Effects of Ischemia–Reperfusion on Physiological Properties of Müller Glial Cells in the Porcine Retina

Abstract

Purpose.: Transient retinal ischemia–reperfusion is associated with neuronal degeneration and activation of Müller glial cells. Reactive gliosis may impede the homeostatic functions of Müller cells. A viable animal model for human ischemic events should display similarities in eye size and retinal blood supply. Therefore, pigs were used in this investigation of physiological alterations in Müller cells after ischemia–reperfusion.

Methods.: Transient retinal ischemia was induced in young adult pigs by high intraocular pressure in one eye for 1 hour. After 3 days of reperfusion, the retinal tissue and isolated Müller cells were used for osmotic swelling recordings, whole-cell patch-clamp experiments, Ca²⁺ microfluorimetry, and immunohistochemistry.

Results.: Müller cells in retinal slices from postischemic eyes but not control cells displayed a significant swelling of the somata when osmotic stress was applied by hypotonic extracellular solution. The amplitude of K⁺ inward currents was significantly reduced (∼60% of the control value). This decrease was accompanied by a depolarization of the cell membrane. The number of Müller cell end feet displaying a Ca²⁺ increase after application of adenosine 5′-triphosphate was increased in the ischemic retina. Moreover, reactive Müller cell gliosis was characterized by an (increased) expression of vimentin, glial fibrillary acidic protein, the phosphorylated mitogen-activated protein kinases extracellular signal-related kinase (ERK) 1 and 2, and the transcription factor c-fos.

Conclusions.: The alterations of reactive Müller cells after transient ischemia of the pig eye were similar to those found in rat and rabbit models, demonstrating that the porcine retina is a suitable model for the investigation of ischemic injury.

Retinal injuries or degenerative diseases not only affect retinal neurons but are accompanied by significant alterations of the Müller glial cells, the dominant macroglia of the retina. In virtually all cases of retinal degeneration, Müller cells undergo reactive gliosis characterized by distinct alterations in gene expression, morphology, and physiology. ¹ Typical reactions of gliotic Müller cells are hypertrophy and increased expression of intermediate filaments, mainly glial fibrillary acidic protein (GFAP). ² Various ocular (and systemic) diseases are associated with retinal ischemia, which is a common cause of visual impairment and blindness. Dysfunctions in retinal blood supply may be caused by occlusions of retinal blood vessels, but also by more general eye diseases (e.g., diabetic retinopathy or glaucoma). Reperfusion injury after transient ischemia is mediated by formation of free oxygen radicals and glutamate excitotoxicity. ³ Reactive gliosis in Müller cells after transient retinal ischemia has been investigated in several species (e.g., rat, ^4,5 rabbit, ⁶ and mouse). ⁷ Similar to the changes in other retinal degenerations, postischemic gliotic Müller cells are characterized by an increase in their membrane capacitance and by certain alterations of their membrane conductance, mainly by a downregulation of K⁺ currents mediated by inwardly rectifying K⁺ (Kir4.1) channels. ^5,6,8 However, the usefulness of rodent and rabbit retinas as valuable animal models for human diseases is limited by certain differences in retinal structure. To deal with this problem, investigators use pigs in ophthalmic research, because size, cone distribution, and retinal blood circulation are similar to those of the human eye. ^9,10 Similar to all other nonprimate mammals, the porcine eye has no fovea centralis; however, the area centralis may be a comparable structure. The porcine eye has been used for retinal ischemia–reperfusion in several studies, mainly focusing on neuronal and vascular degenerations. In a recent paper increased GFAP expression as a marker for glial activation and alterations in the expression of mitogen-activated protein (MAP) kinases have been described. ¹¹ The purpose of the present study was to complement the existing literature by concentrating on physiological alterations of Müller cells in the postischemic pig retina. Moreover, retinal ischemia may cause edema ¹² and is known to be accompanied by inflammatory processes. ¹³ Therefore, we tested the anti-inflammatory corticosteroid triamcinolone acetonide which is in therapeutic use to treat retinal edema. ¹⁴  

Material and Methods

Animals and Ischemia Model

All experiments were performed in accordance with applicable German laws and with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Five young adult domestic white pigs (17–22 kg; both sexes) were used. Preparation for surgery, anesthesia, and killing was conducted as previously described. ¹⁵ Transient retinal ischemia was induced in one eye of each animal, while the other eye served as untreated control. The anterior chamber of the test eye was cannulated through the cornea from the pars plana with a 30-gauge infusion needle connected to a bag containing normal saline. The intraocular pressure was increased to 125 mm Hg for 60 minutes by elevating the saline bag (high intraocular pressure; HIOP). The stability of the HIOP was controlled by a mercury column. Interruption of blood flow was controlled by indirect ophthalmoscopy. After a survival times of 3 days, the animals were anesthetized, the eyes were excised, and the animals were killed. 

Electrophysiology

Retinal pieces were incubated in phosphate-buffered saline (PBS) containing papain (0.2 mg/mL; Roche, Mannheim, Germany) for 30 minutes at 37°C. After washing, the tissue was treated with DNase I (200 U/mL; Sigma-Aldrich, Taufkirchen, Germany), transferred into minimum essential medium (Sigma-Aldrich) and triturated by a pipette. Isolated cells were suspended in a recording chamber on the stage of a microscope (Axioskop; Carl Zeiss Meditec, Oberkochen, Germany). Müller cells were identified by their typical morphology and used for whole-cell, patch-clamp recordings within a few hours as previously described. ¹⁵ Amplitudes of inward currents were recorded at the end of the voltage step from −80 to −140 mV. To isolate fast-inactivating A-type K⁺ currents, we applied depolarizing steps after maximally activating A-type currents by a 500-ms prepulse to −120 mV and after steady state inactivating these currents by a 500-ms prepulse to −40 mV, respectively. A-type currents became visible and delayed rectifier currents were eliminated in the difference between both protocols. Amplitudes were recorded from these differences. 

Müller Cell Swelling

To determine volume changes of Müller glial cells in situ, evoked by hypotonic stress, we measured the cross-sectional area of their somata in the inner nuclear layer (INL) of the retinal slices. The experimental design has been described in detail elsewhere. ¹⁶ Briefly, 1-mm-thick retinal slices were placed in a chamber and loaded with a vital dye (1 μM; Mitotracker orange; Invitrogen, Carlsbad, CA) to stain the Müller cells. The normotonic extracellular solution contained (mM): 136 NaCl, 3 KCl, 2 CaCl₂, 1 MgCl₂, 10 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES), 11 glucose, adjusted to pH 7.4 with Tris(hydroxymethyl)aminomethane (Tris). The hypotonic solution contained 60% of control osmolarity. A gravity-fed system was used to perfuse the chamber continuously with extracellular solution and with the respective test substances. In pharmacologic experiments, adenosine 5′-triphosphate (ATP; Sigma-Aldrich), triamcinolone acetonide (Sigma-Aldrich), and 1,3-dipropyl-8-cyclopentylxanthine (DPCPX; Tocris, Bristol, UK) were used. The slices were examined with a laser scanning microscope (LSM 510 Meta [LSM] and an Achroplan 63×/0.9 water-immersion objective; Carl Zeiss Meditec). The vital dye (Mitotracker Orange; Invitrogen) was excited at 543 nm; emission was recorded with a 560-nm long-pass filter. During the experiment, the dye-stained Müller cell somata were recorded at the plane of their largest extension by continuously adjusting the focal plane. 

Fluorometric Ca²⁺ Imaging

Retinal whole mounts were incubated for 1 hour in extracellular solution containing two different calcium-sensitive fluorescence dyes, Fluo-4/AM (22 μM) and Fura-Red/AM (17 μM; Invitrogen-Molecular Probes, Eugene, OR), which are taken up mainly by Müller cells. The extracellular solution contained (mM) 110 NaCl, 3 KCl, 2 CaCl₂, 1 MgCl₂, 1 Na₂HPO₄, 0.25 glutamine, 10 HEPES-Tris, 11 glucose, 25 NaHCO₃, bubbled with carbogen (95% O₂, 5% CO₂) to pH 7.4. Test substances were added by rapid exchange of the perfusate. Fluorescence images were recorded at the LSM. The dyes were excited at 488 nm; the emission was recorded with a band-pass between 505 and 550 nm (Fluo-4) or with a 650 nm long-pass (Fura-Red). Images were taken from the ganglion cell/nerve fiber layers (GCL/NFL) every 5 seconds (for details, see also Ref. 15). 

Histochemistry

The procedure for immunohistochemical staining of retinal slices has been described. ¹⁵ Briefly, retinal pieces were fixed in 4% paraformaldehyde for 2 hours. After the pieces were washed, 70-μm thick agar slices were cut. The slices (or retinal wholemounts) were incubated in saline containing 5% normal donkey or goat serum plus 0.3% Triton X-100 for 2 (or 3) hours. Incubation with primary antibodies was performed overnight (or for 48 hours) at 4°C, after washing in 1% bovine serum albumin, incubated with secondary antibodies followed for 2 (or 4) hours at room temperature. The following antibodies were used: mouse anti-GFAP (1:200; G-A-5 clone; Sigma-Aldrich), rabbit anti-GFAP (1:500; Dako, Hamburg, Germany), mouse anti-vimentin (1:500; V9 clone; Immunotech, Marseille, France), mouse anti-MAP kinase–activated (diphosphorylated ERK1/2; 1:100; clone MAPK-YT; Sigma-Aldrich), goat anti-aquaporin-4 (1:200; Santa Cruz Biotechnology, Heidelberg, Germany), rabbit anti-Kir4.1 (1:200; Alomone, Jerusalem, Israel), rabbit anti-c-fos (1:40,000; Calbiochem, Darmstadt, Germany), carbocyanine (Cy)2-coupled goat anti-mouse IgG (1:100; Dianova, Hamburg, Germany), Cy2-coupled goat anti-rabbit IgG (1:200; Dianova), Cy3-coupled donkey anti-sheep IgG (1:200; Dianova), Cy3-coupled goat anti-rabbit IgG (1:400; Dianova), Cy3-coupled goat anti-mouse IgG (1:200; Dianova), Cy5-coupled donkey anti-rabbit IgG (Dianova; 1:200). Blood vessels were stained with biotinylated Griffonia simplicifolia isolectin B₄ (1:200; Sigma-Aldrich) and Cy2-coupled streptavidin (1:200; Dianova). Lack of unspecific staining was proven with negative controls omitting the primary antibodies. Images were recorded with the LSM. To compare histochemical data, corresponding images were acquired with the same settings of the microscope. 

Statistical Analysis

In a statistical power calculation using the Cochran-Cox approach, we estimated that a statistical significance of inward current amplitudes should be detectable from comparison of two groups of approximately 25 cells each. The use of five animals was based on this calculation. 

All electrophysiological data from control and ischemic eyes, respectively, were pooled for calculation of the means and standard deviations. For calculation of the percentage decrease of current amplitudes, the mean value of control cells was set to 100% and then the relative value for each cell was calculated for each animal separately. 

For swelling experiments, retinal slices from four animals were used. Soma areas recorded under the same condition in slices from different animals were pooled. Bar diagrams display the mean cross-sectional soma areas with standard errors recorded after a 4-minute perfusion under the respective condition using the analysis software of the LSM. Data are given as a percentage of the soma area before osmotic challenge (100%). 

The data from Ca²⁺ imaging experiments were derived from 8 (control) and 10 (HIOP) wholemounts of three different animals and were analyzed as follows: Fluorescence intensity was recorded in arbitrary units between 0 and 256 (8-bit). To exclude unspecific fluctuations of the fluorescence signal, we defined a threshold at a value of 30 units. Intensities above this threshold were considered to represent specific calcium signals. Averages were calculated from three images before application of ATP and at the peak of the ATP-evoked response, respectively. The difference between the averaged images during and before ATP was determined. Then, the area displaying a response within the total area of 230 × 230 μm was calculated. The mean value for the responding area of each control retina was set to 100%, and the responding area of the respective postischemic retina is given as a percentage (with SE) of the control value. 

Significance was determined with the Mann-Whitney U test or Kruskal-Wallis test followed by Dunn's comparison for multiple groups (Prism; Graphpad Software, San Diego, CA). 

Results

Electrophysiological Membrane Properties

The membrane conductance of porcine Müller cells from untreated retinas was dominated by K⁺ currents (Fig. 1A). A large part of these currents was mediated by inwardly rectifying K⁺ (Kir) channels, because application of the Kir channel blocker BaCl₂ (0.1 and 1 mM) resulted in an almost complete block of inward currents (evoked by hyperpolarization; Fig. 1C). Although there was an obvious interindividual variation of Kir current amplitudes in Müller cells from the control eyes (mean values for individual animals ranging from 636 ± 51 to 1507 ± 691 pA; mean values for all control cells: 1119 ± 455 pA; Table 1), a reduction of these amplitudes to 409 ± 392 pA (P < 0.0001) was observed in Müller cells from the respective ischemic eyes (ranging between 15% and 71% of the control value from the untreated eye in all five animals, mean 38% ± 30%,Table 1; Fig. 1B). 

Figure 1.

 

Representative membrane currents in pig Müller cells from a control eye (A, C, E, F) and from the contralateral ischemic eye of the same animal (B, D). (A–D) Voltage steps were applied from a holding potential of −80 mV to de- and hyperpolarizing potentials between −180 and +20 mV (250 ms, 20-mV increments). Whereas large inward currents were evoked in control cells (A), these currents were substantially reduced after ischemia (B). Application of 0.1 mM Ba²⁺ caused an almost complete block of the inward currents and a reduction of the outward currents (C, D). (E, F) Depolarizing steps to potentials from −60 to +40 mV (20-mV increments) were applied after a hyperpolarizing prepulse to −120 mV (E) or after a depolarizing prepulse to -40 mV (F). Current recordings from the same control cell as shown in (A) and (C). (E) After hyperpolarization fast transient outward currents (A-type K⁺ currents) and inward currents (voltage-dependent Na⁺ currents) could be recorded (shown in a larger timescale in the inset). (F) After the depolarizing prepulse transient currents were steady state inactivated and only sustained delayed rectifier currents were evoked.

Table 1.

 

Electrophysiological Properties of Isolated Müller Cells from Control and Ischemic Retinae

Parameter	Control	Ischemia	P
Inward current at −60-mV hyperpolarization, pA	1119 ± 455 n = 23	409 ± 392 n = 27	<0.0001
Inward current, %	100 ± 37	38 ± 30	<0.0001
Membrane potential, mV	−83 ± 4	−69 ± 19	<0.001
	n = 24	n = 27
Membrane capacitance, pF	51 ± 19	62 ± 15	<0.03
	n = 22	n = 26

 Mean values with standard deviations, derived from n cells from 5 animals.

The high K⁺ conductance of the Müller cell membrane is the prerequisite for the negative membrane potential of these cells. Thus, the reduction of Kir currents after ischemia resulted in a significant depolarization of the cells from −83 ± 4 to −69 ± 19 mV (P < 0.001, Table 1). Moreover, we found an increase in the membrane capacitance from 51 ± 19 to 62 ± 15 pF (P < 0.03, Table 1), which is known to be characteristic of hypertrophied cells in reactive gliosis. ¹⁷  

In Müller cells from control and ischemic eyes, outward currents through voltage-dependent K⁺ (Kv) channels (evoked by depolarizing steps) were less sensitive to Ba²⁺ ions than were Kir currents and, thus, were only partially reduced (Figs. 1C, 1D). The Kv-mediated currents consisted of a fast-inactivating component similar to an A-type current (Fig. 1E) and of a sustained, delayed rectifier-like current (Fig. 1F). In the presence of 0.1 mM Ba²⁺, the peak amplitude of the A-type current was 259 ± 144 pA (n = 17) in control cells and was significantly increased to 452 ± 171 pA (n = 19, P < 0.001) after ischemia. Moreover, voltage-dependent inward currents were elicited in all cells by depolarizing steps in the presence of Ba²⁺ (Fig. 1E, inset). The kinetics of these currents is consistent with that of voltage-dependent Na⁺ currents. Although an increase in the peak amplitude of voltage-dependent A-type currents (and Na⁺ currents) was observed, we evaluate these data with caution, because no specific pharmacologic tools were used to separate the fast transient current types (A-type and Na⁺ currents). 

Müller Cell Swelling

Porcine Müller cells in slices from untreated control eyes did not display a significant increase in the size of their somata during perfusion with hypotonic solution. ¹⁶ To test whether this ability is restricted after ischemia, we investigated soma swelling in retinal slices perfused with a hypotonic solution. Control Müller cells did not swell under this condition (103% ± 1%); however, preincubation with BaCl₂ (1 mM) caused a significant increase in soma size (18% ± 1%; P < 0.001, Fig. 2A). Müller cells from ischemic eyes swelled independently on the absence (14% ± 2%, P < 0.001) or presence (15% ± 1%, P < 0.001; Fig. 2A) of Ba²⁺. 

Figure 2.

 

Osmotic swelling characteristics of Müller cells after transient ischemia. (A) Whereas Müller cells in slices from untreated control eyes did not show a significant swelling in hypotonic solution, cells from ischemic eyes (HIOP) displayed an increase in their soma area. Block of K⁺ currents by Ba²⁺ (1 mM, ■) caused swelling of Müller cells from ischemic and control retinas. (B) Relationship between inward current amplitudes and the extent of osmotic swelling of Müller cell somata (relative soma area). Parallel recordings of cell swelling and membrane currents were performed in tissue from control (circles) and ischemic eyes (triangles) of four animals; each point represents the mean value of data obtained from one eye; same shades of gray label data from one animal. Solid line: the linear regression. There is a negative correlation between both parameters (r = −0.885; P < 0.005). (C) Involvement of the adenosine A₁ receptor in swelling inhibition. Application of the glucocorticoid triamcinolone (Triam; 100 μM) as well as of ATP (200 μM) caused a significant inhibition of the swelling induced by hypotonic solution in Müller cells from HIOP eyes. The effect of ATP demonstrates that P2 receptors are involved in volume regulation. The effect of both substances could be suppressed by DPCPX (100 nM, ), a selective A₁ receptor antagonist, demonstrating that A₁ receptors mediate volume regulatory processes. Data in (A) and (C) are the mean ± SEM, the number of recorded cells is given within the columns. Significant differences: *P < 0.05, ***P < 0.001, ns, not significant.

The reduction of inward currents by Ba²⁺ or by ischemia is supposed to be one reason for altered responses of Müller cells to hypotonic stress. It was tested whether the recorded increases in soma size correlate with the decreases in inward currents. A scatterplot of inward current amplitudes and relative soma areas (Fig. 2B) displays a negative correlation of both parameters (r = −0.885; P < 0.005). 

Triamcinolone has been effectively used in ocular therapeutics in some diseases (e.g., macular edema). ¹⁴ Therefore, we investigated whether this substance may inhibit Müller cell swelling under hypotonic conditions. Acute application of 100 μM triamcinolone significantly reduced swelling of Müller cells from ischemic eyes (from 113% ± 2% to 104% ± 1%; P < 0.05; Fig. 2C). We demonstrated earlier that the swelling-inhibitory effect of triamcinolone in rat Müller cells is mediated by the stimulation of a purinergic signaling cascade, which finally leads to the opening of extrusion pathways for ions. ¹⁸ This cascade is obviously also functional in porcine Müller cells from ischemic eyes, demonstrated by the fact that the swelling was suppressed by application of 200 μM ATP (102% ± 2%; P < 0.05). The effect of ATP was abrogated by preincubation with the adenosine A₁ receptor antagonist, DPCPX (100 nM; 116% ± 2%). Likewise, DPCPX also blocked the effect of triamcinolone (112% ± 1%; Fig. 2C). 

Ca²⁺ Responses of Müller Cells

It is well-known from earlier studies that mammalian Müller cells are endowed with nucleotide (P2Y) receptors which mediate intracellular Ca²⁺ release. ¹⁹ We tested the reaction of P2Y receptors by imaging the intracellular Ca²⁺ concentration during application of 200 μM ATP onto retinal whole mounts. In control retina, this resulted in a transient Ca²⁺ increase in some Müller cell end feet. In about one-fifth of the recorded retinal area, a response was recordable. This value was set 100%, to compare it with data from ischemic tissue. In wholemounts obtained from the postischemic retina the area covered by responding Müller cell end feet displaying a Ca²⁺ response increased significantly more than twofold (P < 0.01; Fig. 3). 

Figure 3.

 

Transient ischemia causes an upregulation in Ca²⁺ responses to extracellular ATP. The images show the fluorescence of Fluo-4 in the ganglion cell layer of a retinal whole mount from an ischemic (HIOP) eye, before and during application of ATP (200 μM). Significantly fewer responses to ATP were observed in untreated control tissue (histogram; **P < 0.01). For quantification, the area showing a fluorescence above a threshold of 12% of the maximum intensity was determined.

Intermediate Filaments

Immunoreactivity for the filament protein GFAP in control retinas was mainly found in the GCL/NFL where the retinal astrocytes (in addition to Müller cell end feet) are located. However, a faint staining of a small number of individual Müller cell stem processes in the inner plexiform layer (IPL) was also observed (Fig. 4, control). After transient ischemia, the GFAP immunoreactivity was strongly increased. In addition to the staining in the GCL, many Müller cells were immunopositive throughout the whole retinal thickness (Fig. 4; HIOP). The filament protein vimentin is constitutively expressed in Müller cells. In HIOP-treated eyes, there was an obvious upregulation of the protein. The vimentin immunoreactivity appeared stronger throughout the retina and immunopositive Müller cell stem processes were thicker than in the control tissue (Fig. 4). Increased immunoreactivity for both intermediate filaments indicates hypertrophy of the gliotic Müller cells. 

Figure 4.

 

Transient ischemia (HIOP) altered the expression of filament proteins in Müller cells. Whereas in the control retina GFAP immunoreactivity was found mainly in astrocytes in the GCL and in a small number of Müller cell stem processes in the IPL, the number of GFAP-positive Müller cells was clearly increased after transient ischemia. Similarly, immunoreactivity of the Müller cell marker vimentin was stronger after ischemia. Retinal tissue from two different animals was used for immunostaining against GFAP and vimentin, respectively. INL, inner nuclear layer; ONL, outer nuclear layer.

pERK and c-Fos Immunoreactivity

Another way to verify glial activation after retinal injury is detection of the transcription factor c-fos and of activation (phosphorylation) of extracellular signal–related kinase 1 and 2 (ERK1/2), which are involved in signaling processes. In the normal retina almost no immunoreactivity for c-fos was found. Three days after transient ischemia, the number of c-fos immunopositive cell nuclei was obviously increased. Double labeling against GFAP strongly suggested that some of these nuclei belong to Müller cells (Fig. 5A). The immunoreactivity against phosphorylated (p)ERK1/2 in the control retina was restricted to a few cell bodies in the GCL and the INL in addition to two horizontal bands in the IPL. After HIOP an additional staining of Müller cells was found. The identity of these cells was verified by their typical morphology and by colocalization of GFAP (Fig. 5B). 

Figure 5.

 

Upregulation of c-fos and pERK after ischemia (HIOP). (A) Whereas no c-fos-immunopositive cells were found in the control retina, immunoreactivity was obvious in some cells in the postischemic retina. Because of the localization in the INL and double-staining against GFAP, it is supposed that Müller cells also express c-fos in their nuclei (arrows). (B) Immunoreactivity against pERK was found in the control retina in some cells in the GCL and INL, as well in two distinct bands in the IPL. After HIOP, immunoreactive cells were observed spanning the whole retinal thickness and displaying GFAP staining, demonstrating that pERK is expressed by Müller cells (arrows). ONL, outer nuclear layer; OPL, outer plexiform layer.

AQP4 and Kir4.1 Immunoreactivity

The water channel protein aquaporin-4 (AQP4) and the K⁺ channel subunit Kir4.1 were specifically expressed by Müller cells within the retina. Both proteins were concentrated in membrane areas abutting blood vessels or the vitreous. Within Müller cell end feet in the GCL of the control retina, the AQP4 immunoreactivity was localized in a specific pattern at the cell borders. Three days after ischemia, this typical staining changed, and the Müller cell end feet displayed an even distribution of AQP4 immunoreactivity (Fig. 6A). In the inner retina, AQP4 immunoreactivity was observed at the surface of blood vessels in control as well as in postischemic tissue without remarkable alterations (Fig. 6B). However, the immunolocalization of the Kir4.1 protein was altered after HIOP. The perivascular Kir4.1 immunoreactivity was strongly decreased around the vessels in the INL (Figs. 6B, 6C). 

Figure 6.

 

Immunolocalization of the water channel protein AQP4 and the K⁺ channel subunit Kir4.1 in retinal whole-mount preparations before (control) and after transient ischemia (HIOP). (A) In the GCL of the control retina, AQP4 was localized in the membranes of Müller cell end feet in a distinct pattern at the cell borders. After ischemia, this distribution was altered, and the AQP4 immunoreactivity was more evenly distributed. (B, C) Glial membranes contacting blood vessels in the inner nuclear layer of control tissue were stained by antibodies against AQP4 and Kir4.1. Immunoreactivity for Kir4.1 is marked by arrows in (C, control). Moreover, the isolectin B₄ bound specifically to the vessels. Isolectin and AQP4 staining remained unaltered after HIOP, whereas the Kir4.1 immunoreactivity at the vessels disappeared, which is unequivocally visible in (C).

Discussion

Retinal ischemia is thought to be one factor involved in the development of retinal edema, which is of particular importance in the case of macular edema in the human eye. ^12,20 Dysfunction of glial cells plays a decisive role in edema formation because removal of excess water from the inner retina is one of the important functions of Müller cells. ^21,22 Transport of water is osmotically coupled to the transport of ions. To mediate both transport functions, AQP4 and Kir4.1 channels are colocalized in Müller cell membrane domains abutting the vitreous and retinal blood vessels. ²³ Müller cells display a significant reduction of their inward K⁺ conductance in cases of reactive gliosis during retinal injury and disease. ^8,24 Using the pig as an animal model for transient retinal ischemia, we confirmed our previous results from ischemia–reperfusion models in rat ⁴ and rabbit. ⁶ Because we observed in an earlier study, using the postischemic rat retina, that maximum downregulation of inward currents occurred after 3 days, ⁵ we used this survival time in the present study on pigs. Moreover, one pig survived 7 days after ischemia. Inward current amplitudes of Müller cells from the ischemic eye of this animal were decreased to 20% ± 8% of the control eye, accompanied by depolarization and increased membrane capacitance. Müller cells in retinal slices from this eye swelled to 115% ± 2% in hypotonic solution. Because these data were derived from only one animal, they were not included in the statistical analysis; however, it is suggested that the physiological alterations are stable for at least 1 week. The reduction of inward currents (probably mediated by Kir4.1 channels) was accompanied by a reduction of Kir4.1 immunoreactivity, mainly around blood vessels. In accordance with our previous data from rat and mouse, ^4,7 the alterations in AQP4 immunostaining were less prominent than for Kir4.1; however, we observed a redistribution of AQP4 immunoreactivity in Müller cell end feet. It remains to be clarified in future studies whether this redistribution causes alterations in the water fluxes between Müller cell end feet and the vitreous. 

The decrease in K⁺ currents correlates with an increased susceptibility of Müller cells to osmotic stress. Müller cells in slices from postischemic tissue lost their ability for efficient volume regulation and displayed swelling of their somata in hypotonic solution. The fact that a swelling of Müller cells could be induced in control tissue by application of the Kir channel blocker Ba²⁺ suggests that the reduction of the K⁺ conductance causes the loss of the volume regulatory function. However, it cannot be excluded that additional, hitherto unknown, alterations in the postischemic tissue impede water and ion movements across glial membranes. For example, we observed in an earlier study that Müller cell swelling in porcine retinal slices can be induced by inflammatory mediators and oxidative stress. Likewise, the swelling of Müller cells from the detached retina could be blocked by inhibiting oxidative stress or the formation of inflammatory mediators. ¹⁶ Although we were not able to repeat a detailed pharmacologic study, the effects of triamcinolone, ATP, and DPCPX suggest that a swelling-inhibitory signaling cascade described before in Müller cells from the (postischemic) rat retina ^18,25 can also be activated in the current pig model. The effect of triamcinolone was prevented by the adenosine A₁ receptor antagonist DPCPX, suggesting that the effect of triamcinolone is mediated by stimulation of endogenous adenosine signaling. 

Next, we observed a significant upregulation of ATP-evoked Ca²⁺ responses. It is well known from a variety of species that Müller cells display P2Y receptor-mediated increases in Ca²⁺. ^{26 –29} In a model of retinal detachment in pig, a transient upregulation of Ca²⁺ responses occurred with amplitudes similar to the data presented here. ¹⁵ Thus, upregulation of ATP-mediated Ca²⁺ reactions may be a general response of Müller cells to various pathologic conditions. 

In addition to physiological alterations, we found changes in the protein expression induced by ischemia–reperfusion. Data from a similar study focusing on MAP kinases were recently published. ¹¹ However, the main difference between the work of Gesslein et al. ¹¹ and ours in the present study is the difference in survival time. Whereas reperfusion time was 3 days in this study, the longest time in their study was 20 hours. Upregulation of intermediate filaments is a typical sign of reactive Müller cell gliosis ^8,30 and has been described before in a rat model of ischemia-reperfusion, ^4,5 in experimental retinal detachment in pig, ¹⁵ and as early as 12 hours after reperfusion in the pig ischemic model. ¹¹ The strong expression of GFAP in reactive Müller cells was used for double immunolabeling with other marker proteins. Thus, we demonstrated that expression of c-fos was induced in Müller cell nuclei after HIOP. Moreover, some Müller cells displayed immunoreactivity for pERK. This confirms and extends the data of Gesslein et al. ¹¹ who found a fast and transient upregulation of pERK after 5 hours. Differences in the duration of ERK1/2 activation have been described before. Thus, an increase in pERK immunoreactivity even after 7 days was found in the detached cat retina, ³¹ whereas a similar increase was observed only during the first 24 hours after transient ischemia of the rat retina. ³² Thus, differences in the experimental procedures (e.g., the age of the pigs) may account for the differences between our study and that of Gesslein et al. ¹¹ Moreover, we did not investigate pERK-expression after shorter times of reperfusion and may thus have missed the maximum activation. In any case, our co-immunolocalization with GFAP demonstrates that pERK is indeed expressed by reactive Müller cells. 

Phosphorylation of ERK1/2 may cause a subsequent activation of transcription factors. The immediate early gene products c-jun and c-fos are components of the activator protein (AP)-1 complex of transcription factors and may be activated in Müller cells by several growth factors (for a review, see Ref. 33). Whereas an increase in c-jun immunoreactivity has been described, ¹¹ in our study, nuclei of GFAP-positive Müller cell displayed c-fos immunoreactivity. It has been demonstrated before that activation of ERK1/2 and c-fos in Müller cells may rescue retinal neurons from cell death. ^32,34 This is obvious support for the fact that reactive gliosis has protective as well as detrimental effects. 

In summary, in this study, transient retinal ischemia in a pig model resulted in distinct physiological alterations of Müller cells, which may reflect a dysregulated water and ion transport. The changes in membrane properties and swelling characteristics were accompanied by alterations in the protein expression, known to be typical of reactive glial cells. The porcine eye may be a valuable model for pharmacologic studies testing the therapeutic potential of modifying gliotic reactions. 

Footnotes

 Supported Deutsche Forschungsgemeinschaft Grants RE 849/10 (SSP1172), RE 849/12 (FOR748), and GRK 1097/1 (AR).

Footnotes

 Disclosure: A. Wurm, None; I. Iandiev, None; S. Uhlmann, None; P. Wiedemann, None; A. Reichenbach, None; A. Bringmann, None; T. Pannicke, None