Oligodendrocytes express gap junction proteins connexin32 and connexin45

Oligodendrocytes, the myelin-forming glia of brain, are connected by gap junctions in situ and in culture. Cultured oligodendrocytes from adult bovine and porcine brains were studied using immunocytochemical, molecular, and electrophysiological techniques in order to characterize the gap junction types. The expression of connexin32 was substantiated by the detection of low, but significant, signals using connexin-specific probes in Northern and Western blot analyses. Connexin43, which comprises gap junctions in astrocytes, was not detectable in pure oligodendrocytic cultures; mRNAs of connexin40 and connexin37 and connexin26 were also not detected. By means of two specific antibodies directed to the recently cloned connexin45 and by RT-PCR we were able to identify this connexin as a second oligodendrocytic gap junction protein. Whole cell voltage clamp recording provided evidence for electrical coupling between pairs of cultured oligodendrocytes (mean junctional conductance 3.9 nS, n = 38 pairs) and intracellular Lucifer Yellow injection indicated that oligodendrocytes were usually only weakly dye coupled, with spread generally being restricted to nearest neighbors. Unitary conductances ranged from >20 to <150 pS with modes of distribution at about 100 to 120pS and 40 to 20 pS, respectively. These unitary conductances are consistent with the channel events expected for connexin32 and connexin45. The low degree of functional coupling between oligodendrocytes in vitro corresponds with the low levels of connexin32 and connexin45 messenger RNAs and protein expression. GLIA 20:101-114, 1997. © 1997 Wiley-Liss Inc.     

Horizontal cell receptive fields are reduced in connexin57-deficient mice

Horizontal cells are coupled by gap junctions; the extensive coupling of the horizontal cells is reflected in their large receptive fields, which extend far beyond the dendritic arbor of the individual cell. In the mouse retina, horizontal cells express connexin57 (Cx57). Tracer coupling of horizontal cells is impaired in Cx57-deficient mice, which suggests that the receptive fields of Cx57-deficient horizontal cells might be similarly reduced. To test this hypothesis we measured the receptive fields of horizontal cells from wildtype and Cx57-deficient mice. First, we examined the synaptic connections between horizontal cells and photoreceptors: no major morphological alterations were found. Moreover, horizontal cell spacing and dendritic field size were unaffected by Cx57 deletion. We used intracellular recordings to characterize horizontal cell receptive fields. Length constants were computed for each cell using the cell's responses to concentric light spots of increasing diameter. The length constant was dependent on the intensity of the stimulus: increasing stimulus intensity reduced the length constant. Deletion of Cx57 significantly reduced horizontal cell receptive field size. Dark resting potentials were strongly depolarized and response amplitudes reduced in Cx57-deficient horizontal cells compared to the wildtype, suggesting an altered input resistance. This was confirmed by patch-clamp recordings from dissociated horizontal cells; mean input resistance of Cx57-deficient horizontal cells was 27% lower than that of wildtype cells. These data thus provide the first quantification of mouse horizontal cell receptive field size and confirm the unique role of Cx57 in horizontal cell coupling and physiology.     

Connexin therapeutics:blocking connexin hemichannel pores is distinct from blocking pannexin channels or gap junctions

Compounds that block the function of connexin and pannexin protein channels have been suggested to be valuable therapeutics for a range of diseases.Some of these compounds are now in clinical trials,but for many of them,the literature is inconclusive about the molecular effect on the tissue,despite evidence of functional recovery.Blocking the different channel types has distinct physiological and pathological implications and this review describes current knowledge of connexin and pannexin protein channels,their function as channels and possible mechanisms of the channel block effect for the latest therapeutic compounds.We summarize the evidence implicating pannexins and connexins in disease,considering their homeostatic versus pathological roles,their contribution to excesive ATP release linked to disease onset and progression.     

Subcellular distribution of connexin45 in OFF bipolar cells of the mouse retina

In the mouse retina, connexin45 (Cx45) participates in the gap junction between ON cone bipolar cells and AII amacrine cells, which constitutes an essential element of the primary rod pathway. Although it has been shown that Cx45 is also expressed in OFF bipolar cells, its subcellular localization and functional role in these cells are unknown. Here, we analyzed the localization of Cx45 on OFF bipolar cells in the mouse retina. For this, we used wild-type mice and a transgenic mouse line that expressed, in addition to native Cx45, a fusion protein consisting of Cx45 and the enhanced green fluorescent protein (EGFP). Cx45-EGFP expression generates an EGFP signal at gap junctions containing Cx45. Combining immunohistochemistry with intracellular injections, we found that Cx45 was present on dendrites and axon terminals of all OFF bipolar cell types. Cx45 was not found at intersections of two terminal processes of the same type, suggesting that Cx45 might not form gap junctions between axon terminals of the same OFF bipolar cell type but rather might connect OFF bipolar cells to amacrine or ganglion cells. In OFF bipolar cell dendrites, Cx45 was found predominantly in the proximal outer plexiform layer (OPL), well below the cone pedicles. Cx45 did not colocalize with Cx36, which is found predominantly in the distal OPL. We conclude that Cx45 is expressed on OFF bipolar cell dendrites, presumably forming gap junctions with cells of the same type, and on OFF bipolar cell axon terminals, presumably forming heterologous gap junctions with other retinal neurons.     

Differential connexin expression,gap junction intercellular coupling,and hemichannel formation in NT2/D1 human neural progenitors and terminally differentiated hNT neurons

Connexin-mediated gap junctions and open hemichannels in nonjunctional membranes represent two biologically relevant mechanisms by which neural progenitors can coordinate their response to changes in the extracellular environment. NT2/D1 cells are a teratocarcinoma progenitor line that can be induced to differentiate terminally into functional hNT neurons and NT-G nonneuronal cells. Clinical transplants of hNT neurons and experimental grafts of NT2/D1 progenitors or hNT neurons have been used in cell-replacement therapy in vivo. Previous studies have shown that NT2/D1 cells express connexin 43 (Cx43) and that NT2/D1 progenitors are capable of dye transfer. To determine whether NT2/D1 progenitors and differentiated hNT cultures express other connexins, Cx26, Cx30, Cx32, Cx36, Cx37, Cx43, and Cx46.6 mRNA and protein were analyzed. NT2/D1 progenitors express Cx30, Cx36, Cx37, and Cx43. hNT/NT-G cultures express Cx36, Cx37, and de novo Cx46.6. Cx26 and Cx32 were not expressed in NT2/D1 or hNT/NT-G cells. NT2/D1 progenitors formed functional gap junctions as assessed by dye coupling as well as open hemichannels in nonjunctional membranes as assessed by dye-uptake studies. Dye coupling was inhibited by the gap junction blocker 18alpha-glycyrrhetinic acid. Hemichannel activity was inhibited by the dual-specificity chloride channel/connexin hemichannel inhibitor flufenamic acid but not by the chloride channel inhibitor 4,4'-diisothiocyanatostilbene-2,2'-disulfonic acid. Both dye coupling and dye uptake were substantially reduced following differentiation of NT2/D1 progenitors. We conclude that the pattern of connexin expression in NT2/D1 cells changes over the course of differentiation corresponding with a reduction in biochemical coupling and hemichannel activity in differentiated cells.     

Evidence for hemi-gap junctional channels in isolated horizontal cells of the skate retina

Prolonged depolarization of isolated, voltageclamped skate retinal horizontal cells produces an outward current that exhibits a late onset and develops slowly with time. This current, which we refer to as the Q-current, is associated with an increase in membrane conductance, and is present when other voltage-gated conductances have been pharmacologically blocked. The reversal potential for the Q-current, obtained using tail current analysis, was close to 0 mV. The magnitude of the current was greatly reduced by superfusion with 25 mM acetate, and by 4 mM cobalt chloride, 2 mM 1-octanol, and a saturated solution of the general anesthetic halothane. In addition, the low-molecular weight fluorescent dye Lucifer yellow, applied extracellularly, entered the cells during activation of the Q-current, whereas a 3 kD dextran-fluorescein complex did not cross the cell membrane. The effects of divalent cations, the nonspecific nature of the ionic current suggested by its reversal potential, the entry of Lucifer yellow, and the ability of acetate, halothane, cobalt, and octanol to block the current lead us to hypothesize that the Q-current results from the opening of hemi-gap junctional channels that mediate electrical coupling between skate horizontal cells. © 1993 Wiley-Liss, Inc.     

Gap junctions among the perikarya, dendrites, and axon terminals of the luminosity-type horizontal cell of the turtle retina

Gap junctions of the H1 horizontal cell of the turtle retina (Leeper, '78) were studied in thin-sectioned material and in freeze-fracture replicas. Perikaryal gap junctions were extremely restricted, 0.02-0.07 micron2 in in area, whereas those of axon terminals were much larger, most being 0.1-1.0 micron2. Both varieties, however, had the usual seven-layered appearance in thin section and measured 15 +/- 1 nm in overall width between cytoplasmic faces. Freeze-fractured views of the perikaryal junctions revealed roughly circular patches of P-face 9-nm particles and E-face pits. The axon terminal gap junctions were seen as large areas of P-face particles and E-face pits containing occasional islands of unspecialized membrane. Particle densities varied from 1,455 to 2,448 microns-2. A serial reconstruction was made of a portion of the axon terminal network in order to measure the surface areas of the axons contained therein and the fraction occupied by gap junctions. These data demonstrated that the fractional area occupied by gap junctions was roughly in inverse proportion to the area of the axon region (tuberous core vs. terminal process). It is argued that this constitutes an impedance matching device to ensure adequate current flow through the axon processes. Assuming that each P-face particle represents a connection having a conductance of 10(-10) S and given the P-face particle density and gap junctional areas determined in this report, we calculated that the gap junction distribution is adequate to account for the spatial properties of the horizontal cell axon network (Lamb, '76).     

Identification and characterization of pannexin expression in the mammalian cochlea

The gap junction in vertebrates is encoded by the connexin gene family. Recently, a new gene family termed pannexin (Panx) has been identified in vertebrates and found to encode gap junctional proteins as well. To date, three pannexin isoforms (Panx1, 2, and 3) have been cloned from mouse and human genomes. In this study, expression of pannexins in the mouse and rat cochlea was investigated. Polymerase chain reaction and Western blot analysis showed that all three pannexin isoforms were expressed in the cochlea. Immunofluorescent staining showed that Panx1 expression was extensive. In the organ of Corti, Panx1 labeling was found in supporting cells, including pillar cells, Hensen cells, Claudius cells, and Boettcher cells. Both surface plaque‐like punctate labeling and diffuse‐cytoplasmic labeling were visible. However, the labeling was weak and rare in Deiters cells. No labeling was found in the hair cells. Intense labeling for Panx1 was also observed in the interdental cells in the spiral limbus, the inner and outer sulcus cells, and the type II fibrocytes in the spiral prominence and central region in the cochlear lateral wall. In addition, Panx1 labeling was detectable in Reissner's membrane and strial blood vessel cells. Panx2 labeling was restricted to the basal cells in the stria vascularis and was also detectable in the spiral ganglion neurons. However, no overlapping labeling for Panx1 and Panx2 was observed. Finally, Panx3 labeling was exclusively observed in the cochlear bone. Thus, Panx1, 2, and 3 are abundantly expressed in the mammalian cochlea and demonstrate distinct cellular distributions. Like connexins, they may play an important role in hearing. J. Comp. Neurol. 512:336–346, 2009. © 2008 Wiley‐Liss, Inc.     

Localization of heterotypic gap junctions composed of connexin45 and connexin36 in the rod pathway of the mouse retina

The primary rod pathway in mammals contains gap junctions between AII amacrine cells and ON cone bipolar cells which relay the rod signal into the cone pathway under scotopic conditions. Two gap junctional proteins, connexin36 (Cx36) and connexin45 (Cx45), appear to play a pivotal role in this pathway because lack of either protein leads to an impairment of visual transmission under scotopic conditions. To investigate whether these connexins form heterotypic gap junctions between ON cone bipolar and AII amacrine cells, we used newly developed Cx45 antibodies and studied the cellular and subcellular distribution of this protein in the mouse retina. Specificity of the Cx45 antibodies was determined, among others, by Western blot and immunostaining of mouse heart, where Cx45 is abundantly expressed. In mouse retina, Cx45 immunosignals were detected in both plexiform layers and the ganglion cell layer. Double staining for Cx45 and Cx36 revealed a partial overlap in the punctate patterns in the ON sublamina of the inner plexiform layer of the retina. We quantified the distributions of these two connexins in the ON sublamina, and detected 30% of the Cx45 signals to be co-localized with or in close apposition to Cx36 signals. Combining immunostaining and intracellular dye injection revealed an overlap or tight association of Cx36 and Cx45 signals on the terminals of injected AII amacrine and two types of ON cone bipolar cells. Our results provide direct evidence for heterotypic gap junctions composed of Cx36 and Cx45 between AII amacrine and certain types of ON cone bipolar cells.     

Extended cocaine self-administration and deprivation produces region-specific and time-dependent changes in connexin36 expression in rat brain

Cocaine addiction is a disease that develops over time, and it is thought that drug-induced neuro-adaptations underlie the changes in behavior seen across the addictive process. While a number of alterations in synaptic transmission have been identified, little is currently known regarding cocaine's effects on gap junctional communication between neurons. Here we examine the effects of a cocaine self-administration regimen, previously shown to increase the reinforcing efficacy of cocaine, on the expression of the neuron-specific gap junction-forming protein connexin36 (C x 36). Using real-time RT-PCR and immunoblotting, we show that binge cocaine self-administration produces region-specific and time-dependent changes in C x 36 mRNA and protein expression in the nucleus accumbens, prefrontal cortex, and hippocampus. A number of changes in C x 36 were present 1 day and 7 days following self-administration, and C x 36 mRNA and protein appeared to be differentially regulated in a region-specific manner. C x 36 protein was significantly decreased in the prefrontal cortex 7 days following self-administration, a time point when behavioral sensitization to the reinforcing effects of cocaine is observed. These results suggest that changes in neuronal gap junction expression may be one mechanism by which cocaine self-administration produces enduring changes in behavior.     

Neuronal connexin36 association with zonula occludens-1 protein (ZO-1) in mouse brain and interaction with the first PDZ domain of ZO-1

Among the 20 members in the connexin family of gap junction proteins, only connexin36 (Cx36) is firmly established to be expressed in neurons and to form electrical synapses at widely distributed interneuronal gap junctions in mammalian brain. Several connexins have recently been reported to interact with the PDZ domain-containing protein zonula occludens-1 (ZO-1), which was originally considered to be associated only with tight junctions, but has recently been reported to associate with other structures including gap junctions in various cell types. Based on the presence of sequence corresponding to a putative PDZ binding motif in Cx36, we investigated anatomical relationships and molecular association of Cx36 with ZO-1. By immunofluorescence, punctate Cx36/ZO-1 colocalization was observed throughout the central nervous system of wild-type mice, whereas labelling for Cx36 was absent in Cx36 knockout mice, confirming the specificity of the anti-Cx36 antibodies employed. By freeze-fracture replica immunogold labelling, Cx36 and ZO-1 in brain were found colocalized within individual ultrastructurally identified gap junction plaques, although some plaques contained only Cx36 whereas others contained only ZO-1. Cx36 from mouse brain and Cx36-transfected HeLa cells was found to coimmunoprecipitate with ZO-1. Unlike other connexins that bind the second of the three PDZ domains in ZO-1, glutathione S-transferase-PDZ pull-down and mutational analyses indicated Cx36 interaction with the first PDZ domain of ZO-1, which required at most the presence of the four c-terminus amino acids of Cx36. These results demonstrating a Cx36/ZO-1 association suggest a regulatory and/or scaffolding role of ZO-1 at gap junctions that form electrical synapses between neurons in mammalian brain.     

Renewal of electrotonic synapses in teleost retinal horizontal cells

In teleost retinas, the somata of same-type cone horizontal cells are electrically coupled via extensive gap junctions, as are the axon terminals of same-type cells. This coupling persists throughout the animal's life and is modulated by dopamine and conditions of light- vs. dark-adaptation. Gap junction particle density in goldfish horizontal cell somata has also been shown to change under these conditions, indicating that these junctions are dynamic. We have used electron microscopy to examine gap junctions in bass horizontal cells with a fixation method that facilitates detection of gap junctions. Annular gap junction profiles were observed in the somatic cytoplasm of all cone horizontal cell types in both light- and dark-adapted animals. Serial sections showed that most profiles represented gap junction vesicles free within the cytoplasm; the remainder represented vesicles still attached to extensive plasma membrane gap junctions by a thin cytoplasmic neck, suggestive of an intermediate stage in endocytosis. Observations of gap junction vesicles containing fragments of gap junctional membrane and/or fused with lysosomal bodies further supported this hypothesis. Because gap junctions persist between the horizontal cells, we propose that gap junction endocytosis and lysosomal degradation are balanced by addition of new junctions. While endocytosis has been widely demonstrated to serve in programmed removal of gap junctions (without subsequent replacement), from both nonneuronal cells and developing neurons, this study indicates that it can also function in the renewal of electrical synapses in the adult teleost retina, where gap junction elimination is not the goal.     

Connexin57 is expressed in dendro‐dendritic and axo‐axonal gap junctions of mouse horizontal cells and its distribution is modulated by light

Mouse horizontal cells are coupled by gap junctions composed of connexin57. These gap junctions are regulated by ambient light via multiple neuromodulators including dopamine. In order to analyze the distribution and structure of horizontal cell gap junctions in the mouse retina, and examine the effects of light adaptation on gap junction density, we developed antibodies that detect mouse retinal connexin57. Using immunohistochemistry in retinal slices, flat‐mounted retinas, and dissociated retinal cells, we showed that connexin57 is expressed in the dendrites and axon terminal processes of mouse horizontal cells. No staining was found in retinas of connexin57‐deficient mice. Significantly more connexin57‐positive puncta were found in the distal than in the proximal outer plexiform layer, indicating a higher level of expression in axon terminal processes than in the dendrites. We also examined the gap junctions using immunoelectron microscopy and showed that connexin57 does not form hemichannels in the horizontal cell dendritic tips. Light adaptation resulted in a significant increase in the number of connexin57‐immunoreactive plaques in the outer plexiform layer, consistent with previously reported effects of light adaptation on connexin57 expression in the mouse retina. This study shows for the first time the detailed location of connexin57 expression within mouse horizontal cells, and provides the first ultrastructural data on mouse horizontal cell gap junctions. J. Comp. Neurol. 513:363–374, 2009. © 2009 Wiley‐Liss, Inc.     

The immunocytochemical localization of connexin 36 at rod and cone gap junctions in the guinea pig retina

Connexin 36 (Cx36) is a channel-forming protein found in the membranes of apposed cells, forming the hexameric hemichannels of intercellular gap junction channels. It localizes to certain neurons in various regions of the brain including the retina. We characterized the expression pattern of neuronal Cx36 in the guinea pig retina by immunocytochemistry using specific antisera against Cx36 and green/red cone opsin or recoverin. Strong Cx36 immunoreactivity was visible in the ON sublamina of the inner plexiform layer and in the outer plexiform layer, as punctate labelling patterns. Double-labelling experiments with antibody directed against Cx36 and green/red cone opsin or recoverin showed that strong clustered Cx36 immunoreactivity localized to the axon terminals of cone or close to rod photoreceptors. By electron microscopy, Cx36 immunoreactivity was visible in the gap junctions as well as in the cytoplasmic matrices of both sides of cone photoreceptors. In the gap junctions between cone and rod photoreceptors, Cx36 immunoreactivity was only visible in the cytoplasmic matrices of cone photoreceptors. These results clearly indicate that Cx36 forms homologous gap junctions between neighbouring cone photoreceptors, and forms heterologous gap junctions between cone and rod photoreceptors in guinea pig retina. This focal location of Cx36 at the terminals of the photoreceptor suggests that rod photoreceptors can transmit rod signals to the pedicle of a neighbouring cone photoreceptor via Cx36, and that the cone in turn signals to corresponding ganglion cells via ON and OFF cone bipolar cells.     

Ablation of connexin30 in transgenic mice alters expression patterns of connexin26 and connexin32 in glial cells and leptomeninges

Expression of connexin26 (Cx26), Cx30 and Cx43 in astrocytes and expression of Cx29, Cx32 and Cx47 in oligodendrocytes of adult rodent brain has been well documented, as has the interdependence of connexin expression patterns of macroglial cells in Cx32- and Cx47-knockout mice. To investigate this interdependence further, we examined immunofluorescence labelling of glial connexins in transgenic Cx30 null mice. Ablation of astrocytic Cx30, confirmed by the absence of immunolabelling for this connexin in all brain regions, resulted in the loss of its coupling partner Cx32 on the oligodendrocyte side of astrocyte-oligodendrocyte (A/O) gap junctions, but had no effect on the localization of astrocytic Cx43 and oligodendrocytic Cx47 at these junctions or on the distribution of Cx32 along myelinated fibres. Surprisingly, gene deletion of Cx30 led to the near total elimination of immunofluorescence labelling for Cx26 in all leptomeningeal tissues covering brain surfaces as well as in astrocytes of brain parenchyma. Moreover northern blot analysis revealed downregulation of Cx26 mRNA in Cx30-knockout brains. Our results support earlier observations on the interdependency of Cx30/Cx32 targeting to A/O gap junctions and further suggest that Cx26 mRNA expression is affected by Cx30 gene expression. In addition, Cx30 protein may be required for co-stabilization of gap junctions or for co-trafficking in cells.     

Modulation of connexin expression and gap junction communication in astrocytes by the gram-positive bacterium S. aureus

Gap junctions establish direct intercellular conduits between adjacent cells and are formed by the hexameric organization of protein subunits called connexins (Cx). It is unknown whether the proinflammatory milieu that ensues during CNS infection with S. aureus, one of the main etiologic agents of brain abscess in humans, is capable of eliciting regional changes in astrocyte homocellular gap junction communication (GJC) and, by extension, influencing neuron homeostasis at sites distant from the primary focus of infection. Here we investigated the effects of S. aureus and its cell wall product peptidoglycan (PGN) on Cx43, Cx30, and Cx26 expression, the main Cx isoforms found in astrocytes. Both bacterial stimuli led to a time-dependent decrease in Cx43 and Cx30 expression; however, Cx26 levels were elevated following bacterial exposure. Functional examination of dye coupling, as revealed by single-cell microinjections of Lucifer yellow, demonstrated that both S. aureus and PGN inhibited astrocyte GJC. Inhibition of protein synthesis with cyclohexamide (CHX) revealed that S. aureus directly modulates, in part, Cx43 and Cx30 expression, whereas Cx26 levels appear to be regulated by a factor(s) that requires de novo protein production; however, CHX did not alter the inhibitory effects of S. aureus on astrocyte GJC. The p38 MAPK inhibitor SB202190 was capable of partially restoring the S. aureus-mediated decrease in astrocyte GJC to that of unstimulated cells, suggesting the involvement of p38 MAPK-dependent pathway(s). These findings could have important implications for limiting the long-term detrimental effects of abscess formation in the brain which may include seizures and cognitive deficits.     

Mutations of connexin32 in charcot-marie-tooth disease type X interfere with cell-to-cell communication but not cell proliferation and myelin-specific gene expression

Connexin32 (Cx32) is a gap junction protein and its mutations are responsible for X-linked Charcot-Marie-Tooth disease. We examined the functional abnormality of C6 glioma cells transfected with mutant (C53S and P172R) Cx32 genes. Nontransfected C6 did not express Cx32. Northern and Western blot analyses showed Cx32 mRNA and protein in cells with the wild-type gene as well as with the mutant Cx32 genes. An immunocytochemical study of cells with the wild-type gene showed the immunoreactive spots in the cell membrane. In cells with C53S or P172R mutant gene, however, the immunoreactivity was found in the cytoplasm. The scrape-loading method produced effective dye transfer in cells with the wild-type gene but not in those with mutant genes. A cell proliferation assay showed no differences in nontransfected cells, cells with the wild-type gene and those with the mutant genes. Messenger RNA expression for proteolipid protein did not change. These findings suggest that Cx32 gene mutation results in loss of cell-to-cell communication because of failure to incorporate Cx32 protein in the cell membrane. The mutations do not, however, interfere with cell proliferation or myelin-specific gene expression, at least myelin proteolipid protein expression in C6 glioma cells. J. Neurosci. Res. 51:154–161, 1998. © 1998 Wiley-Liss, Inc.     

Morphology and tracer coupling pattern of alpha ganglion cells in the mouse retina

Alpha cells are a type of ganglion cell whose morphology appears to be conserved across a number of mammalian retinas. In particular, alpha cells display the largest somata and dendritic arbors at a given eccentricity and tile the retina as independent on- (ON) and off-center (OFF) subtypes. Mammalian alpha cells also express a variable tracer coupling pattern, which often includes homologous (same cell type) coupling to a few neighboring alpha cells and extensive heterologous (different cell type) coupling to two to three amacrine cell types. Here, we use the gap junction-permeant tracer Neurobiotin to determine the architecture and coupling pattern of alpha cells in the mouse retina. We find that alpha cells show the same somatic and dendritic architecture described previously in the mammal. However, alpha cells show varied tracer coupling patterns related to their ON and OFF physiologies. ON alpha cells show no evidence of homologous tracer coupling but are coupled heterologously to at least two types of amacrine cell whose somata lie within the ganglion cell layer. In contrast, OFF alpha cells are coupled to one another in circumscribed arrays as well as to two to three types of amacrine cell with somata occupying the inner nuclear layer. We find that homologous coupling between OFF alpha cells is unaltered in the connexin36 (Cx36) knockout (KO) mouse retina, indicating that it is not dependent on Cx36. However, a subset of the heterologous coupling of ON alpha cells and all the heterologous coupling of OFF alpha cells are eliminated in the KO retina, suggesting that Cx36 comprises most of the junctions made with amacrine cells.