BACKGROUND OF THE INVENTION Voltage-dependent calcium channels have been linked to physiological processes such as neurotransmitter release, secretion of hormones, muscle contraction, and regulation of gene transcription. A functional channel requires at least three subunits, including the al, a28, and P subunits. The channel may also contain a y subunit. There are several known types of voltage-dependent calcium channels that have been defined based on their electrophysiological characteristics and pharmacological properties. These types are L-, N-, P/Q-, R-, and T-type.

Each type is primarily defined by its channel composition. The type of a 1 subunit contained in the channel determines whether the channel is an L-, N-, P/Q-, R-, or T-type channel. The activity of the a 1 subunit is modulated by the a28 and subunits. Channel activity may be further modulated by a fourth subunit, y.

Molecular biological techniques have allowed elucidation of the mechanism of voltage-dependent calcium channel action. Genes for each of the subunits have been isolated and cloned. There are currently nine known genes encoding for different al subunits. The al subunit forms the pore which calcium ions flow through. The al subunit contains the voltage sensor and is also responsible for the binding specificity of certain drugs or toxins that may be associated with the channel type. Channel current through the al pore may be modulated by association of the p, y, or oc26 subunit. There are four known genes

for the intracellular P subunit that may be differentially spliced. There are two known genes for the transmembrane y subunit, one in skeletal muscle and a novel gene expressed in the brain. Only one isoform of a26 was initially identified.

Recently, however, two new a28 genes were identified, a282 and a283. These genes have 55.6 and 30.3% homology with the original a281 gene (Klugbauer, et al., J. Neuroscience 1999; 19 (2): 684-691). The a2 and 8 proteins are expressed by the same gene. The protein product is post-translationally cleaved, and the final a2 and 8 proteins are linked by disulfide bonds. The transmembrane 8 protein secures the a2 protein to the cell membrane.

Studies have shown that the a281 subunit contains a binding site for the anticonvulsant drug, gabapentin [1- (aminomethyl) cyclohexane acetic acid] (Gee, et al., J : Biol. Chem. 1996; 271 (10): 5768-5776). Gabapentin is a y-aminobutyric acid (GABA) analogue. Gabapentin is effective in the treatment of epilepsy and in decreasing seizure frequency in both animal models and in human patients. The precise mechanism of action of gabapentin remains unclear. Recent experiments have shown that gabapentin also binds to the a262 subunit.

Functional channels may be formed by expression of the calcium channel subunits in a cell. This technique is advantageous in determining the effects of various molecules on channel action. US 5,712,158 and US 5,770,447 describe a stable cell line that is useful for investigating gabapentin binding properties to calcium channel subunits. This cell line expresses the ß subunit and the original a28 subunit (now referred to as a281) at high levels. Transfecting the cells with any al subunit results in the formation of functional calcium channels which can be used to evaluate the binding of gabapentin and gabapentin-related compounds.

It is the object of this invention to provide a new cell line that stably expresses a calcium channel a282 subunit. It is a further object of this invention to describe a282 subtype-specific binding of gabapentin, analogues of gabapentin, pregabalin, analogues of pregabalin, and 3-alkyl derivatives of GABA.

SUMMARY OF THE INVENTION The invention provides a method for determining the binding ability of a compound to an a282 subunit of a calcium channel comprising: providing an a282 subunit of a calcium channel, contacting the a282 subunit with the compound, and determining the binding ability of the compound to the a282 subunit.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 diagrams the molecular cloning of human a282 into the pCDNA3.1 expression vector.

Figure 2. RT-PCR Analysis of Human a28 Tissue Distribution. One ng of double-stranded cDNA from different human tissues (CLONTECH) was amplified by PCR with 35 cycles of 94°C for 1 minute, 55°C for 1 minute, and 72°C for 2 minutes. The generated PCR products represent DNA fragments from nucleotide 958 to 2165 (ha281), 2534 to 3643 (ha282), and 1920 to 3272 (ha283).

Figure 3. Northern Blot Analysis Human a28 Tissue Distribution.

Northern blotting was carried out as described in Materials and Methods. Human multiple tissue blots (CLONTECH) were hybridized with Digoxigenin-labeled cDNA synthesized from nucleotide 958 to 2165 (ha281), 2534 to 3643 (ha282), and 1920 to 3272 (ha283). The positions of marker RNA are indicated to the left.

Figure 4. Western Blot Analysis of Human and Mouse a28 Tissue Distribution. Membrane proteins from different human tissues (A, 0.5 jj. g) and mouse tissues (B, 100 Fg) were loaded on 4% to 20% SDS-PAGE (NOVEX) and subjected to Western blot analysis (see Materials and Methods). The blots were probed with anti-a28 monoclonal antibody or polyclonal antibodies against a282 and a283.

Figure 5. Binding of [3H] Gabapentin to Membranes From COS-7 Cells Transfected With a28 cDNA. COS-7 cells were transfected with 20 ag of

pcDNA3.1 (control), pcDNA3.1/porcine a281 construct (pa281), and pcDNA3.1//human a252 construct (ha252). The membranes were prepared for [3H] gabapentin binding assays (see Materials and Methods). Data are an average of three independent assays with triplet in each determination. The same membranes (100 jj, g) were subjected to Western blot analysis with corresponding antibodies as described in Materials and Methods.

Figure 6. Disruption of Disulphide-Linkage Between a2 and 8 Subunits.

An equal amount of membrane protein from each sample (0.5 ig for pa281 and 5 pg for ha282) was incubated in the presence or absence of 100 mM DTT for 10 minutes and resolved on a nonreducing SDS-PAGE and transferred to a PVDF membrane. The blots were probed with either an anti-a251 antibody (left) or an anti-a282 antibody (right). The positions of marker proteins are indicated to the right.

Figure 7. Scatchard Analysis of [3H] gabapentin (GBP) Binding to Membranes Form HEK293 Cells Overproducing Porcine a251 (A) and Human a282 (B). The cell membranes were prepared from GKS02, a stable cell line for porcine a281, and GKS07, a stable cell line for human a2â2. The specific [3H] gabapentin binding was carried out as described in Materials and Methods.

The binding activity was expressed as pmole of gabapentin bound per mg of protein. Each binding reaction contained 20 ug of GKS02 membrane proteins or 10 ig of GKS07 membrane proteins. Data were averages of three assays.

Figure 8. Screening Cell Lines by [3H] Gabapentin (GBP) Binding Activity. HEK293 cells were transfected with human a282. Single clones were selected by G418-resistance."2923,"parental cells HEK293 ;"2L,"HEK293 cells stably expressing porcine a28l.

DETAILED DESCRIPTION OF THE INVENTION As used herein, analogues of gabapentin include but are not limited to alkyl-substituted gabapentin analogues, bridged gabapentin analogues, and heterocyclic gabapentin analogues such as those described by Bryans, et al. in J Med. Chem. 1998; 41: 1838-1845. Analogues are defined as"compounds with similar electronic structures but different atoms" (Grant, et al., Chemical Dictionary, 5th ed., McGraw-Hill, 1987). Gabapentin has the structure:

Examples of gabapentin analogues are described in Bryans, et al., supra, and include, but are not limited to: A molecule with the structure :

This analogue is alkylated at position 3 on the cyclohexane ring. An analogue may be alkylated at any position on a carbon ring with an alkyl group of from 1 to 4 carbon atoms. An analogue may also be a molecule with the structure:

This analogue is alkyl-substituted at the 3-position of the gabapentin ring.

Molecules of this type include pregabalin its analogues, and 3-alkyl derivatives of GABA.

MATERIALS AND METHODS Porcine a281 (pa281) cDNA was from J. Brown (Brown J. P., Dissanayke V. U. K., Briggs A. R., Milic M. R, Gee N., Anal. Biochem., 1998; 255: 236-243).

Mouse a283 (ma283) cDNA was a generous gift from F. Hoffman (Klugbauer N., Lacinova L., Marais E., Hobom M., Hofmann F., R Neurosci., 1999; 19: 684- 691). Monoclonal antibody against oc281 was purchased from Affinity Bioreagents, Inc. Polyclonal antibodies against a282 and a283 were from Sandra Duffy (Pfizer). Human and mouse multiple tissue blots and cDNA were purchased from CLONTECH. Mouse tissues were purchased from Pel-Freez Biologicals.

PCR reagents were from Invitrogen. ECL Western blot kit was from Armersham.

Lipofectamine, growth media, restriction enzymes were from LifeTechnologies.

HEK293 and COS-7 cell lines were from ATCC. All other chemicals were from Sigma.

Cloning of human a282 subunit. Human a282 (ha282) cDNA was amplified by PCR from a human brain cDNA library. Based on the deposited DNA sequence of hot262 subunit from GenBank (accession number AF042792), four overlapped DNA fragments, which covered the whole open reading frame of ha282 cDNA from nt-14 to 994 (fragment H), 845 to 1816 (fragment F), 1517 to 2791 (fragment D), and 2681 to 3790 (fragment C), were generated by PCR and then cloned into expression vector pcDNA3.1 by TA cloning kit. The sequences of the primer pairs used were: 5'-TCTTGAATGGAAACATGGCGGTGC-3' (SEQ ID No. 1) and 5'-TATACCAGGGTCTCCTTCGGACAT-3' (SEQ ID No. 2) (fragment H); 5'-ATGTGTTCATGGAAAACCGCAGAC-3' (SEQ ID No. 3) and 5'-AGCCGTTCAGGTCAATGGCAAACA-3' (SEQ ID No. 4) (fragment F); 5'-CCATCCGCATCAACACACAGGAAT-3' (SEQ ID No. 5) and 5'-GTAAGTCCTCATTGTTAACCTCGC-3' (SEQ ID No. 6) (fragment D); 5'-CTGAGAAGTTCAAGGTGCTAGCCA-3' (SEQ ID No. 7) and 5'-GATGTGATTTGGGTGCCAAACACC-3' (SEQ ID No. 8) (fragment C). The four fragments were cut at internal unique restriction enzyme sites at nt 791

(PflM I), 1395 (Xba I), and 2628 (Hpa I), and assembled into pcDNA3. 1 vector (Invitrogen, Carlsbod, CA) at Hind III/Xho I sites (see Figure 1).

RT-PCR. Double-stranded cDNA preparations from different tissues (CLONTECH) were used for PCR reaction with 35 cycles at 94°C for 1 minute, 55°C for 1 minute, and 72°C for 2 minutes. The reactions were performed in a solution containing 1 ng cDNA, 10 pM primers, 1 mM dNTPs, and 1 x PCR buffer in a volume of 50 pL. Ten microliters of the reaction mix was loaded on 1% agarose gel. The primer pairs for human a281, a2b2, and a263 were 5'-GACGCGGTGAATAATATCACAGCC-3' (SEQ ID No. 9) and 5'-ACAAATCGTGCTTTCACTCCCTTG-3' (nt 958 to 2165; accession number M76559) (SEQ ID No. 10) ; 5'-CTGAGAAGTTCAAGGTGCTAGCCA-3' (SEQ ID No. 11) and 5'-GATGTGATTTGGGTGCCAAACACC-3' (nt 2534 to 3643 ; accession number AF042792) (SEQ ID No. 12); and 5'-CGTGTCCTTGGCAGATGAATGGTC-3' (SEQ ID No. 13) and 5'-CATCTCAGTCAGTGTCACCTTGAG-3' (nt 1920 to 3272; accession number AJ272213) (SEQ ID No. 14), respectively. The expected lengths of PCR products from human a281, a282, and a283 were 1208,1110, and 1352 bp. These primers were specific for each subtype of a28 as determined by sequencing analysis of the corresponding PCR products.

Northern blot analysis. Multiple Tissue Northern Blots (CLONTECH) were hybridized and washed according to the manufacturer's recommendation.

Digoxigenin-labeled probes specific for subtypes of a28 were generated by PCR and hybridized in 10 mL EasyHyb (Boehringer Mennhaim) at 50°C overnight.

The same pairs of primers as those used for RT-PCR were employed to generate the probes. The blots were washed twice, first in 2 x SSC and 0.1% SDS at room temperature for 5 minutes, then in 0.1 x SSC and 0.1% SDS at 68°C for 15 minutes. Detection of expression was in accordance with the manufacturer's instructions (Boehringer Mennhaim).

Cell culture and transfection. COS-7 and HEK293 cells were cultured in DMEM and RPMI 1640 media, respectively. The media were supplemented with

50 units/mL penicillin, 50 pg/mL streptomycin, and 10% heat-inactivated fetal bovine serum (FBS), in a humidified incubator with 95% air and 5% C02 at 37°C.

For transient transfection into COS-7 cells, 20 plg of plasmid DNA (vector or the same vector with a28 insert) was incubated with 30 u. L oflipofectamine. The mixture was overlaid onto the cells in 1.5 mL serum-free medium and incubated for 5 hours. Then FBS was added to the dishes to bring the final concentration to 10%. The medium was changed next morning. Forty-eight hours after the transfection, the cells were harvested for membrane preparation. For stable transfection of porcine a2a 1 and human a282 into HEK 293 cells, the same procedure was applied as that for a transient transfection except for that 800 Lg/mL G418 (gentacin) was added to the cells 48 hours after the transfection.

Two clones, GKS02 and GKS07, showed highest expression of porcine a281 and human a282, respectively, and were selected for further binding studies. The cell line has ATCC No. PTA-1823. In addition, hosts for expression of a282 protein binding assays can also include eukaryotic expression systems such as yeast, insect cells, and mammalian cells (CHO, COS-7, HEK293, etc.).

Membrane preparation. Membranes were prepared from tissues or cultured cells. The cells were washed twice with cold PBS and then scraped off with cold buffer containing Tris (5 mM, pH 7.4), EDTA (5 mM), PMSF (0.1 mM), leupeptin (0.02 mM), and pepstatin (0.02 mM). The cells were incubated on ice for 30 minutes, followed by sonication for 30 to 40 seconds. For membrane preparations from tissues, the tissues were sliced into small pieces and subjected to sonication at interval of 10 seconds 4 times. The resulting homogenates from tissues or cultured cells were centrifuged for 10 minutes at 750 to 1000 x g, and then the supernatants were centrifuged at 50,000 x g for 30 minutes. The resulting pellets were resuspended in the same buffer as described above.

Western blot analysis. The cell membranes (0.5 llg for GKS07 cells, 5 ig for GKS02 cells, 100 pg for transiently transfected cells or tissues) were resolved by 4% to 20% SDS-PAGE and transferred to nitrocellulose membranes using semi-dry transferring unit. The membranes were incubated with either rabbit anti- a281, a282, and a283 antibodies for 1 hour at room temperature, followed by

washing with 1 x PBS. The blots were incubated with anti-rabbit IgG for 1 hour and developed with ECL reaction according to the procedure recommended by manufacturer.

Binding assays. The radioligand-binding assay was done using membrane proteins incubated in the presence of 20 nM [3H] gabapentin. The membranes (100 ug of proteins for transiently transfected cells, 20 pg for GKS02 cell membranes, and 10 u. g for GKS07 cell membranes) were incubated in 10 mM HEPES (N- [2-hydroxyethyl] piperazine-N'- [2-ethanesulfonic acid]) (pH 7.4) for 40 to 50 minutes at room temperature, and then filtered onto pre-wetted GF/C membranes and quickly washed five times with 3 mL of ice cold 50 mM Tris buffer (pH 7.4). The filters were dried and counted in a liquid scintillation counter.

For determining nonspecific binding, the binding assays were performed in the presence of 10 uM pregabalin (Gee NS., Brown J. P., Dissanayake V. U., Offord J., Thurlow R., Woodruff G. N., R Biol. Chenu., 1996; 271: 5768-5776). The specific binding was obtained by subtracting nonspecific binding from the total binding.

Clone #7 was identified as the highest a282 subunit expressing clone. Binding assays can also be performed using recombinant and/or purified a282 protein from human and other mammalian species, for screening a282 subtype-selective inhibitors.

Results Tissue distribution of oc26 transcripts. Tissue distribution of ha281, ha282, and ha283 mRNA was first examined by RT-PCR analysis. These probes were designed to specifically amplify three subtypes of a28. As shown in Figure 2, single PCR products corresponding well to the predicted sizes of ha261, ha282, and ha283 (1208,1110, and 1352 bp) appeared in almost all tissues tested.

A much higher level of ha282 transcript was found in lung than any other tissues including brain. Since the PCR products showed sequences identical to the corresponding a28, the wide scope of tissue distribution revealed the ubiquitous feature of hoc26 mRNA expression. However, the RT-PCR condition used here did not yield quantitative estimation of a28 mRNA levels among different tissues, Northern analysis is necessary for estimating the relative abundance of each

subtype of ha28 mRNA. Northern blots demonstrated that all three ha28 genes were expressed about equally well in brain, heart, and skeletal muscle except for the much higher expression of ha281 in skeletal muscle (Figure 3). In addition to these three tissues, the most abundant ha282 transcript was found in lung. The highest expression of ha282 mRNA in lung was consistent with the above described RT-PCR results and also agreed well with one recent report (Gao B., Sekido Y., Maximov A., Saad M., Forgacs E., Latif F., et al., R Biol. Chemin., 2000; 275: 12237-12242), but differed from an early observation (Klugbauer N., Lacinova L., Marais E., Hobom M., Hofmann, F., J Neurosci., 1999; 19: 684-691).

In the present study we also detected a small amount of ha281 and ha283 mRNAs in liver and kidney, respectively. Results from this and other laboratories (Klugbauer, Supra., 1999; Gao, Supra., 2000, and our unpublished data) have shown that expression of mouse oc263 (ma283) is restricted to the brain. The expression of ha253 also in tissues other than brain suggested species difference in a283 expression.

In the brain, ha281, ha282, and ha253 were detected in every portions of brain tissues tested including cerebellum, cerebral cortex, medulla, occipital pole, frontal lobe, temporal lobe, and putamen. A higher level of ha252 transcript was found in cerebellum than cerebral cortex, while reverse was true for ha253. For ho 261, its mRNA was approximately equally distributed in these two regions. The expression patterns of the three isoforms in these two brain regions were in accordance with previous in situ hybridization results (Klugbauer, Supra., 1999; Hobom M., Dai S., Marais E., Lacinova L., Hofmann F., Klugbauer N., Eur. J.

Neurosci., 2000; 12: 1217-1226). In addition, all three subtypes of a28 mRNA were found in spinal cord, but at lower levels than that found in the brain.

Tissue distribution of oF26 proteins. Although the level of protein is function of the steady-state level of mRNA, the relative abundance of mRNA and protein of specific gene is not always proportional, which may reflect post- transcriptional regulation (Jackson V. N., Price N. T., Carpenter L., Halestrap A. P., Biochem. J, 1997 ; 324: 447-453). To examine the relative levels of human and mouse a28 subunits across tissues, we used antibodies raised against specific subtypes of a26 protein for Western analysis. Equal amounts of proteins were

loaded on SDS polyacrylamide gels. Consistent with the ubiquitous distribution of ha281, Western blots of human and mouse tissues showed that both ha281 and ma281 proteins were widely distributed, although ha281 in lung and jejunum were not detectable. By contrast, ha283 protein was only detected in brain, not in lung, testis, aorta, spleen, jejunum, and kidney (Figure 4A). Similarly, ma283 protein was found only in brain, not in heart, kidney, liver, lung, pancreas, stomach, spleen thymus, ovary, pituitary, thyroid, and prostate. Surprisingly, in contrast to predominant expression of ha282 transcript in lung (Figures 2 and 3), ha282 protein was predominantly found in brain and the level of ha282 protein was not detectable in lung (Figure 4A). In addition to brain, low levels of ha282 protein were also found in aorta, testis, and ventricular muscle. There seemed to be two immunoreactive bands in testis with one equivalent to predicted molecular weight of ha282 (175 kDa) and the other showing slightly lower molecular weight. This lower molecular protein appeared to be similar to the predominant band detected in ventricular muscle. As previously observed with pa281, this lower band may represent the dissociated a2 subunit from the a28 protein or an isoform of 282 (Brown J. P., Dissanayke V. U. K., Briggs A. R., Milic M. R, Gee N., Anal. Biochem., 1998; 255: 236-243; Wang M., Offord J., Oxender D. L., Su, T. Z., Biochem. J, 1999 ; 342: 313-320). In addition, two immunoreactive bands were also detected in mouse heart by anti-a282 antibodies, but the predominant band in this case had molecular weight higher than that found in other tissues (Figure 4B).

Disulphide linkage of oc2 and 8 proteins. It has been shown that a2 and 8 subunits of a281 were linked by disulphide bond (Wang, Supra., 1999). Since the amino acid sequence in 8 region is less conserved between a281 and a282, it is interesting to know if a282 protein is also cleaved into two subunits post translation. To examine such a possibility, cell membranes from HEK 293 cell lines overproducing pa281 (GKS02) and ha282 (GKS07) proteins were treated or untreated with 100 mM DTT before gel electrophoresis. In the presence of DTT, both pa281 and ha282 proteins were shifted to a position predicted for a2,

suggesting that as with pa281, ha252 also consists of two subunits that are linked by disulphide bond (Figure 6).

[3H] Gabapentin Binding. To determine the gabapentin binding properties of the cloned ha282, membranes were isolated from COS-7 cells transiently transfected with pa281, ha282, and vector pcDNA3.1. Expression of the corresponding oc26 proteins was examined by Western blots. As shown in Figure 5, transfection of the cells with pa281 resulted in a prominent increase in gabapentin binding. Similarly, the cells expressing ha252 exhibited about fourfold increase in gabapentin-binding activity. Although a slightly increased binding activity was observed in the cells transfected with pcDNA3.1 vector alone, statistic analysis did not show that this smaller change was significant.

Gabapentin binding KD and the binding properties of pa281 and ha282 were determined in cell lines GSK02 (pa281) and GKS07 (ha282). In HEK293 cells stably expressing pa281, [3H] gabapentin bound to a single population of sites as demonstrated in previous report (Gee, Supra., 1996) with KD value of 72 9 nM (Figure 7A). Similarly, a single population of binding sites were also observed in ha282-containing membranes (Figure 7B), but the KD value was higher than that of pa281 (156 25 nM). To determine pharmacological properties of ha282, several compounds were selected for competition with [3H] gabapentin binding. A similar, but not identical profile of competition was seen in the two subtypes of a28 protein (Table 1). For instance, binding to both subtypes of a28 were stereo-selective because L-leucine was markedly more potent than its D-enantiomer. The affinities of BCH, a model substrate of system L transport (Su T. Z., Lunney E., Campbell G., Oxender D. L., J Neurochem., 1995; 64: 2125-2131), and phenylalanine were weak for both subtype proteins. On the other hand, gabapentin binding to a2b2 was more sensitive to (S+)-3-isobutyl GABA (pregabalin) with ICso value of 96 nM as compared to 149 nM for pa281.

Table 1. ICso Values for Inhibition of [3H] Gabapentin Binding to Membranes From Stable Cell Lines Overproducing Porcine a281 (GKS02) and Human oc282 (GKS07) by Selected Amino Acids CompoundsGKS02 (pa281) GKS07 (ha282) Gabapentin 132 282 Pregabalin 149 96 L-leucine 118 205 L-phenylalanine 825 2,960 D-leucine 198,960 151,510 BCH 1,028 775 Figure 8 also illustrates the screening of stable cell lines that express human a282 protein.

While the forms of the invention herein disclosed constitute presently preferred embodiments, many others are possible. It is not intended herein to mention all of the possible equivalent forms or ramifications of the invention. It is understood that the terms used herein are merely descriptive, rather than limiting, and that various changes may be made without departing from the spirit or scope of the invention.