Basic Science

Gautam Ullal , Margaret Fahnestock , and Ronald Racine
Neurons communicate with each other via chemical neurotransmitters that react with protein receptors. The most common of the excitatory neurotransmitters is glutamate, which reacts with a variety of different receptor subtypes. One of these subtypes (called the kainate receptor, because it responds to the excitatory drug, kainate), itself comes in at least three different forms [referred to as glutamate receptors (or GluR) 5, 6, and 7). It has been suggested that GluR 5 and GluR 6 play an important role in human epilepsy. The role of GluR7 in epilepsy is not yet understood. In this experiment, we monitored the messenger ribonucleic acid (mRNA) and protein expression of GluR5, GluR6, and GluR7 in rat hippocampus, a structure often implicated in epileptic activity, 72 h, 90 days, and 180 days after a strong and prolonged seizure induced by kainate injections. An increase in GluR5 expression, which may render the system more excitable, was seen at 72 h and remained elevated until 180 days, so the change is very long lasting. The GluR7 measures showed only a decrease at 90 days after seizures. No change was found in the GluR6 measures at any of the times after the seizure. These results suggest a rather complex array of seizure-induced changes in these receptor subtypes, but the increase in GluR5 expression may be related to a long-lasting increase in seizure susceptibility that develops after an initial, prolonged seizure. Epilepsia 2005;46(5).     

Enhanced susceptibility of Prnp-deficient mice to kainate-induced seizures, neuronal apoptosis, and death: Role of AMPA/kainate receptors

Normal physiologic functions of the cellular prion protein (PrPc) are still elusive. This GPI-anchored protein exerts many functions, including roles in neuron proliferation, neuroprotection or redox homeostasis. There are, however, conflicting data concerning its role in synaptic transmission. Although several studies report that PrPc participates in NMDA-mediated neurotransmission, parallel studies describe normal behavior of PrPc-mutant mice. Abnormal axon connections have been described in the dentate gyrus of the hippocampi of PrPc-deficient mice similar to those observed in epilepsy. A study indicates increased susceptibility to kainate (KA) in these mutant mice. We extend the observation of these studies by means of several histologic and biochemical analyses of KA-treated mice. PrPc-deficient mice showed increased sensitivity to KA-induced seizures in vivo and in vitro in organotypic slices. In addition, we show that this sensitivity is cell-specific because interference experiments to abolish PrPc expression increased susceptibility to KA in PrPc-expressing cells. We indicate a correlation of susceptibility to KA in cells lacking PrPc with the differential expression of GluR6 and GluR7 KA receptor subunits using real-time RT-PCR methods. These results indicate that PrPc exerts a neuroprotective role against KA-induced neurotoxicity, probably by regulating the expression of KA receptor subunits.     

Localization of kainate receptor subunit GluR5-immunoreactive cells in the rat hypothalamus

Glutamate is the major excitatory neurotransmitter in the hypothalamus, which exerts its effects by activating ion channel-forming (ionotropic) or G-protein-coupled (metabotropic) receptors. Kainate-preferring glutamate receptor subunits (GluR5, GluR6, GluR7, KA1, and KA2) form one of the three ionotropic receptor families. In the present study, we analyzed the distribution of GluR5 subunit protein in the rat hypothalamus with immunohistochemistry. GluR5 immunoreactivity was observed in perikarya and processes of many hypothalamic cells some of which, based upon their morphological differentiation by size and structure, appeared to be neurons and others glial cells. Analyses revealed that higher number of glial cells were GluR5 positive when compared to the moderate number of GluR5-labeled neurons in the anteroventral periventricular nucleus. Numerous GluR5-expressing neurons and similar number of glia were detected in the suprachiasmatic nucleus. In the arcuate nucleus more glial cells were identified with GluR5 immunoreactivity than the number of labeled neurons. Scattered GluR5-positive cells were present in the periventricular nucleus. Specific immunostaining was not seen in the ventromedial nucleus or dorsomedial nucleus. In conclusion, it is suggested that the GluR5 subunits participate in the glutamatergic regulation of several neuroendocrine systems, such as the tubero-infundibular systems as well as in the control of circadian output through neuron-to-neuron and/or neuron-to-glia interactions.     

GluR2(B) knockdown accelerates CA3 injury after kainate seizures

Ca2+ currents are thought to enhance glutamate excitotoxicity. To investigate whether reduced expression of the Ca2+ limiting GluR2(B) subunit enhances seizure-induced vulnerability to either CA1 or CA3 neurons, we delivered GluR2(B) oligodeoxynucleotides (AS-ODNs) to the dorsal hippocampus of adult rats before inducing kainate (KA) seizures. After knockdown, no changes in behavior, electrographic activity, or histology were observed. In contrast, GluR2(B) knockdown and KA-induced status epilepticus produced accelerated histological injury to the ipsilateral CA3a-b and hilar subregions. At 8 to 12 h, the CA3a was preferentially labeled by both silver and TUNEL methods. TUNEL staining revealed 2 types of nuclei. They were round with uniform label, features of necrosis, or had DNA clumping or speckled chromatin deposits within surrounding cytosol, features of apoptosis. At 16 to 24 h, many CA3a-c neurons were shrunken, eosinophilic, argyrophilic, or completely absent. Immunohistochemistry revealed marked decreases in GluR2(B) subunits throughout the hippocampus, NR1 immunoreactivity was also reduced but to a lesser extent. In contrast, GluR1 and NR2A/B immunohistochemistry was relatively uniform except in regions of cell loss or within close proximity to the CA1 infusion site. At 144 h, the CA3 was still preferentially injured although bilateral CA1 injury was also observed in some AS-ODN-, S-ODN-, and KA-only-treated animals. Glutamate receptor antibodies revealed generalized decreases in the CA3 with all probes tested at this delayed time. In contrast, GluR2(B) expression was increased within CA1 irregularly shaped, injured neurons. Therefore, hippocampal deprivation of GluR2(B) subunits is insufficient to induce cell death in mature animals but may accelerate the already known CA3/hilar lesion, possibly by triggering apoptosis within CA3 neurons. CA1 and DG survive the first week despite their loss of GluR2(B) subunits, suggesting that other intrinsic properties such as increased Na+ conductance and reduced ability of the GluR2(B) subunit to interact with certain cytoplasmic proteins may be responsible for the augmented cell death rather than changes in AMPA receptor-mediated Ca2+ permeability. Alternatively, changes in allosteric interactions that affect other receptor classes of high density at the mossy fiber synapse (e.g. KA receptors) may augment KA neurotoxicity. Latent GluR2(B) increases in CA1 injured neurons support a role for AMPA receptor subunit alterations in seizure-induced tolerance.     

红藻氨酸致颞叶癫(癎)大鼠脑内Semaphorin3A、Semaphorin4C基因表达的研究

目的研究轴索导向分子Semaphorin3A(Sema3A)、4C(Sema4C)对癫大鼠海马苔藓纤维重建的调控作用及对皮层神经元的保护作用。方法大鼠侧脑室内注射红藻氨酸制备颞叶癫模型,原位杂交法检测致痫间后1d,1、2、3、4周大鼠脑内Sema3A/Sema4C mRNA表达。结果致痫间后1周Sema3A、Sema4CmRNA分别在齿状回(DG),CA3区表达明显下降(P<0.01),持续至3、4周时恢复至正常(P>0.05);致痫间后1d Sema3A mRNA在皮层表达明显下降(P<0.01),持续至1、2周后恢复至正常(P>0.05)。结论红藻氨酸致痫间后DG及CA3区神经元分别下调Sema3A/Sema4C mRNA的表达,促进癫大鼠苔藓纤维重建;皮层神经元通过下调Sema3A mRNA的表达来维持自身存活。     

Expression of GluR6 kainate receptor subunit in granular layer of weaver mouse cerebellum

The Girk2 ^wv (weaver) mutation impairs migration of cerebellar granule cells from external to internal granular layer and induces neuronal death during the first 2 weeks of postnatal life. Kainate receptors are heteromeric ionotropic receptors of glutamate consisting of five subunits termed GluR5, GluR6, GluR7, KA1 and KA2. In order to investigate whether the weaver gene affects the expression of kainate receptors in weaver cerebellum, we determined mRNA expression levels of GluR6 kainate receptor subunit and [³H]kainic acid specific binding in the developing cerebellum, using in situ hybridization and receptor film autoradiography, respectively. In the weaver postnatal day 10 (P10) cerebellum, our data indicated lower levels of GluR6 mRNA expression and lower [³H]kainic acid specific binding in external granular layer (EGL) compared to normal EGL. Our results are indicative of either down-regulation of kainate receptors or modulation of their functional characteristics in weaver granule cells.     

GluR6-containing KA receptor mediates the activation of p38 MAP kinase in rat hippocampal CA1 region during brain ischemia injury

Our previous study showed that kainate (KA) receptor subunit GluR6 played an important role in ischemia-induced MLK3 and JNK activation and neuronal degeneration through the GluR6-PSD95-MLK3 signaling module. However, whether the KA receptors subunit GluR6 is involved in the activation of p38 MAP kinase during the transient brain ischemia/reperfusion (I/R) in the rat hippocampal CA1 subfield is still unknown. In this present study, we first evaluated the time-course of phospho-p38 MAP kinase at various time-points after 15 min of ischemia and then observed the effects of antagonist of KA receptor subunit GluR6, GluR6 antisence oligodeoxynucleotides on the phosphorylation of p38 MAP kinase induced by I/R. Results showed that inhibiting KA receptor GluR6 or suppressing the expression of KA receptor GluR6 could down-regulate the elevation of phospho-p38 MAP kinase induced by I/R. These drugs also reduced the phosphorylation of MLK3, MKK3/MKK6, MKK4, and MAPKAPK2. Additionally, our results indicated administration of three drugs, including p38 MAP kinase inhibitor before brain ischemia significantly decreased the number of TUNEL-positive cells detected at 3 days of reperfusion and increased the number of the surviving CA1 pyramidal cells at 5 days of reperfusion after 15 min of ischemia. Taken together, we suggest that GluR6-contained KA receptors can mediate p38 MAP kinase activation through a kinase cascade, including MLK3, MKK3/MKK6, and MKK4 and then induce increased phosphorylation of MAPKAPK-2 during ischemia injury and ultimately result in neuronal cell death in the rat hippocampal CA1 region.     

AMPA/kainate receptors in mouse spinal cord cell-specific display of receptor subunits by oligodendrocytes and astrocytes and at the nodes of Ranvier

Spinal cord white matter is susceptible to AMPA/kainate (KA)-type glutamate receptor-mediated excitotoxicity. To understand this vulnerability, it is important to characterize the distribution of AMPA/KA receptor subunits in this tissue. Using immunohistochemistry and laser confocal microscopy, we studied the expression sites of AMPA/KA receptor subunits in mouse spinal cord. The white matter showed consistent immunoreactivity for AMPA receptor subunit GluR2/3 and KA receptor subunits GluR6/7 and KA2. In contrast, antibodies against GluR1, GluR2, GluR4 (AMPA), and GluR5 (KA) subunits showed only weak and occasional labeling of white matter. However, gray matter neurons did express GluR1 and GluR2, as well as GluR2/3. The white matter astrocytes were GluR2/3 and GluR6/7 immunopositive, while the gray matter astrocytes displayed primarily GluR6/7. Both exclusively and abundantly, KA2 labeled oligodendrocytes and myelin, identified by CNPase expression. Interestingly, myelin basic protein, another myelin marker, showed less correlation with KA2 expression, placing KA2 at specific CNPase-containing subdomains. Focal points of dense KA2 labeling showed colocalization with limited, but distinct, axonal regions. These regions were identified as nodes of Ranvier by coexpressing the nodal marker, ankyrin G. Overall, axonal tracts showed little, if any, AMPA/KA receptor expression. The proximity of oligodendrocytic KA2 to the axonal node and the paucity of axonal AMPA/kainate receptor expression suggest that excitotoxic axonal damage may be secondary and, possibly, mediated by oligodendrocytes. Our data demonstrate differential expression of glutamate AMPA and KA receptor subunits in mouse spinal cord white matter and point to astrocytes and oligodendrocytes as potential targets for pharmacological intervention in white matter glutamate excitotoxicity.     

Functional cooperation between KA2 and GluR6 subunits is involved in the ischemic brain injury

We investigated the possible relationships between KA2 subunit and GluR6 subunit, as well as the role of KA2 subunit in neuronal death induced by cerebral ischemia/reperfusion. Our results indicated that intracerebroventricular infusion of KA2 antisense oligodeoxynucleotides (AS) not only knocked down the expressions of KA2 and GluR6, but also suppressed the assembly of the GluR6/KA2-PSD95-MLK3 signaling module, and inhibited JNK activation and phosphorylation of c-jun. In addition, infusion of KA2 AS increased neuronal survival in CA1 region after 5 days of reperfusion. More interestingly, we found that the combination of KA2 and GluR6 AS exerted more significant effects than when pretreated with KA2 AS or GluR6 AS alone. Our results suggest that the KA2 subunit is involved in delayed neuronal death induced by cerebral ischemia, at the same time, it is noteworthy that the functional cooperation between KA2 and GluR6 subunits plays a critical role in the ischemic brain injury by PSD95-MLK3-MKK4/7-JNK3 signal pathway.     

Expression of glutamate receptors in the human and rat basal ganglia: Effect of the dopaminergic denervation on AMPA receptor gene expression in the striatopallidal complex in parkinson's disease and rat with 6-OHDA lesion

The overactivity of subthalamopallidal and corticostriatal glutamatergic neurons observed in Parkinson's disease (PD) suggests that antagonists of glutamate receptor could be used to alleviate the motor symptoms of the disease. In this study, we analysed two features of the striatopallidal complex: (1) the distribution of α-amino-3 hydroxy-5-methyl-4-isoxasol-propionate (AMPA) and kainate receptors and their corresponding mRNA by immunohistochemistry and in situ hybridisation and (2) the effect of dopaminergic denervation on AMPA receptor gene expression in PD patients and rats with 6-hydroxydopamine (6-OHDA)-induced degeneration of the nigrostriatal dopaminergic system. All AMPA receptor mRNAs and proteins (GluR1–4) were detected in the internal segment of the globus pallidus (GPi). Among kainate receptors, only KA1 and KA2 were detectable and only at a low level. Only GluR4 protein was detected in the neuropil of the GPi. In the striatum, GluR1, GluR2, and GluR3 were detected in about 70% of medium-sized and large neurons. By contrast, GluR4 mRNA was detected in only a small number of large and medium-sized neurons. Among kainate receptors, GluR6, GluR7, and KA2 were detected in about 50–60% of medium-sized neurons, whereas GluR5 and KA1 were restricted to 1–2% and 20–30% of these neurons, respectively. These results suggest that antagonists of AMPA and kainate receptors could be effective in alleviating motor symptoms in Parkinson's disease by blocking the overstimulation of pallidal and striatal neurons by glutamate. A significant decrease in GLuR1 gene expression (−33%) was observed in the neurons of the GPi in PD patients and in rat entopeduncular nucleus ipsilateral to the 6-OHDA lesion (−20%). GluR2, GluR3, and GluR4 mRNA levels in the GPi and GluR1–4 mRNA levels in the striatum were unchanged in PD patients and 6-OHDA-lesioned rats compared with their respective controls. These data suggest that dopamine positively regulates only GluR1 gene expression in the GPi. © 1996 Wiley-Liss, Inc.     

Intrathecal injection of GluR6 antisense oligodeoxynucleotides alleviates acute inflammatory pain of rectum in rats

Objective

To investigate whether the kainate (KA) receptor subunit GluR6 is involved in the acute inflammatory pain.

Methods

Formalin was injected into the mucosa of rectum in Sprague-Dawley rats to induce visceral pain. The antisense oligodeoxynucleotides (ODNs) of GluR6 were injected once per day for 3 d before formalin injection, after which GluR6 protein level was examined by immunoblotting method. The change of visceral pain was also investigated.

Results

The expression of GluR6 in the spinal cord of rats increased after the formalin injection. Moreover, pre-treatment of GluR6 antisense ODNs could suppress GluR6 expression in the spinal cord of rats and decrease the scores of visceral pain at 45 min following formalin injection.

Conclusion

Kainate receptor subunit GluR6 plays an important role in the visceral pain induced by injection of formalin into the wall of rectum. GluR6 may serve as a potential target for the treatment of acute inflammatory visceral pain.     

Expression of ionotropic glutamate receptor subunit mRNAs in the hypothalamic paraventricular nucleus of the rat

The hypopthalamic paraventricular nucleus (PVN) coordinates multiple aspects of homeostatic regulation, including pituitary-adrenocortical function, cardiovascular tone, metabolic balance, fluid/electrolyte status, parturition and lactation. In all cases, a substantial component of this function is controlled by glutamate neurotransmission. In this study, the authors performed a high-resolution in situ hybridization analysis of ionotropic glutamate receptor subunit expression in the PVN and its immediate surround. N-methyl-D-aspartate (NMDA) receptor 1 (NMDAR1), NMDAR2A, and NMDAR2B mRNAs were expressed highly throughout the PVN and its perinuclear region as well as in the subparaventricular zone. NMDAR2C/2D expression was limited to subsets of neurons in magnocellular and hypophysiotrophic regions. In contrast with NMDA subunit localization, AMPA (alpha-amino-3-hydroxy-5-methyl-4-isoxazole-propionate)-preferring and kainate (KA)-preferring receptor subunit mRNAs were expressed heterogeneously in the PVN and surround. Glutamate receptor 1 (GluR1) mRNA labeling was most intense in preautonomic subregions, whereas GluR2, GluR4, GluR5, and KA2 were expressed in hypophysiotrophic cell groups. It is noteworthy that GluR5 mRNA expression was particularly robust in the dorsolateral region of the medial parvocellular PVN, suggesting localization in corticotropin-releasing hormone neurons. All four AMPA subunits and GluR6 and GluR7 mRNAs were expressed highly in the perinuclear PVN region and the subparaventricular zone. These data suggest the capacity for multifaceted regulation of PVN function by glutamate, with magnocellular neurons preferentially expressing NMDA subunits, preautonomic neurons preferentially expressing AMPA subunits, and hypophysiotrophic neurons preferentially expressing KA subunits. Localization of all species in the perinuclear PVN suggests that glutamate input to the immediate region of the PVN may modulate its function, perhaps by communication with local gamma-aminobutyric acid neurons.     

Differential regulation of ionotropic glutamate receptor subunits following cocaine self-administration

Previous examination of binge cocaine self-administration and 2 week withdrawal from cocaine self-administration on ionotropic glutamate receptor subunit (iGluRs) protein levels revealed significant alterations in iGluR protein levels that differed between the mesocorticolimbic and nigrostriatal pathways. The present study was undertaken to extend the examination of cocaine-induced alterations in iGluR protein expression by assessing the effects of acute withdrawal (15-16 h) from limited access cocaine self-administration (8 h/day, 15 days). Western blotting was used to compare levels of iGluR protein expression (NR1-3B, GluR1-7, KA2) in the mesolimbic (ventral tegmental area, VTA; nucleus accumbens, NAc; and prefrontal cortex, PFC) and nigrostriatal pathways (substantia nigra, SN and dorsal caudate-putamen, CPu). Within the mesolimbic pathway, reductions were observed in NR1 and GluR5 immunoreactivity in the VTA although no significant alterations were observed in any iGluR subunits in the NAc. In the PFC, NR1 was significantly upregulated while GluR2/3, GluR4, GluR5, GluR6/7, and KA2 were decreased. Within the nigrostriatal pathway, NR1, NR2A, NR2B, GluR1, GluR6/7 and KA2 were increased in the dorsal CPu, whereas no significant changes were observed in the SN. The results demonstrate region- and pathway-specific alterations in iGluR subunit expression following limited cocaine self-administration and suggest the importance for the activation of pathways that are substrates of the reinforcing and motoric effects of cocaine.     

Selective reduction of GluR2 protein in adult hippocampal CA3 neurons following status epilepticus but prior to cell loss

Kainic acid (KA) induces status epilepticus and delayed neurodegeneration of CA3 hippocampal neurons. Downregulation of glutamate receptor 2 (GluR2) subunit mRNA [the α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA) subunit that limits Ca²⁺ permeability] is thought to a play role in this neurodegeneration, possibly by increased formation of Ca²⁺ permeable AMPA receptors. The present study examined early hippocampal decreases in GluR2 mRNA and protein following kainate-induced status epilepticus and correlated expression changes with the appearance of dead or dying cells by several histological procedures. At 12 h, in situ hybridization followed by emulsion dipping showed nonuniform decreases in GluR2 mRNA hybridization grains overlying morphologically healthy-appearing CA3 neurons. GluR1 and N-methyl-D-aspartate receptor mRNAs were unchanged. At 12–16 h, when little argyrophilia or cells with some features of apoptosis were detected by silver impregnation or electron microscopy, single immunohistochemistry with GluR2 and GluR2/3 subunit-specific antibodies demonstrated a pattern of decreased GluR2 receptor protein within CA3 neurons that appeared to predict a pattern of damage, similar to the mRNA observations. Double immunolabeling showed that GluR2 immunofluorescence was depleted and that GluR1 immunofluorescence was sustained in clusters of the same CA3 neurons. Quantitation of Western blots showed increased GluR1:GluR2 ratios in CA3 but not in CA1 or dentate gyrus subfields. Findings indicate that the GluR1:GluR2 protein ratio is increased in a population of CA3 neurons prior to significant cell loss. Data are consistent with the “GluR2 hypothesis” that reduced expression of GluR2 subunits will increase formation of AMPA receptors permeable to Ca²⁺ and predict vulnerability to a particular subset of pyramidal neurons following status epilepticus. Hippocampus 1998;8:511–525. © 1998 Wiley-Liss, Inc.     

Differential expression of kainate receptors in the basal ganglia of the developing and adult rat brain

Glutamate is the principal excitatory transmitter of the mammalian brain and plays a particularly important role in the physiology of the basal ganglia structures responsible for movement regulation. Using in situ hybridization with oligonucleotide probes, we examined the expression patterns of the five known kainate type glutamate receptor subunit genes, KA1, KA2 and GluR5–7, in the basal ganglia of adult and developing rat brain. In the adult rat, a highly organized and selective pattern of expression of the kainate subunits was observed in the basal ganglia and associated structures as well as in other regions of the brain. KA2 mRNA was abundant in the striatum, nucleus accumbens, subthalamic nucleus and substantia nigra pars compacta, and was present at lower levels in the globus pallidus and substantia nigra pars reticulata. Neither KA1 nor GluR5 expression was observed in the basal ganglia of adult rats, although these messages were present in other regions. GluR6 was highly expressed in the striatum and subthalamic nucleus and to a lesser extent in the substantia nigra pars reticulata, while no hybridization signal was detectable in the large, presumably dopaminergic neurons of the substantia nigra pars compacta. In contrast, GluR7 was strongly expressed in the substantia nigra pars compacta, was present at lower levels in the striatum, globus pallidus and substantia nigra pars reticulata, and was not detectable in the subthalamic nucleus. During postnatal development, expression of the kainate receptor subunits was characteristically highest on postnatal day 1 and declined to adult levels by day 20; however, in the globus pallidus we did observe the transient expression of KA1 and GluR5 between day 1 and day 10. These results demonstrate that the neuronal structures comprising the basal ganglia express a distinct combination of kainate receptor subunit genes, suggesting that the pharmacological properties of the resultant glutamate receptors are likely to be regionally specific. The organization of expression of these genes is established early in life, which is consistent with the important role they may play in establishing the functions of the motor system.     

Localization of kainate receptors at the cone pedicles of the primate retina.

In the macaque monkey retina cone pedicles, the output synapses of cone photoreceptors, contain between 20 and 45 ribbon synapses (triads), which are the release sites for glutamate, the cone transmitter. Several hundred postsynaptic dendrites contact individual cone pedicles, and we studied the glutamate receptors expressed and clustered at these contacts, particularly the kainate receptor subunits GluR5, GluR6/7, and KA2. Pre- and postembedding immunocytochemistry and electron microscopy were used to localize GluR5 and GluR6/7 to specific synaptic contacts at the cone pedicle base. The GluR5 subunit was aggregated at bipolar cell flat contacts. The GluR6/7 subunit was aggregated at bipolar cell flat contacts and at the desmosome-like junctions formed by horizontal cell processes underneath the cone pedicles. KA2 immunoreactivity was observed at the invaginating dendritic tips of ON-cone and rod bipolar cells, which we interpret as a cross-reactivity of the KA2 antiserum with some other, unknown protein of the monkey retina. Kainate receptors are preferentially expressed by OFF-cone bipolar cells and to a lesser extent by horizontal cells. We also performed double-labeling experiments with the ribbon-specific marker bassoon and with antibodies against GluR5 and GluR6/7 in order to define the position of the flat bipolar cell contacts with respect to the triads. There was a tendency of GluR6/7 clusters to represent triad-associated contacts, whereas GluR5 clusters represented non-triad-associated contacts. The GluR5 and GluR6/7 subunits were clustered at different bipolar cell contacts. We studied a possible cone-selective expression of the kainate receptor subunits by double labeling cone pedicles for the S-cone opsin and for the different receptor subunits. We observed a reduced expression of both GluR5 and GluR6/7 at the S-cone pedicles. The reduced expression of GluR6/7 was analyzed in more detail and it appears to be a consequence of a horizontal cell-specific expression: H1 horizontal cells express GluR6/7, whereas H2 horizontal cells, which preferentially innervate S-cones, show no expression of GluR6/7.