α2-Macroglobulin Induces Glial Fibrillary Acidic Protein Expression Mediated by Low-Density Lipoprotein Receptor-Related Protein 1 in Müller Cells

Abstract

Purpose.: Although it is known that Müller cells express the glial fibrillary acidic protein (GFAP) in response to acute retinal damage, the regulatory mechanism is not completely understood. α₂-Macroglobulin (α₂M) and its receptor, low-density lipoprotein receptor-related protein 1 (LRP1), have also been found in injured retinas. Herein, the authors examined the involvement of the α₂M/LRP1 system in GFAP expression in Müller cells using in vitro and in vivo experimental models.

Methods.: Using Western blot analysis and immunocytochemistry, the authors evaluated the effect of α₂M* on GFAP expression in the Müller cell line MIO-M1, which constitutively expresses LRP1. Intracellular signaling pathways activated by α₂M* were examined by Western blot analysis. The effect of α₂M* on GFAP expression in the mouse retina was examined by intravitreal microinjection of α₂M* in mouse eyes.

Results.: These data demonstrate that α₂M* induced GFAP expression in the MIO-M1 cell line, which was selectively blocked by RAP, an antagonist of LRP1 binding ligands. In addition, α₂M* induced JAK/STAT pathway activation, determined by STAT3 phosphorylation (p-STAT3), which was also blocked by RAP. Finally, the authors showed that GFAP was expressed in the retinas of mice, preferentially in Müller cells at 3 and 6 days after a single intravitreal α₂M* injection, whereas p-STAT3 staining increased at day 1 in both the ganglion cell layer and the inner nuclear layer.

Conclusions.: These results demonstrate that α₂M* induces GFAP expression in retinal Müller cells through LRP1, which could be mediated by JAK/STAT pathway activation.

Intermediate filaments (IFs) are a heterogeneous group of proteins that form 10-nm–diameter filaments, a highly stable cytoskeletal component occurring in various cell types. The upregulation of one of these IF proteins, glial fibrillary acidic protein (GFAP), has historically been an indicator of “stress” in astrocytes of the central nervous system. ¹ The retina also responds similarly to stress, but the upregulation of IF occurs primarily in the Müller cells, the radial glia of the retina. These cells in the mammalian retina normally contain only a low level of GFAP or none at all. ^2,3 The GFAP level is, however, strongly upregulated in Müller cells in all forms of retinal injury or disease, including retinal trauma, choroidal neovascularization, ⁴ retinal detachment, ^5,6 glaucoma, ^7,8 and diabetic retinopathy. ⁹ Nevertheless, the identity of the regulatory mechanisms responsible for the induction of GFAP in Müller cells is still not completely understood. In addition, considering that Müller cells contain substantial amounts of another IF protein, vimentin, it has been difficult to elucidate the biological function of GFAP in the mammalian retina under determined stress conditions. 

Native α₂-macroglobulin (α₂M) is a 720-kDa homotetrameric glycoprotein that forms complexes with almost all endogenous human proteinases. ¹⁰ The complex formation with proteinases causes a conformational change in the α₂M protein, resulting in an activated form of α₂M, termed α₂M*. ¹¹ This form recognizes the specific cell surface receptor low-density lipoprotein (LDL) receptor-related protein 1 (LRP1), a member of the LDL receptor gene family. However, α₂M* is only recognized by LRP1 and not by other LDL receptor members. ¹² Furthermore, it has been demonstrated that α₂M* is internalized by LRP1; therefore, both molecules are implicated in the modulation of the extracellular activity of several serine- and metalloproteinases. ¹³ It has also been reported that α₂M* induces cellular signal transduction activation mediated by LRP1 in several types of cells, including neurons ^14,15 and macrophages. ¹⁶  

In a previous report on retinas of rats with ischemia-induced neovascularization, we demonstrated an increased expression of LRP1 and α₂M, together with an enhanced activity of metalloproteinases. In the same study, we also showed that LRP1 was expressed in Müller cells. ¹⁷ In addition, in a clinical study, we noted a significant increase in both α₂M and metalloproteinases in the vitreous of human subjects with proliferative diabetic retinopathy. ¹⁸ In both oxygen-induced neovascularization and in retinal ischemia models developed on different animal species, an increased expression of GFAP in Müller cells has been widely reported. ^{19 –22} Recently, it has also been demonstrated that in addition to GFAP, α₂M is upregulated in glaucoma and is a strong mediator of retinal ganglion cell damage because of its ability to bind and clear several neurotrophic growth factors and cytokines. ⁸ On the other hand, in normal retinas, it has previously been demonstrated that α₂M mRNA and protein are present, but obviously without a nocive effect. ²³ We hypothesize that both α₂M* and LRP1 are involved in the activation of Müller cells, which could be related to enhanced GFAP expression. Hence, in the present work, we examined the involvement of the α₂M/LRP1 system in GFAP induction in Müller cells using in vitro and in vivo experimental models. GFAP expression was evaluated in an LRP1-positive human Müller cell line cultured in the presence of α₂M* and in the retinas of mice receiving intravitreal injections of α₂M*. We conclude that α₂M induces GFAP expression in retinal Müller cells through LRP1, which may be mediated by activation of the JAK/STAT intracellular signaling pathway. 

Materials and Methods

Cell Culture and Reagents

A well-characterized Müller cell line, MIO-M1, was used in our studies. ²⁴ Cells were grown in Gibco Dulbecco's modified Eagle's medium (containing 4500 mg/L glucose, sodium pyruvate and stabilized l-glutamine [GlutaMAX; Invitrogen, Paisley, UK]) plus 10% vol/vol fetal bovine serum and penicillin/streptomycin (Invitrogen), at 37°C with 5% CO₂. Experiments designed to investigate the LRP1 expression in mouse retinas were performed on Müller glial cell cultures obtained using a method previously described. ²⁵ Mouse J774A.1 macrophage-derived cells ^16,26 and the rat JHU-4 prostate (MAT-Lu) cell line (this article) were used as positive controls of LRP1 expression, whereas Chinese hamster ovary (CHOK-1) 13–5-1 was used as the LRP1-deficient cell line. ²⁷ α₂M was purified from human plasma following a procedure previously reported, ²⁸ and the activated form of α₂M (α₂M*) was generated by incubating α₂M with 200 mM methylamine-HCl for 6 hours at pH 8.2, as previously described. ²⁹ An expression construct, encoding RAP as a glutathione S-transferase (GST) fusion protein (GST-RAP), was provided by Guojum Bu (Washington University, St. Louis, MO). GST-RAP was expressed and purified as reported previously ³⁰ and was used without further modification. Immunoblots were performed with the following primary antibodies: polyclonal rabbit anti-GFAP (Dako, Carpinteria, CA), mouse monoclonal anti–phosphorylated ERK1/2 (anti–p-ERK1/2), polyclonal rabbit anti–total ERK1/2, monoclonal anti–phosphorylated STAT3 (anti–p-STAT3), anti–calreticulin (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and mouse monoclonal antibody anti–β-subunit LRP1 (clone 5A6), which was kindly provided by Dudley Strickland (University of Maryland School of Medicine, Baltimore, MD). Immunocytochemical studies were performed with the same GFAP and LRP1 antibodies, and additional studies were carried out using a monoclonal mouse anti–vimentin (Dako) and a polyclonal rabbit anti–cellular retinaldehyde-binding protein (CRALBP) kindly donated by John C. Saari (University of Washington, Seattle, WA). 

Western Blot Analysis

To evaluate the effect of α₂M* on GFAP protein expression and to investigate a possible involvement of the JAK/STAT pathway, MIO-M1 cells were plated into six wells and grown to 70% to 80% confluence. After the medium was aspirated, fresh serum-free medium was added and cultured for 18 hours. Then MIO-M1 cells were cultured with α₂M* (60 nM) for different times and lysed using 10 mM phosphate buffer saline (PBS), 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 0.2% sodium azide (NaN3), and 0.1% Nonidet NP-40 (octyl phenoxylpolyethoxylethanol). Fifty micrograms of lysate was separated on 10% SDS-polyacrylamide gels and transferred to a nitrocellulose membrane (Amersham Bioscience, Copenhagen, Denmark). Nonspecific binding was blocked with 5% nonfat dry milk in a Tris-HCl saline buffer containing 0.01% Tween 20 (TBS-T) for 60 minutes at room temperature. The membranes were incubated overnight at 4° to 8°C with polyclonal rabbit anti-GFAP or monoclonal antibody anti–β-subunit LRP1 (both diluted 1:100) or were probed with phosphospecific antibodies against a phosphorylated state of ERK1/2 or STAT3. Treatments were normalized in parallel by assessment of the protein loading using phosphorylation-independent antibodies against total ERK1/2 or calreticulin. The membranes were then incubated with a secondary horseradish peroxidase-conjugated antibody (Amersham Bioscience) for 1 hour at room temperature. The specific bands were revealed by chemiluminescence reaction (Pierce, Northumberland, UK) and were quantified by densitometric analysis using an image analyzer (UVP Life Science, Upland, CA). To inhibit α₂M* effects mediated by LRP1 in MIO-M1 cell cultures, GST-RAP (400 nM) was added in serum-free medium 30 minutes before the addition of ligands. 

Immunofluorescence Microscopy

To evaluate the protein expression of GFAP, vimentin, and LRP1 by immunofluorescence microscopy, MIO-M1 cells were grown in coverslips in 24-well plates and grown to 30% to 70% confluence. After stimulus with α₂M* for different times, coverslips containing the cells were washed twice with PBS, fixed in 4% paraformaldehyde for 30 minutes at 4°C, washed with PBS, permeabilized with 0.1% Triton X-100/200 mM glycine in PBS (10 minutes at 4°C), and incubated in PBS plus 2% BSA for 1 hour at 37°C to block nonspecific binding sites. The coverslips were then incubated overnight at 4°C with primary antibodies, washed five times with PBS plus 1% BSA, and exposed to secondary antibodies for 90 minutes at 37°C. The primary antibodies against LRP1, CRALBP, GFA, and vimentin were diluted 1:50. Secondary antibodies raised in donkey against IgG of rabbit and mouse (Alexa Fluor 488 or 594 nm, 1:1500 dilution; Molecular Probes, Eugene, OR) were applied for 1 hour at room temperature. After a thorough rinse with PBS plus 1% BSA, cell nuclei were stained with Hoechst 33258 (1:1000 dilution; Molecular Probes), and coverslips were mounted (FluorSave; Calbiochem, La Jolla, CA) and visualized using an epifluorescence microscope (TE2000-U; Nikon, Tokyo, Japan). In the same manner as the Western blot assay, to study the inhibitory effect of GST-RAP, 400 nM of this protein was added for 30 minutes in serum-free medium before the addition of α₂M*. 

Animals

Experiments were carried out with C57BL/6 mice of about 6 weeks of age. All animals were maintained and handled according to approved protocols, and experimental procedures were designed to conform to the Research Ethics Committee of the National University of Córdoba and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Animals were maintained in clear plastic cages with standard light cycles (12 hours light/12 hours dark). Before intravitreal injection, animals were anesthetized with 0.5 mL/kg of a rodent anesthesia cocktail containing 100 mg/mL ketamine HCl, 20 mg/mL xylazine HCl, and 10 mg/mL acepromazine administered intraperitoneally and fixed under a microsurgical microscope. After topical instillation of proparacaine, pure solutions of α₂M* (2.15 μg/μL) were prepared in PBS pH 7.4. Intravitreal injections were performed using a 5-μL Hamilton syringe with a 32-gauge needle, according to a previously described method. ³¹ The eye was punctured at the upper nasal limbus, and a volume of 0.5 μL reagent solution (α₂M*) or control vehicle, respectively, was injected into one eye of each mouse. Given that reflux of a certain amount of intraocular fluid is unavoidable when the pipette is removed from the injection site, the pipette was kept in place for 10 seconds to allow diffusion of the solution. After survival times of 1, 3, and 6 days, animals were killed by CO₂ overdose, and their eyes were enucleated and processed for immunocytochemistry. At least three mice were used for each of the survival times examined. 

Immunohistochemistry

Immunostaining was performed on frozen sections using well-established procedures. ³² After enucleation, eyes were placed in fresh 4% paraformaldehyde in PBS and left in a refrigerator for 6 hours. Then the fixative was replaced with 15% sucrose and left overnight. The eyes were embedded in OCT compound, and 10-μm sections were cut on a cryostat and placed on gelatin-subbed slides. These sections were treated with 1% BSA in PBS for 1 hour. Mouse monoclonal anti–β-subunit LRP1 together with vimentin, rabbit polyclonal anti-GFAP, or anti–p-STAT3 at 1:50 dilution were used. The primary antibodies were incubated at 4°C overnight. After the sections were washed with PBS, donkey secondary antibody anti– rabbit IgG or rabbit secondary antibody anti–mouse IgG (conjugated to Alexa Fluor 488 nm, 1:1500 dilution) was applied for 1 hour at room temperature. After a thorough rinse, the sections were coverslipped and the labeling was visualized by means of a epifluorescence microscope (TE2000-U; Nikon). 

Statistical Analysis

Comparison of quantitative data was performed using the nonparametric Mann-Whitney U test. 

Results

α₂M* Induces GFAP Expression in MIO-M1 Cells

To evaluate the effect of α₂M* on GFAP expression, we used an immortalized human Müller cell line, MIO-M1, which constitutively expresses LRP1 (Barcelona PF et al. IOVS 2010;51: ARVO E-Abstract 1900; this study). Cell cultures were treated with α₂M* (60 nM) for different times, and cell extracts were subjected to immunoblotting using an anti–GFAP antibody. Results presented in Figure 1 show that whereas a small amount of GFAP expression was seen in MIO-M1 cell controls, α₂M* induced protein expression of GFAP after 1 hour of incubation. Quantitative analysis showed that the α₂M*-induced GFAP expression reached a maximum of approximately fivefold at 2 hours of incubation, maintaining this high expression at 4 hours of α₂M* stimulation with respect to control. In addition, in other experiments, we observed that GFAP expression was maintained in cells for 24 hours of α₂M* treatment (data not shown). When the GFAP expression was analyzed by immunofluorescence microscopy, we observed that α₂M* induced in MIO-M1 cells a typical distribution of GFAP at the different times evaluated (Figs. 2aB–aD). In agreement with Western blot analysis, weak expression of GFAP was shown in MIO-M1 cells cultured in the absence of α₂M* (Fig. 2aA), indicating that GFAP was poorly expressed in this type of cell. By counting GFAP-positive cells, we observed that whereas approximately 10% of MIO-M1 cells expressed this molecule at 4 hours of α₂M* stimulus (data not shown), LRP1 was expressed in all MIO-M1 cells (Fig. 2a). In addition, CRALBP, a well-documented marker of Müller cells, ^24,33 was also expressed by all cells (Fig. 2b). 

Figure 1.

 

α₂M* induces GFAP protein expression in MIO-M1 cells. Total proteins were extracted from MIO-M1 cells cultured in the presence of 60 nM α₂M* for 1, 2, or 4 hours in serum-free medium and were analyzed by Western blot analysis. GFAP was detected with a polyclonal rabbit anti-GFAP. The protein loading control of calreticulin is also shown. The bars show the relative intensity of GFAP/calreticulin with respect to control 1 (dotted line), representing the mean ± SE from triplicate experiments. *P < 0.01, significantly different from controls.

Figure 2.

 

Expression of GFAP in MIO-M1 cells that constitutively express LRP1 and CRALBP. For immunofluorescence microscopy, cells were grown in coverslips to 30% to 70% confluence. (a) The cells were incubated together with rabbit anti-GFAP (green) and mouse anti–β subunit LRP1 (clone 5A6) (red) antibodies and were revealed by chemiluminescence reaction. Cell nuclei counterstained with Hoechst 33258 are also shown (blue). MIO-M1 cells were, respectively, untreated (aA) or treated (aB–aD) with α₂M* for 1, 2, or 4 hours. Arrows: typical and unmodified cell distribution of LRP1 in MIO-M1 cells cultured in the absence or presence of α₂M*. Scale bar, 10 μm. (b) MIO-M1 cells showing the expression of CRALBP. Scale bar, 25 μm.

LRP1 Mediates α₂M*-Induced GFAP Expression

Considering that α₂M* binds LRP1 but not the other members of the LDL receptor family and that its binding can be blocked by RAP, ^13,34 we examined by Western blot analysis whether the effect of α₂M* on the GFAP expression was mediated by LRP1 in MIO-M1 cells cultured in the presence of GST-RAP (400 nM). Figure 3 shows that pretreatment with GST-RAP significantly reduced the α₂M*-induced GFAP expression. In addition, GST-RAP did not induce GFAP expression when it was added alone to MIO-M1 cell cultures. Thus, we conclude that α₂M* induces GFAP expression in Müller cells through LRP1. 

Figure 3.

 

RAP inhibits α₂M*-induced GFAP expression in MIO-M1 cells. Total proteins were extracted from Müller cells treated with 60 nM α₂M* for 2 or 4 hours and were analyzed by Western blot analysis. Before α₂M* stimulation, cell cultures were incubated with 400 nM GST-RAP for 30 minutes in serum-free medium. GFAP expression was detected with primary antibody. The protein loading control of calreticulin is also shown. The bars show the relative intensity of GFAP/calreticulin respect to control 1 (dotted line), representing the mean ± SE of triplicate experiments. *P < 0.01 in bars 4 and 6 denotes the statistical significance with respect to bars 3 and 5, respectively.

Analysis of GFAP and Vimentin Expression in MIO-M1 Cells Stimulated with α₂M*

As shown, GFAP expression was increased in MIO-M1 cells after stimulation with α₂M*; therefore, we decided to investigate by immunofluorescence microscopy whether the protein expression of vimentin was also modified by the α₂M* stimulus. Simultaneous immunofluorescence labeling of GFAP and vimentin was carried out in MIO-M1 cells stimulated with α₂M* (60 nM) for different times (1, 2, and 4 hours). Figure 4a shows that whereas α₂M* induced GFAP expression, it was unable to modify the expression pattern of vimentin in MIO-M1 cells. Figure 4b demonstrates that vimentin filaments in MIO-M1 cells, in both α₂M*-untreated and treated cells, presented a typical unmodified staining pattern. In addition, pretreatment of MIO-M1 cells with GST-RAP (400 nM) did not modify vimentin expression, whereas α₂M*-induced GFAP expression was blocked (Fig. 5). Thus, taken together, these data demonstrate that α₂M*/LRP1 interaction has a selective effect on GFAP expression without affecting vimentin in MIO-M1 cells, which suggests that α₂M* may differentially regulate the expression of intermediate filament proteins in MIO-M1 cells. 

Figure 4.

 

GFAP and vimentin expression in MIO-M1 cells after α₂M* treatment. For immunofluorescence microscopy, cells were incubated in the presence of 60 nM α₂M* for 1, 2, or 4 hours. After this stimulus, cells were incubated together with rabbit anti-GFAP and mouse antivimentin antibodies and were revealed by chemiluminescence reaction. (a) Upper, middle, and lower panels represent the simultaneous immunodetection for GFAP (green), vimentin (red), and merged image, respectively, in MIO-M1 cells untreated (aA, aE, aI) or treated (aB–aD, aF–aH, aJ–aL) with α₂M* for the times, as indicated. In lower panels, the cell nuclei counterstained with Hoechst 33258 are also shown (blue). Scale bar, 25 μm. (b) Magnified image of MIO-M1cells untreated (bA, bC, bE) or treated (bB, bD, bF) with α₂M* for 4 hours, representing the simultaneous immunodetection of GFAP (top, green), vimentin (middle, red), and merge image (bottom), respectively. Scale bar, 10 μm.

Figure 5.

 

Analysis of RAP effect on α₂M*-induced GFAP and vimentin expression in MIO-M1 cells. Simultaneous immunodetection for GFAP (green) and vimentin (red) in non-pretreated (A, C, E) or pretreated (B, D, F) MIO-M1 cells with 400 nM GST-RAP for 30 minutes Cells were stimulated with 60 nM α₂M* for 2 hours (C, D) or 4 hours (E, F) with respect to control (A, B) in serum-free medium, respectively. After this stimulus, cells were incubated together with rabbit anti-GFAP and mouse antivimentin antibodies. Cell nuclei counterstained with Hoechst 33258 are also shown (blue). Scale bar, 25 μm.

α₂M* Leads to STAT Activation

It has been demonstrated that LRP1 mediates intracellular signaling activation after binding with different receptor-associated ligands, ^{35 –38} including α₂M*-induced Mek1-ERK1/2 activation. ^16,39 To evaluate whether the GFAP expression induced by α₂M* is mediated by Mek1-ERK1/2 activation, we used PD980059 as a pharmacologic inhibitor of this intracellular signaling pathway. In this way, although Mek1-ERK1/2 was activated by α₂M* stimulus for 2 hours in MIO-M1 cells, GFAP expression was not modified after PD980059 pretreatment (Fig. 6a), indicating that this intracellular transduction pathway was not the primary one involved in the induction of GFAP by α₂M* in MIO-M1 cells. Similar results were obtained when MIO-M1 cells were stimulated with α₂M* for 30 minutes, 1 hour, or 4 hours (data not shown). On the other hand, it has been reported that the Janus tyrosine kinase/signal transducer and activator of the transcription (JAK/STAT) signal transduction pathway mediates GFAP expression in retinal Müller cells stimulated with the ciliary neurotrophic factor (CNTF). ⁷ Hence, we evaluated whether α₂M* could induce STAT3 phosphorylation (p-STAT3) in MIO-M1 cells by immunoblot assays using a specific monoclonal anti–p-STAT3 antibody. Figure 6b shows that α₂M* (60 nM) induced the phosphorylation of STAT3 from 0.5 hour of incubation, with quantitative analysis by densitometry demonstrating that p-STAT3 activation increased approximately fourfold at 1 hour and approximately sixfold at 4 hours of incubation with α₂M* in MIO-M1 cell cultures. In addition, pretreatment with GST-RAP significantly reduced the α₂M*-induced STAT3 activation (Fig. 6c). These results, taken together, suggest that the α₂M*/LRP1 interaction induced the activation of the JAK/STAT signal transduction pathway, which may be responsible for promoting GFAP expression in MIO-M1 cells. 

Figure 6.

 

STAT3 but not MAPK-ERK1/2 signaling pathway activated by α₂M*/LRP1 interaction mediates GFAP expression in MIO-M1 cells. (a) Total proteins were extracted from MIO-M1 cells treated with 60 nM α₂M* for 2 hours. Previous to α₂M* stimulation, cell cultures were incubated with 10 μM PD980059 (PD) for 30 minutes in serum-free medium. GFAP and phosphorylated ERK1/2 were detected with primary antibodies and analyzed by Western blot analysis. The protein loading control of total ERK1/2 is also shown. (b) Total proteins were extracted from MIO-M1 cells treated with 60 nM α₂M* for 0.5, 1, or 4 hours, and the p-STAT3 was also analyzed by Western blot analysis using an anti–p-STAT3 (p-Y⁷⁰⁵) primary antibody. The protein loading control of calreticulin is also shown. The bars are the relative intensity of p-STAT3/calreticulin with respect to control 1 (dotted line), representing the mean ± SE of triplicate experiments. *P < 0.01, significantly different from controls (bars 3 and 5, respectively). (c) Before α₂M* stimulation, cell cultures were incubated with 400 nM GST-RAP for 30 minutes in serum-free medium. After electrophoresis and electrotransfer to the nitrocellulose membrane, p-STAT3 was detected with a primary antibody. Bars represent the relative intensity of p-STAT3/calreticulin respect to control 1 (dotted line), for the mean ± SE of triplicate experiments. *P < 0.01, statistical significance with respect to bars 3 and 5, respectively.

α₂M* Induces In Vivo GFAP Expression

Our in vitro results clearly demonstrate that α₂M* induces GFAP expression in MIO-M1–derived Müller glial cells. To examine whether α₂M* had any effect on GFAP expression in the mouse retina, this protein was injected intravitreally in adult mice. A detailed procedure is described in Materials and Methods. For this purpose, we first confirmed that LRP1 was expressed in primary cultures of Müller cells isolated from mice (Fig 7a). In addition, by immunofluorescence it was demonstrated that LRP1 was preferentially expressed in mouse retina at the inner limiting membrane (ILM) and the inner nuclear layer (L), coinciding with the Müller cell localization (Fig 7b). Bearing this in mind, a small quantity of α₂M* (∼1 μg) was injected into the vitreous of adult mice to obtain an estimated α₂M* concentration of approximately 50 ng/μL. This amount was selected based on the α₂M vitreous concentrations previously reported for a glaucomatous rat model (mean value, ∼50 ng/μL) ⁸ and human retinal disorders (mean value of α₂M concentration, 18.6 ng/μL; interval limits, 7.4–54.0 ng/μL). ¹⁸ After survival times ranging from 1 day to 1 week, α₂M*-injected eyes were processed for GFAP-immunocytochemistry, and contralateral PBS-injected eyes were used as control. One day after injection (Fig. 8B), GFAP immunoreactivity was similar to that of the control (Fig. 8A). However, at 3 and 6 days (Figs. 8C and 8D, respectively), an increase in retinal GFAP levels was observed, primarily in astrocytes, with this increase spreading to the glial processes radially oriented from the ILM to the outer retina associated with activated Müller cells. In fact, GFAP-immunostaining increased in intensity throughout the retina and remained at a high level from 1 week to 3 weeks after injection (data not shown). Finally, in contralateral PBS-injected eyes, GFAP immunostaining was weak in the nerve fiber and ganglion cell layer (GCL), both containing astrocytes (Figs. 8A, 8E). 

Figure 7.

 

LRP1 is expressed in Müller cells of the mouse retina. (a) Total proteins extracted from primary cultures of mouse Müller cells were analyzed by Western blot analysis. LRP1 was detected with a mouse anti–β subunit LRP1 (clone 5A6). The protein loading control of calreticulin is also shown. J774A.1 and JHU-4 cell lines were used as positive controls of LRP1 expression, whereas the LRP1-deficient (CHOK-1) 13–5-1 cell line was used as the negative control. (b) Representative immunofluorescence of retinal section of adult mouse stained with mouse anti–β subunit LRP1 antibody. The pattern of immunoreactivity for LRP1 appeared preferentially at the levels of the ILM (arrowheads) and the INL (arrows), with typical punctate labeling. Scale bar, 50 μm. IPL, inner plexiform layer; ONL, outer nuclear layer; OPL, outer plexiform layer.

Figure 8.

 

Localization of GFAP in the retinas of adult mice after intravitreal α₂M* injection. Eyes were processed for immunocytochemistry. Days 1 (B), 3 (C), and 6 (D) after α₂M* injection. (A) Retina of a contralateral PBS-injected eye at day 3. Scale bar, 50 μm. (E, F) Magnified images of (A) and (C), respectively. Scale bar, 25 μm. (E, F) Cell nuclei counterstained with Hoechst 33258 are also shown (blue). Arrows: Müller cell processes. IPL, inner plexiform layer; ONL, outer nuclear layer; OPL, outer plexiform layer.

α₂M* Induces In Vivo STAT3 Activation

Considering that α₂M* induced STAT3 activation in Müller glial cells, we next focused on changes in p-STAT3 in the same α₂M*-treated mice retinas. p-STAT3 labeling was found to be strongest 1 day after a single intravitreal α₂M* injection and was localized primarily in the GCL and in the INL (Fig. 9B). This pattern of staining was maintained for up to 3 days of α₂M* treatment (Fig. 9C). However, a gradual disappearance of labeling in these areas was observed at 6 days (Fig. 9D), similar to that seen in the labeling pattern of the representative contralateral PBS-treated eyes at days 1 and 3 (data not shown) and on day 6 (Fig. 9A). Considering our in vitro results together with the α₂M*-induced STAT3 activation in Müller cells, we suggest that the positive cells for p-STAT3 in the INL are in fact Müller cells. 

Figure 9.

 

Localization of p-STAT3 in the retinas of adult mice after intravitreal α₂M* injection. Representative sections of the same retinas labeled for GFAP (Fig. 8) were probed with anti–p-STAT3 (p-Y⁷⁰⁵) antibody. Retinas at days 1 (B), 3 (C), and 6 (D) after α₂M* injection and a contralateral vehicle-treated eye at day 6 (A). In addition, nonspecific p-STAT3 staining was noted in blood vessels. Scale bar, 50 μm. IPL, inner plexiform layer; ONL, outer nuclear layer; OPL, outer plexiform layer.

Discussion

As is the case in astroglia, GFAP expression in retinal Müller cells is strongly upregulated in response to a variety of retinal degenerative conditions or injuries. ³ Although the reason GFAP upregulation occurs in Müller cells is unknown, it has been suggested that these cells sense abnormal changes in the retina. In this way, GFAP expression in Müller cells might be a secondary response to the loss of retinal neurons or in other situations might occur even when the retina appears to be histologically normal, without any signs of cellular degeneration or loss. ² In the present work, we showed for first time that the α₂M*/LRP1 interaction promotes GFAP expression in MIO-M1 cells and that this expression was significantly blocked by RAP, an antagonist of LRP1-binding ligands. Moreover, α₂M* intraocularly injected in normal mice eyes caused a progressive increase of GFAP in Müller cells at 3 and 6 days after intravitreal injection. 

It is known that α₂M is a soluble protein constitutively and principally expressed by the liver. ⁴⁰ However, we have previously demonstrated that this protein is highly expressed with LRP1 in rat retinas with oxygen-induced neovascularization ¹⁷ and in the retinas of patients with diabetic or Sickle cell retinopathies, with both proteins also detected in Müller cells. ⁴¹ In addition, we showed increased levels of α₂M in the vitreous of patients with diabetic retinopathy. ¹⁸  

Evidence in the literature indicates that α₂M mRNA is highly expressed during experimental glaucoma in rats. ⁴² Related to this, a recent in situ mRNA hybridization study showed that α₂M was preferentially expressed in the inner and outer nuclear layers in glaucoma, whereas α₂M protein was detected in ganglion cells, Müller cell end feet, and astrocytes. ⁸ Although these latest results clearly indicate that α₂M is synthesized by retinal cells, blood-derived α₂M may also enter retinal tissue by the breakdown of blood-retinal barriers in different ischemic diseases. ⁴³ Coincidently, GFAP is also upregulated in injured retinas, with its expression a hallmark of stress response in the mammalian eye. In the present study, we demonstrated by in vitro assays that α₂M* was able to increase GFAP expression in Müller cells. Although the functional links between both proteins have not been clearly defined yet, a number of important publications ^{44 –49} have involved α₂M as a mediator of neurotoxicity in the central nervous system. In addition, previous studies from our laboratory and others have demonstrated that rat and human activated α-macroglobulins inhibit nerve growth factor–promoted neuritogenesis through LRP1 in PC12 cells. ^29,50,51 Independently of the indirect action of α₂M on neuronal death, an interesting finding in our study is that α₂M can directly act on Müller cells, showing a direct link between the upregulation of GFAP and the α₂M*/LRP1 system. 

Previous studies ^{52 –54} have indicated that α₂M is the main glycoprotein able to inhibit proteinases and to transport growth factors and cytokines in the blood and in other extracellular spaces. In addition, it has been reported that α₂M* is also capable of activating multiple intracellular signaling pathways. ^16,55,56 Related to this, we previously reported that α₂M* binding to LRP1 induced MAPK-ERK1/2 activation in a macrophage-derived cell line that was fully blocked by RAP, an antagonist of LRP1-binding ligands, and also by PD980059, a specific inhibitor for the Mek1-ERK1/2 pathway. ^16,39 In this work, although we did not examine the biological response downstream generated by α₂M*-induced MAPK-ERK1/2 activation, PD980059 was unable to inhibit GFAP expression, which indicated that Mek1-ERK1/2 is not the main intracellular transduction pathway involved in the induction of GFAP by α₂M* in MIO-M1 cells. On the other hand, this GFAP induction was completely inhibited by RAP, indicating that the α₂M*/LRP1 interaction is involved in the activation of MIO-M1 cells. 

It is well established that CNTF may induce GFAP expression in retinal Müller cells through the JAK/STAT signaling pathway. ⁷ Interestingly, in the present work, we demonstrated that α₂M*/LRP1 interaction induced the phosphorylation of STAT3 in MIO-M1 cells from 30 minutes of treatment. In the same way, retinas of mice intravitreally injected with α₂M* showed an increased level of STAT3 phosphorylation after 1 day of treatment that was primarily observed in the GCL and INL, indicating that JAK/STAT activation occurred in Müller cells. Considering that α₂M* induced GFAP expression in MIO-M1 cells from 1 hour of stimuli and in the retina from 3 days, these data taken together strongly suggest that the JAK-STAT pathway activated by the α₂M*/LRP1 interaction is involved in the GFAP expression in Müller cells. However, additional experiments must be carried out to demonstrate this possibility. 

Finally, an interesting finding of our study was the substantial difference between the in vitro and in vivo α₂M* effects related to GFAP expression in Müller cells and the time course of these effects. Although in the animal model α₂M* was very effective in inducing GFAP in the whole retina, visualized as an increasing number of radial cell processes, the in vitro model elicited a weak response because only approximately 10% of the MIO-M1 cells expressed GFAP by α₂M* stimulation. This difference in the number of GFAP-positive Müller cells under α₂M* treatment could be due, in part, to the interaction of this protein with growth factors or cytokines. In the same sense, the regulatory effect of α₂M* with growth factors or cytokines might have caused a delay in the effect from the site of injection to the retina with respect to that observed in the in vitro model. In this way, it has been demonstrated that α₂M* interacts with and regulates the biological functions of FGF ^57,58 and CNTF, ^59,60 which have been shown to support the survival of retinal neurons in a variety of damage paradigms and to regulate GFAP expression in Müller cells. ^61,62 However, further studies should be performed to clarify this issue. 

In conclusion, our present results establish a functional relationship between α₂M*/LRP1 interaction and GFAP expression. Furthermore, α₂M is upregulated in degenerative diseases such as retinopathies and glaucoma, which could constitute a potential target for therapeutic intervention. 

Footnotes

 Supported by grants from the Secretaria de Ciencia y Tecnología, Universidad Nacional de Córdoba; FONCyT PICT 01207 and PICT 1642; and CONICET PIP 112–200801-02067, Argentina.

Footnotes

 Disclosure: P.F. Barcelona, None; S.G. Ortiz, None; G.A. Chiabrando, None; M.C. Sánchez, None