NMDH Receptor Proteins and Genes Encoding the Same

This invention relates to novel proteins and genes coding the same, more specifically to protein NMDA (N-methyl-D-aspartic acid) type glutamate receptors which play a central role in nervous information transmission, modified products thereof, and genes (cDNAs) encoding them.

It has been suggested that glutamate receptors, which are receptors of main stimulant nervous transmitter substances in a central nervous system of a higher animal, play a central role in nervous information transmission at a synapse, and also are deeply concerned with appearance of synapse plasticity which is basically required for memory and learning and neuronal cell death caused by a disease such as cerebral ischemia and epilepsy. Thus, it is considered that clarifications of molecular structures and functions of the glutamate receptors are required to understand a transmission mechanism of nervous information in a center, a cerebral structure of higher order and the disease of a brain.

As in the case of receptors of acetylcholine and GABA (.gamma.-aminobutyric acid), the glutamate receptors are roughly classified into ion channel type glutamate receptors and G protein coupled type receptors (metabolism-controlling type receptors). Receptors which can effect long-term reinforcement of synapse transmission at a CA1 region of a hippocampus of which the most advanced study has been made about synapse plasticity are two kinds of ion channel type glutamate receptors. That is, an NMDA type receptor having Ca.sup.2+ permeability and opening depending on membrane potential and a quisqualate/kinate type receptor (or a non-NMDA type receptor). While the non-NMDA type receptor performs general synapse transmission, the NMDA type receptor performs Ca.sup.2+ permeation when a high frequent. stimulus which induces long-term reinforcement of synapse transmission is given. The Ca.sup.2+ permeation is inhibited by Mg.sup.2+ depending on membrane potential.

Although the glutamate receptors have important physiological functions as described above, molecular biological structures thereof had not been clarified for a long term. In recent years, by using a cDNA-producing system of Xenopus oocytes, Hollmann et al. cloned cDNA of a non-NMDA type glutamate receptor ("Nature", 342, pp. 643 to 648 (1989)) and Nakanishi et al. cloned cDNA of a rat-G protein coupled type glutamate receptor ("Nature", 349, pp. 760 to 765 (1991)) and cDNA of a rat-NMDA type receptor ("Nature", 354, pp. 31 to 37 (1991)). Other plural kinds of genes of glutamate receptors have been cloned, and the molecular biological mechanisms thereof have not yet been clarified sufficiently under the present situation.

The present inventors have paid special attention to NMDA type glutamate receptors and researched and studied them intensively, and consequently novel glutamate receptors and glutamate receptors modified by protein engineering have been obtained to accomplish the present invention.

That is, the characteristic features of the present invention reside in a glutamate receptor which is represented by an amino acid sequence described in Sequence ID No. 1, 2, 3 or 4 of the sequence table and a gene encoding the same, a modified glutamate receptor which is represented by an amino acid sequence described in Sequence ID No. 8 or 9 of the sequence table and a gene encoding the same, and a modified glutamate receptor which is represented by a base sequence described in Sequence ID No. 19 of the sequence table and a gene encoding the same.

The drawings relating to the present invention are explained.

FIGS. 1(a) and 1(b) show functional expression of .epsilon.4/.zeta.1 heteromeric NMDA receptor channel cDNAs in frog oocytes. Current responses are measured at -70 mV membrane potential in normal Ringer's solution. FIG. 1 (a) represent a current response to 100 .mu.M NMDA +10 .mu.M glycine (NMDA/Gly) of the .epsilon.4/.zeta.1 heteromeric channel, a current response to 10 .mu.M L-glutamate+10 .mu.M glycine (Glu/Gly) of the same and 1 mM Mg.sup.2+ (Glu/Gly +Mg.sup.2+) on a current response to Glu/Gly of the same, respectively. FIG. 1(b) represent effects of 500 .mu.M D-2-amino-5-phosphonovalerate (APV) (Glu/Gly +APV) and 100 .mu.M 7-chlorokynurenate (7 CK) (Glu/Gly +7 CK) on a current response of the .epsilon.4/.zeta.1 heteromeric channel to Glu/Gly, respectively.

FIG. 2 shows pharmacological properties of the .epsilon.4/.zeta.1 heteromeric NMDA receptor channel. Current responses are measured at -70 mV membrane potential in a Ba.sup.2+ -Ringer's solution. In the figure, dose-response relationship for L-glutamate (.smallcircle.) and glycine ( ) in the presences of 10 .mu.M glycine and 10 .mu.M L-glutamate (each point represents the mean fractional responses obtained from 4 oocytes) are shown.

FIG. 3 shows pharmacological properties of the .epsilon.4/.zeta.1 heteromeric NMDA receptor channel. Current responses are measured at -70 mV membrane potential in a Ba.sup.2+ -Ringer's solution. In the figure, influences of APV and 7 CK on response to 4.7 .mu.M L-glutamate +0.9 .mu.M glycine (the concentrations of agonists were 10-folds the respective EC.sub.50 values) are shown.

FIGS. 4(a) to 4(d) show a current response of the heteromeric NMDA receptor channel to 10 .mu.M L-glutamate +10 .mu.M L-glycine at -70 mV membrane potential in a frog standard Ringer's solution. In the figures, FIGS. 4(a), 4(b), 4(c) and 4(d) represent an .epsilon.2/.zeta.1 NMDA receptor channel, an .epsilon.2/.zeta.1-N598Q (a modified glutamate receptor shown in Sequence ID No. 9 of the sequence table) NMDA receptor channel, an .epsilon.2-N589Q (a modified glutamate receptor shown in Sequence ID No. 8 of the sequence table)/.zeta.l NMDA receptor channel and an .epsilon.2-N589Q/.zeta.1-N598Q NMDA receptor channel, respectively.

FIGS. 5(a) to 5(d) show current-voltage relationships of wild type and mutant heteromeric channels in the presence of 1 mM Mg.sup.2+ ( ) and in the absence thereof (.smallcircle.). In the figures, FIGS. 5(a) , 5(b) , 5(c) and 5(d) represent an .epsilon.2/.zeta.1 NMDA receptor channel, an .epsilon.2/.zeta.1-N598Q NMDA receptor channel, an .epsilon.2-N589Q/.zeta.1 NMDA receptor channel and an .epsilon.2-N589Q/.zeta.1-N598Q NMDA receptor channel, respectively.

FIG. 6 shows the effects of Mg.sup.2+ concentrations on the response at -70 mV membrane potential. In the figure, .largecircle., , , .quadrature. and represent an .epsilon.2/.zeta.1 NMDA receptor channel, an .epsilon.2/.zeta.1-N598Q NMDA receptor channel, an .epsilon.2-N589Q/.zeta.1 NMDA receptor channel, an .epsilon.2-N589Q/.zeta.1-N598Q NMDA receptor channel and an .epsilon.2/.zeta.1-ZAZ (a mutant glutamate receptor shown in Sequence ID No. 10 of the sequence table) NMDA receptor channel, respectively.

FIG. 7 shows suppression after repetitive application of 1 .mu.M (+)-MK 801. In the figure, .largecircle., , , .quadrature. and represent a .epsilon.2/.zeta.1 NMDA receptor channel, an .epsilon.2/.zeta.1-N598Q MNDA receptor channel, an .epsilon.2-N589Q/.zeta.1 NMDA receptor channel, .epsilon.2-N589Q/.zeta.1-N598Q NMDA receptor channel and an .epsilon.2/.zeta.1-ZAZ NMDA receptor channel, respectively.

FIG. 8 shows the sensitivity to various concentrations of Zn.sup.2+. In the figure, .largecircle., , , .quadrature. and represent an .epsilon.2/.zeta.1 NMDA receptor channel, an .epsilon.2/.zeta.1-N598Q NMDA receptor channel, an .epsilon.2-N589Q/.zeta.1 NMDA receptor channel, an .epsilon.2-N589Q/.zeta.1-N598Q NMDA receptor channel and an .epsilon.2/.zeta.1-ZAZ NMDA receptor channel, respectively.

In the following, the present invention is explained in detail.

The glutamate receptor of the present invention has an amino acid sequence as shown in Sequence ID No. 1, 2, 3 or 4 of the sequence table.

The glutamate receptor shown in Sequence ID No. 1 of the sequence table (hereinafter sometimes referred to as ".epsilon.1 subunit") is a protein comprising 1464 amino acids. As a gene encoding such a glutamate receptor, there may be mentioned, for example, a base sequence as shown in Sequence ID No. 5 of the sequence table.

The glutamate receptor shown in Sequence ID No. 2 of the sequence table (hereinafter sometimes referred to as ".epsilon.2 subunit") is a protein comprising 1482 amino acids. As a gene encoding such a glutamate receptor, there may be mentioned, for example, a base sequence as shown in Sequence ID No. 6 of the sequence table.

The glutamate receptor shown in Sequence ID No. 3 of the sequence table (hereinafter sometimes referred to as ".epsilon.3 subunit") is a protein comprising 1239 amino acids. As a gene encoding such a glutamate receptor, there may be mentioned, for example, a base sequence as shown in Sequence ID No. 7 of the sequence table.

The glutamate receptor shown in Sequence ID No. 4 of the sequence table (hereinafter sometimes referred to as ".epsilon.4 subunit") is a protein comprising 1323 amino acids. This protein has a signal sequence, and a mature type thereof is considered to start from phenylalanine which is the 28th amino acid in Sequence ID No. 4 of the sequence table. As a gene encoding such a glutamate receptor, there may be mentioned, for example, a base sequence as shown in Sequence ID No. 16 of the sequence table.

The modified glutamate receptor of the present invention has an amino acid sequence as shown in Sequence ID No. 8, 9 or 10 of the sequence table.

The modified glutamate receptor shown in Sequence ID No. 8 of the sequence table (hereinafter sometimes referred to as ".epsilon.2-N589Q") is a modified product in which asparagine which is the 589th from the N terminal end of the above .epsilon.2 subunit is replaced with glutamine. As a gene encoding such a modified product, there may be mentioned, for example, a base sequence as shown in Sequence ID No. 17 of the sequence table.

The modified glutamate receptor shown in Sequence ID No. 9 of the sequence table (hereinafter sometimes referred to as ".zeta.1-N598Q") is a modified product in which asparagine which is the 598th from the N terminal end of a .zeta.1 subunit comprising 920 amino acids ("FEBS Lett", 300 pp 39 to 45 (1992)) is replaced with glutamine. As a gene encoding such a modified product, there may be mentioned, for example, a base sequence as shown in Sequence ID No. 18 of the sequence table.

The modified glutamate receptor shown in Sequence ID No. 10 of the sequence table (hereinafter sometimes referred to as ".zeta.1-ZAZ") is a modified product in which the 1726th to 1743rd base sequence of the above .zeta.1 subunit is changed to a base sequence as shown in Sequence ID No. 19 of the sequence table.

In the present invention, there may be included modified glutamate receptors in which amino acids or nucleic acids are partially removed, replaced or added within the range which does not impair activity as a glutamate receptor.

DNA fragments of the glutamate receptor of the present invention and the gene coding the same can be obtained by, for example, the following method.

First, cerebral tissues of a mammal, such as mouse, are homogenized in an aqueous solution containing guanidium thiocyanate or the like, and all RNAs are separated as precipitates by cesium chloride equilibrium density gradient centrifugation or sucrose density gradient centrifugation in accordance with the method of Chirgwin et al. ("Biochemistry", 18, pp. 5294 to 5299 (1979)).

After separation, all RNAs are purified by extraction with phenol, precipitation with ethanol or the like, and the resulting RNAs are purified by oligo (dT) cellulose column chromatography to isolate poly (A)-containing mRNAs (poly A.sup.+ mRNAs) including mRNAs of the desired glutamate receptor, whereby an mRNA group can be obtained.

The mRNA group prepared as described above and, for example, primer DNA as described in "FEBS Lett.", 272, pp 73 to 80 (1990) are hybridized, and by using a reverse transcriptase and T4 DNA polymerase, cDNAs with double strands are synthesized and prepared according to a conventional method.

Then, EcoRI linkers are added to both ends of cDNA chains.

The above cDNA chains are inserted into a position of EcoRI cleave site of a .lambda. phage vector such as .lambda.gt10 to obtain a recombinant .lambda. phage DNA group.

By using the recombinant .lambda. phage DNA group obtained as described above and using an in vitro packaging kit such as commercially available Gigapack Gold (trade name, produced by Promega Co.) according to an operating manual, the so-called in vitro packaging was carried out to obtain .lambda. phage particles having the recombinant .lambda. phage DNAs. The .lambda. phage particles obtained are increased by transforming host cells such as Escherichia coli according to a conventional method.

The clone group obtained is taken and collected on a nylon film or a nitrocellulose film such as Gene Screening Plus (trade name, produced by Du Pont), and protein is removed in the presence of an alkali. The .lambda. phage DNAs including cDNAs are hybridized with a .sup.32 P-labeled probe prepared from fragments of cDNA of a mouse glutamate receptor which has been already cloned ("FEBS Lett.", 272, pp. 73 to 80 (1990)) and a DNA group of these cDNA clones, whereby clones having extremely high possibility of coding the desired glutamate receptor can be narrowed down to several clones.

The modified products are prepared by modifying the respective natural substances of the .epsilon.2 subunit and the .zeta.1 subunit by protein engineering or gene technology. Such a modification method is not particularly limited, and may be carried out specifically by the 2-step polymerase chain reaction (PCR) method by using a suitable synthetic oligonucleotide and DNA fragments derived from a plasmid of pBKSA.epsilon.2 ("Nature", 358, pp. 36 to 41 (1992)) or pBKSA.zeta.1 ("FEBS Lett.", 300, pp. 39 to 45 (1992)).

For these clones, activity test is conducted by using translation of Xenopus oocytes to obtain cDNA clones encoding the desired glutamate receptor.

The cDNA obtained as described above can be expressed in, for example, transient in vitro protein translation, specifically translation in Xenopus oocytes as described in "Nature", 329, pp. 836 to 838 (1987), or in a host such as a CHO cell transformed by a plasmid for expressing a protein prepared by connecting such a cDNA to downstream of a promoter of a expression plasmid for animal cells. Then, according to a conventional method, the expressed protein is collected to obtain the glutamate receptor of the present invention.

The present invention is described in detail by referring to Examples, but it is not intended that the present invention be limited by these Examples.

First, a cerebellum of ICR mouse was homogenized in an aqueous solution of guanidium thiocyanate, and then extracted, separated and purified according to the method of Chirgwin et al. ("Biochemistry", 18, PP. 5294 to 5299 (1979)), and subjected to oligo (dT)-cellulose column chromatography according to the method of Aviv et al. ("Proc. Natl. Acad. Sci. USA", 69, pp. 1408 to 1412 (1972)) to isolate poly (A)-containing mRNAs (poly A.sup.+ mRNAs) containing mRNAs of an NMDA type glutamate receptor.

By using a cDNA synthesizing kit produced by Bethesda Research Laboratories Co., cDNAs having double strand were prepared from the poly A.sup.+ mRNAs. Both ends of the cDNAs were blunt-ended with T4 DNA polymerase, and the cDNAs were methylated by using an EcoRI-methylase. Then, EcoRI linkers were added to both ends of the cDNAs by ligase to prepare cDNA fragments having an EcoRI restriction site at both ends by digestion with an EcoRI restriction enzyme.

The cDNAs obtained as described above were subjected to 1.5% agarose gel electrophoresis to select and collect cDNAs having a size of 0.5 Kb or more.

The obtained cDNA fragments having a size of 0.5 Kb or more were linked to .lambda.gt10 phage DNAs by ligase, and the cDNAs were inserted into an EcoRI cleaved site of .lambda.gt10 to constitute a .lambda.gt10 cDNA library according to a conventional method. By using the .lambda.gt10 cDNA library obtained and using an in vitro packaging kit Gigapack Gold (trade name, produced by Promega Co.) according to an operating manual, in vitro packaging was carried out to obtain .lambda. phage particles having recombinant .lambda. phage DNAs. The .lambda. phage particles obtained were cultivated by transforming Escherichia coli of host cells according to a conventional method.

The clone group obtained was taken and collected on a nylon film or Gene Screening Plus (trade name, produced by Du Pont), and protein was removed in the presence of an alkali. Then, screening of the cDNA library was carried out. BamHI DNA fragments (1348 to 1946) and HincII DNA fragments (1279 to 2239) (previously cloned cDNA fragments of mouse glutamate receptors) of a pKCR30 plasmid ("FEBS Lett.", 272, pp. 73 to 80 (1990)) were labeled with .sup.32 p to prepare probes and the plaque hybridization was done in the presence of 30 % formaldehyde at 37.degree. C.

According to the method described above, some cDNA clones encoding the novel glutamate receptor were obtained. Among them, cDNA fragments from a recombinant phage .lambda. A19 having the .epsilon.1 subunit cDNAs which was considered to be a subunit of the NMDA type glutamate receptor ion channel were subcloned at an EcORI site of pBluescript IISK (-) plasmid (trade name) produced by Stratagene Co. to obtain a pGRA19 plasmid. By using the A19 cDNA fragments and other cDNA fragments (AT11) obtained by the same method as that of A19 cDNA fragments as probes, screening was carried out in the presence of 30% formamide at 45.degree. C. to obtain .epsilon.1 subunit cDNA clones. The cDNA-inserted fragments obtained from the total 7 recombinant phages were inserted into an EcoRI site of a pBKSA plasmid, a plasmid obtained by inserting:

5'-CCAGGTGCA-3'

3'-ACGTGGTCC-5'(Sequence ID No. 11 in the sequence table) into a PstI site of pBluescript IIKS (+) (trade name, produced by Stratagene Co.) in the same direction as that of a T3 promoter) to prepare plasmids pAT4, pAT11, pAT12, pAT19, pAT20, pAT201 and pAT202.

All base sequences of A19 (the -327th to 3181st of base pair) and AT19 (the 2859th to 5470th of base pair) were determined according to the dideoxy chain termination method of Sanger et al. ("Proc. Natl. Acad. Sci. USA", 74, pp. 5463 to 5467 (1977)). Base pair numbers were given from the 5' end to the 3' end, and negative numbers were given to base pairs before a codon corresponding to the amino terminal end of the .epsilon.1 subunit. The partial DNA base sequences: base sequences of clone AT4 (the 2550th to 3432nd of base pair), ATll (the 2550th to 3622nd of base pair), AT12 (the 2404th to 3622nd of base pair), AT20 (the 2550th to 3177th of base pair), AT201 (the 3526th to 4401st of base pair) and AT202 (the 3409th to 4371 st of base pair) were completely the same with the corresponding base sequences of A19 and AT19. The base sequences and the amino acid sequences were analyzed by GENETYX Software (SDC). As a result, the amino acid sequence of the .epsilon.1 subunit was a sequence shown in Sequence ID No. 1 of the sequence table, and the base sequence encoding the .epsilon.1 subunit was a sequence shown in Sequence ID No. 5 of the sequence table.

The ends of EcoRI DNA fragments with a size of 3.5 Kb derived from the pGRA19 plasmid were bluntended by using T4 DNA polymerase and inserted into a SmaI site of a pSP64AX plasmid ("FEBS Lett.", 259, pp. 37 to 42 (1989)) in the same direction with respect to an SP6 promoter to obtain a pSPA19 plasmid. SalI/HindIII DNA fragments with a size of 3.0 Kb derived from the pSPA19 plasmid, HindIII/NcoI DNA fragments with a size of 0.36 Kb derived from the same and NcoI/SalI DNA fragments with a size of 5.4 Kb derived from the pAT19 plasmid were linked by ligase to prepare a pBKSA .epsilon.1.

Then, by using the pBKSA .epsilon.1 cut by NotI as a template and using T3 RNA polymerase produced by BRL Co., .epsilon.1-specific mRNAs were synthesized in vitro. Transcription was carried out in a solution containing ATP, TTP and CTP at a concentration of 0.5 mM, respectively and 0.1 mM GTP in the presence of 0.5 mM dinucleotide 7mGpppG having a cap structure (in which 2 Gs were linked at 5' and 7-site of one guanine was methylated, produced by P-L Biochemicals Co.). One or both of .epsilon.1 subunit-specific mRNAs and .zeta.1.NMDA type glutamate receptor subunit-specific mRNAs (mRNAs derived from mice, coding an amino acid sequence different in the 213rd (Glu.fwdarw.Asp) and the 460th (Ile.fwdarw.Val) from the N terminal end from the amino acid sequence which is coded by cDNA.-NMDA R1 of a rat-NMDA type glutamate receptor cloned by Nakanishi et al. ("Nature", 354, pp. 31 to 37 (1991))) were injected into Xenopus oocytes.

The oocytes used were obtained from mature female Xenopus and separated into one oocyte in a Barth medium by using sharp tweezers and scissors. The concentration of the mRNAs was 0.1 .mu.g/.mu.l and 10 .mu.l thereof was injected into about 100 oocytes in an amount of 50 to 100 nl per one oocyte. The mRNAs were injected under a microscope by using a capillary and a micromanipulator. After the injection, the oocytes were incubated at 19.degree. C. for one day in a Barth medium containing 0.1 mg/ml of gentamicin ("The Control of Gene Expression in Animal Development, Clarendon, Oxford (1974)). Then, the oocytes were allowed to stand in 1 mg/ml of collagenase at room temperature for 1 hour, and then extracellular skins were removed under a microscope by using sharp tweezers. These oocytes were returned to the Barth medium containing 0.1 mg/ml of gentamicin, incubated again at 19.degree. C. for one day and then used for an electrophysiological test.

The electrophysiological test was carried out by a conventional micropipet voltage clamp method. Oocytes generally have a membrane potential of about -30 to -40 mV. First, while a frog standard Ringer's solution (115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl.sub.2 and 10 mM HEPES-NaOH (pH: 7.2)) was flown in a chamber, 2 glass microelectrodes charged with 3M KCl were stuck in the oocytes, and the membrane potential was fixed to -70 mV. In these oocytes, an agonist (NMDA) is linked to a glutamate receptor expressed on a cell membrane and an ion channel is opened, the oocytes have an action of returning membrane potential to an original value, whereby inflow and outflow of ions are caused. In the present experiment system, flow of these ions were detected, and opening and closing of the ion channel were observed.

The .epsilon.1 subunit-specific mRNAs synthesized from the cloned cDNAs in vitro were injected into the Xenopus oocytes. The injected oocytes did not show any reaction to 10 .mu.M L-glutamate and 100 .mu.M NMDA even in the presence of 10 .mu.M glycine which was an essential activating factor of the NMDA type receptor channel. However, when the .epsilon.1 subunit-specific mRNAs and the .zeta.1.NMDA type receptor subunit-specific mRNAs were injected simultaneously, significant reaction was observed. In a frog standard Ringer's solution, the inward current at a membrane potential of -70 mV was 364.+-.62 nA (average.+-.standard deviation, n=7) with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 175.+-.30 nA (n=7) with respect to 100 .mu.M NMDA and 10 .mu.M glycine. The current amplitude was extremely larger than that of the case where only the .zeta.1.NMDA type receptor subunit-specific mRNAs were injected to the oocytes (8.+-.2 nA (n= 7) with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 6.+-.1 nA (n=7) with respect to 100 .mu.M NMDA and 10 .mu.M glycine). In a Ba.sup.2+ Ringer's solution (in which Ca.sup.2+ in the standard Ringer's solution described above was replaced, with Ba.sup.2+), the current amplitudes with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine, and 100 .mu.M NMDA and 10 .mu.M glycine were 84.+-.4 nA and 57.+-.3 nA (n=7), respectively, when the .epsilon.1 and .zeta.1 subunits were expressed at the same time, and 2.+-.0.4 nA and 1.+-.0.2 nA (n=7), respectively, when only the .zeta.1 subunit was expressed. From the results described above, it was concluded that the .epsilon.1 protein was a subunit of the NMDA type glutamate receptor ion channel.

Using the 1388 base pair KpnI/HindIII DNA fragments derived from the pGRA 19 plasmid obtained in Example 1 as probes, the mouse brain cDNA library obtained from .lambda.gt10 in Example 1 was screened under mild conditions to obtain .epsilon.2 subunit cDNA clone U9 and two .epsilon.3 subunits, cDNA clone U20 and He49. The cDNA fragments were inserted into the EcoRI site of pBluescript IISK (-) (Stratagene) to obtain pGRU9, pGRU20 and pGRHe49 plasmids. By using 445 base pair DraIII/EcoRI derived from pGRU9 plasmid, 2847 base pair HindIII/EcoRI fragments, 1615 base pair EcoRI fragments derived from pGRHe49 and 314 base pair HindIII/EcoRI fragments derived from pGRUN3 as probes, screening was performed under severe conditions to isolate further some clones (as the .epsilon.2 subunit cDNA clones, U4, U2, U7, U8, U11, U16, U17 and U22, and as the .epsilon.3 subunit cDNA clones, UN6, UN3 and UT5). These inserted cDNAs in recombinant phage were inserted into the EcoRI site of the aforesaid pBSKA plasmid.

All base sequences of the .epsilon.2 subunit cDNA clones U9 (the -417th to 3218th base pair) and U4 (the 1117th to 4454th base pair) were determined according to the dideoxy chain termination method as described above. Base sequences of clones U2 (the 4319th to 4396th base pair), U7 (the -417th to 3616th base pair), U8 (the 4319th to 4396th base pair), U11 (the 4319th to 4396th base pair), U16 (the -205th to -57th base pair), U17 (the -205th to -57th base pair) and U22 (the 1449th to 4365th base pair) were completely coincident with the corresponding base sequences of U9 and U4.

In the .epsilon.3 subunit cDNA clones UN6 (the -232nd to 3283th base pair), He49 (the -101st to 1504th base pair), . UN3 (the 52nd to 4039th base pair), U20 (the 347th to 2016th base pair) and UT5 (the 2487th to 4071st base pair), base sequences were completely identical therewith except for the 711st (T or C), 1485th (A or G), 1494th (A or G) and 2721st (T or C) . Base pair numbers were started from a codon coding an amino terminal end amino acid of a mature type subunits and given a direction from the 5' end to the 3' end. The base sequences and the amino acid sequences were analyzed by GENETYX Software. As a result, the amino acid sequence of the mature type .epsilon.2 subunit was a sequence shown in Sequence ID No. 2 of the sequence table, and the base sequence coding the .epsilon.2 subunit was a sequence shown in Sequence ID No. 6 of the sequence table. Also, the amino acid sequence of the mature type .epsilon.3 subunit was a sequence shown in Sequence ID No. 3 of the sequence table, and the base sequence coding the .epsilon.2 subunit was a sequence shown in Sequence ID No. 7 of the sequence table.

EcoRI/PvuI DNA fragments with a size of 3.5 Kb derived from the pGRU9 and PvuI/EcoRI DNA fragments with a size of 1.6 Kb derived from the pGRU4 (U4 clone derivative) were linked to EcoRI DNA fragments with a size of 3.0 Kb derived from the pBKSA by ligase to obtain pBKSA .epsilon.2 plasmids. SalI DNA fragments with a size of 864 bp derived from the pGRUN6 (UN6 clone derivative) and SalI DNA fragments having a size of 6.4 Kb derived from the pGRUN3 (UN3 clone derivative) were linked by ligase to obtain pBKSA .epsilon.3.

Then, by using the pBKSA .epsilon.2 cut by NotI and the pBKSA .epsilon.3 cut by XbaI as molds and using T3 RNA polymerase produced by BRL Co., .epsilon.2-specific mRNAs and .epsilon.3-specific mRNAs were synthesized in vitro. Transcription was carried out in the same manner as in Example 1.

The .epsilon.l (not more than 22 ng/egg) subunit-specific mRNAs obtained in Example 1, the .epsilon.2 (not more than 19 ng/egg) subunit-specific mRNAs, the .epsilon.3 (not more than 16 ng/egg) subunit-specific mRNAs and .xi.1 (not more than 13 ng/egg) NMDA type glutamate receptor subunit-specific mRNAs described above were used singly or in combination, they were injected into the Xenopus oocytes in the same manner as in Example 1. 10 .mu.l of the mRNA aqueous solution was injected into about 100 oocytes in an amount of 50 to 100 nl per one oocyte. The oocytes after the injection were treated in the same manner as in Example 1 and then used for an electrophysiological test.

The .epsilon.2 or .epsilon.3 subunit-specific mRNAs synthesized from the cloned cDNAs in vitro were injected into the Xenopus oocytes with the .xi.1 subunit-specific mRNAs. In a frog standard Ringer's solution, the inward current of .epsilon.2 and .xi.1 subunits expressing oocytes at a membrane potential of -70 mV was 667.+-.187 nA (average.+-.standard deviation) with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 546.+-.172 nA (N=5) with respect to 100 .mu.M NMDA and 10 .mu.M glycine, and in .epsilon.3 and .xi.1 subunits expressing oocytes, it was 191.+-.67 nA with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 89.+-.22 nA (n=7) with respect to 100 .mu.M NMDA and 10 .mu.M glycine. The current amplitude was extremely larger than that of the case where only the .zeta.1.NMDA type receptor subunit-specific mRNAs were injected to the oocytes (11.+-.l nA, n=12, with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 9.+-.2 nA, n=7, with respect to 100 .mu.M NMDA and 10 .mu.M glycine). No response was observed in the oocytes to which the .epsilon.2 or .epsilon.3 subunit-specific mRNAs were injected singly (<1 nA). In the oocytes in which the .epsilon.2 and .epsilon.1 subunits, or the .epsilon.3 and .xi.1 subunits were expressed, responses to 100 .mu.M kainate and to 100 .mu.M AMPA were less than the measurement limit.

The .epsilon.2/.xi.l heteromeric channel showed a response to 10 .mu.M L-glutamate and 10 .mu.M glycine and to 100 .mu.M L-aspartic acid alone. On the other hand, the .epsilon.3/.xi.1 heteromeric channel showed a response only to 10 .mu.M glycine. The response to 10 .mu.M L-glutamate or 10 .mu.M glycine alone disappeared by 100 .mu.M of D-2-amino-5-phosphonovalerate (APV) which is a specific competitive antagonist of the NMDA receptors and 30 .mu.M 7-chlorokynurenate (7 CK) which had been reported as a competitive antagonist to glycine control site of the NMDA receptors. The response to 10 .mu.M L-glutamate and 10 .mu.M glycine can be suppressed by these competitive antagonists, non-competitive antagonists, 100 .mu.M Mg.sup.2+, 100 .mu.M Zn.sup.2+, or 1 .mu.M (+)-MK-801 (an open channel blocker of the NMDA type receptor channel). The effect of the channel blocker to the .epsilon. 3/.xi.1 channel is weaker than those to the .epsilon.1/.xi.1 and .epsilon.2/.xi.1 channels. The .epsilon.2/.xi.1 and .epsilon.3/.xi.1 channels each showed an inward current in a Na.sup.+ and K.sup.+ -free Ringer's solution containing 20 mM Ca.sup.2+ (Ca.sup.2+ -Ringer's solution), but in a control Na.sup.+ and K.sup.+ -free Ringer's solution, a slightly outward current could be observed. This shows that the heteromeric channels permeate Ca.sup.2+. According to the above, it can be concluded that .epsilon.2 and .epsilon.3 proteins are subunits of the NMDA receptor channels.

In order to minimize effects of a secondary activated Ca.sup.2+ dependent Cl- current on a dose-reaction curve to L-glutamate and glycine of the heteromeric NMDA receptor channel, it was examined in a Ba.sup.2+ Ringer's solution. EC.sub.50 values to .epsilon.1/.xi.1, .epsilon.2/.xi.1 and .epsilon.3/.xi.1 channels were each 1.7 .mu.M, 0.8 .mu.M and 0.7 .mu.M and those to glycine were each 2.1 .mu.M, 0.3 .mu.M and 0.2 .mu.M. Hill coefficient value was 1.2 to 2.2. Effects of the competitive antagonist were examined with 10-fold concentration of the antagonist concentration used when examining the EC.sub.50 values. The strength in sensitivity to APV was in the order of .epsilon.1/.xi.1>.epsilon.2/.xi.1>.epsilon.3/.xi.1 and that to 7 CK was in the order of .epsilon.3/.xi.1>.epsilon.2/.xi.1>.epsilon.1/.xi.1. 0.1 mM and 1 mM of Mg.sup.2+ acted on the .epsilon./.xi. heteromeric NMDA channels voltage-dependently and repressively. However, clear difference can be found between these heteromeric channels. The .epsilon.3/.xi.1 channel showed resistance to Mg.sup.2+ inhibition, and showed activity at membrane potentials of -70 mV and -100 mV in the presence of 1 mM Mg.sup.2+. On the other hand, under the same conditions, the .epsilon.1/.xi.1 and .epsilon.2/.xi.1 channels were strongly repressed. These results suggest that functionally different NMDA receptor channels were formed according to the combination of the subunits.

Amino acid sequences highly preserved of mouse NMDA type receptor subunits, i.e. an oligo nucleotide sense primer, 5'-TGGAAT/CGGA/TATGATG/A/T/CGGG/A/T/CGA-3' (sequence ID No. 14 of the sequence table) corresponding to WNGMI/MGE (sequence ID No. 12 of the sequence table) existing at an upstream end of the membrane spanning region M1 and an oligo nucleotide antisense primer, 5'-GCG/A/TGCT/CAG/AG/ATTG/A/TGCG/A/T/CG/ATG/ATA-3' (sequence ID No. 15 of the sequence table) corresponding to YTANLAA (sequence ID No. 13 of the sequence table) in M3 were synthesized. Polymerase chain reaction (PCR) was carried out using the double strand cDNA as a template and the above synthesized oligo nucleotides as primers. PCR was carried out 30 cycles in total in which, after incubation at 94.degree. C. for 3 minutes in 50 .mu.l of the reaction solution containing 10 mM of Tris-HCl (pH 8.3), 50 mM of KCl, 1.5 mM of MgCl .sub.2, 0.001% of gelatin, 20 ng or less of a mouse cerebrum cDNA, 2 .mu.M of the respective primers, 200 .mu.M of 4 kinds of deoxynucleotide triphosphate and 4 units of Taq polymerase; 94.degree. C. for 1 minute; 50.degree. C. for 1 minute; and 72.degree. C. for 1.5 minutes were as one cycle.

After treating the PCR products by T4 DNA polymerase, they were inserted into the HincII site of pBluescript IISK (+) plasmid (Stratagene).

Subsequently, according to screening and base sequence determination, .epsilon.4 subunit cDNA clone was identified. The thus determined base sequence of the .epsilon.4 subunit cDNA and amino acid sequence expected therefrom were each shown in sequence ID No. 4 of the sequence table.

In the same manner as in Example 1 except for using a mouse cerebrum and cerebellum, cDNA library obtained from .lambda.gt10 was screened by using the .epsilon.4 subunit cDNA as a probe to obtain several clones which code the .epsilon.4 subunit. These phage-derived cDNA fragments were inserted into EcoRI site of the pBluescript IISK (-) plasmid (Stratagene) or pBKSA plasmid.

Base sequences of both chains of cDNA clones SE11 (from the -33rd to 2550th at the 5' end upstream in Sequence ID No. 16 of the sequence table) and SE4 (from the 2515th to +83st at the 3' end downstream in Sequence ID No. 16 of the sequence table) were determined according to the dideoxy chain termination method using various primers synthesized by the DNA autosysnthesizer (available from Applied Biosystem Co.). Partial base sequences of respective clones, SE1 (the -33rd of the 5' upstream region to 2001st of the sequence No. ID No. 16 in the sequence table), TSEE6 (the 1716th to 2291st of the sequence No. ID No. 16 in the sequence table) and K1 (the 2013rd to +83rd of the 3' down stream region of the sequence No. ID No. 16 in the sequence table) were completely identical with the sequences of the cDNA clones SE4 and SE11. The base sequences and the amino acid sequences were analyzed by GENETYX Software (SDC).

All coding area (the 1st to +3rd of the 3' down stream region in the Sequence ID No. 16 of the sequence table, i.e. until stop codon)) of the .epsilon.4 subunit cDNA was inserted by pSP35T plasmid and PCR method according to the PCR method between the NcoI and XbaI portions of the pSP35 plasmid (Cell, 66, pp. 257 to 270 (1991)) to obtain pSPGR .epsilon.4 plasmid.

Then, by using the pSPGR .epsilon.4 plasmid cut by EcoRI as a template and using SP6 RNA polymerase, .epsilon.4-specific mRNAs were synthesized in vitro. Transcription was carried out in the same manner as in Example 1.

The .epsilon.4 subunit-specific mRNAs (not more than 16 ng/egg) synthesized in vitro from a cloned cDNA and .xi.1.NMDA type glutamate receptor subunit-specific mRNAs (not more than 13 ng/egg) described above were used singly or in combination, they were injected into the Xenopus oocytes in the same manner as in Example 1. 10 .mu.l of the mRNA aqueous solution was injected into about 100 oocytes in an amount of 50 to 100 nl per one oocyte. The oocytes after the injection were treated in the same manner as in Example 1 and then used for an electrophysiological test.

In a frog standard Ringer's solution, the inward current of .epsilon.4 and .xi.1 subunits expressing oocytes at -70 mV membrane potential was 70.+-.9 nA (mean.+-.standard deviation, n=13) with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 68.+-.13 nA (n=8) with respect to 100 .mu.M NMDA and 10 .mu.M glycine (FIG. 1(a)). The current amplitude was extremely larger than that of the case where only the .zeta.1.NMDA type receptor subunit-specific mRNAs were injected to the oocytes (17.+-.1 nA, n=10, with respect to 10 .mu.M L-glutamate and 10 .mu.M glycine and 13.+-.2 nA, n=7, with respect to 100 .mu.M NMDA and 10 .mu.M glycine). The response of the heteromeric channel comprising the .epsilon.4/.xi.1 subunit to 10 .mu.M L-glutamate and 10 .mu.M glycine was suppressed by 500 .mu.M APV which is a specific competitive antagonist of the NMDA receptors or 100 .mu.M of 7 CK which had been reported as a competitive antagonist to glycine control site of the NMDA receptors (FIG. 2 and FIG. 3). According to the above, it can be concluded that .epsilon.4 protein is a subunit of the NMDA receptor channels.

Since the suppressive effect of the NMDA receptor competitive antagonist to the e4 subunit was relatively small, the pharmacological characteristics of the .epsilon.4/.xi.1 heteromeric channel were quantitatively examined by using a Ba.sup.2+ Ringer's solution. The EC.sub.50 values to L-glutamate or glycine obtained from the dose-reaction curve were 0.4 .mu.M and 0.09 .mu.M, respectively, and the Hill coefficients were 1.4 and 1.2, respectively (FIG. 2).

Apparent affinity of the .epsilon.4/.xi.1 heteromeric NMDA receptor channel to an agonist was stronger than those of the .epsilon.1/.xi.1, .epsilon.2/.xi.1 and .epsilon.3/.xi.1 heteromeric channels. The strengths in affinity to L-glutamate and glycine were in the order of .epsilon.4/.xi.1>.epsilon.3/.xi.1>.epsilon.2/.xi.1. The effects of APV and 7 CK were examined by 10-fold concentration of the EC.sub.50 value (FIG. 3). Activities of the .epsilon.4/.xi.1 channel were only decreased by 31% in 100 .mu.M APV and 34% in 3 .mu.M 7 CK. The degree of inhibition was smaller than those of the other .epsilon./.xi. channels observed under the same conditions. The sensitivities to APV were .epsilon.1/.xi.1>.epsilon.2/.xi.1>.epsilon.3/.xi.1>.epsilon.4/.xi.1,and those to 7 CK were .epsilon.3/.xi.1>.epsilon.2/.xi.1>.epsilon.1/.xi.1=.epsilon.4/.xi.1. According to the above, the .epsilon.4/.xi.1 heteromeric channel was characterized that it has strong affinity to an agonist and weak sensitivity to a competitive antagonist.

By using the mouse brain cDNA library obtained from .lambda.gt10 in Example 1, screening of the cDNA library was carried out. Then, KpnI/HindIII DNA fragments (1388 base pairs: previously cloned cDNA fragments of mouse glutamate receptors) of a pGRA19 plasmid obtained in Example 1 and mouse .alpha.1 and .alpha.2 subunit cDNAs (FEBS, Lett., 272, pp. 73 to 80 (1990)) were labeled with .sup.32 p and plaque hybridization was done in the presence of 30 % formaldehyde at 37.degree. C.

Partially specific modification was carried out by using a suitable synthesized oligo nucleotide and pBKSA .epsilon.2 obtained in Example 2 and pBKSA .xi.1 (FEBS, Lett., 300, pp. 39 to 45 (1992)) plasmid-derived DNA fragments according to the two-step polymerase chain reaction (PCR) method. The resulting modified 353 base pairs Cfr101SphIand 322 base pairs FspI/BlnI DNA fragments obtained by amplifying according to the PCR were substituted for the corresponding segments of pBKSA .epsilon.2 and pBKSA .xi.1, respectively. The nucleotide sequences of the constructed plasmids are different from those of the original plasmids as follows.

pBKSA .epsilon.2-N589Q (in the .epsilon.2 subunit described in the sequence ID No. 2 and No. 6 of the sequence table, asparagine which is an amino acid at the 589th from the N-terminal is modified to glutamine, A which is the 1765th base sequence to C and C which is the 1767th base sequence to G, respectively. The amino acid sequence and base sequence are shown in sequence ID No. 8 and No. 17 of the sequence table, respectively.)

pBKSA .zeta.1-N598Q (in the .zeta.1 subunit described in "FEBS Lett.", 300, pp. 39 to 45 (1992), asparagine which is an amino acid at the 598th from the N-terminal is modified to glutamine, A which is the 1792nd base sequence to C and C which is the 1794th base sequence to G, respectively. The amino acid sequence and base sequence are shown in sequence ID No. 9 and No. 18 of the sequence table, respectively.)

pBKSA .zeta.1-ZAZ (in the .zeta.1 subunit described in "FEBS Lett.", 300, pp. 39 to 45 (1992), the 1726th to 1743rd base sequence is modified to ACCAGTGACCAGTCAAAT. The amino acid sequence and base sequence are shown in sequence ID No. 10 and No. 19 of the sequence table, respectively.)

Next, by using the pBKSA .epsilon.2 and pBKSA .zeta.1 plasmids and derivatives thereof cut by suitable restriction enzymes as template and using T3 RNA polymerase produced by BRL Co., .epsilon.2, .zeta.1 and their derivatives-specific mRNAs were synthesized in vitro, respectively. Transcription was carried out in the same manner as in Example 1.

In order to examine influence on the modified ion channels, the wild type or modified .epsilon.2 (not more than 19 ng/oocyte) subunit-specific mRNAs and the wild type or modified .zeta.1 (not more than 13 ng/oocyte) subunit-specific mRNAs were injected singly or in combination into the Xenopus oocytes in the same manner as in Example 1. 10 .mu.l of the mRNA aqueous solution was injected into about 100 oocytes in an amount of 50 to 100 nl per one oocyte. The oocytes after the injection were treated in the same manner as in Example 1 and then used for an electrophysiological test.

FIGS. 4(a) to 4(d) show a current response of the heteromeric NMDA receptor channel to 10 .mu.M L-glutamate and 10 .mu.M L-glycine at -70 mV membrane potential in a frog standard Ringer's solution. The wild type .epsilon.2/.zeta.1 NMDA receptor channel was strongly inhibited by 1 mM Mg.sup.2+, 100 .mu.M Zn.sup.2+ and 1 .mu.M (+)-MK-801. MK-801. To the contrary, the modified .epsilon.2/.zeta.1-N598Q channel showed a great current response even in the presence of 1 mM Mg.sup.2+. However, the response was suppressed effectively by 100 .mu.M Zn.sup.2+ and 1 .mu.M (+)-MK-801. Similarly, in the .epsilon.2-N589Q modified product, sensitivity to Mg.sup.2+ inhibition was decreased without changing sensitivity to Zn.sup.2+. To the .epsilon.2/.zeta.1 and .epsilon.2/.zeta.1-N598Q channels, the .epsilon.2-N589Q/.zeta.1 channel showed a great current response after repetitive application of (+)-MK-801. The heteromeric .epsilon.2-N589Q/.zeta.1-N598Q channel showed strong resistance to Mg.sup.2+ and (+)-MK-801 inhibitions, but still had sensitivity to Zn.sup.2+.

In order to minimize effects of a secondary activated Ca.sup.2+ dependent Cl- current, influences of these modified products on Mg.sup.2+ inhibition were measured more quantitatively in a Ba.sup.2+ Ringer's solution. FIGS. 5(a) to 5(d) each represent current-voltage curves of the wild type and modified heteromeric channels in the presence of 1 mM Mg.sup.2+ () and in the absence thereof (.smallcircle.). As observed in the case of the NMDA type receptor channel, Mg.sup.2+ inhibited a current response to the wild type .epsilon.2/.zeta.1 channel depending on voltage (FIG. 5(a). Sensitivity of the heteromeric channel to Mg.sup.2+ was decreased greatly by modification of an asparagine residue(s) of one or both of the subunits, and the modified channel retained activity even at -100 mV membrane potential (FIGS. 5(b) to 5(d). Inhibition degrees at various Mg.sup.2+ concentrations were compared at -70 mV membrane potential (FIG. 6). Activities of the wild type .epsilon.2/.zeta.1 channel and the .epsilon.2/.zeta.1-ZAZ channel were decreased by 50% by not more than 20 .mu.M Mg.sup.2+ and suppressed almost completely by not more than 1 mM Mg.sup.2+ which is a physiological concentration. On the other hand, about 100-fold concentration of Mg.sup.2+ was required to depress the modified .epsilon.2/.zeta.1-N598Q, .epsilon.2-N589Q/.zeta.1 and .epsilon.2-589Q/.zeta.1-N598Q channels.

By administering 1 .mu.M (+)-MK-801 which is an open channel blocker of the NMDA type receptor channel repeatedly, the wild type .epsilon.2/.zeta.1 channel was suppressed almost completely (FIG. 7). Similarly, activities of the modified .epsilon.2/.zeta.1-N598Q and .epsilon.2-N589Q/.zeta.1 channels were strongly inhibited by continuous application of (+)-MK-801. On the other hand, the modified .epsilon.2-N589Q/.zeta.1-N598Q channel retained high activities even after (+)-MK-801 was applied three times repeatedly.

Effect of Zn.sup.2+ which had been reported to induce non-competitive inhibition to the NMDA type receptor not depending on voltage was examined. Sensitivity of the heteromeric channel to 10 .mu.M Zn.sup.2+ was decreased a little by modification. However, the modified channel was strongly inhibited by 100 .mu.M Zn.sup.2+ (FIG. 8).

These results suggested that the NMDA type receptor channels having different functions were formed by modification of the subunits.

The NMDA type glutamate receptor genes of the present invention are not only useful for clarifying nervous information transmission at a synapse, appearance of synapse plasticity which is basically required for memory and learning and neuronal cell death caused by a disease such as cerebral ischemia and epilepsy and understanding a transmission mechanism of nervous information in a center, a cerebral structure of higher order and a disease of the brain, but are also useful for therapy of genetic diseases and preparation of novel pharmaceuticals (e.g. screening of an agonist or an antagonist).