Immunolocalization of Gap Junction Protein Connexin43 (GJA1) in the Human Retina and Optic Nerve

Abstract

Purpose.: Gap junctions are intercellular channels that have been implicated in the pathogenesis of neuronal death after central nervous system injury. This study determines the expression pattern of gap junction protein connexin43 in the human retina and optic nerve.

Methods.: An affinity-isolated polyclonal antibody to the C-terminal segment of the cytoplasmic domain of human connexin43 was used to determine connexin43 localization. Postmortem human eyes were examined by immunohistochemical staining of frozen sections using antibodies to connexin43. Antibody binding was detected using confocal microscopy and fluorochrome-conjugated secondary antibodies. Double-label immunohistochemistry identified the cell types expressing connexin43.

Results.: Connexin43 immunoreactivity was detected in the human retina on glial fibrillary acidic protein (GFAP)-positive astrocytes in the retinal ganglion cell layer and, to a lesser extent, on the processes of glutamine synthetase-labeled Müller cells. The retinal and choroidal circulations showed strong connexin43 immunolabeling. Dense connexin43 immunoreactivity was present between adjacent cells of the retinal pigment epithelium, and there was diffuse connexin43 immunoreactivity on GFAP-positive astrocytes in the optic nerve.

Conclusions.: In the human retina and optic nerve, connexin43 is present on glia, blood vessels, and epithelial cells. An understanding of the distribution of connexin43 in the normal retina and optic nerve may be used to evaluate changes associated with retinal and optic nerve disease.

Gap junctions are intercellular conduits that permit direct intercellular communication through the bidirectional movement of ions, metabolites, and second messengers between neighboring cells. ¹ These channels play an important role in maintaining local metabolic homeostasis and synchronizing cellular activity. ² Each gap junction is formed by the docking of two hemichannels embedded in the plasma membranes of adjacent cells. ³ These hemichannels, or connexons, are formed by the oligomerization of six protein subunits termed connexins and provide a low-resistance pathway for current flow and for the exchange of small cytoplasmic molecules up to 1 kDa. ⁴ To date, 21 connexin genes have been identified in the human genome encoding proteins ranging in predicted molecular mass from 23 to 62 kDa. ⁵ Connexin proteins are designated according to their estimated molecular mass in kilodaltons; alternatively, they may be divided into at least four subgroups based on sequence homology and serially numbered according to order of discovery. ⁵ For example, connexin43 (Cx43) has a theoretical molecular mass of 43 kDa and was the first connexin of the α-group (GJA1). ⁵  

Connexin43 is the most abundant gap junction protein in the central nervous system and is expressed primarily on astrocyte processes surrounding blood vessels and chemical synapses. ¹ The resultant functional syncytium is thought to be important in the maintenance of local homeostasis by buffering and redistributing excess extracellular K⁺ through the cytoplasm of a vast network of contiguous astrocytes coupled by gap junctions. ⁶  

There is increasing interest in the role of gap junction communication in cell death. ^7–9 Evidence is accumulating that after injury to the central nervous system, open gap junctions may allow the passage of apoptotic and necrotic signals to adjacent cells, thereby increasing cell death. ^10–18 Gap junction modulation has been identified as a potential neuroprotective target, and recently it has been shown that gap junction inhibitors protect retinal cells from ischemia in a cell culture model. ¹⁹  

Connexin43 expression has been found at the mRNA and protein levels in rat, mouse, rabbit, carp, and zebrafish retina; fish and rat retinal pigment epithelium and vasculature; and rat nerve fiber layer. ²⁰ However, to date there has been no report on the distribution of connexin43 expression in the human retina or optic nerve; such information is necessary to understand its role in health and disease. In the present study, we used an immunohistochemical approach involving polyclonal antibodies to localize connexin43 immunoreactivity in the human retina and optic nerve. 

Materials and Methods

Human tissue consisted of frozen sections obtained from eyes donated to the New Zealand National Eye Bank. Six eyes from donors aged 29 to 83 (Table 1) were fixed in 1% paraformaldehyde for 30 minutes, washed with phosphate-buffered saline, and cryoprotected in an increasing sucrose gradient. The tissue was subsequently embedded in optimal cutting temperature compound before it was rapidly frozen by immersion in liquid nitrogen. Sagittal 16-μm cryosections were cut and thaw-mounted on slides (SuperFrost; Menzel-Gläser, Braunschweig, Germany). Slide-mounted retinas were rinsed in phosphate-buffered saline and blocked in 6% normal goat serum, 1% bovine serum albumin, 0.4% Triton X-100, and 0.05% thimerosal in phosphate-buffered saline for 1 hour at room temperature. Sections were subsequently incubated with the primary antibodies overnight at 4°C. A list of primary antibodies used in this study is shown in Table 2. All antibodies were obtained from commercial sources and were diluted in phosphate-buffered saline. Negative controls were included by omitting the primary antibody; these did not yield any staining patterns. Immunoreactivity was visualized using either goat anti-rabbit or goat anti-mouse IgG secondary antibodies conjugated to either Alexa-488 or Cy3, except for glial fibrillary acidic protein (GFAP), which was preconjugated to Cy3. Sections were washed and coverslipped using antifade reagent (Prolong Gold; Invitrogen, Eugene, OR) with 4′,6-diamidino-2-phenylindole and sealed with nail polish. Slides were viewed with a confocal microscope (FV1000; Olympus, Tokyo, Japan). Informed consent was obtained from donors' relatives, and the tenets of the Declaration of Helsinki were upheld. 

Table 1.

 

Characterization of Ocular Specimens

Donor	Age (y)	Sex	Ethnicity	Cause of Death	Ocular Disease
1	29	Male	Caucasian	MI	None
2	29	Male	Caucasian	Drowning	None
3	36	Female	Caucasian	MVA	None
4	68	Male	Caucasian	Pneumonia	None
5	72	Female	Caucasian	MI	None
6	83	Male	Caucasian	MI	None

 MI, myocardial infarction; MVA, motor vehicle accident.

Table 2.

 

Primary Antibodies Used in This Study

Antibody	Host	Source	Dilution
Calbindin D28K	Mouse	Sigma (C9848)	1:1000
Calretinin	Mouse	BD Biosciences (610908)	1:1000
Connexin43	Rabbit	Sigma (C6219)	1:2000
GFAP-Cy3	Mouse	Sigma (C9205)	1:1000
Glutamine synthetase	Mouse	BD Biosciences (610518)	1:3000
Islet-1	Mouse	Hybridoma Bank	1:500
Oligodendrocyte-specific protein	Rabbit	Chemicon (ab7474)	1:500
Parvalbumin	Mouse	Sigma (P3088)	1:500
SMI32	Mouse	Sternberger Monoclonals (SMI32)	1:1000
von Willebrand factor	Mouse	Novocastra (NCL-vWF)	1:100

 Sigma: Sigma-Aldrich, St. Louis, MO; BD Biosciences: BD Biosciences, San Jose, CA; Hybridoma Bank: Developmental Studies Hybridoma Bank at the University of Iowa, Iowa City, IA (http://dshb.biology.uiowa.edu/); Chemicon: Chemicon International Inc., Billerica, MA; Sternberger Monoclonals: Sternberger Monoclonals Inc., Baltimore, MD; Novocastra: Novocastra Laboratories Ltd., Newcastle upon Tyne, United Kingdom.

Results

General Observations

Connexin43 immunoreactivity was detected in several locations in the human retina and optic nerve (Table 3). In the retina, strong labeling was detected in the retinal ganglion cell layer, and moderate labeling was present in the inner nuclear and plexiform layers. Beneath the neurosensory retina there was strong labeling in the retinal pigment epithelium and choroid. In the optic nerve there was abundant connexin43 immunoreactivity. 

Table 3.

 

Connexin43 Immunoreactivity in the Human Retina and Optic Nerve

Donor	Age (y)	NFL	GCL	IPL	INL				OPL	ONL	Ph	RPE	Ch	ON
					Amacrine Cells	Bipolar Cells	Horizontal Cells	Müller Cells
1	29	++	++	+	−	−	−	+	+	+	+	++	++	+++
2	29	++	++	+	−	−	−	+	−	−	+	+	+++	+++
3	36	+++	+++	−	−	−	−	+	−	+	++	++	++	+++
4	68	+++	+++	−	−	−	−	+	+	+	++	++	+++	+++
5	72	+++	+++	+	−	−	−	+	+	++	++	++	+++	+++
6	83	+++	+++	+	−	−	−	+	+	+	+	+	+++	+++

 NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; Ph, photoreceptor layer; RPE, retinal pigment epithelium; Ch, choroid; ON, optic nerve. −, no labeling; +, weak labeling; ++, moderate labeling; +++, strong labeling.

Association of Connexin43 Immunoreactivity with Retinal Glia

To determine the relationship between the expression of connexin43 and glial cells in the human retina, cryosections were double-labeled with connexin43 and GFAP, glutamine synthetase, or oligodendrocyte specific protein (Fig. 1). The latter three are markers for astrocytes, Müller cells, and oligodendrocytes, respectively. In the retinal ganglion cell layer, connexin43 immunoreactive puncta were located almost exclusively along GFAP-positive astrocyte processes (Fig. 1A). Where GFAP-positive astrocyte processes extended into the inner plexiform layer, these also colocalized with connexin43 immunoreactive puncta (Fig. 1B). Additionally, connexin43 immunoreactivity was detected on Müller cells labeled with glutamine synthetase (Fig. 1C). These specialized radial glial cells span the entire thickness of the retina from the inner to the outer limiting membrane. Müller cells labeled with connexin43 in the synaptic plexiform layers and both the inner and outer nuclear layers (Fig. 1D). However, in the retinal ganglion cell layer, it was difficult to determine whether connexin43 immunoreactivity was present on Müller cells or retinal astrocytes. No oligodendrocyte-specific, protein-labeled cells were detected in the retinas of any of the subjects examined (not shown). 

Figure 1.

 

Double-labeled immunofluorescence patterns of connexin43 with glial cell markers in the human retina. (A) Connexin43 immunofluorescence was detected in the ganglion cell layer on GFAP-positive astrocyte processes. (B) Retinal astrocyte process extending through the inner plexiform layer colocalizing with connexin43 immunoreactive puncta (arrowhead). (C, D) Faint connexin43 immunoreactive puncta were observed in association with GS-labeled Müller cell processes in the synaptic plexiform layers and in both the inner and outer nuclear layers (arrowheads). Note the perinuclear connexin43 immunoreactivity in a subpopulation of nuclei in the ganglion cell and inner nuclear layers (white arrows). GFAP, glial fibrillary acidic protein; GS, glutamine synthetase; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bar, 20 μm.

Association of Connexin43 Immunoreactivity with Vascular Elements

Blood vessels were visualized by incubation with von Willebrand factor (vWF), a glycoprotein synthesized exclusively by endothelial cells and megakaryocytes that is stored in intracellular granules or constitutively secreted into plasma. Connexin43 colocalized with vWF-positive cells in both the retinal and the choroidal circulations (Figs. 2, 3). Within the retina, punctate connexin43 immunoreactivity was present on retinal blood vessels in the inner plexiform, inner nuclear, and outer plexiform layers (Fig. 2). In the choroidal circulation, connexin43 immunoreactivity was observed in capillaries similar to those seen in Figure 2 and on endothelial cells of larger choroidal blood vessels (Fig. 3). In addition, diffuse punctate labeling occurred, most likely in association with connective tissue fibroblasts. 

Figure 2.

 

Connexin43 immunofluorescence in retinal blood vessels. (A) Sagittal section of the nerve fiber layer and ganglion cell layer double-labeled with connexin43 and vWF. Strong connexin43 immunoreactivity was observed in the inner vascular plexus. (B) Z-stack projection of a retinal capillary at the junction of the inner plexiform and inner nuclear layers (white arrows). Connexin43 immunofluorescence was detected between walls of the vessel lumen labeled with vWF (arrowheads). (C, D) Outer vascular plexus. Connexin43 immunoreactivity exhibited a “train track” pattern (arrowheads). vWF, von Willebrand factor; GCL, ganglion cell layer; INL, inner nuclear layer. Scale bar, 20 μm.

Figure 3.

 

Connexin43 immunoreactivity in the retinal pigment epithelium and choroid. Strong connexin43 immunoreactivity was noted in the retinal pigment epithelium (arrowheads). In the choroid, connexin43 immunoreactivity was associated with endothelial cells of larger blood vessels and formed a radial spoke-like appearance (white arrow). Diffuse punctate connexin43 immunoreactivity was also present in the choroid, most likely in association with connective tissue fibroblasts. RPE, retinal pigment epithelium; BV, blood vessel. Scale bar, 20 μm.

Association of Connexin43 Immunoreactivity with Retinal Neuronal and Epithelial Elements

Double labeling with cell-specific neuronal markers was performed to determine whether retinal neurons express connexin43. No punctate connexin43 label suggestive of gap junctions was found to be associated with any of the neuronal specific markers calretinin, calbindin, parvalbumin, or islet-1. However, a small subpopulation of cell nuclei in the ganglion cell and inner nuclear layers was observed to have diffuse nuclear-associated connexin43 label (Fig. 1C). Occasionally, these nuclei were also calretinin positive. In the retinal pigment epithelium, immunostaining for connexin43 revealed a distinctive pattern. In this layer there was substantial punctate labeling at the margins of adjacent epithelial cells (Fig. 3). 

Connexin43 Immunoreactivity in the Optic Nerve

Within the optic nerve there was very strong connexin43 immunoreactivity. Connexin43 immunoreactive puncta and SMI32-positive retinal ganglion cell axons were compartmentalized into distinct bundles (Fig. 4A). However, the connexin43 label did not appear to be associated with the nerve axons themselves but predominantly colocalized with GFAP-positive astrocytes (Figs. 4B–D). Around optic nerve blood vessels and at the glial limitans, strong connexin43 immunoreactivity was observed. 

Figure 4.

 

Connexin43 immunoreactivity in the human optic nerve. (A) Longitudinal section of the human optic nerve labeled with connexin43 and SMI32. Connexin43 immunoreactive puncta and SMI32-positive axons were compartmentalized into distinct bundles. Strong connexin43 immunoreactivity was present in association with the glial limitans (arrowhead). (B–D) Double-label immunohistochemistry with GFAP showing that connexin43 gap junctions are predominantly astrocytic: connexin 43 (green); GFAP (red); colocalization (yellow). GFAP, glial fibrillary acidic protein. Scale bar, 150 μm.

Discussion

Connexins are a class of proteins that form gap junctions between cells in various mammalian tissues. Gap junction communication may be important in the pathogenesis of neuronal degeneration and has been identified as a potential neuroprotective target. ² The present study establishes that connexin43 is expressed on glia, blood vessels, and epithelial cells in the human retina and optic nerve. The expression pattern in the human shows significant homology with other vertebrates. Janssen-Bienhold et al. ²⁰ characterized the distribution of connexin43 expression in five different vertebrates using immunoblotting and immunofluorescence microscopy. Connexin43 was found to be present in the retinal pigment epithelium of all test species and in blood vessels of vascularized retinas in the fish and rat. ²⁰ In the rat, connexin43 was also localized to the nerve fiber layer, most likely on astrocytes. ²⁰ Connexin43 immunolabeling has also been detected colocalized with GFAP in the ganglion cell layer of the mouse ²¹ and the medullary ray region of the rabbit, ²² suggesting that connexin43 is the major connexin protein of astrocytes in the mammalian retina. Glial connexin43 immunoreactivity has been further characterized by Zahs et al., ²³ who demonstrated homotypic gap junction coupling between astrocytes and heterotypic coupling between astrocytes and Müller cells. Injection of the gap-junction permeant tracer Lucifer yellow combined with simultaneous whole-cell patch recording has confirmed functional astrocyte-astrocyte and astrocyte-Müller cell coupling. ²⁴ Additionally, Müller cell endfoot processes are believed to be coupled through gap junctions at the internal and external limiting membranes. ^25,26 Gold and Dowling ²⁵ reported connexin43 immunoreactivity on the external limiting membrane of the cane toad, whereas Giblin and Christensen ²⁶ described a similar pattern in the catfish retina. This is supported by the work of Mobbs et al., ²⁷ who described strong electrical and dye coupling between Müller cells in the axolotl. In contrast to higher vertebrates, lower vertebrates show extensive coupling of Müller cells through connexin43. ²⁸ As the only glial cells in lower vertebrate retinas, Müller cells are thought to play an important role in K⁺ ion buffering by forming a functional syncytium mediated by connexin43 gap junctions. ²⁹ However, simultaneous whole-cell current-clamp recordings between glial cells in the isolated rat retina would suggest that gap junctions may permit the intercellular spread of ions and small molecules, including messengers mediating Ca²⁺ wave propagation, but are too weak to carry significant K⁺ spatial buffer currents. ³⁰  

Vascular endothelial expression has been described in the retinas of the and fish. ²⁰ This is consistent with the known expression of connexin43 in the aorta ³¹ and coronary arteries ³² of the rat. Electron microscopy and immunohistochemical studies have shown that connexin43 is predominantly expressed on vascular smooth muscle cells. ³³ Although expression is also observed between endothelial cells, the density of connexin43 gap junctions is lower for junctional plaques between endothelial cells than for those between smooth muscle cells. ³³ The presence of endothelial cell-endothelial cell, smooth muscle cell-smooth muscle cell, and endothelial cell-smooth muscle cell coupling has been confirmed by ultrastructural studies. ³⁴ In addition, pericytes have been shown to express connexin43. ³⁵  

The role of gap junctions in blood vessels continues to be elucidated. It has been suggested that intercellular communication through gap junctions between smooth muscle and/or endothelial cells may mediate vasodilation and vasoconstriction through the integration of neural and endothelial signals across the vessel wall. ^36,37 Additionally, gap junctions may allow the diffusion of second-messenger molecules through vascular wall cells. ³⁶ Direct cell-to-cell communication by way of gap junction channels may be particularly important in the maintenance of vascular homeostasis in retinal capillaries. ³⁵ It has been shown that gap junction communication is influenced by cytokines such as TNF-α ³⁸ and that the release of such cytokines may trigger the loss of pericytes through apoptosis, ³⁹ leading to increased vascular permeability. Furthermore, in conditions such as diabetes, there is a significant increase in connexin43-mediated junctional permeability in other vascular tissues. ⁴⁰  

The significance of diffuse connexin43 immunoreactivity surrounding a subpopulation of neuronal cell nuclei in the ganglion cell layer and inner nuclear layer is unclear. It is likely that this pattern of immunolabeling represents nonspecific binding to an evenly distributed protein carrying a connexin43-related epitope. ²⁰ In the human retina, the inner nuclear layer contains horizontal, bipolar, and amacrine cell bodies. ⁴¹ Horizontal cells form a single layer at the outer border of the inner nuclear layer, whereas amacrine cells form a layer two to three cells deep adjacent to the inner plexiform layer. ⁴¹ Gap junctions have been reported on several neuronal cell types in the mammalian retina and are believed to be critical for rapid interneuronal communication and the integration and propagation of signals. ²⁶ Janssen-Bienhold et al. ²⁰ reported connexin43 expression on a subpopulation of amacrine cells in the zebrafish and horizontal cells in the carp retina. Giblin and Christensen ²⁶ observed infrequent labeling of the inner nuclear layer of the mouse. Connexin43 immunoreactivity has been reported in the plexiform layers and may represent synaptic interneuronal gap junctions. ⁴² However, it is possible that the immunoreactivity observed in these studies was caused by the labeling of Müller cells passing through the plexiform layers. Further studies are therefore needed to confirm the presence of connexin43 gap junction coupling between retinal neurons in the inner and outer plexiform layers. 

Connexin43 expression has been reported on epithelial cells in the vertebrate retina. ²⁰ We observed extensive gap junction expression between cells of the retinal pigment epithelium. This is consistent with findings in the rat, mouse, rabbit, chicken, turtle, carp, and zebrafish. ²⁰ Retinal pigment epithelial cells have been shown to be coupled using Lucifer yellow injections in cultured cells of the chick embryo. ⁴³ Furthermore, cultured rat retinal pigment cells have been shown to propagate Ca²⁺ waves, and this propagation can be inhibited by the gap junction blocker halothane. ⁴⁴ It has been suggested that this junctionally coupled epithelial syncytium may also function as a spatial buffering system for extracellular potassium in the outer retina. ²⁰  

In the optic nerve connexin43 immunoreactivity was observed in a diffuse and granular pattern, corresponding with astrocyte processes. This is consistent with studies in the rat that have demonstrated connexin43 on type 1 but not type 2 astrocytes. ⁴⁵ The increased connexin43 immunoreactivity observed at the edge of the nerve likely represents increased numbers of astrocyte processes forming a barrier at the pia mater. ⁴² Connexin43 has been reported on oligodendrocytes in the adult rabbit retina, although connexin32 is considered to be the predominant connexin on these cells. ⁴⁶  

There is a paucity of information regarding the role of connexin43 in the retina and optic nerve; central nervous system research may provide insights into retinal and optic nerve disease. Alterations in connexin43 expression levels have been described after stroke, ⁴⁷ brain ischemia, ^48–50 traumatic brain injury, ⁵¹ and spinal cord injury. ^12,13,52,53 Modulation of connexin43 expression has been shown to be neuroprotective in both brain and spinal cord disease models. Frantseva et al. ¹⁰ demonstrated that acute knockdown of connexin43 using antisense oligodeoxynucleotides reduced cell death using an in vitro model of traumatic brain injury. Similarly, partial reduction of connexin43 expression decreased cell death in an in vitro model of ischemia-induced brain damage. ⁵⁴ In the spinal cord, O'Carroll et al. ¹² showed that connexin43 mimetic peptides reduce swelling, astrogliosis, and neuronal cell death after spinal cord injury. This is consistent with the findings of Cronin et al. ¹³ who, using a different method of connexin43 modulation, demonstrated reduced inflammation and improved functional recovery after spinal cord injury in rats. Similar downregulation of inflammation has been observed in the optic nerve. ⁵⁵ Danesh-Meyer et al. ⁵⁵ found that connexin43 antisense oligodeoxynucleotide treatment downregulates the inflammatory response in an in vitro interphase organotypic culture model of optic nerve ischemia. However, some authors have reported that gap junction blockade may be deleterious after central nervous system injury. ^56,57 Therefore, further studies are required to elucidate the neuroprotective or neurotoxic effects of gap junction modulation. 

In summary, our results show widespread expression of connexin43 in the human retina and optic nerve and can be used as a basis for comparison with disease. Because of their importance to the maintenance of extracellular homeostasis and intercellular communication, tracking them may be useful for understanding the pathogenesis of diseases of the retina and optic nerve. Modulation of connexin43 expression may represent a potential neuroprotective target for minimizing injury to the optic nerve and retina. 

Footnotes

 Supported by the Maurice and Phyllis Paykel Trust and the Neurological Foundation of New Zealand.

Footnotes

 Disclosure: N.M. Kerr, None; C.S. Johnson, None; C.F. de Souza, None; K.-S. Chee, None; W.R. Good, None; C.R. Green, None; H.V. Danesh-Meyer, None