Proteomic similarities in steroid responsiveness in normal and glaucomatous trabecular meshwork cells

TM cell cultures

All human specimens were post-mortem tissues collected with adherence to the principles expressed in the Declaration of Helsinki. Post-mortem human eyes were obtained from the Lions Eye Institute for Tissue and Research in Tampa, FL, which is accredited by the Eye Bank Association of America. TM cells were isolated from TM tissue explants derived from both open angle glaucoma and nonglaucomatous control donors. The glaucoma status was indicated from donor medical histories. The average death to preservation time was 7.7±3.3 h. Eyes were stored at 4 °C until the TM was dissected, generally within 24–36 h. Primary cultures were established and TM cell morphology and purity were characterized as previously described [14,15]. Human TM cells were grown in Dulbecco’s modified Eagle’s medium (HyClone, Logan UT) supplemented with 10% fetal bovine serum (GIBCO, Grand Island, NY), 1% penicillin-streptomycin (HyClone) and 1% L-glutamine (Thermo-Scientific Hyclone, Logan, UT) to confluency in T-25 flasks or in 6-well plates. Primary cultures of human TM cells from 2 POAG and 2 non-glaucomatous donors were treated with or without Dex (100 nM) for 10 days, yielding 4 Dex-treated and 4 untreated TM cell cultures (Table 1).

iTRAQ labeling, SCX chromatography, protein identification, quantitation, and bioinformatics

Detailed methods for sample preparation, iTRAQ labeling, strong cation exchange chromatorgraphy (SCX), protein identification and quantitation and bioinformatics have been described elsewhere [16-18]. Briefly, for proteomic analyses, proteins were extracted from TM cells [18], quantified by amino acid analysis [19], reduced, alkylated, and digested with trypsin [18]. Tryptic peptides from Dex-treated TM cells were labeled with iTRAQ tag 117 and mixed with an equal amount of tryptic peptides from the corresponding untreated cell sample labeled with iTRAQ tag 115. Each peptide mixture was fractionated by strong cation exchange (SCX) chromatography and fractions collected for LC MS/MS. LC MS/MS was performed with a QTOF2 mass spectrometer equipped with a Cap LC system (Waters Corporation, Milford, MA). Protein identification used MASSLYNX 4.1 software (Waters), the Mascot search engine (Matrix Science, Boston, MA), and the SwissProtein human sequence database (version 56.0, ~20,000 total sequences). Two unique peptides per protein and Mascot peptide ion scores ≥25 were required for all protein identification and quantitation. To estimate false discovery rates, all peptide MS/MS spectra were searched (Matrix Science, version 2.2) against a randomized decoy database constructed from the above SwissProtein database with a script provided by Matrix Science [16-18]. Protein quantitation from iTRAQ labeling required ion intensities ≥10 for all iTRAQ tags and was achieved with code written in the statistical program R. To average results over multiple samples, protein ratios were normalized to the protein median per sample, then average protein ratios (unadjusted), standard errors of the mean (SEM) and p-values (two sided t-Test for the null hypothesis that the protein mean=0 in log space) were calculated. Average protein ratios were adjusted to give greater weight to data with lower SEM values then adjusted SEMs and p-values were determined from the adjusted means of the multiple measurements [16-18]. Bioinformatic analyses were performed with The Protein ANalysis THrough Evolutionary Relationships (PANTHER) Classification System, Ingenuity Pathways Analysis 8.5 (Ingenuity Systems, Redwood City, CA), and the SwissProtein database.

Western Analyses

Immunoblots were performed as previously described [18]. Briefly, TM cell cultures were washed with PBS twice, and proteins were extracted with Mammalian Protein Extraction Buffer (ThermoScientific) containing 1% protease inhibitor cocktail (Thermo Fischer Scientific, Rockford, IL). Protein concentrations in TM lysates were determined by the BioRad Dc Protein Assay (Bio-Rad Laboratories, Hercules, CA). SDS-PAGE was performed on 10% acrylamide gels with 30 µg protein applied per lane, and proteins were electrophoretically transferred to polyvinylidene fluoride membranes (EMD Millipore, Billerica, MA). Membranes were blocked with 5% non-fat dry milk in TBST buffer then incubated overnight at 4 °C with primary antibody. The membranes were washed with TBST and probed with horseradish peroxidase-conjugated secondary antibody in 3% non-fat milk in TBST for 1 h at room temperature. Primary antibodies included mouse monoclonal anti-activated leukocyte cell adhesion molecule antigen (CD166; ALCAM; Novacastra/NCL-CD166, diluted 1:40; Abcam, Cambridge, MA) and rabbit polyclonal anti-Tplastin (Plastin-3; #ab45769, diluted 1:1,000; Abcam). Rabbit monoclonal anti-glyceraldehyde-3-phosphate dehydrogenase was used as a loading control (#14C10, diluted 1:1,000; Cell Signaling/Millipore, Billerica, MA). Secondary antibodies included horseradish peroxidase-conjugated secondary antibody (goat anti-mouse IgG; #sc2005, diluted 1:10,000; Santa Cruz, Santa Cruz, CA) or goat anti-rabbit IgG (#7074 diluted 1:10,000; Cell Signaling). Proteins were detected using enhanced chemiluminescence and a Fluor ChemTM 8900 imager (Alpha Innotech/ProteinSimple, Santa Clara, CA). Immunoreactivity was quantified using Image J software (NIH), and statistical analyses were performed using the two-sided t-test.

The majority of the determined proteome was unchanged by Dex-treatment

About 94% of all 718 proteins and ~88% of the 341 proteins quantified in three or more samples appear to be present in similar amounts in Dex-treated and untreated cells (Figure 2 and Appendix 5). Accordingly, the determined proteome largely reflects that of untreated TM cells. The quantified proteins include ~6% secreted, ~48% cytoplasmic, ~22% membrane and membrane-associated, ~15% nuclear, and ~11% mitochondrial proteins. These proportions are very similar to that obtained in quantitative proteomic analysis of the same TM cells treated with transforming growth factor beta 2 (TGFβ₂) [18]. In the current study, Dex-treatment decreased the abundance of more secreted proteins (n=3) than it increased (n=0) and elevated the abundance of more mitochondrial proteins (n=6) than it decreased (n=2).

TM proteins increased by Dex

Twenty proteins were elevated ≥1 SD above the mean ratio (p<0.05) in ≥3 samples by Dex-treatment, including both normal and glaucomatous TM cells (Table 2). Six proteins were elevated ≥2 SD by Dex treatment. Elevated proteins were associated with the cytoskeleton, ECM, mitochondrial metabolism and carbohydrate metabolism. The majority of these proteins have not previously been associated with GC-signaling in TM cells.

TM proteins decreased by Dex

Twenty proteins were significantly decreased ≥1 SD below the mean ratio (p≤0.55) in ≥3 samples by Dex-treatment, including 4 reduced ≥2 SD (Table 3), and all were changed in both normal and glaucomatous TM cells. Decreased proteins were associated with stress response, cellular defenses, protein processing, the cytoskeleton, the extracellular Matrix (ECM), and mitochondrial metabolism. The majority of the Dex-decreased proteins have not previously been considered steroid responsive in TM cells.

Independent evidence supporting the iTRAQ protein quantitation

Western blot analysis (Figure 3) demonstrated Dex-increased amounts of ALCAM/CD166 antigen (p=0.02) and Dex-decreased amounts of plastin-3 (p<0.01), corroborating the iTRAQ protein quantitation. These proteins were selected only because of their presence in Table 2 and Table 3 and the availability of antibodies useful for western analysis. Dex-decreased collagen α1(1) is supported by a previous proteomic study [20] that reported Dex-down-regulation of collagen α1(1) in rat TM (Table 3). Dex-increased amounts of sorbin and SH3 domain containing protein 2 [12] and filamin B [12] are indirectly supported by gene expression studies that reported Dex-induced upregulation of these genes in TM cells.

Upregulated GC-responses in normal and glaucomatous TM cells

Twenty TM proteins were significantly increased in abundance following Dex-treatment (Table 2), supporting steroid-induced, upregulated expression of these polypeptides. Pathway analysis of these proteins highlights cellular movement, cell death, and cell morphology as one of the two highest scoring networks. Several cytoskeletal proteins were elevated, including filamin-B, which connects the cell membrane with the actin cytoskeleton and promotes actin filament branching. These findings are consistent with other studies showing that Dex stimulates an increase in cross-linked actin networks (CLANs) in cultured human TM cells [14]. Such morphological networks are also observed in vivo in both normal and glaucomatous TM [23]. Sorbin and SH3 domain-containing protein 2, elevated about threefold, interacts with a large number of proteins, linking signaling complexes with the cytoskeleton. Increased levels of focal adhesion proteins palladin, zyxin, integrin-linked protein kinase (ILK) and the adherens junction protein nexilin, are consistent with GC-stimulated formation of adherens junctions and tight junctions, previously observed in TM cells [24] and in mammary tumor cells [25]. Notably, ILK functions in the regulation of cell-cell and cell-matrix interactions and appears to have roles in both integrin and growth factor-signaling. Two other elevated proteins potentially impacting cell adhesion and cell motility were CD166 antigen (ALCAM) and basigin. Both are plasma membrane proteins that promote growth of cellular processes and activate leuckocytes. CD166 antigen also functions in cellular interactions in the nervous system [26] and basigin (also known as tumor cell-derived collagenase stimulatory factor) stimulates matrix metalloproteinase production and tumor progression. Finally polymerase 1 and transcript release factor was elevated ~26% by Dex treatment. This protein is essential for caveolae formation in mammals, and caveolae function in cell proliferation, cargo-specific endocytosis, and mechanosensation [27]. Overall, these 9 elevated proteins support the hypothesis that GC-treatment can alter TM cytoskeletal and cell-cell interactions that contribute to AH outflow resistance and elevated IOP [3].

The other high scoring network from pathway analysis highlights cell cycle, carbohydrate, and nucleic acid metabolism. Consistent with GC regulation of gluconeogenesis [28], four enzymes in the pentose-phosphate pathway were upregulated ~30%–40%, namely transaldolase, 6-phosphogluconate dehydrogenase, glucose-6-phosphate dehydrogenase, and triosephosphate isomerase. The consequences of disrupting the pentose shunt in the TM remain to be determined, but genetic defects in these 4 enzymes can cause severe disorders elsewhere in the body, including progressive neuromuscular dysfunction.

Six mitochondrial proteins and a 7th protein that could significantly alter mitochondrial functions were elevated by Dex-treatment. These mitochondrial proteins serve a variety of roles in energy production (ATP synthase δ), signal transduction (GTP:AMP phosphotransferase), oxidation reduction (sulfide:quinone oxidoreductase), chaperone activity (10 kDa heat shock protein), lipid metabolism (3-ketoacyl-CoA thiolase), and carbohydrate metabolism (component of 2-oxoglutarate dehydrogenase complex). These changes could disrupt TM homeostasis by altering oxidative phosphorylation, regulation of apoptosis by reactive oxygen species, and maintenance of intracellular calcium levels [29,30]. Acid ceramidase, the 7th elevated protein, is a cytosolic protein that functions in the sphingosine-ceramide salvage pathway in sphingolipid turnover [31]. Ceramides produced in the salvage pathway in human breast cancer cells accumulate in mitochondria, promote the relocalization of mitochondria around the nuclear envelope and have been linked to apoptosis and mediators of mitochondrial functions [32]. Impaired mitochondria in TM cells has recently been implicated in POAG [18,33,34]. The present results provide clear evidence that GC-signaling can modify mitochondrial protein expression in both normal and glaucomatous TM cells. Dysfunctional TM mitochondria could promote oxidative damage [18,35] and cellular senescence in the aqueous humor outflow pathway [36].

Down-regulated GC-responses in normal and glaucomatous TM cells

Twenty TM proteins were significantly reduced by Dex-treatment (Table 3), suggesting GC-induced down-regulation, and/or degradation. Pathway analysis of these 20 proteins highlight gene expression, free radical scavenging and molecular transport as the highest scoring networks. TM cells from POAG donors reportedly contain higher levels of reactive oxygen species and Ca²⁺ in the cytosol and the mitochondria [33,34]. Interestingly, ~35% (n=7) of the TM proteins reduced by Dex-treatment were calcium-binding proteins (denoted by ** in Table 3), supporting calcium dysregulation in the TM as possibly contributing to elevated IOP [33]. About 50% of the decreased proteins can be associated with cellular defense, stress response, and processing events, consistent with an anti-inflammatory response. Several of these exhibit protein-folding and chaperone functions, including FK506-binding proteins 9 and 10, calreticulin, peptidyl-prolyl cis-trans isomerase B, endoplasmic reticulum resident protein 29, and Serpin H1. Noteworthy, serpin H1 is a stress-induced, collagen-binding chaperone, and calreticulin functions in the GC receptor signaling pathway. Two damage-associated molecular pattern proteins (DAMPs) were also decreased by Dex, namely advanced glycation endproduct receptor 3 (galectin-3) and protein S100-A6. DAMPs are endogenous proteins released by damaged cells that are capable of binding and activating an inflammatory response through pattern recognition receptors such as the complement system, toll-like receptors, and receptor for advanced glycation endproducts (RAGE). Also decreased was proteasome subunit α type-1, important in the processing of MHC peptides, regulation of the cell cycle, and ubiquitin-dependent degradation processes. Others included the 60S acidic ribosomal protein P0 involved in translational elongation, and neutral α-glucosidase AB which functions in conjunction with calreticulin in the release of mono-glycosylated glycoproteins.

Six proteins impacting the cytoskeleton and/or ECM were decreased by Dex-treatment. Among this group were actin-binding proteins transgelin and plastin-3, as well as nestin, involved in intermediate filament assembly/disassembly and cellular remodeling. Altered levels of these proteins could facilitate GC-induced CLAN formation [14]. ECM protein collagen α1(1) was the most significantly reduced, down ~50%. Two enzymes that catalyze proline hydroxylation in collagen were also decreased, namely prolyl 4-hydroxylase α1 and α2. Collagen α1(1) fibrils are a major component of collagen beams in TM in vivo and at physiologic conditions require hydroxyproline to stabilize the triple-helical structure. Altered interactions between collagen fibrils and other ECM components have been reported to trigger collagen degradation and a loss of tissue-specific morphology [37,38]. Whether the reduced collagen α1(1) level we observed in TM cells is due to decreased expression or to degradation is not clear, but perturbed collagen levels in vivo would render the ECM more susceptible to collapse and debris deposition. Collectively, the observed proteomic changes could disrupt the TM cytoskeleton and associated ECM in both normal and glaucomatous TM.

Dex-treatment also decreased the abundance of proteins impacting mitochondrial metabolism (Table 3). Among these were mitochondrial proteins ATP synthase γ, important in energy production and non-specific lipid-transfer protein, functioning in lipid transport across membranes and possibly serving in steroid biosynthesis. Also decreased was serine/threonine-protein phosphatase 2A isoform PR65-α, which appears to be a regulator in cell adhesion, apoptosis and ceramide-associated processes. As noted, ceramides have been linked to apoptosis and may accumulate in the mitochondria [32]. These results are consistent with GC-inducable mitochondrial dysfunction in both normal and glaucomatous TM cells.

Comparison with previous studies of GC-signaling in TM cells

This is the most extensive quantitative proteomic analysis of GC-signaling in TM cells. A previous 2D-gel proteomic study of one immortalized TM cell line identified 163 proteins and reported Dex-down regulation of Rho GDP dissociation inhibitor (RhoGDI) [39]. In the present study, four different primary TM cell cultures were analyzed and RhoGDI was not changed by Dex treatment (Appendix 5). Only one other proteomic study of steroid signaling in the anterior segment is currently published. Proteomic changes induced by in vivo topical application of Dex to three rat eyes was investigated using peptide mass fingerprinting methods and 2D fluorescence difference gel electrophoresis [20]. Four Dex-altered TM proteins were reported, including upregulation of αA-crystallin and βA3-crystallin and down-regulation of collagens α1(1) and α2(1) [20]. We found Dex-decreased amounts of collagen α1(1), detected collagen α2(1) in 2 samples, but did not detect αA- and βA3-crystallins, although several other chaperones were decreased by Dex-treatment (Table 2). The earlier report suggested that Dex may induce impaired collagen processing [20] because only COOH-terminal propeptides were detected. This may occur but our study quantified seven internal peptides from throughout the collagen α1(1) sequence, supporting Dex-induced decreased expression of collagen α1(1).

Hundreds of Dex-induced transcript changes in human TM cells have been described in five gene-profiling studies [9-13]; however, only two genes were reported differentially expressed in all five studies, namely genes encoding myocilin and insulin-like growth factor binding protein 2 (both upregulated). We detected myocilin at elevated levels in two of the four TM cell samples (Appendix 1 and Appendix 3) but did not detect insulin-like growth factor binding protein 2. While myocilin is a glaucoma gene, it is not always elevated at the protein level in TM tissues in open angle glaucoma [20]. Myocilin is a secreted glycoprotein, which may account in part for variable TM levels. In our study, myocilin may have also been in the culture medium, which was not analyzed. Among transcripts reported differentially expressed in the TM by Dex [9-13], we found 27 corresponding gene products, from which reliable quantification was obtained on 11 proteins (ie, proteins found in ≥3 samples). Our proteomic data support upregulation of sorbin and SH3 domain containing protein 2 [12] and filamin B [12] but do not support upregulation of transgelin [11,12] nor the reported down-regulation of 3-ketoacyl-CoA thiolase β [9]. Seven other transcripts reported to be differentially expressed exhibited no significant change at the protein level in this study, namely lactate dehydrogenase A [11], clusterin [10], fibroblast muscle tropomyosin [9], skeletal β-tropomyosin [9], fibulin-1C [9], phosphatidylethanolamine binding protein [9], and thrombospondin [13]. The observed differences between proteomic and genomic data seem reasonable as only ~20% correlation generally exists between mRNA and corresponding protein levels in mammals [40,41]. Similar to our findings from quantitative proteomic analysis of TGFβ₂-signaling in TM cells [18], very few proteases or protease inhibitors were detected in this study (Appendix 5), perhaps because they were below detection limits, and none were found significantly altered by Dex-treatment.

Additional insight to steroid responsiveness in TM cells

To further probe molecular mechanisms of steroid responsiveness, we compared Dex-induced proteomic changes with previously identified protein changes in TM cells induced by TGFβ₂ [18]. TGFβ₂ is elevated in the anterior segment of glaucoma patients, and while the mechanism(s) responsible for elevated expression is unknown, substantial evidence implicates TGFβ₂ as a contributing factor in ocular hypertension [3,18]. Our TGFβ₂ investigation [18] used the same normal and glaucomatous TM cell strains and quantitative proteomic methods as used in the present study. For comparison of Dex- and TGFβ₂-induced proteomic changes, proteins quantified in ≥3 TM cell samples in both studies were sought that exhibited: (i) ratios at least 1 SD above or below the means in both data sets, and (ii) a p-value ≤0.055 in at least one of the treatments (indicating at least one of the treatments significantly changed the protein ratio). Twenty proteins meeting the above criteria were identified (Table 4), including four apparently altered by both TGFβ₂ and Dex (palladin, prolyl 4-hydroxylase α1 and α2, and FK506-binding protein 10). While additional studies with larger samples sizes are warranted to validate proteins altered by both Dex and TGFβ₂, these two modifiers appear to impact several common physiologic processes in both normal and glaucomatous TM cells. These processes include cytoskeletal/cell-cell interactions, cell-matrix/ECM remodeling, and mitochondrial metabolism [18]. Such common physiologic consequences suggest a molecular basis for the increased risk of ocular hypertension in steroid responders.

Conclusions

Based on our analysIs of a small number of TM cell samples, the results expand the repertoire of proteins participating in GC-signaling and support common steroid response mechanisms in both normal and glaucomatous TM cells. While we analyzed more cell samples than in previous studies of GC effects on the TM transcriptome or proteome, the sample size is limiting and further investigations are warranted. The observed proteomic changes implicate as consequences of steroid-treatment remodeling of the extracellular matrix, disorganization of the cytoskeleton, disruption of cell-cell interactions, and mitochondrial dysfunction in the TM. These same physiologic processes appear impacted by TGFβ₂-signaling in both normal and glaucomatous TM cells. Finally, the results contribute to a quantitative database of TM proteins.