Molecular modularity and asymmetry of the molluscan mantle revealed by a gene expression atlas

Molecular modularity and asymmetry of the molluscan mantle revealed by a gene expression atlas Background: Conchiferan molluscs construct a biocalcified shell that likely supported much of their evolutionary success. However, beyond broad proteomic and transcriptomic surveys of molluscan shells and the shell-forming mantle tissue, little is known of the spatial and ontogenetic regulation of shell fabrication. In addition, most efforts have been focused on species that deposit nacre, which is at odds with the majority of conchiferan species that fabricate shells using a crossed-lamellar microstructure, sensu lato. Results: By combining proteomic and transcriptomic sequencing with in situ hybridization we have identified a suite of gene products associated with the production of the crossed-lamellar shell in Lymnaea stagnalis. With this spatial expression data we are able to generate novel hypotheses of how the adult mantle tissue coordinates the deposition of the calcified shell. These hypotheses include functional roles for unusual and otherwise difficult-to-study proteins such as those containing repetitive low-complexity domains. The spatial expression readouts of shell-forming genes also reveal cryptic patterns of asymmetry and modularity in the shell-forming cells of larvae and adult mantle tissue. Conclusions: This molecular modularity of the shell-forming mantle tissue hints at intimate associations between structure, function, and evolvability and may provide an elegant explanation for the evolutionary success of the second largest phylum among the Metazoa. Keywords: mollusc; biomineralization; gene expression; asymmetry; modularity; evolution; shell; matrix protein; transcriptome; alternative splicing throughput study of these molecules are well established and Introduction are technically straightforward. Much progress has been made Due to its evolutionary significance, impressive materials prop- in identifying the components of the shell-forming proteome erties, and aesthetic beauty, the molluscan shell has long re- from a variety of gastropod and (primarily) bivalve species (e.g., ceived attention from a wide variety of scientific disciplines [ 1– [10–17]) largely due to advances in nucleic acid sequencing tech- 6]. Although molluscan shells are constructed from a complex nologies that, when coupled with high-throughput proteomic mixture of calcium carbonate, carbohydrates [7, 8], and lipids [9], surveys of the biomineralized proteome, allow for the rapid gen- proteins have received the most attention arguably for two main eration of extensive lists of shell-associated proteins. However, reasons: they can provide deep insight into the evolutionary his- without further validation, genes identified in this way should tory of this composite structure and the techniques for the high- only be considered as candidate biomineralizing molecules. This Received: 27 November 2017; Revised: 20 March 2018; Accepted: 9 May 2018 The Author(s) 2018. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 2 Modularity and asymmetry of the molluscan mantle problem is often compounded by the fact that these proteins of- Methods ten share little to no sequence similarity with proteins from con- Cultivation of adult L. stagnalis ventional model organisms, making any inference about their function very difficult. This bottleneck represents one of the cur- Lymnaea stagnalis (Mollusca; Gastropoda; Heterobranchia; Eu- rent major challenges for scientists interested in understanding thyneura; Panpulmonata; Hygrophila; Lymnaeoidea; Lymnaei- the mechanisms and evolution of molluscan biomineral forma- dae; Lymnaea) does not fall under the German Animal Protection tion. While knock-down of individual shell-forming genes via Act §8 and is listed as “least concern” under the International RNAi has been reported in some species of bivalves [10, 18, 19], Union for Conservation of Nature (IUCN’s) list of threatened these assays are rarely validated by protein immuno-detection, species. This work was therefore exempt from the University and levels of penetrance or statistical quantitation of knock- ¨ of Gottingen Ethics Committee. Adult specimens of L. stagnalis down phenotypes are rarely reported. derived from animals originally collected from the Northeimer ◦ ◦ Another approach to gain insight into the function of shell- Seenplatte, Germany (51 43’ 26.5368’, 9 57’ 24.75’), and from a forming genes is to characterize their spatial expression pat- pond on the north campus of the University of Gotting ¨ en, Ger- ◦   ◦ terns in situ. We previously adopted this approach in the trop- many (51 33 23.727 ,9 57 25.617 ), were kept in a stand-alone ical abalone Haliotis asinina with a Sanger Expressed Sequence V30 unit (Aqua Schwarz) in demineralized water supplemented Tag (EST) dataset and characterized the spatial expression pat- with ReMineral+ (Dennerle, 7036) to a conductivity of 200–220 terns of more than 20 putative shell-forming genes in juvenile μS, 23 C, apHof7.5 to 7.9and a16:8light regimen.Fiveto10 snails [16]. While a spatial expression pattern in the mantle is individuals were kept in 3- or 5-liter boxes under a constant and not direct evidence of a functional role in calcification, we were low-flow rate. Snails were fed ad libitum with lettuce and a va- able to assign putative functions to genes involved in shell pig- riety of other vegetables. Under this regime, adult snails lay egg mentation [16] and ecological and mineralogical transitions [20]. masses year round. Here, we have combined an next-generation sequencing (NGS) transcriptome analysis of adult mantle tissue with a proteomic Organic matrix extraction from calcified shells survey of the adult shell of the freshwater pulmonate gastro- pod Lymnaea stagnalis in order to both compare the resulting Twelve shells of adult L. stagnalis (larger than 3–4 cm in length) data with other similar datasets and to generate the first in situ- were selected for extraction. Prior to further treatment, the col- validated ontogenetic transcriptome-scale dataset for a species umella was delicately cut and removed from each shell. Super- that forms the most common molluscan shell microstructure, ficial organic contaminants were removed by incubating pooled crossed lamellar [21–23]. The high-order structure of crossed- shell fragments in 10% v/v sodium hypochlorite (NaOCl) for 24 lamella, which allows it to efficiently deflect and arrest cracks hours. Fragments were then thoroughly rinsed with water and [24–27], coupled with its extremely low organic content (typically subsequently ground into a fine powder that was sieved ( >200 <0.5%) has been suggested to be one reason it has enjoyed so μM). This biomineral powder was incubated in 5% v/v NaOCl for much evolutionary success (reviewed in [28]). Recent proteomic 5 hours and rinsed twice with MilliQ water. Powdered samples studies have been reported for molluscs that build crossed- were decalcified overnight at 4 C in cold 5% v/v acetic acid that lamellae shells (Helix aspersa maxima [29]and Cepaea nemoralis was slowly added by an automated burette (Titronic Universal, [14]), however, those studies did not conduct any spatial ex- Mainz, Germany) at a flow rate of 100 μLevery 5seconds.The pression analyses for the shell-forming proteins they identified. solution (final pH ∼4.2) was centrifuged at 3,900 g for 30 min- In addition to characterizing the spatial expression patterns of utes. The resulting acid-insoluble matrix (AIM) pellet was rinsed more than 30 shell-forming candidates in the adult mantle tis- six times with MilliQ water, freeze-dried, and weighed. The su- sue of L. stagnalis, we have also investigated their spatial expres- pernatant containing acetic acid-soluble matrix (ASM) was fil- sion patterns during development. tered (Millipore, 5 μM) and concentrated in an Amicon ultra- Our analyses hint at the potential pleiotropic nature of some filtration stirred cell (model 8400, 400 mL) on a Millipore mem- of these shell-forming genes and highlight the dynamic and brane (10 kDa cutoff). The final solution ( >5 mL) was extensively asymmetric natures of their spatial regulation. A striking result dialyzed against 1 L of MilliQ water (six water changes) before of our analyses in the adult mantle tissue is the degree of spa- being freeze-dried and weighed. tial modularity displayed by distinct sets of genes. This general observation may contribute to an explanation of why the mol- Sample preparation for proteomic analysis luscan shell is apparently so evolvable. With the availability of a draft L. stagnalis genome and transcriptome data from a va- In-solution digestion of unfractionated ASM (0.1 mg) and AIM (1 riety of adult tissues, we have also investigated the genetic ar- mg) material was performed as follows. Samples were reduced chitectures of our biomineralization candidates and explored to with 50 μL of 10 mM dithiothreitol in 50 mM ammonium bicar- what extent alternative splicing plays a role in shell formation bonate (NH HCO ) for 30 minutes at 50 C. Alkylation was per- 4 3 in L. stagnalis. These genes can also be compared with similar formed with 50 μL of 100 mM iodoacetamide in 50 mM NH HCO 4 3 datasets from distantly related molluscs that build shells with for 30 minutes at room temperature in the dark. The solution alternative polymorphs of calcium carbonate (calcite vs. arag- was then treated with 1 μg of trypsin (proteomic grade; Promega) onite) and textures (prismatic vs. nacreous vs. crossed lamel- in 10 μLof50mMNH HCO overnight at 37 C. Samples were 4 3 lae). Such comparisons can generate testable hypotheses regard- then dried in a vacuum concentrator and re-suspended in 30 μL ing which components of the shell-forming toolkit contribute to of 0.1% trifluoroacetic acid and 2% acetonitrile (CH CNCN). these differences and which components are required for more fundamental aspects of shell formation. Peptide fractionation and data acquisition Mass spectrometry (MS) was performed using a Q-Star XL nanospray quadrupole/time-of-flight tandem mass spectrom- eter, nanospray-Qq-TOF-MS/MS (Applied Biosystems, Villebon- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 3 sur-Yvette, France) coupled to an online nanoLC system (Ulti- quences were generated using Gene Structure Draw [40]. Intron- mate Famos Switchos from Dionex, Amsterdam, The Nether- exon boundaries were mapped to a draft genome of L. stagnalis lands). One microliter of each sample was loaded onto a trap originally reported in [41] using Splign [42]. Similar transcripts column (PepMap100 C18; 5 μm; 100 A; 300 μM x 5 mm; Dionex), were retrieved from the assembled transcriptomes of mantle washed for 3 minutes at 25 μL/min with 0.05% trifluoroacetic zones 1–5 combined, mantle zone 5 alone, cephalic tentacle, acid/2% acetonitrile, then eluted onto a C18 reverse phase col- cephalic lobe, central nervous system (CNS), foot, buccal mass, umn (PepMap100 C18; 3 μm; 100 A; 75 μM x 150 mm; Dionex). and larval stages 42 hours post first cleavage (hpfc), 52 hpfc, Peptides were separated at a flow rate of 0.300 μL/min with a and 67 hpfc using BLASTn searches (see below for NGS details). linear gradient of 5–80% acetonitrile in 0.1% formic acid over 120 All transcripts with complete open reading frames were consid- minutes. MS data were acquired automatically using ANALYST ered. Only candidates yielding an mRNA coverage of >98% and QS 1.1 software (Applied Biosystems). Following an MS survey an overall identity of >98% are documented. Scaled schematics scan over m/z 400–1600 range, MS/MS spectra were sequentially of the gene architecture were generated using Gene Structure and dynamically acquired for the three most intense ions over Draw [40]. Protein patterns were searched for using a modified m/z 65–2000 range. The collision energy was set by the software local installation of PatMatch [43]. according to the charge and mass of the precursor ion. MS and MS/MS data were recalibrated using internal reference ions from NGS sequencing a trypsin autolysis peptide at m/z 842.51 [M + H] and m/z 421.76 2+ [M + 2H] . Total RNA was extracted from the mantle edge and the proxi- mal mantle tissue of a single adult L. stagnalis using TriReagent following the manufacturer’s instructions. The resulting RNA Mass spectrometry data analysis was processed by the sequencing center at the IKMB at the Protein identification was performed using the MASCOT search University of Kiel (Germany). Paired end, stranded TrueSeq RNA engine (version 2.1; Matrix Science, London, UK) against trans- libraries were constructed and sequenced for 101 bases from lations in all six frames of our mantle transcriptomes, which both ends using the Illumina HiSeq2000 platform (Illumina, possessed Benchmarking Universal Single-Copy Orthologs com- CA, USA). More than 99 million and 100 million reads were pleteness scores of >98% (see below). Liquid chromatography generated from each of these libraries, respectively. These (LC)-MS/MS data were searched using carbamido-methylation Illumina reads were processed using our pipeline as previously as a fixed modification and methionine oxidation as a variable described [44]. Briefly, raw reads were adapter trimmed and modification. The peptide mass and fragment ion tolerances quality filtered using Trimmomatic V0.32. Filtered reads were were set to 0.5 Da. The peptide hits (protein score >50; false then assembled with Trinity V2.0.3, CLC Genomics Workbench discovery rate <0.05; 1 missed cleavage allowed) were manu- de novo assembler (V8.5), and IDBA-tran. The resulting assem- ally confirmed by the observation of the raw LC-MS/MS spectra blies were then merged and filtered for redundancy using our with ANALYST QS software (version 1.1). Quality criteria were pipeline [44]. Mantle transcriptome assemblies and cDNA and the peptide MS value, the assignment of major peaks to un- protein translations of the 34 shell-forming genes are available interrupted y- and b-ion series of at least three to four con- in the GigaScience Database, GigaDB [45]. In addition, tran- secutive amino acids, and the match with the de novo inter- scriptomes from five adult tissues (cephalic tentacle, cephalic pretations proposed by the software. All MS data has been de- lobe, CNS, foot, and buccal mass) and three larval stages (42 posited with the ProteomeXchange Consortium via PRIDE [30] hpfc, 52 hpfc, and 67 hpfc) were sequenced and assembled as with the dataset identifiers PXD008547 and 10.6019/PXD008547. described above. These transcriptomes were used to assess Shell-forming candidates Lstag-sfc-7, Lstag-sfc-8, and Lstag-sfc-9 the tissue-specific alternative splicing characteristics of all were bioinformatically selected for analysis based on the pres- shell-forming genes. All raw NGS data has been deposited in ence of a signal peptide and their glycine-rich sequences (i.e., the Sequence Read Archive (SRA) with BioSample accession they were not detected using the proteomic methods described numbers SAMN08117214, SAMN08117215, SAMN08709370, above). SAMN08709371, SAMN08709372, SAMN08709373, SAMN08709374, SAMN08709375, SAMN08709376, and SAMN08709377. Bioinformatic analysis of protein sequences Using the peptides identified from the proteomic survey de- In situ hybridization on whole mounts and sections scribed above, partial or, in most cases, full length coding se- quences were isolated by standard or RACE polymerase chain Larvae were prepared for whole mount in situ hybridization as reaction (PCR) as described in [31]. In some cases, Illumina tran- described in [46]. Sections (10 μM) were taken from L. stag- nalis (shell length 10–50 mm) that had been fixed in formalde- scriptome data (see below) were used to clarify the putative com- plete mRNA. Open reading frames were translated with the Ex- hyde for 1 hour and embedded in paraffin. Riboprobes were PASy translate tool [32]. Protein sequences were searched for sig- prepared as described in [16] and were used at concentrations nal sequences with SignalP 4.1 [33]. The theoretical pI, amino of 100–500 ng/mL. Whole mounts and tissue sections were acid composition, and number of amino acids were determined processed for hybridization, the color reaction developed, and using the ExPasy ProtParam tool [34]. Tandem repeats were iden- photo-documented as described in [46]. tified with the T-REKS tool [ 35]. Sequence similarities searches were performed with the Basic Local Alignment Search Tool Comparisons of molluscan shell-forming proteomes (BLAST) algorithm [36] with tBLASTx against nr and dbEST and BLASTx against SwissProt. Domain searches were performed BLASTp-based comparisons of the L. stagnalis shell proteome with CD search [37]. Molecular function was predicted with In- were performed against a variety of calcifying proteomes re- terProScan [38]. GalNAc O-glycosylation sites were predicted us- ported in a wide phylogenetic range of metazoans as described ing the NetOGlyc 4.0 Server [39]. Scaled schematics of protein se- in [14]. These included 42 proteins from the oyster Pinctada max- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 4 Modularity and asymmetry of the molluscan mantle ima reported in [47]; 78 proteins from the oyster Pinctada margar- A brief morphological description of L. stagnalis shell itifera reported in [47]; 94 proteins from the abalone Haliotis asin- ontogeny and the adult mantle ina reported in [17]and [16]; 80 proteins from the abalone H. lae- We previously described the ontogeny of the shell gland and vigata reported in [48]; 63 proteins from the limpet Lottia gigantea shell field in L. stagnalis [57]. In order to aid the interpretation reported in [49]; 53 proteins from the oyster Crassostrea gigas re- of our in situ patterns, the following is a summary of the main ported in [50]; 71 proteins from the mussel Mya truncata reported developmental stages that we focused on. The first visible sign in [51]; 59 proteins from the grove snail Cepaea nemoralis reported of differentiation of the shell-forming tissue in L. stagnalis is a in [14]; 44 proteins from the oyster Pinctada fucata reported in [52]; thickening of the dorsal ectoderm that begins at approximately 53 proteins from the mussel Mytilus coruscus reported in [53]; 66 29 hpfc [57, 58]. These cells subsequently invaginate and by 2 proteins from the brachiopod Magellania venosa reported in [54]; days post first cleavage (dpfc) a clearly visible “shell gland” is 139 proteins from the sea urchin Strongylocentrotus purpuratus re- present [57, 58]. By 3 dpfc, the shell gland has formed a sealed lu- ported in [55]; and 37 proteins from the coral Acropora millepora men and displays the first signs of outward signs of asymmetry reported in [56]. [57]. The marginal cells that border the shell gland remain unin- vaginated and form a ring-like structure, the rosette [2]. During Analysis of the saccharide moieties of the shell matrix this time, the first extracellular organic material is secreted and is clearly visible by Scanning Electron Microscopy (SEM) (Fig. 1; The monosaccharide content of AIM and ASM was obtained [57]). By 3 dpfc, the shell gland has evaginated to form the shell by suspension and homogenization (vortex and ultrasound) of field. The former rosette cells remain highly elongated while the lyophilates in 2 M trifluoroacetic acid and subsequent hydrol- central cells take on a low columnar appearance. Over the next ysis at 105 C for 4 hours under a nitrogen atmosphere. This several days, the shell field continues to expand until it has over- treatment allows for the release of most monosaccharides from grown the visceral mass and will eventually become the adult complex mixtures, except sialic acids, which are destroyed, mantle tissue [2, 57, 58]. and the acetylated forms of glucosamine and galactosamine, The adult mantle covers the inner surface of the shell and which are converted to their respective non-acetylated forms. is responsible for shell growth and repair. The free edge of the Samples were then centrifuged for 5 minutes at 15,000g and mantle is responsible for the growth of the outer lip of the shell. evaporated to dryness (using a SpeedVac) before being dis- Timmermans conducted an extensive histochemical character- solved in 100 μL of 20 mM sodium hydroxide and homoge- ization of the mantle tissue of L. stagnalis and was able to cate- nized. After a short centrifugation (2 minutes), 80 μLofsuper- gorize the free edge of the adult mantle into six distinct zones natant was injected into the chromatograph system. The neu- based on their morphology, enzymatic activities, and biochem- tral, amino, and acidic sugar contents of hydrolysates were de- ical signatures [59]. We largely follow this categorization of the termined using high-pressure anion exchange–pulsed ampero- adult mantle tissue. Parallel to the mantle edge runs the mantle metric detection on a CarboPac PA 100 column (Dionex Corp., groove (also known as the pallial groove) defined as zone 1 (Fig. Sunnyvale, CA, USA). As blank controls, non-hydrolyzed AIMs 2). Several high-resolution microscopy and histological studies were analyzed in order to detect potential free monosaccha- on a variety of molluscs have demonstrated that it is from within rides that may lead to an over-representation of some sugar the pallial groove that the periostracum is formed and secreted residues. [59–64]. We detected a sub-regionalization of the pallial groove (zone 1) into proximal and distal zones. Immediately adjacent to the pallial groove is a broad region of high columnar cells Results referred to by Timmermans [59] as the “belt” that can be sub- Proteomic analysis of the biomineralized matrix of L. divided into three distinct zones (zones 2–4). Zone 2 is imme- stagnalis shells diately adjacent to the posterior wall of the pallial groove and comprises the anterior (or distal) portion of the belt (Fig. 2). Zone More than 1,230 peptides were analyzed by High-performance 3 consists of the posterior portion of the belt, while zone 4 rep- Liquid Chromatography (HPLC)-MS and subsequently used for resents the transitional zone between the high columnar cells protein identification using Mascot against our translated man- of the belt proper and the more posterior low columnar cells of tle transcriptomes. Of these 1,230 peptides, 329 returned sig- the outer epithelium, which comprise zone 5 (Fig. 2)[59]. nificant matches. From these 329 matches, a total of 40 shell- forming candidate transcripts were identified (see Additional file 1). Of these 40 gene products, 31 (78%) could be cloned and ex- Spatial expression patterns and molecular features of hibit in situ hybridization signals compatible with a role in shell shell-forming candidate genes formation (either in larval stages and/or in the adult mantle tis- We performed in situ hybridization for 34 distinct shell-forming sue). Seven of these 40 candidates (18%) could be cloned from genes on four distinct developmental stages and on adult man- L. stagnalis cDNA but did not produce a positive or consistent in tle tissue. The detailed results of these analyses are presented situ signal in any tissue. Three of the 40 candidate genes (8%) in Additional files 2–35, with an extensive summary presented could not be amplified by gene-specific PCR or RACE PCR. In ad- in Additional file 36 (the raw image files that constitute these dition to the 31 proteomically identified candidates that gener- figures are available in the associated GigaDB repository [ 45]. In ated positive in situ signals, three candidates that were identified Fig. 1 (for larvae) and Fig. 2 (for adult mantle tissue) we present via in silico methods (based purely on the presence of a signal se- a selection of these results that highlight some prominent fea- quence and their glycine-rich protein sequences) also generated tures of these expression patterns. In trochophore and veliger in situ signals compatible with a role in shell formation and are larval stages (2–6 dpfc), all genes could be categorized either as reported here. being: expressed in cells that symmetrically or asymmetrically border the shell gland or shell field (15/34); in cells that lay within the shell gland or shell field (9/34); a pattern that did not fit into Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 5 Figure 1: Overview of four developmental stages and representative shell-forming gene expression patterns in L. stagnalis. The first two rows provide a set of reference SEM images and adult shells (top-right-most panel) against which the images of the in situ results can be oriented. All in situ panels are from a dorsal view except the right-most column, which is from a ventral view. Here, we present the expression patterns of a selection of five shell-forming genes. These include genes with expression patterns in shell-forming cells that display evidence of symmetry (sfc-5), right asymmetry (sfc-1), left-asymmetry (sfc-17), expression entirely throughout the shell field and dorsal mantle epithelium ( sfc-20), and expression in additional non-shell-forming cells. This last expression pattern provides evidence of genes involved in shell formation that have pleiotropic functions. The scale bars in the first row are 100 μm. Indicated in the SEM images are the positions of the foot lobe (fl), foot (f), mantle margin (mm), calcified shell (s), stomodeum (st), and insoluble organic material (iom) of the shell. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 6 Modularity and asymmetry of the molluscan mantle Figure 2: Overview of the adult L. stagnalis shell-forming mantle tissue and representative shell-forming gene expression patterns that reveal its molecular modularity. (A) A semi-thin sagittal section of an adult L. stagnalis stained with Giemsa. The foot (f), mantle (m), digestive gland (dg), and radula (r) are indicated. The mantle tissue is a thin sheet of epithelium that covers the dorsal surface of the adult animal and is responsible for fabricating the shell. (B) A magnified view of the red-boxed region in part A reveals some of the cellular morphology of the adult mantle tissue. (C) A schematic representation of the mantle tissue divided into six zones as described by Timmermans [59]. The spatial distribution of enzymatic activities and biochemicals indicated in this schematic are adapted from [59]. We detect a sub-regionalization of the pallial groove (zone 1) into proximal (light green) and distal (dark green) zones. (D) The spatial expression patterns of eight representative shell-forming genes in the adult mantle tissue. The asterisk indicates that sfc-6 was identified using in silico methods rather than proteomic methods. our classification scheme (1/34); or were not expressed in any (10/34); asymmetrically in the outer edge of the mantle (18/34); detectable way (9/34). In later stages (∼7dpfc),all geneswere throughout the entire mantle tissue (2/34); a pattern that did not either: expressed uniformly along the outer edge of the mantle fit into our classification scheme (1/34); or were not expressed in Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 7 Trochophore larva Juvenile Adult mantle dorsal view ventral view sagital section carbonic anhydrase left right malate dehydrogenase ATPase-rich cells Peroxidase RNA rich Peroxidase RNA rich alkaline phosphatase Tyrosine Tyrosinase Tyrosinase zone 5 zone 4 zone 3 zone 2 zone 1 zone 6 Shell-forming Asymmetric Symmetric Left side of Left + right side Entire Right side of Interior Zone 5 Zone 4 Zone 3 Zone 2 Zone 1 Sequence similarity or conserved domain candidate border border outer lip of outer lip mantle outer lip Lstag-sfc-1 right ✓✓✓ Lstag-sfc-2a right Early nodulin-12A (SP) ✓✓✓ Lstag-sfc-2b right ✓✓✓ Lstag-sfc-3 right None ✓✓✓ Lstag-sfc-4 right None ✓✓ Lstag-sfc-5 Animal haem peroxidase (CD) ✓✓ ✓✓ Lstag-sfc-6a* None ✓✓ ✓ Lstag-sfc-6b* None ✓✓ ✓ Lstag-sfc-7* None ✓✓ ✓ Lstag-sfc-8* None ✓✓ Lstag-sfc-9a None ✓✓ ✓ Lstag-sfc-9b None ✓✓ ✓ Lstag-sfc-10a None ✓✓ ✓ Lstag-sfc-10b None ✓✓ ✓ Lstag-sfc-11a None ✓✓ ✓ Lstag-sfc-11b None ✓✓ ✓ Lstag-sfc-11c None ✓✓ ✓ Lstag-sfc-12 left None ✓✓ Lstag-sfc-13 None ✓✓ ✓ Lstag-sfc-14 None ✓✓ ✓ Lstag-sfc-15 None ✓✓ ✓ ✓ ✓ Lstag-sfc-16 None ✓✓✓ Lstag-sfc-17 left None ✓✓ Lstag-sfc-18 Immunoglobulin domain (CD) ?? ? ? ? ? ? ✓✓✓ Lstag-sfc-19 Perlucin-like protein (SP), C-type lectin (CD) Lstag-sfc-20a None ✓✓ ✓ Lstag-sfc-20b None ✓✓ ✓ Lstag-sfc-21 Putative chitin deacetylase-like domain (CD) ✓✓ ✓✓ Lstag-sfc-22 Pif97 Aragonite-binding protein (SP) Lstag-sfc-23a Otoancorin (SP) ✓✓✓✓✓ Lstag-sfc-23b Otoancorin (SP) ✓✓✓✓✓ Lstag-sfc-24a None ✓✓✓ Lstag-sfc-24b None ✓✓✓ Lstag-sfc-25 None ✓✓✓ Lstag-sfc-26 PREDICTED: uncharacterized protein (NR) ✓✓✓ Lstag-sfc-27a PREDICTED: extensin-like isoform X1 (NR) ✓✓✓ Lstag-sfc-27b PREDICTED: formin-like protein 2 (NR) ✓✓✓ Lstag-sfc-28a None ✓✓ ✓ Lstag-sfc-28b None ✓✓ ✓ Lstag-sfc-28c None ✓✓ ✓ Lstag-sfc-29 Galaxin (SP) ✓✓ ✓ Lstag-sfc-30 None ✓✓ ✓ Lstag-sfc-31 SUSHI repeat + short complement-like repeat (CD) ✓ ✓✓✓✓✓ Lstag-sfc-32 Intermediate filament protein (CD) ✓ ✓ ✓✓✓✓✓ Lstag-sfc-33 vWA type A domain + collagen alphaI-XII-like (CD) ✓✓ ????? Lstag-sfc-34 None ✓✓ Figure 3: Summary of the spatial gene expression profiles and conserved features of 34 L. stagnalis shell-forming candidates. Schematically represented in a trochophore larva are genes with an asymmetric expression profile (dark gray), as well as genes expressed broadly across the shell field (light blue). Cells in this r egion of the trochophore are likely to give rise to cells in zone 5 of the adult mantle, and we have maintained that color scheme to suggest this. Although we schematically present a trochophore larva here (2–3 dpfc), the summarized expression patterns also include veliger stages (3–6 dpfc). Cells bordering the larval shell gland and shell field (black ring in the trochophore) are likely to give rise to one or more zones 1–4 in the adult mantle. In juveniles (∼7 dpfc) many genes were expressed in the left, right, or continuously throughout the free edge of the mantle that produces the outer lip of the shell. Question marks indicate expression patterns that could not be categorized according to our scheme. An “x” indicates no expression was detected. A “?” indicates that the expression pattern could not be categorized according to our scheme. The names of enzymes and other molecular features indicated in zones 1–5 on the schematic of the adult mantle are summarized from [59] and [3]. Sequence similarity and conserved domains in the final column of the table are summarized from a number of BLAST searches against SwissProt, the non-redundant N ational Center for Biotechnology Information database, and the Conserved Domain database. See Additional files 40–42 for the results of all BLAST and domain s earches. The lineages of the top BLAST hits are listed in Additional file 43. A version of this figure that includes a more complete summary of the molecular features of each gene is provided in Additional file 36. The asterisks indicate that sfc-6, -7, and -8 were identified using in silico methods rather than proteomic methods (as was the case for all other gene products presented here). Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 alkaline phosphatase 8 Modularity and asymmetry of the molluscan mantle any detectable way (3/34). Finally, in adult mantle tissue, genes 32 (with similarity to C. nemoralis contig 572), which appears were either: expressed in one or more of the five zones described to be an intermediate filament protein. Lstag-sfc-22 (a gene ex- by Timmermans [59] (32/34); a pattern that did not fit into our pressed exclusively in zone 5, Additional file 23) shared relatively classification scheme (1/34); or were not expressed in any de- weak similarity with C. nemoralis contig 821 and shares signifi- tectable way (1/34). We have schematically summarized all of cant sequence similarity with PIF, an aragonite-binding protein these results in Fig. 3. reported to be involved in nacre formation in the oyster P. fucata [19]. Interestingly, of the 12 candidates expressed in the matrix- secreting zone 5, 9 showed similarity with other shell proteins Alternative splicing increases the diversity of (Fig. 4 and Additional files 23–33). Eight of these were shared with shell-forming proteins C. nemoralis (Fig. 4). In contrast, none of the asymmetrically ex- Via alternative splicing of mRNAs, transcripts with a variety pressed or glycine-rich candidates were found in any of the other biomineralizing proteomes (low-complexity filtering was inacti- of functions can be generated from a single genomic locus vated in these comparisons; Fig. 4). [65]. With a draft genome for L. stagnalis available, we were able to perform some preliminary investigations into alternative splicing of our shell-forming candidates. While some candidate Glycosylation patterns of the shell matrix genes displayed the same or very similar exon-splicing patterns in all surveyed tissues (e.g., Additional files 5, 9, 14, 15, and 16), The monosaccharides profiles of ASMs and AIMs extracted from most candidates are apparently alternatively spliced depending adult L. stagnalis shells were peculiar, with less than half of the on the tissue they are expressed in (Additional files 4, 11, 12, 18, dozen standard monosaccharides represented. For ASM, these 19, 22, 23, 25, and 33). Striking examples include Lstag-sfc-21 and include galactosamine, glucosamine, galactose, glucuronic acid, Lstag-sfc-24, which are expressed in many tissues but display and glucose, while AIM lacked glucuronic acid (Table 1). We also significant alternative splicing in the adult mantle (Additional found marked differences in the glycosylation rates of ASMs and files 22 and 25). All splice variants of candidate Lstag-sfc-24 en- AIMs. In general, theabsolutedegreeofglycosylation of theASM code proteins with the same aspartic acid-rich motif (Additional was 1–3 orders of magnitude greater that of the AIM. The largest files 25, 37, and 38). Aspartic acid-rich proteins have been sug- difference was in the amount of galactosamine (10.6 ng/μLvs. gested to act as an organic template for epitaxial crystal growth 0.03 ng/μL). While glucosamine was more abundant absolutely [66, 67]. It is tempting to speculate that the three additional do- in the ASM (9.41 ng/μL vs. 0.13 ng/μL), the proportional differ- mains only present in adult mantle Lstag-sfc-24 contigs confer ence was not so extreme (34.8% vs. 54.2%). a specific shell-forming function to this protein. The putative chitin-interacting candidate Lstag-sfc-21 presented in Additional file 22 carries a signal sequence and is predicted to possess a Discussion catalytic activity. Intriguingly, a number of splice variants of this Molecular modularity of the adult molluscan mantle gene within the adult mantle are predicted to lack a signal se- Two of the most striking features of the phylum Mollusca are quence, the chitin-binding or catalytic ability (Additional file 37). A number of shell-forming gene candidates produce alter- its size (in terms of number of species) and its diversity. Widely accepted to be second only to the Arthropoda in terms of num- natively spliced transcripts that encode proteins with differ- ences regarding the presence/absence of a signal sequence (Ad- beroflivingspecies [73, 74], molluscs arguably display the great- est diversity of body forms of all metazoan phyla and have suc- ditional files 11, 12, 22, 25, and 37). Some shell-forming genes also produce alternatively spliced transcripts that encode pro- cessfully colonized all kinds of environments. While there cur- teins with similar coding features but radically different 5’ or rently exists no consensus as to why molluscs have enjoyed such deep evolutionary success (one interesting suggestion includes 3’ untranslated regions (UTRs) (Additional files 12, 19, 33, and 37). While UTRs do not contain protein-coding information, they a plastic nervous system [75]) we believe the mantle tissue (an apomorphy of the phylum) and its ability to prolifically evolve can be critical for localization of the mRNA [68, 69] and post- transcriptional gene regulation by molecules such as miRNAs new shell phenotypes must contribute to an explanation of this success. A logical extension of this question would therefore be, [70]. Indeed, several miRNAs have now been associated with the targeting and regulation of biomineralizing proteins [71, 72]. “what is it about the molluscan mantle tissue that makes it so evolutionarily plastic?” For arthropods, segmentation and body plan modularization (and the underlying gene regulatory net- Comparisons of molluscan shell-forming proteomes works that control appendage identity within each segment) are We conducted a broad comparison of our L. stagnalis shell- widely thought to have played leading roles in supporting the forming genes against a wide phylogenetic range of 12 other diversification of insects, spiders, and crustaceans [ 76]. The im- biomineralizing proteomes comprising 879 proteins (sequences portance of establishing segmentation at a very early develop- used in this analysis are provided in Additional file 39). Of all L. mental age in prominent phyla such as annelids, chordates, and stagnalis shell proteins, 27 shared significant sequence similar- arthropods has caused much effort to be spent on identifying ity with 1 or more of these 12 proteomes (Fig. 4). The highest the causal molecular mechanisms that may have common evo- degree of overall similarity was found with the shell-forming lutionary histories [77–79]. As recently reviewed by Esteve-Altava proteome of the common groove snail C. nemoralis (Fig. 4), the [80], the presence of morphological modules can help us to un- closest phylogenetic relative to L. stagnalis of all species in this derstand the evolvability of body form, but the identification of comparison. The next highest level of similarity shared with the such modules has so far been biased toward mammals, arthro- L. stagnalis shell-forming proteome (15.9% of the L. gigantea shell- pods, and plants. Following Esteve-Altava’s and Eble’s [81]defi- forming proteome) was markedly lower. Lymnaea stagnalis shell- nition of a morphological module (a group of body parts that are forming proteins that shared significant similarity with biomin- more integrated among themselves than they are to other parts eralizing proteins from other species and that also returned a outside the group), we propose that the molluscan mantle is a significant match against a SwissProt entry included Lstag-sfc- prime example of such a morphological module. This modular Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 9 L. stagnalis 58.7% (27/46) M. venosa 3.0% (2/66) M. truncata 4.2% S. purpuratus (3/71) 0.7% (1/139) C. nemoralis 28.8% (17/59) P. maxima A. millepora 2.4% 8.1% (1/42) (3/37) L. gigantea P. margaratifera 15.9% 7.7% (10/63) (6/78) P. fucata 9.1% H. asinina 1.1% (4/44) (1/94) M. coruscus 9.4% Top quartile of global similarity (5/53) 3rd quartile of global similarity H. laevigata C. gigas 2nd quartile of global similarity 3.75% 11.3% Lowest quartile of global similarity (3/80) (6/53) Threshold e-value = 10e-⁶ Candidate Adult mantle zone expression Conserved domains/similarity to Features Lstag-sfc-4 1 - Gly, Cys rich Lstag-sfc-5 1, 2 Animal haem peroxidase (CD) - Lstag-sfc-10a 3 - RLCD; Gly, Ser rich Lstag-sfc-10b 3 - Gly, Ser rich Lstag-sfc-11a/b 3 - RLCD; Ala, Thr, Ser, Pro, Leu rich Lstag-sfc-11c 3 - RLCD; Ala, Thr, Ser, Pro, Leu rich Lstag-sfc-18 2, 3, 4 Immunoglobulin domain (CD) Ser rich Lstag-sfc-20a 4 - - Lstag-sfc-20b 4 - Leu rich Lstag-sfc-21 1, 2, 5 Putative chitin deacetylase-like domain (CD) Thr rich Lstag-sfc-22 5 Pif97 Aragonite-binding protein (SP) Thr rich Lstag-sfc-23a/b 3, 4, 5 Otoancorin (SP) RLCD; Asp, Leu rich Lstag-sfc-24a/b 5 - RLCD; Asp, Ser, Asn rich Lstag-sfc-26 5 - RLCD; Asn, Gly, Gln Lstag-sfc-27a 5 PREDICTED: extensin-like isoform X1 (NR) RLCD; Pro, Gly Leu rich Lstag-sfc-27b 5 PREDICTED: formin-like prottein 2 (NR) RLCD; Pro, Gly rich Lstag-sfc-28a 5 - Gln, Pro Ala rich Lstag-sfc-28b 5 - Ala, Leu, Pro, Ser, Thr rich Lstag-sfc-28c 5 - RLCD; Gln, Pro rich Lstag-sfc-29 4, 5 Galaxin (SP) RLCD; Pro, Gln Gly rich Lstag-sfc-30 5 - - Lstag-sfc-31 2, 3, 4, 5 SUSHI repeat + short complement-like repeat (CD) RLCD; Gly, Ala, Leu, Pro rich Lstag-sfc-32 1, 2, 3, 4, 5 Intermediate filament protein (CD) Leu, Glu, Ser, Ala rich Lstag-sfc-33 NA vWA type A domain + collagen alphaI-XII-like (CD) RLCD; Val, Ala rich Figure 4: BLASTp comparisons of the L. stagnalis shell proteome against 879 biocalcifying proteins derived from six bivalves, four gastropods, one brachiopod, one −6 sea urchin, and one coral. Individual lines spanning the ideogram connect proteins that share significant similarity (e-values <10e ). Transparent red lines connect −6 proteins with the lowest quartile of similarity (with a threshold of 10e ) and green lines with the highest quartile of similarity. The percentage of each shell proteome that shared similarity with the L. stagnalis proteome is indicated. The table provides further information for those candidates that share sequence similarity with another species. Abbreviations: CD: conserved domain database; NR: GenBank non-redundant protein database; SP: SwissProt database. nature of the molluscan mantle is not unique to Lymnaea [82–84]. spatial expression patterns of shell-forming proteomes from a Although the precise functions of these zones (and of the indi- selection of other molluscan lineages will contribute to a more vidual gene products that define them) await the development refined understanding of molluscan shell evolution. of targeted genome editing methods, it is clear that they must act in a coordinated way to deposit the shell. We predict that there are related modules of gene regulatory networks (GRNs) Ontogenetic expression of shell-forming candidates that act to specify each zone of the molluscan mantle and that A prominent outcome of our survey of the adult shell proteome it is the modular nature of these GRNs and the resulting mor- is that many of the genes that encode these proteins are not only phological modularity of the mantle tissue that supported the regulated spatially but also temporally. Many shell-forming can- diversification of the phylum Mollusca. Characterization of the didates are expressed in the invaginated larval shell gland of the Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 S. purpuratus A. millepora L. gigantea H. asinina C. gigas M. coruscus P. fucata P. margaratifera P. maxima C. nemoralis M. truncata M. venosa H. laevigata 10 Modularity and asymmetry of the molluscan mantle Table 1: Glycosylation analysis of acid-soluble and acid-insoluble matrices extracted from adult L. stagnalis shells ASM AIM Monosaccharide ng/μg%ng/μg% Fucose ND - TR - Rhamnose ND - TR - Galactosamine 10.60 39.2 0.03 12.5 Arabinose ND - ND - Glucosamine 9.41 34.8 0.13 54.2 Galactose 4.34 16.0 0.04 16.7 Glucose 0.32 1.2 0.04 16.7 Mannose ND - TR - Xylose ND - ND - Galacturonic acid ND - ND - Glucuronic acid 2.39 8.8 ND - Total 27.06 100.0 0.24 100.0 ND = not detected, TR = trace. trochophore (Additional files 14, 15, 21, 24, 25, 27, and 28) or in (Fig. 1). We therefore suggest that Lstag-sfc-1, -2,and -3 are in cells that border it (Additional files 2, 3, 6, 7, 8, 10, 11, and 18). some way associated with producing thinner, more rapidly pro- Only two candidates were solely expressed in the adult man- duced shell at the outer shell lip than in the thicker parietal re- tle tissue and not in any larval stage (Additional files 17 and 23). gion, while Lstag-sfc-17 may inhibit the rapid deposition of shell. Timmermans [59] concluded that the spatial patterning of larval Exactly how this is achieved awaits the development of gene- shell-forming cells persists throughout development and fore- specific function assays. shadows the zonation observable in the adult mantle. We also In addition to the trochophore left/right asymmetry corre- observed this phenomenon at the molecular level. All candidate sponding to the left + parietal/right + outer lip regions of the genes that were expressed in the margin of the shell gland or shell, there is a second axis of symmetry that becomes appar- the shell field were expressed in the belt (zones 2, 3, and 4) of ent in 7-day-old juvenile snails. Many shell-forming candidates the adult mantle (summarized in Fig. 3, Additional files 2–8, 10– are initially symmetrically expressed in or surrounding the shell 12, and 18). Most candidates expressed in the invaginated cells gland of 2- to 3-dpfc trochophores but then become asymmet- of the shell gland or throughout the developing shell field were rically expressed in the mantle of older animals. For example, subsequently expressed in the low columnar outer epithelium Lstag-sfc-6, -7, -8, -12, -14, -15, -17, -18, -20, -23, -24, -26, -27, -29 of zone 5 in adult mantle tissue (summarized in Fig. 3, Addi- and -31 are expressed in the left side of the free mantle edge tional files 24, 25, 27– 29, and 31). However, three genes conspicu- in 7-dpfc juveniles (summarized in Fig. 3). In contrast, relatively ously deviate from this pattern. Lstag-sfc-13, -14,and -20 display few shell-forming gene candidates (Lstag-sfc-5, -9,-10,-21, and a broad expression pattern in the invaginated cells of the shell -25)are expressedevenlyalong thefreeedgeofthe mantlein gland and throughout the entire shell field in larvae but were not 7-dpfc juveniles (summarized in Fig. 3). detected in the low columnar outer epithelium (zone 5) of the adult mantle tissue (Additional files 14, 15, and 21). However, we The spatial expression of a peroxidase in the adult should point out that for all candidate shell-forming genes, we did not consider the potential effect of a diurnal rhythm on gene mantle allows a model of shell formation to be expression. All samples for in situ hybridization were taken dur- developed ing daylight hours, and so genes with activity during the night In agreement with Timmermans histochemical study of perox- would be missed. idase activity [59], the expression of Lstag-sfc-5, a shell-forming candidate with an “animal heme-dependent peroxidase” do- main (Pfam PF03098; Additional file 41B) is localized to zones Asymmetric expression of shell-forming genes 1 and 2. Peroxidases may be involved in periostracum forma- The expression of Lstag-sfc-1, Lstag-sfc-2,and Lstag-sfc-3 in tion by cross-linking fibrous proteins rich in reactive quinones to zones 1 and 2 of the adult mantle suggests they may be in- form water insoluble, protease-resistant polymers [85–87]. This volved in the formation of the periostracum (Fig. 1 and Ad- process, also referred to as tanning or sclerotization, can also ditional files 2–4); however, it is their larval expression pat- be catalyzed by tyrosinase (also known as catechol oxidase, cat- terns that are more striking. Lstag-sfc-1, -2,and -3 display a echolase, polyphenoloxidase, phenoloxidase, and phenolase). right-sided asymmetric expression pattern in cells bordering Within the molluscan biomineralization literature, sclerotiza- the shell gland and shell field. In contrast, Lstag-sfc-17 is ex- tion by tyrosinase appears to be the more commonly assumed pressed on the left side (Fig.1and Additional file 18). Following mechanism, rather than by peroxidase. Nonetheless, Timmer- the expression of these genes ontogenetically into older lar- mans demonstrated that heat inactivation clears the periostra- vae that begin to display the coiled phenotype of the adult, it cal groove and belt of both peroxidase activity and the ability to is apparent that right-sided cells in the trochophore are likely form melanin (a typical assay used to test for tyrosinase activ- to be those that give rise to the right + anterior region of ity), while specific tyrosinase inhibitors sodium bisulphite and the adult mantle that will produce the outer lip of the shell, potassium cyanide (NaHSO and KCN) did not affect its ability while left-sided cells will give rise to posterior mantle tissue to produce melanin [59]. The spatial expression pattern of Lstag- responsible for forming the left + parietal region of the shell sfc-5, coupled with the observations that newly secreted perios- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 11 tracum itself also displays peroxidase activity [57] and Timmer- damaged, these IGFs would be released and subsequently stim- mans experiments [59], strongly suggests that the peroxidase we ulate the underlying mantle epithelium to re-calcify. One line have identified here plays a key role in cross-linking the perios- of evidence that strongly supports this hypothesis is provided tracum in L. stagnalis rather than a tyrosinase, as also supposed by the osteogenic activity of mollusc shells [97]. This hypothe- for other gastropods such as Lottia [15]. sis implies that although the shell is acellular, it is able to ac- tively communicate and provide real-time feedback to the man- tle epithelium [98]. This interesting hypothesis awaits the devel- Glycine-rich shell-forming candidates are likely to be opment of gene-specific knock-down or knock-out assays. substrates for the peroxidase An important aspect of scleroprotein formation is its spatial RLCDs are an abundant feature of L. stagnalis shell coordination. The cross-linking reaction often generates cyto- proteins toxic intermediates, and the end products cannot be easily de- graded or resorbed [88]. Furthermore, the uncontrolled forma- Proteins containing RLCDs are a prominent feature of mollus- tion of extensive scleroprotein polymers prior to secretion would can shell-forming proteomes [15, 99, 100], and L. stagnalis is no exception. More than half of the L. stagnalis shell-forming can- clearly be detrimental to the cell. One common strategy to avoid these events is to compartmentalize the scleroprotein precur- didates we identified possess RLCDs. Proteins containing these domains were present in the belt and the low columnar outer sor (that is unable to spontaneously polymerize) away from the cross-linking enzyme. Following secretion, the precursors are epithelium of the adult mantle and in a wide variety of pat- terns of the larval stages we investigated. The motif complex- activated and enzymatically cross-linked [88]. Such a scenario would suggest that the substrate upon which the peroxidase ity, motif length, and number of motif repeats can vary greatly, from stretches consisting of a single amino acid (e.g., Additional acts is not located within the same cells. Three candidates expressed in zone three (Lstag-sfc-6, -7,and file 24) to motifs that exceed 10 amino acids (e.g., Additional files 3, 4, 29 and 34). In some cases, almost the whole protein -8) encode secreted, basic proteins that are dominated by repet- itive low-complexity domains (RLCDs) and anomalous amino is composed of RLCDs (Additional files 7–10). Repeated motifs are a common feature of structural proteins such as collagens, acid contents (high glycine, tyrosine, asparagine, and leucine keratins, silk, and cell wall proteins, as well as structural mod- contents; Additional files 37 and 38). All of these glycine-rich proteins carry tyrosine residues flanked by glycine. This arrange- ules in functional proteins such as receptors, histones, ion chan- nels, and transcription factors [101, 102]. RLCDs are often part ment has been shown to be favorable for the formation of cross- links between tyrosine residues by peroxidase [89]. Waite, in of intrinsically unstructured regions that lack a fixed or ordered three-dimensional structure [101]. In some cases, these regions his review of natural quinone-tanned glues [85], highlighted the typical L-3,4-dihydroxyphenylalanine (DOPA) -containing con- define the functionality of the protein. As a general rule, un- structured proteins interact readily with other proteins [103], sensus precursor peptide sequences from a number of marine invertebrates. Allowing for a single mismatch, these substrate and the highly repetitive, modular, and biased amino acid com- positions can confer strength and elasticity [104]. It will be ex- peptides (VGGYGYGK, GGGFGGYGK, and GGGYGGYGK, cross- linking tyrosine residues in bold) can be found within Lstag-sfc- tremely informative to selectively remove RLCDs from shell- forming proteins and to study the resulting shell phenotypes 6, Lstag-sfc-7, and Lstag-sfc-8. Interestingly, these glycine-rich candidates are expressed exclusively in zone 3 (Additional files once genome modification tools become broadly available to molluscs. 7–9) immediately adjacent to zone 2, the region in which the per- oxidase Lstag-sfc-5 is expressed (Additional file 6). Theoretically, once these proteins are secreted, the secreted peroxidase would Broad sequence similarity comparisons of metazoan be in very close proximity to the glycine-rich proteins and could biomineralizing proteomes act on the favored tyrosine residues to form di-tyrosine cross- The crossed-lamellar microstructure is fabricated by phyloge- links extracellularly. netically diverse molluscan taxa and is by far the most com- monly employed shell design of the Conchifera [21, 22, 28]. A role for immunity and signaling in shell formation While much attention has been dedicated to the characteri- Lstag-sfc-18 contains two Ig superfamily domains (Additional zation of nacre-forming bivalve shell proteomes, technical ad- file 19; [ 90]) and displays sequence similarity with the IMP- vances in nucleic acid sequencing and proteome-scale surveys L2–like proteins (Additional files 40 and 41), an insulin- have seen a rapid growth in the number and diversity of mol- like growth factor–binding protein (IGF-BP) that carries two luscan shell-forming proteomes and allow broad comparisons immunoglobulin-like domains and is able to bind IGF [91]. Sev- of these datasets to be performed. These comparisons can pro- eral studies by Dogterom and colleagues demonstrated the in- vide insight into the degree of evolutionary conservation that fluence of a growth hormone secreted by the cerebral ganglia exists across shell-forming proteomes [50]. In general, mollus- specifically on shell formation in L. stagnalis [92–95]. The authors can shell-forming proteomes are markedly different, with some conclude that this growth hormone acts on cells in the belt re- deeply conserved elements such as alkaline phosphatases, per- gion to control shell extension and periostracum formation but oxidases, and carbonic anhydrases [28, 57, 105, 106]. The signif- not on shell thickening. Interestingly, perlustrin, a protein asso- icant diversity of molluscan shell ultrastructures, crystal tex- ciated with nacre in abalone shells, contains an IGF-BP domain tures, colors, and materials properties therefore cannot be ex- and was also shown to bind IGFs and insulin [96]. An intrigu- plained by the use of the same genes in different ways. Rather, ing idea for the presence of IGF-BP in the abalone shell is that each lineage has uniquely evolved a large fraction of its shell- it would allow the shell to signal to the underlying mantle ep- forming proteome [14–16, 100, 107]. To expand on this compara- ithelium. According to this hypothesis, IGFs present in the ex- tive theme, we collected 879 biomineralizing proteins validated trapallial fluid are bound by IGF-BP during calcification and in- by proteomics from 10 molluscs, 1 brachiopod, 1 sea urchin, and corporated into the shell. Should the shell dissolve or be locally 1 coral and performed sequence similarity comparisons against Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 12 Modularity and asymmetry of the molluscan mantle our L. stagnalis dataset. Two of the 10 molluscs, Cepaea nemoralis this phylum-specific organ and allude to an explanation as to and Mya truncata [51, 108], construct shells that contain crossed- why the Mollusca have evolved so many successful shell mor- lamellar texture. Interestingly, our comparative analyses show phologies. While gene co-option, domain shuffling, and gene that L. stagnalis and M. truncata have only three proteins that family expansion are mechanisms that have clearly contributed share relatively low degrees of sequence similarity, while L. stag- to the great diversity of molluscan shell-forming proteins, our nalis and C. nemoralis share 17 proteins (some of these with very analyses also suggest that alternative splicing acts to signifi- high degrees of sequence similarity), the highest extent of simi- cantly expand the shell-forming molecular repertoire. Compar- larity between all species surveyed (Fig. 4). Both L. stagnalis and C. ing the results of spatial gene expression surveys focused on nemoralis inhabit non-marine environments, and the similarities shell-formation from a broad range of molluscan taxa will shed in their shell proteomes may either be a manifestation of this further light on the evolutionary story of this fascinating struc- and/or a reflection of their crossed-lamellar shells. The similar- ture. ity of their shell protein content may also reflect the relatively recent divergence time (Meso-Cenozoic) of these two clades (Sty- Availability of supporting data lommatophoran, i.e., C. nemoralis vs. hygrophilid, i.e., L. stag- nalis) within the monophyletic order of pulmonate gastropods, All raw NGS data has been deposited with the SRA with in comparison to the other species. One of the most striking ob- BioSample accession numbers SAMN08117214, SAMN08117215, servations we made in these comparisons was that almost all L. SAMN08709370, SAMN08709371, SAMN08709372, stagnalis shell-forming candidates expressed in zone 5 share se- SAMN08709373, SAMN08709374, SAMN08709375, quence similarity with C. nemoralis. Conversely, L. stagnalis shell- SAMN08709376, and SAMN08709377. Individual image files forming candidates expressed asymmetrically on the right side for the in situ hybridization gene expression patterns and in larvae were not present in any other biomineralizing pro- the sense strand cDNA sequences used to generate the teomes. in situ hybridization riboprobes can be accessed from the Some L. stagnalis shell-forming proteins contain domains associated GigaDB repository [45]. The mantle transcrip- found in a number of the biomineralizing proteins present in tome assemblies are also available via GigaDB [45] (file the dataset we assembled or are known to play a role in pro- names: C2844 CLC idba Trinity for annotation.fasta and cesses other than biomineralization such as the Sushi domain, C2845 CLC idba Trinity for annotation.fasta). All MS data have the von Willebrand factor A domain, the immunoglobulin do- been deposited with the ProteomeXchange Consortium with main, and the filament protein domain [ 14, 109]. The Pif-like pro- the dataset identifiers PXD008547 and 10.6019/PXD008547. tein is prevalent in both bivalve and gastropod nacreous shell Other supporting data are available from additional files, also proteomes and is known to bind aragonite crystals and to reg- including an extended description of the in situ hybridization ulate nacre formation [110]. However, limpets, which construct results (see additional file 47). crossed-lamellar structures, also contain Pif in their shells [15, 110, 111]. Our results further demonstrate that Pif is not limited Additional files to nacreous matrices and that it is likely to be a deeply conserved element of the molluscan biomineralizing proteome. Additional file 1 . Summarized results of MASCOT searches. Additional files 2–35 . Whole mount in situ hybridisation results Glycosylation patterns and molecular features of 34 shell-forming gene candidates. Additional file 36 . A more comprehensive summary of the re- Our preliminary analysis of the sugar moieties associated with sults presented in Fig. 3. shell-forming proteins revealed an interesting dichotomy be- Additional file 37 . Detailed table of the molecular features of all tween the ASM and AIM; the population of ASM proteins ap- shell-forming protein candidates. pear to be far more glycosylated than AIM proteins (Table 1). Additional file 38 . Results of repetitive motif searches using Whether this difference is generated by a heavily glycosylated Xstream. subset of the ASM or if it reflects a general trend of most ASM Additional file 39 . A FASTA formatted file of the 879 protein se- proteins being glycosylated remains unknown. We also cannot quences used to construct Fig. 4. determine whether there are any spatial biases within the adult Additional file 40 . Detailed results of tBLASTx similarity mantle tissue with regards to the location of glycosylated pro- searches for all shell-forming candidate genes against nr teins. The high percentage of glucosamine identified in AIM and database. ASM suggests that chitin, or its deacetylated derivative chitosan, Additional file 41 . Detailed results of protein family and protein is present in both extracts, but this hypothesis requires further domain similarity searches for all shell-forming candidate genes testing. Despite their likely importance to the functional mech- against CDD database. anisms of shell formation, post-translational modifications of Additional file 42 . Detailed results of BLAST similarity searches molluscan shell-forming proteins remain relatively understud- for all shell-forming candidate genes against SwissProt ied, and we predict that research efforts in these directions database. would yield interesting functional insights into the mechanisms Additional file 43 . Lineages for all shell-forming candidates that of shell fabrication. returned positive BLAST results. Additional file 44 . Nucleotide sequences of 34 families of shell Conclusion forming candidate genes. Additional file 45 . Translated sequences of 34 families of shell By characterizing the spatial expression patterns of 34 genes forming candidate genes. associated with shell formation, we have revealed patterns of Additional file 46 . mRNA regions targeted by riboprobes. asymmetry that presumably contribute to the coiled phenotype Additional file 47 . Extended Results and Discussion. of Lymnaea’s shell. Our broad survey of these genes in the adult mantle tissue also highlight the morphological modularity of Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 13 tion and molecular mechanism exploration of three gran- Abbreviation ulin epithelin precursor splice variants in biomineraliza- AIM: acid-insoluble matrix; ASM: acid-soluble matrix; BLAST: tion of the pearl oyster Pinctada fucata. Mol Genet Genom Basic Local Alignment Search Tool; CNS: central nervous sys- 2016;291:399–409. tem; dpfc: days post first cleavage; GRN: gene regulatory net- 11. Wang J, Gao J, Xie J, et al. Cloning and mineralization- work; hpfc: hours post first cleavage; IGF-BP: insulin-like growth related functions of the calponin gene in Chlamys farreri. factor–binding protein; LC: Liquid chromatography; MS: mass Comp Biochem Physiol B 2016;201:53–58. spectrometry; NGS: next-generation sequencing; PCR: poly- 12. YarraT,GharbiK,BlaxterM,etal. Characterizationofthe merase chain reaction; RLCD: repetitive low-complexity do- mantle transcriptome in bivalves: Pecten maximus, Mytilus main; SEM: Scanning Electron Microscopy; SRA: Sequence Read edulis and Crassostrea gigas. Mar Geonomics 2016;27:9–15. Archive; UTR: untranslated region. 13. Gao P, Liao Z, Wang X-X, et al. Layer-by-layer pro- teomic analysis of Mytilus galloprovincialis shell. PLoS One 2015;10:e0133913. Competing interests 14. Mann K, Jackson DJ. Characterization of the pigmented The authors declare that they have no competing interests. shell-forming proteome of the common grove snail Cepaea nemoralis. BMC Genom 2014;15:249. 15. Marie B, Jackson DJ, Ramos-Silva P, et al. The shell-forming Author contributions proteome of Lottia gigantea reveals both deep conservations I.H. carried out the molecular work, bioinformatic analyses, and and lineage-specific novelties. FEBS J 2013; 280:214–32. co-wrote and drafted the manuscript. F.M. and B.M. performed 16. Jackson DJ, McDougall C, Green K, et al. A rapidly evolv- the proteomic analyses and drafted the manuscript. D.J.J. con- ing secretome builds and patterns a sea shell. BMC Biol ceived and supervised the study, contributed to the molecular 2006;4:40. work, contributed to the bioinformatic analyses, performed the 17. Marie B, Marie A, Jackson DJ, et al. Proteomic analysis of the histological sections, and co-wrote and drafted the manuscript. organic matrix of the abalone Haliotis asinina calcified shell. All authors read and approved the final manuscript. Prot Sci 2010;8:54. 18. Zhao M, He M, Huang X, et al. A homeodomain transcription factor gene, PfMSX, activates expression of Pif gene in the Acknowledgements pearl oyster Pinctada fucata. PLoS One 2014;9:e103830. 19. Suzuki M, Saruwatari K, Kogure T, et al. An acidic matrix We are grateful to Wolfgang Drose ¨ for assistance and advice with protein, Pif, is a key macromolecule for nacre formation. in situ sectioning, Isabelle Zanella-Cleon from IBCP (Lyon) for Science 2009;325:1388–90. MS analysis, and Jennifer Hohagen and Dorothea Hause-Reitner 20. Jackson DJ, Worheide ¨ G, Degnan BM. Dynamic expression of who generated SEM images. Illumina sequencing was per- ancient and novel molluscan shell genes during ecological formed by Markus B. Schilhabel and his team at the Institute of transitions. BMC Evol Biol 2007;7:160. Clinical Molecular Biology, Christian-Albrechts-University Kiel. 21. Dauphin Y, Denis A. Structure and composition of the arag- This work was funded by DFG (JA 2108/2-1 and JA 2108/6-1) and onitic crossed lamellar layers in six species of Bivalvia and VolkswagenStiftung (92075) grants to D.J.J. Gastropoda. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology 2000;126:367–77. References 22. de Paula SM, Silveira M. Studies on molluscan shells: contri- butions from microscopic and analytical methods. Micron 1. Mao L-B, Gao H-L, Yao H-B, et al. Synthetic nacre 2009;40:669–90. by predesigned matrix-directed mineralization. Science 23. Almagro I, Drzymala P, Berent K, et al. New crystallographic 2016;354:107–10. relationships in biogenic aragonite: the crossed-lamellar 2. Kniprath E. Ontogeny of the molluscan shell field: a review. microstructures of mollusks. Crystal Growth & Design 2016. Zool Script 1981;10:61–79. 24. Kuhn-Spearing LT, Kessler H, Chateau E, et al. Fracture 3. Jackson DJ, Degnan BM. The importance of Evo-Devo to an mechanisms of the Strombus gigas conch shell: implications integrated understanding of molluscan biomineralisation. for the design of brittle laminates. J Mater Sci 1996;31:6583– J Struct Biol 2016;196:67–74. 4. Okabe T, Yoshimura J. Optimal designs of mollusk shells 25. Kamat S, Su X, Ballarini R, et al. Structural basis for the frac- from bivalves to snails. Sci Rep 2017;7:42445. ture toughness of the shell of the conch Strombus gigas. Na- 5. Aguilera F, McDougall C, Degnan BM. Co-option and de novo ture 2000;405:1036–40. gene evolution underlie molluscan shell diversity. Mol Biol 26. Pokroy B, Zolotoyabko E. Microstructure of natural Evol 2017;34:779–92. plywood-like ceramics: a study by high-resolution electron 6. Liang J, Xie J, Gao J et al. Identification and characteriza- microscopy and energy-variable X-ray diffraction. J Mater tion of the lysine-rich matrix protein family in Pinctada Chem 2003;13:682–8. fucata: indicative of roles in shell formation. Mar Biotech 27. Rodriguez-Navarro AB, Checa A, Willinger M-G, et al. 2016;18:645–58. Crystallographic relationships in the crossed lamellar mi- 7. Marxen JC, Hammer M, Gehrke T, et al. Carbohydrates of crostructure of the shell of the gastropod Conus marmoreus. the organic shell matrix and the shell-forming tissue of the Acta Biomater 2012;8:830–5. snail Biomphalaria glabrata (Say). Biol Bull 1998;194:231–40. 28. Marin F, Luquet G, Marie B, et al. Molluscan shell proteins: 8. Arias JL, Fernandez ´ MS. Polysaccharides and proteoglycans primary structure, origin, and evolution. Curr Top Dev Biol in calcium carbonate-based biomineralization. Chem Rev 2008;80:209–76. 2008;108:4475–82. 29. Pavat C, Zanella-Cleon ´ I, Becchi M, et al. The shell matrix 9. Farre B, Dauphin Y. Lipids from the nacreous and prismatic of the pulmonate land snail Helix aspersa maxima. Comp layers of two Pteriomorpha mollusc shells. Comp Biochem Physiol B 2009;152:103–9. 10. Zhao M, He M, Huang X, et al. Functional characteriza- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 14 Modularity and asymmetry of the molluscan mantle Biochem Physiol B 2012;161:303–14. 50. Feng D, Li Q, Yu H, et al. Identification of conserved pro- 30. Vizca´ıno JA, Csordas A, Del-Toro N, et al. 2016 update of teins from diverse shell matrix proteome in Crassostrea gi- the PRIDE database and its related tools. Nucleic Acids Res gas: characterization of genetic bases regulating shell for- 2016;44:D447–56. mation. Sci Rep 2017;7:45754. 31. Jackson DJ, Ellemor N, Degnan BM. Correlating gene ex- 51. Arivalagan J, Marie B, Sleight VA, et al. Shell matrix pro- pression with larval competence, and the effect of age and teins of the clam, Mya truncata: roles beyond shell formation parentage on metamorphosis in the tropical abalone Halio- through proteomic study. Mar Geonomics 2016;27:69–74. tis asinina. Mar Biol 2005;147:681–97. 52. Liu C, Li S, Kong J, et al. In-depth proteomic analysis of shell 32. Gasteiger E, Gattiker A, Hoogland C, et al. ExPASy: the pro- matrix proteins of Pinctada fucata. Sci Rep 2015;5:17269. teomics server for in-depth protein knowledge and analy- 53. Liao Z, Bao L-F, Fan M-H, et al. In-depth proteomic analy- sis. Nuc Acids Res 2003;31:3784–8. sis of nacre, prism, and myostracum of Mytilus shell. J Pro- 33. Petersen TN, Brunak S, von Heijne G, et al. SignalP 4.0: dis- teomics 2015;122:26–40. criminating signal peptides from transmembrane regions. 54. Jackson DJ, Mann K, Haussermann ¨ V, et al. The Magellania Nat Meth 2011;8:785–6. venosa biomineralizing proteome: a window into brachio- 34. Gasteiger E, Hoogland C, Gattiker A, et al. Protein identi- pod shell evolution. Gen Biol Evol 2015;7:1349–62. fication and analysis tools on the ExPASy server. Humana 55. Mann K, Poustka AJ, Mann M. In-depth, high-accuracy Press. Totowa, NJ; 2005. proteomics of sea urchin tooth organic matrix. Prot Sci 35. Jorda J, Kajava AV. T-REKS: identification of tandem REpeats 2008;6:33. in sequences with a K-meanS based algorithm. Bioinfor- 56. Ramos-Silva P, Kaandorp J, Herbst F, et al. The skeleton of matics 2009;25:2632–8. the staghorn coral Acropora millepora: molecular and struc- 36. Altschul SF, Gish W, Miller W, et al. Basic local alignment tural characterization. PLoS One 2014;9:e97454. search tool. J Mol Biol 1990;215:403–10. 57. Hohagen J, Jackson DJ. An ancient process in a modern mol- 37. Marchler-Bauer A, Derbyshire MK, Gonzales NR, et al. lusc: early development of the shell in Lymnaea stagnalis. CDD: NCBI’s conserved domain database. Nuc Acids Res BMC Dev Biol 2013;13:27. 2015;43:D222–6. 58. Kniprath E. Zur Ontogenese des Schalenfeldes von Lymnaea 38. Mitchell A, Chang H-Y, Daugherty L, et al. The InterPro pro- stagnalis. Wilhelm Roux’s Archives of Developmental Biol- tein families database: the classification resource after 15 ogy 1977;181:11–30. years. Nuc Acids Res 2015;43:D213–21. 59. Timmermans LPM. Studies on shell formation in molluscs. 39. Steentoft C, Vakhrushev SY, Joshi HJ, et al. Precision map- Neth J Zool 1969;19:413–523. ping of the human O-GalNAc glycoproteome through Sim- 60. Saleuddin ASM. An electron microscopic study on the for- pleCell technology. EMBO J 2013;32:1478–88. mation of the periostracum in Helisoma (Mollusca). Calcif 40. Gene Structure Draw. [ http://www.compgen.uni-muenster Tissue Int 1975;18:297–310. .de/tools/strdraw/index.hbi?]. Accessed 10th April 2018 61. Kniprath E. Formation and structure of the periostracum in 41. Davison A, McDowell GS, Holden JM, et al. Formin is asso- Lymnaea stagnalis. Cal Tiss Res 1972;9:260–71. ciated with left-right asymmetry in the pond snail and the 62. Bubel A. An electron-microscope study of periostracum for- frog. Curr Biol 2016;26:654–60. mation in some marine bivalves. I. The origin of the perios- ˝ ´ 42. Kapustin Y, Tozser J, Souvorov A, et al. Splign: algorithms tracum. Mar Biol 1973;20:213–21. for computing spliced alignments with identification of 63. Bevelander G, Nakahara H. An electron microscope study of paralogs. Biology Direct 2008;3:20. the formation of the periostracum of Macrocallista maculata. 43. Yan T, Yoo D, Berardini TZ, et al. PatMatch: a program for Cal Tiss Res 1967;1:55–67. finding patterns in peptide and nucleotide sequences. Nuc 64. Bubel A. An electron-microscope study of periostracum for- Acids Res 2005;33:W262–6. mation in some marine bivalves. II. The cells lining the pe- 44. Cerveau N, Jackson DJ. Combining independent de novo as- riostracal groove. Mar Biol 1973;20:222–34. semblies optimizes the coding transcriptome for noncon- 65. Ast G. How did alternative splicing evolve. Nat Rev Gen ventional model eukaryotic organisms. BMC Bioinformatics 2004;5:773–82. 2016;17:525. 66. Weiner S, Traub W, Parker SB. Macromolecules in mollusc 45. Herlitze I, Marie B, Marin F, et al. Supporting data for shells and their functions in biomineralization [and Discus- “Molecular modularity and asymmetry of the molluscan sion]. Philosophical Transactions of the Royal Society B: Bi- mantle revealed by a gene expression atlas.” GigaScience ological Sciences 1984;304:425–34. Database 2018. http://dx.doi.org/10.5524/100436. 67. Addadi L, Weiner S. Interactions between acidic proteins 46. Jackson DJ, Herlitze I, Hohagen J. A whole mount in situ hy- and crystals: stereochemical requirements in biomineral- bridization method for the gastropod mollusc Lymnaea stag- ization. Proc Natl Acad Sci 1985;82:4110–4. nalis. J Vis Exp 2016;(109). doi: 10.3791/53968. 68. Kingsley EP, Chan XY, Duan Y, et al. Widespread RNA seg- 47. Marie B, Joubert C, Tayalea ´ A, et al. Different secretory reper- regation in a spiralian embryo. Evol Dev 2007;9:527–39. toires control the biomineralization processes of prism and 69. Lambert JD, Nagy LM. Asymmetric inheritance of centroso- nacre deposition of the pearl oyster shell. Proc Natl Acad mally localized mRNAs during embryonic cleavages. Nature Sci 2012;109:20986–91. 2002;420:682–6. 48. Mann K, Cerveau N, Gummich M, et al., In-depth proteomic 70. Bartel DP, Chen C-Z. Micromanagers of gene expression: the analyses of Haliotis laevigata (greenlip abalone) nacre and potentially widespread influence of metazoan microRNAs. prismatic organic shell matrix, Proteome Science, 2018. Nat Rev Gen 2004;5:396–400. Submitted. 71. Zheng Z, Jiao Y, Du X, et al. Computational prediction of 49. Mann K, Edsinger E. The Lottia gigantea shell matrix pro- candidate miRNAs and their potential functions in biomin- teome: re-analysis including MaxQuant iBAQ quantitation eralization in pearl oyster Pinctada martensii. Saudi Journal and phosphoproteome analysis. Prot Sci 2014;12:28. of Biological Sciences 2016;23:372–8. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 15 72. Jiao Y, Zheng Z, Tian R, et al. MicroRNA, pm-miR-2305, par- mation. Comparative Biochemistry and Physiology Part A: ticipates in nacre formation by targeting pearlin in pearl Physiology 1980;66:687–90. oyster Pinctada martensii. Int J Molec Sci 2015;16:21442–53. 94. Dogterom AA, van Loenhout H, van der Schors RC. The ef- 73. Brusca RC, Brusca GJ. Invertebrates. 2nd edition. Sinauer fect of the growth hormone of Lymnaea stagnalis on shell Associates: Sunderland, Massachusetts; 2002. calcification. Gen Comp Endocrinol 1979; 39:63–68. 74. Rosenberg G. A new critical estimate of named species- 95. Dogterom AA, van der Schors RC. The effect of the growth level diversity of the recent Mollusca. Am Malacol Bull hormone of Lymnaea stagnalis on (bi) carbonate movements, 2014;32:308–22. especially with regard to shell formation. Gen Comp En- 75. Hochner B, Glanzman DL. Evolution of highly diverse forms docrinol 1980;41:334–9. of behavior in molluscs. Curr Biol 2016;26:R965–71. 96. Weiss IM, Gohring ¨ W, Fritz M, et al. Perlustrin, a Haliotis lae- 76. Williams TA, Nagy LM. Developmental modularity and the vigata (abalone) nacre protein, is homologous to the insulin- evolutionary diversification of arthropod limbs. Journal of like growth factor binding protein N-terminal module of Experimental Zoology Part A: Ecological Genetics and Phys- vertebrates. Biochem Biophys Res Commun 2001;285:244– iology 2001;291:241–57. 9. 77. Peel AD, Chipman AD, Akam M. Arthropod segmentation: 97. Zhang G, Willemin AS, Brion A, et al. A new method for the beyond the Drosophila paradigm. Nat Rev Gen 2005;6:905– separation and purification of the osteogenic compounds of 16. nacre ethanol soluble matrix. J Struct Biol 2016;196:127–37. 78. Tautz D. Segmentation. Dev Cell 2004;7:301–12. 98. Marin F, Luquet G. Molluscan shell proteins. CR Palevol 79. Raff RA. The shape of life: genes, development, and the evo- 2004;3:469–92. lution of animal form. University of Chicago Press: Chicago, 99. Shen X, Belcher AM, Hansma PK, et al. Molecular cloning Illinois; 1996. and characterization of lustrin A, a matrix protein from 80. Esteve-Altava B. In search of morphological modules: a sys- shell and pearl nacre of Haliotis rufescens. J Biol Chem tematic review. Biol Rev 2017;92:1332–47. 1997;272:32472–81. 81. Eble GJ. Morphological modularity and macroevolution. In: 100. Kocot KM, Aguilera F, McDougall C, et al. Sea shell diversity Callebaut W, Rasskin-Gutman D , eds. Modularity: Under- and rapidly evolving secretomes: insights into the evolution standing the Development and Evolution of Natural Com- of biomineralization. Front Zool 2016;13:23. plex Systems. MIT Press, Cambridge, Massachusetts; 2005. 101. Luo H, Nijveen H. Understanding and identifying amino pp. 221–38. acid repeats. Brief Bioinform 2014;15:582–91. 82. McDougall C, Green K, Jackson DJ, et al. Ultrastructure of the 102. Alba` M, Tompa P, Veitia R. Amino acid repeats and the mantle of the gastropod Haliotis asinina and mechanisms of structure and evolution of proteins. In: Volff J-N (ed) Basel, shell regionalization. Cells Tissues Organs 2011;194:103–7. Karger. Gene and Protein Evolution. Karger Publishers; 2007. 83. Jolly C, Berland S, Milet C, et al. Zonal localization of shell pp. 119–30. matrix proteins in mantle of Haliotis tuberculata (Mollusca, 103. Alberts B, Johnson A, Lewis J, et al. Molecular Biology of the Gastropoda). Mar Biotech 2004;6:541–51. Cell. 5th revised edition. Garland Science: New York; 2007. 84. Sleight VA, Marie B, Jackson DJ, et al. An Antarctic mollus- 104. McDougall C, Woodcroft BJ, Degnan BM. The widespread can biomineralisation tool-kit. Sci Rep 2016;6:36978. prevalence and functional significance of silk-like struc- 85. Waite JH. The phylogeny and chemical diversity of quinone- tural proteins in metazoan biological materials. PLoS One tanned glues and varnishes. Comp Biochem Physiol B 2016;11:e0159128. 1990;97:19–29. 105. Le Roy N, Jackson D, Marie B, et al. Carbonic anhydrase and 86. Waite JH. Evidence for the mode of sclerotization in a mol- metazoan biocalcification: a focus on molluscs. In: Marin luscan periostracum. Comp Biochem Physiol B 1977;58:157– F, Brummer ¨ F, Checa A, Furtos G, Lesci IG, Siller L , eds. 62. Biomineralization: From Fundamentals to Biomaterials & 87. Waite JH, Wilbur KM. Phenoloxidase in the periostracum Environmental Issues. Trans Tech Publications Ltd; 2015. of the marine bivalve Modiolus demissus Dillwyn. J Exp Zool pp. 151–7. 1976;195:359–67. 106. Sun X, Yang A, Wu B, et al. Characterization of the man- 88. Waite JH. Quinone-tanned scleroproteins. In: Saleuddin tle transcriptome of Yesso scallop (Patinopecten yessoensis): ASM, Wilbur KM , eds. Physiology. Academic Press: New identification of genes potentially involved in biomineral- York; 1983. pp. 467–504. ization and pigmentation. PLoS One 2015;10:e0122967. 89. Michon T, Chenu M, Kellershon N, et al. Horseradish 107. Jackson DJ, McDougall C, Woodcroft B, et al. Parallel evolu- peroxidase oxidation of tyrosine-containing peptides and tion of nacre building gene sets in molluscs. Mol Biol Evol their subsequent polymerization: a kinetic study. Biochem 2010;27:591–608. 1997;36:8504–13. 108. Sleight VA, Thorne MAS, Peck LS, et al. Characterisation 90. Halaby DM, Mornon JPE. The immunoglobulin superfamily: of the mantle transcriptome and biomineralisation genes an insight on its tissular, species, and functional diversity. in the blunt-gaper clam, Mya truncata. Mar Geonomics J Mol Evol 1998;46:389–400. 2016;27:47–55. 91. Honegger B, Galic M, Kohler ¨ K, et al. Imp-L2, a putative ho- 109. Arivalagan J, Yarra T, Marie B, et al. Insights from the shell molog of vertebrate IGF-binding protein 7, counteracts in- proteome: biomineralization to adaptation. Mol Biol Evol sulin signaling in Drosophila and is essential for starvation 2017;34(1):66–77. resistance. Journal of Biology 2008;7:1. 110. Suzuki M, Kogure T, Weiner S, et al. Formation of aragonite 92. Dogterom AA, Doderer A. A hormone dependent calcium- crystals in the crossed lamellar microstructure of limpet binding protein in the mantle edge of the freshwater snail shells. Crystal Growth & Design 2011;11:4850–9. Lymnaea stagnalis. Calcif Tissue Int 1981;33:505–8. 111. Mann K, Edsinger-Gonzales E, Mann M. In-depth proteomic 93. Dogterom AA, Jentjens T. The effect of the growth hormone analysis of a mollusc shell: acid-soluble and acid-insoluble of the pond snail Lymnaea stagnalis on periostracum for- matrix of the limpet Lottia gigantea. Prot Sci 2012;10:28. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png GigaScience Oxford University Press

Molecular modularity and asymmetry of the molluscan mantle revealed by a gene expression atlas

Free
15 pages

Loading next page...
 
/lp/ou_press/molecular-modularity-and-asymmetry-of-the-molluscan-mantle-revealed-by-nWc8YlwSUk
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press.
eISSN
2047-217X
D.O.I.
10.1093/gigascience/giy056
Publisher site
See Article on Publisher Site

Abstract

Background: Conchiferan molluscs construct a biocalcified shell that likely supported much of their evolutionary success. However, beyond broad proteomic and transcriptomic surveys of molluscan shells and the shell-forming mantle tissue, little is known of the spatial and ontogenetic regulation of shell fabrication. In addition, most efforts have been focused on species that deposit nacre, which is at odds with the majority of conchiferan species that fabricate shells using a crossed-lamellar microstructure, sensu lato. Results: By combining proteomic and transcriptomic sequencing with in situ hybridization we have identified a suite of gene products associated with the production of the crossed-lamellar shell in Lymnaea stagnalis. With this spatial expression data we are able to generate novel hypotheses of how the adult mantle tissue coordinates the deposition of the calcified shell. These hypotheses include functional roles for unusual and otherwise difficult-to-study proteins such as those containing repetitive low-complexity domains. The spatial expression readouts of shell-forming genes also reveal cryptic patterns of asymmetry and modularity in the shell-forming cells of larvae and adult mantle tissue. Conclusions: This molecular modularity of the shell-forming mantle tissue hints at intimate associations between structure, function, and evolvability and may provide an elegant explanation for the evolutionary success of the second largest phylum among the Metazoa. Keywords: mollusc; biomineralization; gene expression; asymmetry; modularity; evolution; shell; matrix protein; transcriptome; alternative splicing throughput study of these molecules are well established and Introduction are technically straightforward. Much progress has been made Due to its evolutionary significance, impressive materials prop- in identifying the components of the shell-forming proteome erties, and aesthetic beauty, the molluscan shell has long re- from a variety of gastropod and (primarily) bivalve species (e.g., ceived attention from a wide variety of scientific disciplines [ 1– [10–17]) largely due to advances in nucleic acid sequencing tech- 6]. Although molluscan shells are constructed from a complex nologies that, when coupled with high-throughput proteomic mixture of calcium carbonate, carbohydrates [7, 8], and lipids [9], surveys of the biomineralized proteome, allow for the rapid gen- proteins have received the most attention arguably for two main eration of extensive lists of shell-associated proteins. However, reasons: they can provide deep insight into the evolutionary his- without further validation, genes identified in this way should tory of this composite structure and the techniques for the high- only be considered as candidate biomineralizing molecules. This Received: 27 November 2017; Revised: 20 March 2018; Accepted: 9 May 2018 The Author(s) 2018. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 2 Modularity and asymmetry of the molluscan mantle problem is often compounded by the fact that these proteins of- Methods ten share little to no sequence similarity with proteins from con- Cultivation of adult L. stagnalis ventional model organisms, making any inference about their function very difficult. This bottleneck represents one of the cur- Lymnaea stagnalis (Mollusca; Gastropoda; Heterobranchia; Eu- rent major challenges for scientists interested in understanding thyneura; Panpulmonata; Hygrophila; Lymnaeoidea; Lymnaei- the mechanisms and evolution of molluscan biomineral forma- dae; Lymnaea) does not fall under the German Animal Protection tion. While knock-down of individual shell-forming genes via Act §8 and is listed as “least concern” under the International RNAi has been reported in some species of bivalves [10, 18, 19], Union for Conservation of Nature (IUCN’s) list of threatened these assays are rarely validated by protein immuno-detection, species. This work was therefore exempt from the University and levels of penetrance or statistical quantitation of knock- ¨ of Gottingen Ethics Committee. Adult specimens of L. stagnalis down phenotypes are rarely reported. derived from animals originally collected from the Northeimer ◦ ◦ Another approach to gain insight into the function of shell- Seenplatte, Germany (51 43’ 26.5368’, 9 57’ 24.75’), and from a forming genes is to characterize their spatial expression pat- pond on the north campus of the University of Gotting ¨ en, Ger- ◦   ◦ terns in situ. We previously adopted this approach in the trop- many (51 33 23.727 ,9 57 25.617 ), were kept in a stand-alone ical abalone Haliotis asinina with a Sanger Expressed Sequence V30 unit (Aqua Schwarz) in demineralized water supplemented Tag (EST) dataset and characterized the spatial expression pat- with ReMineral+ (Dennerle, 7036) to a conductivity of 200–220 terns of more than 20 putative shell-forming genes in juvenile μS, 23 C, apHof7.5 to 7.9and a16:8light regimen.Fiveto10 snails [16]. While a spatial expression pattern in the mantle is individuals were kept in 3- or 5-liter boxes under a constant and not direct evidence of a functional role in calcification, we were low-flow rate. Snails were fed ad libitum with lettuce and a va- able to assign putative functions to genes involved in shell pig- riety of other vegetables. Under this regime, adult snails lay egg mentation [16] and ecological and mineralogical transitions [20]. masses year round. Here, we have combined an next-generation sequencing (NGS) transcriptome analysis of adult mantle tissue with a proteomic Organic matrix extraction from calcified shells survey of the adult shell of the freshwater pulmonate gastro- pod Lymnaea stagnalis in order to both compare the resulting Twelve shells of adult L. stagnalis (larger than 3–4 cm in length) data with other similar datasets and to generate the first in situ- were selected for extraction. Prior to further treatment, the col- validated ontogenetic transcriptome-scale dataset for a species umella was delicately cut and removed from each shell. Super- that forms the most common molluscan shell microstructure, ficial organic contaminants were removed by incubating pooled crossed lamellar [21–23]. The high-order structure of crossed- shell fragments in 10% v/v sodium hypochlorite (NaOCl) for 24 lamella, which allows it to efficiently deflect and arrest cracks hours. Fragments were then thoroughly rinsed with water and [24–27], coupled with its extremely low organic content (typically subsequently ground into a fine powder that was sieved ( >200 <0.5%) has been suggested to be one reason it has enjoyed so μM). This biomineral powder was incubated in 5% v/v NaOCl for much evolutionary success (reviewed in [28]). Recent proteomic 5 hours and rinsed twice with MilliQ water. Powdered samples studies have been reported for molluscs that build crossed- were decalcified overnight at 4 C in cold 5% v/v acetic acid that lamellae shells (Helix aspersa maxima [29]and Cepaea nemoralis was slowly added by an automated burette (Titronic Universal, [14]), however, those studies did not conduct any spatial ex- Mainz, Germany) at a flow rate of 100 μLevery 5seconds.The pression analyses for the shell-forming proteins they identified. solution (final pH ∼4.2) was centrifuged at 3,900 g for 30 min- In addition to characterizing the spatial expression patterns of utes. The resulting acid-insoluble matrix (AIM) pellet was rinsed more than 30 shell-forming candidates in the adult mantle tis- six times with MilliQ water, freeze-dried, and weighed. The su- sue of L. stagnalis, we have also investigated their spatial expres- pernatant containing acetic acid-soluble matrix (ASM) was fil- sion patterns during development. tered (Millipore, 5 μM) and concentrated in an Amicon ultra- Our analyses hint at the potential pleiotropic nature of some filtration stirred cell (model 8400, 400 mL) on a Millipore mem- of these shell-forming genes and highlight the dynamic and brane (10 kDa cutoff). The final solution ( >5 mL) was extensively asymmetric natures of their spatial regulation. A striking result dialyzed against 1 L of MilliQ water (six water changes) before of our analyses in the adult mantle tissue is the degree of spa- being freeze-dried and weighed. tial modularity displayed by distinct sets of genes. This general observation may contribute to an explanation of why the mol- Sample preparation for proteomic analysis luscan shell is apparently so evolvable. With the availability of a draft L. stagnalis genome and transcriptome data from a va- In-solution digestion of unfractionated ASM (0.1 mg) and AIM (1 riety of adult tissues, we have also investigated the genetic ar- mg) material was performed as follows. Samples were reduced chitectures of our biomineralization candidates and explored to with 50 μL of 10 mM dithiothreitol in 50 mM ammonium bicar- what extent alternative splicing plays a role in shell formation bonate (NH HCO ) for 30 minutes at 50 C. Alkylation was per- 4 3 in L. stagnalis. These genes can also be compared with similar formed with 50 μL of 100 mM iodoacetamide in 50 mM NH HCO 4 3 datasets from distantly related molluscs that build shells with for 30 minutes at room temperature in the dark. The solution alternative polymorphs of calcium carbonate (calcite vs. arag- was then treated with 1 μg of trypsin (proteomic grade; Promega) onite) and textures (prismatic vs. nacreous vs. crossed lamel- in 10 μLof50mMNH HCO overnight at 37 C. Samples were 4 3 lae). Such comparisons can generate testable hypotheses regard- then dried in a vacuum concentrator and re-suspended in 30 μL ing which components of the shell-forming toolkit contribute to of 0.1% trifluoroacetic acid and 2% acetonitrile (CH CNCN). these differences and which components are required for more fundamental aspects of shell formation. Peptide fractionation and data acquisition Mass spectrometry (MS) was performed using a Q-Star XL nanospray quadrupole/time-of-flight tandem mass spectrom- eter, nanospray-Qq-TOF-MS/MS (Applied Biosystems, Villebon- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 3 sur-Yvette, France) coupled to an online nanoLC system (Ulti- quences were generated using Gene Structure Draw [40]. Intron- mate Famos Switchos from Dionex, Amsterdam, The Nether- exon boundaries were mapped to a draft genome of L. stagnalis lands). One microliter of each sample was loaded onto a trap originally reported in [41] using Splign [42]. Similar transcripts column (PepMap100 C18; 5 μm; 100 A; 300 μM x 5 mm; Dionex), were retrieved from the assembled transcriptomes of mantle washed for 3 minutes at 25 μL/min with 0.05% trifluoroacetic zones 1–5 combined, mantle zone 5 alone, cephalic tentacle, acid/2% acetonitrile, then eluted onto a C18 reverse phase col- cephalic lobe, central nervous system (CNS), foot, buccal mass, umn (PepMap100 C18; 3 μm; 100 A; 75 μM x 150 mm; Dionex). and larval stages 42 hours post first cleavage (hpfc), 52 hpfc, Peptides were separated at a flow rate of 0.300 μL/min with a and 67 hpfc using BLASTn searches (see below for NGS details). linear gradient of 5–80% acetonitrile in 0.1% formic acid over 120 All transcripts with complete open reading frames were consid- minutes. MS data were acquired automatically using ANALYST ered. Only candidates yielding an mRNA coverage of >98% and QS 1.1 software (Applied Biosystems). Following an MS survey an overall identity of >98% are documented. Scaled schematics scan over m/z 400–1600 range, MS/MS spectra were sequentially of the gene architecture were generated using Gene Structure and dynamically acquired for the three most intense ions over Draw [40]. Protein patterns were searched for using a modified m/z 65–2000 range. The collision energy was set by the software local installation of PatMatch [43]. according to the charge and mass of the precursor ion. MS and MS/MS data were recalibrated using internal reference ions from NGS sequencing a trypsin autolysis peptide at m/z 842.51 [M + H] and m/z 421.76 2+ [M + 2H] . Total RNA was extracted from the mantle edge and the proxi- mal mantle tissue of a single adult L. stagnalis using TriReagent following the manufacturer’s instructions. The resulting RNA Mass spectrometry data analysis was processed by the sequencing center at the IKMB at the Protein identification was performed using the MASCOT search University of Kiel (Germany). Paired end, stranded TrueSeq RNA engine (version 2.1; Matrix Science, London, UK) against trans- libraries were constructed and sequenced for 101 bases from lations in all six frames of our mantle transcriptomes, which both ends using the Illumina HiSeq2000 platform (Illumina, possessed Benchmarking Universal Single-Copy Orthologs com- CA, USA). More than 99 million and 100 million reads were pleteness scores of >98% (see below). Liquid chromatography generated from each of these libraries, respectively. These (LC)-MS/MS data were searched using carbamido-methylation Illumina reads were processed using our pipeline as previously as a fixed modification and methionine oxidation as a variable described [44]. Briefly, raw reads were adapter trimmed and modification. The peptide mass and fragment ion tolerances quality filtered using Trimmomatic V0.32. Filtered reads were were set to 0.5 Da. The peptide hits (protein score >50; false then assembled with Trinity V2.0.3, CLC Genomics Workbench discovery rate <0.05; 1 missed cleavage allowed) were manu- de novo assembler (V8.5), and IDBA-tran. The resulting assem- ally confirmed by the observation of the raw LC-MS/MS spectra blies were then merged and filtered for redundancy using our with ANALYST QS software (version 1.1). Quality criteria were pipeline [44]. Mantle transcriptome assemblies and cDNA and the peptide MS value, the assignment of major peaks to un- protein translations of the 34 shell-forming genes are available interrupted y- and b-ion series of at least three to four con- in the GigaScience Database, GigaDB [45]. In addition, tran- secutive amino acids, and the match with the de novo inter- scriptomes from five adult tissues (cephalic tentacle, cephalic pretations proposed by the software. All MS data has been de- lobe, CNS, foot, and buccal mass) and three larval stages (42 posited with the ProteomeXchange Consortium via PRIDE [30] hpfc, 52 hpfc, and 67 hpfc) were sequenced and assembled as with the dataset identifiers PXD008547 and 10.6019/PXD008547. described above. These transcriptomes were used to assess Shell-forming candidates Lstag-sfc-7, Lstag-sfc-8, and Lstag-sfc-9 the tissue-specific alternative splicing characteristics of all were bioinformatically selected for analysis based on the pres- shell-forming genes. All raw NGS data has been deposited in ence of a signal peptide and their glycine-rich sequences (i.e., the Sequence Read Archive (SRA) with BioSample accession they were not detected using the proteomic methods described numbers SAMN08117214, SAMN08117215, SAMN08709370, above). SAMN08709371, SAMN08709372, SAMN08709373, SAMN08709374, SAMN08709375, SAMN08709376, and SAMN08709377. Bioinformatic analysis of protein sequences Using the peptides identified from the proteomic survey de- In situ hybridization on whole mounts and sections scribed above, partial or, in most cases, full length coding se- quences were isolated by standard or RACE polymerase chain Larvae were prepared for whole mount in situ hybridization as reaction (PCR) as described in [31]. In some cases, Illumina tran- described in [46]. Sections (10 μM) were taken from L. stag- nalis (shell length 10–50 mm) that had been fixed in formalde- scriptome data (see below) were used to clarify the putative com- plete mRNA. Open reading frames were translated with the Ex- hyde for 1 hour and embedded in paraffin. Riboprobes were PASy translate tool [32]. Protein sequences were searched for sig- prepared as described in [16] and were used at concentrations nal sequences with SignalP 4.1 [33]. The theoretical pI, amino of 100–500 ng/mL. Whole mounts and tissue sections were acid composition, and number of amino acids were determined processed for hybridization, the color reaction developed, and using the ExPasy ProtParam tool [34]. Tandem repeats were iden- photo-documented as described in [46]. tified with the T-REKS tool [ 35]. Sequence similarities searches were performed with the Basic Local Alignment Search Tool Comparisons of molluscan shell-forming proteomes (BLAST) algorithm [36] with tBLASTx against nr and dbEST and BLASTx against SwissProt. Domain searches were performed BLASTp-based comparisons of the L. stagnalis shell proteome with CD search [37]. Molecular function was predicted with In- were performed against a variety of calcifying proteomes re- terProScan [38]. GalNAc O-glycosylation sites were predicted us- ported in a wide phylogenetic range of metazoans as described ing the NetOGlyc 4.0 Server [39]. Scaled schematics of protein se- in [14]. These included 42 proteins from the oyster Pinctada max- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 4 Modularity and asymmetry of the molluscan mantle ima reported in [47]; 78 proteins from the oyster Pinctada margar- A brief morphological description of L. stagnalis shell itifera reported in [47]; 94 proteins from the abalone Haliotis asin- ontogeny and the adult mantle ina reported in [17]and [16]; 80 proteins from the abalone H. lae- We previously described the ontogeny of the shell gland and vigata reported in [48]; 63 proteins from the limpet Lottia gigantea shell field in L. stagnalis [57]. In order to aid the interpretation reported in [49]; 53 proteins from the oyster Crassostrea gigas re- of our in situ patterns, the following is a summary of the main ported in [50]; 71 proteins from the mussel Mya truncata reported developmental stages that we focused on. The first visible sign in [51]; 59 proteins from the grove snail Cepaea nemoralis reported of differentiation of the shell-forming tissue in L. stagnalis is a in [14]; 44 proteins from the oyster Pinctada fucata reported in [52]; thickening of the dorsal ectoderm that begins at approximately 53 proteins from the mussel Mytilus coruscus reported in [53]; 66 29 hpfc [57, 58]. These cells subsequently invaginate and by 2 proteins from the brachiopod Magellania venosa reported in [54]; days post first cleavage (dpfc) a clearly visible “shell gland” is 139 proteins from the sea urchin Strongylocentrotus purpuratus re- present [57, 58]. By 3 dpfc, the shell gland has formed a sealed lu- ported in [55]; and 37 proteins from the coral Acropora millepora men and displays the first signs of outward signs of asymmetry reported in [56]. [57]. The marginal cells that border the shell gland remain unin- vaginated and form a ring-like structure, the rosette [2]. During Analysis of the saccharide moieties of the shell matrix this time, the first extracellular organic material is secreted and is clearly visible by Scanning Electron Microscopy (SEM) (Fig. 1; The monosaccharide content of AIM and ASM was obtained [57]). By 3 dpfc, the shell gland has evaginated to form the shell by suspension and homogenization (vortex and ultrasound) of field. The former rosette cells remain highly elongated while the lyophilates in 2 M trifluoroacetic acid and subsequent hydrol- central cells take on a low columnar appearance. Over the next ysis at 105 C for 4 hours under a nitrogen atmosphere. This several days, the shell field continues to expand until it has over- treatment allows for the release of most monosaccharides from grown the visceral mass and will eventually become the adult complex mixtures, except sialic acids, which are destroyed, mantle tissue [2, 57, 58]. and the acetylated forms of glucosamine and galactosamine, The adult mantle covers the inner surface of the shell and which are converted to their respective non-acetylated forms. is responsible for shell growth and repair. The free edge of the Samples were then centrifuged for 5 minutes at 15,000g and mantle is responsible for the growth of the outer lip of the shell. evaporated to dryness (using a SpeedVac) before being dis- Timmermans conducted an extensive histochemical character- solved in 100 μL of 20 mM sodium hydroxide and homoge- ization of the mantle tissue of L. stagnalis and was able to cate- nized. After a short centrifugation (2 minutes), 80 μLofsuper- gorize the free edge of the adult mantle into six distinct zones natant was injected into the chromatograph system. The neu- based on their morphology, enzymatic activities, and biochem- tral, amino, and acidic sugar contents of hydrolysates were de- ical signatures [59]. We largely follow this categorization of the termined using high-pressure anion exchange–pulsed ampero- adult mantle tissue. Parallel to the mantle edge runs the mantle metric detection on a CarboPac PA 100 column (Dionex Corp., groove (also known as the pallial groove) defined as zone 1 (Fig. Sunnyvale, CA, USA). As blank controls, non-hydrolyzed AIMs 2). Several high-resolution microscopy and histological studies were analyzed in order to detect potential free monosaccha- on a variety of molluscs have demonstrated that it is from within rides that may lead to an over-representation of some sugar the pallial groove that the periostracum is formed and secreted residues. [59–64]. We detected a sub-regionalization of the pallial groove (zone 1) into proximal and distal zones. Immediately adjacent to the pallial groove is a broad region of high columnar cells Results referred to by Timmermans [59] as the “belt” that can be sub- Proteomic analysis of the biomineralized matrix of L. divided into three distinct zones (zones 2–4). Zone 2 is imme- stagnalis shells diately adjacent to the posterior wall of the pallial groove and comprises the anterior (or distal) portion of the belt (Fig. 2). Zone More than 1,230 peptides were analyzed by High-performance 3 consists of the posterior portion of the belt, while zone 4 rep- Liquid Chromatography (HPLC)-MS and subsequently used for resents the transitional zone between the high columnar cells protein identification using Mascot against our translated man- of the belt proper and the more posterior low columnar cells of tle transcriptomes. Of these 1,230 peptides, 329 returned sig- the outer epithelium, which comprise zone 5 (Fig. 2)[59]. nificant matches. From these 329 matches, a total of 40 shell- forming candidate transcripts were identified (see Additional file 1). Of these 40 gene products, 31 (78%) could be cloned and ex- Spatial expression patterns and molecular features of hibit in situ hybridization signals compatible with a role in shell shell-forming candidate genes formation (either in larval stages and/or in the adult mantle tis- We performed in situ hybridization for 34 distinct shell-forming sue). Seven of these 40 candidates (18%) could be cloned from genes on four distinct developmental stages and on adult man- L. stagnalis cDNA but did not produce a positive or consistent in tle tissue. The detailed results of these analyses are presented situ signal in any tissue. Three of the 40 candidate genes (8%) in Additional files 2–35, with an extensive summary presented could not be amplified by gene-specific PCR or RACE PCR. In ad- in Additional file 36 (the raw image files that constitute these dition to the 31 proteomically identified candidates that gener- figures are available in the associated GigaDB repository [ 45]. In ated positive in situ signals, three candidates that were identified Fig. 1 (for larvae) and Fig. 2 (for adult mantle tissue) we present via in silico methods (based purely on the presence of a signal se- a selection of these results that highlight some prominent fea- quence and their glycine-rich protein sequences) also generated tures of these expression patterns. In trochophore and veliger in situ signals compatible with a role in shell formation and are larval stages (2–6 dpfc), all genes could be categorized either as reported here. being: expressed in cells that symmetrically or asymmetrically border the shell gland or shell field (15/34); in cells that lay within the shell gland or shell field (9/34); a pattern that did not fit into Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 5 Figure 1: Overview of four developmental stages and representative shell-forming gene expression patterns in L. stagnalis. The first two rows provide a set of reference SEM images and adult shells (top-right-most panel) against which the images of the in situ results can be oriented. All in situ panels are from a dorsal view except the right-most column, which is from a ventral view. Here, we present the expression patterns of a selection of five shell-forming genes. These include genes with expression patterns in shell-forming cells that display evidence of symmetry (sfc-5), right asymmetry (sfc-1), left-asymmetry (sfc-17), expression entirely throughout the shell field and dorsal mantle epithelium ( sfc-20), and expression in additional non-shell-forming cells. This last expression pattern provides evidence of genes involved in shell formation that have pleiotropic functions. The scale bars in the first row are 100 μm. Indicated in the SEM images are the positions of the foot lobe (fl), foot (f), mantle margin (mm), calcified shell (s), stomodeum (st), and insoluble organic material (iom) of the shell. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 6 Modularity and asymmetry of the molluscan mantle Figure 2: Overview of the adult L. stagnalis shell-forming mantle tissue and representative shell-forming gene expression patterns that reveal its molecular modularity. (A) A semi-thin sagittal section of an adult L. stagnalis stained with Giemsa. The foot (f), mantle (m), digestive gland (dg), and radula (r) are indicated. The mantle tissue is a thin sheet of epithelium that covers the dorsal surface of the adult animal and is responsible for fabricating the shell. (B) A magnified view of the red-boxed region in part A reveals some of the cellular morphology of the adult mantle tissue. (C) A schematic representation of the mantle tissue divided into six zones as described by Timmermans [59]. The spatial distribution of enzymatic activities and biochemicals indicated in this schematic are adapted from [59]. We detect a sub-regionalization of the pallial groove (zone 1) into proximal (light green) and distal (dark green) zones. (D) The spatial expression patterns of eight representative shell-forming genes in the adult mantle tissue. The asterisk indicates that sfc-6 was identified using in silico methods rather than proteomic methods. our classification scheme (1/34); or were not expressed in any (10/34); asymmetrically in the outer edge of the mantle (18/34); detectable way (9/34). In later stages (∼7dpfc),all geneswere throughout the entire mantle tissue (2/34); a pattern that did not either: expressed uniformly along the outer edge of the mantle fit into our classification scheme (1/34); or were not expressed in Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 7 Trochophore larva Juvenile Adult mantle dorsal view ventral view sagital section carbonic anhydrase left right malate dehydrogenase ATPase-rich cells Peroxidase RNA rich Peroxidase RNA rich alkaline phosphatase Tyrosine Tyrosinase Tyrosinase zone 5 zone 4 zone 3 zone 2 zone 1 zone 6 Shell-forming Asymmetric Symmetric Left side of Left + right side Entire Right side of Interior Zone 5 Zone 4 Zone 3 Zone 2 Zone 1 Sequence similarity or conserved domain candidate border border outer lip of outer lip mantle outer lip Lstag-sfc-1 right ✓✓✓ Lstag-sfc-2a right Early nodulin-12A (SP) ✓✓✓ Lstag-sfc-2b right ✓✓✓ Lstag-sfc-3 right None ✓✓✓ Lstag-sfc-4 right None ✓✓ Lstag-sfc-5 Animal haem peroxidase (CD) ✓✓ ✓✓ Lstag-sfc-6a* None ✓✓ ✓ Lstag-sfc-6b* None ✓✓ ✓ Lstag-sfc-7* None ✓✓ ✓ Lstag-sfc-8* None ✓✓ Lstag-sfc-9a None ✓✓ ✓ Lstag-sfc-9b None ✓✓ ✓ Lstag-sfc-10a None ✓✓ ✓ Lstag-sfc-10b None ✓✓ ✓ Lstag-sfc-11a None ✓✓ ✓ Lstag-sfc-11b None ✓✓ ✓ Lstag-sfc-11c None ✓✓ ✓ Lstag-sfc-12 left None ✓✓ Lstag-sfc-13 None ✓✓ ✓ Lstag-sfc-14 None ✓✓ ✓ Lstag-sfc-15 None ✓✓ ✓ ✓ ✓ Lstag-sfc-16 None ✓✓✓ Lstag-sfc-17 left None ✓✓ Lstag-sfc-18 Immunoglobulin domain (CD) ?? ? ? ? ? ? ✓✓✓ Lstag-sfc-19 Perlucin-like protein (SP), C-type lectin (CD) Lstag-sfc-20a None ✓✓ ✓ Lstag-sfc-20b None ✓✓ ✓ Lstag-sfc-21 Putative chitin deacetylase-like domain (CD) ✓✓ ✓✓ Lstag-sfc-22 Pif97 Aragonite-binding protein (SP) Lstag-sfc-23a Otoancorin (SP) ✓✓✓✓✓ Lstag-sfc-23b Otoancorin (SP) ✓✓✓✓✓ Lstag-sfc-24a None ✓✓✓ Lstag-sfc-24b None ✓✓✓ Lstag-sfc-25 None ✓✓✓ Lstag-sfc-26 PREDICTED: uncharacterized protein (NR) ✓✓✓ Lstag-sfc-27a PREDICTED: extensin-like isoform X1 (NR) ✓✓✓ Lstag-sfc-27b PREDICTED: formin-like protein 2 (NR) ✓✓✓ Lstag-sfc-28a None ✓✓ ✓ Lstag-sfc-28b None ✓✓ ✓ Lstag-sfc-28c None ✓✓ ✓ Lstag-sfc-29 Galaxin (SP) ✓✓ ✓ Lstag-sfc-30 None ✓✓ ✓ Lstag-sfc-31 SUSHI repeat + short complement-like repeat (CD) ✓ ✓✓✓✓✓ Lstag-sfc-32 Intermediate filament protein (CD) ✓ ✓ ✓✓✓✓✓ Lstag-sfc-33 vWA type A domain + collagen alphaI-XII-like (CD) ✓✓ ????? Lstag-sfc-34 None ✓✓ Figure 3: Summary of the spatial gene expression profiles and conserved features of 34 L. stagnalis shell-forming candidates. Schematically represented in a trochophore larva are genes with an asymmetric expression profile (dark gray), as well as genes expressed broadly across the shell field (light blue). Cells in this r egion of the trochophore are likely to give rise to cells in zone 5 of the adult mantle, and we have maintained that color scheme to suggest this. Although we schematically present a trochophore larva here (2–3 dpfc), the summarized expression patterns also include veliger stages (3–6 dpfc). Cells bordering the larval shell gland and shell field (black ring in the trochophore) are likely to give rise to one or more zones 1–4 in the adult mantle. In juveniles (∼7 dpfc) many genes were expressed in the left, right, or continuously throughout the free edge of the mantle that produces the outer lip of the shell. Question marks indicate expression patterns that could not be categorized according to our scheme. An “x” indicates no expression was detected. A “?” indicates that the expression pattern could not be categorized according to our scheme. The names of enzymes and other molecular features indicated in zones 1–5 on the schematic of the adult mantle are summarized from [59] and [3]. Sequence similarity and conserved domains in the final column of the table are summarized from a number of BLAST searches against SwissProt, the non-redundant N ational Center for Biotechnology Information database, and the Conserved Domain database. See Additional files 40–42 for the results of all BLAST and domain s earches. The lineages of the top BLAST hits are listed in Additional file 43. A version of this figure that includes a more complete summary of the molecular features of each gene is provided in Additional file 36. The asterisks indicate that sfc-6, -7, and -8 were identified using in silico methods rather than proteomic methods (as was the case for all other gene products presented here). Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 alkaline phosphatase 8 Modularity and asymmetry of the molluscan mantle any detectable way (3/34). Finally, in adult mantle tissue, genes 32 (with similarity to C. nemoralis contig 572), which appears were either: expressed in one or more of the five zones described to be an intermediate filament protein. Lstag-sfc-22 (a gene ex- by Timmermans [59] (32/34); a pattern that did not fit into our pressed exclusively in zone 5, Additional file 23) shared relatively classification scheme (1/34); or were not expressed in any de- weak similarity with C. nemoralis contig 821 and shares signifi- tectable way (1/34). We have schematically summarized all of cant sequence similarity with PIF, an aragonite-binding protein these results in Fig. 3. reported to be involved in nacre formation in the oyster P. fucata [19]. Interestingly, of the 12 candidates expressed in the matrix- secreting zone 5, 9 showed similarity with other shell proteins Alternative splicing increases the diversity of (Fig. 4 and Additional files 23–33). Eight of these were shared with shell-forming proteins C. nemoralis (Fig. 4). In contrast, none of the asymmetrically ex- Via alternative splicing of mRNAs, transcripts with a variety pressed or glycine-rich candidates were found in any of the other biomineralizing proteomes (low-complexity filtering was inacti- of functions can be generated from a single genomic locus vated in these comparisons; Fig. 4). [65]. With a draft genome for L. stagnalis available, we were able to perform some preliminary investigations into alternative splicing of our shell-forming candidates. While some candidate Glycosylation patterns of the shell matrix genes displayed the same or very similar exon-splicing patterns in all surveyed tissues (e.g., Additional files 5, 9, 14, 15, and 16), The monosaccharides profiles of ASMs and AIMs extracted from most candidates are apparently alternatively spliced depending adult L. stagnalis shells were peculiar, with less than half of the on the tissue they are expressed in (Additional files 4, 11, 12, 18, dozen standard monosaccharides represented. For ASM, these 19, 22, 23, 25, and 33). Striking examples include Lstag-sfc-21 and include galactosamine, glucosamine, galactose, glucuronic acid, Lstag-sfc-24, which are expressed in many tissues but display and glucose, while AIM lacked glucuronic acid (Table 1). We also significant alternative splicing in the adult mantle (Additional found marked differences in the glycosylation rates of ASMs and files 22 and 25). All splice variants of candidate Lstag-sfc-24 en- AIMs. In general, theabsolutedegreeofglycosylation of theASM code proteins with the same aspartic acid-rich motif (Additional was 1–3 orders of magnitude greater that of the AIM. The largest files 25, 37, and 38). Aspartic acid-rich proteins have been sug- difference was in the amount of galactosamine (10.6 ng/μLvs. gested to act as an organic template for epitaxial crystal growth 0.03 ng/μL). While glucosamine was more abundant absolutely [66, 67]. It is tempting to speculate that the three additional do- in the ASM (9.41 ng/μL vs. 0.13 ng/μL), the proportional differ- mains only present in adult mantle Lstag-sfc-24 contigs confer ence was not so extreme (34.8% vs. 54.2%). a specific shell-forming function to this protein. The putative chitin-interacting candidate Lstag-sfc-21 presented in Additional file 22 carries a signal sequence and is predicted to possess a Discussion catalytic activity. Intriguingly, a number of splice variants of this Molecular modularity of the adult molluscan mantle gene within the adult mantle are predicted to lack a signal se- Two of the most striking features of the phylum Mollusca are quence, the chitin-binding or catalytic ability (Additional file 37). A number of shell-forming gene candidates produce alter- its size (in terms of number of species) and its diversity. Widely accepted to be second only to the Arthropoda in terms of num- natively spliced transcripts that encode proteins with differ- ences regarding the presence/absence of a signal sequence (Ad- beroflivingspecies [73, 74], molluscs arguably display the great- est diversity of body forms of all metazoan phyla and have suc- ditional files 11, 12, 22, 25, and 37). Some shell-forming genes also produce alternatively spliced transcripts that encode pro- cessfully colonized all kinds of environments. While there cur- teins with similar coding features but radically different 5’ or rently exists no consensus as to why molluscs have enjoyed such deep evolutionary success (one interesting suggestion includes 3’ untranslated regions (UTRs) (Additional files 12, 19, 33, and 37). While UTRs do not contain protein-coding information, they a plastic nervous system [75]) we believe the mantle tissue (an apomorphy of the phylum) and its ability to prolifically evolve can be critical for localization of the mRNA [68, 69] and post- transcriptional gene regulation by molecules such as miRNAs new shell phenotypes must contribute to an explanation of this success. A logical extension of this question would therefore be, [70]. Indeed, several miRNAs have now been associated with the targeting and regulation of biomineralizing proteins [71, 72]. “what is it about the molluscan mantle tissue that makes it so evolutionarily plastic?” For arthropods, segmentation and body plan modularization (and the underlying gene regulatory net- Comparisons of molluscan shell-forming proteomes works that control appendage identity within each segment) are We conducted a broad comparison of our L. stagnalis shell- widely thought to have played leading roles in supporting the forming genes against a wide phylogenetic range of 12 other diversification of insects, spiders, and crustaceans [ 76]. The im- biomineralizing proteomes comprising 879 proteins (sequences portance of establishing segmentation at a very early develop- used in this analysis are provided in Additional file 39). Of all L. mental age in prominent phyla such as annelids, chordates, and stagnalis shell proteins, 27 shared significant sequence similar- arthropods has caused much effort to be spent on identifying ity with 1 or more of these 12 proteomes (Fig. 4). The highest the causal molecular mechanisms that may have common evo- degree of overall similarity was found with the shell-forming lutionary histories [77–79]. As recently reviewed by Esteve-Altava proteome of the common groove snail C. nemoralis (Fig. 4), the [80], the presence of morphological modules can help us to un- closest phylogenetic relative to L. stagnalis of all species in this derstand the evolvability of body form, but the identification of comparison. The next highest level of similarity shared with the such modules has so far been biased toward mammals, arthro- L. stagnalis shell-forming proteome (15.9% of the L. gigantea shell- pods, and plants. Following Esteve-Altava’s and Eble’s [81]defi- forming proteome) was markedly lower. Lymnaea stagnalis shell- nition of a morphological module (a group of body parts that are forming proteins that shared significant similarity with biomin- more integrated among themselves than they are to other parts eralizing proteins from other species and that also returned a outside the group), we propose that the molluscan mantle is a significant match against a SwissProt entry included Lstag-sfc- prime example of such a morphological module. This modular Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 9 L. stagnalis 58.7% (27/46) M. venosa 3.0% (2/66) M. truncata 4.2% S. purpuratus (3/71) 0.7% (1/139) C. nemoralis 28.8% (17/59) P. maxima A. millepora 2.4% 8.1% (1/42) (3/37) L. gigantea P. margaratifera 15.9% 7.7% (10/63) (6/78) P. fucata 9.1% H. asinina 1.1% (4/44) (1/94) M. coruscus 9.4% Top quartile of global similarity (5/53) 3rd quartile of global similarity H. laevigata C. gigas 2nd quartile of global similarity 3.75% 11.3% Lowest quartile of global similarity (3/80) (6/53) Threshold e-value = 10e-⁶ Candidate Adult mantle zone expression Conserved domains/similarity to Features Lstag-sfc-4 1 - Gly, Cys rich Lstag-sfc-5 1, 2 Animal haem peroxidase (CD) - Lstag-sfc-10a 3 - RLCD; Gly, Ser rich Lstag-sfc-10b 3 - Gly, Ser rich Lstag-sfc-11a/b 3 - RLCD; Ala, Thr, Ser, Pro, Leu rich Lstag-sfc-11c 3 - RLCD; Ala, Thr, Ser, Pro, Leu rich Lstag-sfc-18 2, 3, 4 Immunoglobulin domain (CD) Ser rich Lstag-sfc-20a 4 - - Lstag-sfc-20b 4 - Leu rich Lstag-sfc-21 1, 2, 5 Putative chitin deacetylase-like domain (CD) Thr rich Lstag-sfc-22 5 Pif97 Aragonite-binding protein (SP) Thr rich Lstag-sfc-23a/b 3, 4, 5 Otoancorin (SP) RLCD; Asp, Leu rich Lstag-sfc-24a/b 5 - RLCD; Asp, Ser, Asn rich Lstag-sfc-26 5 - RLCD; Asn, Gly, Gln Lstag-sfc-27a 5 PREDICTED: extensin-like isoform X1 (NR) RLCD; Pro, Gly Leu rich Lstag-sfc-27b 5 PREDICTED: formin-like prottein 2 (NR) RLCD; Pro, Gly rich Lstag-sfc-28a 5 - Gln, Pro Ala rich Lstag-sfc-28b 5 - Ala, Leu, Pro, Ser, Thr rich Lstag-sfc-28c 5 - RLCD; Gln, Pro rich Lstag-sfc-29 4, 5 Galaxin (SP) RLCD; Pro, Gln Gly rich Lstag-sfc-30 5 - - Lstag-sfc-31 2, 3, 4, 5 SUSHI repeat + short complement-like repeat (CD) RLCD; Gly, Ala, Leu, Pro rich Lstag-sfc-32 1, 2, 3, 4, 5 Intermediate filament protein (CD) Leu, Glu, Ser, Ala rich Lstag-sfc-33 NA vWA type A domain + collagen alphaI-XII-like (CD) RLCD; Val, Ala rich Figure 4: BLASTp comparisons of the L. stagnalis shell proteome against 879 biocalcifying proteins derived from six bivalves, four gastropods, one brachiopod, one −6 sea urchin, and one coral. Individual lines spanning the ideogram connect proteins that share significant similarity (e-values <10e ). Transparent red lines connect −6 proteins with the lowest quartile of similarity (with a threshold of 10e ) and green lines with the highest quartile of similarity. The percentage of each shell proteome that shared similarity with the L. stagnalis proteome is indicated. The table provides further information for those candidates that share sequence similarity with another species. Abbreviations: CD: conserved domain database; NR: GenBank non-redundant protein database; SP: SwissProt database. nature of the molluscan mantle is not unique to Lymnaea [82–84]. spatial expression patterns of shell-forming proteomes from a Although the precise functions of these zones (and of the indi- selection of other molluscan lineages will contribute to a more vidual gene products that define them) await the development refined understanding of molluscan shell evolution. of targeted genome editing methods, it is clear that they must act in a coordinated way to deposit the shell. We predict that there are related modules of gene regulatory networks (GRNs) Ontogenetic expression of shell-forming candidates that act to specify each zone of the molluscan mantle and that A prominent outcome of our survey of the adult shell proteome it is the modular nature of these GRNs and the resulting mor- is that many of the genes that encode these proteins are not only phological modularity of the mantle tissue that supported the regulated spatially but also temporally. Many shell-forming can- diversification of the phylum Mollusca. Characterization of the didates are expressed in the invaginated larval shell gland of the Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 S. purpuratus A. millepora L. gigantea H. asinina C. gigas M. coruscus P. fucata P. margaratifera P. maxima C. nemoralis M. truncata M. venosa H. laevigata 10 Modularity and asymmetry of the molluscan mantle Table 1: Glycosylation analysis of acid-soluble and acid-insoluble matrices extracted from adult L. stagnalis shells ASM AIM Monosaccharide ng/μg%ng/μg% Fucose ND - TR - Rhamnose ND - TR - Galactosamine 10.60 39.2 0.03 12.5 Arabinose ND - ND - Glucosamine 9.41 34.8 0.13 54.2 Galactose 4.34 16.0 0.04 16.7 Glucose 0.32 1.2 0.04 16.7 Mannose ND - TR - Xylose ND - ND - Galacturonic acid ND - ND - Glucuronic acid 2.39 8.8 ND - Total 27.06 100.0 0.24 100.0 ND = not detected, TR = trace. trochophore (Additional files 14, 15, 21, 24, 25, 27, and 28) or in (Fig. 1). We therefore suggest that Lstag-sfc-1, -2,and -3 are in cells that border it (Additional files 2, 3, 6, 7, 8, 10, 11, and 18). some way associated with producing thinner, more rapidly pro- Only two candidates were solely expressed in the adult man- duced shell at the outer shell lip than in the thicker parietal re- tle tissue and not in any larval stage (Additional files 17 and 23). gion, while Lstag-sfc-17 may inhibit the rapid deposition of shell. Timmermans [59] concluded that the spatial patterning of larval Exactly how this is achieved awaits the development of gene- shell-forming cells persists throughout development and fore- specific function assays. shadows the zonation observable in the adult mantle. We also In addition to the trochophore left/right asymmetry corre- observed this phenomenon at the molecular level. All candidate sponding to the left + parietal/right + outer lip regions of the genes that were expressed in the margin of the shell gland or shell, there is a second axis of symmetry that becomes appar- the shell field were expressed in the belt (zones 2, 3, and 4) of ent in 7-day-old juvenile snails. Many shell-forming candidates the adult mantle (summarized in Fig. 3, Additional files 2–8, 10– are initially symmetrically expressed in or surrounding the shell 12, and 18). Most candidates expressed in the invaginated cells gland of 2- to 3-dpfc trochophores but then become asymmet- of the shell gland or throughout the developing shell field were rically expressed in the mantle of older animals. For example, subsequently expressed in the low columnar outer epithelium Lstag-sfc-6, -7, -8, -12, -14, -15, -17, -18, -20, -23, -24, -26, -27, -29 of zone 5 in adult mantle tissue (summarized in Fig. 3, Addi- and -31 are expressed in the left side of the free mantle edge tional files 24, 25, 27– 29, and 31). However, three genes conspicu- in 7-dpfc juveniles (summarized in Fig. 3). In contrast, relatively ously deviate from this pattern. Lstag-sfc-13, -14,and -20 display few shell-forming gene candidates (Lstag-sfc-5, -9,-10,-21, and a broad expression pattern in the invaginated cells of the shell -25)are expressedevenlyalong thefreeedgeofthe mantlein gland and throughout the entire shell field in larvae but were not 7-dpfc juveniles (summarized in Fig. 3). detected in the low columnar outer epithelium (zone 5) of the adult mantle tissue (Additional files 14, 15, and 21). However, we The spatial expression of a peroxidase in the adult should point out that for all candidate shell-forming genes, we did not consider the potential effect of a diurnal rhythm on gene mantle allows a model of shell formation to be expression. All samples for in situ hybridization were taken dur- developed ing daylight hours, and so genes with activity during the night In agreement with Timmermans histochemical study of perox- would be missed. idase activity [59], the expression of Lstag-sfc-5, a shell-forming candidate with an “animal heme-dependent peroxidase” do- main (Pfam PF03098; Additional file 41B) is localized to zones Asymmetric expression of shell-forming genes 1 and 2. Peroxidases may be involved in periostracum forma- The expression of Lstag-sfc-1, Lstag-sfc-2,and Lstag-sfc-3 in tion by cross-linking fibrous proteins rich in reactive quinones to zones 1 and 2 of the adult mantle suggests they may be in- form water insoluble, protease-resistant polymers [85–87]. This volved in the formation of the periostracum (Fig. 1 and Ad- process, also referred to as tanning or sclerotization, can also ditional files 2–4); however, it is their larval expression pat- be catalyzed by tyrosinase (also known as catechol oxidase, cat- terns that are more striking. Lstag-sfc-1, -2,and -3 display a echolase, polyphenoloxidase, phenoloxidase, and phenolase). right-sided asymmetric expression pattern in cells bordering Within the molluscan biomineralization literature, sclerotiza- the shell gland and shell field. In contrast, Lstag-sfc-17 is ex- tion by tyrosinase appears to be the more commonly assumed pressed on the left side (Fig.1and Additional file 18). Following mechanism, rather than by peroxidase. Nonetheless, Timmer- the expression of these genes ontogenetically into older lar- mans demonstrated that heat inactivation clears the periostra- vae that begin to display the coiled phenotype of the adult, it cal groove and belt of both peroxidase activity and the ability to is apparent that right-sided cells in the trochophore are likely form melanin (a typical assay used to test for tyrosinase activ- to be those that give rise to the right + anterior region of ity), while specific tyrosinase inhibitors sodium bisulphite and the adult mantle that will produce the outer lip of the shell, potassium cyanide (NaHSO and KCN) did not affect its ability while left-sided cells will give rise to posterior mantle tissue to produce melanin [59]. The spatial expression pattern of Lstag- responsible for forming the left + parietal region of the shell sfc-5, coupled with the observations that newly secreted perios- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 11 tracum itself also displays peroxidase activity [57] and Timmer- damaged, these IGFs would be released and subsequently stim- mans experiments [59], strongly suggests that the peroxidase we ulate the underlying mantle epithelium to re-calcify. One line have identified here plays a key role in cross-linking the perios- of evidence that strongly supports this hypothesis is provided tracum in L. stagnalis rather than a tyrosinase, as also supposed by the osteogenic activity of mollusc shells [97]. This hypothe- for other gastropods such as Lottia [15]. sis implies that although the shell is acellular, it is able to ac- tively communicate and provide real-time feedback to the man- tle epithelium [98]. This interesting hypothesis awaits the devel- Glycine-rich shell-forming candidates are likely to be opment of gene-specific knock-down or knock-out assays. substrates for the peroxidase An important aspect of scleroprotein formation is its spatial RLCDs are an abundant feature of L. stagnalis shell coordination. The cross-linking reaction often generates cyto- proteins toxic intermediates, and the end products cannot be easily de- graded or resorbed [88]. Furthermore, the uncontrolled forma- Proteins containing RLCDs are a prominent feature of mollus- tion of extensive scleroprotein polymers prior to secretion would can shell-forming proteomes [15, 99, 100], and L. stagnalis is no exception. More than half of the L. stagnalis shell-forming can- clearly be detrimental to the cell. One common strategy to avoid these events is to compartmentalize the scleroprotein precur- didates we identified possess RLCDs. Proteins containing these domains were present in the belt and the low columnar outer sor (that is unable to spontaneously polymerize) away from the cross-linking enzyme. Following secretion, the precursors are epithelium of the adult mantle and in a wide variety of pat- terns of the larval stages we investigated. The motif complex- activated and enzymatically cross-linked [88]. Such a scenario would suggest that the substrate upon which the peroxidase ity, motif length, and number of motif repeats can vary greatly, from stretches consisting of a single amino acid (e.g., Additional acts is not located within the same cells. Three candidates expressed in zone three (Lstag-sfc-6, -7,and file 24) to motifs that exceed 10 amino acids (e.g., Additional files 3, 4, 29 and 34). In some cases, almost the whole protein -8) encode secreted, basic proteins that are dominated by repet- itive low-complexity domains (RLCDs) and anomalous amino is composed of RLCDs (Additional files 7–10). Repeated motifs are a common feature of structural proteins such as collagens, acid contents (high glycine, tyrosine, asparagine, and leucine keratins, silk, and cell wall proteins, as well as structural mod- contents; Additional files 37 and 38). All of these glycine-rich proteins carry tyrosine residues flanked by glycine. This arrange- ules in functional proteins such as receptors, histones, ion chan- nels, and transcription factors [101, 102]. RLCDs are often part ment has been shown to be favorable for the formation of cross- links between tyrosine residues by peroxidase [89]. Waite, in of intrinsically unstructured regions that lack a fixed or ordered three-dimensional structure [101]. In some cases, these regions his review of natural quinone-tanned glues [85], highlighted the typical L-3,4-dihydroxyphenylalanine (DOPA) -containing con- define the functionality of the protein. As a general rule, un- structured proteins interact readily with other proteins [103], sensus precursor peptide sequences from a number of marine invertebrates. Allowing for a single mismatch, these substrate and the highly repetitive, modular, and biased amino acid com- positions can confer strength and elasticity [104]. It will be ex- peptides (VGGYGYGK, GGGFGGYGK, and GGGYGGYGK, cross- linking tyrosine residues in bold) can be found within Lstag-sfc- tremely informative to selectively remove RLCDs from shell- forming proteins and to study the resulting shell phenotypes 6, Lstag-sfc-7, and Lstag-sfc-8. Interestingly, these glycine-rich candidates are expressed exclusively in zone 3 (Additional files once genome modification tools become broadly available to molluscs. 7–9) immediately adjacent to zone 2, the region in which the per- oxidase Lstag-sfc-5 is expressed (Additional file 6). Theoretically, once these proteins are secreted, the secreted peroxidase would Broad sequence similarity comparisons of metazoan be in very close proximity to the glycine-rich proteins and could biomineralizing proteomes act on the favored tyrosine residues to form di-tyrosine cross- The crossed-lamellar microstructure is fabricated by phyloge- links extracellularly. netically diverse molluscan taxa and is by far the most com- monly employed shell design of the Conchifera [21, 22, 28]. A role for immunity and signaling in shell formation While much attention has been dedicated to the characteri- Lstag-sfc-18 contains two Ig superfamily domains (Additional zation of nacre-forming bivalve shell proteomes, technical ad- file 19; [ 90]) and displays sequence similarity with the IMP- vances in nucleic acid sequencing and proteome-scale surveys L2–like proteins (Additional files 40 and 41), an insulin- have seen a rapid growth in the number and diversity of mol- like growth factor–binding protein (IGF-BP) that carries two luscan shell-forming proteomes and allow broad comparisons immunoglobulin-like domains and is able to bind IGF [91]. Sev- of these datasets to be performed. These comparisons can pro- eral studies by Dogterom and colleagues demonstrated the in- vide insight into the degree of evolutionary conservation that fluence of a growth hormone secreted by the cerebral ganglia exists across shell-forming proteomes [50]. In general, mollus- specifically on shell formation in L. stagnalis [92–95]. The authors can shell-forming proteomes are markedly different, with some conclude that this growth hormone acts on cells in the belt re- deeply conserved elements such as alkaline phosphatases, per- gion to control shell extension and periostracum formation but oxidases, and carbonic anhydrases [28, 57, 105, 106]. The signif- not on shell thickening. Interestingly, perlustrin, a protein asso- icant diversity of molluscan shell ultrastructures, crystal tex- ciated with nacre in abalone shells, contains an IGF-BP domain tures, colors, and materials properties therefore cannot be ex- and was also shown to bind IGFs and insulin [96]. An intrigu- plained by the use of the same genes in different ways. Rather, ing idea for the presence of IGF-BP in the abalone shell is that each lineage has uniquely evolved a large fraction of its shell- it would allow the shell to signal to the underlying mantle ep- forming proteome [14–16, 100, 107]. To expand on this compara- ithelium. According to this hypothesis, IGFs present in the ex- tive theme, we collected 879 biomineralizing proteins validated trapallial fluid are bound by IGF-BP during calcification and in- by proteomics from 10 molluscs, 1 brachiopod, 1 sea urchin, and corporated into the shell. Should the shell dissolve or be locally 1 coral and performed sequence similarity comparisons against Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 12 Modularity and asymmetry of the molluscan mantle our L. stagnalis dataset. Two of the 10 molluscs, Cepaea nemoralis this phylum-specific organ and allude to an explanation as to and Mya truncata [51, 108], construct shells that contain crossed- why the Mollusca have evolved so many successful shell mor- lamellar texture. Interestingly, our comparative analyses show phologies. While gene co-option, domain shuffling, and gene that L. stagnalis and M. truncata have only three proteins that family expansion are mechanisms that have clearly contributed share relatively low degrees of sequence similarity, while L. stag- to the great diversity of molluscan shell-forming proteins, our nalis and C. nemoralis share 17 proteins (some of these with very analyses also suggest that alternative splicing acts to signifi- high degrees of sequence similarity), the highest extent of simi- cantly expand the shell-forming molecular repertoire. Compar- larity between all species surveyed (Fig. 4). Both L. stagnalis and C. ing the results of spatial gene expression surveys focused on nemoralis inhabit non-marine environments, and the similarities shell-formation from a broad range of molluscan taxa will shed in their shell proteomes may either be a manifestation of this further light on the evolutionary story of this fascinating struc- and/or a reflection of their crossed-lamellar shells. The similar- ture. ity of their shell protein content may also reflect the relatively recent divergence time (Meso-Cenozoic) of these two clades (Sty- Availability of supporting data lommatophoran, i.e., C. nemoralis vs. hygrophilid, i.e., L. stag- nalis) within the monophyletic order of pulmonate gastropods, All raw NGS data has been deposited with the SRA with in comparison to the other species. One of the most striking ob- BioSample accession numbers SAMN08117214, SAMN08117215, servations we made in these comparisons was that almost all L. SAMN08709370, SAMN08709371, SAMN08709372, stagnalis shell-forming candidates expressed in zone 5 share se- SAMN08709373, SAMN08709374, SAMN08709375, quence similarity with C. nemoralis. Conversely, L. stagnalis shell- SAMN08709376, and SAMN08709377. Individual image files forming candidates expressed asymmetrically on the right side for the in situ hybridization gene expression patterns and in larvae were not present in any other biomineralizing pro- the sense strand cDNA sequences used to generate the teomes. in situ hybridization riboprobes can be accessed from the Some L. stagnalis shell-forming proteins contain domains associated GigaDB repository [45]. The mantle transcrip- found in a number of the biomineralizing proteins present in tome assemblies are also available via GigaDB [45] (file the dataset we assembled or are known to play a role in pro- names: C2844 CLC idba Trinity for annotation.fasta and cesses other than biomineralization such as the Sushi domain, C2845 CLC idba Trinity for annotation.fasta). All MS data have the von Willebrand factor A domain, the immunoglobulin do- been deposited with the ProteomeXchange Consortium with main, and the filament protein domain [ 14, 109]. The Pif-like pro- the dataset identifiers PXD008547 and 10.6019/PXD008547. tein is prevalent in both bivalve and gastropod nacreous shell Other supporting data are available from additional files, also proteomes and is known to bind aragonite crystals and to reg- including an extended description of the in situ hybridization ulate nacre formation [110]. However, limpets, which construct results (see additional file 47). crossed-lamellar structures, also contain Pif in their shells [15, 110, 111]. Our results further demonstrate that Pif is not limited Additional files to nacreous matrices and that it is likely to be a deeply conserved element of the molluscan biomineralizing proteome. Additional file 1 . Summarized results of MASCOT searches. Additional files 2–35 . Whole mount in situ hybridisation results Glycosylation patterns and molecular features of 34 shell-forming gene candidates. Additional file 36 . A more comprehensive summary of the re- Our preliminary analysis of the sugar moieties associated with sults presented in Fig. 3. shell-forming proteins revealed an interesting dichotomy be- Additional file 37 . Detailed table of the molecular features of all tween the ASM and AIM; the population of ASM proteins ap- shell-forming protein candidates. pear to be far more glycosylated than AIM proteins (Table 1). Additional file 38 . Results of repetitive motif searches using Whether this difference is generated by a heavily glycosylated Xstream. subset of the ASM or if it reflects a general trend of most ASM Additional file 39 . A FASTA formatted file of the 879 protein se- proteins being glycosylated remains unknown. We also cannot quences used to construct Fig. 4. determine whether there are any spatial biases within the adult Additional file 40 . Detailed results of tBLASTx similarity mantle tissue with regards to the location of glycosylated pro- searches for all shell-forming candidate genes against nr teins. The high percentage of glucosamine identified in AIM and database. ASM suggests that chitin, or its deacetylated derivative chitosan, Additional file 41 . Detailed results of protein family and protein is present in both extracts, but this hypothesis requires further domain similarity searches for all shell-forming candidate genes testing. Despite their likely importance to the functional mech- against CDD database. anisms of shell formation, post-translational modifications of Additional file 42 . Detailed results of BLAST similarity searches molluscan shell-forming proteins remain relatively understud- for all shell-forming candidate genes against SwissProt ied, and we predict that research efforts in these directions database. would yield interesting functional insights into the mechanisms Additional file 43 . Lineages for all shell-forming candidates that of shell fabrication. returned positive BLAST results. Additional file 44 . Nucleotide sequences of 34 families of shell Conclusion forming candidate genes. Additional file 45 . Translated sequences of 34 families of shell By characterizing the spatial expression patterns of 34 genes forming candidate genes. associated with shell formation, we have revealed patterns of Additional file 46 . mRNA regions targeted by riboprobes. asymmetry that presumably contribute to the coiled phenotype Additional file 47 . Extended Results and Discussion. of Lymnaea’s shell. Our broad survey of these genes in the adult mantle tissue also highlight the morphological modularity of Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 13 tion and molecular mechanism exploration of three gran- Abbreviation ulin epithelin precursor splice variants in biomineraliza- AIM: acid-insoluble matrix; ASM: acid-soluble matrix; BLAST: tion of the pearl oyster Pinctada fucata. Mol Genet Genom Basic Local Alignment Search Tool; CNS: central nervous sys- 2016;291:399–409. tem; dpfc: days post first cleavage; GRN: gene regulatory net- 11. Wang J, Gao J, Xie J, et al. Cloning and mineralization- work; hpfc: hours post first cleavage; IGF-BP: insulin-like growth related functions of the calponin gene in Chlamys farreri. factor–binding protein; LC: Liquid chromatography; MS: mass Comp Biochem Physiol B 2016;201:53–58. spectrometry; NGS: next-generation sequencing; PCR: poly- 12. YarraT,GharbiK,BlaxterM,etal. Characterizationofthe merase chain reaction; RLCD: repetitive low-complexity do- mantle transcriptome in bivalves: Pecten maximus, Mytilus main; SEM: Scanning Electron Microscopy; SRA: Sequence Read edulis and Crassostrea gigas. Mar Geonomics 2016;27:9–15. Archive; UTR: untranslated region. 13. Gao P, Liao Z, Wang X-X, et al. Layer-by-layer pro- teomic analysis of Mytilus galloprovincialis shell. PLoS One 2015;10:e0133913. Competing interests 14. Mann K, Jackson DJ. Characterization of the pigmented The authors declare that they have no competing interests. shell-forming proteome of the common grove snail Cepaea nemoralis. BMC Genom 2014;15:249. 15. Marie B, Jackson DJ, Ramos-Silva P, et al. The shell-forming Author contributions proteome of Lottia gigantea reveals both deep conservations I.H. carried out the molecular work, bioinformatic analyses, and and lineage-specific novelties. FEBS J 2013; 280:214–32. co-wrote and drafted the manuscript. F.M. and B.M. performed 16. Jackson DJ, McDougall C, Green K, et al. A rapidly evolv- the proteomic analyses and drafted the manuscript. D.J.J. con- ing secretome builds and patterns a sea shell. BMC Biol ceived and supervised the study, contributed to the molecular 2006;4:40. work, contributed to the bioinformatic analyses, performed the 17. Marie B, Marie A, Jackson DJ, et al. Proteomic analysis of the histological sections, and co-wrote and drafted the manuscript. organic matrix of the abalone Haliotis asinina calcified shell. All authors read and approved the final manuscript. Prot Sci 2010;8:54. 18. Zhao M, He M, Huang X, et al. A homeodomain transcription factor gene, PfMSX, activates expression of Pif gene in the Acknowledgements pearl oyster Pinctada fucata. PLoS One 2014;9:e103830. 19. Suzuki M, Saruwatari K, Kogure T, et al. An acidic matrix We are grateful to Wolfgang Drose ¨ for assistance and advice with protein, Pif, is a key macromolecule for nacre formation. in situ sectioning, Isabelle Zanella-Cleon from IBCP (Lyon) for Science 2009;325:1388–90. MS analysis, and Jennifer Hohagen and Dorothea Hause-Reitner 20. Jackson DJ, Worheide ¨ G, Degnan BM. Dynamic expression of who generated SEM images. Illumina sequencing was per- ancient and novel molluscan shell genes during ecological formed by Markus B. Schilhabel and his team at the Institute of transitions. BMC Evol Biol 2007;7:160. Clinical Molecular Biology, Christian-Albrechts-University Kiel. 21. Dauphin Y, Denis A. Structure and composition of the arag- This work was funded by DFG (JA 2108/2-1 and JA 2108/6-1) and onitic crossed lamellar layers in six species of Bivalvia and VolkswagenStiftung (92075) grants to D.J.J. Gastropoda. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology 2000;126:367–77. References 22. de Paula SM, Silveira M. Studies on molluscan shells: contri- butions from microscopic and analytical methods. Micron 1. Mao L-B, Gao H-L, Yao H-B, et al. Synthetic nacre 2009;40:669–90. by predesigned matrix-directed mineralization. Science 23. Almagro I, Drzymala P, Berent K, et al. New crystallographic 2016;354:107–10. relationships in biogenic aragonite: the crossed-lamellar 2. Kniprath E. Ontogeny of the molluscan shell field: a review. microstructures of mollusks. Crystal Growth & Design 2016. Zool Script 1981;10:61–79. 24. Kuhn-Spearing LT, Kessler H, Chateau E, et al. Fracture 3. Jackson DJ, Degnan BM. The importance of Evo-Devo to an mechanisms of the Strombus gigas conch shell: implications integrated understanding of molluscan biomineralisation. for the design of brittle laminates. J Mater Sci 1996;31:6583– J Struct Biol 2016;196:67–74. 4. Okabe T, Yoshimura J. Optimal designs of mollusk shells 25. Kamat S, Su X, Ballarini R, et al. Structural basis for the frac- from bivalves to snails. Sci Rep 2017;7:42445. ture toughness of the shell of the conch Strombus gigas. Na- 5. Aguilera F, McDougall C, Degnan BM. Co-option and de novo ture 2000;405:1036–40. gene evolution underlie molluscan shell diversity. Mol Biol 26. Pokroy B, Zolotoyabko E. Microstructure of natural Evol 2017;34:779–92. plywood-like ceramics: a study by high-resolution electron 6. Liang J, Xie J, Gao J et al. Identification and characteriza- microscopy and energy-variable X-ray diffraction. J Mater tion of the lysine-rich matrix protein family in Pinctada Chem 2003;13:682–8. fucata: indicative of roles in shell formation. Mar Biotech 27. Rodriguez-Navarro AB, Checa A, Willinger M-G, et al. 2016;18:645–58. Crystallographic relationships in the crossed lamellar mi- 7. Marxen JC, Hammer M, Gehrke T, et al. Carbohydrates of crostructure of the shell of the gastropod Conus marmoreus. the organic shell matrix and the shell-forming tissue of the Acta Biomater 2012;8:830–5. snail Biomphalaria glabrata (Say). Biol Bull 1998;194:231–40. 28. Marin F, Luquet G, Marie B, et al. Molluscan shell proteins: 8. Arias JL, Fernandez ´ MS. Polysaccharides and proteoglycans primary structure, origin, and evolution. Curr Top Dev Biol in calcium carbonate-based biomineralization. Chem Rev 2008;80:209–76. 2008;108:4475–82. 29. Pavat C, Zanella-Cleon ´ I, Becchi M, et al. The shell matrix 9. Farre B, Dauphin Y. Lipids from the nacreous and prismatic of the pulmonate land snail Helix aspersa maxima. Comp layers of two Pteriomorpha mollusc shells. Comp Biochem Physiol B 2009;152:103–9. 10. Zhao M, He M, Huang X, et al. Functional characteriza- Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 14 Modularity and asymmetry of the molluscan mantle Biochem Physiol B 2012;161:303–14. 50. Feng D, Li Q, Yu H, et al. Identification of conserved pro- 30. Vizca´ıno JA, Csordas A, Del-Toro N, et al. 2016 update of teins from diverse shell matrix proteome in Crassostrea gi- the PRIDE database and its related tools. Nucleic Acids Res gas: characterization of genetic bases regulating shell for- 2016;44:D447–56. mation. Sci Rep 2017;7:45754. 31. Jackson DJ, Ellemor N, Degnan BM. Correlating gene ex- 51. Arivalagan J, Marie B, Sleight VA, et al. Shell matrix pro- pression with larval competence, and the effect of age and teins of the clam, Mya truncata: roles beyond shell formation parentage on metamorphosis in the tropical abalone Halio- through proteomic study. Mar Geonomics 2016;27:69–74. tis asinina. Mar Biol 2005;147:681–97. 52. Liu C, Li S, Kong J, et al. In-depth proteomic analysis of shell 32. Gasteiger E, Gattiker A, Hoogland C, et al. ExPASy: the pro- matrix proteins of Pinctada fucata. Sci Rep 2015;5:17269. teomics server for in-depth protein knowledge and analy- 53. Liao Z, Bao L-F, Fan M-H, et al. In-depth proteomic analy- sis. Nuc Acids Res 2003;31:3784–8. sis of nacre, prism, and myostracum of Mytilus shell. J Pro- 33. Petersen TN, Brunak S, von Heijne G, et al. SignalP 4.0: dis- teomics 2015;122:26–40. criminating signal peptides from transmembrane regions. 54. Jackson DJ, Mann K, Haussermann ¨ V, et al. The Magellania Nat Meth 2011;8:785–6. venosa biomineralizing proteome: a window into brachio- 34. Gasteiger E, Hoogland C, Gattiker A, et al. Protein identi- pod shell evolution. Gen Biol Evol 2015;7:1349–62. fication and analysis tools on the ExPASy server. Humana 55. Mann K, Poustka AJ, Mann M. In-depth, high-accuracy Press. Totowa, NJ; 2005. proteomics of sea urchin tooth organic matrix. Prot Sci 35. Jorda J, Kajava AV. T-REKS: identification of tandem REpeats 2008;6:33. in sequences with a K-meanS based algorithm. Bioinfor- 56. Ramos-Silva P, Kaandorp J, Herbst F, et al. The skeleton of matics 2009;25:2632–8. the staghorn coral Acropora millepora: molecular and struc- 36. Altschul SF, Gish W, Miller W, et al. Basic local alignment tural characterization. PLoS One 2014;9:e97454. search tool. J Mol Biol 1990;215:403–10. 57. Hohagen J, Jackson DJ. An ancient process in a modern mol- 37. Marchler-Bauer A, Derbyshire MK, Gonzales NR, et al. lusc: early development of the shell in Lymnaea stagnalis. CDD: NCBI’s conserved domain database. Nuc Acids Res BMC Dev Biol 2013;13:27. 2015;43:D222–6. 58. Kniprath E. Zur Ontogenese des Schalenfeldes von Lymnaea 38. Mitchell A, Chang H-Y, Daugherty L, et al. The InterPro pro- stagnalis. Wilhelm Roux’s Archives of Developmental Biol- tein families database: the classification resource after 15 ogy 1977;181:11–30. years. Nuc Acids Res 2015;43:D213–21. 59. Timmermans LPM. Studies on shell formation in molluscs. 39. Steentoft C, Vakhrushev SY, Joshi HJ, et al. Precision map- Neth J Zool 1969;19:413–523. ping of the human O-GalNAc glycoproteome through Sim- 60. Saleuddin ASM. An electron microscopic study on the for- pleCell technology. EMBO J 2013;32:1478–88. mation of the periostracum in Helisoma (Mollusca). Calcif 40. Gene Structure Draw. [ http://www.compgen.uni-muenster Tissue Int 1975;18:297–310. .de/tools/strdraw/index.hbi?]. Accessed 10th April 2018 61. Kniprath E. Formation and structure of the periostracum in 41. Davison A, McDowell GS, Holden JM, et al. Formin is asso- Lymnaea stagnalis. Cal Tiss Res 1972;9:260–71. ciated with left-right asymmetry in the pond snail and the 62. Bubel A. An electron-microscope study of periostracum for- frog. Curr Biol 2016;26:654–60. mation in some marine bivalves. I. The origin of the perios- ˝ ´ 42. Kapustin Y, Tozser J, Souvorov A, et al. Splign: algorithms tracum. Mar Biol 1973;20:213–21. for computing spliced alignments with identification of 63. Bevelander G, Nakahara H. An electron microscope study of paralogs. Biology Direct 2008;3:20. the formation of the periostracum of Macrocallista maculata. 43. Yan T, Yoo D, Berardini TZ, et al. PatMatch: a program for Cal Tiss Res 1967;1:55–67. finding patterns in peptide and nucleotide sequences. Nuc 64. Bubel A. An electron-microscope study of periostracum for- Acids Res 2005;33:W262–6. mation in some marine bivalves. II. The cells lining the pe- 44. Cerveau N, Jackson DJ. Combining independent de novo as- riostracal groove. Mar Biol 1973;20:222–34. semblies optimizes the coding transcriptome for noncon- 65. Ast G. How did alternative splicing evolve. Nat Rev Gen ventional model eukaryotic organisms. BMC Bioinformatics 2004;5:773–82. 2016;17:525. 66. Weiner S, Traub W, Parker SB. Macromolecules in mollusc 45. Herlitze I, Marie B, Marin F, et al. Supporting data for shells and their functions in biomineralization [and Discus- “Molecular modularity and asymmetry of the molluscan sion]. Philosophical Transactions of the Royal Society B: Bi- mantle revealed by a gene expression atlas.” GigaScience ological Sciences 1984;304:425–34. Database 2018. http://dx.doi.org/10.5524/100436. 67. Addadi L, Weiner S. Interactions between acidic proteins 46. Jackson DJ, Herlitze I, Hohagen J. A whole mount in situ hy- and crystals: stereochemical requirements in biomineral- bridization method for the gastropod mollusc Lymnaea stag- ization. Proc Natl Acad Sci 1985;82:4110–4. nalis. J Vis Exp 2016;(109). doi: 10.3791/53968. 68. Kingsley EP, Chan XY, Duan Y, et al. Widespread RNA seg- 47. Marie B, Joubert C, Tayalea ´ A, et al. Different secretory reper- regation in a spiralian embryo. Evol Dev 2007;9:527–39. toires control the biomineralization processes of prism and 69. Lambert JD, Nagy LM. Asymmetric inheritance of centroso- nacre deposition of the pearl oyster shell. Proc Natl Acad mally localized mRNAs during embryonic cleavages. Nature Sci 2012;109:20986–91. 2002;420:682–6. 48. Mann K, Cerveau N, Gummich M, et al., In-depth proteomic 70. Bartel DP, Chen C-Z. Micromanagers of gene expression: the analyses of Haliotis laevigata (greenlip abalone) nacre and potentially widespread influence of metazoan microRNAs. prismatic organic shell matrix, Proteome Science, 2018. Nat Rev Gen 2004;5:396–400. Submitted. 71. Zheng Z, Jiao Y, Du X, et al. Computational prediction of 49. Mann K, Edsinger E. The Lottia gigantea shell matrix pro- candidate miRNAs and their potential functions in biomin- teome: re-analysis including MaxQuant iBAQ quantitation eralization in pearl oyster Pinctada martensii. Saudi Journal and phosphoproteome analysis. Prot Sci 2014;12:28. of Biological Sciences 2016;23:372–8. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018 Herlitze et al. 15 72. Jiao Y, Zheng Z, Tian R, et al. MicroRNA, pm-miR-2305, par- mation. Comparative Biochemistry and Physiology Part A: ticipates in nacre formation by targeting pearlin in pearl Physiology 1980;66:687–90. oyster Pinctada martensii. Int J Molec Sci 2015;16:21442–53. 94. Dogterom AA, van Loenhout H, van der Schors RC. The ef- 73. Brusca RC, Brusca GJ. Invertebrates. 2nd edition. Sinauer fect of the growth hormone of Lymnaea stagnalis on shell Associates: Sunderland, Massachusetts; 2002. calcification. Gen Comp Endocrinol 1979; 39:63–68. 74. Rosenberg G. A new critical estimate of named species- 95. Dogterom AA, van der Schors RC. The effect of the growth level diversity of the recent Mollusca. Am Malacol Bull hormone of Lymnaea stagnalis on (bi) carbonate movements, 2014;32:308–22. especially with regard to shell formation. Gen Comp En- 75. Hochner B, Glanzman DL. Evolution of highly diverse forms docrinol 1980;41:334–9. of behavior in molluscs. Curr Biol 2016;26:R965–71. 96. Weiss IM, Gohring ¨ W, Fritz M, et al. Perlustrin, a Haliotis lae- 76. Williams TA, Nagy LM. Developmental modularity and the vigata (abalone) nacre protein, is homologous to the insulin- evolutionary diversification of arthropod limbs. Journal of like growth factor binding protein N-terminal module of Experimental Zoology Part A: Ecological Genetics and Phys- vertebrates. Biochem Biophys Res Commun 2001;285:244– iology 2001;291:241–57. 9. 77. Peel AD, Chipman AD, Akam M. Arthropod segmentation: 97. Zhang G, Willemin AS, Brion A, et al. A new method for the beyond the Drosophila paradigm. Nat Rev Gen 2005;6:905– separation and purification of the osteogenic compounds of 16. nacre ethanol soluble matrix. J Struct Biol 2016;196:127–37. 78. Tautz D. Segmentation. Dev Cell 2004;7:301–12. 98. Marin F, Luquet G. Molluscan shell proteins. CR Palevol 79. Raff RA. The shape of life: genes, development, and the evo- 2004;3:469–92. lution of animal form. University of Chicago Press: Chicago, 99. Shen X, Belcher AM, Hansma PK, et al. Molecular cloning Illinois; 1996. and characterization of lustrin A, a matrix protein from 80. Esteve-Altava B. In search of morphological modules: a sys- shell and pearl nacre of Haliotis rufescens. J Biol Chem tematic review. Biol Rev 2017;92:1332–47. 1997;272:32472–81. 81. Eble GJ. Morphological modularity and macroevolution. In: 100. Kocot KM, Aguilera F, McDougall C, et al. Sea shell diversity Callebaut W, Rasskin-Gutman D , eds. Modularity: Under- and rapidly evolving secretomes: insights into the evolution standing the Development and Evolution of Natural Com- of biomineralization. Front Zool 2016;13:23. plex Systems. MIT Press, Cambridge, Massachusetts; 2005. 101. Luo H, Nijveen H. Understanding and identifying amino pp. 221–38. acid repeats. Brief Bioinform 2014;15:582–91. 82. McDougall C, Green K, Jackson DJ, et al. Ultrastructure of the 102. Alba` M, Tompa P, Veitia R. Amino acid repeats and the mantle of the gastropod Haliotis asinina and mechanisms of structure and evolution of proteins. In: Volff J-N (ed) Basel, shell regionalization. Cells Tissues Organs 2011;194:103–7. Karger. Gene and Protein Evolution. Karger Publishers; 2007. 83. Jolly C, Berland S, Milet C, et al. Zonal localization of shell pp. 119–30. matrix proteins in mantle of Haliotis tuberculata (Mollusca, 103. Alberts B, Johnson A, Lewis J, et al. Molecular Biology of the Gastropoda). Mar Biotech 2004;6:541–51. Cell. 5th revised edition. Garland Science: New York; 2007. 84. Sleight VA, Marie B, Jackson DJ, et al. An Antarctic mollus- 104. McDougall C, Woodcroft BJ, Degnan BM. The widespread can biomineralisation tool-kit. Sci Rep 2016;6:36978. prevalence and functional significance of silk-like struc- 85. Waite JH. The phylogeny and chemical diversity of quinone- tural proteins in metazoan biological materials. PLoS One tanned glues and varnishes. Comp Biochem Physiol B 2016;11:e0159128. 1990;97:19–29. 105. Le Roy N, Jackson D, Marie B, et al. Carbonic anhydrase and 86. Waite JH. Evidence for the mode of sclerotization in a mol- metazoan biocalcification: a focus on molluscs. In: Marin luscan periostracum. Comp Biochem Physiol B 1977;58:157– F, Brummer ¨ F, Checa A, Furtos G, Lesci IG, Siller L , eds. 62. Biomineralization: From Fundamentals to Biomaterials & 87. Waite JH, Wilbur KM. Phenoloxidase in the periostracum Environmental Issues. Trans Tech Publications Ltd; 2015. of the marine bivalve Modiolus demissus Dillwyn. J Exp Zool pp. 151–7. 1976;195:359–67. 106. Sun X, Yang A, Wu B, et al. Characterization of the man- 88. Waite JH. Quinone-tanned scleroproteins. In: Saleuddin tle transcriptome of Yesso scallop (Patinopecten yessoensis): ASM, Wilbur KM , eds. Physiology. Academic Press: New identification of genes potentially involved in biomineral- York; 1983. pp. 467–504. ization and pigmentation. PLoS One 2015;10:e0122967. 89. Michon T, Chenu M, Kellershon N, et al. Horseradish 107. Jackson DJ, McDougall C, Woodcroft B, et al. Parallel evolu- peroxidase oxidation of tyrosine-containing peptides and tion of nacre building gene sets in molluscs. Mol Biol Evol their subsequent polymerization: a kinetic study. Biochem 2010;27:591–608. 1997;36:8504–13. 108. Sleight VA, Thorne MAS, Peck LS, et al. Characterisation 90. Halaby DM, Mornon JPE. The immunoglobulin superfamily: of the mantle transcriptome and biomineralisation genes an insight on its tissular, species, and functional diversity. in the blunt-gaper clam, Mya truncata. Mar Geonomics J Mol Evol 1998;46:389–400. 2016;27:47–55. 91. Honegger B, Galic M, Kohler ¨ K, et al. Imp-L2, a putative ho- 109. Arivalagan J, Yarra T, Marie B, et al. Insights from the shell molog of vertebrate IGF-binding protein 7, counteracts in- proteome: biomineralization to adaptation. Mol Biol Evol sulin signaling in Drosophila and is essential for starvation 2017;34(1):66–77. resistance. Journal of Biology 2008;7:1. 110. Suzuki M, Kogure T, Weiner S, et al. Formation of aragonite 92. Dogterom AA, Doderer A. A hormone dependent calcium- crystals in the crossed lamellar microstructure of limpet binding protein in the mantle edge of the freshwater snail shells. Crystal Growth & Design 2011;11:4850–9. Lymnaea stagnalis. Calcif Tissue Int 1981;33:505–8. 111. Mann K, Edsinger-Gonzales E, Mann M. In-depth proteomic 93. Dogterom AA, Jentjens T. The effect of the growth hormone analysis of a mollusc shell: acid-soluble and acid-insoluble of the pond snail Lymnaea stagnalis on periostracum for- matrix of the limpet Lottia gigantea. Prot Sci 2012;10:28. Downloaded from https://academic.oup.com/gigascience/article-abstract/7/6/giy056/4997018 by Ed 'DeepDyve' Gillespie user on 21 June 2018

Journal

GigaScienceOxford University Press

Published: May 17, 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