Quantitative proteomic comparison of myofibroblasts derived from bone marrow and cornea

Myofibroblasts are fibroblastic cells that function in wound healing, tissue repair and fibrosis, and arise from bone marrow (BM)-derived fibrocytes and a variety of local progenitor cells. In the cornea, myofibroblasts are derived primarily from stromal keratocytes and from BM-derived fibrocytes after epithelial-stromal and endothelial-stromal injuries. Quantitative proteomic comparison of mature alpha-smooth muscle actin (α-SMA)+ myofibroblasts (verified by immunocytochemistry for vimentin, α-SMA, desmin, and vinculin) generated from rabbit corneal fibroblasts treated with transforming growth factor (TGF) beta-1 or generated directly from cultured BM treated with TGF beta-1 was pursued for insights into possible functional differences. Paired cornea-derived and BM-derived α-SMA+ myofibroblast primary cultures were generated from four New Zealand white rabbits and confirmed to be myofibroblasts by immunocytochemistry. Paired cornea- and BM-derived myofibroblast specimens from each rabbit were analyzed by LC MS/MS iTRAQ technology using an Orbitrap Fusion Lumos Tribrid mass spectrometer, the Mascot search engine, the weighted average quantification method and the UniProt rabbit and human databases. From 2329 proteins quantified with ≥ 2 unique peptides from ≥ 3 rabbits, a total of 673 differentially expressed (DE) proteins were identified. Bioinformatic analysis of DE proteins with Ingenuity Pathway Analysis implicate progenitor-dependent functional differences in myofibroblasts that could impact tissue development. Our results suggest BM-derived myofibroblasts may be more prone to the formation of excessive cellular and extracellular material that are characteristic of fibrosis.

Myofibroblast primary cell cultures. Four separate primary keratocyte-derived corneal fibroblast cultures were generated from both eyes of the four New Zealand white rabbit corneas, as previously described 16 with Dulbecco's Modified Eagle Medium (DMEM, Gibco, Grand Island, NY). Briefly 16 ,primary keratocytes were isolated from each cornea by first removing the epithelial and endothelial layers using 0.12 mm forceps and a #64 scalpel blade (BD Beaver, Franklin Lakes, NJ) under a dissecting microscope using the sterile technique. Keratocytes were isolated from the corneal stroma by digestion in sterile Dulbecco's Modified Eagle Medium (DMEM; Gibco, Grand Island, NY) containing 2.0 mg/ml collagenase (Gibco, Grand Island, NY) and 0.5 mg/ml hyaluronidase (Worthington, Lakewood, NJ) overnight at 37 °C. Cells were spun down and cultured in DMEM (Gibco) with 1% FBS and 20 ng/ml TGF-b1 (R&D, Minneapolis, MN). The medium was changed every 48 h in all cultures.
BM-derived myofibroblast primary cultures were generated from the same four rabbits as described previously to generate cornea-derived myofibroblasts. Briefly, [13 = the tibias and femurs of rabbits were removed and BM cells were harvested by flushing medium and scratching the bone marrow cavity with the end of an 18-gauge needle. BM cells were collected in a petri dish and clumps were teased out and gently dissociated with a one ml pipette to form a single-cell suspension. The suspension of cells was centrifuged at 1500 rpm for 10 min at 4 °C to obtain a cell pellet. Red blood cells were lysed by adding sterile Milli-Q water at 4 °C, followed by dilution with 10X PBS at 4 °C at a ratio of one part PBS to nine parts cell solution, with immediate mixing. Cell suspensions were centrifuged again at 1500 rpm for 10 min at 4 °C and re-suspended in 1 × PBS at 4 °C. Cell viability in the range of 90% to 95% was verified by staining with 0.4% trypan blue. Cells were suspended in PBS at 4 °C at a final concentration of 2 × 10 6 cells/ml. BM-derived cells, were cultured in DMEM (Gibco) with 1% FBS and 20 ng/ml TGF-β1 (R&D Systems, Minneapolis, MN). The medium was changed every 48 h in all cultures.
Cornea-and BM-derived myofibroblast cultures were each harvested from 6 T-75 flasks after 14 days of culture with 20 ng/ml TGF-β1, the cells washed with PBS, and cell pellets frozen at − 80 °C until analysis.

Sample preparation.
Individual pellets from each of the cornea-and BM-derived myofibroblast preparations were homogenized in 100 mM triethylammonium bicarbonate containing 2% SDS, the protein extracted three times from the cell debris and quantified by AccQ-Tag amino acid analysis 17 . Approximately 290 µg protein per rabbit was recovered from the corneal myofibroblast cultures and 850 µg protein per rabbit from the BM myofibroblast cultures. Soluble myofibroblast protein (100 µg) from each of the specimens was reduced with tris-(2-carboxyethyl) phosphine, cysteines alkylated with methyl methanethiosulfonate, then the protein was precipitated with acetone overnight. Protein pellets were washed two times with ice cold 67% acetone, gently blown-dry with argon and re-suspended in 50 mM mM triethylammonium bicarbonate containing 0.5 mM CaCl 2 and digested overnight at 37 °C with trypsin (initially with 2% trypsin (w/w), followed in 2 h with another 2% (w/w), and the next day with another 1% (w/w) for 2 h additional incubation). Following proteolysis, soluble peptides were quantified by AccQ-Tag amino acid analysis. ITRAQ labeling and peptide fractionation. iTRAQ labeling with an 8-plex iTRAQ kit were performed as previously described [17][18][19][20][21] . In this study, tryptic digests of each myofibroblast preparation (100 µg/specimen) were labeled individually with a different iTRAQ tag and the labeled specimens mixed together in equal amounts and fractionated by reverse phase high performance liquid chromatography (RPHPLC) at pH 10 on a Waters xBridge BEH300 C18 column (3.5µ particle size, 2.1 × 100 mm). Chromatography was performed at a flow rate . The UniProt rabbit database is currently incomplete therefore protein identification utilized both the rabbit and human databases as rabbits are closely related phylogenetically to primates 22 . Proteins were identified in four categories including: (1) proteins characterized only in the rabbit database; (2) proteins characterized in both the rabbit and human databases; (3) proteins uncharacterized in the rabbit database but characterized in the human database; and (4) proteins characterized only in the human database. Sequence Identity between identified rabbit and human proteins was determined using Blast 2.9.0 23 . Protein identification required detection of a minimum of two unique peptides per protein and a database gene symbol. Database search parameters were restricted to three missed tryptic cleavage sites, a precursor ion mass tolerance of 10 ppm, a fragment ion mass tolerance of 20 mmu and a false discovery rate of ≤ 1%.

Results
Overview. Cornea and BM were isolated from four New Zealand white rabbits and four myofibroblast primary cultures were generated from each tissue. Myofibroblast identity and homogeneity were confirmed by immunocytochemistry ( Fig. 1) and paired cornea-and BM-derived myofibroblast specimens were analyzed by LC MS/MS iTRAQ technology, yielding a total of 2420 proteins quantified, of which 2329 were quantified in ≥ 3 rabbits. Proteomic results are summarized in Table 1 and presented in detail for each rabbit in Supplemental Tables S1-S4. These results include protein accession numbers and descriptions, gene symbols, protein ratios (cornea/BM), number of unique peptides, number of summed peptide intensities, percent sequence coverage, database identification category, and percent identity between rabbit and human proteins. The proteomic results from all four paired samples were similar in quality and exhibited near-to-normal distributions (Fig. 2) and therefore the quantitation was suitable for averaging. The mean relative abundance of proteins quantified in all the myofibroblast samples is presented in Supplemental Table S5, including sample frequency, standard error of the mean, and moderated p values adjusted for multiple testing, A total 673 DE proteins were identified, as illustrated by Volcano plot (Fig. 3) and itemized in Supplemental www.nature.com/scientificreports/ blast samples. Three hundred sixty proteins (~ 15%) were found significantly more abundant in cornea-than BM-derived myofibroblasts and are considered to be differentially expressed (Table S5). Thirty-three of these 360 proteins exhibited ratios ≥ 2 SD from the mean. Three hundred thirteen DE proteins were found significantly more abundant in BM-than cornea-derived myofibroblasts (Table S5), including 63 proteins exhibiting ratios ≥ 2 SD from the mean. Proteomic differences between these two types of myofibroblasts are suggested by a comparison of 30 proteins that are significantly more abundant in cornea-( Table 2) or BM- (Table 3)      www.nature.com/scientificreports/ Table 2. Differentially expressed proteins proteins significantly more abundant in rabbit myofibroblasts from cornea than from BM. The above 30 proteins were selected from 360 differentially expressed proteins more abundant in rabbit myofibroblasts from cornea than from BM. Each exhibited a protein ratio ≥ 1 SD from the mean and an adjusted pvalues ≤ 0.05 in ≥ 3 paired myofibroblast samples. All differentially expressed proteins are illustrated in Fig. 2 and identified in Supplemental Table 5.  (Table 5) support significant differences in the molecular and cellular functions of these two types of myofibroblasts.

Independent evidence supporting the iTRAQ protein quantitation. Western blot analysis was
used to independently corroborate the iTRAQ protein quantitation by analysis of three proteins likely to be involved in fibrosis mediated by myofibroblasts (collagen type III, collagen type VII and collagen type XI.) Consistent with the iTRAQ quantitation, immunoblot results confirmed that collagen type III and collagen type XI were decreased (Fig. 4A, B) and collagen type VII was elevated (Fig. 4C) in cornea-derived myofibroblasts compared to BM-derived myofibroblasts, respectively.

Discussion
Myofibroblasts are critical mediators of fibrosis that may occur in most organs of animals after injury 1 . Typically, at least two progenitor cells to myofibroblasts have been found to participate in the pathophysiology of fibrosis in each organ evaluated [7][8][9]13 . In the present study, we have compared the proteome of corneal keratocyte-derived myofibroblasts to that of BM-derived myofibroblasts-both of which have been shown to be major progenitors to myofibroblasts in fibrosis that occurs after corneal injury 6,13 . The keratocyte-derived and BM-derived myofibroblasts used in this study were at similar stages of myofibroblast differentiation as cell culture procedures were carefully coordinated and 100% of the cells in all cultures were α-SMA+ 16 .
The present study used quantitative proteomics technology to identify DE proteins in myofibroblasts derived from cornea and BM progenitors. The data indicates about 29% of the proteins quantified were differentially expressed between these two types of myofibroblasts. Proteomic differences were confirmed by Western blot analysis of three proteins likely to be involved in the pathophysiology of fibrosis, namely collagen type III, collagen type VII and collagen type XI. Clues to progenitor-dependent differences in myofibroblasts were suggested by bioinformatic analysis of the DE proteins. Canonical pathways involving mitochondrial dysfunction, oxidative phosphorylation and sirtuin signaling were implicated as most predominant in cornea-derived cells (Table 4), and pathways involving glycolysis I, integrin signaling and remodeling of epithelial adherens junctions as most predominant in BM-derived cells (Table 5). Molecular and cellular functions of cornea-and BM-derived myofibroblasts as projected from Ingenuity Pathway Analysis of the DE proteins, also were significantly different. RNA processing, mRNA splicing, transport and export, protein synthesis, and homologous cell recombination were the top functions implicated by bioinformatic analysis of DE proteins more abundant in cornea-derived myofibroblasts (Table 4). Top functions of DE proteins more abundant in BM-derived myofibroblasts were bioinformatically centered on cellular organization, degranulation of cells, microtubule dynamics, fibrogenesis, cell movement, and formation of filaments and cell protrusions (Table 5). For example, since BM-derived myofibroblasts produce much more collagen type XI and collagen type III, they likely contribute greatly to structure and strength of fibrotic tissue in the cornea and may contribute most of the collagen type III deposited in the cornea after injuries 28 . Conversely, since corneal keratocyte-derived myofibroblasts produce more collagen type VII, they may be more likely to modulate cytokine production by adjacent fibroblasts in the healing stroma 29 .
The bioinformatic results support the likelihood of progenitor-dependent functional differences in myofibroblasts, however, potentially limiting factors warrant acknowledgement. These factors include the relatively small sample size employed, namely only four rabbits, and the fact that the rabbit protein sequence databases are not well developed and are incompletely curated. Nevertheless, rabbits are close phylogenetic relatives of humans 22 and utilization of the human UniProt sequence database, one of the most complete and well-curated databases available, supports the reliability of the protein identifications. Notably, the overall level of significant proteomic differences (~ 29%) observed between cornea-and BM-derived myofibroblasts is 2-3× greater than the variability reported from repetitive proteomic analysis of a variety of normal tissues 30,31 . Furthermore, the levels of three different collagens thought to contribute to corneal fibrosis, were confirmed to be different between keratocytederived myofibroblasts and BM-derived myofibroblasts. It's possible that the BM-derived myofibroblasts analyzed in this study could have been heterogeneous if they were derived from both bone marrow fibrocytes and bone marrow stromal cells after stimulation with TGF beta-1 32,33 . Another possible limitation is that these two types of myofibroblasts were cultured and characterized in vitro, and it is not certain that the cells generated in vivo and in vitro are the same. However, the expression of vimentin, α-SMA and desmin found for both myofibroblast types in vitro was similar to findings after fibrosis-producing injuries in situ 34 .
The results of this study suggest that the myofibroblasts derived from different progenitors contribute differentially, and perhaps additively, to the fibrosis response to injury in the cornea. This further suggests that the character of the fibrotic tissue may vary depending on the relative contributions of the myofibroblast progenitors. For example, after anterior corneal injury produced by irregular phototherapeutic keratectomy (PTK) to inhibit epithelial basement membrane regeneration in mice, 30 to 70% of myofibroblasts were derived from BM-derived progenitors, with the remaining myofibroblasts developed from keratocyte-derived progenitors 6 , although some possibly developed from Schwann cells 35 . How the variation in myofibroblasts would affect properties of fibrosis such as contractility, opacity or persistence in the cornea remains unknown. A recent in vitro study 36 showed that the numbers of alpha-smooth muscle actin+ myofibroblasts generated from either keratocyte-derived precursors or BM-derived precursor cells were higher when both cells were co-cultured together in a culture flask Table 3. Differentially expressed proteins proteins significantly more abundant in rabbit myofibroblasts from bm than from cornea. The above 30 proteins were selected from 313 differentially expressed proteins more abundant in rabbit myofibroblasts derived from BM than from cornea. Each exhibits a protein ratio ≥ 1 SD from the mean and an adjusted p values ≤ 0.05 in ≥ 3 paired myofibroblast samples. All differentially expressed proteins are illustrated in Fig. 2 and identified in Supplemental Table 5. www.nature.com/scientificreports/ (juxtacrine) as compared to when BM-derived precursor cells and keratocyte-derived precursor cells were coculture in different compartments of a Transwell System (paracrine). This suggests that the presence of the two different myofibroblasts cells in the stroma after injury may potentiate the overall fibrosis response. Our current proteomic and bioinformatic results suggest that BM-derived myofibroblasts may be more prone than corneaderived precursors to impact cellular organization and the formation of excessive cellular and extracellular material characteristic of fibrosis. Hopefully, the findings of this study will stimulate future research to better understand the contributions of myofibroblasts from different precursors to the fibrosis response, and toward the development of more effective therapies to fibrotic tissue damage.    a, b, and c). A representative full-length western blot of the three performed for each protein is shown. Beta actin western blots were used to demonstrate equivalent loading (lower panels in a, b, and c). Each blot underwent densitometric analysis using Image J software (NIH, Bethesda, MD) of each of the three western blots from different experiments (n = 3); error bars reflect mean ± SD. All three differences were statistically significant (P < 0.05). CM corneal keratocytederived myofibroblast; BM BM-derived myofibroblasts.