A Developmental Switch in the Expression of Aquaporin-4 and Kir4.1 from Horizontal to Müller Cells in Mouse Retina

Alejandra Bosco; Karen Cusato; Grazia P. Nicchia; Antonio Frigeri; David C. Spray

doi:10.1167/iovs.05-0385

Abstract

purpose. In adult retina, aquaporin-4 (AQP4) and inwardly rectifying K⁺ (Kir4.1) channels localize to astrocyte and Müller cell membranes facing vascular and vitreous compartments, optimizing clearance of extracellular K⁺ and water from the synaptic layers. However, it is unknown whether these channels are expressed at early developmental stages, before gliogenesis or angiogenesis take place in the neural retina. This study was conducted to determine the presence of AQP4 and Kir4.1 proteins in the developing mouse retina.

methods. Simultaneous AQP4 and Kir4.1 immunodetection was performed in postnatal mice 1, 9, 15, and 30 days of age. Confocal microscopy was used to identify the cellular distribution of AQP4 and Kir4.1 proteins, as well as their coexpression with the cell-selective immunomarkers Prox-1, calbindin, and neurofilament.

results. AQP4 and Kir4.1 proteins were coexpressed in calbindin- and Prox1-expressing retinal neurons at birth. These neurons were identified as horizontal cells based on their position and morphology. By P15, when vision starts, AQP4 and Kir4.1 localization coordinately switched from horizontal cells to Müller glial cells.

conclusions. The findings showed that AQP4 and Kir4.1 protein expression is confined to differentiating horizontal cells before its expression in Müller cells. The finding of AQP4 in neurons is novel, since AQP4 expression within the central nervous system is restricted to glia. Also, the results demonstrated that AQP4 is a horizontal cell-specific immunomarker in neonatal retina. The transitory coexpression of AQP4 and Kir4.1 proteins by differentiating horizontal interneurons suggests that these cells mediate K⁺ and water transcellular uptake until the initiation of phototransduction, when glial cells assume these functions.

Ion and water homeostasis of the central nervous system (CNS) neuronal environment is primarily controlled by astroglia. Extracellular potassium ion (K⁺) efflux from active neurons is passively taken up by astrocytes through strong, inwardly rectifying K⁺ (Kir) channels, redistributed throughout the astrocytic syncytium via gap junctions, and cleared away into the blood or the cerebrospinal fluid via weak Kir channels. This glial K⁺ spatial buffering prevents K⁺ build-up in the narrow CNS extracellular space, which would affect diverse neuronal processes.¹ ² ³ ⁴ ⁵Water flux via aquaporin-4 (AQP4) channels⁶is thought to be linked to K⁺ spatial buffering.⁷ ⁸The association between K⁺ and osmotic balance is structurally supported by the coenrichment of Kir4.1 and AQP4 proteins in perivascular membranes⁹ ¹⁰and is functionally corroborated by the impaired K⁺ clearance concomitant with perivascular AQP4 loss.¹¹

In the adult retina, astrocytes are confined to the superficial nerve fiber and ganglion cell layers (NFL and GCL), where their end feet envelop vitreal capillaries.¹²The end feet membrane domains contacting basal lamina express AQP4,¹³as well as Kir4.1 channels.¹⁴ ¹⁵Within the inner retina, Müller glial cells assume several astrocytic functions. Indeed, Müller cells redistribute (siphon) excess K⁺ from the extraneuronal space toward fluid reservoirs of low K⁺ (sinks), such as vitreous body, subretinal space, and blood vessels.² ⁵ ¹⁶ ¹⁷Müller cell membrane domains surrounding neurons express strongly rectifying Kir2.1 channels, which may mediate K⁺ influx into these glia.¹⁸In contrast, membrane areas facing K⁺ sinks contain arrays of weakly rectifying Kir4.1¹⁹ ²⁰ ²¹and AQP4 channels,²² ²³ ²⁴ ²⁵optimizing outward K⁺ and water flux.¹³Recent studies of retinal cytotoxic edema further elucidate the coupling of K⁺ currents to water flux.²⁶

During development, the onset of spontaneous retinal activity²⁷ ²⁸precedes Müller cell genesis,²⁹ ³⁰astrocyte arrival to the retinal surface,³¹and retinal angiogenesis³² ³³by at least five days. However, K⁺ and water uptake via AQP4 and Kir4.1 channels may already be operational in retinal cells of early genesis. Toward supporting this hypothesis, we examined AQP4 and Kir4.1 immunoexpression in the developing mouse retina. Surprisingly, AQP4 and Kir4.1 proteins were specifically coexpressed in differentiating horizontal cells at birth, whereas by the time of eye opening, the coexpression of this pair of proteins became enriched in Müller cells. The expression of the dominant regulators of retinal K⁺ spatial buffering and water flux by differentiating neurons, as well as the cytoarchitecture of these interneurons, suggest that horizontal cells may contribute to early retinal homeostasis.

Some of the present findings have been preliminarily reported (Bosco A, et al. IOVS 2004;45:ARVO E-Abstract 5324).³⁴ ³⁵

Methods

Histology

Newborn (postnatal day [P]0), P9, P15, and P30 C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) were maintained and treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Two pups from two different litters were used at each time point. Neonates and P9 pups were decapitated under deep isoflurane anesthesia. Neural retinas were dissected and flattened on membrane inserts (Millicell-CM; Millipore, Bedford, MA) and fixed in 4% (wt/vol) paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline (PBS; pH 7.4) for 30 minutes. P15 and P30 mice, anesthetized similarly, were perfused transcardially with saline solution followed by PFA, and their eyes were further postfixed for 2 hours. Either eye cups or flatmounted retinas of all ages studied were embedded in a common block in 20% sucrose and 7% gelatin solution (300 Bloom; Sigma-Aldrich, St. Louis, MO), frozen, and cryosectioned (20 μm). Sections were mounted on slides (SuperFrost Plus; Fisher Scientific, Pittsburgh, PA) and stored frozen. A subset of eyes had retinas flatmounted for immediate processing.

Immunofluorescence

Whole retinas and sections were simultaneously processed to assure equal treatment and reliable comparison of the staining patterns during development. After thawing of cryosections and rinsing with PBS, autofluorescence was decreased by incubation with 50 mM ammonium chloride in PBS for 15 minutes. After rinsing, cells were permeabilized with 0.25% Triton X-100 in PBS for 5 minutes, and finally the tissue was blocked with 5% normal donkey serum plus 1% bovine serum albumin (BSA) for 1 hour. Specimens were immunolabeled by incubation with one or three antibodies, diluted in 1% BSA, for 18 to 72 hours at 4°C. The primary antibodies used include a monoclonal antibody to bovine calbindin D-28k (1:1000 dilution; Sigma-Aldrich), rabbit polyclonal antibodies to: mouse Prox1 C terminus (1:5000; Covance, Berkely, CA), rat Kir4.1 C terminus (1:50; Alomone, Jerusalem, Israel), rat AQP4 (1:50; AQP41.A; Alpha Diagnostic, San Antonio, TX), and goat polyclonal antibody to human AQP4 C terminus (1:50; C-19; Santa Cruz Biotechnology, Santa Cruz, CA). Single immunodetections indicated the optimal working conditions for each antibody. The specificity of both anti-Kir4.1 and anti-AQP4C19 antibodies was confirmed by preabsorption with the corresponding antigenic peptides (data not shown; Frigeri et al.³⁶ ). Further, the identical patters detected at P0 with the AQP4C19 and AQP41.A antibodies, confirmed their specificity (data not shown), although C19 signal was stronger and therefore was chosen for this study. Background fluorescence was nominal after omitting each primary antibody. Primary antibodies were detected with one or three mixed affinity-purified secondary antibodies raised in donkey against IgG of goat, rabbit, and mouse (Alexa Fluor 594, 488, and 647 nm, respectively; 1:400 dilution; Molecular Probes, Eugene, OR). Secondary antibodies diluted in 1% BSA buffer were incubated for 2 hours at room temperature. After a thorough rinse, nuclei were counterstained with 10 μm 4′,6-diamidino-2-phenylindole (DAPI) in PBS and coverslipped (Fluoromount G; Fisher Scientific).

Photomicrography

Fluorescence microscopy was conducted on a confocal upright microscope (BX50WI, Fluoview 500; Olympus, Melville, NY), using a 40× water-immersion objective, appropriate filter settings, and sequential image acquisition for multilabeled specimens. Fluorescent data from both retinal wholemounts and cryosections were collected at 0.31 μm/pixel, with a z-step ranging from 0.5 to 1 μm, and a 5 to 20 μm depth of field (see details in figure legends). The figures show slices merged into one plane (Fluoview; Olympus) and pseudocolored red, green, or white (representing 594, 488, and 647 nm, respectively). Multiple channel overlay revealed fluorophore colocalization (yellow). Before conversion to CMYK, images were edited for brightness, contrast, and color tone (Photoshop; Adobe Systems, Mountain View, CA), and assembled (Freehand; Macromedia, San Francisco, CA).

Results

Expression of AQP4 and Kir4.1 in Müller Cells in the Adult Retina

Immunolabeling of adult retinas showed AQP4 (Figs. 1A 1B , red) and Kir4.1 coenrichment (Figs. 1A 1C , green) in perivascular compartments within the vitreous margin and inner layers, with an increasing intensity gradient inward from the outer plexiform layer (OPL). Labeling reached peak levels toward the ganglion cell (GCL) and nerve fiber (NFL) layers. Merged AQP4 and Kir4.1 images (Figs. 1A 1E , yellow) revealed colocalization next to the inner retinal surface, along cell processes either radiating perpendicularly from it, or flanking vessels crossing the inner nuclear layer (INL). Furthermore, the separate visualization of AQP4 (Fig. 1B , red) and Kir4.1 labeling (Fig. 1C , green) confirmed their distribution in Müller cell end feet (Figs. 1B 1C , arrow) and processes (arrowhead) crossing the inner plexiform layer (IPL). Within the outer nuclear layer (ONL), Müller cell processes appeared so weakly labeled for both AQP4 and Kir4.1 that they were hardly visible with optical conditions optimized to the intense inner retinal expression (Fig. 1A) . Intense calbindin labeling identified horizontal cell somata in the outer INL and their processes within the OPL (Fig. 1A , white). Also, anti-calbindin faintly labeled amacrine cells within the INL (Fig. 1A)and displaced amacrine and ganglion cells in the GCL (Figs. 1A , white; 1D, asterisk), appearing encased by AQP4/Kir4.1-rich Müller cell end feet (Figs. 1A ; 1E , asterisk). Amacrine, ganglion and horizontal cells showed neither AQP4 nor Kir4.1 immunoreactivity. The vascular lumen showed some background crossreactivity to the anti-mouse secondary antibody used (Fig. 1A , white). Whether astrocytes located in the superficial layers contribute to the AQP4 and/or Kir4.1 labeling in surrounding vessels within the NFL was unclear, given their contiguity with Müller cell end feet.

Expression of AQP4 and Kir4.1 at P15 in Müller Cells

At P15, Müller cells expressed AQP4 (Figs. 1F 1G , red) and Kir4.1 (Figs. 1F 1H , green), both distributed with a manifest outer-to-inner gradient of intensity starting at the OPL. Overall, APQ4/Kir4.1 labeling intensity was lower than observed in adult tissue. AQP4 antibody weakly labeled Müller cell processes (Fig. 1G) , but perivascular end feet (Fig. 1G , arrow) in the IPL (Fig. 1G , arrowhead) and INL (Fig. 1J , arrow) were strongly labeled. In contrast, Kir4.1 labeled Müller cells with robust intensity and notable OPL-to-NFL polarization, closely resembling the adult pattern (Fig. 1F , green). Müller cell end feet (Fig. 1H , arrow), and both radial and lateral processes in the IPL were labeled with antibodies to Kir4.1; however, perivascular areas in the IPL did not express Kir4.1 (Figs. 1G 1H , arrowheads). Despite the relatively lower expression levels detected for AQP4, its colocalization with Kir4.1 in Müller cell end feet was unambiguous (Fig. 1F , yellow). Within the central retina, calbindin-positive horizontal cell somata (Figs. 1F , white; 1I, arrowhead) displayed faint Kir4.1 labeling (Fig. 1J , green, arrowhead), whereas AQP4 was absent. However, AQP4 was detectable in processes from horizontal cells positioned at the far peripheral retina (data not shown). Note that, within the OPL, calbindin-labeled horizontal cell somata and processes (Figs. 1F 1I , white) were contiguous with blood vessels outlined by AQP4-expressing Müller cell processes (Figs. 1F 1I 1J , red). Both AQP4 and Kir4.1 weakly labeled the outer limiting membrane (Fig. 1F) .

Coexpression of AQP4 and Kir4.1 at P9 in Horizontal and Müller Cells

At P9, Kir4.1 labeling showed polarization, with the IPL, GCL, and NFL exhibiting the most intense labeling (Figs. 1K 1M , green). Whereas antibodies to calbindin faintly labeled amacrine (Figs. 1K 1L , white, arrowhead) and horizontal cells (Figs. 1K , white; 1N , arrowheads), anti-Kir4.1 did not specifically label amacrine cells (compare Fig. 1Lwith 1M , arrowheads), but weakly labeled horizontal cells (Figs. 1K ; 1N 1O , arrowheads). Müller cell processes spanning the INL clearly expressed Kir4.1 (Fig. 1K , green). The identity of the Kir4.1-labeled processes within the IPL is unclear. Conversely, AQP4 expression was restricted to the NFL (Figs. 1K 1M , red), IPL perivascular areas (Fig. 1M , arrow), and horizontal cell processes in the OPL (Figs. 1K 1O) . Blood vessels near the OPL nonspecifically labeled with the anti-mouse IgG antibody used against calbindin (Fig. 1N , arrows), but were not surrounded by AQP4 labeling (Fig. 1O , arrows), although it is not clear if the Müller cell processes have not developed yet or simply fail to express AQP4. Some AQP4 expression was detectable within the NFL and IPL, whereas AQP4 and Kir4.1 showed little overlap in Müller cells at the inner retina.

Horizontal Cell–Specific Coexpression of AQP4 and Kir4.1 at P0

In the newborn retina, AQP4 (Figs. 1P 1Q , red) and Kir4.1 (Figs. 1P 1R , green) proteins colocalized to cells within the outer neuroblastic layer (NBL). Moreover, the detection of calbindin (Fig. 1S , white) within these AQP4/Kir4.1-positive cells, together with their morphology and position identified them as presumptive horizontal cells. AQP4 staining was highly restricted to horizontal cell somata and processes (Figs. 1P 1Q , red), and Kir4.1 (Fig. 1P 1R , green) was coexpressed in most AQP4-labeled cells. The subcellular distribution of AQP4 and Kir4.1 was highly coincident (Fig. 1T , yellow), except for process endings labeled more intensely for AQP4 than for Kir4.1. Also, some Kir4.1 labeling was observed in the NFL and GCL (Fig. 1P , green), along with a calbindin-positive plexus (Fig. 1P , white), only overlapping at the vitreous margin. Although the identity of the cells coexpressing Kir4.1/calbindin is unknown, their confinement to the NFL suggests they could be ganglion cell axons. The strong IPL calbindin labeling most likely represents developing amacrine cell processes. AQP4/Kir4.1-colabeling revealed the diverse morphologies of horizontal cells corresponding to their eccentric position and maturation level (Fig. 1P , central-to-peripheral, right-to-left). Accordingly, more mature, multipolar cells (Figs. 1P ; 1Q 1R 1S , arrowheads) were nearly aligned toward central locations, whereas less-differentiated horizontal cells, showing radial, bipolar silhouettes, remained scattered at far peripheral regions (Figs. 1Q 1R 1S , arrows).

To confirm further the identity of the AQP4/Kir4.1/calbindin-expressing cells, we coimmunostained neonatal retinas for Prox1 and AQP4. In the outer NBL, round nuclei within AQP4-positive cells (red) showed Prox1 labeling (Fig. 2white and arrows in the NBL). The soma shape, processes and position of these Prox1/AQP4-coexpressing cells corroborated their identification as horizontal cells. Within the inner NBL, adjacent to the IPL, Prox1 labeled elongated nuclei from AQP4-negative cells (Fig. 2 , arrowheads), presumably representing amacrine or bipolar cells. Unlike Prox1 and calbindin, which recognize multiple neuronal types at birth (i.e., amacrine and horizontal cells), AQP4 specifically labeled horizontal cells, regardless of their degree of maturation (Fig. 2 , retinal edge at left). AQP4 punctate labeling outlined horizontal cell inward projections reaching the GCL (Fig. 2 , asterisks), as well as outward processes oriented toward the subretinal space. Of interest, some cells within the GCL displayed cytoplasmic rather than nuclear Prox1 expression, though they lacked AQP4 labeling (Fig. 2 , arrows in the GCL).

Furthermore, we studied AQP4 distribution in retinal wholemount preparations. The horizontal cells within the central retina (Fig. 3)revealed remarkably homogeneous levels of AQP4 expression (Fig. 3A , red) throughout the neurofilament-stained cell mosaic (Fig. 3B , green). The localization of AQP4 to horizontal cell somata was evident, whereas AQP4 content at the neurofilament-positive distal tips of processes was weak (Fig. 3A 3B , arrows; 3C ), suggesting some subcellular polarization of AQP4 in horizontal cells.

Discussion

The key finding of this study is that two glial proteins, Kir4.1 and AQP4, essential for efficient extracellular K⁺ and water clearance within the CNS, are specifically coexpressed by retinal neurons, namely horizontal cells, during early development. Furthermore, we report a remarkable switch of AQP4 and Kir4.1 localization from neurons to glia coincident with eye opening. The timing of this switch to glial cells matches the reported upregulation of inward-rectifying currents in differentiating Müller cells³⁷and confirms the postnatal expression of Kir4.1 protein by Müller glia.¹⁸As such, AQP4 and Kir4.1 proteins are enriched in Müller cells by P15, whereas Kir4.1-expression was reduced in horizontal cells, and only less differentiated cells expressed AQP4 within their processes. The adult patterns of AQP4/Kir4.1 expression detected in the current study reproduced the asymmetrical patterns reported for adult mouse and rat retina.¹³ ¹⁸ ¹⁹ ²⁰ ²⁵Unlike these previous reports, however, neither AQP4 nor Kir4.1 was detected in the OLM. This discrepancy may be attributable to differences in specimen preparation. This is the first report of neuronal AQP4 expression, since AQP4 expression within the CNS is restricted to glia, such as astrocytes, ependymal, and supporting cells.²³ ²⁴Likewise, Kir4.1 expression is largely limited to astrocyte subpopulations, oligodendrocytes, Bergman glia, and satellite cells.¹⁵ ³⁸ ³⁹ ⁴⁰ ⁴¹However, Kir4.1 mRNA has also been detected within brain stem⁴²and cortical neurons.⁴³Neither AQP4 nor Kir4.1 proteins were detected in other retinal neurons. The absence of Kir4.1 expression in retinal ganglion cells at adulthood matches the findings of Chen et al.⁴⁴who detected multiple Kir channel subtypes, except for Kir4.1.

Calbindin is detectable in presumptive amacrine and horizontal cells in mouse retina at E18.5.⁴⁵In adult retina, horizontal cell somata and processes display robust calbindin immunoreactivity, whereas ganglion and amacrine cells show variable expression.⁴⁶ The Prox1 homeodomain protein is detectable in mouse retinal progenitor cell nuclei between E12.5 and P0, coincident with amacrine and horizontal cell genesis.³⁰ ⁴⁷As development continues, Prox1 expression increases in these interneuron’s nuclei, and in the adult, Prox1 expression remains intense in horizontal and AII amacrine cells, but faint in bipolar cells.⁴⁷Given that the time courses of amacrine and horizontal cell genesis overlap³⁰and that both calbindin and Prox1 are expressed in both cell types, these markers cannot unequivocally identify horizontal cells during retinogenesis. Thus, AQP4 particularly, and to a lesser extent Kir4.1, specifically identify horizontal cells in the newborn mouse retina. AQP4 is apparently expressed in the entire horizontal cell mosaic, which in mouse is formed by a single population of axon-bearing cells.⁴⁸

The transient Kir4.1/AQP4 coexpression by horizontal cells suggests that these channels may be functional during development. Immature horizontal cells may mediate K⁺ spatial buffering and/or osmoregulation during retinogenesis, before Müller cells and astrocytes assume these roles at adulthood. Whereas horizontal cells are generated by embryonic day (E)10 to E12³⁰as a stable population lacking cell overproduction and death,⁴⁹ ⁵⁰at birth, when we detected AQP4/Kir4.1 expression, few or no glia are present in the mouse retina. Müller cells, native to the retina, are generated last during retinogenesis, at P2 to P4,²⁹ ³⁰whereas astrocytes invade the vitreal surface during the first ten postnatal days.⁵¹Hence, the perinatal mouse retina is avascular, since the primary plexus forms underneath the NFL after astrocyte migration.³¹A deep plexus grows within the OPL by P15, and interconnects to the vitreal plexus by P21.³² ³³The OPL capillaries (Fig. 1F , red) emerge in striking proximity to horizontal cell somata (Fig. 1F , white), suggesting an interaction between neurons and vasculature during development. Interestingly, horizontal cells directly associate with capillaries in the adult tree shrew retina.⁵² ⁵³

By birth, when horizontal cell AQP4/Kir4.1 coexpression is intense and cell specific, these relatively scarce interneurons are distributed at all eccentricities, showing elongated somata mostly aligned in the outermost level of the NBL.⁵⁴Horizontal cell AQP4–content exposes their radial shape, with processes spanning the NBL (Figs. 1P 1T 2A 2B) . Although horizontal and Müller cell densities differ, both cell types share a radial morphology, at least during horizontal cell differentiation. After P9, the aligned somata of horizontal cells project almost coplanar processes toward the OPL (Fig. 1K) . The array of horizontal cells and nascent capillaries resembles the known association between vasculature and Müller cells, which optimizes K⁺ and water transglial transport. The remaining expression of Kir4.1 within horizontal cell somata and of AQP4 in their least differentiated processes raises the novel and intriguing possibility of a developmental neuron–vascular metabolic communication, contemporaneous with the occurrence of spontaneous retinal activity,²⁸later supplanted by gliovascular interactions in the adult retina. It will be interesting to determine whether AQP4 or Kir4.1 subcellular distribution in developing horizontal cells shows any asymmetry as angiogenesis within the OPL progresses, which may indicate a transcellular pattern of water and K⁺ redistribution. The possibility that differentiating horizontal cells can fulfill an archetypal glial function contributes to the enigmatic nature of these retinal interneurons.

Supported by Grant MH65495 from the National Institute of Mental Health, National Institute of Neurological Disorders and Stroke Grant NS41282 (DCS), and National Heart, Lung, and Blood Institute Grant HL07675 (KC).

Submitted for publication March 25, 2005; revised May 10, 2005; accepted July 25, 2005.

Disclosure: A. Bosco, None; K. Cusato, None; G.P. Nicchia, None; A. Frigeri, None; D.C. Spray, None

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

Corresponding author: Alejandra Bosco, Albert Einstein College of Medicine, Department of Neuroscience, Rose F. Kennedy Center, 1300 Morris Park Avenue, Room 712, Bronx, NY 10461; [email protected].

Figure 1.

View Original Download Slide

During retinal development, AQP4 and Kir4.1 protein expression coordinately switches from neuronal to glial cells. Temporal expression and distribution of AQP4 (red), Kir4.1 (green), and calbindin (white) in triple-immunolabeled radial sections of mouse retina. (A) At adulthood, both AQP4 and Kir4.1 expression showed a conspicuous polarization along Müller glial cells, with an intensity gradient increasing from the OPL toward the NFL. At the vitreous margin, Müller cell compartments adjacent to blood vessels and to the vitreous body exhibited the highest AQP4/Kir4.1 colocalization (yellow). The INL contained calbindin-positive amacrine cells and horizontal cells, lining the IPL and OPL, respectively, which did not colabel with AQP4 or Kir4.1. (B–E) Detail of the vitreous margin (A, box), displaying two channels, individually and merged. (B) AQP4 showed a continuous distribution along Müller cell end feet (arrow) and punctate labeling at the inner processes (arrowhead). (C) Kir4.1 staining clearly exposed Müller glia end feet (arrow) within the GCL, as well as radial processes (arrowhead). (D) Calbindin stained both ganglion cells ( Image not available

) and displaced amacrine cells in the GCL. (E) The overlay of AQP4 and Kir4.1 labeling demonstrated strong colocalization in Müller cell processes encasing amacrine and ganglion cells and confirmed their absence in ganglion cells ( Image not available

). (F) At P15, the Müller cell population displayed patterns of AQP4 and Kir4.1 expression similar to mature retina. Kir4.1 strongly labeled the inner retina, prominently colocalizing with AQP4 in the Müller cell end feet (yellow). Outlining the OPL, calbindin-positive horizontal cells alternated with blood vessels wrapped in AQP4-labeled Müller cell processes. (G, H) Detailed view of the INL as separate channels (F, right box). (G) AQP4 weakly labeled the Müller cell inner processes (arrow) and perikarya (arrowhead). (H) Kir4.1 labeling more clearly delineated the same Müller cell compartments. (I, J) High-magnification view of separate and merged images of an OPL area (F, left box). (I) Horizontal cell somata and processes stained with antibodies to calbindin. (J) Intense AQP4 labeling was concentrated in perivascular processes attributable to Müller cells (I, J, arrows), given the absence of calbindin staining. Horizontal cells, intercalated between deep vessels, maintain a subtle Kir4.1 expression (arrowheads), but no AQP4 labeling. (K) At P9, Kir4.1 expression coincides in horizontal and Müller cells (green). (L, M) Inner retinal region viewed as separate channels (K, white box). (L) Calbindin weakly labels amacrine cells (arrowhead) and the IPL (arrow). (M) Both AQP4 (red) and Kir4.1 (green) expression lack an intense colocalization. AQP4 is enriched around the nascent vessels (arrow). (N, O) Detailed view of outer INL and OPL (K, blue box). (N) Calbindin labels horizontal cells (arrowheads) and the secondary antibodies nonspecifically stain the vasculature (arrow). (O) Kir4.1 (green) lightly labeled the horizontal cell somata (N, arrowheads), whereas AQP4 intensely labeled horizontal cell processes in the OPL. At this stage, AQP4 did not label the perivascular regions in the OPL (arrows). (P) At birth, AQP4 and Kir4.1 were exclusively coexpressed in horizontal cells, which in far peripheral regions were still undergoing interkinetic migration. (Q–T) Detailed outer NBL (P, box) viewed as single or merged images. (Q) AQP4 labeled both horizontal cell somata and processes. (R) Although, Kir4.1 distribution greatly matched AQP4 labeling, it did not label intensely all AQP4-positive horizontal cells (arrow). (S) Calbindin faintly labeled the differentiating horizontal cells. (T) The merged view of AQP4 and Kir4.1 staining denote their intense colocalization to horizontal cells (yellow). Projection of optical stacks with the following depth of field (in micrometers): 23 (A–E), 7 (F, J), and 5.5 (K–T). Scale bars: (A, F, K, P) 20 μm; (B–E, G–J, L–O, Q–T) 10 μm.

Figure 1.

View Original Download Slide

Figure 2.

View Original Download Slide

AQP4 is confined to horizontal cells in the newborn (P0) mouse retina. (A, B) Prox1 and AQP4 double-immunolabeled radial section of neural retina, with nuclei counterstained with DAPI. (A) The overlay of DAPI (blue), Prox1 (white), and AQP4 (red) labelings identified nuclei in both AQP4-positive (arrows) and negative cells (arrowheads) in the NBL. Some cells in the GCL contained cytoplasmic Prox1 labeling (arrow). (B) The merged view of Prox1 (white) and AQP4 (red) stainings shows Prox1-labeled cells positioned at the GCL (displaying cytoplasmic expression, arrow) and both within the inner and outer NBL (arrows, arrowheads). AQP4 showed a compact distribution along horizontal cell processes and somata, regardless of their radial position, always coexpressed with nuclear Prox1. Also, AQP4 as discrete puncta (asterisk), was detected in the GCL, often connected to horizontal cell processes, outlined by radial arrangements of AQP4-puncta within the INL. Images represent the projection of a 9-μm-thick optical stack. Abbreviations as in Figure 1 . Scale bar, 20 μm.

Figure 3.

View Original Download Slide

At birth (P0), AQP4 was uniformly expressed throughout the horizontal cell mosaic. (A–C) Tangential view of the outer NBL of a wholemount retina, double immunostained with AQP4 (red) and neurofilament (NF, green) antibodies. (A) AQP4 localized to both horizontal cell somata and processes. (B) Neurofilament stained the horizontal cell cytoskeleton, showing primary and even higher-order processes. (C) The merged image of AQP4 and neurofilament yields a yellow, signal showing that both markers coincide in nearly all horizontal cell processes, except the distal tips (arrows). Projection of an 8.5-μm-thick optical stack. Scale bar, 10 μm.

The authors thank Amanda Lee, Marcia U. Maldonado, and Joseph Zavilowitz for excellent technical assistance and Tiffany Cook (Cincinnati Children’s Hospital) for valuable comments on the manuscript.

OrkandRK, NichollsJG, KufflerSW. Effect of nerve impulses on the membrane potential of glial cells in the central nervous system of amphibia. J Neurophysiol. 1966;29:788–806. [PubMed]

NewmanEA, FrambachDA, OdetteLL. Control of extracellular potassium levels by retinal glial cell K⁺ siphoning. Science. 1984;225:1174–1175. [CrossRef] [PubMed]

WalzW. Role of astrocytes in the clearance of excess extracellular potassium. Neurochem Int. 2000;36:291–300. [CrossRef] [PubMed]

ScemesE, SprayDC. The astrocytic syncytium. Adv Mol Cell Biol. 2004;31:165–179.

KofujiP, NewmanEA. Potassium buffering in the central nervous system. Neurosci. 2004;129:1045–1056.

HasegawaH, MaT, SkachW, MatthayMA, VerkmanAS. Molecular cloning of a mercurial-insensitive water channel expressed in selected water-transporting tissues. J Biol Chem. 1994;269:5497–5500. [PubMed]

NielsenS, NagelhusEA, Amiry-MoghaddamM, BourqueC, AgreP, OttersenOP. Specialized membrane domains for water transport in glial cells: high-resolution immunogold cytochemistry of aquaporin-4 in rat brain. J Neurosci. 1997;17:171–180. [PubMed]

RashJE, YasumuraT. Direct immunogold labeling of connexins and aquaporin-4 in freeze-fracture replicas of liver, brain, and spinal cord: factors limiting quantitative analysis. Cell Tissue Res. 1999;296:307–321. [CrossRef] [PubMed]

ConnorsNC, AdamsME, FroehnerSC, KofujiP. The potassium channel Kir4.1 associates with the dystrophin-glycoprotein complex via alpha-syntrophin in glia. J Biol Chem. 2004;279:28387–28392. [CrossRef] [PubMed]

NagelhusEA, MathiisenTM, OtterseenOP. Aquaporin-4 in the central nervous system cellular and subcellular distribution and coexpression with Kir4.1. Neurosci. 2004;129:905–913. [CrossRef]

Amiry-MoghaddamM, WilliamsonA, PalombaM, et al. Delayed K⁺ clearance associated with aquaporin-4 mislocalization: phenotypic defects in brains of alpha-syntrophin-null mice. Proc Natl Acad Sci USA. 2003;100:13615–13620. [CrossRef] [PubMed]

ZahsKR, WuT. Confocal microscopic study of glial-vascular relationships in the retinas of pigmented rats. J Comp Neurol. 2001;429:253–269. [CrossRef] [PubMed]

NagelhusEA, HorioY, InanobeA, et al. Immunogold evidence suggests that coupling of K⁺ siphoning and water transport in rat retinal Müller cells is mediated by a coenrichment of Kir4.1 and AQP4 in specific membrane domains. Glia. 1999;26:47–54. [CrossRef] [PubMed]

TianM, ChenL, XieJX, YangXL, ZhaoJW. Expression patterns of inwardly rectifying potassium channel subunits in rat retina. Neurosci Lett. 2003;345:9–12. [CrossRef] [PubMed]

KalsiAS, GreenwoodK, WilkinG, ButtAM. Kir4.1 expression by astrocytes and oligodendrocytes in CNS white matter: a developmental study in the rat optic nerve. J Anat. 2004;204:475–485. [CrossRef] [PubMed]

NewmanEA, ReichenbachA. The Müller cell: a functional element of the retina. Trends Neurosci. 1996;19:307–312. [CrossRef] [PubMed]

NewmanEA. Regulation of potassium levels by glial cells in the retina. Trends Neurosci. 1985;8:156–159. [CrossRef]

KofujiP, BiedermannB, SiddharthanV, et al. Kir potassium channel subunit expression in retinal glial cells: implications for spatial potassium buffering. Glia. 2002;39:292–303. [CrossRef] [PubMed]

IshiiM, HorioY, TadaY, et al. Expression and clustered distribution of an inwardly rectifying potassium channel, KAB-2/Kir4.1, on mammalian retinal Müller cell membrane: their regulation by insulin and laminin signals. J Neurosci. 1997;17:7725–7735. [PubMed]

IshiiM, FujitaA, IwaiK, et al. Differential expression and distribution of Kir5.1 and Kir4.1 inwardly rectifying K⁺ channels in retina. Am J Physiol. 2003;285:C260–C267. [CrossRef]

KofujiP, CeelenO, ZahsKR, SurbeckLW, LesterHA, NewmanEA. Genetic inactivation of an inwardly potassium channel (Kir4.1 subunit) in mice: phenotypic impact in the retina. J Neurosci. 2000;21:15:5733–5740.

JungJS, BhatRV, PrestonGM, GugginoWB, BarabanJM, AgreP. Molecular characterization of an aquaporin cDNA from brain: candidate osmoreceptor and regulator of water balance. Proc Natl Acad Sci USA. 1994;91:13052–13056. [CrossRef] [PubMed]

FrigeriA, GropperMA, TurckCW, VerkmanAS. Immunolocalization of the mercurial-insensitive water channel and glycerol intrinsic protein in epithelial cell plasma membranes. Proc Natl Acad Sci USA. 1995;92:4328–4331. [CrossRef] [PubMed]

FrigeriA, GropperMA, UmenishiF, KawashimaM, BrownD, VerkmanAS. Localization of MIWC and GLIP water channel homologs in neuromuscular, epithelial and glandular tissues. J Cell Sci. 1995;108:2993–3002. [PubMed]

NagelhusEA, VerukiML, TorpR, et al. Aquaporin-4 water channel protein in the rat retina and optic nerve: polarized expression in Müller cells and fibrous astrocytes. J Neurosci. 1998;18:2506–2519. [PubMed]

PannickeT, IandievI, UckermannO, et al. A potassium channel-linked mechanism of glial cell swelling in the postischemic retina. Mol Cell Neurosci. 2004;26:493–502. [CrossRef] [PubMed]

GalliL, MaffeiL. Spontaneous impulse activity of rat retinal ganglion cells in prenatal life. Science. 1988;242:90–91. [CrossRef] [PubMed]

BansalA, SingerJH, HwangBJ, XuW, BeaudetA, FellerMB. Mice lacking specific nicotinic acetylcholine receptor subunits exhibit dramatically altered spontaneous activity patterns and reveal a limited role for retinal waves in forming ON and OFF circuits in the inner retina. J Neurosci. 2000;20:7672–7681. [PubMed]

BlanksJC, BokD. An autoradiographic analysis of postnatal cell proliferation in the normal and degenerative mouse retina. J Comp Neurol. 1977;174:317–327. [CrossRef] [PubMed]

RapaportDH, WongLL, WoodED, YasumuraD, LaVailMM. Timing and topography of cell genesis in the rat retina. J Comp Neurol. 2004;474:304–324. [CrossRef] [PubMed]

LingTL, MitrofanisJ, StoneJ. Origin of retinal astrocytes in the rat: evidence of migration from the optic nerve. J Comp Neurol. 1989;286:345–352. [CrossRef] [PubMed]

DorrellMI, AguilarE, FriedlanderM. Retinal vascular development is mediated by endothelial filopodia, a preexisting astrocytic template and specific R-cadherin adhesion. Invest Ophthalmol Vis Sci. 2002;43:3500–3510. [PubMed]

FruttigerF. Development of the mouse retinal vasculature: angiogenesis versus vasculogenesis. Invest Ophthalmol Vis Sci. 2002;43:522–527. [PubMed]

BoscoA, CusatoK, NicchiaGP, FrigeriA, SprayDC. Glial markers are expressed in neurons: unexpected presence of glial proteins in differentiating retinal horizontal cells from the mouse retina. Soc Neurosci. 2004.Abstract 268.15.35

BoscoA, CusatoK, NicchiaGP, FrigeriA, SprayDC. Neuronal expression of aquaporin-4: switch from horizontal cells to Müller glia during retinal development. American Soc Cell Biol. 2003.Abstract 1242

FrigeriA, NicchiaGP, BalenaR, NicoB, SveltoM. Aquaporins in skeletal muscle: reassessment of the functional role of aquaporin-4. FASEB J. 2004;18:905–907. [PubMed]

PannickeT, BringmannA, ReichenbachA. Electrophysiological characterization of retinal Müller glial cells from mouse during postnatal development: comparison with rabbit cells. Glia. 2002;38:268–272. [CrossRef] [PubMed]

TakumiT, IshiiT, HorioY, et al. A novel ATP-dependent inward rectifier potassium channel expressed predominantly in glial cells. J Biol Chem. 1995;270:16339–16346. [CrossRef] [PubMed]

HibinoH, HorioY, FujitaA, et al. Expression of an inwardly rectifying K⁺ channel, Kir4.1, in satellite cells of rat cochlear ganglia. Am J Physiol. 1999;277:C638–C644. [PubMed]

PoopalasundaramS, KnottC, ShamotienkoOG, et al. Glial heterogeneity in expression of the inwardly rectifying K⁺ channel, Kir4.1, in adult rat CNS. Glia. 2000;30:362–372. [CrossRef] [PubMed]

HigashiK, FujitaA, InanobeA, et al. An inwardly rectifying K(+) channel, Kir4.1, expressed in astrocytes surrounds synapses and blood vessels in brain. Am J Physiol. 2001;281:C922–C931.

WuJ, XuH, ShenW, JiangC. Expression and coexpression of CO₂-sensitive Kir channels in brainstem neurons of rats. J Membr Biol. 2004;197:179–191. [CrossRef] [PubMed]

LiL, HeadV, TimpeLC. Identification of an inward rectifier potassium channel gene expressed in mouse cortical astrocytes. Glia. 2001;33:57–71. [CrossRef] [PubMed]

ChenL, YuYC, ZhaoJW, YangXL. Inwardly rectifying potassium channels in rat retinal ganglion cells. Eur J Neurosci. 2004;20:956–964. [CrossRef] [PubMed]

SharmaRK, O’LearyTE, FieldsCM, JohnsonDA. Development of the outer retina in the mouse. Dev Brain Res. 2003;145:93–105. [CrossRef]

HaverkampS, WässleH. Immunocytochemical analysis of the mouse retina. J Comp Neurol. 2000;424:1–23. [CrossRef] [PubMed]

DyerMA, LiveseyFJ, CepkoCL, OliverG. Prox1 function controls progenitor cell proliferation and horizontal cell genesis in the mammalian retina. Nat Genetics. 2003;34:53–58. [CrossRef]

PeichlL, González-SorianoJ. Morphological types of horizontal cell in rodent retinae: a comparison of rat, mouse, gerbil, and guinea pig. Vis Neurosci. 1994;11:501–517. [CrossRef] [PubMed]

YoungRW. Cell death during differentiation of the retina in the mouse. J Comp Neurol. 1984;229:362–373. [CrossRef] [PubMed]

KarlssonM, MayordomoR, ReichardtLF, CatsicasS, KartenH, HallböökF. Nerve growth factor is expressed by postmitotic avian retinal horizontal cells and supports their survival during development in an autocrine mode of action. Development. 2001;128:471–479. [PubMed]

WatanabeT, RaffMC. Retinal astrocytes are immigrants from the optic nerve. Nature. 1988;332:834–837. [CrossRef] [PubMed]

KnabeW, KuhnHJ. Capillary-contacting horizontal cells in the retina of the tree shrew Tupaia belangeri belong to the mammalian type A. Cell Tissue Res. 2000;299:307–311. [PubMed]

OchsM, MayhewTM, KnabeW. To what extent are the retinal capillaries ensheathed by Müller cells?—a stereological study in the tree shrew Tupaia belangeri. J Anat. 2000;196:453–461. [CrossRef] [PubMed]

ReeseBE, NecessaryBD, TamPPL, Faulkner-JonesB, TanSS. Clonal expansion and cell dispersion in the developing mouse retina. Eur J Neurosci. 1999;11:2965–2978. [CrossRef] [PubMed]

Figure 1.

View Original Download Slide