Animal maintenance

Zebrafish wild type strains were reared at the fish facility of Wageningen University at 27.1°C with a 14:10 light dark cycle. All fish were raised in a density of 5 fish/L. For the experiments, matings were set up with two males and three females. Eggs were kept at 28.5°C at the breeding facility, and transferred to the fish facility at 20 days post fertilization (dpf). All fish were fed ad libitum three times a day. Zebrafish larvae were fed in-house cultured paramecium from 120 hours post fertilization. Between 10 and 60 dpf the fish were fed both paramecium and artemia, after which the paramecium was exchanged for Tetramin flakes. Zebrafish were kept under standard conditions until they reached 14, 28, or 358 dpf, at which point the fish were euthanized with 0.1% (w/v) tricaine methane sulphonate (TMS, Crescent Research chemicals, USA) buffered with 0.08% (w/v) sodium bicarbonate (Gibco, Paisley, Scotland).

Ethic statement

The experiment was approved by the Wageningen University Animal Experiments Committee (protocol nr. 2011026.a and 2012020.a).

Tissue preparation and extraction

Craniofacial, axial, and caudal fin skeletal elements were isolated from 500 zebrafish of 14 dpf, 100 zebrafish of 28 dpf, and 15 zebrafish of 358 dpf. First, the head region was separated from the axial skeleton between the head and the Weberian vertebrae. The axial skeleton included the Weberian vertebrae, the precaudal, and the caudal vertebrae [19]. The caudal fin was excised at the start of the caudal fin vertebrae. An overview of skeletal elements exhibited by the 3 developmental stages, obtained by acid-free bone and cartilage staining as described previously [20], is shown in Fig. S1. Skeletal structures were isolated from 14 day old larvae by removal of excess tissue using Accumax solution (Millipore). Larvae were incubated for 1 hour at room temperature while pipetting the sample up and down every 10 minutes using a syringe with a 23G needle. After 1 hour, the skeletal structures were collected using a 70 µm cell strainer (BD Falcon). The bone structures from 28 and 358 dpf zebrafish were excised manually, trimmed free of excess tissue and incubated in Accumax solution for 2 hours at room temperature under vigorous shaking to remove cell remnants.

Protein extraction

For protein extraction the sequential protein extraction method developed by Jiang et al. [21] was adapted for optimal protein extraction from the limited sample sizes used here. The main modifications included the removal of the pre-extraction step in 4M guanidine-HCl (GdmCl) to prevent protein loss from the bone structures of zebrafish larvae that were minimal in their degree of ossification as compared to juvenile and adult bone structures. Secondly, the extraction volumes were kept to a minimum so that processing of the extracts could be performed using spin filters [22], [23].

Before protein extraction, the isolated material was washed in phosphate buffered saline (PBS) containing a protease inhibitor cocktail (Roche), pH 7.4, for several hours at 4°C. For the protein extraction, the material from the initially separated craniofacial and axial region was used to create triplicates containing 5 mg (wet weight) each. Skeletal material was transferred to 1.5 ml Eppendorf tube containing 6 stainless steel balls (diameter, 3.9 mm), snap frozen in liquid nitrogen, and pulverized using a Retsch mill (Retsch GmbH, Hann, Germany) for 3 times 15 seconds at 30 Hz. Since not enough skeletal material could be isolated from the caudal fin region of zebrafish larva (14 dpf), this sample point was excluded. The isolation from the juvenile (28 dpf) caudal fin region resulted in enough material for one sample only (5 mg), and was therefore extracted as one sample and measured in duplicate. After pulverization the samples were sequentially extracted in order to improve the protein extraction from the mineralized matrix of bone. (I) First, the samples were demineralized by incubation in 1.2 M HCl for 2 hours (larvae), or overnight (juvenile and adult) at 4°C. The supernatant was collected after centrifugation; the pellet was washed with PBS containing protease inhibitor cocktail (pH 7.4) and also collected after centrifugation. The HCl and PBS fractions were pooled and stored at –80°C. (II) The pelleted skeletal material was incubated in 50 mM Tris-HCl, 4M GdmCl, 0.5 M tetrasodium EDTA (pH 7.4) containing protease inhibitor cocktail, for 72 h at 4°C. The supernatant was collected after centrifugation, and the EDTA was removed by dialysis overnight in Slide-A-Lyzer mini dialysis units (Thermo scientific) against 50 mM Tris-HCl, 4 M GdmCl at 4°C. The dialyzed solution was subsequently stored at –80°C. (III) The remaining pellet was incubated for 24 h at 4°C in 50 mM Tris-HCl, 6 M GdmCl (pH 7.4) containing protease inhibitor cocktail. The solution collected after centrifugation was stored at –80°C.

Filter-aided sample preparation

Filter aided-sample preparation was employed to prevent protein loss during the subsequent precipitation steps. Since two of the extraction steps were performed in high chaotropic salt (GdmCl) concentrations and all extraction steps were performed without a detergent like sodium dodecyl sulfate (SDS), the filter-aided sample preparation method allowed for the use of sample volumes up to 400 µl. To include all the proteins extracted in the first step (HCl extraction), the subsequent extracts of each sample were pooled to obtain a sample containing a high chaotropic salt concentration. Cysteine residues were reduced with 7.5 mM dithiothreitol (DTT) for 30 min at room temperature. 400 µl of the samples was added to a Pall 3K omega filter and centrifuged at 15,871 g for 30 min. The flow-through was discarded and this step was repeated until the complete sample volume was filtered. 400 µl 8 M urea in 100 mM Tris/HCl, pH 8 was added to the filter and centrifuged at 15,871 g for 30 min. Subsequently, 400 µl 0.05 M iodoacetamide/8 M urea in 100 mM Tris/HCl, pH 8 was added to the filter, mixed, incubated for 30 min in the dark at RT, and centrifuged at 15,871 g for 30 min. The filter was washed by the addition of 110, 120, and 130 µl 8 M urea in 100 mM Tris/HCl, pH 8 in subsequent steps with a centrifuge step in between each step, 15,871 g for 30 min. 140 µl 0.05 M NH₄HCO₃ (ABC) solution was added to the filter unit and centrifuged at 15,871 g for 30 min. This was then followed by the addition of 100 µl ABC solution containing 0.5 µg trypsin to the filter and incubated overnight at room temperature. A final centrifugation step was then performed at 15,871 g for 30 min and the flow through was acidified using 10% (v/v) trifluoroacetatic acid to pH 2-4. The samples were stored at –20°C until the nLC-MS/MS measurement was performed.

Mass spectrometry measurement

The analysis was performed by injecting 18 µl of sample over a 0.10×32 mm Prontosil 300-5-C18H (Bischoff, Germany) pre-concentration column (prepared in-house) at a maximum pressure of 270 bar (Thermo Proxeon Easy nLC). The peptides were eluted from the pre-concentration column onto a 0.10×200 mm Prontosil 300-3-C18H analytical column with an acetonitrile gradient at a flow rate of 0.5 µl/min. The gradient that was used here increased the acetonitrile in water percentage from 9 to 21% during 100 min, 21 to 34% in 26 min, 34 to 50% in 3 min and constant at 50% for 5 min. Acetic acid was added to the eluent at 5 ml/l. The column was cleaned by increasing the acetonitrile up to 80% in 3 min. Between the pre-concentration and analytical column, an electrospray potential of 3.5 kV was applied directly to the eluent via a metal needle electrode fitted into a P777 Upchurch micro cross. Spectra were obtained between m/z 380 and 1400 on an LTQ-Orbitrap XL (Thermo electron, San José, CA, USA). MS/MS scans of the 10 most abundant doubly and triply charged peaks in the FTMS scan were recorded in data-dependent mode in the linear trap (MS/MS threshold =  5000, 60 s exclusion duration).

Data analysis and label-free quantification

MaxQuant software version 1.2.2.5 was used to analyze the raw files from the LTQ-Orbitrap. MS/MS spectra were searched against the Uniprot zebrafish database downloaded from www.uniprot.org (2012), and a contaminant database including frequently observed contaminants such as human keratins and trypsin. For label-free quantification, the MaxQuant settings were kept default (using a 1% false discovery rate (FDR) on both peptide and protein level) except for the following changes. Asparagine or glutamine de-amidation was specified as variable modification. Label-free protein quantification was switched on, together with the “match between runs” option used with a maximal retention time window set to 2 min, and the intensity based absolute quantification (iBAQ) on. Normalized intensity values (LFQ intensity) were used for relative quantification. The MaxQuant results are shown in Table S1 in File S1. Filtering of the MaxQuant protein groups result file (Table S1 in File S1) was performed using the Perseus module (version 1.3.0.4). Reverse hits were removed from the data together with proteins with only 1 identified peptide, no unique peptide, or no unmodified peptide. Database search and quantification results were grouped by combining results for the different injections of the craniofacial, axial, and caudal fin region for each of the developmental stages.

Gene ontology (GO)-terms were specified for each protein using Software Tool for Researching Annotations of Proteins (STRAP) program analysis [24]. For genes with missing zebrafish ontology information, human orthologs were used instead and the same STRAP analysis was performed. Proteins known to be located in the (proteinaceous) extracellular region or matrix were used for further analysis. For quantification purposes, the proteins were additionally filtered and required to be detected in 2 or more out of each triplicate LC-MS/MS runs in at least 2 out of 3 sample points per region. Proteins without enough label-free quantification values were not selected for this procedure. Selected proteins were further analyzed using the Perseus module. Relative protein abundances were compared by performing a pair-wise t-test (both sides), with a permutation-based on FDR, threshold values of 0.05 or 0.01 and S0  =  1, with 250 number of randomizations and –log10 as selected parameters. The obtained protein ratios were used to assess differences in protein abundance between the different regions and developmental stages. Proteins that changed significantly in abundance were subsequently analyzed based on previous identification and/or characterization in the vertebrate skeleton by assessing literature and zebrafish-specific data available at ZFIN [25]. Proteins that passed the threshold value of 0.05 were explored by Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA). Protein abundance ratios (Table S5 in File S1) and corresponding t-test values were uploaded into IPA and used to predict biological processes differing in abundance during the complete time course from larvae to adult.

In situ mRNA hybridization

Four of the proteins that differed in abundance (Table S5 in File S1) were selected for spatial gene expression analysis with in situ hybridization. Primers were designed using Primer 3 software [26] and extended with a T7, or Sp6 transcription initiation sequence on the reverse primer (Table S6 in File S1). cDNA obtained from zebrafish of 15 dpf was used to perform a PCR with these primers to generate the specific coding sequences of the genes (0.2–1.0 kb) to use as templates for the DIG labeling reaction. DIG-labeled probes were synthesized according to the manufacturer’s instructions (Roche, DIG-labeling) and purified with the Qiagen RNeasy kit (Protocol RNA Cleanup).

Wild type zebrafish (28 dpf) were euthanized as described above and fixed overnight in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) at 4°C. After washing with PBS (2×5 min), fish were dehydrated in 100% methanol at RT (room temperature, 2×5 min) and stored at –20°C until further use. For sectioning, fish were rehydrated (75%, 50% and 25% methanol/PBS, 1×5 min), and washed in PBST (PBS + 0.1% v/v Tween 20, 3×5 min). Fish were embedded in 1.5% agar/5% sucrose and incubated overnight in 30% sucrose at 4°C. Embedded fish were frozen in liquid nitrogen and stored at –80°C. Longitudinal sections of 20 µm were made with the cryostat (Leica CM3050S) and transferred onto polylysine coated slides (Menzel-Gläser, Thermo scientific). Slides were stored in a box with silica gel at –20°C or if used immediately, dried overnight at RT.

In situ hybridization was performed as described in Smith et al. [27] and Schulte-Merker [28] with the following modifications. Probes were diluted to a final concentration of 0.5 ng/µl in hybridization buffer (50% formamide, 5x salt solution; 0.75 M NaCl and 0.075 M sodium citrate pH = 7.0, 0.1% v/v Tween 20, 9.2 µM citric acid, 0.5 mg/ml tRNA yeast (Invitrogen), heparin (45 U/ml). Two hundred µl of the probe solution was added to each slide and covered with Nescofilm (Bando Chemical IND., Kobe, Japan). Slides were incubated overnight at 70°C in a humid chamber. Slides were washed in solution A (50% v/v formamide, 1 x salt solution, 0.1% v/v Tween 20), 2×30 min, 70°C) and in 1xTBST (0.14 M NaCl, 2.7 mM KCl, 0.025 M Tris HCl, 0.1% v/v Tween 20, pH = 7.5), 2×30 min, RT. Subsequently, slides were incubated in 2% blocking buffer (Blocking reagent, Roche, Mannheim, Germany) in 1 x TBST for at least 1 hour at RT. The blocking buffer was replaced with 200 µl of a 1∶2000 dilution of anti-DIG AP FAB fragments (Roche) in 2% blocking buffer. The slides were covered with Nescofilm and incubated overnight at 4°C. Then, slides were washed with TBST and equilibrated with AP staining buffer (0.1 M Tris, 0.1% v/v Tween 20, 0.1 M NaCl, 0.05 M MgCl₂), 2×10 min. Slides were incubated in 400 µl of staining solution (1 ml AP staining buffer, 4.5 µl NBT and 3.5 µl BCIP) in the dark. Staining was monitored regularly and stopped by washing in distilled water (2×10 min). After clearing with methanol, washing with PBS and fixation with 4% PFA/PBS (20 min at RT), slides were mounted in Aquatex (Merck, Darmstadt, Germany) and incubated overnight at RT to dry. Images were taken with an Olympus DP50 digital camera mounted on a Nikon Microphot-FXA microscope and Analysis^D software (Soft Imaging System GmbH, Germany).

Zebrafish skeletal extracellular proteome

A total of 262 different extracellular proteins (Fig. 1A) were identified and quantified in protein extracts of the zebrafish skeleton at three developmental stages of the craniofacial, axial and caudal fin regions (see Fig. 2, 3 and 4 respectively). Extracellular proteins were selected based on available GO information of the zebrafish (n =  126), and GO information based on human orthologs (n =  136) as described in the Materials and Methods section. Based on STRAP gene ontology analysis, the extracellular proteins were associated mainly with binding (n =  111), but also with processes like catalytic activity (n =  40), structural molecule activity (n =  36), and enzyme regulatory activity (n =  27) suggesting that next to major structural proteins, the selected protein set also included proteins with regulatory functions in skeletal development. A large number of proteins (n =  116) however lacked gene ontology information (Fig. 1A). The full list is shown in Table S1 in File S1 while an overview of all identified peptides is presented in Table S2 in File S1. One hundred and forty-eight cellular proteins were also identified but, together with proteins that could not be properly annotated, these were not further analyzed (Table S4 in File S1). As expected, the most abundant proteins within the list of extracellular proteins include major bone (e.g. collagen type I isoforms, extracellular matrix protein 2) and cartilage proteins (e.g. collagen type II, matrilin 1, apolipoprotein A-I, epyphican), previously described in MS based analyses of these tissues/structures in other vertebrate species, including humans [29], [30], [31], [32]. Also the serum-derived protein alpha-2-HS-glycoprotein (Ahsg), reported to bind to the mineral content of bone, was found among the most abundant proteins (Fig. 1B) [33], [34].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Composition of the zebrafish extracellular protein profile.

(A) Distribution of molecular functions for the proteins identified as extracellular proteins. ^a Gene Ontology information based on the human ortholog of a protein in case this information was absent in the zebrafish database. The diagram contains 351 entries corresponding to 262 different proteins of which some fall into multiple categories (B) Ten most abundant extracellular proteins within the obtained protein profile based on their intensity.

https://doi.org/10.1371/journal.pone.0090568.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Quantitative analysis of the zebrafish craniofacial skeleton by MS-based proteomics.

(A) Alcian blue/alizarin red stain of cartilage/bone structures in the craniofacial skeleton. Lateral (top) and ventral (bottom) images of all three time point used for protein extraction (left to right: 14 dpf, 28 dpf, 358 dpf). (B) VENN diagram of proteins selected for label-free quantification, with areas drawn to represent the number of proteins. Total number of proteins as well as distinct and common proteins are indicated for each time point. Proteins that did not qualify for label-free quantification are depicted in grey. (C-E) Ratio abundance plots showing log total iBAQ intensities versus log protein abundance ratio of the 28/14 dpf (C), the 358/28 dpf (D) and the 358/14 dpf (E) craniofacial skeleton ratios of all the proteins that met the strict criteria for label-free quantification (black circles, significantly differential abundant proteins at FDR =  0.01; dark grey circles, significantly differential abundant proteins at FDR =  0.05; light grey circles, no significant change in abundance). The numbers correspond to the proteins listed in 2F. (F) Table containing several of the significant differentially abundant proteins within the cranial skeleton. Specific proteins are discussed in the text.

https://doi.org/10.1371/journal.pone.0090568.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Quantitative analysis of the zebrafish axial skeleton by MS-based proteomics.

(A) Alcian blue/alizarin red stain of cartilage/bone structures in the axial skeleton. Lateral images of all three time point used for protein extraction (left to right: 14 dpf, 28 dpf, 358 dpf). (B) VENN diagram of proteins selected for label-free quantification, with areas drawn to represent the number of proteins. Total number of proteins as well as distinct and common proteins are indicated for each time point. Proteins that did not qualify for label-free quantification are depicted in grey. (C-E) Ratio abundance plots showing log total iBAQ intensities versus log protein abundance ratio of the 28/14 dpf (C), the 358/28 dpf (D) and the 358/14 dpf (E) axial skeleton ratios of all the proteins that met the strict criteria for label-free quantification (black circles, significantly differential abundant proteins at FDR =  0.01; dark grey circles, significantly differential abundant proteins at FDR =  0.05; light grey circles, no significant change in abundance). The numbers correspond to the proteins listed in 3F. (F) Table containing several of the significant differentially abundant proteins within the axial skeleton. Specific proteins are discussed in the text.

https://doi.org/10.1371/journal.pone.0090568.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Quantitative analysis of the zebrafish caudal fin by MS-based proteomics.

(A) Alcian blue/alizarin red stain of cartilage/bone structures in the caudal fin skeleton. Lateral images of two time points used for protein extraction (left to right: 28 dpf, 358 dpf). (B) VENN diagram of proteins selected for label-free quantification, with areas drawn to represent the number of proteins. Total number of proteins as well as distinct and common proteins are indicated for each time point. Proteins that did not qualify for label-free quantification are depicted in grey. (C) Ratio abundance plot showing log total iBAQ intensities versus log protein abundance ratio of the 358/28 dpf caudal fin region ratios of all the proteins that met the strict criteria for label-free quantification (dark grey circles, significantly differential abundant proteins at FDR =  0.05; light grey circles, no significant change in abundance). The numbers correspond to the proteins listed in 4D. (D) Table containing several of the significant differentially abundant proteins within the cranial skeleton. Specific proteins are discussed in the text.

https://doi.org/10.1371/journal.pone.0090568.g004

In addition to these highly abundant proteins, the total protein profile contained multiple other ECM structural constituents among which various types of collagens, as well as several proteins from the small leucine-rich proteoglycan (SLRP) family (e.g. biglycan, opticin, osteoglycin, asporin), growth factors (e.g. bone morphogenic proteins, transforming growth factors, fibroblast growth factors), growth factor regulator proteins (e.g., insulin-like growth factor binding proteins, transforming growth factor binding proteins), various peptidases (e.g. matrix metallopeptidases, serine peptidases, carboxypeptidases), peptidase inhibitors (e.g. serpin peptidase inhibitors, timp metallopeptidase inhibitors), and serum proteins (e.g., transferrin, thrombin, apolipoproteins, complement factors) (Table S3 in File S1).

Analysis of extracellular matrix proteins during skeletal development

The relative contribution of cartilage and bone to the skeletal structures in the craniofacial sections changes rapidly (Fig. 2A). To identify extracellular matrix proteins associated with the different skeletal tissues during zebrafish development, relative quantification was performed separately for the craniofacial skeleton, the axial skeleton, and the caudal fin region of three developmental stages; larvae (14 dpf), juveniles (28 dpf) and adult fish (358 dpf). In total, 188 of the proteins identified in the craniofacial skeleton were selected for the relative quantification based analysis. Of these, 127 were found at all three time points (Fig. 2B) while limited numbers were found at two or only one time point. Normalized abundance ratios were compared between the three different stages at a threshold of FDR  =  0.01. Seventy-four proteins significantly differed in abundance between larval, juvenile and adult stages (Fig. 2C-E, Table S5 in File S1). Three different analyses are presented, providing a pair-wise comparison between the three developmental stages in terms of protein abundance compared with iBAQ intensity [35]. A selection of the proteins that differ most in abundance for the three different stages is given in Fig. 2F, and their positions indicated in the diagrams of Fig. 2C-E. Among the proteins strongly reduced in abundance upon development between two of the time points, collagen type IV alpha 6 (Col4a6), laminin beta 2-like (Lamb2l), leukocyte cell-derived chemotaxin 1 (Lect1) and serum paraoxonase/arylesterase 2 (Pon2) were found while a strong increase in abundance was noted for osteonectin (Sparc), midkine-related growth factor a (Mdka), midkine-related growth factor b (Mdkb), unique cartilage matrix-associated protein a (Ucmab), secreted phosphoprotein 24 (Spp2) and platelet-derived growth factor receptor-like (Pdgfr1). The entire list of craniofacial extracellular proteins, including the significantly differing proteins, is given in Table S5 in File S1.

A similar analysis is presented for the axial region (Fig. 3). There, 163 proteins were found to be present at all three time points and again a minority was found to be associated only with one or two time points. In comparing the most significant reduction or increase in protein abundance there are striking differences with the list shown for the head region. For example the proteins Lamb2l, Nidogen 2 (Nid2), microfibrillar-associated protein 2 (Mfap2), collagen type V alpha 3b (Col5a3b) and collagen type IV alpha 5 (Col4a5) show a reduced 358/14 dpf ratio. Although the analysis is limited to juvenile and adult stages of the caudal fin (Fig. 4A), due to the limited yield of skeletal material from the larvae, the list of proteins significantly changed in abundance is again different from the previous two. Of the 8 proteins differing in abundance between the juvenile and adult stages of caudal fin development (Fig. 4D), myocilin (Myoc) and prostaglandin D2 synthase b (Ptgdsb) were found to among the proteins that exhibited the highest difference in abundance in the axial protein abundance comparison at the same stages. None of the caudal fin proteins strongly differing in abundance are found in the list of protein abundance comparison of the head region.

Multiple cartilage matrix proteins, such as Lect1, hyaluronan and proteoglycan link protein 1b (Hapln1b) and opticin (Optc) were found to be reduced in abundance during development. This observation fits with the gradual replacement of cartilage by ossified structures during development. Other proteins including the collagen type IV isoform alpha 6 (Col4a6), Lamb2l, laminin beta 4 (Lamb4), and nidogen 1b (Nid1b) are conventional basement membrane proteins [36], [37]. The same proteins are also expressed by chondrocytes and in mice become less abundant during the aging process [38]. Other proteins were connected to either mineralization (ectonucleotide pyrophosphatase/ phosphodiesterase family member 1)[39], or remodeling of the bone matrix (annexin A2b) [40]. The extracellular protein slit homolog 3 (Slit3), and otospiralin (Otos) have not previously been associated with vertebrate skeletogenesis.

A number of extracellular proteins are highly abundant in the juvenile skeleton but not in the larval and adult stages, for example several Wnt signaling regulator proteins such as secreted frizzled-related protein 1b (Sfrp1b) and frizzled-related protein (Frzb). Both have a documented role in chondral ossification [41], [42] thereby corresponding with the process of chondral ossification which is most prominent at the juvenile stage, but no longer evident in the adult skeleton. Additional proteins most abundant at the juvenile stage in the head region include the cartilage protein ‘upper zone of growth plate and cartilage matrix associated b’ (Ucmab), as well as lysyl oxidase-like 2b (Loxl2b), and inter-alpha-trypsin inhibitor heavy chain family, member 6 (Itih6) previously identified in cartilage tissues during MS-based approaches [29], [30]. The protein tissue factor pathway inhibitor a (Tfpia) was the one extracellular protein considered as novel in vertebrate skeletogenesis that showed highest abundance at 28 dpf (Table S5 in File S1).

The remaining proteins were shown to increase during the complete developmental sequence analyzed (Table S5 in File S1). The list includes proteins with an implicated function in mineralization such as Ahsg and Spp2 [43], [44], in bone formation such as osteopontin (Spp1), and Sparc, but also in the process of cartilage maturation, including cartilage intermediate layer protein 2 (Cilp2), and tenascin Xb (Tnxb) [29]. Other proteins significantly increased in abundance included growth factor regulators such as fibroblast growth factor-binding protein 2 (Fgfbp2), and Insulin-like growth factor-binding protein 5b (Igfbp5b). This last protein has been implicated to be processed by the proteinase htra serine peptidase 1b (Htra1b) that is able to induce bone formation by regulating transforming growth factor beta (Tgfβ) signaling [45], but can also regulate IGF signaling by cleaving IGFBP5 [46]. From the list of 123 extracellular matrix proteins that were significantly differing in abundance (P =  0.01), 52 were novel in the context of zebrafish skeletal development or have only been studied based on their gene expression pattern by in situ hybridization at stages up to 5 dpf (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Extracellular matrix proteins novel in the zebrafish skeleton.

https://doi.org/10.1371/journal.pone.0090568.t001

Finally, a small number of extracellular proteins were found that can be regarded as novel in the context of vertebrate skeletogenesis in general. This list includes tissue factor pathway inhibitor a (Tfpia), hyaluronan binding protein 2 (Habp2), otospiralin (Otos), tectorin alpha (Tecta), meteorin, and glial cell differentiation regulator-like (Metrnl). While there is no clear picture emerging from this small list of proteins in relation to the signaling processes or the structural proteins involved in bone formation, there seem to be a number of proteins acting as growth inhibitors or representing elements of other tissues such as blood vessels or neurons. For instance, the protein Metrnl is known to act on glial cell differentiation of mice [47].

The obtained protein abundance ratios were used for analysis using the Ingenuity Pathway Analysis (IPA) tool. As shown in Fig. 5, the network consisting of 29 proteins is enriched in proteins linked to processes that contribute to formation of the extracellular matrix, including cell-matrix interactions, organization of collagen type I fibrils, mineralization of bone, and bone remodeling. During development, proteins contributing to the cartilage matrix (e.g. Hapln1) decreased in abundance whereas most proteins involved in the formation and remodeling of bone increased.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Ingenuity Pathway Analysis (IPA).

Proteins that differ in abundance are involved in connective tissue, skeletal and muscular development. Green shading indicates an increase in abundance during development whereas red shading indicates a decrease. Increased intensity in colors indicates a higher differential abundance.

https://doi.org/10.1371/journal.pone.0090568.g005

To determine whether the extracellular proteins as identified here are secreted by either chondrocytes or osteoblasts, in situ mRNA hybridization was employed for the cartilage-related genes lect1 and ucmab and the bone-related genes bmper and col1a2 (Fig. 6). Sagittal sections of the head region of 28 dpf zebrafish juveniles were used, as both cartilage and bone formation are actively taking place at this stage (cf. Fig. 2A). The results show that lect1 and ucmab expression is found in chondrocytes of the parachordal (pch) cartilage (arrows) whereas bmper is expressed in ossification sites as confirmed by col1a2 expression. This confirms that these extracellular proteins are deposited close to their respective sites of secretion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. In situ mRNA hybridization of genes encoding extracellular proteins implicated in cartilage and bone formation in the head region.

Sagittal sections of the head region of 28 dpf zebrafish juveniles stained by acid-free bone (red) and cartilage (blue) staining (A-B), or used for in situ hybridization with an antisense RNA probe corresponding to lect1 (C), ucmab (D), bmper (E), col1a2 (F). As a control, hybridization with the lect1 sense RNA probe is shown (G). (B-G) Magnification of boxed area in A focusing on the parachordal (pch) cartilage. Images are from consecutive sections at the same position. (C-D) Expression of lect1 and ucmab is located in chondrocytes of the parachordal (pch) cartilage (arrows). (E-F) Transcripts of bmper and col1a2 were detected in ossification sites surrounding the parachondral chondrocytes (arrow heads). Scale bars indicate 0.5 mm (black), or 100 µm (white). Abbreviations: ep, ethmoid plate; Mk, Meckel’s cartilage; oc, otic capsule; pch, parachordal.

https://doi.org/10.1371/journal.pone.0090568.g006