The Effect of TGF-β2 on Elastin, Type VI Collagen, and Components of the Proteolytic Degradation System in Human Optic Nerve Astrocytes

Abstract

purpose. To investigate the effect of transforming growth factor (TGF)-β2 on the expression of (1) elastin and type VI collagen (ColVI), (2) extracellular matrix (ECM)–degrading matrix-metalloproteinases (MMPs) and regulators of their activity/activation, and (3) the involvement of connective tissue growth factor (CTGF) in the TGF-β2-mediated regulations in cultured human type-1A and -1B optic nerve astrocytes.

methods. Astrocytes were isolated from the optic nerves of 11 donors aged 19 to 62 years without a history of eye disease from the prelaminar (type 1B, five explants) or postlaminar (type 1A, six explants) region. Cultures of passages 3 to 5 were treated with 1 ng/mL recombinant human TGF-β2 for 72 hours, and regulatory effects on the expression of elastin and the ColVI chains α1, α2, and α3; MMP-1, -2, -3, -7, -9, -12, and -13; tissue inhibitors of MMPs (TIMPs) -1, -2, and -3; plasminogen activator inhibitor 1 (PAI-1); and urokinase and tissue plasminogen activators (uPA, tPA) were initially analyzed by RT-PCR and confirmed and quantified by real-time PCR (rtPCR). The regulation of proteins was studied by Western blot analysis, and MMP-2 activity was assessed by gelatin zymography. The involvement of CTGF was tested by knockdown experiments with CTGF-small interfering (si)RNA.

results. TGF-β2 increased the expression of elastin (5×[rtPCR]/6×[WB]), ColVIα2 (3×/5×), ColVIα3 (7×/9×), MMP-2 (2×/2×), TIMP-1/-3 (1.5×/2×), and PAI-1 (8×/4×) compared to untreated controls. tPA was reduced to 0.5×. MMP-1, -3, -7, and -12 and TIMP-2 were expressed but were not responsive to TGF-β2. MMP-9 and -13 and uPA were marginally expressed and close to the detection threshold. MMP-2 activity was significantly reduced in gelatin zymography. Transfection of CTGF-siRNA blocked TGF-β2–mediated activation of elastin and ColVI but had no effect on MMP-2 and PAI-1 induction. Type 1A and 1B astrocytes reacted identically.

conclusions. TGF-β2 induces expression of elastin and ColVI and thereby could contribute to the increase of type VI collagen fibers in the tissue septae and the elastotic changes typically observed in POAG. With the concurrent activation of TIMP-1 and -3 and PAI-1 and the repression of tPA, TGF-β2 could negatively regulate the activity and activation of MMPs. This effect could further amplify ECM accumulation and elastosis.

Immunohistochemical and ultrastructural investigations of eyes with primary open-angle glaucoma (POAG) over the past decades have identified distinct morphologic alterations defining the disease and allowing discrimination of POAG from other types of glaucoma (e.g., pseudoexfoliation glaucoma; PEXG). With respect to the optic nerve, all forms of glaucoma show a significant reduction and degradation of optic nerve axons. This effect leads to an initial reduction of the lateral visual field and progressive loss of sight, finally resulting in complete blindness. ^{1 2 3 4 5} In POAG there are at least two more prominent pathologic changes in the optic nerve. The prelaminar and laminar regions frequently show a massive increase in elastic fibers, with a disorganized distribution and deposition, embraced by the term elastosis. ⁶ In the postlaminar compartment, one typically finds a pronounced thickening of the connective tissue septae surrounding the optic nerve fibers. ^{7 8 9 10 11} In contrast, PEXG eyes with a comparable degree of axon loss do not show this connective tissue reaction. ¹¹ These phenotypical differences could be judged as indications for different pathomechanisms underlying POAG and PEXG development, respectively. This is further corroborated by the finding that in PEXG, axon loss correlates significantly with the intraocular pressure (IOP), whereas this correlation does not apply in POAG. ^{12 13} From that, it was hypothesized that in POAG, there are factors that circulate in the aqueous humor (AH) and are present in the vitreous that can induce glaucomatous changes in the optic nerve, either alone or synergistically with increased IOP. A major candidate fulfilling these demands is transforming growth factor (TGF)-β2, a factor that is significantly elevated in the AH of approximately 50% of POAG eyes, but not in PEXG. ^{14 15} In a recent study, we showed that TGF-β2 induces expression of different components of the ECM and ECM-modifying enzymes via connective tissue growth factor (CTGF) in cultured human optic nerve astrocytes. ¹⁶ In the present study, we addressed three distinct questions about the capacities of TGF-β2 with respect to the induction of POAG-associated changes. 

First, we analyzed the effect of TGF-β2 on the expression of elastin, the protein being deregulated in elastosis, and collagen type VI, which we recently identified as a component of the thickened connective tissue septae ¹¹ and an anchor of the elastic fibers in the adjacent BM. ^{17 18}  

The ECM is not a static formation, but rather the result of a fine-tuned balance of proteolytic degradation and new synthesis. To analyze whether the thickening of the connective tissue septae may be the result of a reduced proteolysis, we then investigated the effects of TGF-β2 on the expression of members of the proteolytic degradation machinery in astrocytes. 

According to the morphologic data, elastotic changes seem to be restricted to the prelaminar and laminar regions, whereas the connective tissue reaction seems to take place mainly in the laminar and postlaminar regions. The compartments of the optic nerve are populated by different types of astrocytes: type 1B astrocytes in the prelaminar–laminar region and type 1A astrocytes in the postlaminar region. ^{19 20 21 22} The third question we therefore addressed was whether both types of astrocytes differ in their response to TGF-β2 treatment. 

Materials and Methods

Human Optic Nerve Astrocyte Explant Cultures

Explant cultures of human optic nerve astrocytes were obtained from our collaborators in the Department of Ophthalmology, Maximilians University. Monolayer cultures were established from eyes of 11 human donors (19–62 years old, obtained 4 to 8 hours after death) without a history of eye disease. Explants were prepared either from the prelaminar (five explants) or the postlaminar (six explants) regions of the optic nerve. Preparation of eyes and propagation of astrocytes was performed as described elsewhere. ^{16 20 21 23} Cells were shipped from Munich to Erlangen at passage 2 and propagated for another 1 to 3 passages. Astrocytes were characterized by immunohistochemical staining against α-smooth muscle actin (α-smA), A2B5, S100, paired box gene (PAX)-2, glial fibrillary acidic protein (GFAP), and neural cell adhesion molecule (NCAM)-1 (Table 1) . According to the literature, this allowed discrimination of type-1A and -1B astrocytes ^{19 20 22 24 25 26} (Table 2) . Methods of securing human tissue were humane, included proper consent and approval, and complied with the Declaration of Helsinki. 

To test the effects of TGF-β2, astrocytes of passages 3 to 5 were plated in 35-mm Petri dishes and grown to confluence in DMEM/F-12 plus 10% FBS (Invitrogen, Karlsruhe, Germany) at standard conditions. Before treatment, the cells were starved for 24 hours in serum-free DMEM/F-12, and then the medium was changed to serum-free DMEM/F-12 containing 1.0 ng/mL active TGF-β2 ¹⁶ (Roche, Basel, Switzerland), a concentration that resembles that of the active TGF-β2 measured in AH samples of POAG eyes with increased TGF-β2. ^{14 15 27} Treatment duration was 72 hours, to allow analysis of eventual secondary effects of TGF-β2-responsive degradation system components, which were expected to require longer incubation to be detectable. Control cultures were treated analogously but in TGF-β2-free medium. 

RNA Isolation and Complementary DNA Synthesis

Total RNA was phenol-chloroform extracted from TGF-β2-treated and control astrocytes (TRIzol reagent; Invitrogen). Structural integrity of the RNA samples was confirmed by electrophoresis in 1% Tris-acetate-EDTA (TAE)-agarose gels. Yield and purity were determined photometrically. 

First-strand complementary DNA (cDNA) was prepared from 2.5 μg total RNA by reverse transcription (RT; Superscript II reverse transcriptase; Invitrogen) and oligo(dT)-17 primer according to standard protocols. The total assay volume was 20 μL, which was diluted to 200 μL after RT in low TE (10 mM Tris [pH 7.4], 1 mM EDTA). 

Semiquantitative PCR

Gene-specific PCRs were performed in a total volume of 25 μL containing 5 μL cDNA, 2.5 μL 10× PCR buffer (Mg²⁺ free), 0.5 μL 10 mM dNTP mix, 0.5 μL 10 μM primer (forward and reverse each), 0.75 μL 50 mM MgCl₂, 0.1 μL (5 U/μL) Taq polymerase (all from Invitrogen), and H₂O. The 25-μL PCR steps were 30 seconds of denaturation at 96°C, 30 seconds of annealing, and 45 seconds of extension at 72°C, followed by an end-extension step of 5 minutes at 72°C after the last cycle. Primer sequences, specific annealing temperatures, cycle numbers, and product sizes are given in Table 3 . The functionality of primers was tested on cDNAs obtained from different tissues before the experiments to exclude false-negative results (not shown). 

Real-Time PCR

Quantitative real-time PCR was performed (LightCycler System; Roche Diagnostics, Germany) according to the manufacturer’s specifications. Primers and probes were selected by computer (ProbeFinder program, ver. 2.04; Roche, Table 4 ). To normalize for differences in the amount of total RNA added to each reaction, we processed 18S rRNA simultaneously in the same sample as the internal control. The level of mRNA was determined as the relative ratio (RR), which was calculated by dividing the level of mRNA by the level of the 18S rRNA housekeeping gene in the same samples. 

Protein Extraction and Western Blot Analysis

TGF-β2-treated and control astrocytes were directly lysed in RIPA lysis buffer (150 mM NaCl, 1% NP-40, 0.5% DOC, 0.1% SDS, 50 mM Tris [pH 8]), the cell debris was pelleted by centrifugation for 30 minutes at 14,000 rpm at 4°C, and the protein contents were determined by the Bradford protein-detection assay (Bio-Rad, München, Germany). For analysis of secreted proteins, media of treated and control cells were collected and concentrated 100-fold by centrifugation through filter membranes (Vivaspin 20, 10,000 MWCO; Vivascience, Hannover, Germany), according to the manufacturer’s instructions. Probes were supplemented with SDS-loading buffer (4× RotiLoad; Roth, Karlsruhe, Germany) and denaturized at 60°C for 7 minutes. Samples (containing 5 μg protein for RIPA and 25 μL of concentrated (cc) medium, respectively) were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred onto a nitrocellulose membrane (Protran BA83, 0.2 μm; Schleicher & Schüll, Dassel, Germany) by the tank blot method at 70 V for 0.75 to 1.25 hours in 1× transfer buffer (10 mM CAPS [pH 11; 3-(cyclohexylamino)-1-propanesulfonic acid], 20% methanol, and 0.1% SDS). Membranes were blocked in TBST, 5% Blotto, and 1% primary serum (tris-buffered saline, 0.1% [vol/vol] Tween-20, 5% [wt/vol]) nonfat dry milk, 1% [vol/vol] primary serum of the host in which secondary antibodies were raised; pH 7.2) for 1 hour. After one rinse and once wash for 15 minutes in TBST, the primary antibodies were added in TBST, 1.5% Blotto, and 0.5% primary serum (for dilutions, see Table 1 ) and allowed to react for 1 hour at RT or overnight at 4°C. After the membranes were washed twice for 5 minutes with TBST, secondary antibodies were added in TBST/1.5% nonfat dry milk at the appropriate dilution (see Table 1 ) for 30 minutes at room temperature, followed by three to four washing steps in TBST, 5 minutes each. For detection, one volume ECL A (125 μg/mL luminol [Sigma-Aldrich, Munich, Germany], 0.5 M Tris-HCl, and 0.5× PBS [pH 7.2]), plus 1:15 volumes ECL B (1.1 mg/mL para-hydroxy-coumaric acid [Sigma-Aldrich] in dimethylsulfoxide [DMSO]) were activated with 6‰ (vol/vol) hydrogen peroxide (H₂O₂), added to membranes, and incubated for 5 minutes at room temperature. Chemiluminescence signals were visualized by exposure to light-sensitive films (Hyperfilm ECL; GE Healthcare, Little Chalfont, UK) for 1 to 10 minutes. Quantification was performed on computer (Lumi-Analyst software; Roche, Mannheim, Germany). 

Immunohistochemistry

Cultured astrocytes were grown on microscope chamber slides and treated with TGF-β2, as described earlier. Control cells were kept in TGF-β2-free medium. After treatment, the cells were washed three times with PBS, fixed in methanol for 4 minutes, and air dried. The slides were blocked in PBS with 5% dry nonfat milk for 30 minutes before the primary antibodies were added in appropriate dilutions (Table 1)in PBS and 1% BSA and allowed to bind overnight at 4°C. The slides were washed once in PBS, and fluorescein-conjugated secondary antibodies were added (Table 1)for 1 hour at room temperature. Excess antibodies were rinsed off three times with PBS, and the nuclei were stained with 4′,6-diamidino-2-phenylindol (DAPI) for 3 minutes. After three final washes with PBS, the cells were mounted with fluorescent mounting medium (Dako, Glostrup, Denmark). The slides were analyzed by fluorescence microscope (Aristoplan; Leitz, Wetzlar, Germany). Negative controls to estimate nonspecific binding of secondary antibodies were incubated in PBS/BSA but without primary antibody and showed no signals (not shown). 

Gelatin Zymography

For analysis of MMP-2 activity, medium of treated and control astrocytes was collected and concentrated as described. Probes were diluted 1:2 (vol/vol) in zymogram sample buffer (BioRad, Munich, Germany) and 50 μL were loaded on a zymogram gel containing 10% gelatin (Ready Gel; Bio-Rad, Munich, Germany). After electrophoresis, the gels were renatured and developed according to the manufacturer’s instructions. The gels were counterstained in 0.5% Coomassie brilliant blue R-250 solution, destained, and images were obtained (Lumi-Imager; Roche). Quantification was performed with the accompanying software (Lumi-Analyst software; Roche, Boehringer). 

Generation and Transfection of siCTGF

Human CTGF siRNA was generated with an siRNA construction kit (Silencer; Ambion, Austin, TX) as previously described. ¹⁶ For transfection, astrocytes were seeded as already described, transfected with 10 nM CTGF siRNA (Lipofectamine reagent; Invitrogen, Karlsruhe, Germany), according to the manufacturer’s instructions. On control astrocytes and astrocytes to be treated with TGF-β2, only the transfection procedure was performed equally but without 10 nM siCTGF. After 4 hours, the medium was changed to medium containing 1.0 ng/mL TGF-β2 or to normal culture medium in the controls, respectively. Treatment was continued for 72 hours before the cells were harvested for RNA isolation. The specificity of siCTGF’s effects was tested on three different astrocyte cultures, which were transfected with 10 nM of a nontargeted siRNA species (siGLO Lamin A/C Dharmafect; Dharmacon, Lafayette, CO) before TGF-β2 treatment. In these experiments, the TGF-β2-mediated activations of elastin and ColVIα3 were not affected (data not shown). By the use of this fluorescence-conjugated siRNA we determined a transfection efficiency of ≥75%, on average (n = 3, data not shown). 

Statistical Analysis

The statistical significance of TGF-β2-mediated regulation and siCTGF’s effects were computed by paired two-tailed Student’s t-tests. 

Results

Characterization of Human Astrocyte Cultures

Astrocyte cultures of pre- and postlaminar origin used for these studies were classified by immunofluorescence staining (Fig. 1) . All lines were negative for α-smA and A2B5, which excluded potential muscle cell and oligodendrocyte or type-2 astrocyte contaminations ^{20 24} (Fig. 1 , left). Cells showed positive signals for S100, a glia cell marker ^{25 26} (Fig. 1 , left). Expression of PAX-2 was present in the nucleus and weak, indicating a mature state of the astrocytes ¹⁹ (Fig. 1 , middle). Astrocytes were also all positive for GFAP, characterizing them as type 1 astrocytes (Fig. 1 , middle). Cells derived from the postlaminar region were NCAM-1 negative, whereas cells originating from the prelaminar compartment were NCAM-1 positive (Fig. 1 , right). From that, NCAM⁺ cells were termed type 1B, and NCAM⁻ cells, type-1A astrocytes ^{20 22} (Fig. 1 , Table 2 ). 

Elastin and Collagen Type VI

All of the astrocyte cell lines expressed elastin and the α1, -2, and -3 chains of collagen type VI (Fig. 2A) . Treatment with TGF-β2 activated the expression of elastin, ColVIα2 and α3 (Fig. 2A) , whereas ColVIα1 showed no changes (not shown). There was no difference between type-1A and -1B astrocytes. 

Quantitative analysis of the TGF-β2-mediated inductions by real-time PCR revealed an induction of 5.2 ± 0.7-fold for elastin, 3.6 ± 0.6-fold for ColVIα2, and 7.5 ± 1.2-fold for ColVIα3 (Fig. 2B , Table 5 ). Data are expressed as the relative ratio of normalized data of treated versus control samples ± SD (n = 11). These results were consistent in both astrocyte cell types. 

In Western blot analyses, the transcriptional activation induced by TGF-β2 resulted in a significant elevation of the signals for elastin (66 kDa) and the ColVI subchains (α1, 140 kDa; α2, 140 kDa; and α3, 200 kDa ²⁸ ; Fig. 2C ). Densitometric quantifications revealed inductions of 6.2 ± 0.5-fold for elastin, 5.1 ± 0.2-fold for ColVIα2, and 10.3 ± 1.8-fold for ColVIα3 (Table 5) . 

In the cell cultures, treatment with TGF-β2 led to a significant increase in elastin and ColVI immunoreactivity, respectively (Fig. 2D) . The elastin signal was mainly located in the cytoplasm of the cells, whereas staining for ColVI was pronounced in the cell membranes and in the fibrillar structures surrounding the cell processes. Also within the extracellular spaces, a more pronounced signal was detectable in the astrocyte cultures that were exposed to TGF-β2. 

MMPs and Regulators of MMP-Activity/Activation

From the selected set of MMPs, all tested astrocyte cell lines expressed MMP-1, -2, -3, -7, -12, and -13 (Fig. 3A) . MMP-9 signals were only visible when high amounts of cDNA were used and a high number of cycles were run, indicating marginal expression (Fig. 3A) . Expression of TIMP-1, -2, and -3 was detectable in all cell lines, and the same was true for tPA and PAI-1 (Fig. 3B) . uPA in contrast was only detectable at background level, similar to MMP-9. Qualitative TGF-β2-mediated upregulation was seen in MMP-2, TIMP-1 and -3, and PAI-1 and downregulation in tPA (Figs. 3A 3B) . Because of the weak expression of MMP-9 and uPA, assertions about putative regulatory effects are not possible. All other factors (MMP-1, -3, -7, -9, -12, and -13 and TIMP-2) were unaffected by TGF-β2. Signal intensities of GAPDH amplicons served as the control for equal cDNA amounts and were considered in judging the regulatory effects. Type-1A and -1B astrocytes showed no significant differences in expression profiles and response to TGF-β2. 

Quantification of the regulatory effects by real-time PCR revealed inductions of 1.8 ± 0.3-fold for MMP-2, 1.5 ± 0.1-fold for TIMP-1, 1.5 ± 0.2-fold for TIMP-3, 9.2 ± 1.6-fold for PAI-1, and a repression of 0.5 ± 0.2-fold for tPA (Fig. 3C , Table 5 ). Data are given as the relative ratio of normalized data of treated versus control samples ± SD (n = 11). 

The general expression pattern as well as TGF-β2-mediated regulations were also confirmed on the protein level (Fig. 3D) . For MMP-2, a clear band of 72 kDa, corresponding to the predicted mass of pro-MMP-2, was detected, which was significantly increased in treated astrocytes. Significant inductions were also detected for TIMP-1 (28 kDa), TIMP-3 (24 kDa), and PAI-1 (45 kDa). The demonstrated transcriptional repression of tPA was reflected as a reduced signal intensity of the corresponding 63-kDa band. Densitometric quantification revealed inductions of 2.0 ± 0.5-fold for MMP-2, 2.0 ± 0.5-fold for TIMP-1, 1.8 ± 0.3-fold for TIMP-3, 4.2 ± 0.5-fold for PAI-1, and a repression of 0.6 ± 0.2-fold for tPA (Table 5) . 

The upregulation of PAI-1 was also demonstrated by IHC-stainings in living cells. The PAI-1 signal was significantly increased throughout the cytoplasm of TGF-β2-treated astrocytes compared with that in untreated control cells (Fig. 3E) . Because of the mild activations or repressions, differences of the signal intensities of the other factors were not observed (not shown). 

Gelatin Zymography

Gelatin zymography was performed to assess directly MMP-2 activity in response to TGF-β2 treatment. For the 70-kDa band of pro-MMP-2 (nonreduced conformation), no regulatory effect was seen when this method was used (Fig. 4A) . The 66-kDa band of the active MMP-2 in contrast was significantly reduced in intensity (Fig. 4A) . Densitometric quantification of the signal intensities showed a 25% reduction of act-MMP-2 in the TGF-β2-treated cells, whereas pro-MMP-2 remained constant (Fig. 4B) . 

Effect of CTGF-siRNA

The effect of siCTGF on the TGF-β2-mediated regulations of elastin, ColVIα3, MMP-2, and PAI-1 was analyzed by RT-PCR in astrocytes that were transfected with 10 nM siCTGF before TGF-β2 treatment. Data were compared with those of untreated control cells and TGF-β2-treated astrocytes, which were mock transfected (without siCTGF). The TGF-β2-mediated activations were again detected in the mock-transfected, TGF-β2 exposed astrocytes. In the siCTGF-transfected astrocytes in contrast, the induction of elastin and ColVIa3 was almost completely blocked, resulting in an expression level comparable to expression in untreated control cells. MMP-2 and PAI-1 activation instead remained unaffected by siCTGF (Fig 5A) . These effects were confirmed by semiquantitative analysis of band intensities normalized to GAPDH (Fig. 5B) . 

Discussion

Our study clearly demonstrated the strong potency of TGF-β2 to induce expression of elastin and collagen type VI in cultured human optic nerve astrocytes. Both factors were induced by three- to fivefold on the transcriptional level and even more (six- to ninefold) on the protein level. It is well documented that elastosis, which is the result of increased elastin synthesis, and the thickening of the laminar beams and the connective tissue septae, in which elevated amounts of type VI collagen were detected, ^{11 29} are more pronounced in POAG than in PEXG with a similar degree of axon loss. ¹¹ Coincidentally, both forms of glaucoma also differ in the AH concentrations of TGF-β isoforms. In POAG, TGF-β2 levels are about twice as high as in control cells (≈750 pg/mL vs. ≈300 pg/mL) whereas TGF-β1 levels are comparable to those in control cells. In PEXG in contrast, TGF-β2 levels are not altered but TGF-β1 is elevated to approximately 30 pg/mL. ^{14 27} However, levels of TGF-β1 are still only approximately 10% of levels of TGF-β2 in PEXG. TGF-β1 can induce elastin expression, but the inductions have been significantly lower (twofold) than those we report for TGF-β2. ³⁰ It is necessary to mention that these results were obtained by microarray analysis in lamina cribrosa (LC) cells, and so a direct comparison of induction is not adequate, but could be judged indicative. Taking this into account, the differences in TGF-β2 and -β1 levels in both forms of glaucoma and the potentially different inductive abilities of both TGF-β isoforms could correlate with the lower extent of the connective tissue reaction in PEXG compared with POAG. This correlation would further support the hypothesis that TGF-β2 may be a factor that induces or at least propagates biochemical and morphologic changes typically, and in a manner specifically observed in POAG. 

In a previous study, we showed that induction of collagen type I and IV as well as fibronectin expression in astrocytes is mediated by CTGF, which itself is induced by TGF-β2. ¹⁶ In the present study, elastin and type VI collagen induction were also dependent on CTGF, promoting its role as a putative central factor in POAG pathogenesis. 

ECM accumulation can also be the result of disturbed homeostasis in the direction of reduced degradation. For that, we analyzed the expression of selected MMPs which correspond to the ECM components shown to be increased in POAG as well as regulators of MMP activity and activation. Our results show that astrocytes basally expressed MMP-1, -2, -3, -7, -12; TIMP-1 to -3; tPA; and PAI-1 but not MMP-9 and -13 or uPA. MMP-2, a known TGF-β2-responsive MMP, ³¹ and TIMP-1 and -3 were moderately activated by TGF-β2 whereas tPA was slightly downregulated. This result argues for a more or less intact degradation system that is not significantly affected by TGF-β2. However, the PAI-1 protein, a very potent inhibitor of the plasminogen activation pathway, ³² was more than fivefold activated by TGF-β2. By this, activation of proMMPs could be efficiently inhibited as active plasmin, the cleavage product of plasminogen is required for this process. ³² Our zymography results support this idea, as we found significantly reduced levels of active MMP-2 on TGF-β2 treatment. However, further studies on the plasminogen–plasmin dependent proMMP-activation pathway are necessary to test this hypothesis. But, without overestimating our data, the fact that the activation of elastin, type VI collagen and, as we previously described, type I and IV collagen was significantly stronger on protein level as on transcriptional level ¹⁶ supports this model of a reduced degradation induced by TGF-β2 via increase of PAI-1. 

Compared with the literature, our results with respect to basal MMP and TIMP expression disagree in some aspects with the findings of Agapova et al. ³³ They likewise detected expression of MMP-1 and -2 and TIMP-1 and -2 and absence of MMP-9, but contrarily to our findings did not detect MMP-3, -7, and -12. ³³ This finding may be explained by the fact that they detected the proteins directly by Western blot, gelatin/casein zymography, or immunohistochemical staining. We applied RT-PCR to detect the corresponding mRNA or cDNA and therefore could still detect weakly expressed mRNA species. In agreement with their findings, we also did not detect MMP-3, -7 and -12 in Western blot analyses and gelatin/casein zymography (not shown). 

The accumulation of ECM is not uniform within the optic nerve, but seems to be concentrated at different areas of the optic nerve compartments. These areas are the connective tissue septae in the postlaminar region, the lamina cribrosa itself and the lateral portions of the prelaminar region adjacent to the sclera. As the different compartments are populated by different types of astrocytes—namely, prelaminar type 1B and postlaminar type 1A ^{19 20 21 22} —we sought to determine whether both populations differ in the basal expression of the analyzed factors and in response to TGF-β2. Our data did not show differences between the two types of optic nerve astrocytes. The most plausible reason for that may be the lack of external, environment-specific factors in a cell culture system, that in vivo may be responsible for the differences in ECM production in the optic nerve regions. 

Induction of elastin and other ECM components in astrocytes and LC cells have also been described for increased hydrostatic pressure, ³⁴ mechanical stretch, ^{35 36} and oxidative stress. ²³ Kirwan et al. ^{35 36} also reported that mechanical stretch concurrently induced TGF-β1 ³⁵ and -β2 ³⁶ release in LC cells, indicating a potential correlation between ECM activation and TGF-β isoforms. It is possible that these factors support the TGF-β2 effects in glaucomatous eyes. In a recent publication, Yu et al. ²³ showed that astrocytes responded with increased TGF-β1 and -2 release after hypoxia/reoxygenation and that TGF-β2 mediates activation of αB-crystallin. However, the reason that TGF-β2 is so frequently increased, especially in POAG, is still not clear. 

In summary, the data we have presented highlight the strong potential of TGF-β2 to induce glaucoma-relevant changes in the extracellular matrix and the optic nerve by directly activating synthesis of type VI collagen and elastin. Moreover, we have provided evidence that TGF-β2 could mediate a repressive effect on extracellular matrix degradation by reducing activation and in consequence activity of MMPs via the strong activation of PAI-1. However, further studies are needed, to determine why TGF-β2 is increased, particularly in eyes with POAG. 

 Supported by Grant SFB 539 (Glaukome) from Deutsche Forschungsgemeinschaft, Bonn.

 Submitted for publication August 13, 2007; revised October 17 and December 29, 2007; accepted February 21, 2008.

 Disclosure: C. Neumann, None; A. Yu, None; U. Welge-Lüssen, None; E. Lütjen-Drecoll, None; M. Birke, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Antibodies Used for Western Blot Analyses and Immunohistochemistry in the Present Study

Antibody	Abbreviation	Application	Dilution	Supplier
Mouse monoclonal anti-human smooth muscle α-actin	mc m-α-hu-α-smA	IHC	1:150	Dako
Mouse monoclonal anti-human A2B5	mc m-α-hu A2B5	IHC	1:200	Chemicon
Rabbit polyclonal anti-human S100	pc rb-α-hu S100	IHC	1:20	Zymed
Rabbit polyclonal anti-human PAX-2	pc rb-α-hu PAX-2	IHC	1:100	Abcam
Rabbit polyclonal anti-human GFAP	pc rb-α-hu GFAP	IHC	1:200	Sigma-Aldrich
Mouse monoclonal anti-human NCAM-1	mc m-α-hu NCAM-1	IHC	1:25	Serotec
Rabbit polyclonal anti-human elastin	pc rb-α-hu elastin	WB/IHC	1:80/1:40	Chemicon
Rabbit polyclonal anti-human Collagen VI	pc rb-α-hu ColVI	WB/IHC	1:1,500/1:500	Abcam
Goat polyclonal anti-human MMP-2	pc g-α-hu MMP-2	WB	1:250	Santa Cruz
Rabbit polyclonal anti-human TIMP-1	pc rb-α-hu TIMP-1	WB	1:2,500	Chemicon
Rabbit polyclonal anti-human TIMP-3	pc rb-α-hu TIMP-3	WB	1:2,500	Chemicon
Mouse monoclonal anti-human tPA	mc m-α-hu tPA	WB	1:250	Acris
Rabbit polyclonal anti-human PAI-1	pc rb-α-hu PAI-1	WB/IHC	1:125/1:50	Santa Cruz
Mouse monoclonal anti-chicken α-actin	mc ms-α-ck α-actin	WB	1:1,000	Santa Cruz
HRP-conjugated goat anti-rabbit IgG	g-α-rb IgG-HRP	WB	1:10,000	Caltag
HRP-conjugated swine anti-goat IgG	s-α-g IgG-HRP	WB	1:10,000	Caltag
HRP-conjugated goat anti-mouse IgG	g-α-m IgG-HRP	WB	1:10,000	Caltag
Alexa Fluor 488-conjugated goat anti-rabbit IgG	g-α-rb IgG-Alexa488	IHC	1:2,000	Mobitec
Alexa Fluor 488-conjugated goat anti-mouse IgG	g-α-m IgG-Alexa488	IHC	1:500	Mobitec
Alexa Fluor 488-conjugated rabbit anti-goat IgG	rb-α-g IgG-Alexa488	IHC	1:2,000	Mobitec

 Dako, Glostrup, Denmark; Chemicon, Hofheim, Germany; Zymed, Berlin, Germany; Abcam, Hiddenhausen, Germany; Sigma-Aldrich, Munich, Germany; Serotec, Dusseldorf, Germany; Santa Cruz Biotechnology, Heidelberg, Germany; Acris, Herford; Invitrogen-Caltag, Karlsruhe; MoBiTec, Göttingen, Germany.

Table 2.

 

Characterization of Human Astrocytes

Cell Line	Preparation	smA	A2B5	S100	Pax-2	GFAP	NCAM	Type
1	Pre	—	—	+	(+)	+	+	1B
2	Pre	—	—	+	(+)	+	+	1B
3	Pre	—	—	+	(+)	+	+	1B
4	Pre	—	—	+	—	+	+	1B
5	Pre	—	—	+	—	+	+	1B
6	Post	—	—	+	(+)	+	—	1A
7	Post	—	—	+	(+)	+	—	1A
8	Post	—	—	+	(+)	+	—	1A
9	Post	—	—	+	—	+	—	1A
10	Post	—	—	+	(+)	+	—	1A
11	Post	—	—	+	—	+	—	1A

Table 3.

 

Primers Used for Reverse Transcription–PCR Amplification in the Present Study

Gene	Sequence	Position	Annealing Temp. (°C)	Cycles (sqPCR)	Fragment Size (bp)
Elastin	Fwd.: 5′-gctttggcccgggagtagtt-3′	983–1602	60	30	619
	Rev.: 5′-caccttggcagcggattttg-3′
ColVIa2	Fwd.: 5′-gttctacctggaccaggtg-3′	325–1067	54	32	742
	Rev.: 5′-gtccatcggtcccgttc-3′
ColVla3	Fwd.: 5′-tttcgactcctccctggtgttc-3′	2043–2581	58	30	538
	Rev.: 5′-cacaaaaagtcaggatgcccg-3′
MMP-1	Fwd.: 5′-gccagatttgccaagagcaga-3′	437–1077	55	29	620
	Rev.: 5′-cggcaaattcgtaagcagcttc-3′
MMP-2	Fwd.: 5′-aaccctcagagccaccccta-3′	2520–2805	55	27	286
	Rev.: 5′-gtgcatacaaagcaaactgc-3′
MMP-3	Fwd.: 5′-agctctgaaagtctgggaagaggtg-3′	473–972	55	34	499
	Rev.: 5′-tcagagtgctgacaggatcaaagg-3′
MMP-7	Fwd.: 5′-gtttagaagccaaactcaagg-3′	208–440	ND	ND	232
	Rev.: 5′-ctttgacactaatcgatccac-3′
MMP-9	Fwd.: 5′-tgggctacgtgacctatgac-3′	2100–2290	60	33	190
	Rev.: 5′-caaaggtgagaagagagggc-3′
MMP-12	Fwd.: 5′-gtggaatcctagcccatgctt-3′	578–1285	62	30	707
	Rev.: 5′-aaccagggtccatcatctgtc-3′
MMP-13	Fwd.: 5′-tgccattaccagtctccgagga-3′	901–1415	ND	ND	514
	Rev.: 5′-ggcatgacgcgaacaatacggt-3′
TIMP-1	Fwd.: 5′-aattccgacctcgtcatcag-3′	301–676	55.5	25	375
	Rev.: 5′-gtttgcaggggatggataaa-3′
TIMP-2	Fwd.: 5′-ctggacgttggaggaaagaa-3′	606–950	55.5	28	344
	Rev.: 5′-gtcgagaaactcctgcttgg-3′
TIMP-3	Fwd.: 5′-acatttaaagaaaggtctat-3′	354–1021	55.5	28	667
	Rev.: 5′-ccaggacgccttctgcaact-3′
uPA	Fwd.: 5′-aacgtaccatgcccacagat-3′	246–842	50	40	596
	Rev.: 5′-ggcaggcagatggtctgtat-3′
tPA	Fwd.: 5′-cccagatcgagactcaaagc-3′	641–1224	52	30	583
	Rev.: 5′-atgttctgcccaagatcacc-3′
PAI-1	Fwd.: 5′-aggaccgcaacgtggttttctc-3′	104–609	59	34	505
	Rev.: 5′-agtgctgccgtctgatttgtg-3′
GAPDH	Fwd.: 5′-gaaggtgaaggtcggagtc-3′	108–333	57	23	225
	Rev.: 5′-gaagatggtgatgggatttc-3′

Table 4.

 

Primers Used for Real-Time PCR

Gene	Sequence	Position	Annealing Temp. (°C)	Fragment Size (bp)
Elastin	Fwd.: 5′-caaggctgccaagtacgg-3′	1341–1427	59	87
	Rev.: 5′-ccaggaactaacccaaactgg-3′
ColVIa2	Fwd.: 5′-agaccttccctgccaaaca-3′	841–952	59	112
	Rev.: 5′-cttgtggaagttctgctcacc-3′
ColVIa3	Fwd.: 5′-cattcatccgtgagtccaga-3′	396–476	59	81
	Rev.: 5′-aataatgtcagcagagtcttgtgc-3′
MMP-2	Fwd.: 5′-ataacctggatgccgtcgt-3′	2148–2210	52	63
	Rev.: 5′-aggcacccttgaagaagtagc-3′
TIMP-1	Fwd.: 5′-gggcttcaccaagacctaca-3′	603–673	60	71
	Rev.: 5′-tgcaggggatggataaacag-3′
TIMP-2	Fwd.: 5′-gaagagcctgaaccacaggt-3′	725–809	59	85
	Rev.: 5′-cggggaggagatgtagcac-3′
TIMP-3	Fwd.: 5′-ccttaagctggaggtcaacaa-3′	1462–1532	59	71
	Rev.: 5′-ccgtgtacatcttgccatca-3′
tPA	Fwd.: 5′-cgggtggaatattgctggt-3′	401–490	53	90
	Rev.: 5′-cccgttgaaacaccttgg-3′
PAI-1	Fwd.: 5′-ctcctggttctgcccaagt-3′	958–1023	60	66
	Rev.: 5′-caggttctctaggggcttcc-3′

Figure 1.

 

Characterization of human astrocyte cultures. Astrocyte cultures derived from the pre- and postlaminar region were characterized by immunofluorescence staining against smA, to exclude muscle cell contamination; A2B5 to exclude oligodendrocytes; S-100, to confirm astrocytic lineage; Pax-2, to confirm the mature differentiation state; GFAP, to confirm type 1 astrocytic lineage; and NCAM-1, to differentiate between type 1A and type 1B astrocytes. Accordingly, postlaminar type 1A astrocytes were characterized as smA⁻, A2B5⁻, S-100⁺, Pax-2^low, GFAP⁺, and NCAM⁻ and prelaminar type 1B astrocytes by smA⁻, A2B5⁻, S-100⁺, Pax-2^low, GFAP⁺, NCAM⁺.

Figure 2.

 

Effect of TGF-β2 on elastin and type VI collagen. (A) Reverse transcription–PCR analysis of elastin and Col6α2 and -α3 expression in untreated (co) and TGF-β2-treated astrocytes. (B) Real-time PCR analysis of TGF-β2-mediated activations. Data are expressed as the mean ± SD of results in 11 tested astrocyte lines. Dotted line: normalized expression of untreated astrocytes (**P ≤ 0.01). (C) Representative Western blot on concentrated. medium of TGF-β2-treated astrocytes compared with controls (co), to confirm the transcriptional regulation of the protein level. Coomassie staining of gels demonstrates equal loading and protein contents of media probes. (D) Immunofluorescence staining for elastin and ColVI on TGF-β2-treated astrocytes compared with untreated control cells. The nuclei were counterstained with DAPI. Scale bar, 100 μm.

Table 5.

 

Summarized Quantifications

	Elastin	ColVIa2	ColVIa3	MMP-2	TIMP-1	TIMP-3	tPA	PAI-1
rtPCR	5.2 ± 0.7, **	3.6 ± 0.6, **	7.5 ± 1.2, **	1.8 ± 0.3*	1.5 ± 0.1*	1.5 ± 0.2, **	0.5 ± 0.2, **	9.2 ± 1.6, **
WB	6.2 ± 0.5, **	5.1 ± 0.2, **	10.3 ± 1.8, ***	2.0 ± 0.5*	2.0 ± 0.4*	1.8 ± 0.3, **	0.6 ± 0.2, **	4.2 ± 0.5, **

 Data are expressed as the mean ± SD (n = 11).

*   P ≤ 0.05,

**   P ≤ 0.01,

***   P ≤ 0.001.

Figure 3.

 

Effect of TGF-β2 on MMPs and regulators of MMP-activity/activation. (A) Reverse transcription–PCR analysis of MMP-1, -2, -3, -7, -9, -12, and -13 expression in untreated (co) and TGF-β2-treated astrocytes. (B) Reverse transcription–PCR analysis of TIMP-1, -2, and -3; uPA and tPA; and PAI-1 expression in untreated (co) and TGF-β2-treated astrocytes. (C) Real-time PCR analysis of observed TGF-β2-mediated regulations. Data are expressed as the mean ± SD of results in 11 tested astrocyte lines. Dotted line: normalized expression of untreated astrocytes (*P ≤ 0.05, **P ≤ 0.01). (D) Representative Western blot on protein extracts of TGF-β2-treated astrocytes compared with the controls (co), to confirm the transcriptional regulations on the protein level. Actin detection demonstrates equal loading and protein contents of extracts. (E) Immunofluorescence staining for PAI-1 on TGF-β2-treated astrocytes compared with untreated control cells. Nuclei were counterstained with DAPI. Scale bar, 100 μm.

Figure 4.

 

Gelatin zymography of MMP-2 activity. Twenty-five microliters of 100-fold concentrated medium of either control astrocytes or TGF-β2-treated astrocytes was loaded on a precast zymogram gel. After incubation and staining and destaining a 70-kDa band corresponding to the inactive pro-MMP-2 protein and a 66-kDa band corresponding to the active MMP-2 protein were visible (A). Densitometric measurements revealed a significant reduction of the active form in treated cells, whereas the amount of the pro form remained constant (B). Experiments were run in triplicate on each astrocyte line. Data are expressed as the mean ± SD (n = 11) and normalized to proMMP-2 expression in control cells (*P ≤ 0.05).

Figure 5.

 

Effect of CTGF siRNA on regulation of TGF-β2-mediated expression. Expression of MMP-2, PAI-1, elastin, and ColVIα3 was analyzed by PCR in control astrocytes, TGF-β2-treated astrocytes (both mock-transfected, i.e., transfected but without siCTGF), and astrocytes that were transfected with a CTGF-targeting siRNA before TGF-β2 treatment. siCTGF had no effect on TGF-β2-mediated MMP-2 and PAI-1 activation; elastin and ColVIα3 induction, in contrast, was significantly blocked (A). Quantification confirmed this observation (B). Experiments were performed on all astrocyte lines in triplicate (n = 11) (*P ≤ 0.05, **P ≤ 0.01).

The authors thank Julia Mausolf, Heide Wiederschein, and Katja Obholzer for expert technical assistance.