kainate receptor-operated channels with high and low Ca2+ permeability in single rat retinal ganglion cells

Neuroscience Vol. 67, No. 1, pp. 177 188, 1995

Pergamon

0306-4522(94)00627-X

Elsevier ScienceLtd Copyright ~) 1995 IBRO Printed in Great Britain. All rights reserved 0306-4522/95 $9.50 + 0.00

CO-EXPRESSION OF AMPA/KAINATE RECEPTOR-OPERATED CHANNELS WITH HIGH A N D LOW Ca 2+ PERMEABILITY IN SINGLE RAT RETINAL GANGLION CELLS D. Z H A N G , * N. J. S U C H E R * t and S. A. L I P T O N { *Laboratory of Cellular and Molecular Neuroscience, Department of Neurology, Children's Hospital, and {Departments of Neurology, Beth Israel Hospital, Brigham and Women's Hospital, Massachusetts General Hospital, and Program in Neuroscience, Harvard Medical School, Boston, MA 02115, U.S.A. Abstract--The patch-clamp technique was used to record whole-cell currents induced by ~-amino-3hydroxyl-5-methyl-isoxazol-4-propionic acid (AMPA) or kainate in solitary rat retinal ganglion cells (n = 125) in vitro. Two groups of retinal ganglion cells could be distinguished according to their responses to kainate or AMPA in extracellular solutions with Ca 2÷ as the only permeant cation. The ratio of the steady-state currents evoked by a given concentration of AMPA compared to kainate was low (0.08) in the first group and high (0.61) in the second group of retinal ganglion cells. The Ca 2+ permeability through AMPA/kainate receptor-operated channels was low (Pca,./Pcs+ < 0.1) in the first group (n = 74, 59%) and moderate (Pca~ +/Pcs- = 0.53) in the second group (n = 51,41%) of retinal ganglion cells. The fraction of the total current induced by stimulation of non-N-methyl-D-aspartate receptors that is flowing through Ca 2+ permeable AMPA/kainate channels in single cells with high Ca 2+ permeability was estimated by comparing the current voltage relationship in extracellular solutions with either Ca 2+ or Na + as the sole charge carrier. The contribution of Ca2+-permeable channels to the non-N-methyl-t)-aspartate receptor induced whole-cell current in single C a 2+ permeable cells (n = 12) ranged from 40 to 70%, correlating with the intermediate level of Ca 2+ permeability (Pca2+/Pcs + = 0.22q).80) measured by an independent method in these cells. Thus, single CaZ+-permeable cells appear to express at least two types of AMPA/kainate receptor-operated channels with high or low Ca 2+ permeability. Using the polymerase chain reaction, transcripts for the glutamate receptor subunits 1-4, including their "flip" and "flop" versions, were identified in retinal ganglion cells. Together, these findings suggest that among rat retinal ganglion cells there are differences in the pattern of expression of AMPA/kainate receptor-operated channels. Moreover, individual cells co-express multiple heterologous non-N-methyl-D-aspartate receptors with distinct functional properties. The functional diversity of these receptors may play an important role in controlling Ca 2+ entry into neurons. We speculate that the low Ca 2+ permeability and the preference for kainate in one group of retinal ganglion cells may be due to the predominant expression of non-N-methyl-D-aspartate receptors containing the edited form of the glutamate receptor subunit 2 flop splice variant.

Glutamate is thought to be the major excitatory neurotransmitter in the CNS. Glutamate receptor (GluR)-operated ion channels mediate fast cellular information signaling between neurons and are important for processes underlying neuronal plasticity, outgrowth and survival. 8'3L33 Consequently, GluR-operated channels are mediators not only of normal intercellular communication but also of neuronal injury and death. The ionotropic glutatTo whom correspondence shonld be addressed. Abbreviations: AMPA, ~-amino-3-hydroxyl-5-methyl-isoxazol-4-propionic acid; D-AP5, D-2-amino-5-phosphonovalerate; CNQX, 6-cyano-7-nitroquinoxaline-2,3dione; EGTA, ethylene glycol bis (amino ethylether) tetro-acetate; GluR, glutamate receptor; HCP, retinal ganglion cells with high Ca 2+ permeability; HEPES, N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; LCP, retinal ganglion cells with low Ca 2÷-permeability; NMDA, N-methyl-D-aspartate; PCR, polymerase chain reaction; RGC, rat retinal ganglion cell.

mate receptors have been classified according to their preferred agonists as N-methyl-D-aspartate ( N M D A ) , ~-amino-3-hydroxyl-5-methyl-isoxazol-4propionic acid (AMPA) and kainate subtypes. 44,52 The latter two types are also referred to as nonN M D A receptors since the functional properties between the two groups are not totally distinct. It has been shown that N M D A receptors, whose channels are highly permeable to Ca 2+, play key roles in plasticity, e.g. long-term potentiation and neurite outgrowth. 9'3~ However, recent evidence suggests that n o n - N M D A receptors may also play important roles in these phenomena, at least in some C N S neurons. 34 Of particular importance is the finding that some n o n - N M D A receptors are permeable to Ca 2 4. Since the intracellular Ca 2 ÷ levels help determine the functional state of a particular neuron, and excessive intracellular Ca 2÷ levels may contribute to neurotoxicity, it is very important to characterize the 177

178

D. Zhang et al.

Ca 2+ permeability on non-NMDA receptors in specific neurons from various areas of the CNS. Studies with expressed recombinant receptors suggest that the functional properties of the GluR-operated channels critically depend on their molecular structure. Ca 2+ permeability of the AMPA-preferring receptors is determined by a single amino acid residue [arginine (R) vs glutamine (Q)] in the second membrane segment of the receptor subu n i t s . 7'l°A2A3,19,21,27'43,5° When glutamine is present at this critical site, as in GluR1, 3 and 4, the permeability ratio of Ca 2+ vs Cs ÷ (Pca2+/Pcs+) is around 1.2. This ratio drops to 0.05 if arginine replaces glutamine at this site, such as in the edited form of the GluR2 [GluR2(R)]. When several different subunits are co-expressed, the GluR2(R) subunit plays a dominant role in controlling Ca: ÷ permeability, since cells with 1:1 expression of GIuR2(R) vs other subunits [GluR1, 3, 4 or the unedited form of GIuR2, GIuR2(Q)] contain channels with very low Ca 2+ permeability.7'19 Furthermore, with an expression ratio of 1:10 for GluR2(R) vs other subunits, half of the cells show an intermediate level of Ca 2+ permeability (Pca2+/Pcs+ = 0.47). Assuming that the assembly of channels depends only on the relative amounts of each subunit, these studies suggest that a non-NMDA receptor operatedchannel becomes barely permeable to Ca 2+ even if only one of the ~five subunits of an individual receptor is GIuR2(R). Another property of non-NMDA receptors that is dependent on the molecular structure of their subunits is the amplitude of the steady-state currents evoked by glutamate (or AMPA) vs kainate. The relative magnitude of these steady-state currents appears to be determined by alternative splicing of the transcripts. The presence of either the "flip" or "flop" splice variant between the third and fourth transmembrane segments of the subunits influences the ratio of the agonist-induced currents. 42The "flop" version of GluR2 subunits forms channels with a small sustained response to glutamate (or AMPA) compared to that induced by kainate. On the other hand, glutamate (or AMPA) activates the "flip" version of these channels four to five times more effectively than kainate. A similar difference has also been found between the "flip" and the "flop" versions of other subunits (GIuR1, 3, 4). Recombinant AMPA/kainate receptor-operated channels obtained by pair-wise co-expression of the various subunits differ in their biophysical and pharmacological properties. This finding suggests either formation of heteromeric channels or cellular mixing of two types of homomeric channels.7 Given that the distribution of different non-NMDA receptor subunits is distinct in different brain a r e a s , 4'5'11'17'22'26'35'51 it is conceivable that native AMPA/kainate receptoroperated channels consist of various mixtures of homomers and heteromers. In fact, Ca 2+-permeable AMPA/kainate receptor-operated channels have

been identified in several areas of the CNS including retina, although the degree of Ca 2+ permeability of these channels appears to be different in neurons from various regions. 6'13'14'23'3°'38'39'41Similarly, it has been shown that the ratio of sustained currents evoked by glutamate (or AMPA) vs kainate is different between neurons in the CA1 and CA3 regions of hippocampus. 24However, relatively little quantitative information regarding the heterogeneity of AMPA/ kainate-operated channels in identified neurons is available, and it is not yet known how the subunit composition of non-NMDA channels relates to the functional properties of primary neurons. Transcripts for all subunits of the AMPA family plus the GluR5-7 subunits have been found in rat retinal ganglion cells (RGCs) by in situ hybridization. Interestingly, individual RGCs appear to co-express transcripts for multiple subunits. 15'2°,37We conducted this electrophysiological and molecular biological study in order to investigate if the co-expression of multiple AMPA/kainate receptor subunit gene mRNAs might be correlated with specific functional properties of non-NMDA receptors in single RGCs.

EXPERIMENTAL PROCEDURES

Preparation of retinal ganglion cells The experimental procedures for RGC labeling with the fluorescent dye Granular Blue, followed by dissociation from the retina, and placement in culture have been described in detail previously by this laboratoryfl9,32.47 Neonatal Lon~Evans rats (postnatal days 8-12; Charles River Breeding Laboratory, Boston, MA) were used in these experiments. Briefly,following retrograde labeling of RGCs in situ, retinas were dissected and enzymatically dissociated in Hanks' balanced salt solution with papain. Dissociated retinal cells were plated onto poly-L-lysine-coated glass coverslips in 35 mm tissue culture dishes that contained 2 ml of Eagle's minimum essential medium supplemented with 1/~g/ml gentamicin, 16 mM glucose, 2 mM glutamine and 5% (v/v) rat serum. Retinal cultures were incubated in a humidified atmosphere of 5% CO2 in air at a temperature of 36°C. RGCs maintained in culture dishes for 3 24 h were used for physiological recordings. Electrophysiological experiments Whole-cell recordings with patch electrodes were performed on RGCs using standard procedures,t't6'47The cultures were continuously perfused (~0.8 ml/min) at room temperature. Three types of bath solutions were used in the experiments: (i) control saline: based on Hanks' balanced salts (in mM): NaC1, 137.6; NaHCO3, 1; Na2HPO4, 0.34; KCI, 5.36; KH2PO4, 0.44; CaC12,2.5; glucose, 22.2; glycine, 0.001, HEPES, 5; with Phenol Red indicator (0.001% v/v), and adjusted to pH 7.2 with 0.3 N NaOH; (ii) Na+-saline: NaC1, 140; glucose, 22.2; HEPES, 5; with Phenol Red indicator (0.001% v/v), and adjusted to pH 7.2 with 0.3 N NaOH; (iii) Ca2+-saline: CaCI2 (0-70 mM) and N-methylD-glucamine (0-140mM) were used to replace NaC1 isotonically in Na+-saline, which was adjusted to pH 7.2 with 1N HC1. All drugs were prepared by dissolving concentrated stock solutions into bath solution and were then applied with pneumatic pipettes. Voltage protocols were administered once the drug-induced responses had reached a steady-state level and thus only the sustained currents were studied. The patch electrodes were made of hard glass and had resistance of 2 4 MfL The patch electrode-

Calcium permeability of GluRs in retinal ganglion cells

179

filling solution contained (in mM): CsC1, 120; tetraethylammonium chloride, 20; MgC12, 2; CaCI2, I; E(iTA, 2.25; HEPES-NaOH I0 (adjusted to pH 7.2).

Data analysis The reversal potential of each RGC was obtained by determining the interception of the current trace with the voltage axis during ramp voltage protocols. The permeability ratio of Ca 2÷ vs Cs ÷ for experiments employing a single Ca 2÷ concentration was calculated using the constant field equation: 23'24

Window

48"~

32" o ~,

Pc~2./Pc~. = ([Cs]i/4[Ca]o){exp(EF/RT) × [exp(EF/RT) + 11}, (1)

16"

or by the constant field voltage equation for experiments using multiple Ca 2+ concentrations: I

Erev = (RF/T)In{0.5[- 1

16

+ (1 + 16Pc,2+[Ca]o/Pcs+[Cs]i)ln]}.

(2)

For calculation of the percentage of high Ca 2+ permeable non-NMDA receptors in a given cell, the kainate-evoked currents were measured at - 5 0 mV in Ca 2+-saline (70 mM) and Na+-saline (140mM). The assumption is that, when measured at - 5 0 mV, the current evoked in Na+-saline is due to influx of Na + through all non-NMDA receptors in a given cell, whereas current evoked in CaZ+-saline is due to Ca 2+ influx through only those that are highly permeable to Ca 2+ . Any Ca z+ entering through voltagedependent channels will be subtracted in the ramp voltage protocol. With these assumptions, the following equations are valid: INa = NteNagNa(Vm -- Erev.Na)

(3)

lea = NcaPcagca(Vm - Erev.ca),

(4)

where INa and/ca are currents measured at Vm(mV) in Na ÷and Ca2+-saline, respectively. Er~,Na and Ere~.c, are reversal potentials of the kainate-induced currents (in units of mV), and PNagNa and Pc~gca are open probability times single channel conductance in Na ÷- and Ca 2÷ -saline, respectively. N t and Nc~ represent the total number of non-NMDA receptors and the number of high Ca2+-permeable nonNMDA receptors, respectively. By rearranging equations (3) and (4), we obtain the following equation:

Nca/ N t = (Ica/ INa)( PN~gNJ Pc~gc.) (VII I --

Erev,Na)/(V

m --

Erev,Ca ).

(5)

The currents and reversal potentials were measured in the experimental protocols. The value of the term (PN.gNJPcagca) was determined from data presented by Burnashev et al., 7 who used experimental conditions similar to ours except that a homogeneous population of recombinant channels with high Ca 2+ permeability was investigated. The Ps.gNJPc.gc. value was estimated by measuring Ic./IN~ at a holding potential of 100 mV and calculated by using equation (5) (with N~ = Nc~). The value of PNagN./Pc.gc. obtained in those studies was 0.35 for the unedited form of GluR2 (from Fig. 1E of Ref. 7) and 0.55 for GIuR4 (from Fig. 2B of Ref. 7). We used the value of 0.35 for PNagNa/Pcagca in our calculations to avoid overestimation of high Ca2+-permeable channels in RGCs.

Purification o f R N A and reverse transcription After enzymatic dissociation, retinal cells were resuspended in Hanks' balanced salt solution at 0.2~).5 x 106 cells/ml. Cells were subjected to flow cytometric analysis using a FACS IV sorting instrument equipped with a 6 W argon-ion laser (Innova 90-6), adjusted to emit at 351 and 364 nm lines. Emission spectra of RGCs labeled with the fluorescent dye Granular Blue were analysed after passing through a 460 nm long-pass filter. RGCs were selected according to both size and intensity of labeling with fluor-

I

32

I

48

Relative Cell Size Fig. 1. Enrichment of RGCs by fluorescence-activated celt sorter analysis. RGCs were sorted according to their relative size (abscissa) and intensity of labeling with the fluorescent dye, Granular Blue (ordinate). The presence of fluorescence signal outside of the box represents small labeled cells and non-specific background that is caused by dye-positive cell debris. Cells inside the window were selected and collected in Hanks' balanced salt solution. In this representative experiment, cells from the retinas of two rats (postnatal day 10) were dissociated, and 2.5% of the total cells were selected. After sorting, cells maintained their integrity and could be used for cell culture and electrophysiologieal experiments.

escent dye (Fig. 1). After sorting, cells maintained their integrity and could be used for cell culture and electrophysiological experiments. Using these procedures, approximately 2.5% of total retinal cells were isolated in three sorting experiments (two to four rats were used for each sorting experiment), and their total RNA was isolated and reverse transcribed as detailed previously. 46

Polymerase chain reaction amplification o f glutamate receptor subtype 1-4 fragments Two rounds of polymerase chain reaction (PCR) experiments were conducted as described previously by Lambolez and co-workers with some modificationsY Briefly, the first round of PCR was aimed at amplification of mixed fragments of GluR1-4 cDNAs of both flip and flop forms. The PCR reaction mix contained: 10 mM Tris-HC1, 50 mM KC1, l mM MgC12, 0.1% gelatin, 200/~M dNTP, 2 # M forward and reverse primers, 2.5 U/100 #1 Taq Polymerase and template (10 times dilution of Reverse transcription product). Reactions were performed in a thermocycler for 35 cycles (95°C, 1 min; 49°C, 1 min; 7Z'C, 1 min). The products of the first PCR were then diluted 1000 times and used as templates for the second round of PCR to amplify specifically the flip and flop forms of GIuRI~I. We designed two additional primers for specific amplification of GluR1 flop, GluR3 flip and GluR3 flop. The new Rlflop primer (TCAATTTGTCCAAAAGCCCC, corresponding to bases 2329 2310 of the GluR1 flop cDNA) was used with the RI primer (as described by Lambolez and co-workers) to amplify a fragment of the GluR1 flop gene. The new R3 primer (GCCTCGTGACCCACAAAGCC, corresponding to bases 1749-1768 of the GIuR3 cDNA) was used with either the R3flop or the R3flip primer (as described by Lambolez and co-workers). The reaction conditions for the second PCR were the same as for the first, except the annealing temperature was increased from 49 to 52°C. Twenty microliters of each PCR product were analysed by 1.5% agarose gel electrophoresis. The identity of the PCR

D. Zhang et al.

180

products obtained in the second round was confirmed by DNA sequencing. RESULTS

Ca 2+ permeability of AMPA /kainate receptoroperated channels defines two types of retinal ganglion cells The possible heterogeneity of AMPA/kainate receptor-operate d channels in RGCs was investigated by recording whole-cell currents evoked by kainate in an extracellular solution with Ca 2+ as the only permeant cation. With this CaZ+-saline as the extracellular solution, application of kainate (250/~M) activated an inward current in a subpopulation of RGCs at a holding potential of - 6 0 mV (Fig. 2). Given that Ca 2+ was the only permeant catiOn in the external solution, the inward current observed had to be carried by Ca 2+ . Qualitatively similar results were obtained in parallel Ca 2+imagining experiments using the Ca 2+-sensitive fluorescent dye fura-2 (data not shown). In order to rule out the possibility that the kainate-evoked currents were due to non-specific activation of other Ca 2+permeable channels (such as those operated by NMDA receptors), we tested the ability of specific antagonists to block the evoked current. Recordings of RGCs (n = 5) in Ca 2+-saline at 0 and - 6 0 mV showed that kainate-evoked currents were completely blocked by the non-NMDA antagonist 6-cyano-7nitroquinoxaline-2,3-dione (CNQX, 20/~M), 45 but

were not affected by the specific NMDA antagonist D-2-amino-5-phosphonovalerate (D-AP5, 200ktM; Fig. 2). For a more quantitative analysis of the observed differences in the Ca 2+ permeability of AMPA/ kainate receptor-operated channels in RGCs, the reversal potential of kainate-evoked currents in single neurons was measured in the presence of different concentrations of extracellular Ca 2+ . The membrane potential was held at 0mV and a ramp voltage pulse (0 to - 1 2 0 mV, 0.2 mV/ms) was applied). Five to 10 such current responses in the presence and absence of kainate were averaged, and the kainate-evoked currents were obtained by subtraction of the resulting traces. Figure 3A and B illustrates kainate-induced responses from two different RGCs that displayed outward rectification in Ca 2+ -saline. The reversal potentials of the kainateevoked responses were shifted to more depolarized values when Ca 2+ concentrations were increased in the saline. However, the magnitude of the shift varied considerably from cell to cell (e.g., compare Fig. 3A and B). The differences observed in the reversal potentials of different RGCs are most likely explained by different Ca 2+ permeabilities of the non-NMDA receptors in those cells. For the two cells illustrated in Fig. 3A and B, the relative permeabilities of Ca 2+ vs Cs + (Pc,2+/Pcs+, calculated from the constant field equation by using reversal potentials observed in 30 mM Ca 2+ saline) were 0.04 and 0.65, respectively.

250 pM Kainat~

VH = 0 mV

VH = -60 mV

+ ZO wM CNQX

IlOOp^

[ 5op^

983 ms

983 ms

Fig. 2. Ca2+ permeates kainate-activated channels in RGCs. The inward and outward currents evoked by kainate (250 gM in 10 mM Ca2+-saline) are shown at holding potentials of - 6 0 and 0 mV. The currents evoked at both holding potentials were blocked by CNQX (20#M), a relatively selective non-NMDA receptor antagonist in the presence of glycine, but not by D-AP5 (200 #M), a selective NMDA receptor antagonist. These findings indicate that the inward current (presumably carried by Ca2+) evoked by kainate was due to specific activation of kainate-operated receptor channels.

181

Calcium permeability of GluRs in retinal ganglion cells

Analysis of a population of RGCs (n = 60), tested in 70mM Ca2+-sallne, revealed two major types of RGCs with respect to the Ca 2÷ permeability of

their non-NMDA receptors (Fig. 3C). The distribution of the measured reversal potentials was fitted with two Gaussian components. One group

B

A I (pA)

I (pA)

0 Ca2+ 1 ~

V (mV) ~lmmmlm~'*~"~"

I

I

-100

I

-60

,

I

~

,ttfl

-20

-500

-125 30 Ca 2+

D

C

O-

20

f - / " ~ ( 9 )

~- -20-

15 m

L~ .-q

~ -41110

=

--~ -60"

z ,~ -80"

-80

-60

-40

-20

Reversal Potential (mV)

0

20

411

60

Calcium C~mcentration (raM)

Fig. 3. Differential Ca 2+ permeabilities through n o n - N M D A receptors separate RGCs into two subclasses. The permeability ratio of Ca 2÷ vs Cs ÷ (Pca2+/Pc~+) for kainate receptor-operated channels was obtained in different extracellular Ca 2+ concentrations. Ramp voltage protocols (0 to - 1 2 0 m V , 0.2 mV/ms) were given before (representing the leak current) and during application of kainate (250/~ M) to determine the reversal potential of the agonist-~voked current. The mean current of I0 traces in each group was subtracted to yield kainate-evoked curre'nts. Examples of recordings are shown for two types of neurons, an LCP RGC in A and an HCP RGC in B (leak current has already been subtracted). A progressive positive shift in the reversal potential for kainate-activated current was observed with increasing extracellular Ca 2+ concentration in both RGCs. The shift, however, was much smaller for the LCP neuron (A) than for the HCP neuron (B). The small inward current negative to the reversal potential in the presence of N-methyl-D-glucamine is due to the low permeability of this substance through these c h a n n e l s / ( C ) Distribution of the reversal potentials of the kainate-evoked currents recorded in 70 mM extracellular Ca 2 ÷ -saline. The reversal potential for each cell (binned at 10 mV intervals) is plotted vs the number of cells. The distribution of the reversal potentials was fitted with two Gaussian components. (D) Reversal potentials of kainate-activated currents for a group of HCP cells are plotted against Ca 2÷ concentration o f the extracellular solution. Values represent mean + S.D. The smooth curve was fitted using the constant field voltage equation with a Pc~,2+/Pcs+value of 0.534 and coefficient of variation of 14%.

80

182

D. Zhang et al.

(n = 30) of RGCs had a very low Ca 2+ permeability (Pca2+/Pcs+ <0.09), with a reversal potential of --64.67_+ 14.89 mV (mean_+ S.D.). The second group (n =30) had a higher Ca 2+ permeability (Pca2+/Pc~+>0.22), with a reversal potential of -15.57 _+ 7.32 inV. The reversal poteltials of these two groups were significantly different (t-test, P < 0.05). The RGCs with high and low Ca 2+ permeability will be referred to as HCP and LCP cells, respectively. Similar differences in Ca 2+ permeability of n o n - N M D A receptors between the two groups were also observed in external solutions with lower Ca 2+ concentrations (2.5-30mM). Of 125 cells analysed in various Ca 2+-salines, 51 cells (41% were classified as HCP cells and 74 cells (59%) as LCP cells. No correlation was found between the Ca 2+ permeability of the non-NMDA receptors in these cells to their size, shape or age under our experimental conditions. By recording in different Ca 2+ concentrations, it was possible to fit the data obtained in HCP cells using a modified form of the constant field equation (weighted with the reciprocal of the standard deviation of each data point). This yielded a mean Pca~+/Pc~+ value of 0.53 (Fig. 3D). At higher Ca 2+ concentrations, the data points began to deviate from the fitted line (coefficient of variation = 14%). This may indicate that, at higher Ca 2+ concentrations, the permeation of ions through these channels deviates from the independence principle due to increasing repulsion between ions and membrane surface charge. 3'~8It is also predicted from the fitted line that the permeability ratio tended to be underestimated if it was calculated using higher Ca 2+ concentrations, and vice versa. The PCa~+/Pcs+ value for non-NMDA receptors in HCP cells was intermediate compared to that formed by recombinant GIuRI, 3 or 4 subunits in expression systems (Pc,2+/Pc~+ about 1.2),7 suggesting that these RGCs might have a mixture of non-NMDA receptors with high and low Ca 2+ permeability. LCP cells, in contrast, might predominantly express receptors containing the edited form of the GluR2 subunit, since their Pcaz+/Pcs+ value ( < 0.09) was close to that of recombinant GIuR2(R) receptors (Pc,~+/Po+ = 0.05). 7

Co-expression of A M P A /kainate receptors with high and low Ca 2+ permeability The foregoing experiments suggested that HCP cells co-express AMPA/kainate receptors with high and low Ca 2+ permeability. The following experiment was designed to obtain a more quantitative estimate of the contribution of high Ca 2+-permeable AMPA/kainate receptor-operated channels to the kainate-induced current in individual RGCs. This was accomplished by comparing kainate-evoked inward currents of a given cell in Ca 2+- and Na +saline (see Experimental Procedures). We assumed that under our conditions the inward current seen in Ca 2+-saline was due largely to Ca 2+ flowing through

the high Ca 2+-permeable channels, while the inward current seen in Na+-saline was due to Na + flowing through both high and low CaZ+-permeable channels. Examples of recordings are shown for an LCP (Fig. 4A) and an HCP cell (Fig. 4B), each cell tested in 70mM Ca 2+- and 140mM Na+-saline. In Na +saline, the reversal potentials were similar and close to 0 mV for both RGC types, suggesting that the permeability of Na + through high and low Ca 2+permeable channels was similar. In contrast, the reversal potential in Ca2+-saline was shifted to a more depolarized level for an HCP cell ( - 12 mV) compared with an LCP cell ( - 53 mV). By measuring currents in both Ca 2+- and Na+-salines with the membrane potential clamped to - 50 mV, to limit the contribution of Ca 2+ current through low Ca 2+permeable channels, the fraction of whole-cell current flowing through Ca 2+-permeable channels in a given HCP cell could be calculated (see Experimental Procedures). In addition, the Ca 2+ permeability of the channels in these cells was estimated independently by applying the constant field equation using the reversal potentials measured in Ca 2+ saline for each cell. As shown in Fig. 4C, the percentage of current carried through Ca2+-permeable channels ranged from 40 to 70% in a group of HCP cells (n = 12), correlating well with their intermediate level of Ca 2+ permeability (Pca2+/Pcs+ = 0.22--0.80). Without knowledge of the single channel conductance and opening probabilities of the various channel types, the numbers of each channel type present in the cell membrane cannot be estimated. An important assumption underlying our calculation of the contribution of channels with high Ca 2+ permeability to the whole-cell currents is that any possible effect of the extracellular Ca 2+ concentration on the gating of recombinant non-NMDA channels is similar to that in native channels in RGCs. If such an effect were significantly different between the two systems, this would lead to an over- or underestimation of the actual contribution of the non-NMDA receptor channels with high Ca 2+ permeability to the wholecell current under physiological conditions. Even taking this possibility into account, our results clearly suggest that HCP cells co-express at least two types of AMPA/kainate operated channels with high and low Ca 2+ -permeability.

Ratio of A M P A vs kainate evoked steady-state currents correlates with Ca 2+ permeability in high-Ca 2+permeable and low-Ca 2+-permeable retinal ganglion cells Another interesting difference among functional properties of disparate non-NMDA receptors is the ratio of AMPA- vs kainate-evoked sustained currents. Variation in this ratio occurs because of differences in the molecular structure generated by the two alternative splice variants (flip and flop) of each subunit.42 We thus investigated the sustained currents evoked by saturating concentrations of

Calcium permeability of GluRs in retinal ganglion cells

183

B

A I (pA)

I (pA)

1000"-

333

V (mY)

V (mY) ,

j

3

-333

-

"

C 9080-

O

70-

~ Q,I

~ +

O

O

60o

5O 40

o

o

30

20

0.0

6.2

6.4

6.6

6.8

1.0

P e r m e a b i l i t y Ratio (Ca2+/Cs+) Fig. 4. Single RGCs have a mixture of high- and low-Ca2+ permeable non-NMDA receptor-operated channels. Kainate-evoked currents were studied in Na+-saline (140 mM) and Ca 2+-saline (70 raM) using ramp voltage protocols (0 to -120 mV, 0.2 mV/ms). The reversal potential in Ca2+-saline was - 8 mV for a representative HCP cell, illustrated in B, and - 5 3 mV for a typical LCP cell shown in A. The reversal potential in Na+-saline was near 0 mV for both cell types. (C) For all RGCs studied, the percentage of high-Ca2+ -permeable channels in a given cell (left ordinate) is plotted against the Pca2*/Pcs* value of the non-NMDA receptors (abscissa). The percentage of high-Ca2÷-permeable channels in a given RGC neuron was calculated based on the amplitude of currents evoked by kainate in Na ÷- and Ca2+-salines at - 5 0 mV (see Experimental Procedures). The Pca2+/Pcs + value was calculated for each RGC using the reversal potential and the constant field equation. A M P A (500/zM) vs kainate ( 2 m M ) in RGCs. By applying ramp voltage pulses and recording the ligand-evoked currents, the reversal potential and the amplitude of the evoked currents over a range of potentials could be measured. Since the recordings were made in high Ca z+-saline (70 mM), the Ca 2+

permeability of n o n - N M D A receptors for a given R G C could also be determined to identify it as an LCP or HCP cell. As shown in Fig. 5A, even at positive holding potentials A M P A evoked only a small current in an LCP cell (reversal potential - 58 mV), while kainate elicited a considerably larger

184

D. Zhang et al.

current in the same cell. In contrast, A M P A a n d k a i n a t e evoked currents o f c o m p a r a b l e m a g n i t u d e in H C P cells (reversal potential - 1 2 mV; Fig. 5B). A least-squares fit to the amplitudes of the ligand-gated currents, m e a s u r e d a t + 1 0 0 m V , showed t h a t the

ratio of A M P A - to kainate-evoked currents was 0.08 for L C P cells (n = 10) and 0.62 for H C P cells (n = 8) (Fig. 5C). This result indicated that, besides the difference in the Ca 2+ permeability, the nonN M D A receptors in R G C s were also different in their

A

B I (pA)

I (pA) KA

KA

/

666-

250

i -40

I

.

.

.

.

- -'"i

-2.--

-r'

!

"" "-" !

=~-= I

60

i

~

I

"¢

V (mV)

I

I

60

I

I

V (mY)

-666

-250

C 700 600 <

5OO

~,

400

<

300

<

200 100 0

0

200

400

600

800

1000

Kainate Current (pA) Fig. 5. The ratio of AMPA- vs kainate-evoked currents is correlated to the Ca 2+ permeability of non-NMDA receptor-operated channels in the two RGC subclasses. Sustained currents evoked by AMPA (500 # M) and kainate (2 mM) in RGCs were recorded in Ca 2+ -saline (70 mM Ca 2+) using ramp voltage protocols ( - 6 0 to + 120 mV, 0.3 mV/ms), (A) The amplitude of AMPA-evoked current was much smaller than that evoked by kainate in a representative LCP cell (Ca 2+ permeability determined by very negative reversal potential). (B) In contrast, in a typical HCP cell (reversal potential, - 12 mV), the AMPA-evoked current was almost as large as that of the kainate-evoked current. (C) The amplitude of AMPA-evoked currents is plotted against that of kainate-evoked currents for HCP (n = 8) and LCP (n = 10) cells. The straight lines were obtained by least-squares linear fitting and indicate that the ratio of magnitudes of AMPA to kainate currents is 0.08 and 0.62 for LCP and HCP cells, respectively.

Calcium permeability of GluRs in retinal ganglion cells compositions of alternative splice variants. These data are consistent with the notion that the LCP cells express predominantly the flop version of the GluR2 subunit, while HCP cells express more flip variants of the GluR1, 3, or 4 subunits.

Polymerqse chain reaction analysis of non-N-methylD-aspartate receptor subunits in retinal ganglion cells The electrophysiological data predict that RGCs display specific repertoires of non-NMDA receptor subunits, including their flip and flop splice variants. As a first step leading to the characterization of the molecular composition of non-NMDA receptors in RGCs, we used the PCR to determine which flip or flop versions of non-NMDA receptor subunit mRNAs are expressed in these neurons. Following dissociation of the retina, fluorescence-activated cell sorter analysis was used to enrich for RGCs (see Experimental Procedures). The total RNA from the sorted cells was then purified and reverse transcribed. The cDNA thus obtained was used as a template for two consecutive rounds of PCR. Mixed fragments of G I u R I - 4 with sizes of approximately 750 bp were identified as the products of the first round of PCR by agarose gel electrophoresis. These mixed fragments then served as templates for the second round of PCR with primers specifically designed to amplify the flip and flop forms of the GluR1-4 subunits. The amplified fragments corresponding to the flip and flop forms of GluR1-4 were all present in their predicted sizes (lanes 1-8 in Fig. 6). The identity of these amplified fragments was confirmed by DNA sequencing. Similar results were obtained with cDNA samples from three separate sorting exper-

185

iments. No differences were encountered using cells from 10 or 17-day-old rats. DISCUSSION Our results show that RGCs possess AMPA/ kainate receptor-operated channels with different degrees of Ca 2+ permeability. With Ca 2~ permeability as the criterion, RGCs can be broadly separated into two classes: in the first group of cells (LCP cells) the non-NMDA receptors have a very low Ca 2+ permeability (Pca2+/Pcs+ = 0.09), similar to that reported in expression studies with the edited form of recombinant GIuR2 subunits (Pc~2+/Pc~+= 0.05). 7 In a second group of cells (HCP cells), AMPA/kainate receptor-operated channels have an intermediate Ca 2+ permeability (Pca2~/Pc~+ = 0.53) compared to recombinant receptors composed of the GIuR1, 3, or 4 receptor subunits (Pc,2+/Pc~+ = 1.2).7 Furthermore, we found that HCP and LCP cells differed in their relative sensitivity to AMPA and kainate: LCP compared to HCP cells generated smaller steady-state currents in response to a given concentration of AMPA relative to kainate. A similar difference in the AMPA-induced steady-state currents was previously found when the flip/flop variants of recombinant non-NMDA receptor subunits were expressed in mammalian cells. 42 One possible explanation for our data is that the intermediate level of Ca 2+ permeability observed in HCP cells is due to the presence of a mixture of heteromeric receptors with and without the edited form of the GIuR2 subunit. LCP cells, in contrast, appear to express mainly non-NMDA receptors containing the edited form

M 1 2 3 4 5 6 7 8 9 900bp 692bp 501bp

Fig. 6. The flip and flop forms of the GluRI-4 Subunits are present in enriched RGCs. The DNA fragments corresponding to flip and flop forms of GI'uR1-4 in enriched RGCs (obtained by fluorescenceactivated cell sorter analysis) were amplified by reverse transcription-PCR and analysed by agarose gel electrophoresis. The lanes represent DNA fragments obtained after the second round of PCR: lane 1, GIuRI, flip; lane 2, GluR1, flop; lane 3, GIuR2, flip; lane 4, GIuR2, flop; lane 5, GluR3, flip; lane 6, GluR3, flop; lane 7, GluR4, flip; lane 8, GIuR4, flop. The size of the DNA fragments were estimated by comparison with DNA markers (lane M) and corresponded to the predicted sizes. The identity of these fragments was further confirmed by DNA sequencing. Both flip and flop forms of GluRI ~ were detected in enriched RGCs.

186

D. Zhang et al.

of GIuR2 subunits, since the Ca 2+ permeability of the receptors in these cells is very low. While this study was in progress, 54 similar results regarding the distinction between non-NMDA receptors of high or low Ca 2+-permeability were reported in a parallel investigation of rat RGCs in a retinal whole-mount preparation. 4~ Glutamate activated Ca 2+ currents in Na+-free saline in 75% of the RGCs studied, but only 50% of all RGCs tested showed an inward rectification indicative of high Ca 2+ permeability (Pc,2+/Pcs+ value of 1.33 for kainate-activated currents). The higher percentage of RGCs with nonN M D A receptors with very low Ca 2+ permeability in our sample may result from the difference in preparations (whole mount vs dissociated cells) and/or the resulting patch-clamp conditions (neurons with long neurites vs solitary cells with almost no neurites). However, the novel, major results of the present study are the findings that single RGCs express both Ca2+-permeable and -impermeable non-NMDA receptors and that a high Ca 2+ permeability is correlated with large sustained currents in these receptors. The presence of the flip and flop splice variants of the GIuR1-4 subunits, as predicted by our electrophysiological findings, was confirmed by reverse transcription-PCR experiments. Together, the data suggest that the expression of specific non-NMDA subtypes of GluRs is controlled at the cellular level, imparting individual cells with a precisely tuned Ca 2÷ permeability. Transcripts for all subunits of the A M P A family plus the GluR5-7 subunits and receptors of the A M P A family have been found in RGCs by in situ hybridization and immunocytochemical methods, including co-expression of multiple subunits in single RGCs. 15'20'36A°'53It is unlikely, however, that homomeric channels of the kainate family of nonN M D A receptors contribute significantly to the nonN M D A responses in RGCs, since these receptors have a Kd for A M P A in the millimolar range, while the saturating concentration of A M P A in rat RGCs is 60 pM. 2 Thus, it is more likely that GluR1, 3 and 4 subunits, along with the unedited form of GIuR2, form high Ca2+-permeable homomeric or heteromeric channels in the HCP type of RGCs. If this interpretation holds true, the relative amount of these subunits in HCP cells must be considerably larger than that in LCP cells. The presence of one edited GluR2 subunit in a presumptive pentameric receptor would make the associated channel almost impermeable to Ca 2+. However, it is not sufficient to know which subunit mRNAs are present in a single cell to predict the electrophysiological observations; it will be necessary first to define how the assembly of the different subunits is controlled--are there mixtures of homomers and/or heteromers? 48 Moreover, since the stoichiometry of non-NMDA receptor subunits is not yet known, it is not sufficient to quantify the subunit mRNAs that are present in an individual RGC in order to predict the Ca 2+

permeability or effectiveness of A M P A vs kainate in evoking steady-state currents. Without knowing the rules of non-NMDA channel assembly and stoichiometry, we cannot determine if the electrophysiological differences are the result of qualitative and/or quantitative differences in the subunit composition of individual receptors. It has been shown that the flop version of GluRs generate channels with very small sustained responses to glutamate (or AMPA) compared to kainate. 42 In contrast, glutamate (or AMPA) activates the flip version of the channels four to five times more effectively, similar to the response to kainate. The amplitude of the fast-desensitizing component of the glutamate-evoked currents also depends on which flip and flop versions of the GIuR subunits are present. In LCP RGCs, the low Ca 2÷ permeability of non-NMDA receptors predicts the predominant expression of receptors containing the edited form of GIuR2 subunits. These GluR2 subunits in LCP cells are likely to be the flop version, as implied by the very small sustained AMPA-evoked current in these cells. On the other hand, HCP cells would be predicted to express predominantly the flip versions of GluR1, 3 or 4, in accordance with the observed larger sustained AMPA-evoked currents and the high Ca 2÷ permeability. It cannot be ruled out, however, that the LCP cells may express flip versions of other subunits besides the flop version of GluR2, since our voltage ramp protocol did not allow us to study the fast desensitizing components of AMPA-evoked currents. Along the same lines, it remains to be seen whether the HPC cells express combinations of flip and flop splice variants or only the flip versions. Since our results showed that both flip and flop versions of G l u R I - 4 subunits were present in enriched fractions of RGCs, these neurons have the capacity to assemble receptors with the characteristics found in our electrophysiological experiments. It is, however, not possible at present to give a quantitative prediction of the ratio of flip/flop versions of subunits in these cells since it is unknown whether the amplitude of sustained A M P A current is controlled by one or more flop versions of the subunit(s) in the presumptive pentameric receptors. Similarly, it remains unknown how RGCs or any other neuron regulate the subunit composition of their n o n - N M D A receptors. The Ca 2+ permeability coupled with the rate of desensitization of the non-NMDA receptors determine how much Ca 2+ enters the cell via these channels. A longer sustained response to glutamate and a higher permeability to Ca 2+ will work synergistically to allow greater influx of Ca 2+ into RGCs. The kainate-evoked current is generally much larger than that evoked by N M D A in RGCs. 1'25 Therefore, the amount of Ca 2+ entering through AMPA/kainate receptor-operated channels in RGCs might be considerable, even though the PCa2+/Pcs+ value for nonN M D A receptors is about 10 times lower than that

Calcium permeability of GluRs in retinal ganglion cells for N M D A receptors in RGCs. 49 N o n e the less, excitotoxicity observed after exposure of cultured R G C s to kainate has been shown to be largely due to kainate-induced release of glutamate and subsequent stimulation of N M D A receptors. 45 Thus, it appears unlikely that a large influx of Ca 2 + through n o n - N M D A receptors contributes significantly to glutamate-induced neurotoxicity in our culture system. While our data clearly establish the existence of at least two types of n o n - N M D A channels, they are also compatible with the possibility that there is a much larger number of these channel types. It remains to be seen if the differentiation of R G C s in vitro into H C P and LCP subtypes correlates with existing functional and anatomical classifications of R G C s in vivo. In individual RGCs, the combination of multiple subunits to form single channels, as well as the co-expression of the resulting multiple channel types, most likely determine the response to glutamate.

187

CONCLUSIONS Our results indicate that among rat retinal ganglion cells there are differences in the pattern of expression of A M P A / k a i n a t e receptor-operated channels. Importantly, individual cells co-express multiple heterologous n o n - N M D A receptors with distinct functional properties. The functional diversity of these receptors may play an important role in controlling Ca2+-entry into neurons. Using the PCR, transcripts for the G I u R subunits 1-4, including their "flip" and "flop" versions, were identified in retinal ganglion cells. It is tempting to speculate that the low Ca2+-permeability and the preference for kainate in some retinal ganglion cells may be due to the expression of n o n - N M D A receptors containing the edited form of the GluR2 flop splice variant. Acknowledgements--This work was supported by NIH

grants R01 EY05477, R0! EY09024 and P01 HD29587 (to S. A. L. and N. J. S.):

REFERENCES

1. Aizenman E., Frosch M. P. and Lipton S. A. (1988) Responses mediated by excitatory amino acid receptors in solitary retinal ganglion cells from rat. J. Physiol. 396, 75-91. 2. Aizenman E., Karschin A. and Lipton S. A. (1989) Two pharmacological classes of quisqualate-induced electrical responses in rat retinal ganglion cells in vitro. Eur. J. Pharmac. 174, 9-22. 3. Ascher P. and Nowak L. (1988) The role of divalent cations in the N-methyl-D-aspartate responses of mouse central neurones in culture. J. Physiol. 399, 247566. 4. Bettler B., Boulter J., Hermans-Borgmeyer I., O'Shea-Greenfield A., Deneris E. S., Moll C., Borgmeyer U., Hollmann M. and Heinemann S. (1990) Cloning of a novel glutamate receptor subunit, GIuR5: expression in the nervous system during development. Neuron 5, 583-595. 5. Bettler B., Egebjerg J., Sharma G., Pecht G., Hermans-Borgmeyer I., Moll C., Stevens C. F. and Heinemann S. (1992) Cloning of a putative glutamate receptor: a low affinity kainate-binding subunit. Neuron 8, 257-265. 6. Brorson J. R., Bleakman D., Chard P. S. and Miller R. J. (1992) Calcium directly permeates kainate/~-amino-3hydroxy-5-methyl-4-isoxazolepropionic acid receptors in cultured cerebellar Purkinje neurons. Molec. Pharmae. 41, 603-608. 7. Burnashev N., Monyer H., Seeburg P. H. and Sakmann B. (1992) Divalent ion permeability of AMPA receptor channels is dominated by the edited form of a single subunit. Neuron 8, 189-198. 8. Choi D. W. (1988) Glutamate neurotoxicity and diseases of the nervous system. Neuron 1, 623-634. 9. Collingridge G. L. and Singer W. (1990) Excitatory amino acid receptors and synaptic plasticity. Trends pharmac. Sci. ll, 290-296. 10. Dingledine R., Hume R. I. and Heinemann S. F. (1992) Structural determinants of barium permeation and rectification in non-NMDA glutamate receptor channels. J. Neurosci. 12, 4080-4087. 11. Egebjerg J., Bettler B., Hermans-Borgmeyer I. and Heinemann S. (1991) Cloning of a cDNA for a glutamate receptor subunit activated by kainate but not AMPA. Nature 351, 745 748. 12. Egebjerg J. and Heinemann S. F. (1993) Ca 2+ permeability of unedited and edited version of the kainate selective glutamate receptor GluR6. Proc. nam. Acad. Sci. U.S.A. 90, 755-759. 13. Gilbertson T. A., Scobey R. and Wilson M. (1991) Permeation of calcium ions through non-NMDA glutamate channels in retinal bipolar cells. Science 251, 1613-1615. 14. Gu Y. and Huang L.-Y. M. (1991) Block of kainate receptor channels by Ca 2+ in isolated spinal trigeminal neurons of rat. Neuron 6, 777 784. 15. Hamassaki-Britto D. E., Hermans-Borgmeyer I., Heinemann S. and Hughes T. E. (1993) Expression of glutamate receptor genes in the mammalian retina: the localization of GluRI through GluR7 mRNAs. J. Neurosci. 13, 1888-1898. 16. Hamill O. P., Marty A., Neher E., Sakmann B. and Sigworth F. J. (1981) Improved patch-clamp techniques for high resolution current recording from cells and cell-free membrane patches. Pflfigers Arch. 391, 85 100. 17. Herb A., Burnashev N., Sakmann B., Wisden W. and Seeburg P. H. (1992) The KA-2 subunit of excitatory amino acid receptors shows widespread expression in brain and forms ion channels with distantly related subunits. Neurons 8, 775-785. 18. Hille B. (1992) Ionic Channels of Excitable Membranes, pp. 362-389. Sinauer Associates, Sunderland, MA. 19. Hollmann M., Hartley M. and Heinemann S. (1991) Ca 2+ permeability of KA AMPA-gated glutamate receptor channels depends on subunit composition. Science 252, 851-853. 20. Hughes T. E., Hermans-Borgmeyer I. and Heinemann S. (1992) Differential expression of glutamate receptor genes (gluR 1-5) in the rat retina. Vis. Neurosci. 8, 49 55. 21. Hume R. I., Dingledine R. and Heinemann S. F. (1991) identification of a site in glutamate receptor subunits that controls calcium permeability. Science 253, 1028-1031. 22. Huntley G. W., Rogers S. W., Moran T., Janssen W., Archin N., Vickers J. C., Cauley K., Heinemann S. F. and Morrison J. H. (1993) Selective distribution of kainate receptor subunit immunoreactivity in monkey

188

23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51.

52. 53. 54.

D. Zhang et al. neocortex revealed by a monoclonal antibody that recognizes glutamate receptor subunits GluR5/6/7. d. Neurosci. 13, 2965-2981. Iino M., Ozawa S. and Tsuzuki K. (1990) Permeation of calcium through excitatory amino acid receptor channels in cultured rat hippocampal neurones. J. Physiol. 424, 151-165. Jonas P. and Sakmann B. (1992) Glutamate receptor channels in isolated patches from CA1 and CA3 pyrimidal cells of rat hippocampal slices. J. Physiol. 455, 143-171. Karschin A., Aizenmann E. and Lipton S. A. (1988) The interaction of agonists and non-competitive antagonists at the excitatory amino acid receptors in rat retinal ganglion cells in vitro. J. Neurosci. 8, 2895-2907. Kein~inen K., Wisden W., Sommer B., Werner P., Herb A., Verdoorn T. A., Sakmann B. and Seeburg P. H. (1990) A family of AMPA-selective glutamate receptors. Science 249, 556-560. K6hler M., Burnashev N., Sakmann B. and Seeburg P. H. (1993) Determinants of Ca 2÷ permeability in both TMI and TM2 of high affinity kainate receptor channels: diversity by RNA editing. Neuron 10, 491 500. Lambolez B., Audinat E., Bochet P., Cr6pel F. and Rossier J. (1992) AMPA receptor subunits expressed by single Purkinje cells. Neuron 9, 247-258. Leifer D., Lipton S. A., Barnstable C. J. and Masland R. H. (1984) Monoclonal antibody to Thy-1 enhances regeneration of processes by rat retinal ganglion cells in culture. Science 224, 303-306. Linn C. P. and Christensen B. N. (1992) Excitatory amino acid regulation of intracellular Ca 2+ in isolated catfish cone horizontal cells measured under voltage- and concentration-clamp conditions. J. Neurosci. 12, 2156-2164. Lipton S. A. and Kater S. B. (1989) Neurotransmitter regulation of neuronal outgrowth, plasticity and survival. Trends Neurosci. 12, 265-270. Lipton S. A. and Tauck D. L. (1987) Voltage-dependent conductances of solitary ganglion cells dissociated from the rat retina. J. Physiol. 385, 361-391. Meldrum B. and Garthwaite J. (1990) Excitatory amino acid neurotoxicity and neurodegenerative disease. Trends pharmac. Sci. 11, 379-387. Miller R. J. (1991) The revenge of the kainate receptor. Trends Neurosci. 14, 477~479. Monyer H., Seeburg P. H. and Wisden W. (1991) Glutamate-operated channels: developmentally early and mature forms arise by alternative splicing. Neuron 6, 799-810. Mfiller F., Greferath U., W~issle H., Wisden W. and Seeburg P. (1992) Glutamate receptor expression in the rat retina. Neurosci. Lett. 138, 179-182. Mfiller T., M611er T., Berger T., SchnRzer J. and Kettenman H. (1992) Calcium entry through kainate receptors and resulting potassium-channel blockade in Bergmann glial cells. Science 256, 1563 1566. Murphy S. N. and Miller R. J. (1989) Regulation of Ca ÷ ÷ influx into striatal neurons by kainate acid. J. -Pharmac. exp. Ther. 249, 184~193. Murphy S. N., Thayer S. A. and Miller R. J. (1987) The effects of excitatory amino acids on intracellular calcium in single mouse striatal neurons in vitro. J. Neurosci. 7, 4145-4158. Petralia R. S. and Wenthold R. J. (1992) Light and electron immunocytochemical localization of AMPA-selective glutamate receptors in the rat brain. J. comp. Neurol. 318, 329-354. R6rig B. and Grantyn R. (1993) Rat retina ganglion cells express Ca2+-permeable non-NMDA glutamate receptors during the period of histogenetic cell death. Neurosci. Lett. 153, 32-36. Sommer B., Kein~inen K., Verdoorn T. A., Wisden W., Burnashev N., Herb A., K6hler M., Takagi T., Sakmann B. and Seeburg P. H. (1990) Flip and flop: a cell-specific functional switch in glutamate-operated channels of the CNS. Science 249, 1580-1585. Sommer B., K6hler M., Sprengel R. and Seeburg P. H. (1991) RNA editing in brain controls a determinant of ion flow in glutamate-gated channels. Cell 67, 11-19. Sommer B. and Seeburg P. H. (1992) Glutamate receptor channels: novel properties and new clones. Trends pharmac. Sci. 13, 291-296. Sucher N. J., Aizenman E. and Lipton S. A. (1991) N-Methyl-o-aspartate antagonists prevent kainate neurotoxicity in rat retinal ganglion cells in vitro. J. Neurosci. l l , 966-971. Sucher N. J., Brose N., Deitcher D. L., Awobuluyi M., Gasic G. P., Bading H., Cepko C. L., Greenberg M. E., Jahn R., Heinemann S. F. and Lipton S. A. (1993) Expression of endogenous NMDAR1 transcripts without receptor protein suggests post-transcriptional control in PC12 cells. J. biol. Chem. 268, 22299-22304. Sucher N. J., Cheng T. P. O. and Lipton S. A. (1991) Cryopreservation of postnatal rat retinal ganglion cells: persistence of voltage- and ligand-gated ionic currents. Neuroscience 43, 135-150. Sucher N. J., Deitcher D. L., Cepko C. L. and Lipton S. A. (1991) Patch clamp and PCR on a single rat retinal ganglion cell: neuronal nicotinic acetylcholine receptor (nAChR) subunits. Invest. Ophthal. Vis. Sci. 32, 1591. Sucher N. J., Wong L. A. and Lipton S. A. (1990) Redox modulation of NMDA receptor-mediated Ca 2+ flux in mammalian central neurons. NeuroReport l, 29 32. Verdoorn T. A., Burnashev N., Monyer H., Seeburg P. H. and Sakmann B. (1991) Structural determinants of ion flow through recombinant glutamate receptor channels. Science 252, 1715-1718. Vickers J. C., Huntley G. W., Edwards A. M., Moran T., Rogers S. W., Heinemann S. F. and Morrison J. H. (1993) Quantitative localization of AMPA/kainate and kainate glutamate receptor subunit immunoreactivity in neurochemically identified subpopulations of neurons in the prefrontal cortex of the macaque monkey. J. Neurosci. 13, 2982-2992. Watkins J. C., Krogsgaard-Larsen P. and Honore T. (1990) Structure-activity relationships in the development of excitatory amino acid receptor agonists and competitive antagonists. Trends pharmac. Sci. 11, 25-33. Wenthold R. J., Yokotani N., Doi K. and Wada K. (1992) Immunochemical characterization of the non-NMDA glutamate receptor using subunit-specific antibodies. J. biol. Chem. 267, 501-507. Zhang D. and Lipton S. A. (1992) Permeation of calcium ions through kainate receptor-coupled channels in rat retinal ganglion cells. Soc. Neurosci. Abstr. 18, 395. (Accepted 1 December 1994)