Quantitative Proteomics: TGFβ₂ Signaling in Trabecular Meshwork Cells

Abstract

Purpose.: Transforming growth factor beta 2 (TGFβ₂) is often elevated in the aqueous humor (AH) and trabecular meshwork (TM) of patients with primary open-angle glaucoma (POAG) and appears to contribute to POAG pathogenesis. To better understand TGFβ₂ signaling in the eye, TGFβ₂-induced proteomic changes were identified in cells cultured from the TM, a tissue involved in intraocular pressure (IOP) elevation in glaucoma.

Methods.: Primary cultures of human TM cells from four donors were treated with or without TGFβ₂ (5 ng/mL) for 48 hours; then cellular protein was analyzed by liquid chromatography–mass spectrometry iTRAQ (isobaric tags for relative and absolute quantitation) technology.

Results.: A total of 853 proteins were quantified. TGFβ₂ treatment significantly altered the abundance of 47 proteins, 40 of which have not previously been associated with TGFβ₂ signaling in the eye. More than half the 30 elevated proteins support growing evidence that TGFβ₂ induces extracellular matrix remodeling and abnormal cytoskeletal interactions in the TM. The levels of 17 proteins were reduced, including four cytoskeletal and six regulatory proteins. Both elevated and decreased regulatory proteins implicate TGFβ₂-altered processes involving transcription, translation, and the glutamate/glutamine cycle. Altered levels of eight mitochondrial proteins support TGFβ₂-induced mitochondrial dysfunction in the TM that in POAG could contribute to oxidative damage in the AH outflow pathway, TM senescence, and elevated IOP.

Conclusions.: The results expand the repertoire of proteins known to participate in TGFβ₂ signaling, provide new molecular insight into POAG, and establish a quantitative proteomics database for the TM that includes candidate glaucoma biomarkers for future validation studies.

Primary open-angle glaucoma (POAG) is the most common form of the primary glaucomas and affects approximately 3 million Americans and more than 70 million people worldwide. ¹ Visual loss in POAG results from damage to retinal ganglion cells and the optic nerve; advanced age and elevated intraocular pressure (IOP) are risk factors. ^2,3 A significant proportion of POAG patients have what is termed normal-tension glaucoma in which IOP remains in the normal range and the cause of the neuropathy is unclear. ⁴ Nevertheless, IOP is still a major risk factor in normal-tension glaucoma because further lowering of IOP decreases disease progression. Despite the high prevalence of POAG and the identification of glaucoma susceptibility genes, ⁵ the molecular mechanisms of glaucoma are poorly understood. Elevated IOP in POAG appears related to pathologic changes in the aqueous humor (AH) outflow pathway that cause increased outflow resistance, particularly in the trabecular meshwork (TM). Abnormal accumulation of extracellular matrix (ECM), ^6,7 abnormal protein expression, ⁸ and changes in cytoskeletal interactions ^9,10 within the TM have been associated with increased AH outflow resistance and elevated IOP. Oxidative damage may also contribute to elevated IOP ¹¹ and play a role in POAG. ^12,13  

Transforming growth factor beta 2 (TGFβ₂) is an immunosuppressive factor in normal human AH that helps maintain the immune privilege of the eye. ^14,15 In addition to elevated levels of TGFβ₂ in the AH of POAG patients, as shown in numerous studies, ^{16 –18} TGFβ2 is elevated in glaucomatous TM tissues and cultured glaucomatous TM cells. ¹⁹ In vitro studies in the TM show that TGFβ₂ can induce ECM remodeling, ^20,21 inhibit cell proliferation, ²² induce senescence-like changes, ²³ and alter the actin cytoskeleton. Elsewhere in the body, TGFβ₂ mediates increased ECM deposition and has been implicated in fibrosis of the liver, ²⁴ kidney, ²⁵ and lung. ²⁶ Gene expression studies have reported changes in TGFβ₂-treated TM cells, ^21,27,28 including increased levels of transcripts encoding ECM and cytoskeletal components. Organ culture studies in which the anterior segment of human and porcine eyes is perfused with TGFβ₂ have reported increased IOP and ECM deposition in the AH drainage pathway. ^21,29,30 More recently, adenoviral gene transfer of TGFβ₂ in vivo to rodent eyes has led to reduced AH outflow and increased IOP. ³¹ This growing body of evidence implicates TGFβ₂ in POAG pathology. 

To better understand the molecular consequences of TGFβ₂ signaling in the anterior segment, we compared human TM cells with and without TGFβ₂ treatment using global quantitative proteomics methods. Liquid chromatography–mass spectrometry (LC MS/MS) isobaric tags for relative and absolute quantitation (iTRAQ; Applied Biosystems, Foster City, CA) technology was used to quantify TGFβ₂-induced protein changes in TM cells. Forty-seven significantly altered proteins were identified, 40 of which have not previously been associated with TGFβ₂ signaling in the TM. The results provide new molecular insight into the consequences of TGFβ₂ signaling in ocular hypertension and POAG. 

Methods

TM Cell Cultures

TM cells were isolated from postmortem human TM tissue explants derived from open-angle glaucoma and nonglaucomatous control donor eyes. Glaucoma status was indicated from donor medical histories. The average death to preservation time was 7.75 ± 3.3 hours. Eyes were stored at 4°C until the TM was dissected, generally within 24 to 36 hours. Primary cultures were established, and TM cell morphology and purity of these cultures were characterized as previously described. ²² 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, Waltham, MA) to confluence in T-25 flasks or in six-well plates. Primary cultures of human TM cells from two POAG and two nonglaucomatous donors were treated with or without TGFβ₂ (5 ng/mL) for 48 hours, yielding four TGFβ₂-treated and four4 untreated TM cultures. 

Sample Preparation, iTRAQ Labeling, SCX Chromatography, Protein Identification and Quantitation, and Bioinformatic and Western Blot Analysis

Detailed methods are provided in the Supplementary Methods available online. Briefly, for proteomic analyses, proteins were extracted from TM cells, quantified by amino acid analysis, reduced, alkylated, and digested with trypsin. Tryptic peptides from TGFβ₂-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. ^32,33 Each peptide mixture was fractionated by strong cation exchange (SCX) chromatography, and fractions were collected for LC MS/MS. LC MS/MS was performed with a QTOF2 mass spectrometer equipped with a capillary column system (Cap LC; Waters Corporation, Milford, MA). ^32,33 Protein identification used MASSLYNX 4.1 software (Waters), the Mascot search engine (Matrix Science), and the Swiss-Protein human sequence database. Protein quantitation from iTRAQ labeling was achieved with code written in the statistical program R. ^32,33 Bioinformatic analyses were performed with The Protein ANalysis THrough Evolutionary Relationships (PANTHER) Classification System, pathway analysis (Ingenuity Pathways Analysis 8.5; Ingenuity Systems, Redwood City, CA), and the Swiss Protein database. Western blot analyses were performed on polyvinylidene difluoride membranes using established protocols and commercially available antibodies. 

Results

From four primary cultures of human TM cells treated or untreated with TGFβ₂, a total of 853 proteins were identified and quantified with two or more peptides by LC MS/MS and iTRAQ quantification code written in R. The TM cell samples were from normal (n = 2) and glaucomatous (n = 2) donor eyes (Table 1). Quantitative results from each cell sample are presented in Supplementary Tables S1–S4, including protein ratios, SDs, P values, number of unique peptides quantified, percentage sequence coverage for each protein, and peptide false discovery rates. The data from all four TM samples were of comparable quality and appropriate for averaging based on consistently low peptide false discovery (average rates, 2.1% identity, 3.0% homology) and similar distributions of protein ratios. The average relative abundance of the 853 quantified proteins is presented in Supplementary Table S5, and the distribution of mean protein ratios for all proteins and those quantified in three or more TM samples (n = 450) are shown in Figure 1. The distributions in Figure 1 are slightly skewed, indicating that more proteins were increased in abundance than were decreased by TGFβ₂ treatment. Criteria for determining whether a protein was elevated or decreased included the average adjusted protein ratio and P value, with only proteins quantified in three or more cell samples used for comparative purposes. Proteins exhibiting average protein ratios (adjusted by SEM) above or below the mean by at least 1 SD (Fig. 1) and P < 0.06 were considered of higher or lower abundance. 

Table 1.

 

Trabecular Meshwork Cell Samples

Experiment	Cell Culture	Donor Age and Sex	Cell Passage	Protein Analyzed (μg)
1	NTM416-07	78/M	4	47
2	NTM496-05	82/F	3	54
3	GTM304-4	75/F	3	38
4	GTM477-02	85/F	4	67

 Each of the above primary TM cell cultures was treated with or without TGFβ₂, then analyzed by LC MS/MS iTRAQ technology. NTM, normal trabecular meshwork; GTM, glaucomatous trabecular meshwork; M, male; F, female.

Figure 1.

 

Distribution of protein ratios. The log₂ mean distribution of protein ratios (TGFβ₂-treated TM/untreated TM) are shown for (A) all 853 quantified proteins and (B) 450 proteins quantified in three or more cell samples. Median, mean, and SD values are indicated; protein ratios between 1 and 2 SD from the mean are shaded. The distribution of protein ratios is near normal for all proteins but with some spread above 1.0 (log₂ mean ratio 0), suggesting the increased abundance of several proteins after TGFβ₂ treatment.

Majority of the Determined Proteome Reflects That of Untreated TM Cells

Approximately 95% of the 853 quantified proteins and ∼90% of the 450 proteins quantified in three or more samples (Supplementary Table S5) were present in similar amounts in TGFβ₂-treated and untreated TM cells. Accordingly, the overall determined proteome largely reflects that of untreated TM cells. The quantified proteins include ∼5% secreted, ∼46% cytoplasmic, ∼22% membrane and membrane-associated, ∼15% nuclear, and ∼12% mitochondrial proteins. Among the significant proteomic changes after TGFβ₂ treatment, more secreted proteins were elevated (∼20%) than decreased (0%) and twice as many mitochondrial proteins were decreased (∼41%) than elevated (∼20%). 

TM Proteins Increased by TGFβ₂

Thirty proteins were elevated ≥1 SD (P < 0.06) above the mean ratio after TGFβ₂ treatment, including 11 proteins elevated ≥2 SD (Table 2). The majority of these proteins have not previously been associated with TGFβ₂ signaling in the TM; however, more than 50% can be grouped by function into processes previously associated with TGFβ₂ treatment, such as ECM remodeling and cytoskeletal interactions. For example, among the most elevated proteins were ECM-associated SPARC (secreted protein acidic and rich in cysteine), CTGF (connective tissue growth factor), cytoskeletal-associated myosin regulatory light chain 9, and caldesmon. Approximately 20% of the elevated proteins can be grouped into protein binding, folding, and chaperone processes, and another ∼26% can be grouped into regulatory processes involving transcription, translation, steroid binding, and mitochondrial metabolism. Proteins significantly increased in abundance by TGFβ₂ treatment account for ∼7% of those quantified in three or more samples. 

Table 2.

 

Elevated Proteins in TGFβ₂-Treated Human TM Cells

SwissProt Accession	Protein	Subcellular Source	Sample Frequency (Total = 4)	Mean Protein Ratio	SEM	P
ECM Remodeling
P09486	SPARC†	AC	3	6.70*	0.14	0.006
P29279	Connective tissue growth factor	A	4	2.94*	0.18	0.009
P02461	Collagen α1 (III)	A	4	2.89*	0.13	0.004
O00469	Procollagen-lysine,2-oxoglutarate 5-dioxygenase 2†	C	4	2.58*	0.09	0.002
P07996	Thrombospondin-1†	A	3	2.41*	0.15	0.030
Q16270	Insulin-like growth factor-binding protein 7	A	3	2.40*	0.17	0.036
O15460	Prolyl 4-hydroxylase subunit α2†	C	4	2.11*	0.22	0.042
Q96CG8	Collagen triple helix repeat-containing protein 1	A	3	2.04*	0.14	0.036
P13674	Prolyl 4-hydroxylase subunit α1†	C	4	1.53	0.14	0.057
P55145	Mesencephalic astrocyte-derived neurotrophic factor	A	4	1.45	0.08	0.020
Cytoskeletal Interactions
P24844	Myosin regulatory light chain 9†	B	4	2.12*	0.11	0.006
Q05682	Caldesmon	B	4	2.06*	0.21	0.042
Q8WX93	Palladin	B	4	1.70	0.06	0.003
Q9HBL0	Tensin-1	B	3	1.67	0.07	0.017
P37802	Transgelin-2	B	4	1.56	0.08	0.012
Q9Y2B0	Protein canopy homolog 2	C	3	1.45	0.03	0.008
Protein Folding and Chaperone Processes
P02511	α-Crystallin B	B	4	3.37*	0.16	0.004
Q8NBS9	Thioredoxin domain-containing protein 5	C	4	1.66	0.06	0.003
Q9Y4L1	Hypoxia upregulated protein 1	C	4	1.49	0.09	0.020
Q96AY3	FK506-binding protein 10†	C	4	1.48	0.05	0.005
P11021	78 kDa glucose-regulated protein	C	4	1.43	0.06	0.008
Q15084	Protein disulfide-isomerase A6	C	4	1.41	0.10	0.041
Regulatory Processes
P50479	PDZ and LIM domain protein 4†	B	4	1.68	0.14	0.036
Q14192	Four and a half LIM domain protein 2†	BD	4	1.67	0.09	0.012
Q9P2E9	Ribosome-binding protein 1	C	4	1.65	0.07	0.006
O00264	Membrane-associated progesterone receptor component 1	C	4	1.53	0.14	0.059
P62851	40S ribosomal protein S25	B	4	1.50	0.12	0.045
O94925	Glutaminase, kidney isoform	E	3	1.43	0.07	0.037
P24534	Elongation factor 1-β	B	4	1.42	0.06	0.009
O15173	Membrane-associated progesterone receptor component 2	C	4	1.39	0.10	0.052

 Adjusted mean protein ratios (TGFβ₂-treated/control), standard error of the mean (SEM). P values, and subcellular source for abundant proteins (from Supplementary Table S5). Abundant proteins exhibited mean ratios at least 1 SD above the mean and P < 0.06 (Fig. 2). All proteins were quantified in ≥3 TM cell samples (median ratio = 1.00; mean = 1.01; SD = 0.31; n = 450).

*  Ratios at least 2 SD above the mean. Protein subcellular source from the Swiss Protein database: A, secreted; B, cytoplasmic; C, membrane; D, nuclear; E, mitochondrial.

†  Calcium or metal ion-binding proteins.

TM Proteins Decreased by TGFβ₂

In response to TGFβ₂ treatment, 17 proteins were decreased in abundance ≥1 SD (P < 0.06) below the mean ratio (Table 3); none of these proteins have previously been associated with TGFβ₂ signaling in TM cells. CD9 antigen and mitochondrial superoxide dismutase 2 (SOD2) were the most significantly reduced proteins (decreased ∼64% and ∼46%, respectively). Four of the reduced proteins are associated with cytoskeletal interactions and six with a variety of regulatory processes, including transcription, translation, and the immune response. Notably, more than 40% of the decreased proteins were from the mitochondria and are involved in critical metabolic functions. Proteins significantly reduced in abundance accounted for ∼4% of those quantified in three or more samples. 

Table 3.

 

Decreased Proteins in TGFβ₂-Treated Human TM Cells

SwissProt Accession	Protein	Subcellular Source	Sample Frequency (Total = 4)	Mean Protein Ratio	SEM	P
Signaling/Regulation
P39023	60S ribosomal protein L3	B	4	0.69	0.11	0.044
Q71DI3	Histone H3.2	D	4	0.68	0.08	0.014
Q99536	Synaptic vesicle membrane protein VAT-1 homolog	B	4	0.66	0.05	0.004
P01891	HLA class I histocompatibility antigen, A-68 α	C	3	0.65	0.08	0.032
P04899	Guanine nucleotide-binding protein G(i), α2	B	4	0.59	0.10	0.012
P21926	CD9 antigen	C	3	0.36*	0.24	0.051
Cytoskeletal Interactions
P07737	Profilin-1	B	4	0.70	0.12	0.059
Q9ULV4	Coronin-1C	B	3	0.60	0.05	0.011
O00159	Myosin-1c	BC	3	0.58	0.12	0.047
P68371	Tubulin β-2C	B	3	0.57	0.05	0.007
Mitochondrial Metabolism
P45880	Voltage-dependent anion-selective channel protein 2	CE	4	0.73	0.05	0.011
P25705	ATP synthase subunit α	CE	4	0.72	0.05	0.006
P55084	Trifunctional enzyme subunit β	E	4	0.70	0.06	0.008
P24539	ATP synthase subunit β	CE	4	0.68	0.11	0.037
P21796	Voltage-dependent anion-selective channel protein 1	CE	4	0.66	0.13	0.044
P00367	Glutamate dehydrogenase 1	E	4	0.60	0.10	0.012
P04179	Superoxide dismutase 2 [Mn]	E	4	0.054*	0.06	0.002

 Adjusted mean protein ratios (TGFβ₂-treated/control), SEM, P values, and subcellular source for less abundant proteins (Supplementary Table S5). Less abundant proteins exhibited mean ratios at least 1 SD below the mean and P < 0.06 (Fig. 2). All proteins were quantified in ≥3 TM cell samples (median ratio = 1.00; mean = 1.01; SD = 0.31; n = 450).

*  Ratios at least 2 SD below the mean. Protein subcellular source from the Swiss Protein database and Ingenuity Pathway Analysis 8.5: A, secreted; B, cytoplasmic; C, membrane; D, nuclear; E, mitochondrial.

Independent Evidence Supporting the ITRAQ Protein Quantitation

Western blot analysis was used to independently evaluate proteomic changes in the TM cell samples with and without TGFβ₂ treatment. Densitometric results (Fig. 2) from immunoblots corroborated increased amounts of caldesmon and decreased amounts of SOD2 after TGFβ₂ treatment and were in reasonable agreement with the iTRAQ data in both ratios and P values. Reports in the literature support TGFβ₂-induced changes observed for at least six other proteins. Specifically, Western blot analyses published by others show that TGFβ₂ treatment of human TM cells results in increased amounts of SPARC, ³⁴ CTGF, ³⁵ thrombospondin-1, ³⁵ and α-crystallin B ³⁶ and that CTGF treatment of human TM cells results in elevated levels of collagen α1 type III. ³⁷ In addition, a previous proteomic study reported TGFβ₂ induced increased amounts of transgelin-2, α-crystallin B, and possibly 78 kDa glucose-regulated protein in TM cells based on 2D gel spot staining intensity. ²⁷  

Figure 2.

 

Western blot analysis. Western blot analysis was used to evaluate the relative amounts of proteins in TGFβ₂-treated and untreated TM cells (n = 4 samples each). Average densitometry ratios (TGFβ₂-treated/untreated) and P values (two-sided t-test) for caldesmon and SOD2 support the quantitative data shown in Tables 2 and 3. Previous protein analyses corroborate at least six additional TGFβ₂-induced proteomic changes detected in this study (see Results).

Discussion

To better understand TGFβ₂ signaling in the anterior segment, we quantified proteomic changes in four primary cultures of human TM cells after TGFβ₂ treatment using LC MS/MS iTRAQ technology. The study was not designed to reveal differences between glaucomatous and nonglaucomatous cell cultures but rather to probe the impact of TGFβ₂ signaling on TM cells. Although glaucomatous and normal TM cells may exhibit differences in TGFβ₂ signaling yet to be identified, all TGFβ₂-altered proteins in this study were derived from both TM donor populations and definitively demonstrate common consequences of TGFβ₂ signaling in normal and glaucomatous TM cells. Most of the 853 quantified proteins (∼90%) were present in similar amounts in TGFβ₂-treated and untreated cells, indicating that the determined proteome largely reflects that of untreated TM cells. TGFβ₂ treatment significantly altered the abundance of 47 proteins; independent data from Western blot analysis and 2D-PAGE corroborated eight or more (∼17%) of these TGFβ₂-induced proteomic changes. Bioinformatic analysis suggested that major biological processes associated with the quantified proteins (Fig. 3) include cellular metabolism (∼28% of the 853 proteins), signal transduction (∼8%), cell structure/motility (∼8%), immunity and defense (∼5%). and cell adhesion (2%). Biological analysis and interpretation (Ingenuity Pathway Analysis [IPA]; Ingenuity) of these proteins implicated gene expression and cellular assembly and organization as the highest scoring networks. 

Figure 3.

 

Biological functions of TM proteins. Functional classification of the 853 proteins quantified in human TM cells was performed with the PANTHER Classification System.

Molecular Clues to POAG from TM Proteins Elevated by TGFβ₂ Treatment

Thirty TM proteins were significantly increased in abundance after TGFβ₂ treatment (Table 2), supporting the upregulated expression of these polypeptides. Analysis (IPA; Ingenuity) of the 30 elevated proteins suggested that TGFβ₂ impacted two major networks, one involving amino acid metabolism, posttranslational modification, and small molecule biochemistry and the other involving gene expression, cellular compromise, and development. Seven of the elevated proteins (or their transcripts) were previously reported to be upregulated by TGFβ₂; the other 23 elevated proteins provide an expanded profile of TGFβ₂ signaling in TM cells. Though yet unconfirmed, calcification and mineralization of the TM have been hypothesized to contribute to elevated IOP in glaucoma, ³⁸ and 30% of the elevated proteins bind either calcium or zinc, as noted in Table 2. Consistent with previous reports, ^6,7,20,21 10 elevated proteins are involved in the structure and regulation of the ECM, and eight of these were increased more than twofold. SPARC, elevated more than sixfold, is a matricellular protein that binds multiple proteins and cell membranes and that has been hypothesized to contribute to elevated IOP in POAG in part because SPARC-null mice exhibit lower IOPs than wild-type controls. ³⁹ The SPARC gene, highly expressed in nonglaucomatous TM, ⁴⁰ has also been reported to be upregulated by mechanical stretching. ⁴¹ CTGF, a mediator of the effects of TGFβs on ECM remodeling, and thrombospondin-1, a major activator of latent TGFβs and elevated in the TM of some POAG patients, ⁴² were both found significantly elevated. Also elevated in this study were insulin-like growth factor-binding protein 7, which stimulates cell adhesion and mesencephalic astrocyte-derived neurotrophic factor (also known as Protein ARMET), which binds unfolded proteins and inhibits stress-induced cell death. In addition, levels of collagen α1 type III and several other proteins that modify collagen or collagen deposition in the ECM—namely procollagen-lysine 2-oxogluturate 5-dioxygenase 2, prolyl 4-hydroxylase α2 and α1, and collagen triple helix repeat-containing protein 1—were elevated. 

Previous studies suggest that the TM cytoskeleton is involved in regulating AH outflow and IOP, perhaps in part through the formation of cross-linked actin networks (CLANs). ^9,10,43 CLANs are morphologically diverse structures in TM cells, appearing as tangles to geodesic dome-like, and observed in both normal and glaucomatous TM in vivo. ⁴³ In this study, TGFβ₂ induced apparent increased expression of the actin-binding proteins caldesmon, palladin, tensin-1, and transgelin-2, and cytoskeletal regulators like myosin regulatory light chain 9 and its binding partner protein canopy homolog 2. Altered expression of these proteins could promote CLAN formation and impact TM cell shape, motility, proliferation, contractile and tone properties, and cell-cell and cell-ECM interactions. 

Six proteins increased in abundance by TGFβ₂ were associated with protein folding and chaperone activities, with α-crystallin B the most significantly elevated (>3-fold). Others in this group included thioredoxin domain-containing protein 5, hypoxia upregulated protein 1, FK506-binding protein 10 (also known as peptidyl-prolyl cis-trans isomerase FKBP10), 78 kDa glucose-regulated protein, and protein disulfide-isomerase A6. Overall, about twice as many proteins were elevated than decreased by TGFβ₂ treatment, and the increased abundance of chaperones and folding proteins appear to be a stress response to metabolic changes, such as in protein synthesis and oxidation reduction. 

A variety of regulatory proteins were elevated ∼40% to 70% after TGFβ₂ treatment. The possibility that TGFβ₂ alters translational elongation is suggested by elevated levels of ribosome-binding protein 1, 40S ribosomal protein S25, and elongation factor 1β. TGFβ₂-altered transcriptional regulation is also suggested by elevated levels of PDZ and LIM domain protein 4 and four and a half LIM domain protein 2, which bind zinc and a zinc finger binding protein. Also elevated were two steroid-binding proteins, membrane-associated progesterone receptor components 1 and 2, suggesting the possibility that TGFβ₂ may play a role in steroid responsiveness in glucocorticoid-stimulated secondary glaucoma. ⁴⁴ Mitochondrial glutaminase, an enzyme catalyzing the conversion of glutamine to glutamate, was elevated more than 40% after TGFβ₂ treatment; notably, increased glutaminase activity in the rat retina has been associated with ocular hypertension. ⁴⁵ Seven other mitochondrial proteins were altered by TGFβ₂ treatment, as discussed. 

Molecular Clues to POAG from TM Proteins Decreased by TGFβ₂ Treatment

The levels of 17 TM proteins were significantly reduced after TGFβ₂ treatment (Table 3), implying downregulated expression, possible degradation, or both. Little is known about TGFβ₂-downregulated processes in the TM. The highest scoring network from pathway analysis (IPA; Ingenuity) of the 17 decreased proteins concerns developmental disorders, renal and urological diseases, and gene expression. TGFβ₂ lowered the abundance of four cytoskeletal components by ∼30% to 40%, including the actin-binding proteins profilin-1, coronin-1C, and myosin-1c and microtubule component tubulin β-2C. As noted, altered ratios of cytoskeletal proteins could promote the formation of CLANs and adversely impact TM cell properties and processes. ⁴³ In addition to cytoskeletal proteins, a variety of regulatory proteins were decreased by ∼30% to 60%. CD9 antigen, a membrane protein involved in cell adhesion and motility, was the most significantly reduced (>60%). Decreased amounts of 60S ribosomal protein L3 and histone H3.2 implicate TGFβ₂-altered translational elongation and transcription, respectively; all these processes are also potentially altered by other proteins elevated by TGFβ₂ treatment, as discussed. Altered GTPase and ATPase activities are implied by TGFβ₂-decreased levels of guanine nucleotide-binding protein Gi α2 and synaptic vesicle membrane protein VAT-1 homolog, respectively. Reduction of HLA class 1 histocompatability antigen A68α and CD9 antigen suggests TGFβ₂-altered immune responses in TM cells, as has been proposed in POAG for retinal glial cells and the optic nerve head. ⁴⁶  

Mitochondrial Dysfunction in the TM

In addition to elevated mitochondrial glutaminase, seven other mitochondrial proteins were significantly reduced by TGFβ₂ treatment (Table 3). These decreased proteins serve important roles in energy production (e.g., ATP synthases), transmembrane transport (e.g., voltage-dependent anion channel proteins), and oxidation reduction (e.g., glutamate dehydrogenase 1 [GLUD1] and SOD2), and their decreased levels could severely impair cellular functions. Mitochondrial dysfunction in retinal ganglion cells has long been associated with glaucoma. ^47,48 However, it is only recently that evidence suggests that elevated reactive oxygen species and impaired calcium regulation in TM cell mitochondria and TM mitochondrial DNA deletions may play a role in POAG. ^{49 –51} The identification of eight mitochondrial proteins altered in abundance by TGFβ₂ adds strong support to the hypothesis that mitochondrial dysfunction in the TM contributes to POAG. For example, TGFβ₂ has been reported to induce senescence in the TM, ²³ and elevated IOP has been reported to contribute to glutamate toxicity in the rat retina (by increasing glutaminase activity and reducing glutamate uptake). ⁴⁵ We found mitochondrial glutaminase elevated by ∼43% and mitochondrial GLUD1 decreased by ∼40% in TM cells after TGFβ₂ treatment. GLUD1 reduces the amount of glutamate by catalyzing its oxidation to 2-oxoglutarate. TGFβ₂ may induce TM senescence and cell death through mitochondrial dysfunction that in part promotes glutamate imbalance. Elevated IOP has also been reported to decrease total SOD activity in the rat retina, ⁵² and we observed ∼46% reduction in mitochondrial SOD2 in TM cells after TGFβ₂ treatment. Previous work has shown that ∼50% knockdown of mitochondrial SOD2 in mouse retinal pigment epithelium correlates with increased levels of oxidative protein modifications ⁵³ and that TGFβ₂ increases lipid peroxidation in human TM cells. ²³ We and others (Bhattacharya SK, et al. IOVS. 2004;45:ARVO E-Abstract 4560) ⁵⁴ have also observed increased levels of oxidative protein modifications in trabeculectomy tissues from POAG patients. Accordingly, the present data support the hypothesis that elevated IOP in POAG is due in part to TGFβ₂-induced mitochondrial dysfunction in the TM, with concomitant oxidative damage in the AH outflow pathway. 

Comparison with Other Studies of TGFβ₂-Induced Changes in TM Protein and Gene Expression

This is the most extensive quantitative proteomic analysis to date of changes in human TM cells in response to TGFβ₂. The only other global proteomic study of TGFβ₂-treated TM cells used peptide mass fingerprinting for protein identification and 2D gel staining intensity for protein quantification. ²⁷ Of the 56 proteins reported in the previous study, we detected 42 and obtained reliable quantitation (i.e., proteins found in three or more TM samples) for 12 of the 24 proteins reported to exhibit a significant change in 2D gel spot intensity. ²⁷ For these 12 proteins, two to three exhibited changes in agreement with the present results, namely increased amounts of transgelin-2, α-crystallin B, and possibly 78 kDa glucose-regulated protein (which was reported as 1 of 2 proteins in a gel spot). ²⁷ Comparison of these proteomic studies is complicated by the different methods used, including the analysis of pooled rather than individual donor TM samples and a lower concentration of TGFβ₂ for treating TM cells in the earlier study. Comparison of the present results with gene expression studies of TGFβ₂-treated TM cells ^21,27,28 reveals limited correlation, however; only ∼20% correlation generally exists between transcript and cognate protein levels in mammals. ⁵⁵ From microarray studies, more than 20 transcripts have been reported to be differentially expressed by TGFβ₂ in human TM cells. ^21,27,28 Significant proteomic changes in this study are consistent with reported TGFβ₂ upregulation of transcripts for thrombospondin 1, ^21,27 SPARC, ²³ and prolyl 4 hydroxylase subunit alpha 2. ²⁸ Proteases and protease inhibitors function in ECM remodeling in normal TM, ⁵⁶ and TGFβ₂ has been reported to upregulate the expression of the inactive proform of matrix metalloproteinase-2 ⁵⁷ and plasminogen activator inhibitor-1 in TM cells. ^21,35,57 Surprisingly, few proteases or protease inhibitors were among the 853 proteins quantified in this study, perhaps because they were below detection limits. Nevertheless, we found almost twice as many proteins increased rather than decreased in abundance, suggesting that TGFβ₂ signaling may inhibit proteolytic processes in the TM. ⁵⁷ Several proteins previously reported to modify or to be associated with TGFβ₂ signaling in the TM or to be overexpressed in POAG TM, including bone morphogenetic protein family members gremlin, Smad 7, and tissue transglutaminase and the lysyl oxidase family of enzymes, were not detected in this study. ^{28,35,58 –60} Again, these proteins may be below the detection limits of this study. In addition, no known glaucoma susceptibility gene products were detected. We did detect two proteins reportedly upregulated in TM by TGFβ₂, namely senescence-associated-β-galactosidase ²³ and fibronectin, ^21,35 but the abundance of these proteins was not significantly different between TGFβ₂-treated and untreated cells. Other TGFβ₂-regulated proteins certainly exist in the TM, and some may be among those (Supplementary Table S5) that narrowly miss the criteria used for a significant change in this study. 

The mechanism or mechanisms responsible for elevated expression of TGFβ₂ in glaucoma patients are unknown. TGFβ₂ in the aqueous humor could be made and secreted from one or more tissues surrounding the anterior segment, including the ciliary epithelium, iris, lens epithelium, corneal endothelium, and TM. Elevated levels of TGFβ₂ in glaucomatous TM tissues could be derived in part from the aqueous humor and bound to the TM extracellular matrix. ¹⁹ However, glaucomatous TM cells express a latent isoform of TGFβ₂ of higher molecular weight that must be activated to become biologically active. ¹⁹ How latent TGFβ₂ is activated in TM cells remains unclear but may involve thrombospondin-1, ⁴² which was found elevated in this study. 

In summary, the present results add strong support to evidence that TGFβ₂ can induce ECM remodeling and alter cytoskeletal interactions in the TM. In addition, the results implicate TGFβ₂-altered regulation in processes such as transcription, translation, steroid metabolism, and the glutamate/glutamine cycle. Finally, the data provide new evidence that TGF-β₂-induces mitochondrial dysfunction in the TM that could contribute to oxidative damage in the AH outflow pathway, TM senescence, and ocular hypertension in POAG. Overall, the results expand the repertoire of proteins known to participate in TGFβ₂ signaling and establish a quantitative proteomic database for the TM that includes candidate glaucoma biomarkers for future validation studies. 

Supplementary Materials

Text s1, PDF - Text s1, PDF 

Footnotes

 Supported in part by National Institutes of Health Grants EY018147, EY14239, and EY15638; a Research to Prevent Blindness Challenge grant; American Health Assistance Foundation Grant 34-0714585 (KEB); an RPB Senior Investigator Award (JWC); a Steinbach Award (JWC); and The Cleveland Clinic Foundation.

Footnotes

 Disclosure: K.E. Bollinger, None; J.S. Crabb, None; X. Yuan, None; T. Putliwala, None; A.F. Clark, None; J.W. Crabb, Alcon (C), Allergan (C)

The authors thank Phillip Howe, Joe Hollyfield, Edward Rockwood, Andrew Schachat, and Elias Traboulsi for valuable discussions.