide from the mammalian cell-derived enzyme, whereas the other was not (Figure 1C). In addition, the linker region, which is rich in threonines, was detected in QSOX1 produced in bacteria but not in mammalian cells (Figure 1C). The region spanning a reported phosphorylation site (Ser426) (Tagliabracci et al. 2015) was not detected in either of the enzyme versions. This set of results was consistent with a previous MS analysis of native QSOX1 secreted from fibroblasts, which showed that the peptide containing the first N-glycosylation site could readily be identified, but the threonine-rich peptides in the linker and the peptide containing the site of phosphorylation were not detected in their unmodified versions (Ilani et al. 2013). This preliminary PTM mapping pointed to sites meriting additional analysis. Mutagenesis confirms utilization of PTM sites To further investigate the modifications of QSOX1 produced in mammalian cells, consensus sequences or functional groups required for modification were eliminated by mutagenesis. The Asn–Gly–Ser sequences at the first (Asn130) and second (Asn243) sites of potential N-glycosylation were separately mutated to Gln–Gly–Ser to generate the mutants at sites hereafter referred to as N1 and N2, respectively. A quadruple mutant, designated O1, was constructed to eliminate four threonine residues in the linker (Thr276, Thr277, Thr281 and Thr282) (Figure 2A). Electrophoresis of these QSOX1 variants purified from HEK cell supernatants showed changes in migration pattern compared with wild type. Wild-type QSOX1 migrated on a 15% acrylamide gel as a doublet at approximately 65 kD (Figure 2B). Mutant N1 was observed as a single band migrating similarly to the lower band of wild-type QSOX1 (Figure 2B), suggesting that the N1 site is partially utilized. Mutant O1, calculated to be only 49 Da (i.e. <0.1%) lower in protein mass than wild-type QSOX1, retained the doublet pattern but migrated slightly faster than wild-type QSOX1 in the gel (Figure 2B). This slight but detectable change in migration may indicate removal of modification sites, though formally such a shift may also arise from a difference in sodium dodecyl sulfate (SDS) binding or a change in the hydrodynamic radius of the denatured protein. Mutant N2 was not secreted at sufficient levels from cells in this experiment, so its migration pattern was not analyzed. Fig. 2. View largeDownload slide QSOX1 mutants lacking putative post-translational modification sites. (A) Sequence of the flexible linker between QSOX1 modules showing the amino acid changes constituting the O1 mutant. The specific mutations were chosen to eliminate sites for O-glycosylation while retaining some degree of side chain β-branching (T→V) and hydrophilicity (T→N). Other possible choices of mutations could also have satisfied these requirements. (B) Plasmids encoding wild-type QSOX1 and the N1 and O1 mutants were transiently transfected into HEK cells. Proteins were purified from supernatant, separated on a 15% SDS-PAGE gel run under reducing conditions, and analyzed by western blot using an anti-QSOX1 antibody. The observed migration pattern was seen also in another cell type (Figure 6D). The positions of molecular weight markers are shown and labeled in kilodaltons. (C) VVL binding to QSOX1 produced in bacteria (bacteria) or HEK cells (WT) was compared to binding of the O1, N1 and phosphorylation-site (S to A) mutants. Except for the lane labeled “bacteria”, all proteins were produced in HEK cells and purified from culture supernatant. The top panel shows VVL binding to a blot from a reducing gel (12%) of the indicated proteins, the bottom panel shows Coomassie staining of a gel run in parallel. Panels B and C are from separate experiments using the same protein preparations. Fig. 2. View largeDownload slide QSOX1 mutants lacking putative post-translational modification sites. (A) Sequence of the flexible linker between QSOX1 modules showing the amino acid changes constituting the O1 mutant. The specific mutations were chosen to eliminate sites for O-glycosylation while retaining some degree of side chain β-branching (T→V) and hydrophilicity (T→N). Other possible choices of mutations could also have satisfied these requirements. (B) Plasmids encoding wild-type QSOX1 and the N1 and O1 mutants were transiently transfected into HEK cells. Proteins were purified from supernatant, separated on a 15% SDS-PAGE gel run under reducing conditions, and analyzed by western blot using an anti-QSOX1 antibody. The observed migration pattern was seen also in another cell type (Figure 6D). The positions of molecular weight markers are shown and labeled in kilodaltons. (C) VVL binding to QSOX1 produced in bacteria (bacteria) or HEK cells (WT) was compared to binding of the O1, N1 and phosphorylation-site (S to A) mutants. Except for the lane labeled “bacteria”, all proteins were produced in HEK cells and purified from culture supernatant. The top panel shows VVL binding to a blot from a reducing gel (12%) of the indicated proteins, the bottom panel shows Coomassie staining of a gel run in parallel. Panels B and C are from separate experiments using the same protein preparations. Lectin detection and MS mapping of glycosylation in QSOX1 To confirm the presence of O-linked glycosylation suggested by the results shown above, we compared the binding of Vicia villosa lectin (VVL) to wild-type QSOX1 and the O1 mutant, both expressed in mammalian cells, as well as to QSOX1 expressed in bacteria. VVL binds to the Tn antigen (Porter et al. 2010), the single N-acetylgalactosamine (GalNAc) linked to a serine or threonine residue resulting from the first step of O-linked glycosylation in the Golgi apparatus. Wild-type QSOX1 expressed in HEK cells was detected by VVL, whereas the bacterially expressed QSOX1 was not (Figure 2C). VVL binding was also observed in QSOX1 mutants N1 and a serine-to-alanine mutant (S426A) that removes the reported phosphorylation site. The O1 mutant showed weaker but detectable VVL binding, perhaps due to another modified serine or threonine that remains in this mutant. A previous, proteome-wide MS study of O-glycosylation revealed O-linked GalNAc modifications of five threonines in the QSOX1 linker, including the four mutated in O1 (Steentoft et al. 2013) and an additional residue (Thr289). To further define the modifications in the enzyme and potentially identify additional sites, we performed an in-depth glycopeptide and phosphosite MS analysis of the N1 QSOX1 variant purified from HEK cell supernatants (Table I). In addition to the five O-glycosylated threonines in the QSOX1 linker (Steentoft et al. 2013), which we observed to carry Tn antigens (T289 also carries T antigen, with and without sialyation), a few serine residues were found to be glycosylated. At these sites, a substantial amount of sialyl Tn antigen was detected (Figure 3A). The phosphorylation at Ser426 (Tagliabracci et al. 2015) was not detected, but other phosphorylated serines, threonines, and tyrosines were observed. Finally, the MS results of the N1 mutant revealed the nature of the N-linked glycan at the N2 site (Table I and Figure 3B). The N-glycan population at N2 appeared heterogeneous and showed complex glycosylation modifications, including sialylation, indicating that this site undergoes extensive processing in the Golgi following its initial glycosylation in the ER. Fig. 3. View largeDownload slide Examples of mass spectra for QSOX1 peptides with glycan modifications. Annotation diagrams were made according to the best match in the UniCarbKB database (Campbell et al. 2011). A minus symbol shows that the indicated group has been lost from the ion. (A) A sialylated peptide. (B) A peptide detailing N-linked glycosylation at the N2 site. Fig. 3. View largeDownload slide Examples of mass spectra for QSOX1 peptides with glycan modifications. Annotation diagrams were made according to the best match in the UniCarbKB database (Campbell et al. 2011). A minus symbol shows that the indicated group has been lost from the ion. (A) A sialylated peptide. (B) A peptide detailing N-linked glycosylation at the N2 site. Table I. Modified QSOX1 peptides Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  aAmino acids indicated in lowercase in peptide sequences are sites of modification. Flanking residues are separated by periods from the identified peptides. bETD is electron-transfer dissociation, HCD is high-energy collisional dissociation, and EThcD is a combination of the two. View Large Table I. Modified QSOX1 peptides Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  aAmino acids indicated in lowercase in peptide sequences are sites of modification. Flanking residues are separated by periods from the identified peptides. bETD is electron-transfer dissociation, HCD is high-energy collisional dissociation, and EThcD is a combination of the two. View Large O-glycosylation protects QSOX1 from proteolytic degradation but is not required for sulfhydryl oxidase activity Based on previous observations that QSOX1 produced in bacteria was sensitive to proteolysis at the intermodule linker (Alon et al. 2010), we assessed the effect of linker PTMs in a limited proteolysis assay. The O1 mutant and wild-type QSOX1, both expressed in HEK cells, were compared. Serial dilutions of trypsin were applied to aliquots of the two proteins, and cleavage was analyzed by SDS polyacrylamide gel electrophoresis (SDS-PAGE) (Figure 4). The O1 mutant was found to be more sensitive to trypsin cleavage in the linker region, despite the fact that no new trypsin recognition sites were introduced, and overall the amino acid changes were minimal in terms of size and polarity (Figure 2A). This observation implies that glycosylation in the QSOX1 linker region affects accessibility of the polypeptide backbone to protease. Fig. 4. View largeDownload slide Glycan modification of the QSOX1 linker protects the enzyme from proteolysis. Wild-type QSOX1 and the O1 mutant were incubated with serial dilutions of trypsin, and the products of the limited proteolysis were separated on a 15% acrylamide SDS gel and stained with Coomassie. The positions of molecular weight markers are shown and labeled in kilodaltons. The bands indicated by yellow arrowheads are the products of QSOX1 proteolytic cleavage by trypsin. Chymotrypsin cleavage using the same protein preparations showed similar results, and QSOX1 produced in HEK cells was more resistant than QSOX1 produced in bacteria to cleavage with either trypsin or chymotrypsin (not shown). Fig. 4. View largeDownload slide Glycan modification of the QSOX1 linker protects the enzyme from proteolysis. Wild-type QSOX1 and the O1 mutant were incubated with serial dilutions of trypsin, and the products of the limited proteolysis were separated on a 15% acrylamide SDS gel and stained with Coomassie. The positions of molecular weight markers are shown and labeled in kilodaltons. The bands indicated by yellow arrowheads are the products of QSOX1 proteolytic cleavage by trypsin. Chymotrypsin cleavage using the same protein preparations showed similar results, and QSOX1 produced in HEK cells was more resistant than QSOX1 produced in bacteria to cleavage with either trypsin or chymotrypsin (not shown). Wild-type, N1, O1 and a S426D mutant of QSOX1 were also subjected to activity assays monitoring oxygen consumption upon oxidation of the model substrate dithiothreitol (DTT) as previously described (Grossman et al. 2013). The serine mutant was not expected to affect activity and was included in the experiment as an additional positive control. All tested QSOX1 variants showed similar rates of oxygen consumption when supplied with reducing substrate (Figure 5); the small variation was within the range of error for enzyme concentration determination. Thus, these PTMs do not affect steady-state turnover rates of QSOX1 in vitro. Fig. 5. View largeDownload slide QSOX1 glycosylation mutants are catalytically active on model substrates. Oxygen consumption during DTT oxidation was monitored as a measure of QSOX1 sulfhydryl oxidase activity. Representative reaction progress traces are shown in color (WT, orange; S426D, green; N1, red; O1, blue), and turnover numbers are presented in the bar graph insert as the average and standard deviation of five measurements. DTT was injected at the points indicated by the arrowheads to a final concentration of 1 mM. The inset shows a Coomassie-stained gel of the QSOX1 variants preparations subjected to the activity assays. Molecular weight markers are in kilodaltons. Fig. 5. View largeDownload slide QSOX1 glycosylation mutants are catalytically active on model substrates. Oxygen consumption during DTT oxidation was monitored as a measure of QSOX1 sulfhydryl oxidase activity. Representative reaction progress traces are shown in color (WT, orange; S426D, green; N1, red; O1, blue), and turnover numbers are presented in the bar graph insert as the average and standard deviation of five measurements. DTT was injected at the points indicated by the arrowheads to a final concentration of 1 mM. The inset shows a Coomassie-stained gel of the QSOX1 variants preparations subjected to the activity assays. Molecular weight markers are in kilodaltons. Conserved N-linked glycosylation is essential for Golgi export After analyzing the secreted QSOX1 variants that were successfully obtained, we questioned whether the lack of secretion of the N2 mutant was due to a trafficking defect. To obtain high quality spatial differentiation of organelles, intracellular staining of QSOX1 was conducted in HeLa cells, which are larger and more spread out than HEK cells. As observed for HEK cells, secretion of N2 from HeLa cells was minimal, whereas the other QSOX1 variants could be detected in cell culture supernatants (Figure 6A). Intracellular staining of QSOX1 was stronger in cells transfected with each of the QSOX1 variants than in mock transfected cells (Figure 6B and C), indicating that the expression from plasmids was greater than the production of endogenous, wild-type QSOX1. Remarkably, immunofluorescence staining of QSOX1 and organelle markers in HeLa cells indicated Golgi localization for all variants, including N2. Furthermore, Golgi staining of N2 was observed at intensities comparable to the other mutants (Figure 6B), and total intracellular QSOX1 staining was quantitatively similar (Figure 6C). To test whether the lack of secretion of the N2 mutant was specific to cell lines that do not normally secrete significant amounts of endogenous QSOX1, we transfected the various QSOX1 constructs into fibroblast cells, known for their secretion of QSOX1 during quiescence (Coppock et al. 2000; Ilani et al. 2013). Similarly to the observations described above, all QSOX1 variants except for the N2 mutant were found in the fibroblast culture supernatants (Figure 6D). In this experiment, mock transfected fibroblasts appeared to secrete more of the endogenous protein than cells transfected with the various short QSOX1 constructs, probably because the mock transfected cells were more confluent by the end of the experiment than the cells transfected with QSOX1 constructs and had dedicated no cellular resources to production of plasmid-encoded protein. Fig. 6. View largeDownload slide A highly conserved N-linked glycan is essential for QSOX1 trafficking. (A) Plasmids encoding wild-type, the N1, N2 and O1 QSOX1 mutants, or an empty vector (Mock) were transiently transfected into HeLa cells. The culture supernatant was subjected to TCA precipitation before it was analyzed by western blot using QSOX1 antibody. (B) The transfected cells were fixed and immunofluorescently stained with ER-specific (PDI) or Golgi-specific (GM130) antibodies (red) and QSOX1 antibody (green). (C) Quantification of per-cell QSOX1 fluorescence, corresponding to the experiment shown in panel B. Error bars indicate one standard deviation. (D) The same plasmids as in panel A were transfected into WI-38 fibroblast cells, and culture supernatants were TCA-precipitated and analyzed by western blot using QSOX1 antibody. Asterisks denote the endogenous protein. Arrows indicate the protein produced from the transfected plasmid. Fig. 6. View largeDownload slide A highly conserved N-linked glycan is essential for QSOX1 trafficking. (A) Plasmids encoding wild-type, the N1, N2 and O1 QSOX1 mutants, or an empty vector (Mock) were transiently transfected into HeLa cells. The culture supernatant was subjected to TCA precipitation before it was analyzed by western blot using QSOX1 antibody. (B) The transfected cells were fixed and immunofluorescently stained with ER-specific (PDI) or Golgi-specific (GM130) antibodies (red) and QSOX1 antibody (green). (C) Quantification of per-cell QSOX1 fluorescence, corresponding to the experiment shown in panel B. Error bars indicate one standard deviation. (D) The same plasmids as in panel A were transfected into WI-38 fibroblast cells, and culture supernatants were TCA-precipitated and analyzed by western blot using QSOX1 antibody. Asterisks denote the endogenous protein. Arrows indicate the protein produced from the transfected plasmid. All experiments described above were performed using a QSOX1 expression construct lacking the transmembrane region and thus resembling the shorter of the two QSOX1 splice variants. To determine whether the conserved N2 glycosylation site is important to the secretion of luminal QSOX1 only, HeLa cells were transfected with QSOX1 constructs that include the transmembrane region and correspond to the longer of the splice variants. The presence of the transmembrane region and short cytosolic tail of QSOX1 did not alter the secretion patterns observed: the QSOX1 N2 mutant containing the transmembrane region was poorly secreted, as seen for its luminal counterpart (Figure 7). Fig. 7. View largeDownload slide The transmembrane region of QSOX1 does not alter N2 secretion. Plasmids encoding an empty vector (Mock), QSOX1 constructs that include its transmembrane region (Full length): WT, N1 and N2 QSOX1 mutants, or QSOX1 lacking its transmembrane region (Short) were transiently transfected into HeLa cells. The culture supernatants were precipitated using TCA, separated on SDS-page, and analyzed by western blot using QSOX1 antibody. The upper arrow indicates the longer form of QSOX1 after processing for secretion, the lower arrow indicates the position of the shorter QSOX1 variant. This experiment was performed once and showed consistency with Figure 6A. Fig. 7. View largeDownload slide The transmembrane region of QSOX1 does not alter N2 secretion. Plasmids encoding an empty vector (Mock), QSOX1 constructs that include its transmembrane region (Full length): WT, N1 and N2 QSOX1 mutants, or QSOX1 lacking its transmembrane region (Short) were transiently transfected into HeLa cells. The culture supernatants were precipitated using TCA, separated on SDS-page, and analyzed by western blot using QSOX1 antibody. The upper arrow indicates the longer form of QSOX1 after processing for secretion, the lower arrow indicates the position of the shorter QSOX1 variant. This experiment was performed once and showed consistency with Figure 6A. Intracellular QSOX1 N2 mutant is catalytically active and shows Golgi glycan modifications Proteins missing essential PTMs may be retained in the cell due to folding defects. To test whether the QSOX1 N2 mutant retained in cells is misfolded, we transfected large-scale (0.5–1 L) suspension-adapted HEK cell cultures with plasmids encoding the QSOX1 variants, purified intracellular enzyme on the basis of the polyhistidine tag, and tested activity. Due to the lesser amounts of enzyme that could be obtained from cell lysates compared to supernatants, a discontinuous colorimetric assay was used instead of the continuous oxygen consumption assay shown in Figure 5. The colorimetric assay showed qualitatively that the N2 mutant purified from cells was active, as was wild-type QSOX1 purified in parallel (Figure 8A). The dependence of the observed activity on the transfected QSOX1 constructs was demonstrated in two ways. First, comparable fractions collected from affinity chromatography of lysates from mock transfected cells showed no sulfhydryl oxidase activity. Second, the activity observed for wild-type and N2 QSOX1 purified from cell lysates was completely eliminated upon addition of a QSOX1 inhibitory antibody (Grossman et al. 2013). This experiment demonstrates that the N2 mutant protein present in cells is folded and functional. The trace amount of N2 mutant secreted into the medium was also purified separately, concentrated, and found to be catalytically active and thermally stable within 2°C of wild-type QSOX1 and the N1 variant in a SYPRO Orange thermal shift assay (transition temperatures for wild type, N1 and N2 were 66.4, 65.9 and 64.4°C, respectively). Fig. 8. View largeDownload slide N2 purified from cell lysates is active and shows Golgi processing of the remaining N-glycan. (A) N2 and wild-type QSOX1 purified from HEK 293F cells were assayed for sulfhydryl oxidase activity. Both wild-type and N2 showed robust oxidation of the model substrate DTT. DTT depletion was quantified by addition of DTNB, the reduced version of which shows increased absorbance at 412 nm. The sulfhydryl oxidase activity was completely inhibited by QSOX1 inhibitory antibodies (Ab), confirming that the activities of both the wild-type and N2 QSOX1 preparations were due to QSOX1 and not to a contaminant. Furthermore, mock transfected cells subjected to the same purification procedure (Mock) showed no activity in the same chromatographic elution fractions. Controls (cont) lacked enzyme. Error bars are standard deviations. (B) The N2 QSOX1 variant purified from cells was subjected to mass spectrometric analysis of peptides and glycans, and the spectrum of a peptide containing the N1 glycan is shown. The annotation diagram was made according to the best match in the UniCarbKB database (Campbell et al. 2011). “M” indicates the mass of the parent peptide, and “H” is a hydrogen in this context. A minus symbol shows that the indicated group has been lost from the ion. Fig. 8. View largeDownload slide N2 purified from cell lysates is active and shows Golgi processing of the remaining N-glycan. (A) N2 and wild-type QSOX1 purified from HEK 293F cells were assayed for sulfhydryl oxidase activity. Both wild-type and N2 showed robust oxidation of the model substrate DTT. DTT depletion was quantified by addition of DTNB, the reduced version of which shows increased absorbance at 412 nm. The sulfhydryl oxidase activity was completely inhibited by QSOX1 inhibitory antibodies (Ab), confirming that the activities of both the wild-type and N2 QSOX1 preparations were due to QSOX1 and not to a contaminant. Furthermore, mock transfected cells subjected to the same purification procedure (Mock) showed no activity in the same chromatographic elution fractions. Controls (cont) lacked enzyme. Error bars are standard deviations. (B) The N2 QSOX1 variant purified from cells was subjected to mass spectrometric analysis of peptides and glycans, and the spectrum of a peptide containing the N1 glycan is shown. The annotation diagram was made according to the best match in the UniCarbKB database (Campbell et al. 2011). “M” indicates the mass of the parent peptide, and “H” is a hydrogen in this context. A minus symbol shows that the indicated group has been lost from the ion. The purification of the N2 mutant from cells allowed us to examine its PTMs as well as its activity. MS analysis of the remaining (N1) glycosite of the N2 mutant revealed Golgi modifications of the glycan at the N1 position, including mannosidase trimming and GlcNAc transferase additions (Figure 8B). Fucosylation was also detected. These MS results support the immunofluorescence staining (Figure 6) showing colocalization with Golgi markers of all tested QSOX1 variants, including the N2 mutant. Discussion Inspection of the sequence context of N2 in human QSOX1 reveals that it corresponds to a structural motif termed the “enhanced aromatic sequon”, which has been recognized to increase the stability of proteins when glycosylated (Culyba et al. 2011) and to promote the efficiency and homogeneity of glycosylation (Murray et al. 2015). Notably, a glycan at the N2 position is not essential for QSOX1 protein folding, since QSOX1 produced in bacteria and thus lacking all glycans, including this one, is well-folded and functional (Alon et al. 2012; Ilani et al. 2013). Furthermore, the N2 mutant isolated from supernatants of large-scale mammalian cell cultures or from the intracellular organelle fraction was stable and showed sulfhydryl oxidase activity, demonstrating that a glycan at this position in QSOX1 does not contribute critically to the structural or functional integrity of the enzyme. However, the QSOX1 N2 site is clearly glycosylated more consistently than is the N1 site, which deviates in its sequence and structural context from the enhanced aromatic sequon consensus (Figure 9). The main feature of the enhanced aromatic sequon is the presence of an aromatic amino acid, typically phenylalanine, two residues before the glycosylated asparagine in a hairpin turn connecting two β-strands in a β-sheet. The phenylalanine two residues before the asparagine in the N2 site of QSOX1 is part of a cluster of partially surface-exposed hydrophobic and aromatic amino acids on a poorly hydrated face of the protein (Figure 9). In addition, on the other side of the β-sheet from the phenylalanine, a hydrophobic groove is found on the domain surface. A glycan at N2 is expected to at least partially obscure these hydrophobic surfaces. The glycan itself, perhaps in the context of surrounding surface features of the protein, may serve as a handle for trafficking out of the Golgi for productive secretion. Another hypothesis is that the extensive hydrophobic surfaces exposed in the absence of the glycan in the N2 mutant are recognized as a signal for Golgi retention and eventual lysosomal degradation rather than secretion. Comparison of structural data with cell biological observations of glycan modification and trafficking will continue to shed light on the physical features of proteins that enhance glycosylation, as well as on the function of glycans once they have been added in particular structural contexts. Fig. 9. View largeDownload slide The N2 region contains exposed hydrophobic and aromatic residues. Structures displayed are from the amino-terminal fragment of human QSOX1 (PDB ID 3Q6O). In the top panels, the asparagines of the N1 and N2 sites are shown in space-filling representation. Cyan spheres representing water molecules bound to the protein in the crystal structure show that the region around the N1 site is more hydrated than the region around N2. Exposed hydrophobic and aromatic side chains, which are much more prevalent near N2, are shown as yellow sticks. The region labeled with a circled red “1” constitutes a cluster of surface-exposed hydrophobic side chains. The region labeled with a circled red “2” is a hydrophobic groove on the surface of the protein. The bottom panels show the secondary structure contexts of the two glycosylation sites. A five-residue region spanning each asparagine is colored lime green, and the side chains near the asparagines are in stick format. Fig. 9. View largeDownload slide The N2 region contains exposed hydrophobic and aromatic residues. Structures displayed are from the amino-terminal fragment of human QSOX1 (PDB ID 3Q6O). In the top panels, the asparagines of the N1 and N2 sites are shown in space-filling representation. Cyan spheres representing water molecules bound to the protein in the crystal structure show that the region around the N1 site is more hydrated than the region around N2. Exposed hydrophobic and aromatic side chains, which are much more prevalent near N2, are shown as yellow sticks. The region labeled with a circled red “1” constitutes a cluster of surface-exposed hydrophobic side chains. The region labeled with a circled red “2” is a hydrophobic groove on the surface of the protein. The bottom panels show the secondary structure contexts of the two glycosylation sites. A five-residue region spanning each asparagine is colored lime green, and the side chains near the asparagines are in stick format. QSOX1 belongs to a set of enzymes that introduce cross-links into ECM proteins. Other enzymes with ECM cross-linking functions are lysyl oxidase and its paralogs (Robins 2007), peroxidasin (Bhave et al. 2012), and transglutaminases (Aeschlimann and Thomazy 2000). Interestingly, these other enzymes, with the possible exception of a transglutaminase associated with the male reproductive tract (Dubbink et al. 1998; Cho et al. 2010), are not reported to be Golgi localized. Of the secreted cross-linking enzymes that have been studied, only QSOX1 appears to be concentrated in the Golgi in its intracellular form. This observation suggests a different mechanism for localization and secretion of QSOX1 compared with the other enzymes. Secreted QSOX1 is found in diverse biological contexts. The enzyme is localized to the Golgi apparatus of many cell types, but it is upregulated and secreted from confluent cultured fibroblasts. QSOX1 is found in blood serum, milk and other biological fluids (Janolino and Swaisgood 1975; Ostrowski et al. 1979; Ostrowski and Kistler 1980; Hoober et al. 1996; Israel et al. 2014). It was originally purified from semen and chicken eggs (Ostrowski and Kistler 1980; Hoober et al. 1996). Higher levels of QSOX1 or its degradation products were found in serum of pancreatic cancer patients compared to healthy controls (Antwi et al. 2009; Pan et al. 2011), suggesting that greater amounts of the protein may be secreted in the patients, though this possibility remains to be investigated. These observations call for a better understanding of QSOX1 secretion, about which little is known. A report by Bullied and colleagues described the proteolytic cleavage of Golgi-localized QSOX1 and the secretion of the enzyme into the culture supernatant (Rudolf et al. 2013). That study also noted that QSOX1 is modified post-translationally by glycosylation, but the link between its modification and trafficking was not investigated. Prior to analyzing the effects of QSOX1 PTMs on secretion of the enzyme, we made a detailed catalog of these modifications. On the basis of the QSOX1 amino acid sequence, potential sites of N-linked glycosylation, i.e. N-X-S/T motifs, were identified at positions 130, 243, 575 and 591. However, only the first two sites, corresponding to N1 and N2, were predicted using NetNGlyc (Gupta and Brunak 2002) to be modified. These two sites are evolutionarily conserved, with N2 showing stronger conservation. The other two N-X-S/T motifs, at positions 575 and 591, are located outside the region spanned by the four-domain QSOX1 construct used for much of the current study. Since the secretion pattern of mutants based on the full-length QSOX1 construct showed similar behavior as the truncated version (Figure 7), sites 575 and 591 were not further investigated. Regarding O-linked glycans, the residues mutated in the O1 quadruple mutant were among those predicted using NetOGlyc to be modified and found experimentally to be glycosites (Steentoft et al. 2013). To validate these putative sites of PTMs in our system, we first compared QSOX1 expressed in bacteria to QSOX1 expressed in cultured mammalian cells. MS analysis showed that the peptides containing the N2 and O1 sites were detected in the protein produced in bacteria but not that produced in mammalian cells, suggesting that they are indeed modified in the latter. The peptide containing N1 was detected in both QSOX1 versions, indicating that at least a fraction of the protein molecules remains unmodified at this position. Gel migration analysis of QSOX1 mutants produced in mammalian cells was used to further analyze PTMs. Whereas wild-type QSOX1 and the O1 mutant appeared as double bands on SDS-PAGE gels, the N1 mutant was observed as a single band. Partial utilization of N1 could explain both the migration pattern and the detection by MS of unmodified peptides spanning this site. Peptides containing an unmodified N1 site were also observed in the MS analysis of QSOX1 secreted from quiescent fibroblasts (Ilani et al. 2013). According to band intensities on gels, roughly half of the QSOX1 protein molecules contain a glycan at the N1 position, whether produced in fibroblasts or other cells (Figures 2, 4 and 6D). It remains to be determined whether a physical difference distinguishes QSOX1 molecules during their maturation in the ER so that only half of them are modified at N1, or if inherent inefficiencies in oligosaccharyltransferase modification of nonoptimal sequons are responsible. Moreover, the physiological importance of the partial utilization of this site and the role it plays in the function of QSOX1 remain to be revealed. One function identified for QSOX1 PTMs in this study is protection against proteases. QSOX1 undergoes large conformational changes during its catalytic cycle, facilitated by a flexible linker between its two redox-active modules (Alon et al. 2012). Avian QSOX1 was cleaved into two fragments at this linker, enabling the study of the properties of each fragment (Raje and Thorpe 2003), and we have previously observed that mammalian QSOX1 produced in bacteria is also sensitive to degradation at the flexible linker. In the current study, the O1 mutant proved to be more sensitive than wild-type QSOX1 to proteolytic cleavage when both enzyme variants were produced in mammalian cells, suggesting that one purpose of the O-linked glycans in the QSOX1 linker is to protect this vulnerable region of the protein. A threonine-rich linker is a conserved feature of mammalian QSOX1 enzymes, whereas avian, amphibian and fish QSOX1 enzymes often contain an N-linked glycosylation consensus site in this region instead. Glycosylation may be a general mechanism to shield the polypeptide backbone of the linker from proteases, while retaining its flexibility. This concept was discussed by Steentoft and colleagues, as linker domains of various proteins were found to be enriched with O-linked glycosylation (Steentoft et al. 2013). De-glycosylation of the QSOX1 linker region by mutagenesis does not interfere with enzymatic activity, however, since both the N1 and O1 mutants showed similar activity as wild-type QSOX1 on a model dithiol substrate (Figure 5). This observation was not surprising, since QSOX1 expressed in bacteria and lacking all glycosylation is functional in in vitro sulfhydryl oxidase assays (Grossman et al. 2013; Ilani et al. 2013) and rescues the ECM-related phenotypes caused by the knockdown of QSOX1 in cell culture (Ilani et al. 2013). Whereas trafficking and secretion were robust to mutation of the N1 and O1 sites, we found that mutation of the N2 site prevented QSOX1 from being secreted from cells (Figures 6 and 7). Even fibroblast cells, which naturally secrete large quantities of QSOX1 (Coppock et al. 2000; Ilani et al. 2013), did not secrete the N2 mutant while still secreting the endogenous wild-type protein (Figure 6D). Upon investigation of the defect in N2, we expected to find this mutant retained in or retrotranslocated from the ER for failing to pass ER quality control. However, as shown in Figure 6B and C, the N2 mutant was observed in the Golgi apparatus instead, indicating that at least a substantial fraction had successfully exited the ER. Purification of the QSOX1 N2 mutant from cells allowed us to test sulfhydryl oxidase activity, revealing that this enzyme variant was as active as wild type and thus is folded and loaded with its FAD cofactor. A similar observation of trafficking defects was made for a glycan variant of ADAM8, a transmembrane protein normally expressed on the cell surface that is retained in the Golgi when one of its N-linked glycosylation sites is mutated (Srinivasan et al. 2014). A surprising finding from this study is that QSOX1 variants with and without a transmembrane anchor and cytosolic tail are trafficked similarly. Despite the importance of transmembrane regions and linear cytosolic motifs in selection of proteins for Golgi residence or transport out of the compartment (Guo et al. 2014), these features do not appear to determine whether secretion of QSOX1 will occur. QSOX1 constructs comprising only the catalytic domains were secreted from cells similarly to QSOX1 constructs corresponding to the longer QSOX1 splice variant with the transmembrane anchor, and secretion of both versions was sensitive to the presence of the N2 glycosylation site. The similar behavior observed for the long and short QSOX1 constructs may reflect the possibility that packaging for secretion occurs subsequent to the proteolytic cleavage that releases the transmembrane region, such that both constructs lack transmembrane and cytosolic signals at this stage. It should be noted, though, that they still differ after cleavage by about 125 residues at the carboxy-terminus, including two putative N-linked glycosylation sites. In any event, our findings place emphasis on Golgi luminal features for selection of QSOX1 for secretion. While the identity of these features and how they are recognized by the Golgi-localized transport machinery are still unclear, the importance of a conserved N-linked glycan for QSOX1 secretion is evident from our results. Quality control in the late secretory pathway is an emerging field of research, and the molecular mechanisms involved are beginning to be addressed. The link we have uncovered between a PTM site and the ability of QSOX1 to be secreted suggests that this enzyme may provide a useful system with which to further explore quality assessment and trafficking control in the Golgi apparatus. Materials and methods Cell lines and maintenance WI-38 fibroblasts were purchased from Coriell and were maintained in Minimal Essential Medium (MEM) supplemented with 10% fetal bovine serum (FBS), l-glutamine, antibiotics, nonessential amino acids and sodium pyruvate. HEK 293T and HeLa cells were obtained from the laboratories of Dr. Ron Diskin and Prof. Moshe Oren, respectively, and maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS, l-glutamine, antibiotics, nonessential amino acids and sodium pyruvate. Cells were checked for mycoplasma approximately once a month. Expression of QSOX1 and mutants The pcDNA3.1 plasmids expressing human QSOX1 and its variants were transfected into adherent HEK 293T, HeLa or WI-38 cells using jetPEI transfection reagent (Polyplus Transfections) according to manufacturer’s instructions. Suspension HEK 293F cells (Thermo Fisher) were cultured in FreeStyle 293 expression medium and transfected using the PEI Max reagent (Polysciences Inc.). For purification of QSOX1 and QSOX1 mutants, the portion of the enzyme spanning residues 1–546, with six histidines appended to the carboxy terminus, was produced by transient transfection of HEK 293T cells or HEK 293F cells. HEK 293T cells were transferred after transfection to serum-free medium. Four days after transfection of adherent cells, culture supernatant containing secreted QSOX1 (which consisted, after signal sequence cleavage, of residues 30–546 and the His6 tag) was dialyzed against phosphate buffered saline to lower the glutamine concentration. HEK 293F cells were harvested 2 days post transfection by centrifuging at 1000 rpm for 10 min, and organelle proteins were isolated (Holden and Horton 2009). QSOX1 was then purified by nickel–nitrilotriacetic acid chromatography, aliquoted and stored at −80°C after addition of glycerol to 10%. Concentrations of QSOX1 variant stocks produced from cell culture supernatants were determined by absorbance at 280 nm using an extinction coefficient of 112,000 M−1 cm−1, which includes the contribution of the bound FAD cofactor. Relative concentrations of enzyme purified from HEK 293F lysates were estimated based on western blotting, since absorbance readings were unreliable due to low protein concentrations. For western blot analyses shown in the figures, culture supernatants from transient transfections were subjected to trichloroacetic acid (TCA) precipitation and cold acetone washes before SDS-PAGE. QSOX1 was expressed in bacteria according to published protocol (Alon et al. 2012). Mass spectrometry Each MS experiment was conducted on one sample. The preliminary peptide fingerprinting analysis of QSOX1 produced in bacteria vs. HEK cells was performed in the Mass Spectrometry Unit of the Weizmann Institute Life Sciences Core Facilities according to published procedures (Ilani et al. 2013). For analysis of intact QSOX1, the samples were acidified to a final concentration of 1% formic acid and desalted on a 4.6 × 50 mm2 ProSwift RP-2S column using 1 mL/min flow on a Waters Acquity UPLC. Samples were loaded on the column on 20% B (80% acetonitrile (MeCN), 0.1% formic acid) for 5 min, followed by elution at 80% B for 5 min. Samples were dried using a speed-vac and analyzed using a nanoAquity nUPLC (Waters) coupled to a Tribrid Orbitrap Fusion Lumos instrument (Thermo Fisher). Sample was loaded and desalted on column using a 0.1 mm×50 cm ProSwift column (Thermo) for trapping and separation. Separation was conducted using a gradient of 2–50% B (80% MeCN, 0.1% formic acid) over 50 min. Sample was analyzed using the following parameters: spray voltage 1.8 kV, ion transfer tube temperature 275°C, and data were acquired alternating between orbitrap and ion trap detection. Oribtrap scans were performed at a range of 700–1500 m/z in 15,000 resolution (@ 200 m/z), 70% RF, in source dissociation of 35 eV, AGC target of 4e5 in maximum of 50 ms, summing up 20 microscans. Ion trap scans were done with the same parameters except using enhanced mode and AGC target of 3e4 with maximum injection time of 35 ms. For detailed glycopeptide analysis, protein samples were loaded onto 3 kDa molecular weight cut-off spin columns. Volume was reduced to 25 μL by centrifugation at 14,000 ×g for 10 min. The 175 μL 8 M urea was added, and the sample was centrifuged at 14,000× g for 10 min. Filters were reversed and centrifuged to extract the proteins. Proteins were reduced with 5 mM DTT for 1 h at room temperature and then alkylated with 10 mM iodoacetamide (Sigma) in the dark for 45 min at room temperature. Samples were diluted to 2 M urea with 50 mM ammonium bicarbonate. Proteins were then subjected to digestion with trypsin (Promega; Madison, WI, USA) overnight at 37°C at 50:1 protein:trypsin ratio, followed by a second trypsin digestion for 4 h. The digestions were stopped by addition of trifluroacetic acid (1% final concentration). Following digestion, peptides were desalted using Oasis HLB, μElution format (Waters, Milford, MA, USA). The samples were vacuum dried and stored in −80°C until further analysis. Samples were analyzed using the nanoAquity nUPLC coupled to the Tribrid Orbitrap Fusion Lumos instrument. Samples were loaded on a 0.18×20 mm2, 5 μm C18 Symmetry (Waters) column at 10 μL/min for 1 min in 2% B. Sample was then separated on a 0.075×250 mm2, 1.8 μm HSS T3 C18 column, using a gradient of 4–30% B over 50 min. Mass spectrometry was performed using the following parameters: spray voltage 1.6 kV, ion transfer tube temperature 275°C. MS1 was acquired at 120,000 resolution (@ 200 m/z) at a range of 300–1800 m/z, AGC target 4e5 with maximum of 50 ms inject time. MS2 was set to 3 s cycle time, with MIPS on, selecting precursors of intensity higher than 5e4, with charges between +2 and +8. Each precursor was selected for fragmentation once, and then excluded for 30 s. Precursors were isolated using a 1 m/z window and AGC target 5e4 with maximum inject time of 100 ms, fragmented using either HCD set to 30NCE, EThcD with ETD reaction time of 60 ms and 3e5 reagent target ion and HCD set to 15NCE or EThcD with ETD reaction time of 60 ms and 3e5 reagent target ion and HCD set to 22NCE. MS2 acquisition was set at a fixed first mass of 130 m/z, acquired at 15,000 resolution (@ 200 m/z). Data were analyzed using Byonic search engine (Protein Metrics Inc.), against the human proteome database. The search was performed iteratively, first constructing a limited focused database, then searching common modifications such as oxidations and deamidations, then searching against the Human O- and N-glycosylation libraries provided by Byonic. Western blot and lectin staining Samples were separated on SDS-PAGE and transferred to a nitrocellulose membrane. After protein transfer, the membrane was blocked with 5% bovine serum albumin (BSA) in phosphate buffered saline (PBS) containing 0.1% Tween (PBS-T). Following blocking, the membrane was blotted using a polyclonal rabbit anti human QSOX1 antibody (Ilani et al. 2013) with 5% BSA in PBS-T overnight at 4°C. The membrane was then washed three times in PBS-T and blotted with a secondary antibody conjugated to horseradish peroxidase (HRP) (Abcam). Following washes with PBS-T, HRP reaction was carried out using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) and visualized using a ChemiDoc XRS+ System (Bio-Rad). For lectin staining, a similar procedure was followed except that biotin-conjugated VVL (Vector Laboratories) was incubated with the membrane for 1 h at room temperature, and streptavidin-conjugated HRP was used instead of secondary antibody. Immunofluorescence Cells were grown on coverslips in a 24-well plate, fixed for 20 min using 3.7% formaldehyde, and permeabilized in 0.1% Triton X-100 in PBS for 5 min. Cells were then blocked with 5% BSA in PBS-T and incubated overnight at 4°C with primary antibody in 5% BSA in PBS-T. Staining was performed using polyclonal rabbit anti human QSOX1 and either mouse anti GM130 (Abcam) or mouse anti PDI (Abcam). The cells were then washed three times with PBS-T and incubated with a fluorescently labeled secondary antibody before being placed face down onto 5 μL Fluoroshield antifade reagent (Sigma) on coverslips. Samples were observed on an Olympus CKX31SF2 imaging system and analyzed using Fiji software. Per-cell total fluorescence for QSOX1 immuno-staining was measured by integrating the intensity within boxes of uniform size positioned to enclose entire single cells. At least 40 cells were quantified for each QSOX1 variant. These cells were selected randomly from imaged fields, except that cells in close proximity to a neighbor such that they could not be boxed individually were avoided. Limited proteolysis Trypsin and was dissolved at a concentration of 1 mg/mL in 20 mM Tris, pH 8.0, 200 mM NaCl. Four-fold serial dilutions were made into the same buffer. One microliter of protease was added to 7 μL of QSOX1 (wild-type or O1 mutant) from a 3.8 μM stock in PBS plus 10% glycerol. Reactions were incubated at RT for 30 min. Oxygen consumption assays QSOX1 variants were assayed at a concentration of 100 nM with 1 mM DTT in a Clarke-type oxygen electrode (Hansatech Instruments) at 25°C. Reactions were initiated by DTT injection. The oxygen electrode buffer (OEB) was 50 mM potassium phosphate buffer, pH 7.5, 65 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA). Colorimetric sulfhydryl oxidase assay Intracellular QSOX1 was incubated in OEB with 200 μM DTT with or without 250 nM of a QSOX1 inhibitory antibody (Grossman et al. 2013). At each time point, an aliquot of the reaction was diluted 1:10 into 500 μM 5,5′-dithio-bis-[2-nitrobenzoic acid] (DTNB) in OEB, and, after 5 min incubation, absorbance was recorded at 412 nm in an Infinite 200 Pro plate reader (TECAN). The assay was done in triplicate. SYPRO Orange thermal shift assay QSOX1 and glycan variants at concentrations of 1–7 μM were incubated with a 1:250 dilution of SYPRO Orange 5000× stock (Thermofisher) in a 96-well PCR plate in a volume of 20 μL per well, and fluorescence data were collected on an Applied Biosystems Real-Time PCR System using the fluorescein amidite channel. The temperature was held for 1 min per 0.5° from 24 to 95°C. Funding This work was supported by the European Research Council (ERC) under the European Union’s Seventh Framework Programme [Grant number 310649]. Acknowledgements The authors thank Tevie Mehlman and Dalit Merhav for mass spectrometry protein fingerprinting. Conflict of interest statement None declared. Abbreviations BSA bovine serum albumin DTT dithiothreitol ECM extracellular matrix ER endoplasmic reticulum; ETD electron-transfer dissociation EThcD electron-transfer/higher-energy collision dissociation FAD flavin adenine dinucleotide GalNAc N-acetylgalactosamine HCD higher-energy collision dissociation HEK human embryonic kidney HRP horseradish peroxidase LC–MS-MS liquid chromatography followed by tandem mass spectrometry MeCN acetonitrile MEM minimal essential medium MS mass spectrometry N1 Asn-to-Gln mutation at position 130 in QSOX1 N2 Asn-to-Gln mutation at position 243 in QSOX1 O1 QSOX1 mutant with the threonines at positions 276, 277, 281 and 282 replaced by other amino acids OEB oxygen electrode buffer PBS phosphate buffered saline PTMs post-translational modifications QSOX1 quiescin sulfhydryl oxidase 1 SDS sodium dodecyl sulfate SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis TCA trichloroacetic acid VVL Vicia villosa lectin References Aeschlimann D, Thomazy V. 2000. Protein crosslinking in assembly and remodeling of extracellular matrices: The role of transglutaminases. Connect Tissue Res . 41: 1– 27. Google Scholar CrossRef Search ADS PubMed  Alon A, Grossman I, Gat Y, Kodali VK, DiMaio F, Mehlman T, Haran G, Baker D, Thorpe C, Fass D. 2012. The dynamic disulphide relay of quiescin sulphydryl oxidase. Nature . 488: 414– 418. Google Scholar CrossRef Search ADS PubMed  Alon A, Hekler EJ, Thorpe C, Fass D. 2010. QSOX contains a pseudo-dimer of functional and degenerate sulfhydryl oxidase domains. FEBS Lett . 584: 1521– 1525. Google Scholar CrossRef Search ADS PubMed  Antwi K, Hostetter G, Demeure MJ, Katchman BA, Decker GA, Ruiz Y, Sielaff TD, Koep LJ, Lake DF. 2009. Analysis of the plasma peptidome from pancreas cancer patients connects a peptide in plasma to overexpression of the parent protein in tumors. J Proteome Res . 8: 4722– 4231. Google Scholar CrossRef Search ADS PubMed  Bauer P, Leshinsky-Silver E, Blumkin L, Schlipf N, Schröder C, Schicks J, Lev D, Riess O, Lerman-Sagie T, Schöls L. 2012. Mutation in the AP4B1 gene cause hereditary spastic paraplegia type 47 (SPG47). Neurogenetics . 13: 73– 76. Google Scholar CrossRef Search ADS PubMed  Bhave G, Cummings CF, Vanacore RM, Kumagai-Cresse C, Ero-Tolliver IA, Rafi M, Kang JS, Pedchenko V, Fessler LI, Fessler JH et al.  . 2012. Peroxidasin forms sulfilimine chemical bonds using hypohalous acids in tissue genesis. Nat Chem Biol . 8: 784– 790. Google Scholar CrossRef Search ADS PubMed  Bonifacino JS. 2004. The GGA proteins: Adaptors on the move. Nat Rev Mol Cell Biol . 5: 23– 32. Google Scholar CrossRef Search ADS PubMed  Briant K, Johnson N, Swanton E. 2017. Transmembrane domain quality control systems operate at the endoplasmic reticulum and Golgi apparatus. PLoS One . 12: e0173924. Google Scholar CrossRef Search ADS PubMed  Campbell MP, Haves CA, Struwe WB, Wilkins MR, Aoki-Kinoshita KF, Harvey DJ, Rudd PM, Kolarich D, Lisacek F, Karsson NG et al.  . 2011. UniCarbKB: Putting the pieces together for glycomics research. Proteomics . 11: 4117– 4121. Google Scholar CrossRef Search ADS PubMed  Canuel M, Korkidakis A, Konnyu K, Morales CR. 2008. Sortilin mediates the lysosomal targeting of cathepsins D and H. Biochem Biophys Res Commun . 373: 292– 297. Google Scholar CrossRef Search ADS PubMed  Castillon GA, Michon L, Watanabe R. 2013. Apical sorting of lysoGPI-anchored proteins occurs independent of association with detergent-resistant membranes but dependent on their N-glycosylation. Mol Biol Cell . 24: 2021– 2033. Google Scholar CrossRef Search ADS PubMed  Chavez CA, Bohnsack RN, Kudo M, Gotschall RR, Canfield WM, Dahms NM. 2007. Domain 5 of the cation-independent mannose 6-phosphate receptor preferentially binds phosphodiesters (mannose 6-phosphate N-acetylglucosamine ester). Biochemistry . 46: 12604– 12617. Google Scholar CrossRef Search ADS PubMed  Chen C, Ma J, Lazic A, Backovic M, Colley KJ. 2000. Formation of insoluble oligomers correlates with ST6Gal I stable localization in the Golgi. J Biol Chem . 275: 13819– 13826. Google Scholar CrossRef Search ADS PubMed  Cho SY, Jeon JH, Kim CW, Shin DM, Jang GY, Jeong EM, Lee SE, Song KY, Kim IG. 2010. Monoclonal antibodies to human transglutaminase 4. Hybridoma . 29: 263– 267. Google Scholar CrossRef Search ADS PubMed  Coppock D, Kopman C, Gudas J, Cina-Poppe DA. 2000. Regulation of the quiescence-induced genes: Quiescin Q6, decorin, and ribosomal protein S29. Biochem Biophys Res Commun . 269: 604– 610. Google Scholar CrossRef Search ADS PubMed  Coppock DL, Kopman C, Scandalis S, Gilleran S. 1993. Preferential gene expression in quiescent human lung fibroblasts. Cell Growth Differ . 4: 483– 493. Google Scholar PubMed  Cordat E, Kittanakom S, Yenchitsomanus PT, Li J, Du K, Lukacs GL, Reithmeier RA. 2006. Dominant and recessive distal renal tubular acidosis mutations of kidney anion exchanger 1 induce distinct trafficking defects in MDCK cells. Traffic . 7: 117– 128. Google Scholar CrossRef Search ADS PubMed  Culyba EK, Price JL, Hanson SR, Dhar A, Wong CH, Gruebele M, Powers ET, Kelly JW. 2011. Protein native-state stabilization by placing aromatic side chains in N-glycosylated reverse turns. Science . 331: 571– 575. Google Scholar CrossRef Search ADS PubMed  Dell’Angelica EC, Shotelersuk V, Aguilar RC, Gahl WA, Bonifacino JS. 1999. Altered trafficking of lysosomal proteins in Hermansky-Pudlak syndrome due to mutations in the beta 3A subunit of the AP-3 adaptor. Mol Cell . 3: 11– 21. Google Scholar CrossRef Search ADS PubMed  Dubbink HJ, de Waal L, van Haperen R, Verkaik NS, Trapman J, Romijn JC. 1998. The human prostate-specific transglutaminase gene (TGM4): Genomic organization, tissue-specific expression, and promoter characterization. Genomics . 51: 434– 444. Google Scholar CrossRef Search ADS PubMed  Grossman I, Alon A, Ilani T, Fass D. 2013. An inhibitory antibody blocks the first step in the dithiol/disulfide relay mechanism of the enzyme QSOX1. J Mol Biol . 425: 4366– 4378. Google Scholar CrossRef Search ADS PubMed  Guo Y, Sirkis DW, Schekman R. 2014. Protein sorting at the trans-Golgi network. Annu Rev Cell Dev Biol . 30: 169– 206. Google Scholar CrossRef Search ADS PubMed  Gupta R, Brunak S. 2002. Prediction of glycosylation across the human proteome and the correlation to protein function. Pac Symp Biocomput . 7: 310– 322. Gut A, Kappeler F, Hyka N, Balda MS, Hauri HP, Matter K. 1998. Carbohydrate-mediated Golgi to cell surface transport and apical targeting of membrane proteins. EMBO J . 17: 1919– 1929. Google Scholar CrossRef Search ADS PubMed  Holden P, Horton WA. 2009. Crude subcellular fractionation of cultured mammalian cell lines. BMC Res Notes . 2: 243. Google Scholar CrossRef Search ADS PubMed  Hoober KL, Joneja B, White HB 3rd, Thorpe C. 1996. A sulfhydryl oxidase from chicken egg white. J Biol Chem . 271: 30510– 30516. Google Scholar CrossRef Search ADS PubMed  Ilani T, Alon A, Grossman I, Horowitz B, Kartvelishvily E, Cohen SR, Fass D. 2013. A secreted disulfide catalyst controls extracellular matrix composition and function. Science . 341: 74– 76. Google Scholar CrossRef Search ADS PubMed  Israel BA, Jiang L, Gannon SA, Thorpe C. 2014. Disulfide bond generation in mammalian blood serum: Detection and purification of quiescin-sulfhydryl oxidase. Free Radic Biol Med . 69: 129– 135. Google Scholar CrossRef Search ADS PubMed  Janolino VG, Swaisgood HE. 1975. Isolation and characterization of sulfhydryl oxidase from bovine milk. J Biol Chem . 250: 2532– 2538. Google Scholar PubMed  Kienzle C, von Blume J. 2014. Secretory cargo sorting at the trans-Golgi network. Trends Cell Biol . 24: 584– 593. Google Scholar CrossRef Search ADS PubMed  Martínez-Maza R, Poyatos I, López-Corcuera B, Núñez E, Giménez C, Zafra F, Aragón C. 2001. The role of N-glycosylation in transport to the plasma membrane and sorting of the neuronal glycine transporter GLYT2. J Biol Chem . 276: 2168– 2173. Google Scholar CrossRef Search ADS PubMed  Munro S. 1995. An investigation of the role of transmembrane domains in Golgi protein retention. EMBO J . 14: 4695– 4704. Google Scholar PubMed  Munro S, Nichols BJ. 1999. The GRIP domain—A novel Golgi-targeting domain found in several coiled-coil proteins. Curr Biol . 9: 377– 380. Google Scholar CrossRef Search ADS PubMed  Murray AN, Chen W, Antonopoulos A, Hanson SR, Wiseman RL, Dell A, Haslam SM, Powers DL, Powers ET, Kelly JW. 2015. Enhanced aromatic sequons increase oligosaccharyltransferase glycosylation efficiency and glycan homogeneity. Chem Biol . 22: 1052– 1062. Google Scholar CrossRef Search ADS PubMed  Ostrowski MC, Kistler WS. 1980. Properties of a flavoprotein sulfhydryl oxidase from rat seminal vesicle secretion. Biochem . 19: 2639– 2645. Google Scholar CrossRef Search ADS   Ostrowski MC, Kistler WS, Williams-Ashman HG. 1979. A flavoprotein responsible for the intense sulfhydryl oxidase activity of rat seminal vesicle secretion. Biochem Biophys Res Commun . 87: 171– 176. Google Scholar CrossRef Search ADS PubMed  Pan S, Chen R, Crispin DA, May D, Stevens T, McIntosh MW, Bronner MP, Ziogas A, Anton-Culver H, Brentnall TA. 2011. Protein alterations associated with pancreatic cancer and chronic pancreatitis found in human plasma using global quantitative proteomics profiling. J Proteome Res . 10: 2359– 2376. Google Scholar CrossRef Search ADS PubMed  Porter A, Yue T, Heeringa L, Day S, Suh E, Haab BB. 2010. A motif-based analysis of glycan array data to determine the specificities of glycan-binding proteins. Glycobiology . 20: 369– 380. Google Scholar CrossRef Search ADS PubMed  Raje S, Thorpe C. 2003. Inter-domain redox communication in flavoenzymes of the quiescin/sulfhydryl oxidase family: Role of a thioredoxin domain in disulfide bond formation. Biochem . 42: 4560– 4568. Google Scholar CrossRef Search ADS   Reczek D, Schwake M, Schröder J, Hughes H, Blanz J, Jin X, Brondyk W, Van Patten S, Edmunds T, Saftig P. 2007. LIMP-2 is a receptor for lysosomal mannose-6-phosphate-independent targeting of beta-glucocerebrosidase. Cell . 131: 770– 783. Google Scholar CrossRef Search ADS PubMed  Robins SP. 2007. Biochemistry and functional significance of collagen cross-linking. Biochem Soc Trans . 35: 849– 852. Google Scholar CrossRef Search ADS PubMed  Rudolf J, Pringle MA, Bulleid NJ. 2013. Proteolytic processing of QSOX1A ensures efficient secretion of a potent disulfide catalyst. Biochem J . 454: 181– 190. Google Scholar CrossRef Search ADS PubMed  Scheiffele P, Peränen J, Simons K. 1995. N-glycans as apical sorting signals in epithelial cells. Nature . 378: 96– 98. Google Scholar CrossRef Search ADS PubMed  Srinivasan S, Romagnoli M, Bohm A, Sonenshein GE. 2014. N-Glycosylation regulates ADAM8 processing and activation. J Biol Chem . 289: 33676– 33688. Google Scholar CrossRef Search ADS PubMed  Steentoft C, Vakhrushev SY, Joshi HJ, Kong Y, Vester-Christensen MB, Schjoldager KT, Lavrsen K, Dabelsteen S, Pedersen NB, Marcos-Silva L et al.  . 2013. Precision mapping of the human O-GalNAc glycoproteome through SimpleCell technology. EMBO J . 32: 1478– 1488. Google Scholar CrossRef Search ADS PubMed  Tagliabracci VS, Wiley SE, Guo X, Kinch LN, Durrant E, Wen J, Xiao J, Cui J, Nguyen KB, Engel JL et al.  . 2015. A single kinase generates the majority of the secreted phosphoproteome. Cell . 161: 1619– 1632. Google Scholar CrossRef Search ADS PubMed  Thorpe C, Hoober KL, Raje S, Glynn NM, Burnside J, Turi GK, Coppock DL. 2002. Sulfhydryl oxidases: Emerging catalysts of protein disulfide bond formation in eukaryotes. Arch Biochem Biophys . 405: 1– 12. Google Scholar CrossRef Search ADS PubMed  Verkerk AJ, Schot R, Dumee B, Schellekens K, Swagemakers S, Bertoli-Avella AM, Lequin MH, Dudink J, Govaert P, van Zwol AL et al.  . 2009. Mutation in the AP4M1 gene provides a model for neuroaxonal injury in cerebral palsy. Am J Hum Genet . 85: 40– 52. Google Scholar CrossRef Search ADS PubMed  von Blume J, Alleaume AM, Kienzle C, Carreras-Sureda A, Valverde M, Malhotra V. 2012. Cab45 is required for Ca(2+)-dependent secretory cargo sorting at the trans-Golgi network. J Cell Biol . 199: 1057– 1066. Google Scholar CrossRef Search ADS PubMed  Yeaman C, Le Gall AH, Baldwin AN, Monlauzeur L, Le Bivac A, Rodriguez-Boulan E. 1997. The O-glycosylated stalk domain is required for apical sorting of neurotrophin receptors in polarized MDCK cells. J Cell Biol . 139: 929– 940. Google Scholar CrossRef Search ADS PubMed  Zhang L, Cheng Q, Zhang L, Wang Y, Merrill GF, Ilani T, Fass D, Arnér ES, Zhang J. 2016. Serum thioredoxin reductase is highly increased in mice with hepatocellular carcinoma and its activity is restrained by several mechanisms. Free Radic Biol Med . 99: 426– 435. Google Scholar CrossRef Search ADS PubMed  Zhang L, Tran DT, Ten Hagen KG. 2010. An O-glycosyltransferase promotes cell adhesion during development by influencing secretion of an extracellular matrix integrin ligand. J Biol Chem . 285: 19491– 19501. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Glycobiology Oxford University Press

Quiescin sulfhydryl oxidase 1 (QSOX1) glycosite mutation perturbs secretion but not Golgi localization

Loading next page...
 
/lp/ou_press/quiescin-sulfhydryl-oxidase-1-qsox1-glycosite-mutation-perturbs-ZbR7ceVp4i
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0959-6658
eISSN
1460-2423
D.O.I.
10.1093/glycob/cwy044
Publisher site
See Article on Publisher Site

Abstract

Abstract Quiescin sulfhydryl oxidase 1 (QSOX1) catalyzes the formation of disulfide bonds in protein substrates. Unlike other enzymes with related activities, which are commonly found in the endoplasmic reticulum, QSOX1 is localized to the Golgi apparatus or secreted. QSOX1 is upregulated in quiescent fibroblast cells and secreted into the extracellular environment, where it contributes to extracellular matrix assembly. QSOX1 is also upregulated in adenocarcinomas, though the extent to which it is secreted in this context is currently unknown. To achieve a better understanding of factors that dictate QSOX1 localization and function, we aimed to determine how post-translational modifications affect QSOX1 trafficking and activity. We found a highly conserved N-linked glycosylation site to be required for QSOX1 secretion from fibroblasts and other cell types. Notably, QSOX1 lacking a glycan at this site arrives at the Golgi, suggesting that it passes endoplasmic reticulum quality control but is not further transported to the cell surface for secretion. The QSOX1 transmembrane segment is dispensable for Golgi localization and secretion, as fully luminal and transmembrane variants displayed the same trafficking behavior. This study provides a key example of the effect of glycosylation on Golgi exit and contributes to an understanding of late secretory sorting and quality control. enhanced aromatic sequon, N-linked glycosylation, O-linked glycosylation, subcellular localization, Tn antigen Introduction Successful protein folding and appropriate post-translational modifications (PTMs) allow trafficking along the secretory pathway, whereas defects in maturation result in protein degradation by quality control mechanisms. One major stage for quality assessment is at the exit from the endoplasmic reticulum (ER). Immature proteins or proteins not designated to progress along the secretory pathway are retrieved to or retained in the ER, and misfolded proteins are extracted from the ER for degradation. However, later decision points also exist in the secretory pathway (Briant et al. 2017). These points may be more sensitive to protein assembly state and PTMs than to folding per se. Furthermore, late secretory quality control is likely to occur by sorting and vesicular trafficking to the lysosome (Briant et al. 2017), rather than by direct retrotranslocation into the cytosol, though the latter has not been ruled out. Features by which proteins are recognized for quality control in the late secretory pathway are poorly understood, but a number of mechanisms are known to contribute to post-ER protein routing. The length and nature of transmembrane regions may influence whether proteins are retained in the Golgi or trafficked onwards (Munro 1995). Oligomerization of luminal domains (Chen et al. 2000) or the presence of certain cytosolic segments (Munro and Nichols 1999), can also promote Golgi localization. Other cytosolic motifs, typically short canonical amino acid sequences such as the acidic cluster/dileucine motif, contribute to cargo-selection for further transport out of the Golgi (reviewed in Bonifacino 2004; Kienzle and von Blume 2014). Cargo proteins lacking their own cytosolic selection motifs can also be recruited by interacting with transmembrane mediators that contain recognition motifs for the trafficking machinery (Chavez et al. 2007; Reczek et al. 2007; Canuel et al. 2008). Calcium is involved in certain Golgi cargo sorting events (von Blume et al. 2012). Lastly, glycan recognition appears to be an additional mechanism by which proteins can be recognized for trafficking out of the Golgi and directed to subsequent destinations. For example, N-linked glycosylation has been reported to be critical for apical vs. basolateral targeting of various proteins in polarized epithelial cells (Scheiffele et al. 1995; Gut et al. 1998; Martínez-Maza et al. 2001; Castillon et al. 2013). Other studies have shown a role for O-linked glycosylation in sorting and secretion (Yeaman et al. 1997; Zhang et al. 2010). Defective decisions regarding sorting and trafficking of cargo proteins play an important role in various diseases. Mutations within individual proteins that affect their trafficking lead to specific loss-of-function defects. For example, mutations in or near the transmembrane segment of the kidney Cl–/HCO3– anion exchanger result in retention of the protein in the Golgi and cause distal renal tubular acidosis (Cordat et al. 2006). Mutations in adaptors or other trafficking machinery components result in mislocalization of cargo proteins and produce diverse disease phenotypes (Dell’Angelica et al. 1999; Verkerk et al. 2009; Bauer et al. 2012). Proper intracellular trafficking and secretion is clearly essential for both cellular and organismal health, and insights into the mechanisms behind productive transport decisions in cell biology are often gained by the documentation and analysis of trafficking failures. Quiescin sulfhydryl oxidase 1 (QSOX1) (Sulfhydryl oxidase 1; UniProt O00391) is a catalyst of disulfide bond formation localized to the Golgi apparatus of many cell types (Ilani et al. 2013). Increased QSOX1 expression occurs in quiescent fibroblasts (Coppock et al. 1993, 2000), and the enzyme is then secreted (Coppock et al. 2000; Ilani et al. 2013). Secreted QSOX1 participates in assembly of the extracellular matrix (ECM) of cultured cells (Ilani et al. 2013). The presence of extracellular QSOX1 in serum (Israel et al. 2014; Zhang et al. 2016) and the correlation of extracellular QSOX1 and its degradation products with pancreatic cancer (Antwi et al. 2009; Pan et al. 2011) motivate a better understanding of QSOX1 trafficking. We therefore investigated QSOX1 glycan PTMs and how these influence trafficking and enzyme function. Results Identification of potential PTMs in QSOX1 QSOX1 is a four-domain protein consisting of two functional modules connected by a flexible linker (Alon et al. 2012) (Figure 1A). The carboxy-terminal of the two modules binds flavin adenine dinucleotide (FAD) (Thorpe et al. 2002). QSOX1 is naturally produced as two splice variants, one containing, and one lacking, a transmembrane segment downstream of the catalytic domains (Coppock et al. 1993; Thorpe et al. 2002; Rudolf et al. 2013). Both splice variants can be secreted, the longer following proteolytic cleavage to release the ectodomain from the transmembrane anchor (Ilani et al. 2013; Rudolf et al. 2013). We over-expressed in human embryonic kidney (HEK) cells a version of QSOX1 lacking both the transmembrane region and a poorly conserved segment of about 160 amino acids downstream of the catalytic modules. The enzyme was purified from cell culture supernatant and compared by mass spectrometry (MS) with a construct spanning the same region expressed in Escherichia coli (Figure 1B). An intact mass consistent with the amino acid sequence was readily obtained from the bacterial version. Specifically, a mass of 58,601.6 Da was measured, compared to a calculated value of 58,601.2 Da after considering the loss of 10 hydrogens due to formation of five disulfide bonds. In contrast, the QSOX1 version produced in mammalian cells was too heterogeneous in its PTMs to allow resolution of the charge states of the various sub-species (Figure 1B). Fig. 1. View largeDownload slide Potential post-translational modification sites in QSOX1. (A) The structure of QSOX1 is shown with domains colored and labeled. The flexible linker is shown as a dashed curve. The image is a composite of PDB IDs 3LLK and 3Q6O. Cysteine side chains are shown as spheres with yellow sulfur atoms, and asparagine side chains in N-X-T/S motifs are magenta (carbon) and blue (nitrogen). The bound flavin adenine dinucleotide (FAD) cofactor is orange. The two redox-active CXXC motifs are indicated. Glycosylation sites discussed in the text are labeled. (B) Comparison of intact mass analysis of QSOX1 produced in bacteria (top) or HEK cells (bottom). Extensive and heterogeneous post-translational modification of QSOX1 produced in mammalian cells led to a poorly resolved spectrum. (C) Sequence of QSOX1 without the transmembrane and transmembrane-proximal regions. The amino terminus after signal peptide cleavage is indicated by the bent arrow. Residues in peptides identified by MS in QSOX1 produced in bacteria but not observed in QSOX1 produced in mammalian cells are shown in red on a pink background. Residues observed in enzyme produced in both hosts are on a gray background. Residues observed only in the HEK version are on a gold background. Putative N-linked glycosylation sites are boxed, threonine residues in the linker are underlined, and a site of phosphorylation is circled. Fig. 1. View largeDownload slide Potential post-translational modification sites in QSOX1. (A) The structure of QSOX1 is shown with domains colored and labeled. The flexible linker is shown as a dashed curve. The image is a composite of PDB IDs 3LLK and 3Q6O. Cysteine side chains are shown as spheres with yellow sulfur atoms, and asparagine side chains in N-X-T/S motifs are magenta (carbon) and blue (nitrogen). The bound flavin adenine dinucleotide (FAD) cofactor is orange. The two redox-active CXXC motifs are indicated. Glycosylation sites discussed in the text are labeled. (B) Comparison of intact mass analysis of QSOX1 produced in bacteria (top) or HEK cells (bottom). Extensive and heterogeneous post-translational modification of QSOX1 produced in mammalian cells led to a poorly resolved spectrum. (C) Sequence of QSOX1 without the transmembrane and transmembrane-proximal regions. The amino terminus after signal peptide cleavage is indicated by the bent arrow. Residues in peptides identified by MS in QSOX1 produced in bacteria but not observed in QSOX1 produced in mammalian cells are shown in red on a pink background. Residues observed in enzyme produced in both hosts are on a gray background. Residues observed only in the HEK version are on a gold background. Putative N-linked glycosylation sites are boxed, threonine residues in the linker are underlined, and a site of phosphorylation is circled. To obtain further insight into possible sites of modification, the versions of QSOX1 produced in bacterial and mammalian cells were subjected to proteolysis and compared using liquid chromatography followed by tandem mass spectrometry (LC–MS-MS). When searching against unmodified amino acid sequences, peptide identifications missing only from mammalian cell-derived QSOX1 LC–MS/MS suggest sites of modification. Of the two potential N-linked glycosylation sites in this QSOX1 construct, one was detected in a peptide from the mammalian cell-derived enzyme, whereas the other was not (Figure 1C). In addition, the linker region, which is rich in threonines, was detected in QSOX1 produced in bacteria but not in mammalian cells (Figure 1C). The region spanning a reported phosphorylation site (Ser426) (Tagliabracci et al. 2015) was not detected in either of the enzyme versions. This set of results was consistent with a previous MS analysis of native QSOX1 secreted from fibroblasts, which showed that the peptide containing the first N-glycosylation site could readily be identified, but the threonine-rich peptides in the linker and the peptide containing the site of phosphorylation were not detected in their unmodified versions (Ilani et al. 2013). This preliminary PTM mapping pointed to sites meriting additional analysis. Mutagenesis confirms utilization of PTM sites To further investigate the modifications of QSOX1 produced in mammalian cells, consensus sequences or functional groups required for modification were eliminated by mutagenesis. The Asn–Gly–Ser sequences at the first (Asn130) and second (Asn243) sites of potential N-glycosylation were separately mutated to Gln–Gly–Ser to generate the mutants at sites hereafter referred to as N1 and N2, respectively. A quadruple mutant, designated O1, was constructed to eliminate four threonine residues in the linker (Thr276, Thr277, Thr281 and Thr282) (Figure 2A). Electrophoresis of these QSOX1 variants purified from HEK cell supernatants showed changes in migration pattern compared with wild type. Wild-type QSOX1 migrated on a 15% acrylamide gel as a doublet at approximately 65 kD (Figure 2B). Mutant N1 was observed as a single band migrating similarly to the lower band of wild-type QSOX1 (Figure 2B), suggesting that the N1 site is partially utilized. Mutant O1, calculated to be only 49 Da (i.e. <0.1%) lower in protein mass than wild-type QSOX1, retained the doublet pattern but migrated slightly faster than wild-type QSOX1 in the gel (Figure 2B). This slight but detectable change in migration may indicate removal of modification sites, though formally such a shift may also arise from a difference in sodium dodecyl sulfate (SDS) binding or a change in the hydrodynamic radius of the denatured protein. Mutant N2 was not secreted at sufficient levels from cells in this experiment, so its migration pattern was not analyzed. Fig. 2. View largeDownload slide QSOX1 mutants lacking putative post-translational modification sites. (A) Sequence of the flexible linker between QSOX1 modules showing the amino acid changes constituting the O1 mutant. The specific mutations were chosen to eliminate sites for O-glycosylation while retaining some degree of side chain β-branching (T→V) and hydrophilicity (T→N). Other possible choices of mutations could also have satisfied these requirements. (B) Plasmids encoding wild-type QSOX1 and the N1 and O1 mutants were transiently transfected into HEK cells. Proteins were purified from supernatant, separated on a 15% SDS-PAGE gel run under reducing conditions, and analyzed by western blot using an anti-QSOX1 antibody. The observed migration pattern was seen also in another cell type (Figure 6D). The positions of molecular weight markers are shown and labeled in kilodaltons. (C) VVL binding to QSOX1 produced in bacteria (bacteria) or HEK cells (WT) was compared to binding of the O1, N1 and phosphorylation-site (S to A) mutants. Except for the lane labeled “bacteria”, all proteins were produced in HEK cells and purified from culture supernatant. The top panel shows VVL binding to a blot from a reducing gel (12%) of the indicated proteins, the bottom panel shows Coomassie staining of a gel run in parallel. Panels B and C are from separate experiments using the same protein preparations. Fig. 2. View largeDownload slide QSOX1 mutants lacking putative post-translational modification sites. (A) Sequence of the flexible linker between QSOX1 modules showing the amino acid changes constituting the O1 mutant. The specific mutations were chosen to eliminate sites for O-glycosylation while retaining some degree of side chain β-branching (T→V) and hydrophilicity (T→N). Other possible choices of mutations could also have satisfied these requirements. (B) Plasmids encoding wild-type QSOX1 and the N1 and O1 mutants were transiently transfected into HEK cells. Proteins were purified from supernatant, separated on a 15% SDS-PAGE gel run under reducing conditions, and analyzed by western blot using an anti-QSOX1 antibody. The observed migration pattern was seen also in another cell type (Figure 6D). The positions of molecular weight markers are shown and labeled in kilodaltons. (C) VVL binding to QSOX1 produced in bacteria (bacteria) or HEK cells (WT) was compared to binding of the O1, N1 and phosphorylation-site (S to A) mutants. Except for the lane labeled “bacteria”, all proteins were produced in HEK cells and purified from culture supernatant. The top panel shows VVL binding to a blot from a reducing gel (12%) of the indicated proteins, the bottom panel shows Coomassie staining of a gel run in parallel. Panels B and C are from separate experiments using the same protein preparations. Lectin detection and MS mapping of glycosylation in QSOX1 To confirm the presence of O-linked glycosylation suggested by the results shown above, we compared the binding of Vicia villosa lectin (VVL) to wild-type QSOX1 and the O1 mutant, both expressed in mammalian cells, as well as to QSOX1 expressed in bacteria. VVL binds to the Tn antigen (Porter et al. 2010), the single N-acetylgalactosamine (GalNAc) linked to a serine or threonine residue resulting from the first step of O-linked glycosylation in the Golgi apparatus. Wild-type QSOX1 expressed in HEK cells was detected by VVL, whereas the bacterially expressed QSOX1 was not (Figure 2C). VVL binding was also observed in QSOX1 mutants N1 and a serine-to-alanine mutant (S426A) that removes the reported phosphorylation site. The O1 mutant showed weaker but detectable VVL binding, perhaps due to another modified serine or threonine that remains in this mutant. A previous, proteome-wide MS study of O-glycosylation revealed O-linked GalNAc modifications of five threonines in the QSOX1 linker, including the four mutated in O1 (Steentoft et al. 2013) and an additional residue (Thr289). To further define the modifications in the enzyme and potentially identify additional sites, we performed an in-depth glycopeptide and phosphosite MS analysis of the N1 QSOX1 variant purified from HEK cell supernatants (Table I). In addition to the five O-glycosylated threonines in the QSOX1 linker (Steentoft et al. 2013), which we observed to carry Tn antigens (T289 also carries T antigen, with and without sialyation), a few serine residues were found to be glycosylated. At these sites, a substantial amount of sialyl Tn antigen was detected (Figure 3A). The phosphorylation at Ser426 (Tagliabracci et al. 2015) was not detected, but other phosphorylated serines, threonines, and tyrosines were observed. Finally, the MS results of the N1 mutant revealed the nature of the N-linked glycan at the N2 site (Table I and Figure 3B). The N-glycan population at N2 appeared heterogeneous and showed complex glycosylation modifications, including sialylation, indicating that this site undergoes extensive processing in the Golgi following its initial glycosylation in the ER. Fig. 3. View largeDownload slide Examples of mass spectra for QSOX1 peptides with glycan modifications. Annotation diagrams were made according to the best match in the UniCarbKB database (Campbell et al. 2011). A minus symbol shows that the indicated group has been lost from the ion. (A) A sialylated peptide. (B) A peptide detailing N-linked glycosylation at the N2 site. Fig. 3. View largeDownload slide Examples of mass spectra for QSOX1 peptides with glycan modifications. Annotation diagrams were made according to the best match in the UniCarbKB database (Campbell et al. 2011). A minus symbol shows that the indicated group has been lost from the ion. (A) A sialylated peptide. (B) A peptide detailing N-linked glycosylation at the N2 site. Table I. Modified QSOX1 peptides Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  aAmino acids indicated in lowercase in peptide sequences are sites of modification. Flanking residues are separated by periods from the identified peptides. bETD is electron-transfer dissociation, HCD is high-energy collisional dissociation, and EThcD is a combination of the two. View Large Table I. Modified QSOX1 peptides Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  Peptidea  Modification site  Frag. method  Glycans  -.APRsALYSPSDPLTLLQADTVR.G  S33 Phospho  HCDb    R.SALYsPSDPLTLLQADTVR.G  S37 Phospho  HCD    R.SALySPSDPLTLLQADTVR.G  Y38 Phospho  EThcDb    R.LIDALESHHDtWPPAcPPLEPAKLEEIDGFFAR.N  T160 Phospho  HCD    R.NNEEyLALIFEK.G  Y187 Phospho  HCD    K.FGVtDFPScYLLFR.N  T232 HexNAc  EThcD    R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(4)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(5)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(1)Fuc(1)  R.nGSVSR.V  N243 N-glycans  EThcD  HexNAc(5)Hex(4)Fuc(1)NeuAc(2)  R.VPVLmEsRsFYTAYLQR.L  S255 O-glycan S257 O-glycan  EThcD  HexNAc, HexNAc(3)Hex(1)  R.SFYTAyLQR.L  Y262 Phospho  EThcD    R.EAAQttVAPttANK.I  T276 O-HexNAc T277 O-HexNAc T281 O-HexNAc T282 O-HexNAc  EThcD  HexNAc(1), HexNAc(1), HexNAc(1), HexNAc(1)  R.EAAQTtVAPTTANK.I  T277 O-HexNAc  EThcD  HexNAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  K.IAPtVWK.L  T289 O-glycan  EThcD  HexNAc(1)Hex(1)  K.IAPtVWK.L  T289 O-HexNAc  EThcD  HexNAc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETDb  HexNAc(2)Hex(1)Fuc(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(3)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)  R.DcAsHFEQMAAASMHR.V  S455 O-glycan  ETD  HexNAc(1)Hex(1)NeuAc(3)  R.VGSPNAAVLWLWsSHNR.V  S479 Phospho  EThcD    R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(2)  R.LAGAPsEDPQFPK.V  S493 O-glycan  EThcD  HexNAc(1)Hex(1)NeuAc(1)  aAmino acids indicated in lowercase in peptide sequences are sites of modification. Flanking residues are separated by periods from the identified peptides. bETD is electron-transfer dissociation, HCD is high-energy collisional dissociation, and EThcD is a combination of the two. View Large O-glycosylation protects QSOX1 from proteolytic degradation but is not required for sulfhydryl oxidase activity Based on previous observations that QSOX1 produced in bacteria was sensitive to proteolysis at the intermodule linker (Alon et al. 2010), we assessed the effect of linker PTMs in a limited proteolysis assay. The O1 mutant and wild-type QSOX1, both expressed in HEK cells, were compared. Serial dilutions of trypsin were applied to aliquots of the two proteins, and cleavage was analyzed by SDS polyacrylamide gel electrophoresis (SDS-PAGE) (Figure 4). The O1 mutant was found to be more sensitive to trypsin cleavage in the linker region, despite the fact that no new trypsin recognition sites were introduced, and overall the amino acid changes were minimal in terms of size and polarity (Figure 2A). This observation implies that glycosylation in the QSOX1 linker region affects accessibility of the polypeptide backbone to protease. Fig. 4. View largeDownload slide Glycan modification of the QSOX1 linker protects the enzyme from proteolysis. Wild-type QSOX1 and the O1 mutant were incubated with serial dilutions of trypsin, and the products of the limited proteolysis were separated on a 15% acrylamide SDS gel and stained with Coomassie. The positions of molecular weight markers are shown and labeled in kilodaltons. The bands indicated by yellow arrowheads are the products of QSOX1 proteolytic cleavage by trypsin. Chymotrypsin cleavage using the same protein preparations showed similar results, and QSOX1 produced in HEK cells was more resistant than QSOX1 produced in bacteria to cleavage with either trypsin or chymotrypsin (not shown). Fig. 4. View largeDownload slide Glycan modification of the QSOX1 linker protects the enzyme from proteolysis. Wild-type QSOX1 and the O1 mutant were incubated with serial dilutions of trypsin, and the products of the limited proteolysis were separated on a 15% acrylamide SDS gel and stained with Coomassie. The positions of molecular weight markers are shown and labeled in kilodaltons. The bands indicated by yellow arrowheads are the products of QSOX1 proteolytic cleavage by trypsin. Chymotrypsin cleavage using the same protein preparations showed similar results, and QSOX1 produced in HEK cells was more resistant than QSOX1 produced in bacteria to cleavage with either trypsin or chymotrypsin (not shown). Wild-type, N1, O1 and a S426D mutant of QSOX1 were also subjected to activity assays monitoring oxygen consumption upon oxidation of the model substrate dithiothreitol (DTT) as previously described (Grossman et al. 2013). The serine mutant was not expected to affect activity and was included in the experiment as an additional positive control. All tested QSOX1 variants showed similar rates of oxygen consumption when supplied with reducing substrate (Figure 5); the small variation was within the range of error for enzyme concentration determination. Thus, these PTMs do not affect steady-state turnover rates of QSOX1 in vitro. Fig. 5. View largeDownload slide QSOX1 glycosylation mutants are catalytically active on model substrates. Oxygen consumption during DTT oxidation was monitored as a measure of QSOX1 sulfhydryl oxidase activity. Representative reaction progress traces are shown in color (WT, orange; S426D, green; N1, red; O1, blue), and turnover numbers are presented in the bar graph insert as the average and standard deviation of five measurements. DTT was injected at the points indicated by the arrowheads to a final concentration of 1 mM. The inset shows a Coomassie-stained gel of the QSOX1 variants preparations subjected to the activity assays. Molecular weight markers are in kilodaltons. Fig. 5. View largeDownload slide QSOX1 glycosylation mutants are catalytically active on model substrates. Oxygen consumption during DTT oxidation was monitored as a measure of QSOX1 sulfhydryl oxidase activity. Representative reaction progress traces are shown in color (WT, orange; S426D, green; N1, red; O1, blue), and turnover numbers are presented in the bar graph insert as the average and standard deviation of five measurements. DTT was injected at the points indicated by the arrowheads to a final concentration of 1 mM. The inset shows a Coomassie-stained gel of the QSOX1 variants preparations subjected to the activity assays. Molecular weight markers are in kilodaltons. Conserved N-linked glycosylation is essential for Golgi export After analyzing the secreted QSOX1 variants that were successfully obtained, we questioned whether the lack of secretion of the N2 mutant was due to a trafficking defect. To obtain high quality spatial differentiation of organelles, intracellular staining of QSOX1 was conducted in HeLa cells, which are larger and more spread out than HEK cells. As observed for HEK cells, secretion of N2 from HeLa cells was minimal, whereas the other QSOX1 variants could be detected in cell culture supernatants (Figure 6A). Intracellular staining of QSOX1 was stronger in cells transfected with each of the QSOX1 variants than in mock transfected cells (Figure 6B and C), indicating that the expression from plasmids was greater than the production of endogenous, wild-type QSOX1. Remarkably, immunofluorescence staining of QSOX1 and organelle markers in HeLa cells indicated Golgi localization for all variants, including N2. Furthermore, Golgi staining of N2 was observed at intensities comparable to the other mutants (Figure 6B), and total intracellular QSOX1 staining was quantitatively similar (Figure 6C). To test whether the lack of secretion of the N2 mutant was specific to cell lines that do not normally secrete significant amounts of endogenous QSOX1, we transfected the various QSOX1 constructs into fibroblast cells, known for their secretion of QSOX1 during quiescence (Coppock et al. 2000; Ilani et al. 2013). Similarly to the observations described above, all QSOX1 variants except for the N2 mutant were found in the fibroblast culture supernatants (Figure 6D). In this experiment, mock transfected fibroblasts appeared to secrete more of the endogenous protein than cells transfected with the various short QSOX1 constructs, probably because the mock transfected cells were more confluent by the end of the experiment than the cells transfected with QSOX1 constructs and had dedicated no cellular resources to production of plasmid-encoded protein. Fig. 6. View largeDownload slide A highly conserved N-linked glycan is essential for QSOX1 trafficking. (A) Plasmids encoding wild-type, the N1, N2 and O1 QSOX1 mutants, or an empty vector (Mock) were transiently transfected into HeLa cells. The culture supernatant was subjected to TCA precipitation before it was analyzed by western blot using QSOX1 antibody. (B) The transfected cells were fixed and immunofluorescently stained with ER-specific (PDI) or Golgi-specific (GM130) antibodies (red) and QSOX1 antibody (green). (C) Quantification of per-cell QSOX1 fluorescence, corresponding to the experiment shown in panel B. Error bars indicate one standard deviation. (D) The same plasmids as in panel A were transfected into WI-38 fibroblast cells, and culture supernatants were TCA-precipitated and analyzed by western blot using QSOX1 antibody. Asterisks denote the endogenous protein. Arrows indicate the protein produced from the transfected plasmid. Fig. 6. View largeDownload slide A highly conserved N-linked glycan is essential for QSOX1 trafficking. (A) Plasmids encoding wild-type, the N1, N2 and O1 QSOX1 mutants, or an empty vector (Mock) were transiently transfected into HeLa cells. The culture supernatant was subjected to TCA precipitation before it was analyzed by western blot using QSOX1 antibody. (B) The transfected cells were fixed and immunofluorescently stained with ER-specific (PDI) or Golgi-specific (GM130) antibodies (red) and QSOX1 antibody (green). (C) Quantification of per-cell QSOX1 fluorescence, corresponding to the experiment shown in panel B. Error bars indicate one standard deviation. (D) The same plasmids as in panel A were transfected into WI-38 fibroblast cells, and culture supernatants were TCA-precipitated and analyzed by western blot using QSOX1 antibody. Asterisks denote the endogenous protein. Arrows indicate the protein produced from the transfected plasmid. All experiments described above were performed using a QSOX1 expression construct lacking the transmembrane region and thus resembling the shorter of the two QSOX1 splice variants. To determine whether the conserved N2 glycosylation site is important to the secretion of luminal QSOX1 only, HeLa cells were transfected with QSOX1 constructs that include the transmembrane region and correspond to the longer of the splice variants. The presence of the transmembrane region and short cytosolic tail of QSOX1 did not alter the secretion patterns observed: the QSOX1 N2 mutant containing the transmembrane region was poorly secreted, as seen for its luminal counterpart (Figure 7). Fig. 7. View largeDownload slide The transmembrane region of QSOX1 does not alter N2 secretion. Plasmids encoding an empty vector (Mock), QSOX1 constructs that include its transmembrane region (Full length): WT, N1 and N2 QSOX1 mutants, or QSOX1 lacking its transmembrane region (Short) were transiently transfected into HeLa cells. The culture supernatants were precipitated using TCA, separated on SDS-page, and analyzed by western blot using QSOX1 antibody. The upper arrow indicates the longer form of QSOX1 after processing for secretion, the lower arrow indicates the position of the shorter QSOX1 variant. This experiment was performed once and showed consistency with Figure 6A. Fig. 7. View largeDownload slide The transmembrane region of QSOX1 does not alter N2 secretion. Plasmids encoding an empty vector (Mock), QSOX1 constructs that include its transmembrane region (Full length): WT, N1 and N2 QSOX1 mutants, or QSOX1 lacking its transmembrane region (Short) were transiently transfected into HeLa cells. The culture supernatants were precipitated using TCA, separated on SDS-page, and analyzed by western blot using QSOX1 antibody. The upper arrow indicates the longer form of QSOX1 after processing for secretion, the lower arrow indicates the position of the shorter QSOX1 variant. This experiment was performed once and showed consistency with Figure 6A. Intracellular QSOX1 N2 mutant is catalytically active and shows Golgi glycan modifications Proteins missing essential PTMs may be retained in the cell due to folding defects. To test whether the QSOX1 N2 mutant retained in cells is misfolded, we transfected large-scale (0.5–1 L) suspension-adapted HEK cell cultures with plasmids encoding the QSOX1 variants, purified intracellular enzyme on the basis of the polyhistidine tag, and tested activity. Due to the lesser amounts of enzyme that could be obtained from cell lysates compared to supernatants, a discontinuous colorimetric assay was used instead of the continuous oxygen consumption assay shown in Figure 5. The colorimetric assay showed qualitatively that the N2 mutant purified from cells was active, as was wild-type QSOX1 purified in parallel (Figure 8A). The dependence of the observed activity on the transfected QSOX1 constructs was demonstrated in two ways. First, comparable fractions collected from affinity chromatography of lysates from mock transfected cells showed no sulfhydryl oxidase activity. Second, the activity observed for wild-type and N2 QSOX1 purified from cell lysates was completely eliminated upon addition of a QSOX1 inhibitory antibody (Grossman et al. 2013). This experiment demonstrates that the N2 mutant protein present in cells is folded and functional. The trace amount of N2 mutant secreted into the medium was also purified separately, concentrated, and found to be catalytically active and thermally stable within 2°C of wild-type QSOX1 and the N1 variant in a SYPRO Orange thermal shift assay (transition temperatures for wild type, N1 and N2 were 66.4, 65.9 and 64.4°C, respectively). Fig. 8. View largeDownload slide N2 purified from cell lysates is active and shows Golgi processing of the remaining N-glycan. (A) N2 and wild-type QSOX1 purified from HEK 293F cells were assayed for sulfhydryl oxidase activity. Both wild-type and N2 showed robust oxidation of the model substrate DTT. DTT depletion was quantified by addition of DTNB, the reduced version of which shows increased absorbance at 412 nm. The sulfhydryl oxidase activity was completely inhibited by QSOX1 inhibitory antibodies (Ab), confirming that the activities of both the wild-type and N2 QSOX1 preparations were due to QSOX1 and not to a contaminant. Furthermore, mock transfected cells subjected to the same purification procedure (Mock) showed no activity in the same chromatographic elution fractions. Controls (cont) lacked enzyme. Error bars are standard deviations. (B) The N2 QSOX1 variant purified from cells was subjected to mass spectrometric analysis of peptides and glycans, and the spectrum of a peptide containing the N1 glycan is shown. The annotation diagram was made according to the best match in the UniCarbKB database (Campbell et al. 2011). “M” indicates the mass of the parent peptide, and “H” is a hydrogen in this context. A minus symbol shows that the indicated group has been lost from the ion. Fig. 8. View largeDownload slide N2 purified from cell lysates is active and shows Golgi processing of the remaining N-glycan. (A) N2 and wild-type QSOX1 purified from HEK 293F cells were assayed for sulfhydryl oxidase activity. Both wild-type and N2 showed robust oxidation of the model substrate DTT. DTT depletion was quantified by addition of DTNB, the reduced version of which shows increased absorbance at 412 nm. The sulfhydryl oxidase activity was completely inhibited by QSOX1 inhibitory antibodies (Ab), confirming that the activities of both the wild-type and N2 QSOX1 preparations were due to QSOX1 and not to a contaminant. Furthermore, mock transfected cells subjected to the same purification procedure (Mock) showed no activity in the same chromatographic elution fractions. Controls (cont) lacked enzyme. Error bars are standard deviations. (B) The N2 QSOX1 variant purified from cells was subjected to mass spectrometric analysis of peptides and glycans, and the spectrum of a peptide containing the N1 glycan is shown. The annotation diagram was made according to the best match in the UniCarbKB database (Campbell et al. 2011). “M” indicates the mass of the parent peptide, and “H” is a hydrogen in this context. A minus symbol shows that the indicated group has been lost from the ion. The purification of the N2 mutant from cells allowed us to examine its PTMs as well as its activity. MS analysis of the remaining (N1) glycosite of the N2 mutant revealed Golgi modifications of the glycan at the N1 position, including mannosidase trimming and GlcNAc transferase additions (Figure 8B). Fucosylation was also detected. These MS results support the immunofluorescence staining (Figure 6) showing colocalization with Golgi markers of all tested QSOX1 variants, including the N2 mutant. Discussion Inspection of the sequence context of N2 in human QSOX1 reveals that it corresponds to a structural motif termed the “enhanced aromatic sequon”, which has been recognized to increase the stability of proteins when glycosylated (Culyba et al. 2011) and to promote the efficiency and homogeneity of glycosylation (Murray et al. 2015). Notably, a glycan at the N2 position is not essential for QSOX1 protein folding, since QSOX1 produced in bacteria and thus lacking all glycans, including this one, is well-folded and functional (Alon et al. 2012; Ilani et al. 2013). Furthermore, the N2 mutant isolated from supernatants of large-scale mammalian cell cultures or from the intracellular organelle fraction was stable and showed sulfhydryl oxidase activity, demonstrating that a glycan at this position in QSOX1 does not contribute critically to the structural or functional integrity of the enzyme. However, the QSOX1 N2 site is clearly glycosylated more consistently than is the N1 site, which deviates in its sequence and structural context from the enhanced aromatic sequon consensus (Figure 9). The main feature of the enhanced aromatic sequon is the presence of an aromatic amino acid, typically phenylalanine, two residues before the glycosylated asparagine in a hairpin turn connecting two β-strands in a β-sheet. The phenylalanine two residues before the asparagine in the N2 site of QSOX1 is part of a cluster of partially surface-exposed hydrophobic and aromatic amino acids on a poorly hydrated face of the protein (Figure 9). In addition, on the other side of the β-sheet from the phenylalanine, a hydrophobic groove is found on the domain surface. A glycan at N2 is expected to at least partially obscure these hydrophobic surfaces. The glycan itself, perhaps in the context of surrounding surface features of the protein, may serve as a handle for trafficking out of the Golgi for productive secretion. Another hypothesis is that the extensive hydrophobic surfaces exposed in the absence of the glycan in the N2 mutant are recognized as a signal for Golgi retention and eventual lysosomal degradation rather than secretion. Comparison of structural data with cell biological observations of glycan modification and trafficking will continue to shed light on the physical features of proteins that enhance glycosylation, as well as on the function of glycans once they have been added in particular structural contexts. Fig. 9. View largeDownload slide The N2 region contains exposed hydrophobic and aromatic residues. Structures displayed are from the amino-terminal fragment of human QSOX1 (PDB ID 3Q6O). In the top panels, the asparagines of the N1 and N2 sites are shown in space-filling representation. Cyan spheres representing water molecules bound to the protein in the crystal structure show that the region around the N1 site is more hydrated than the region around N2. Exposed hydrophobic and aromatic side chains, which are much more prevalent near N2, are shown as yellow sticks. The region labeled with a circled red “1” constitutes a cluster of surface-exposed hydrophobic side chains. The region labeled with a circled red “2” is a hydrophobic groove on the surface of the protein. The bottom panels show the secondary structure contexts of the two glycosylation sites. A five-residue region spanning each asparagine is colored lime green, and the side chains near the asparagines are in stick format. Fig. 9. View largeDownload slide The N2 region contains exposed hydrophobic and aromatic residues. Structures displayed are from the amino-terminal fragment of human QSOX1 (PDB ID 3Q6O). In the top panels, the asparagines of the N1 and N2 sites are shown in space-filling representation. Cyan spheres representing water molecules bound to the protein in the crystal structure show that the region around the N1 site is more hydrated than the region around N2. Exposed hydrophobic and aromatic side chains, which are much more prevalent near N2, are shown as yellow sticks. The region labeled with a circled red “1” constitutes a cluster of surface-exposed hydrophobic side chains. The region labeled with a circled red “2” is a hydrophobic groove on the surface of the protein. The bottom panels show the secondary structure contexts of the two glycosylation sites. A five-residue region spanning each asparagine is colored lime green, and the side chains near the asparagines are in stick format. QSOX1 belongs to a set of enzymes that introduce cross-links into ECM proteins. Other enzymes with ECM cross-linking functions are lysyl oxidase and its paralogs (Robins 2007), peroxidasin (Bhave et al. 2012), and transglutaminases (Aeschlimann and Thomazy 2000). Interestingly, these other enzymes, with the possible exception of a transglutaminase associated with the male reproductive tract (Dubbink et al. 1998; Cho et al. 2010), are not reported to be Golgi localized. Of the secreted cross-linking enzymes that have been studied, only QSOX1 appears to be concentrated in the Golgi in its intracellular form. This observation suggests a different mechanism for localization and secretion of QSOX1 compared with the other enzymes. Secreted QSOX1 is found in diverse biological contexts. The enzyme is localized to the Golgi apparatus of many cell types, but it is upregulated and secreted from confluent cultured fibroblasts. QSOX1 is found in blood serum, milk and other biological fluids (Janolino and Swaisgood 1975; Ostrowski et al. 1979; Ostrowski and Kistler 1980; Hoober et al. 1996; Israel et al. 2014). It was originally purified from semen and chicken eggs (Ostrowski and Kistler 1980; Hoober et al. 1996). Higher levels of QSOX1 or its degradation products were found in serum of pancreatic cancer patients compared to healthy controls (Antwi et al. 2009; Pan et al. 2011), suggesting that greater amounts of the protein may be secreted in the patients, though this possibility remains to be investigated. These observations call for a better understanding of QSOX1 secretion, about which little is known. A report by Bullied and colleagues described the proteolytic cleavage of Golgi-localized QSOX1 and the secretion of the enzyme into the culture supernatant (Rudolf et al. 2013). That study also noted that QSOX1 is modified post-translationally by glycosylation, but the link between its modification and trafficking was not investigated. Prior to analyzing the effects of QSOX1 PTMs on secretion of the enzyme, we made a detailed catalog of these modifications. On the basis of the QSOX1 amino acid sequence, potential sites of N-linked glycosylation, i.e. N-X-S/T motifs, were identified at positions 130, 243, 575 and 591. However, only the first two sites, corresponding to N1 and N2, were predicted using NetNGlyc (Gupta and Brunak 2002) to be modified. These two sites are evolutionarily conserved, with N2 showing stronger conservation. The other two N-X-S/T motifs, at positions 575 and 591, are located outside the region spanned by the four-domain QSOX1 construct used for much of the current study. Since the secretion pattern of mutants based on the full-length QSOX1 construct showed similar behavior as the truncated version (Figure 7), sites 575 and 591 were not further investigated. Regarding O-linked glycans, the residues mutated in the O1 quadruple mutant were among those predicted using NetOGlyc to be modified and found experimentally to be glycosites (Steentoft et al. 2013). To validate these putative sites of PTMs in our system, we first compared QSOX1 expressed in bacteria to QSOX1 expressed in cultured mammalian cells. MS analysis showed that the peptides containing the N2 and O1 sites were detected in the protein produced in bacteria but not that produced in mammalian cells, suggesting that they are indeed modified in the latter. The peptide containing N1 was detected in both QSOX1 versions, indicating that at least a fraction of the protein molecules remains unmodified at this position. Gel migration analysis of QSOX1 mutants produced in mammalian cells was used to further analyze PTMs. Whereas wild-type QSOX1 and the O1 mutant appeared as double bands on SDS-PAGE gels, the N1 mutant was observed as a single band. Partial utilization of N1 could explain both the migration pattern and the detection by MS of unmodified peptides spanning this site. Peptides containing an unmodified N1 site were also observed in the MS analysis of QSOX1 secreted from quiescent fibroblasts (Ilani et al. 2013). According to band intensities on gels, roughly half of the QSOX1 protein molecules contain a glycan at the N1 position, whether produced in fibroblasts or other cells (Figures 2, 4 and 6D). It remains to be determined whether a physical difference distinguishes QSOX1 molecules during their maturation in the ER so that only half of them are modified at N1, or if inherent inefficiencies in oligosaccharyltransferase modification of nonoptimal sequons are responsible. Moreover, the physiological importance of the partial utilization of this site and the role it plays in the function of QSOX1 remain to be revealed. One function identified for QSOX1 PTMs in this study is protection against proteases. QSOX1 undergoes large conformational changes during its catalytic cycle, facilitated by a flexible linker between its two redox-active modules (Alon et al. 2012). Avian QSOX1 was cleaved into two fragments at this linker, enabling the study of the properties of each fragment (Raje and Thorpe 2003), and we have previously observed that mammalian QSOX1 produced in bacteria is also sensitive to degradation at the flexible linker. In the current study, the O1 mutant proved to be more sensitive than wild-type QSOX1 to proteolytic cleavage when both enzyme variants were produced in mammalian cells, suggesting that one purpose of the O-linked glycans in the QSOX1 linker is to protect this vulnerable region of the protein. A threonine-rich linker is a conserved feature of mammalian QSOX1 enzymes, whereas avian, amphibian and fish QSOX1 enzymes often contain an N-linked glycosylation consensus site in this region instead. Glycosylation may be a general mechanism to shield the polypeptide backbone of the linker from proteases, while retaining its flexibility. This concept was discussed by Steentoft and colleagues, as linker domains of various proteins were found to be enriched with O-linked glycosylation (Steentoft et al. 2013). De-glycosylation of the QSOX1 linker region by mutagenesis does not interfere with enzymatic activity, however, since both the N1 and O1 mutants showed similar activity as wild-type QSOX1 on a model dithiol substrate (Figure 5). This observation was not surprising, since QSOX1 expressed in bacteria and lacking all glycosylation is functional in in vitro sulfhydryl oxidase assays (Grossman et al. 2013; Ilani et al. 2013) and rescues the ECM-related phenotypes caused by the knockdown of QSOX1 in cell culture (Ilani et al. 2013). Whereas trafficking and secretion were robust to mutation of the N1 and O1 sites, we found that mutation of the N2 site prevented QSOX1 from being secreted from cells (Figures 6 and 7). Even fibroblast cells, which naturally secrete large quantities of QSOX1 (Coppock et al. 2000; Ilani et al. 2013), did not secrete the N2 mutant while still secreting the endogenous wild-type protein (Figure 6D). Upon investigation of the defect in N2, we expected to find this mutant retained in or retrotranslocated from the ER for failing to pass ER quality control. However, as shown in Figure 6B and C, the N2 mutant was observed in the Golgi apparatus instead, indicating that at least a substantial fraction had successfully exited the ER. Purification of the QSOX1 N2 mutant from cells allowed us to test sulfhydryl oxidase activity, revealing that this enzyme variant was as active as wild type and thus is folded and loaded with its FAD cofactor. A similar observation of trafficking defects was made for a glycan variant of ADAM8, a transmembrane protein normally expressed on the cell surface that is retained in the Golgi when one of its N-linked glycosylation sites is mutated (Srinivasan et al. 2014). A surprising finding from this study is that QSOX1 variants with and without a transmembrane anchor and cytosolic tail are trafficked similarly. Despite the importance of transmembrane regions and linear cytosolic motifs in selection of proteins for Golgi residence or transport out of the compartment (Guo et al. 2014), these features do not appear to determine whether secretion of QSOX1 will occur. QSOX1 constructs comprising only the catalytic domains were secreted from cells similarly to QSOX1 constructs corresponding to the longer QSOX1 splice variant with the transmembrane anchor, and secretion of both versions was sensitive to the presence of the N2 glycosylation site. The similar behavior observed for the long and short QSOX1 constructs may reflect the possibility that packaging for secretion occurs subsequent to the proteolytic cleavage that releases the transmembrane region, such that both constructs lack transmembrane and cytosolic signals at this stage. It should be noted, though, that they still differ after cleavage by about 125 residues at the carboxy-terminus, including two putative N-linked glycosylation sites. In any event, our findings place emphasis on Golgi luminal features for selection of QSOX1 for secretion. While the identity of these features and how they are recognized by the Golgi-localized transport machinery are still unclear, the importance of a conserved N-linked glycan for QSOX1 secretion is evident from our results. Quality control in the late secretory pathway is an emerging field of research, and the molecular mechanisms involved are beginning to be addressed. The link we have uncovered between a PTM site and the ability of QSOX1 to be secreted suggests that this enzyme may provide a useful system with which to further explore quality assessment and trafficking control in the Golgi apparatus. Materials and methods Cell lines and maintenance WI-38 fibroblasts were purchased from Coriell and were maintained in Minimal Essential Medium (MEM) supplemented with 10% fetal bovine serum (FBS), l-glutamine, antibiotics, nonessential amino acids and sodium pyruvate. HEK 293T and HeLa cells were obtained from the laboratories of Dr. Ron Diskin and Prof. Moshe Oren, respectively, and maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS, l-glutamine, antibiotics, nonessential amino acids and sodium pyruvate. Cells were checked for mycoplasma approximately once a month. Expression of QSOX1 and mutants The pcDNA3.1 plasmids expressing human QSOX1 and its variants were transfected into adherent HEK 293T, HeLa or WI-38 cells using jetPEI transfection reagent (Polyplus Transfections) according to manufacturer’s instructions. Suspension HEK 293F cells (Thermo Fisher) were cultured in FreeStyle 293 expression medium and transfected using the PEI Max reagent (Polysciences Inc.). For purification of QSOX1 and QSOX1 mutants, the portion of the enzyme spanning residues 1–546, with six histidines appended to the carboxy terminus, was produced by transient transfection of HEK 293T cells or HEK 293F cells. HEK 293T cells were transferred after transfection to serum-free medium. Four days after transfection of adherent cells, culture supernatant containing secreted QSOX1 (which consisted, after signal sequence cleavage, of residues 30–546 and the His6 tag) was dialyzed against phosphate buffered saline to lower the glutamine concentration. HEK 293F cells were harvested 2 days post transfection by centrifuging at 1000 rpm for 10 min, and organelle proteins were isolated (Holden and Horton 2009). QSOX1 was then purified by nickel–nitrilotriacetic acid chromatography, aliquoted and stored at −80°C after addition of glycerol to 10%. Concentrations of QSOX1 variant stocks produced from cell culture supernatants were determined by absorbance at 280 nm using an extinction coefficient of 112,000 M−1 cm−1, which includes the contribution of the bound FAD cofactor. Relative concentrations of enzyme purified from HEK 293F lysates were estimated based on western blotting, since absorbance readings were unreliable due to low protein concentrations. For western blot analyses shown in the figures, culture supernatants from transient transfections were subjected to trichloroacetic acid (TCA) precipitation and cold acetone washes before SDS-PAGE. QSOX1 was expressed in bacteria according to published protocol (Alon et al. 2012). Mass spectrometry Each MS experiment was conducted on one sample. The preliminary peptide fingerprinting analysis of QSOX1 produced in bacteria vs. HEK cells was performed in the Mass Spectrometry Unit of the Weizmann Institute Life Sciences Core Facilities according to published procedures (Ilani et al. 2013). For analysis of intact QSOX1, the samples were acidified to a final concentration of 1% formic acid and desalted on a 4.6 × 50 mm2 ProSwift RP-2S column using 1 mL/min flow on a Waters Acquity UPLC. Samples were loaded on the column on 20% B (80% acetonitrile (MeCN), 0.1% formic acid) for 5 min, followed by elution at 80% B for 5 min. Samples were dried using a speed-vac and analyzed using a nanoAquity nUPLC (Waters) coupled to a Tribrid Orbitrap Fusion Lumos instrument (Thermo Fisher). Sample was loaded and desalted on column using a 0.1 mm×50 cm ProSwift column (Thermo) for trapping and separation. Separation was conducted using a gradient of 2–50% B (80% MeCN, 0.1% formic acid) over 50 min. Sample was analyzed using the following parameters: spray voltage 1.8 kV, ion transfer tube temperature 275°C, and data were acquired alternating between orbitrap and ion trap detection. Oribtrap scans were performed at a range of 700–1500 m/z in 15,000 resolution (@ 200 m/z), 70% RF, in source dissociation of 35 eV, AGC target of 4e5 in maximum of 50 ms, summing up 20 microscans. Ion trap scans were done with the same parameters except using enhanced mode and AGC target of 3e4 with maximum injection time of 35 ms. For detailed glycopeptide analysis, protein samples were loaded onto 3 kDa molecular weight cut-off spin columns. Volume was reduced to 25 μL by centrifugation at 14,000 ×g for 10 min. The 175 μL 8 M urea was added, and the sample was centrifuged at 14,000× g for 10 min. Filters were reversed and centrifuged to extract the proteins. Proteins were reduced with 5 mM DTT for 1 h at room temperature and then alkylated with 10 mM iodoacetamide (Sigma) in the dark for 45 min at room temperature. Samples were diluted to 2 M urea with 50 mM ammonium bicarbonate. Proteins were then subjected to digestion with trypsin (Promega; Madison, WI, USA) overnight at 37°C at 50:1 protein:trypsin ratio, followed by a second trypsin digestion for 4 h. The digestions were stopped by addition of trifluroacetic acid (1% final concentration). Following digestion, peptides were desalted using Oasis HLB, μElution format (Waters, Milford, MA, USA). The samples were vacuum dried and stored in −80°C until further analysis. Samples were analyzed using the nanoAquity nUPLC coupled to the Tribrid Orbitrap Fusion Lumos instrument. Samples were loaded on a 0.18×20 mm2, 5 μm C18 Symmetry (Waters) column at 10 μL/min for 1 min in 2% B. Sample was then separated on a 0.075×250 mm2, 1.8 μm HSS T3 C18 column, using a gradient of 4–30% B over 50 min. Mass spectrometry was performed using the following parameters: spray voltage 1.6 kV, ion transfer tube temperature 275°C. MS1 was acquired at 120,000 resolution (@ 200 m/z) at a range of 300–1800 m/z, AGC target 4e5 with maximum of 50 ms inject time. MS2 was set to 3 s cycle time, with MIPS on, selecting precursors of intensity higher than 5e4, with charges between +2 and +8. Each precursor was selected for fragmentation once, and then excluded for 30 s. Precursors were isolated using a 1 m/z window and AGC target 5e4 with maximum inject time of 100 ms, fragmented using either HCD set to 30NCE, EThcD with ETD reaction time of 60 ms and 3e5 reagent target ion and HCD set to 15NCE or EThcD with ETD reaction time of 60 ms and 3e5 reagent target ion and HCD set to 22NCE. MS2 acquisition was set at a fixed first mass of 130 m/z, acquired at 15,000 resolution (@ 200 m/z). Data were analyzed using Byonic search engine (Protein Metrics Inc.), against the human proteome database. The search was performed iteratively, first constructing a limited focused database, then searching common modifications such as oxidations and deamidations, then searching against the Human O- and N-glycosylation libraries provided by Byonic. Western blot and lectin staining Samples were separated on SDS-PAGE and transferred to a nitrocellulose membrane. After protein transfer, the membrane was blocked with 5% bovine serum albumin (BSA) in phosphate buffered saline (PBS) containing 0.1% Tween (PBS-T). Following blocking, the membrane was blotted using a polyclonal rabbit anti human QSOX1 antibody (Ilani et al. 2013) with 5% BSA in PBS-T overnight at 4°C. The membrane was then washed three times in PBS-T and blotted with a secondary antibody conjugated to horseradish peroxidase (HRP) (Abcam). Following washes with PBS-T, HRP reaction was carried out using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) and visualized using a ChemiDoc XRS+ System (Bio-Rad). For lectin staining, a similar procedure was followed except that biotin-conjugated VVL (Vector Laboratories) was incubated with the membrane for 1 h at room temperature, and streptavidin-conjugated HRP was used instead of secondary antibody. Immunofluorescence Cells were grown on coverslips in a 24-well plate, fixed for 20 min using 3.7% formaldehyde, and permeabilized in 0.1% Triton X-100 in PBS for 5 min. Cells were then blocked with 5% BSA in PBS-T and incubated overnight at 4°C with primary antibody in 5% BSA in PBS-T. Staining was performed using polyclonal rabbit anti human QSOX1 and either mouse anti GM130 (Abcam) or mouse anti PDI (Abcam). The cells were then washed three times with PBS-T and incubated with a fluorescently labeled secondary antibody before being placed face down onto 5 μL Fluoroshield antifade reagent (Sigma) on coverslips. Samples were observed on an Olympus CKX31SF2 imaging system and analyzed using Fiji software. Per-cell total fluorescence for QSOX1 immuno-staining was measured by integrating the intensity within boxes of uniform size positioned to enclose entire single cells. At least 40 cells were quantified for each QSOX1 variant. These cells were selected randomly from imaged fields, except that cells in close proximity to a neighbor such that they could not be boxed individually were avoided. Limited proteolysis Trypsin and was dissolved at a concentration of 1 mg/mL in 20 mM Tris, pH 8.0, 200 mM NaCl. Four-fold serial dilutions were made into the same buffer. One microliter of protease was added to 7 μL of QSOX1 (wild-type or O1 mutant) from a 3.8 μM stock in PBS plus 10% glycerol. Reactions were incubated at RT for 30 min. Oxygen consumption assays QSOX1 variants were assayed at a concentration of 100 nM with 1 mM DTT in a Clarke-type oxygen electrode (Hansatech Instruments) at 25°C. Reactions were initiated by DTT injection. The oxygen electrode buffer (OEB) was 50 mM potassium phosphate buffer, pH 7.5, 65 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA). Colorimetric sulfhydryl oxidase assay Intracellular QSOX1 was incubated in OEB with 200 μM DTT with or without 250 nM of a QSOX1 inhibitory antibody (Grossman et al. 2013). At each time point, an aliquot of the reaction was diluted 1:10 into 500 μM 5,5′-dithio-bis-[2-nitrobenzoic acid] (DTNB) in OEB, and, after 5 min incubation, absorbance was recorded at 412 nm in an Infinite 200 Pro plate reader (TECAN). The assay was done in triplicate. SYPRO Orange thermal shift assay QSOX1 and glycan variants at concentrations of 1–7 μM were incubated with a 1:250 dilution of SYPRO Orange 5000× stock (Thermofisher) in a 96-well PCR plate in a volume of 20 μL per well, and fluorescence data were collected on an Applied Biosystems Real-Time PCR System using the fluorescein amidite channel. The temperature was held for 1 min per 0.5° from 24 to 95°C. Funding This work was supported by the European Research Council (ERC) under the European Union’s Seventh Framework Programme [Grant number 310649]. Acknowledgements The authors thank Tevie Mehlman and Dalit Merhav for mass spectrometry protein fingerprinting. Conflict of interest statement None declared. Abbreviations BSA bovine serum albumin DTT dithiothreitol ECM extracellular matrix ER endoplasmic reticulum; ETD electron-transfer dissociation EThcD electron-transfer/higher-energy collision dissociation FAD flavin adenine dinucleotide GalNAc N-acetylgalactosamine HCD higher-energy collision dissociation HEK human embryonic kidney HRP horseradish peroxidase LC–MS-MS liquid chromatography followed by tandem mass spectrometry MeCN acetonitrile MEM minimal essential medium MS mass spectrometry N1 Asn-to-Gln mutation at position 130 in QSOX1 N2 Asn-to-Gln mutation at position 243 in QSOX1 O1 QSOX1 mutant with the threonines at positions 276, 277, 281 and 282 replaced by other amino acids OEB oxygen electrode buffer PBS phosphate buffered saline PTMs post-translational modifications QSOX1 quiescin sulfhydryl oxidase 1 SDS sodium dodecyl sulfate SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis TCA trichloroacetic acid VVL Vicia villosa lectin References Aeschlimann D, Thomazy V. 2000. Protein crosslinking in assembly and remodeling of extracellular matrices: The role of transglutaminases. Connect Tissue Res . 41: 1– 27. Google Scholar CrossRef Search ADS PubMed  Alon A, Grossman I, Gat Y, Kodali VK, DiMaio F, Mehlman T, Haran G, Baker D, Thorpe C, Fass D. 2012. The dynamic disulphide relay of quiescin sulphydryl oxidase. Nature . 488: 414– 418. Google Scholar CrossRef Search ADS PubMed  Alon A, Hekler EJ, Thorpe C, Fass D. 2010. QSOX contains a pseudo-dimer of functional and degenerate sulfhydryl oxidase domains. FEBS Lett . 584: 1521– 1525. Google Scholar CrossRef Search ADS PubMed  Antwi K, Hostetter G, Demeure MJ, Katchman BA, Decker GA, Ruiz Y, Sielaff TD, Koep LJ, Lake DF. 2009. Analysis of the plasma peptidome from pancreas cancer patients connects a peptide in plasma to overexpression of the parent protein in tumors. J Proteome Res . 8: 4722– 4231. Google Scholar CrossRef Search ADS PubMed  Bauer P, Leshinsky-Silver E, Blumkin L, Schlipf N, Schröder C, Schicks J, Lev D, Riess O, Lerman-Sagie T, Schöls L. 2012. Mutation in the AP4B1 gene cause hereditary spastic paraplegia type 47 (SPG47). Neurogenetics . 13: 73– 76. Google Scholar CrossRef Search ADS PubMed  Bhave G, Cummings CF, Vanacore RM, Kumagai-Cresse C, Ero-Tolliver IA, Rafi M, Kang JS, Pedchenko V, Fessler LI, Fessler JH et al.  . 2012. Peroxidasin forms sulfilimine chemical bonds using hypohalous acids in tissue genesis. Nat Chem Biol . 8: 784– 790. Google Scholar CrossRef Search ADS PubMed  Bonifacino JS. 2004. The GGA proteins: Adaptors on the move. Nat Rev Mol Cell Biol . 5: 23– 32. Google Scholar CrossRef Search ADS PubMed  Briant K, Johnson N, Swanton E. 2017. Transmembrane domain quality control systems operate at the endoplasmic reticulum and Golgi apparatus. PLoS One . 12: e0173924. Google Scholar CrossRef Search ADS PubMed  Campbell MP, Haves CA, Struwe WB, Wilkins MR, Aoki-Kinoshita KF, Harvey DJ, Rudd PM, Kolarich D, Lisacek F, Karsson NG et al.  . 2011. UniCarbKB: Putting the pieces together for glycomics research. Proteomics . 11: 4117– 4121. Google Scholar CrossRef Search ADS PubMed  Canuel M, Korkidakis A, Konnyu K, Morales CR. 2008. Sortilin mediates the lysosomal targeting of cathepsins D and H. Biochem Biophys Res Commun . 373: 292– 297. Google Scholar CrossRef Search ADS PubMed  Castillon GA, Michon L, Watanabe R. 2013. Apical sorting of lysoGPI-anchored proteins occurs independent of association with detergent-resistant membranes but dependent on their N-glycosylation. Mol Biol Cell . 24: 2021– 2033. Google Scholar CrossRef Search ADS PubMed  Chavez CA, Bohnsack RN, Kudo M, Gotschall RR, Canfield WM, Dahms NM. 2007. Domain 5 of the cation-independent mannose 6-phosphate receptor preferentially binds phosphodiesters (mannose 6-phosphate N-acetylglucosamine ester). Biochemistry . 46: 12604– 12617. Google Scholar CrossRef Search ADS PubMed  Chen C, Ma J, Lazic A, Backovic M, Colley KJ. 2000. Formation of insoluble oligomers correlates with ST6Gal I stable localization in the Golgi. J Biol Chem . 275: 13819– 13826. Google Scholar CrossRef Search ADS PubMed  Cho SY, Jeon JH, Kim CW, Shin DM, Jang GY, Jeong EM, Lee SE, Song KY, Kim IG. 2010. Monoclonal antibodies to human transglutaminase 4. Hybridoma . 29: 263– 267. Google Scholar CrossRef Search ADS PubMed  Coppock D, Kopman C, Gudas J, Cina-Poppe DA. 2000. Regulation of the quiescence-induced genes: Quiescin Q6, decorin, and ribosomal protein S29. Biochem Biophys Res Commun . 269: 604– 610. Google Scholar CrossRef Search ADS PubMed  Coppock DL, Kopman C, Scandalis S, Gilleran S. 1993. Preferential gene expression in quiescent human lung fibroblasts. Cell Growth Differ . 4: 483– 493. Google Scholar PubMed  Cordat E, Kittanakom S, Yenchitsomanus PT, Li J, Du K, Lukacs GL, Reithmeier RA. 2006. Dominant and recessive distal renal tubular acidosis mutations of kidney anion exchanger 1 induce distinct trafficking defects in MDCK cells. Traffic . 7: 117– 128. Google Scholar CrossRef Search ADS PubMed  Culyba EK, Price JL, Hanson SR, Dhar A, Wong CH, Gruebele M, Powers ET, Kelly JW. 2011. Protein native-state stabilization by placing aromatic side chains in N-glycosylated reverse turns. Science . 331: 571– 575. Google Scholar CrossRef Search ADS PubMed  Dell’Angelica EC, Shotelersuk V, Aguilar RC, Gahl WA, Bonifacino JS. 1999. Altered trafficking of lysosomal proteins in Hermansky-Pudlak syndrome due to mutations in the beta 3A subunit of the AP-3 adaptor. Mol Cell . 3: 11– 21. Google Scholar CrossRef Search ADS PubMed  Dubbink HJ, de Waal L, van Haperen R, Verkaik NS, Trapman J, Romijn JC. 1998. The human prostate-specific transglutaminase gene (TGM4): Genomic organization, tissue-specific expression, and promoter characterization. Genomics . 51: 434– 444. Google Scholar CrossRef Search ADS PubMed  Grossman I, Alon A, Ilani T, Fass D. 2013. An inhibitory antibody blocks the first step in the dithiol/disulfide relay mechanism of the enzyme QSOX1. J Mol Biol . 425: 4366– 4378. Google Scholar CrossRef Search ADS PubMed  Guo Y, Sirkis DW, Schekman R. 2014. Protein sorting at the trans-Golgi network. Annu Rev Cell Dev Biol . 30: 169– 206. Google Scholar CrossRef Search ADS PubMed  Gupta R, Brunak S. 2002. Prediction of glycosylation across the human proteome and the correlation to protein function. Pac Symp Biocomput . 7: 310– 322. Gut A, Kappeler F, Hyka N, Balda MS, Hauri HP, Matter K. 1998. Carbohydrate-mediated Golgi to cell surface transport and apical targeting of membrane proteins. EMBO J . 17: 1919– 1929. Google Scholar CrossRef Search ADS PubMed  Holden P, Horton WA. 2009. Crude subcellular fractionation of cultured mammalian cell lines. BMC Res Notes . 2: 243. Google Scholar CrossRef Search ADS PubMed  Hoober KL, Joneja B, White HB 3rd, Thorpe C. 1996. A sulfhydryl oxidase from chicken egg white. J Biol Chem . 271: 30510– 30516. Google Scholar CrossRef Search ADS PubMed  Ilani T, Alon A, Grossman I, Horowitz B, Kartvelishvily E, Cohen SR, Fass D. 2013. A secreted disulfide catalyst controls extracellular matrix composition and function. Science . 341: 74– 76. Google Scholar CrossRef Search ADS PubMed  Israel BA, Jiang L, Gannon SA, Thorpe C. 2014. Disulfide bond generation in mammalian blood serum: Detection and purification of quiescin-sulfhydryl oxidase. Free Radic Biol Med . 69: 129– 135. Google Scholar CrossRef Search ADS PubMed  Janolino VG, Swaisgood HE. 1975. Isolation and characterization of sulfhydryl oxidase from bovine milk. J Biol Chem . 250: 2532– 2538. Google Scholar PubMed  Kienzle C, von Blume J. 2014. Secretory cargo sorting at the trans-Golgi network. Trends Cell Biol . 24: 584– 593. Google Scholar CrossRef Search ADS PubMed  Martínez-Maza R, Poyatos I, López-Corcuera B, Núñez E, Giménez C, Zafra F, Aragón C. 2001. The role of N-glycosylation in transport to the plasma membrane and sorting of the neuronal glycine transporter GLYT2. J Biol Chem . 276: 2168– 2173. Google Scholar CrossRef Search ADS PubMed  Munro S. 1995. An investigation of the role of transmembrane domains in Golgi protein retention. EMBO J . 14: 4695– 4704. Google Scholar PubMed  Munro S, Nichols BJ. 1999. The GRIP domain—A novel Golgi-targeting domain found in several coiled-coil proteins. Curr Biol . 9: 377– 380. Google Scholar CrossRef Search ADS PubMed  Murray AN, Chen W, Antonopoulos A, Hanson SR, Wiseman RL, Dell A, Haslam SM, Powers DL, Powers ET, Kelly JW. 2015. Enhanced aromatic sequons increase oligosaccharyltransferase glycosylation efficiency and glycan homogeneity. Chem Biol . 22: 1052– 1062. Google Scholar CrossRef Search ADS PubMed  Ostrowski MC, Kistler WS. 1980. Properties of a flavoprotein sulfhydryl oxidase from rat seminal vesicle secretion. Biochem . 19: 2639– 2645. Google Scholar CrossRef Search ADS   Ostrowski MC, Kistler WS, Williams-Ashman HG. 1979. A flavoprotein responsible for the intense sulfhydryl oxidase activity of rat seminal vesicle secretion. Biochem Biophys Res Commun . 87: 171– 176. Google Scholar CrossRef Search ADS PubMed  Pan S, Chen R, Crispin DA, May D, Stevens T, McIntosh MW, Bronner MP, Ziogas A, Anton-Culver H, Brentnall TA. 2011. Protein alterations associated with pancreatic cancer and chronic pancreatitis found in human plasma using global quantitative proteomics profiling. J Proteome Res . 10: 2359– 2376. Google Scholar CrossRef Search ADS PubMed  Porter A, Yue T, Heeringa L, Day S, Suh E, Haab BB. 2010. A motif-based analysis of glycan array data to determine the specificities of glycan-binding proteins. Glycobiology . 20: 369– 380. Google Scholar CrossRef Search ADS PubMed  Raje S, Thorpe C. 2003. Inter-domain redox communication in flavoenzymes of the quiescin/sulfhydryl oxidase family: Role of a thioredoxin domain in disulfide bond formation. Biochem . 42: 4560– 4568. Google Scholar CrossRef Search ADS   Reczek D, Schwake M, Schröder J, Hughes H, Blanz J, Jin X, Brondyk W, Van Patten S, Edmunds T, Saftig P. 2007. LIMP-2 is a receptor for lysosomal mannose-6-phosphate-independent targeting of beta-glucocerebrosidase. Cell . 131: 770– 783. Google Scholar CrossRef Search ADS PubMed  Robins SP. 2007. Biochemistry and functional significance of collagen cross-linking. Biochem Soc Trans . 35: 849– 852. Google Scholar CrossRef Search ADS PubMed  Rudolf J, Pringle MA, Bulleid NJ. 2013. Proteolytic processing of QSOX1A ensures efficient secretion of a potent disulfide catalyst. Biochem J . 454: 181– 190. Google Scholar CrossRef Search ADS PubMed  Scheiffele P, Peränen J, Simons K. 1995. N-glycans as apical sorting signals in epithelial cells. Nature . 378: 96– 98. Google Scholar CrossRef Search ADS PubMed  Srinivasan S, Romagnoli M, Bohm A, Sonenshein GE. 2014. N-Glycosylation regulates ADAM8 processing and activation. J Biol Chem . 289: 33676– 33688. Google Scholar CrossRef Search ADS PubMed  Steentoft C, Vakhrushev SY, Joshi HJ, Kong Y, Vester-Christensen MB, Schjoldager KT, Lavrsen K, Dabelsteen S, Pedersen NB, Marcos-Silva L et al.  . 2013. Precision mapping of the human O-GalNAc glycoproteome through SimpleCell technology. EMBO J . 32: 1478– 1488. Google Scholar CrossRef Search ADS PubMed  Tagliabracci VS, Wiley SE, Guo X, Kinch LN, Durrant E, Wen J, Xiao J, Cui J, Nguyen KB, Engel JL et al.  . 2015. A single kinase generates the majority of the secreted phosphoproteome. Cell . 161: 1619– 1632. Google Scholar CrossRef Search ADS PubMed  Thorpe C, Hoober KL, Raje S, Glynn NM, Burnside J, Turi GK, Coppock DL. 2002. Sulfhydryl oxidases: Emerging catalysts of protein disulfide bond formation in eukaryotes. Arch Biochem Biophys . 405: 1– 12. Google Scholar CrossRef Search ADS PubMed  Verkerk AJ, Schot R, Dumee B, Schellekens K, Swagemakers S, Bertoli-Avella AM, Lequin MH, Dudink J, Govaert P, van Zwol AL et al.  . 2009. Mutation in the AP4M1 gene provides a model for neuroaxonal injury in cerebral palsy. Am J Hum Genet . 85: 40– 52. Google Scholar CrossRef Search ADS PubMed  von Blume J, Alleaume AM, Kienzle C, Carreras-Sureda A, Valverde M, Malhotra V. 2012. Cab45 is required for Ca(2+)-dependent secretory cargo sorting at the trans-Golgi network. J Cell Biol . 199: 1057– 1066. Google Scholar CrossRef Search ADS PubMed  Yeaman C, Le Gall AH, Baldwin AN, Monlauzeur L, Le Bivac A, Rodriguez-Boulan E. 1997. The O-glycosylated stalk domain is required for apical sorting of neurotrophin receptors in polarized MDCK cells. J Cell Biol . 139: 929– 940. Google Scholar CrossRef Search ADS PubMed  Zhang L, Cheng Q, Zhang L, Wang Y, Merrill GF, Ilani T, Fass D, Arnér ES, Zhang J. 2016. Serum thioredoxin reductase is highly increased in mice with hepatocellular carcinoma and its activity is restrained by several mechanisms. Free Radic Biol Med . 99: 426– 435. Google Scholar CrossRef Search ADS PubMed  Zhang L, Tran DT, Ten Hagen KG. 2010. An O-glycosyltransferase promotes cell adhesion during development by influencing secretion of an extracellular matrix integrin ligand. J Biol Chem . 285: 19491– 19501. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

GlycobiologyOxford University Press

Published: May 11, 2018

There are no references for this article.

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