Regional Alterations in Müller Cell Protein Expression in Human and a Rat Model of Geographic Atrophy

Abstract

Purpose: Despite being crucial to neuronal survival, the role Müller cells play in geographic atrophy (GA) has only recently been considered. We investigated whether Müller cells retain their normal functional profile or form a fibrotic scar when remodeling in human GA eyes and our subretinal sodium iodate (NaIO₃) model.

Methods: Sprague Dawley rats given subretinal injections of NaIO₃ (5 mg/mL) were sacrificed at 3 and 12 weeks. Cryosections and retinal flatmounts from rats and cryosections from human GA eyes were stained with antibodies against the Müller cell proteins glutamine synthetase (GS), inwardly rectifying potassium channel 4.1 (Kir4.1), aquaporin 4 (AQP4), cellular retinaldehyde-binding protein 1 (CRALBP), and glial fibrillary acidic protein (GFAP), as well as alpha smooth muscle actin (α-SMA), fibronectin, and collagens I and IV. The immunofluorescence intensity of AQP4 and Kir4.1 was quantified using Image J, and Kir4.1 protein levels were verified by western blot.

Results: In both human GA eyes and NaIO₃-injected rats, Müller cell processes at the external limiting membrane (ELM) descent and in the subretinal membrane exhibited increased GS expression. GFAP was elevated throughout the Müller cells. AQP4 staining at the ELM descent was particularly pronounced throughout the radial processes, including those extending into the subretinal space. In NaIO₃-injected rats, perivascular Kir4.1 expression significantly decreased in the atrophic retina, but expression increased in the subretinal glial membrane. α-SMA and extracellular matrix proteins were not detected in the subretinal membrane.

Conclusions: Our findings underscore the persistence of homeostatic proteins, albeit altered, in Müller cells as they remodel and extend into the subretinal space.

Müller cells, the primary glial cells within the retina, are the first cells to respond to injury. As Müller cells become reactive, they release both neurotrophic and pro-inflammatory cytokines aimed at neuroprotection.^1,2 Over time, however, this response can be detrimental as Müller cells become gliotic, a process involving hypertrophy, proliferation, and migration.^1–4 It is not fully understood how these changes affect the normal function of Müller cells. Given that retinal neurons rely on Müller cells for numerous functions, including maintaining retinal homeostasis and synaptic function, understanding how these glial cells are affected is crucial to understanding retinal disease.² 

One disease in which the role of Müller cells has recently garnered increased attention is age-related macular degeneration (AMD), the leading cause of blindness in the elderly.^6–10 Geographic atrophy (GA), the advanced form of dry AMD, is characterized by degenerative lesions that involve the loss of retinal pigment epithelium (RPE) cells, dropout of choroidal capillaries, and death of photoreceptors.⁵ We and others have identified subretinal glial membranes composed primarily of Müller cell processes occupying the atrophic area in GA.^6,7,11,13 These membranes also contained sporadic astrocytes. Similar glial membranes were also reported in human donor eyes with Stargardt disease, retinitis pigmentosa, and choroideremia, as well as animal models of retinal degeneration.^14–18 Similar membranes are also observed following retinal detachment.¹⁹ 

In other areas of the central nervous system, such as spinal cord injury, astrocytes create a glial scar with extracellular matrix (ECM) components.^20–22 These glial scars can protect healthy tissue from damage but at the same time prevent neuronal regeneration and axonal growth.²³ In the retina, Müller cells and astrocytes create similar scars in response to optic nerve injury and ischemia.^24,25 Müller cells also undergo mesenchymal transition, reducing glial marker expression and increasing expression of fibroblast markers, such as smooth muscle actin, when forming epiretinal membranes.²⁶ Although this process is often referred to as glial scarring, it has not been demonstrated that Müller cells indeed form true scars with ECM components in the subretinal space.¹⁸ Considering the critical role of Müller cells in maintaining retinal health, it is crucial to determine whether these cells undergo fibroblastic transformation or retain their typical functional properties when remodeling to form subretinal glial membranes in GA.^6,7,11 

In the present study, we investigated the expression of proteins associated with some key Müller cell functions crucial to retinal health in human GA eyes and the rat subretinal sodium iodate (NaIO₃) model, described in detail below.^27,28 To this end, we examined glutamine synthetase (GS), an enzyme that deaminates glutamate to glutamine. Müller cells osmoregulate via inwardly rectifying potassium channel 4.1 (Kir4.1) and aquaporin 4 (AQP4), which are responsible for retinal potassium and water homeostasis, respectively; hence, we assessed their expression. Cellular retinaldehyde-binding protein 1 (CRALBP), a carrier of 11-cis-retinoids that is essential for cone PR dark adaptation, was also analyzed. Additionally, we analyzed glial fibrillary acidic protein (GFAP), a marker of glial activation. Finally, we investigated the presence of alpha smooth muscle actin (α-SMA), a marker of myofibroblasts, and the ECM proteins collagens I and IV and fibronectin within the glial membranes. 

The subretinal NaIO₃ model is unique among other delivery methods in that we create a very focal region of atrophy surrounded by normal retina and RPE. This model mimics other key features of human GA such as choriocapillaris and photoreceptor loss after RPE atrophy, subretinal glial membrane formation, development of basal laminar deposits (BLamDs), and outer retinal tubulations (ORTs).^27,28 A key feature that stands out in GA and the subretinal NaIO₃ model is the external limiting membrane (ELM) descent.^28–30 According to Sarks and others,^29,30 the ELM descent is the curved line created by the ELM descending toward Bruch's membrane (BrM) that delineates the atrophic border signified by RPE loss, outer segment absence, and inner segment shortening at the descent. This anatomical feature is observed clinically on optical coherence tomography (OCT) and has been defined histologically.³⁰ Müller cell processes creating the ELM descent appear to create a barrier between the healthy retina and that already affected in GA.^12,30 

We assessed protein expression at 3 and 12 weeks post-injection in the rat model to study temporal changes and compared these observations to human eyes with GA. Since our goal was to determine the Müller cell state in later stages of degeneration, after a glial membrane has formed, we focused on later time points in our model. Understanding whether Müller cells maintain their homeostatic functions or transition to a more fibroblastic state is essential for developing treatments for GA, as healthy Müller cells are required to support neurons and photoreceptors. 

Materials and Methods

Donor Eyes

Human tissue was used in accordance with the tenets of the Declaration of Helsinki with the approval of the Joint Committee on Clinical Investigation at Johns Hopkins University School of Medicine. Eyes from three elderly control subjects, 81 to 91 years of age, with no ocular pathology, were obtained through the National Disease Research Interchange in Philadelphia, PA. Two male patients, 92 and 98 years of age, with GA (referred to as “GA eyes”) were included in the study. As observed in the gross images, both GA donors had severe GA (Supplementary Fig. S1). The retinal and choroidal pathology in these patients was extensively studied in our previous study.⁷ Donor information is summarized in Table 1. All eyes were received within 22 to 48 hours of death after shipping on wet ice and processed as described below. 

Animals

All animal experiments adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the protocols were approved by the Animal Care and Use Committee at Johns Hopkins University. Experiments were conducted with adult male (6 to 8 weeks old) Sprague Dawley rats (Envigo, Indianapolis, IN, USA). Rats were fed standard laboratory chow (LabDiet; PMI Nutrition International, Arden Hills, MN, USA) and allowed free access to water with a 12-hour light/12-hour dark cycle in a climate-controlled animal facility. 

Subretinal NaIO₃ Injection

Rats were anaesthetized via intraperitoneal injection of a mixture containing 100 mg/mL ketamine (VetOne; MWI Veterinary Supply, Boise, ID, USA) and 20 mg/mL xylazine (Akorn, Lake Forest, IL, USA). Sodium iodate (S-4007; Sigma-Aldrich, St. Louis, MO, USA) was freshly prepared in phosphate-buffered saline (PBS; Quality Biological, Gaithersburg, MD, USA) to achieve a concentration of 5 mg/mL. Each rat received a single 1-µL subretinal injection of sterilized NaIO₃ solution in both eyes, as previously described.^27,28 Briefly, subretinal injections were given using sterilized glass microneedles back-filled with injection solution and connected to a controlled pressure delivery device (PLI-100 Pico-Injector; Harvard Apparatus, Holliston, MA, USA). The success of subretinal injections was clearly indicated by the creation of a bleb. As a control, an equivalent volume of PBS was injected into other animals. Animals were euthanized at 3 and 12 weeks post-injection. A minimum of four rats per time point were analyzed using immunohistochemistry (both flatmounts and cryosections). 

Tissue Processing for Cryosectioning

Human donor eyes were processed as described previously.⁷ The eyes were opened at the limbus, the anterior segments were removed, and the posterior eyecups were examined using a ZEISS Stemi 2000-C Stereo Microscope (Carl Zeiss Microscopy, Oberkochen, Germany). The posterior eyecups were fixed in 2% paraformaldehyde (PFA) in 0.1-M phosphate buffer with 5% sucrose for 2 hours at room temperature. Rat tissues were processed as described previously.^27,28 Briefly, rats were euthanized at 3 and 12 weeks after PBS and NaIO₃ injections. Eyes were enucleated immediately, and the anterior structures were removed, creating an eyecup. Eyes exhibiting vitreous or subretinal hemorrhage were excluded from the study after being examined under a stereo dissecting microscope. Eyecups were then fixed for 2 hours in 2% PFA in 0.1-M phosphate buffer with 5% sucrose. Human and rat tissues were then washed in 0.1-M phosphate buffer with 5% sucrose and increased gradients of sucrose before cryopreservation in 20% sucrose in 0.1-M phosphate buffer with Tissue-Tek O.C.T. compound (Sakura Finetek, Torrance, CA, USA). 

Flatmount Immunohistochemistry

At 12 weeks post-injection, the eyes were enucleated, with one eye from each rat being cryopreserved and the other used for retinal flatmounts. The retinas were immunostained following the methods described previously.^27,28 Briefly, the tissues were fixed overnight in 2% PFA in Tris-buffered saline (TBS; Quality Biological) at 4°C. Afterward, retinas were washed and blocked with 5% normal goat serum (Jackson ImmunoResearch, West Grove, PA, USA) in TBS containing 0.1% bovine serum albumin (BSA; Signa-Aldrich, St. Louis, MO, USA) and 1% Triton X-100 (TBST-BSA) for 6 hours at 4°C. The retinas were then incubated overnight at 4°C with a primary antibody cocktail, followed by three washes in TBST and another overnight incubation at 4°C with a secondary antibody cocktail. Before imaging, four cuts were made to flatten the retina. Z-stack images were captured at 20× magnification using a ZEISS LSM 710 confocal microscope and Zen software. 

Immunolabeling

For immunostaining, cryosections were permeabilized with chilled methanol, air dried, and blocked for 20 minutes with 2% goat serum (Jackson ImmunoResearch). Cryosections were then incubated for 2 hours at room temperature with primary antibodies at the dilutions specified in Table 2. Primary antibodies were detected with anti-chicken Alexa Fluor 647 (Invitrogen, Carlsbad, CA, USA), anti-rabbit Alexa Fluor 647 (Invitrogen), or anti-mouse Cy3 (Jackson ImmunoResearch). Sections were incubated with secondary antibodies at a 1:500 dilution for 30 minutes at room temperature. The secondary cocktail contained 2-(4-amidinophenyl)-1H -indole-6-carboxamidine (DAPI) diluted 1:1000. Griffonia simplicifolia isolectin (GS-isolectin; Invitrogen) was used in rat tissue sections and Ulex europaeus agglutinin (UEA) lectin (GeneTex, Irvine, CA) was used in human tissue sections to stain blood vessels. Nonimmune IgG controls were diluted to 1 mg/mL and run at 1:100 (matching our most concentrated antibodies) to verify the specificity of our staining. Sections were imaged using a ZEISS 710 confocal microscope. The imaging settings, including laser power, pinhole, gain, and other capture parameters, were saved and consistently applied for each antibody combination to ensure uniform conditions while imaging the tissues. 

Table 1.

 

Human Donor Eyes

 

Table 2.

 

List of Primary Antibodies Used in This Study

 

Quantification of AQP4 and Kir4.1 Intensity

As we observed drastic changes in the localization for these proteins, we quantitatively assessed area-specific differences in AQP4 and Kir4.1 immunoreactivity between atrophic and non-atrophic regions in NaIO₃-injected rats using confocal images from cryosections captured using identical confocal laser settings (e.g., laser power, gain, pinhole) within 3 days. The images for different regions were collected from the same sections, and all sections for analysis were stained at the same time, eliminating any potential for variations due to immunohistochemistry. In Fiji (ImageJ; National Institutes of Health, Bethesda, MD, USA), the single-channel TIFF file exported from the Zen software was opened. The straight-line tool was selected and set to a width of 10 pixels. A straight line was drawn through the full thickness of the retina on the image. Under the “Analyze” menu, “Plot Profile” was selected to generate a plot of the grayscale values along the line. In the “Plot Profile” window, the “List” button was clicked to display the grayscale values, and the peak gray values were identified. The data were saved as a CSV file. The CSV file was opened in Excel, where the top 10 peak grayscale values were copied and averaged. For each time point and protein, we analyzed three sections from three different rats. Protein intensity measurements were taken at different regions within each section, including non-atrophic, atrophic, and border areas. For AQP4, we took a total of 18 measurements from the control sections and three measurements from each section within each pathological region (for a total of nine data points). For Kir4.1, given the marked reduction observed at the inner limiting membrane (ILM) and an increase at the ELM, we performed additional quantification, taking 10 measurements per section for each region (atrophic, non-atrophic, and border). 

Western Blot

For protein isolation, the lens and cornea were removed before cutting the eyecup roughly in half, with one half containing the atrophic area, which was clearly visible under the dissecting microscope. The retina was then separated from the RPE/choroid. Protein was isolated using T-PER Tissue Protein Extraction Reagent (Thermo Fisher Scientific, Waltham, MA, USA) with protease inhibitor. Protein was quantified using the Rapid Gold BCA Protein Assay Kit (Thermo Fisher Scientific). Protein samples for 3-week post-NaIO₃ retinas and controls were prepared by mixing each sample (15 mg) with 5 µL of Novex 4× Bolt loading dye (Thermo Fisher Scientific) and 2 µL of 10× sample reducing agent β-mercaptoethanol (M3148; Sigma-Aldrich). The samples were denatured at 70°C for 10 minutes followed by brief centrifugation. Electrophoresis was conducted using an Invitrogen system with a 4% to 12% Invitrogen gradient gel. Samples (20 µL) were loaded into the gel. Electrophoresis was performed at 200 V for 50 minutes. Following electrophoresis, proteins were transferred to polyvinylidene fluoride membranes using the iBlot System (Life Technologies, Carlsbad, CA, USA). The membranes were then blocked in Blocker FL Fluorescent Blocking Buffer (37565; Thermo Fisher Scientific) for 1 hour at room temperature followed by incubation with primary antibody Kir4.1 (1:1000; Abcam, Cambridge, UK) in blocking buffer overnight at 4°C. Following primary incubation, the membrane underwent three sequential 20-minute washes in wash buffer. Subsequently, it was incubated with secondary antibodies diluted 1:3000 in wash buffer for 1 hour, followed by three 20-minute washes in wash buffer before being imaged using an Invitrogen iBright FL1500 Imaging System. The same membrane was stripped using Restore PLUS Western Blot Stripping Buffer (Thermo Fisher Scientific) for 5 to 10 minutes. After stripping, the membrane was washed three times and subsequently reblocked followed by incubation with β-actin. The protein abundance was quantified using standard densitometry analysis with ImageJ software using β-actin as the normalization protein. 

Statistical Analysis

All statistical analyses were performed using Prism 9.3.1 (GraphPad, Boston, MA, USA). Descriptive statistics are shown as the mean ± standard error of the mean (SEM). The differences in variables between and across the groups were assessed by one-way ANOVA with post hoc Bonferroni correction. 

Results

Glutamine Synthetase is Strongly Expressed by Müller Cells Creating the ELM Descent and Subretinal Membrane

Sections of rats given subretinal injections of NaIO₃ were assessed at 3 weeks and 12 weeks and compared to PBS-injected controls. As we observed no difference between 3- and 12-week PBS-injected control tissue, only 3-week control eyes are reported herein. In the control retina, GS was observed within the Müller cell radial processes spanning from the ILM to the ELM as well as the cell body in the inner nuclear layer (INL) (Figs. 1A, 1B). The strongest expression was observed within endfeet and processes in the inner plexiform layer. In NaIO₃-injected eyes, the atrophic border was well demarcated by the ELM descent, as indicated by the arrows in Figures 1C, 1D, 1G, and 1H. In 3-week NaIO₃ injected eyes, GS immunostaining was more intense throughout the retina on the non-atrophic aspect of the border. Increased GS expression was noted at the ELM descent, as indicated by the arrows, and in the subretinal membrane at 3 and 12 weeks (Figs. 1C, 1D, 1G, 1H). In the atrophic area at 12 weeks, we observed horizontally orientated GS⁺ cell processes in the subretinal space running parallel with BrM/choroid (Figs. 1I, 1J). GS expression in the inner retina was decreased in the atrophic region at 12 weeks (Figs. 1I, 1J). The expression pattern of GS in NaIO₃-injected eyes distant from the border of atrophy was similar to controls. 

Figure 1.

 

GS localization in rats. (A, B) In PBS-injected rat retinas, GS staining was localized exclusively to Müller cells, highlighting their endfeet, radial processes, cell bodies, and termination at the ELM. (C, D, G, H) In the NaIO₃-injected eyes, increased GS expression was observed in Müller cells at the ELM descent (arrows) at 3 and 12 weeks. (E, F, I, J) GS expression was also localized to Müller cell processes of the glial membrane in the subretinal space within the atrophic areas (arrowheads). By 12 weeks, GS staining in the inner retina was markedly reduced in the atrophic region. Scale bar: 100 µm.

In human control eyes, GS staining was observed in Müller cell processes, extending from the ILM to the ELM (Figs. 2A, 2B). A particularly strong signal was observed in the Müller cell endfeet located at the ILM. As observed in NaIO₃-injected rats, cryosections from eyes with GA demonstrated increased GS expression at the ELM descent (Figs. 2C, 2D, arrow). In one eye, an ORT was observed adjacent to the non-atrophic aspect of the border (Figs. 2C, 2D, asterisk) with GS⁺ Müller cells around the photoreceptor nuclei present in the outer nuclear layer (ONL). Outer retinal tubulations have a circular or oval appearance in cross-sections, with photoreceptors that completely encircle the lumen. Compared to the atrophic area adjacent to the ELM descent, the central atrophic area, away from the ELM descent, exhibited a thicker GS⁺ membrane parallel to BrM (Figs. 2E, 2F). At higher magnification, GS immunostaining demonstrated intense staining of Müller cell processes in the glial membrane (Figs. 2G, 2H). 

Figure 2.

 

GS immunoreactivity in human. (A, B) In the controls, GS immunoreactivity in Müller cells spanned the entire retinal thickness from the ILM to the ELM. (C, D) In a section from a GA eye, the ELM descent toward BrM is shown (arrow). The atrophic area is recognized by a preserved INL and loss of ONL. Glutamine synthetase expression diminished at the ILM near the atrophic border, while the ELM descent was intensely stained (arrows). (E, F) The subretinal glial membrane in the atrophic area exhibited markedly increased GS expression compared to the control. (G, H) High-magnification images revealed weak staining of Müller cell processes in the inner retina and strong GS immunoreactivity in the glial membrane (arrowheads). Scale bar: 100 µm. An asterisk (*) indicates outer retinal tubulation.

Müller Cell Polarized AQP4 Expression is Altered and Becomes Diffuse Throughout Radial Processes

Immunofluorescence labeling for AQP4 in PBS-injected eyes was primarily localized to Müller cell endfeet and astrocytes in the inner retina, gradually declining in intensity toward the outer plexiform layer (Figs. 3A, 3B). Additionally, strong perivascular staining was observed surrounding the deep capillary plexus, which comprises the capillary beds on either side of the inner nuclear layer (Figs. 3A, 3B). Weak AQP4 immunostaining was also observed in the RPE. At both 3 and 12 weeks post-NaIO₃ injection, there was increased AQP4 expression in the Müller cell radial processes of the INL in the non-atrophic aspect of the ELM descent. At the ELM descent, there was particularly strong AQP4 staining throughout the radial processes and processes extending into the subretinal space (Figs. 3D, 3E, 3G, 3H, arrows). In the atrophic area, increased AQP4 expression was observed in Müller cell processes of the subretinal glial membrane (Figs. 3J, 3K, 3M, 3N, arrowheads). The expression pattern of AQP4 in NaIO₃-injected eyes distant from the border of atrophy was similar to controls. Quantification of immunofluorescence intensity revealed higher levels of AQP4 in the atrophic region and ELM descent compared to the non-atrophic area (both P = 0.01) and 3-week controls (P < 0.0001) (Fig. 4A). At 12 weeks, the intensity difference was significantly higher at the border (P = 0.0004) and in the atrophic area (P < 0.0001) compared to the non-atrophic region (Fig. 4B). Again, 3- and 12-week controls were similar in expression pattern and staining intensity. 

Figure 3.

 

Localization of AQP4 in rat. (A–C) In the PBS-injected control, AQP4 was localized to glial cell processes of the inner retina and around the lectin-positive deep capillary plexus. (D–I) In NaIO₃-injected eyes at both 3 and 12 weeks after injection, AQP4 labeling was increased in Müller cell processes on either side of the ELM descent (arrows). (J–O) AQP4 staining was markedly increased in the subretinal glial membrane at both time points (arrowheads). Scale bar: 100 µm.

Figure 4.

 

Quantification of AQP4 fluorescence intensity was measured throughout the retina for 3(A) and 12(B) week post NaIO₃ (n = 3). Fluorescence intensity analysis was performed using a minimum of three representative images of the control, non-atrophic, and atrophic regions and ELM descent from each animal per age group. Fluorescent intensity is expressed as arbitrary units (AU). Comparison within the different pathologic regions was done using a two-way ANOVA multiple-comparisons test. *P < 0.05, ***P < 0.001, and ****P ≤ 0.0001; ns, non-significant.

In human control eyes, AQP4 expression was polarized, showing the highest concentration at the endfeet of Müller cells and astrocytes in the inner retina. This expression gradually diminished within Müller cell processes toward the ELM (Figs. 5A, 5B). Additionally, moderate AQP4 labeling was observed around the deep capillary plexuses as highlighted by UEA lectin staining (Figs. 5B, 5C). In contrast, eyes with GA exhibited a marked increase in AQP4 immunolabeling in the inner retina at the non-atrophic and atrophic aspects of ELM descent (Figs. 5D, 5E, arrows). Similar to the rat model, the AQP4⁺ glial membrane terminated near the non-atrophic border (Figs. 5D, 5E). Increased AQP4 expression was observed in the subretinal glial membrane occupying the atrophic area (Figs. 5G, 5H, arrowheads). AQP4⁺ Müller cell processes were also observed enveloping drusen (Figs. 5G, 5H, asterisks). 

Figure 5.

 

AQP4 immunoreactivity in humans. (A–C) Immunolabeling of AQP4 in the control eye was predominantly localized to the inner retina, with faint staining observed in the deep capillary plexus and Müller cell processes of the INL. (D–F) In the GA eye, AQP4 expression was increased in the INL and was also markedly elevated at the ELM descent (arrow). (G–I) AQP4 was also observed in the subretinal glial membrane encasing drusen (asterisk). Scale bar: 100 µm.

Polarized Kir4.1 Expression is Disrupted in NaIO₃-Injected Retinas

In control rat eyes, Kir4.1 was predominantly localized to the Müller cell endfeet at the ILM (Figs. 6A–C). Similar to AQP4, Kir4.1 immunofluorescence labeling was seen in perivascular processes surrounding the deep capillary plexus that was stained with GS-isolectin (Figs. 6A–C, arrowheads). At 3 weeks post-NaIO₃ injection, Kir4.1 expression increased in Müller cell radial processes throughout the retina at both the non-atrophic and atrophic aspects of the ELM descent and in the ELM descent itself (Figs. 6D–F). Perivascular staining was decreased in the non-atrophic aspect of the ELM descent and was weak or absent in the atrophic area (Figs. 6D–F, arrowheads). At 12 weeks, diffuse Kir4.1 staining throughout Müller cell radial processes replaced perivascular staining in both the non-atrophic and atrophic aspects of the ELM descent (Figs. 6G–I). In the central atrophic area, Müller cell Kir4.1 expression within the retina was reduced but that in the subretinal glial membrane increased (Figs. 6J, 6K, 6M, 6N, paired arrows). Kir4.1 expression in Müller cell endfeet was markedly reduced at both time points in the atrophic areas (Figs. 6J–O). We conducted a semiquantitative analysis of overall Kir4.1 immunofluorescence intensity in control and NaIO₃-injected rat eyes. Analysis of the immunofluorescence intensity, as shown in Figure 7A, revealed a significant reduction in Kir4.1 staining at the atrophic border within the atrophic area compared to the control and non-atrophic regions (P < 0.0001) at 3 weeks post-NaIO₃. Subsequently, we assessed the intensity specifically at the ELM descent and the subretinal membrane (Fig. 7B). Despite an overall decrease in Kir4.1 expression, higher levels were observed at the subretinal space of the atrophic area (P < 0.0001) compared to the non-atrophic regions of the same sections. 

Figure 6.

 

Kir4.1 localization in rat. (A–C) In control rat retinas, Kir4.1 staining was prominent in the inner retina and around the deep capillary plexus (arrowheads) and only weakly at the ELM (paired arrows). (D–F) At 3 weeks post NaIO₃ injection, Kir4.1 immunostaining was strong in the inner retina and around the deep capillary plexus (arrowheads) on the non-atrophic aspect of the ELM descent. Capillaries on the atrophic aspect of the ELM descent were either weakly immunoreactive or unlabeled. Müller cells on both aspects of the ELM descent showed increased expression compared to controls as did the subretinal glial membrane (paired arrows). (G–I) At 12 weeks post NaIO₃ injection, Kir4.1staining was intense in Müller cell processes on both the non-atrophic and atrophic aspects of the ELM descent (arrow) and in the subretinal glial membrane in the atrophic region (paired arrows). Perivascular staining was markedly reduced in the non-atrophic and atrophic aspects of the ELM descent at 12 weeks. (J–O) Expression of inner retinal Kir4.1 was markedly reduced at 3 and 12 weeks within the atrophic areas compared to that at the borders. The subretinal glial membrane showed strong Kir4.1 staining in the atrophic area (paired arrows). Scale bar: 100 µm.

Figure 7.

 

Analysis of Kir4.1 expression by fluorescence intensity and western blot. The overall intensity of Kir4.1 immunofluorescence and its intensity at the ELM and subretinal glial membrane were quantified in rats 3 weeks post NaIO₃ injection (n = 3). Fluorescence intensity is expressed in arbitrary units (AU). (A) The overall expression of Kir4.1 was reduced in the ELM descent and the atrophic areas of rats 3 weeks post NaIO₃ injection compared to the non-atrophic areas of the same eye and control. (B) Kir4.1 expression at the ELM was increased in the non-atrophic side and the subretinal glial membrane of the atrophic area versus control. Comparisons within the different pathological regions were performed using a two-way ANOVA multiple-comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, non-significant. (C) Western blot analysis using retinal lysates from control and 3-week post-NaIO₃–injected rats (n ≥ 3 for each group) showed decreased expression in the atrophic area compared to the non-atrophic area of the same eye and control. Equal amounts of protein were loaded into the wells, as confirmed by β-actin levels. (D) The intensities of the bands were quantified by densitometric analysis and presented as a ratio to β-actin. Data are presented as mean ± SEM. ***P < 0.001; ns, non-significant.

We further validated the diminished Kir4.1 expression in rats through western blot analysis conducted on the atrophic and the non-atrophic retina of the same eye. We also assessed a separate control sample. As illustrated in the representative results in Figure 7C, Kir4.1 protein expression at 3 weeks post-NaIO₃ was notably reduced in the atrophic area compared to both the non-atrophic area and the control. Kir4.1 protein levels were normalized to respective β-actin levels and were plotted (Fig. 7D). The difference in band intensity was calculated for atrophic versus non-atrophic versus control at 3 weeks. Statistical analysis demonstrated significant variations in Kir4.1 band intensities between atrophic retina and control (P < 0.0001). Even though Kir4.1 expression was increased at the ELM descent, cumulatively Kir4.1 expression in the controls was much higher than the non-atrophic area (P = 0.0001). 

Müller Cells Increase GFAP Expression While Redistributing CRALBP

We assessed the expression of GFAP, which indicates Müller cell activation, in combination with the Müller cell/RPE marker CRALBP (Fig. 8). In control rats, the Müller cell inner processes and cell bodies in the INL were positive for CRALBP (Figs. 8A, 8B), but GFAP was restricted to the astrocytes in the ganglion cell layer (Figs. 8A, 8C). Compared to controls, we observed increased CRALBP expression in the RPE of NaIO₃-injected rat eyes 3 and 12 weeks post-injection in the non-atrophic aspect of the ELM descent (Figs. 8D, 8E, 8G, 8H, arrows). At 3 weeks post-injection, CRALPB⁺ Müller cell somata in the INL were clearly distinguishable (Fig. 8E), as were radial fibers extending toward the outer retina. The increased expression in the outer retina was associated with reduced CRALBP expression within the inner plexiform layer. Faint immunostaining for CRALBP was observed in parts of the innermost Müller cell processes. The expression pattern of CRALBP in NaIO₃-injected eyes distant from atrophy was similar to controls (data not shown). GFAP⁺ Müller cell processes were observed spanning the entire retinal thickness in both the non-atrophic and atrophic regions (Figs. 8D, 8F). By 12 weeks, the CRALBP immunoreactivity was reduced in the Müller cell soma but remained strong in the processes extending through the ONL (non-atrophic region), as well as in the ELM descent and subretinal membrane. At this same time point, some radial processes were stained with GFAP in the non-atrophic area (Figs. 8G, 8I). These processes extended from the ILM to the ELM, and the labeled processes within the retina appeared slightly thickened. GFAP staining at both time points was most intense near the ILM (Figs. 8F, 8I). At the ELM descent, GFAP immunoreactivity of outer Müller cell processes was upregulated (Figs. 8E, 8H). At both time points, CRALBP⁺/GFAP⁺ Müller cell processes occupied the subretinal space within the atrophic regions, forming a glial membrane (Figs. 8D–O, arrowheads). 

Figure 8.

 

CRALBP expression in rat. (A, B) In PBS-injected control eyes, CRALBP immunolabeling was observed in the Müller cell processes of the inner retina, INL, and RPE. The processes in the ONL were only weakly labeled. (C) GFAP immunostaining was limited to astrocytes in the nerve fiber layer near the ILM. (D, E, G, H) Strong CRALBP staining was observed in the RPE of the non-atrophic region near the border and at the ELM descent (arrows) at both 3 weeks (D, E) and 12 weeks (G, H) post NaIO₃ injection. (F, I) Müller cell processes displayed strong labeling for GFAP spanning the entire retina in the non-atrophic and atrophic regions at 3 and 12 weeks post NaIO₃ injection. (J–O) Within the atrophic regions, the subretinal glial membranes (arrowheads) were double-labeled for both CRALBP and GFAP at 3 and 12 weeks. Scale bar: 100 µm.

In control human eyes, GFAP was primarily confined to astrocytes in the inner retina, but CRALBP-labeled Müller cell processes spanned the entire retinal thickness (Figs. 9A–C). In GA eyes, although the overall CRALBP expression was not significantly changed, we noted an increase at the ELM descent (arrows) and in the subretinal glial membrane (Figs. 9D, 9E). Similarly, GFAP was increased throughout the retina on both aspects of the ELM descent (Figs. 9D, 9F, arrow). In the atrophic area, a GFAP⁺/CRALBP⁺ glial membrane occupied the subretinal space in place of the degenerated photoreceptors (Figs. 9G–I, arrowheads). 

Figure 9.

 

CRALBP and GFAP in human eyes. (A, C) CRALBP and GFAP minimally overlapped at the ILM, and weak staining was seen in the other retinal layers. (B) CRALBP weakly labeled Müller cell bodies in the INL and processes of endfeet at the vitreal surface and outer processes at the ELM. (C) In the control, only astrocytes at the inner margin of the retina contained GFAP. (D–F) In the GA eye, CRALBP and GFAP intensity was increased throughout the Müller cell processes of the retina and at the ELM descent (arrows). (G–I) CRALBP and GFAP co-localization was evident in the subretinal glial membranes (arrowhead). Scale bar: 100 µm.

Müller Cell Processes Invade the Choroid in the NaIO₃ Model

As we reported previously, Müller cell processes traverse BrM within atrophic regions in NaIO₃-injected rats and invade the choroidal stroma.²² To examine this further, sections were double labeled for AQP4 and GFAP (Fig. 10), as well as with CRALBP and GS (data not shown). These double-labeled sections confirmed the invasion of Müller cell processes into the underlying choroid at both 3 weeks (Figs. 10A–F, arrows) and 12 weeks (Figs. 10G–J, paired arrows) in NaIO₃-injected rats. In some cases, the processes enveloped or appeared to invade the choroidal capillaries (Fig. 10, arrowheads). Notably, these intrachoroidal Müller cell processes continued to express their characteristic markers, such as AQP4, GFAP, CRALBP, and GS. The control rats had no immunostaining for glial markers in the choroid (Supplementary Fig. S2). 

Figure 10.

 

Müller cell processes invade choroid in rat NaIO₃ model. (A–F) Section from the atrophic region of a rat eye at 3 weeks post NaIO₃ injection showing AQP4 and GFAP immunoreactivity in Müller cell processes surrounding lectin-positive retinal capillaries (arrowheads) and extending through BrM (dashed line), where they enveloped the lectin-positive choriocapillaris (arrows). (G–J) Section from the atrophic region of a rat eye at 12 weeks after NaIO₃ injection showing AQP4 and GFAP immunoreactivity in the radial processes of Müller cells. Extensive AQP4 and GFAP-positive Müller cell processes were observed extending beyond BrM (dashed line) into the choroidal stroma (paired arrows). Scale bars: 25 µm (A–F) and 50 µm (G–J). CC, choriocapillaris.

Do Müller Cells Undergo Mesenchymal Transition to Form a True Scar?

To determine whether Müller cells within subretinal glial membranes transform into a fibroblastic phenotype, we used α-SMA as a marker of activated fibroblasts. Both human GA and rat retinal sections were immunolabeled with α-SMA antibody and GS (Fig. 11). In both cases, α-SMA labeling was observed in the retinal blood vessels (presumably arteries and arterioles) (Fig. 11A–J, arrowheads) and choroidal vessels. The GS-positive glial membranes in both NaIO₃-injected rat and human GA sections were negative for α-SMA (Figs. 11C–J, paired arrows). In atrophic regions of rat retinal flatmounts at 12 weeks after NaIO₃ injection, some retinal vessels were labeled with α-SMA, but the processes within the membranes only expressed glial cell markers (Fig. 11K). 

Figure 11.

 

Fibrosis marker α-SMA in subretinal glial membrane of GA and NaIO₃-injected rats. (A, B) Large retinal vessels (arrowheads) were positive for α-SMA in both control and GA sections. (C, D) The subretinal glial membrane (paired arrows) adjacent to the ELM descent (arrows) was negative for α-SMA. (E, F) PBS-injected rat eye showed positive staining for α-SMA in the retinal blood vessels (arrowheads). (G, H, J) In rat eyes at both 3 weeks and 12 weeks after NaIO₃ injection, the subretinal membranes in the atrophic area (paired arrows) adjacent to the ELM descent (arrows) were negative for α-SMA. (I) In flatmounted retinas 12 weeks after NaIO₃ injection (right), the glial scar showed α-SMA labeling of retinal blood vessels; however, the membrane with disorganized glial processes positive for GFAP was negative for α-SMA. Scale bars: 100 µm (A–J) and 50 µm (K).

We then examined the localization of ECM protein in the glial membranes. In the rat model, we examined fibronectin (Supplementary Figs. S3A, S3B), collagen I (Supplementary Figs. S3C, S3D), and collagen IV (Supplementary Figs. S3E, S3F) in rat retinal sections, all of which were negative in the subretinal glial membrane. We then investigated the localization of collagen I in human sections (Fig. 12). Collagen I was detected in the blood vessels of the retina, the ILM, and throughout the choroid in both aged controls and GA eyes (Fig. 12). The GFAP⁺ subretinal glial membrane in the atrophic area of the GA eyes did not exhibit collagen I expression in the atrophic aspects of ELM descent or in the central atrophic area (Figs. 12D–I, arrow). Notably, we observed two weakly labeled bands of collagen I. One band was continuous with a band underlying RPE in the non-atrophic region and the other was observed along the inner aspect of drusen. These bands merged at the ELM descent. Non-immune IgG controls to match these antibodies are shown in Supplementary Figure S4. 

Figure 12.

 

ECM marker collagen I in human eyes. (A–C) In the control eye retinal vessels, the ILM and the choroidal stroma stained for collagen I. (D–I) In GA sections, as in the controls, collagen I labeled the ILM and retinal vessels; however, the subretinal glial membrane, both in the atrophic aspect of the ELM descent (D–F, arrows) and in the center of the atrophic area (G–I, arrowheads), was negative for collagen I. BLamDs overlying calcified drusen (asterisks) displayed weak collagen I immunoreactivity. Scale bars: 100 µm (A–I).

In summary, our findings reveal significant alterations in the expression and localization of Müller cell proteins in both the human and the rat model at the border and atrophic regions compared to controls. Table 3 summarizes the key observations. 

Table 3.

 

Summary of Altered Protein Expression

 

Discussion

In the present study, we set out to determine whether the Müller cells creating subretinal membranes in GA and our rat NaIO₃ model retained their normal profiles or took on more fibroblastic characteristics, as is observed with astrocytes creating a glial scar following spinal cord injury.²⁰ Observations were identical between the rat model and human GA eyes so these are discussed together below unless otherwise specified. We found that Müller cells preserved their characteristic proteins (GS, CRALBP, Kir4.1, and AQP4). The cells within the glial membrane did not contain α-SMA or ECM proteins. Together, these data indicate that Müller cells did not become fibrotic or create a true scar as seen in other areas of the central nervous system. The localization of these proteins, however, shifted within the Müller cells associated with atrophy and at the atrophic border. These changes are discussed in more detail below. 

Confirming previous results, we observed significant increases in GFAP in Müller cells within the atrophic area, as well as in the non-atrophic retina adjacent to atrophy.^6,7,27,28 The increased GFAP in the non-atrophic retina indicates Müller cell activation in advance of RPE and photoreceptor atrophy. It seems likely that, although RPE and photoreceptors were still present, their cellular functions may be impaired, thereby stimulating Müller cell activation. Indeed, recent OCT evidence indicates photoreceptor changes and inner retinal neurodegeneration beyond the border of RPE loss in eyes with GA.^13,31,32 It has been suggested that the photoreceptor degeneration outside the GA area may serve as a predictor of disease progression and could be a factor in identifying patients that will respond best to certain treatments.³² Therefore, Müller cell activation and other changes outside the GA area could also impact disease progression. The ELM descent is a central pathological feature of GA in humans but has not been reported in the systemic NaIO₃ model.³³ Since GA progresses at the atrophic border, the ability to study changes at the ELM descent in an animal model could prove crucial to understanding GA expansion. 

GS was highly expressed within the subretinal glial membrane. This increased GS expression in the subretinal glia may be explained by previous findings suggesting its role in detoxifying elevated ammonia levels in hepatic retinopathy.^34,35 Increased ammonia levels could occur due to changes in the subretinal space as well as vessel leakage.³⁶ This increased GS expression may also serve as a compensatory mechanism to protect inner retinal neurons from excess glutamate that may accumulate in the subretinal space due to photoreceptor death. In retinitis pigmentosa, findings from Jones et al.¹⁷ showed increased GS in Müller cells; however, this expression was reduced as they formed a subretinal glial membrane. The difference between this study and ours could be due to the disease pathologies and progression. It is likely that, in GA, Müller cells retain their GS expression to protect surviving photoreceptors and inner retinal neurons. Astrocytes exposed to lipopolysaccharides show a rapid increase in GS expression.³⁷ Therefore, it is also possible that GS is increased in the subretinal space in response to chronic inflammation. 

AQP4 polarization within Müller cells was significantly altered in the non-atrophic side of the border, as well as within the atrophic area. We observed increased AQP4 throughout Müller cell radial processes and within the subretinal membranes. The staining throughout radial processes would increase Müller cell water permeability while decreasing water transport to the plasma, potentially causing retinal edema. It is possible that astrocytes, which we have observed sporadically in subretinal membranes, contribute to some of the AQP4 staining in the subretinal space. Based on our previous characterization of these membranes, however, we believe Müller cell processes are the main component.^6,7,27,28 A similar shift in Müller cell AQP4 expression was reported in retinas following exposure to blue light injury.³⁸ This shift could be a compensatory mechanism to remove extracellular fluid and prevent subretinal edema that could be caused by breakdown of the outer blood–retinal barrier with RPE degeneration and death. In addition, damaged and dying cells may release ions and water, contributing to the need for increased AQP4 in the region.^39,40 

AQP4 is highly expressed by astrocytes surrounding amyloid plaques in Alzheimer's disease, where they are speculated to assist in the clearance of amyloid while also potentially creating a barrier around the plaque.^42,43 AQP4⁺ glial cells could have a similar function in the subretinal space, assisting in the clearance of remnant photoreceptor debris and other toxins where the RPE is lost. Interestingly, we also observed increased AQP4 in Müller cell processes surrounding and, in some cases, projecting into drusen (Fig. 5). In this regard, Müller cell AQP4 could be important for clearing drusen components or by ensheathing them to create a barrier. 

In the brain, shifts in AQP4 expression are crucial for astrocyte migration and scar formation following injury.⁴⁴ Therefore, we propose that the shift in Müller cell AQP4 expression may promote extension and migration of their processes into the subretinal space, driving membrane formation. Indeed, we do see strong AQP4 expression at the ELM descent. Further research is needed to investigate the role that AQP4 plays in Müller cell migration and subretinal membrane formation. 

In the subretinal NaIO₃ model, we observed a notable decrease in Kir4.1 expression at the ILM within the atrophic area, coupled with an increase in the subretinal glial membrane. This increase was particularly striking at the ELM descent. The shift in Kir4.1 protein expression in the subretinal space, along with that of AQP4 and GS, indicates a potential Müller cell polarity change. Although this has not been previously reported, we propose the novel concept that, when Müller cells proliferate in the early stages of gliosis, they orient themselves with reversed polarity. Further support of this lies in our previous observation that Müller cells create endfeet-like structures on BrM in our rat model.²⁸ Although additional research is required to investigate this novel concept, it is also possible that the localization of these proteins simply shifts. 

Another prominent observation in the NaIO₃ model was the decrease in perivascular Kir4.1 expression within the retina, particularly in atrophic regions. A similar reduction in overall, perivascular, and endfeet Kir4.1 expression has been reported in retinas exposed to ischemia-reperfusion and blue light.^38,45,46 Kir4.1 is also reduced in the Royal College of Surgeons rat, which experiences slow retinal degeneration, as well as a transgenic rat with polycystrin 2 mutation leading to photoreceptor loss.^39,47 In these models, Kir4.1 was observed throughout the entire Müller cell processes, similar to our observations at the ELM descent and in the non-atrophic retina adjacent to the border. Importantly, dystrophin, which is known to regulate the Kir4.1 polarity within Müller cells, was also disrupted following ischemia.^46,48,49 Kir4.1 is observed throughout the Müller cell processes early in development, suggesting that this redistribution could be a sign of dedifferentiation.^38,50 Reduced Kir4.1 expression surrounding retinal vessels would also impair the ability of Müller cells to release K⁺, leading to its accumulation in Müller cells and an altered osmotic gradient with the plasma. Ultimately, this could lead to cell swelling as water enters the retina, depending on AQP4 channel regulation.⁴⁶ Diffuse Kir4.1 distribution throughout the Müller cells processes also would allow increased K⁺ into the extracellular space surrounding neurons, potentially leading to neuronal hyperexcitation and glutamate production, which could be toxic.³⁹ Therefore, the redistribution of Kir4.1 could contribute to neurodegeneration that occurs secondary to RPE and photoreceptor loss in AMD and has recently been reported to occur in areas without RPE loss in GA.^30,31,39 Finally, the reduced K⁺ influx into the blood could stimulate vascular changes leading to the reduced retinal vessel density that has been reported in GA.^7,51,52 

Not only did Müller cells maintain their CRALBP expression within their radial processes, but this expression also extended into the processes creating the subretinal membrane. Similar increased CRALBP expression was reported in light-induced retinal degeneration (LIRD) rats.⁵³ In the LIRD rats, however, the CRALBP was reduced in later stages of atrophy, along with GS. It seems likely that Müller cells may increase CRALBP to ensure that retinoids are available to the neural retina. Since Müller cell CRALBP is important for the cone visual cycle, the preserved CRALBP could explain why cones survive longer in GA than rods.⁵⁴ Since retinoic acid is also used for adult neurogenesis in other parts of the CNS, perhaps this increased expression is part of the remodeling process.⁵⁵ We also observed increased CRALBP expression by RPE on the non-atrophic side of the ELM descent, potentially indicating that these are stressed cells. 

We and others have previously demonstrated that Müller cell processes not only extend beyond the ELM and abut BrM but some transverse BrM to reach the choriocapillaris and choroidal stroma, as shown in our rat model, as well as in GA.^6,7,28 In the current study, we further demonstrated that Müller cell processes within the choroidal stroma surrounding choroidal capillaries were positive for AQP4, CRALBP (data not shown), and GS (data not shown), in addition to GFAP as shown in our earlier study.²⁸ Furthermore, a study by Sullivan et al.⁵⁶ demonstrated that, in addition to Müller cells, amacrine and rod bipolar cells were present in the underlying choroid in AMD. Their study also suggested that Müller cells in the choroid might serve as conduits, potentially forming synaptic connections. The maintenance of GS and CRALBP in the choroid supports this idea. The high expression of AQP4 surrounding choroidal vessels in atrophic regions could be an attempt to maintain the outer blood–retinal barrier in the absence of RPE or for osmoregulation. 

The Müller cell processes creating the subretinal glial membrane in GA eyes were negative for α-SMA and collagen I. We did, however, observe two bands of collagen I. One band was continuous with the labeling observed under the RPE in the non-atrophic region. We believe this represents persistent Basal laminar deposits, which can remain after the RPE is lost.⁵⁷ Of importance, GFAP⁺ processes were seen on both the retina and BrM side of this collagen I band in the atrophic and non-atrophic sides of the ELM descent. In the atrophic area, a second collagen I band was observed on the choroidal aspect of the glial membrane, suggesting that Müller cells may secrete collagen 1. In vitro studies have shown that activated Müller cells express and secrete collagen I.⁵⁸ It is important to note, however, that, even in the areas containing collagen I, Müller cells retain their signature proteins, indicating that they have not transformed into another cell type. Further research on additional GA eyes and perhaps with later time points in the NaIO_3, model is required to fully understand this observation. 

Despite the important findings in this study, there are a few limitations to be addressed. The characterization of Müller cell profiles was based on a small sample size of only two human GA subjects. Due to the challenges in obtaining human tissues, we supplemented our findings with experiments conducted using our subretinal NaIO₃ model. This study further validates how closely the pathology in our model mimics that in GA at the border, where disease progression occurs. Unfortunately, we have yet to find a Kir4.1 antibody that works well on human retinal sections, so we were unable to assess this protein in our GA eyes. There are also additional Müller cell functional proteins to investigate in our tissue, including those important for synaptic support. We cannot rule out that some Müller cells may undergo epithelial transition, and we are looking into this possibility. Although beyond the scope of the present study, we are currently investigating how early in disease pathology, in both AMD and our rat model, Müller cell protein changes occur. 

In conclusion, our results are promising, as they indicate that Müller cells retain their metabolic and functional proteins rather than taking on a more fibroblastic phenotype. The diffuse expression of Kir4.1 and AQP4, however, indicates that Müller ion and water regulation is impaired. These changes likely contribute to neurodegeneration occurring in GA. Glial changes adjacent to atrophy where RPE and PR are intact indicate that the retina may not be as healthy here as currently thought. It is crucial that we fully understand changes happening in the retina, as these are key to restoring and/or retaining vision. The glial membrane itself also creates a strong adhesion between the retina and choroid, which likely makes the administration of treatments subretinally difficult. Given the lack of improvement in vision in response to current treatments, new alternatives are essential. Perhaps new treatments should target chronic gliosis and/or be combined with treatments to reduce glial membrane formation. 

Acknowledgments

The authors are grateful to the donors and their families for their generous gift to science. The authors thank Shreya Jolly and Olivia Wanex for assistance with immunohistochemistry and manuscript review and Thomas Freddo, OD, PhD, for critical review of the manuscript. 

Supported by a grant from the National Eye Institute, National Institutes of Health (R01EY031044 to MME), a Wilmer Core Grant for Vision Research (EY001765), the Bright Focus Foundation (MME), Tom Clancy Professorship funds (ME), the Altsheler-Durelll Foundation (MME), and Research to Prevent Blindness unrestricted funds to the Wilmer Eye Institute. 

This work was partially presented at the 2024 ARVO Annual Meeting, Seattle, WA. 

Disclosure: P. Naik, None; D.S. McLeod, None; I.A. Bhutto, None; M.M. Edwards, None