BACKGROUND OF THE INVENTION There are three major classes of GABA receptors in the central nervous system (CNS): GABAA, GABAB and GABAC receptors. The pharmacology of GABAA and GABAB receptors has been extensively investigated, but GABAC receptors have been only recognised recently, and their pharmacological potential is still unknown (Johnston, 1996b). y-Aminobutyric acid (GABA) is the main inhibitory neurotransmitter in the central nervous system (CNS), and activates three major subtypes of GABA receptors, the GABAA, GABAB and GABAC receptors. GABAA receptors are ligand-gated Cl channels which are inhibited by the alkaloid bicuculline (Johnston, 1996a). These are heterooligomeric receptors made up of oc, ß, y, and 8 subunits. GABAB receptors are transmembrane receptors 2+ + coupled to second messenger systems and Ca and K channels via G-proteins. These receptors are not blocked by bicuculline, but are activated by ) baclofen and 3-aminopropylphosphinic acid (CGP27492) and blocked by phaclofen and saclofen (Kerr and Ong, 1995).

GABAC receptors (sometimes called GABANANB or r receptors) were first proposed when a series of conformationally restricted GABA analogues, including cis- 4-aminocrotonic acid (CACA), that had bicuculline- insensitive depression actions on neuronal activity, showed no affinity for [3H] baclofen binding sites in rat

cerebellar membranes (Drew et al, 1984). GABAC receptors with similar pharmacology were first found in neurons from rat retina (Feigenspan et al, 1993) and white perch retina (Qian et al, 1993). In rat retina, rod bipolar cells contain bicuculline-insensitive, baclofen-insensitive receptors that were activated by CACA (Feigenspan et al, 1993). These were detected by the co-application of GABA with 100 uM bicuculline to abolish the GABAA component (Feigenspan et al, 1993). In white perch retina, rod- driven horizontal cells (H4) and not bipolar cells showed GABAC receptor-like pharmacology. Application of GABA on bipolar cells showed rapid desensitisation, while on rod- driven horizontal cells, desensitisation was not observed (Qian et al, 1993). Subsequently, GABAC receptors were found on cone-driven horizontal cells in catfish (Dong et al, 1994) and bipolar terminals in tiger salamander (Lukasiewicz et al, 1994).

The expression of mRNA from bovine retina in Xenopus oocytes showed that GABA activated two distinct GABA receptors. Both receptors activated Cl currents.

One was mediated by GABAA receptors and was blocked by bicuculline, and the other was mediated by GABAC receptors and was insensitive to both bicuculline and baclofen (Polenzani et al, 1991). Subsequently, two cDNAs that have 30-38% sequence identity with GABAA receptor subunits were cloned from human retinal mRNA (Cutting et al, 1991; 1992).

These subunits have been termed r1 and r2, and have 74% sequence identity (Cutting et al, 1991; 1992).

At least two major subtypes of GABAC receptors are now known, namely rho-1 and rho-2. As is known for other neurotransmitter receptor subtypes, different subtypes of GABAC receptors are likely to be involved in different aspects of nervous system function. As the rho-2 subunit is found in the hippocampus and neocortex, and these areas of the brain are important for memory, potent and selective agents for the rho-2 GABAC receptor are key compounds.

The species equivalents of the human r 1 and r 2 subunits have been cloned from rat (Enz et al, 1995).

These show 88-99% homology with the respective human sequences. The use of PCR and in Si tu hybridisation have shown high expression of both the r 1 and r 2 subunits in rod bipolar cells. However, only the r 2 subunit is expressed in the CNS, particularly in the hippocampus and cortex (Enz et al, 1995). Recently, a third r subunit was cloned from rat retina cDNA (Ogurusu and Shingai, 1996).

This subunit exhibits 63% and 61% sequence homology to the human r 1 and rat r 2 sequences respectively (Ogurusu and Shingai, 1996).

Expression of human r subunits in Xenopus oocytes generates homooligomeric GABA receptors with intrinsic Cl channels. These receptor ion channels are activated by GABA and CACA, but are insensitive to bicuculline, (-) baclofen, barbiturates and benzodiazepines. They have been shown to be sensitive to picrotoxin, and have been classified as GABAC receptors (Cutting et al, 1991; 1992; Polenzani et al, 1991; Shimada et al, 1992; Kusama et al, 1993a; 1993b; Wang et al, 1994; Bormann and Feigenspan, 1995; Johnston, 1996b).

The most potent GABAC receptor agonists known so far are trans-4-aminocrotonic acid (TACA, KD = 0.6 uM) and GABA (KD = 1.7 uM) (Woodward et al, 1993). TACA, a conformationally restricted analogue of GABA in an extended conformation, is also a GABAA receptor agonist (Johnston, 1996a). CACA, a conformationally-restricted analogue of GABA in a folded conformation, has moderate partial agonist activity at GABAC receptors (KD = 74 uM), and may be the most selective agonist for this receptor subtype (Johnston, 1996b).

Selective agonists and antagonists are needed to determine the physiological role of GABAC receptors and to provide more specific therapeutic agents with a lower risk of unwanted side-effects. GABA is a flexible compound, due to its rotation about the C2-C3 and C3-C4 bonds. It can

exist in a range of low energy conformations (Johnston et : al, 1978; Allan and Johnston, 1983). Two of these conformations have been restricted by the introduction of unsaturation in the form of a double bond at the C2-C3 position, and two compounds that represent these restricted conformations are CACA and TACA (Johnston et al, 1975).

CACA and TACA have fewer degrees of rotational freedom than GABA, and can only rotate about the C3-C4 bond (Johnston et al, 1978; Allan and Johnston, 1983). CACA is a partially folded analogue of GABA. It has moderate activity at GABAC receptors expressed in Xenopus oocytes, and although its agonist activity is weak, it is to date the most selective agonist at these receptors, having minimal activity on GABAA and GABAB receptors (Johnston, 1996b). TACA is an extended analogue of GABA. It has potent agonist activity at GABAC receptors expressed in Xenopus oocytes; however, it is not selective, as it is also a potent GABAA receptor agonist (Johnston, 1996b).

Woodward et al (1993), using poly (A) + RNA from mammalian retina expressed in Xenopus oocytes, tested many GABAA and GABAB receptor agonists and antagonists to determine a pharmacological profile for GABAC receptors.

From this study, it was found that the phosphinic and methylphosphinic analogues of GABA, which are known to be potent GABAB receptor agonists, were potent antagonists at GABAC receptors.

A series of GABA analogues was tested for agonist and antagonist activity at GABAC receptors, using poly (A) +. RNA from mammalian retina injected into Xenopus oocytes. Several potent GABAC receptor antagonists were identified, including (3-aminopropyl) methylphosphinic acid (CGP35024; KB = 0.8 uM), 3-aminopropylphosphinic acid (CGP27492; KB = 1.8 uM), and 3-aminopropylphosphonic acid (3-APA, KB = 10 uM) (Woodward et al, 1993). These agents are not selective for GABAC receptors, as CGP35024 and CGP27492 are also very potent GABAB receptor agonists, while 3-APA is a GABAB receptor antagonist.

To date, only one specific GABAC receptor antagonist has been described. A more recently synthesised compound, 1,2,5,6-tetrahydropyridine-4-yl) methylphosphinic acid (TPMPA), does show potent and selective GABAC receptor antagonist activity (KD = 2.1 uM) (Murata et al, 1996; Ragozzino et al, 1996). TPMPA produces 50% inhibition of GABAC receptor activation at 2.1 uM, and has the following structure: TPMPA The effects of TPMPA on cognition are unknown.

As a result of the structure-activity relationship study and the selectivity of CACA for GABAC receptors, we have investigated the methylphosphinic acid and phosphinic acid analogues of CACA and the closely related analogue, TACA, as potential GABAC receptor antagonists. In this study, we demonstrate that the phosphinic and methylphosphinic acid derivatives of CACA and TACA, and 3-aminopropyl-n-butyl-phosphinic acid (CGP36742), an orally-active GABAB receptor antagonist, are GABAC receptor antagonists, and we have linked GABAC receptors with cognitive function. Extensive structure- activity studies were carried out on recombinant GABAC receptors from human retina expressed in frog oocytes.

Among the compounds studied were a variety of compounds known to interact with GABAB receptors, provided by Ciba- Geigy AG, Basle.

The most interesting of the Ciba-Geigy compounds were a series of GABAB receptor antagonists that had been investigated in various memory and learning tests in rats and mice. Only one compound of the series reversed age-

related deficits of old rats (Froestl, 1995b). The cognition-enhancing effects of this compound were confirmed in learning experiments in monkeys. This compound had good oral bioavailability in rats and dogs, and in healthy young and elderly male volunteers. On this basis it was selected as a development compound for the treatment of cognition deficits.

The cognition-enhancing compound, (3- aminopropyl)-n-butylphosphinic acid, code-named CGP36742, has the following structure: The GABAB antagonist properties of CGP36742 do not satisfactorily explain its cognition-enhancing properties, since much more potent GABAB antagonists have been described that lack these properties.

We have now surprisingly found that CGP36742 has similar potency as a GABAC antagonist to its potency as a GABAB antagonist (50% inhibition of receptor activation being found at 38 uM and 62 uM against GABAB and GABAC receptors respectively). None of the other potent GABAB antagonists showed activity against GABAC receptors. These findings indicate a likely role for GABAC receptor antagonism in the cognition-enhancing properties of CGP36742.

SUMMARY OF THE INVENTION In one aspect the invention provides a method of enhancing the cognitive activity of an animal in need of such treatment, comprising the step of administering an effective amount of a compound which has GABAC receptor antagonist activity to said animal.

In a second aspect the invention provides a method of stimulating memory capacity, comprising the step of administering an effective amount of a compound which has GABAC receptor antagonist activity to an animal in need of such treatment.

The methods of the invention are suitable for the treatment of a variety of cognitive deficit conditions, dementias, and memory impairment conditions, including but not limited to those associated with Alzheimer's disease, AIDS, and schizophrenia.

Preferably the compound has selective antagonist activity against GABAC receptors compared with GABAB receptors. More preferably, the compound has selective antagonist activity against GABAC receptors compared with GABAA receptors. Even more preferably, the compound is substantially inactive against both GABAA and GABAB receptors.

More preferably the compound comprises a phosphinic acid group, and even more preferably comprises an alkyl-substituted phosphinic acid group in which the alkyl group is of 1 to 6 carbon atoms, such as a methyl or ethyl phosphinic acid group. Most preferably the compound also comprises a double bond which imposes a conformational restriction on rotation about the bond corresponding to the C3-C4 bond of GABA. Particularly preferred compounds include, but are not limited to, conformationally- restricted analogues of CGP44530 in which rotation about the C3-C4 bond is restricted, such as TPMPA and analogues thereof.

Thus preferred compounds of the invention are represented by general formula I or general formula II,

in which X represents hydrogen, an alkyl group optionally substituted with a halogen, or a hydroxyalkyl group, and Y represents hydrogen, a halogen, or an alkyl, alkenyl, alkynyl or acyl group, optionally substituted with halogen, nitrile, or NO2.

In general formula I, Y may also be an alkoxy group, optionally substituted with halogen, nitrile or NOs.

By"alkyl"is meant a straight or branched, saturated or unsaturated, substituted or unsubstituted alkyl chain of 1 to 6, preferably 1 to 4 carbon atoms, and includes alicyclic alkyl chains such as cyclopropylethyl. Alkenyl, alkynyl and acyl also refer to groups of 1-6, preferably 1- 4 carbon atoms. The halogen is preferably chlorine or fluorine.

It will be clearly understood that some of the compounds which are useful for the purposes of the invention are novel, and form part of the invention. Thus in a third aspect the invention provides a compound having GABAC antagonist activity and selectivity for the rho-2 subtype of GABA receptors of general formula II as defined above. Thus in a third aspect the invention provides a compound having GABAC antagonist activity and selectivity for the rho-2 subtype of GABAC receptors of general

formula II as defined above.

In a fourth aspect, the invention provides a composition comprising a compound of general formula II, together with a pharmaceutically-acceptable carrier.

While the invention is not in any way restricted to treatment of any particular animal species, in general the animal will be a human.

The compounds may be administered at any suitable dose and by any suitable route. Oral administration is preferred because of its greater convenience and acceptability. The effective dose will depend on the nature of the condition to be treated, and the age, weight and underlying state of health of the individual to be treated, and will be at the discretion of the attending physician or veterinarian. Suitable dosage levels may readily be determined by trial and error experimentation, using methods which are well known in the art. Similarly, suitable formulations for administration by any desired route may be prepared by standard methods, for example by reference to well-known texts such Remington: The Science and Practice of Pharmacy, Volume II, 1995 (luth edition), A. R. Gennaro (Ed), Mack Publishing Company, Easton, Pennsylvania 18042, USA., or Australian Prescription Products Guide, Volume 1,1995 (24th edition), J Thomas (Ed), Australian Pharmaceutical Publishing Company Limited, Victoria, Australia.

Throughout the description and claims of this specification, the word"comprise"and variations of the word, such as"comprising"and"comprises", means "including but not limited to"and is not intended to exclude other additives, components, integers or steps.

BRIEF DESCRIPTION OF THE FIGURES Figure 1 shows that expression of human ri receptors in Xenopus oocytes produces homooligomeric GABA receptors (GABAC receptors) with intrinsic Cl channels.

GABA (1 uM) activates the Cl channels (duration indicated

by filled bar) and produces an inward current when the oocyte is clamped at-60 mV. (A) CGP38593 (100 uM), (B) CGP44530 (100 uM), (C) CGP70523 (100 uM), (D) CGP36742 (100 uM), and (E) CGP70522 (300 uM) do not activate the receptor (duration indicated by hatched bar). However, when (A) CGP38593 (100 uM), (B) CGP44530 (100 uM), (C) CGP70523 (100 uM), (D) CGP36742 (100 uM), and (E) CGP70522 (3-00 uM) are co-applied with GABA (1 uM), the GABA response is blocked or reduced.

Figure 2 shows (A) Structures of compounds that show agonist activity at GABAC receptors. (B) Structures of compounds that show antagonist activity at GABAC receptors.

Figure 3 shows structures of orally active GABAB receptor antagonists with no cognitive enhancement effects.

These compounds show no affinity for GABAC receptors as either agonists or antagonists when tested at 100 uM.

Figure 4 summarizes the synthesis of PMPA by reduction fo a precursor of TMPA and subsequent hydrolysis.

Figure 5 shows the effect of TPMPA on memory formation in chicks.

Figure 6 shows the dose response relationship for the effects of TPMPA on discrimination ratio.

Figure 7 shows the effect of time after injection of TPMPA on memory formation in chicks.

Figure 8 shows the effect of TPMPA on memory for an elevated plus maze in male Swiss mice.

DETAILED DESCRIPTION OF THE INVENTION The invention is described in detail by way of reference only to the following non-limiting general methods and experimental examples, and to the figures.

Materials [(E)-3-Aminopropen-1-yl] methylphosphinic acid (CGP44530), [(E)-3-aminopropen-1-yl] phosphinic acid (CGP38593), [ (Z)-3-aminopropen-1-yl] methylphosphinic acid (CGP70523), [ (Z)-3-aminopropen-1-yl] phosphinic acid (CGP70522),

3-aminopropyl-n-butyl-phosphinic acid (CGP36742), 3-aminopropyl (diethoxymethyl) phosphinic acid (CGP35348), 3-aminopropyl (cyclohexylmethyl) phosphinic acid (CGP46381), (2S)-3-amino-2-hydroxypropyl (cyclohexylmethyl) phosphinic acid (CGP51176) and (2R, 1'S)- (3-N- [1' (3,4-dichlorophenyl) ethyl]) amino-2-hydrox ypropyl) benzylphosphinic acid (CGP55845A) were synthesised as described previously by Froestl et al, (1992; 1995a; 1995b). CACA and TACA were prepared as previously described (Johnston et al, 1975). GABA was purchased from Sigma Chemical Co (St Louis, MO, USA).

Electrophysiological Recording Human r1 cDNA in pcDNA (Invitrogen, San Diego, CA, USA) was obtained from Dr. George Uhl (National Institute for Drug Abuse, Baltimore, USA). The plasmid was linearized with XbaI and cRNA made using the"Message Machine"kit from Ambion Inc. (Austin, Texas, USA). 50 ng of cRNA was injected into defolliculated Stage V Xenopus oocytes. Two to seven days later, receptor activity was measured by two-electrode voltage clamp recording, using a Geneclamp 500 amplifier (Axon Instruments Inc., Foster City, CA., USA) and a MacLab 2e recorder (ADInstruments, Sydney, NSW, Australia). Oocytes were voltage clamped at- 60 mV and continuously superfused with ND96 buffer (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2,1 mM MgCl2 and 5 mM HEPES, pH 7.5). For receptor activation measurements, the indicated concentrations of agonist and antagonist were added to ND96.

Analysis of Kinetic Data Current (I) as a function of agonist concentration ( [A]) was fitted by least squares to I=Imax [A]/(EC50 + [A]), where Imax is the maximal current, the EC5o is the effective dose that activates 50% of the maximal current and nH is the hill coefficient. EC50 values are expressed as meanS. E. M. (n=3-6) and were

determined by fitting data from individual oocytes using Kaleidagraph 2.1 (1990). Current (I) as a function of antagonist concentration ( [Ant]) was fitted by least squares to I=Imax-(Imax[Ant]/(IC5o+[Ant])},where the IC50 is the inhibition dose that blocks 50% of the current generated by 1 uM GABA and nH is the hill coefficient.

IC50 values are expressed as meanS. E. M. (n=3-6). KB values are the apparent dissociation constants for the antagonists, and were determined using Schild plot analysis (Arunlakshana and Schild, 1959).-logKg values were determined using the following equation: log { (A)/ (A*)- 1} =m. log [Ant]-logKB, where A is the EC50 of GABA in the presence of a known antagonist concentration, A* is the ECso of GABA in the absence of an antagonist, [Ant] is the concentration of the antagonist, and'm'is the slope of the curve. For simple competitive antagonism,'m'is 1.- logKB values were determined by fitting data to the above function using Kaleidagraph 2.1 (1990). Schild analyses were carried out for compounds that had IC50 values of less than 30 uM.

Example 1 GABAC Receptor Antagonists Block Activation of Chloride Channels by GABA Expression of human ri mRNA in Xenopus oocytes generated GABAC receptors which showed a dose-dependent GABA activated inward current when the cell was voltage clamped at-60 mV. This could be blocked by compounds such as CGP44530, CGP38593, CGP70523, CGP70522 and CGP36742, as shown in Figure 1. The structures of the compounds are shown in Figure 2 and Figure 3. These compounds were first screened at 100 uM to determine agonist activity, by activation of Cl channels, or antagonist activity, by blocking the activation of the channels by 1 uM GABA.

Figure 2 shows the active compounds that had some effect at 100 uM as agonists (Figure 2A) or antagonists (Figure 2B) at GABAC receptors, and Figure 3 shows the compounds that

had no effect at 100 uM as agonists or antagonists at GABAC receptors.

Only the carboxylic acids, TACA, GABA and CACA activated the C1 channel. TACA was more potent than GABA, with an EC50 of 0.44 0.02 uM, and was almost a full agonist, with a maximal TACA dose generating 95% of the maximal GABA activated current. GABA was found to have an EC5o value of 0.82 0.09 uM. CACA was less potent than GABA, with an EC5o value of 37.4 6.1 uM, and was a partial agonist, with a maximal CACA dose generating 75% of the maximal GABA activated current. These results are summarised in Table 1. The Hill Coefficients (nH) as shown in Tables 1 and 2 were greater or equal to 2, which suggests that more than one molecule of the agonist is required to bind before the Cl-channels can open. These findings are in agreement with those of Woodward et al (1993).

Table 1 Summary of EC50, IC50, KB and Hill Coefficients of various agonists and antagonists at the GABAC Receptor Expressed in Xenopus oocytes.

EC50 (, UM) a IC50 (, UM) b nHC KB (, UM) d GABA 2.60.2 TACA 2.40.2 CACA 2.20.3 Isoguvacinee 99 CGP35024 0.580.14 CGP44530 8.61.6 CGP70523 1.60.1 CGP27492 3.21.0 CGP38593 15.51.7 CGP70522 >100 CGP36742 3.00.4 TPMPA 2.1

a ECso is the effective dose that activates 50 % of the maximal current when tested at ri receptors expressed in Xenopus oocytes. b IC 50 is the concentration that inhibits 50% of the response produced by 1 uM GABA. Data are the meanS. E. M.

(n = 3-6 oocytes). c nH is the Hill Coefficient. d KB is the binding constant for the antagonist. These were determined using Schild plot analysis assuming competitive antagonism over the tested concentrations (Table 2).

Data from Murata et al, 1996.

Table 2 Results of Schild Analyses of CGP35024, CGP27492, CGP44530 and CGP38593 at the GABAC Receptor Expressed in Xenopus oocytes.

Antagonist [Antagonis ECso (, uM) nHb Slope of t] (uM) of GABA Schild PlotC CGP35024 3 1.14 10 2.20.2 30 1.90.2 CGP27492 10 0.99 30 2.40.3 100 2.50.2 CGP44530 10 1.01 30 3.00.1 100 3.70.4 CGP38593 30 0.95 60 2.50.4 100 6. 80.3 3.00.1

a ECso is the effective dose that activates 50 % of the maximal current when tested at r1 receptors expressed in Xenopus oocytes. ECso values are expressed as meanS. E. M.

(n=3-6) and are determined by fitting data from individual oocytes using Kaleidagraph 2.1 (1990). ECso values of GABA have shifted to the right in the presence of a known concentration of the antagonist.-log KB values were determined as described in Materials and Methods section.

The KB values are shown in Table 1. b nH is the Hill Coefficient. These are greater than 1 indicating that more than 1 molecule of GABA is required for the channel to open. c Slope of Schild plot analysis indicating competitive antagonism over the tested concentrations.

GP35024, CGP27492, CGP44530, CGP38593, CGP70523 and CGP70522 did not activate any current on their own (Figure 1). They acted as GABAC receptor antagonists, inhibiting the current activated by 1 uM GABA (Figure 1).

IC50 values were obtained for these compounds (Table 1) and Schild analyses were carried out for the active compounds (Table 2). KB (binding constant) values for CGP27492, CGP44530, CGP35024, and CGP38593 are shown in Table 1.

The methylphosphinic analogue, CGP44530, and phosphinic analogue, CGP38593 of TACA were antagonists with ICso values of uM and uM respectively.

These compounds had lower affinity for the GABAC receptor expressed in Xenopus oocytes than that of the corresponding methylphosphinic analogue, CGP35024, and phosphinic analogue, CGP27492, of GABA. CGP35024 had an ICso of uM and CGP27492 had an ICso of uM. The methylphosphinic analogue, CGP70523 and phosphinic analogue, CGP70522, of CACA were antagonists, with IC50 values of uM and >100 uM respectively. These

compounds had lower affinity for GABAC receptors than the methylphosphinic and phosphinic analogues of GABA and TACA.

The order of potency of the methyl phosphinic acids and phosphinic acids is CGP35024 > CGP27492 > CGP44530 > CGP38593 > CGP70523 >> CGP70522.

The new compounds CGP44530, CGP38593, CGP70523 and CGP70522 were weaker at the GABAC receptor than the existing phosphinic acid, CGP27492 and the methylphosphinic acid, CGP35024.

CGP35024, CGP27492, CGP44530 and CGP38593 were found to be competitive antagonists. The gradients of the Schild regression plots were not significantly different from 1 over the concentrations tested, indicating that these compounds compete for the same site as GABA.

CGP36742 was found to be an antagonist with moderate potency at the GABAC receptor, with an IC5p value of uM. This compound is orally active, showing cognitive enhancement effects. Other related compounds, such as CGP35348, CGP46381, CGP51176 and CGP55845A (Figure 3), are also orally active, but do not show cognitive enhancement effects. These were screened at 100 uM, and had no effect as either agonists or antagonists at GABAC receptors. These compounds show high selectivity as GABAB receptor antagonists.

Example 2 Relative Effects of Compounds on GABAA, GABAB and GABAC Receptors The development of many alkylphosphinic and phosphinic analogues of GABA has yielded novel GABAB receptor agonists and antagonists (Olpe et al, 1990; 1993; Bittiger et al, 1992; 1993; Froestl et al, 1992; 1995a; 1995b), including the methylphosphinic and phosphinic analogues of TACA and CACA, ie. CGP44530, CGP38593, CGP70522 and CGP70523. In this study, we tested these compounds on GABAC receptors expressed in Xenopus oocytes, and found them to be competitive antagonists. The antagonist potencies of CGP44530, CGP38593, CGP70522 and

CGP70523 were found to be lower than that of the methylphosphinic and phosphinic analogues of GABA, CGP35024 and CGP27492.

The relative effects of the compounds at GABAA, GABAB and GABAC receptors are shown in Table 3. Three compounds, CGP38593, CGP70522 and CGP27492, were moderately potent at GABAA receptors when tested using radioligand binding assays (IC5p = 6.8 uM; ICso = 6.6 uM and ICso = 1. 7 uM, respectively) (Froestl et al, 1995a). However, the compounds were more potent at GABAB receptors than at GABAA receptors using this assay. Similarly, these compounds appear more potent at GABAB receptors than at GABAC receptors.

Table 3 Affinities of the compounds used in this study at GABA receptors.

Receptor Affinitya Compound GABAA (, uM) b GABAB (uM) C GABAC (uM) d GABA 0.128k 0.033 ECso = 0.82e CGP27492 1.7k 0.005 2.47 CGP35024 inactive at 10k 0.016 0.75 CGP36742 508 38 62 TACA 0.14f, k inactive at EC50 = 0.44e 1009 CGP38593 6.8 0.28 7.68 CGP44530 inactive at 100 0.65 5.53 CACA 25f, inactive at EC50 = 37e 1009 CGP70522 6.6 4.4 > 300 CGP70523 242 16 38 Isoguvacine 1.4ft k inactive at EC50 = ggi 500h TPMPA Kb = 320i ECso-500 Kb = 2. li PMPA > 100 > 10001 6.0

a Receptor affinities are ICSo values unless otherwise stated. b ICso values ie. concentration that inhibits 50% of the total [3H] muscimol binding using rat cortical membranes (Froestl et al, 1995a; 1995b). c ICso values for the inhibition of [3H] CGP27492 binding using rat cortical membranes (Froestl et al, 1995a; 1995b). d ICso values for the inhibition of the response of 1 uM GABA using human r1 mRNA expressed in Xenopus oocytes as described in the Materials and Methods section. s EC5Q values ie. the effective dose that activates 50 % of the maximal current when tested at ri receptors expressed in Xenopus oocytes as described herein. f ICso values for the inhibition of the total Na- independent [3H] GABA binding using rat brain membranes (Johnston et al, 1978).

9 Data from Kerr and Ong (1995) using guinea pig ileum, in the presence of bicuculline, against baclofen-depression of twitch contractions. h Data from Ragozzino et al (1996) using whole cell patch recordings from pyrimidal neurons in hippocampal slices in the presence of bicuculline (20 uM).

Data from Murata et al (1996) using human r1 mRNA expressed in Xenopus oocytes.

3 Data from Ragozzino et al (1996) using poly (A) + RNA from rat cortex expressed in Xenopus oocytes. k ECso values for GABA, CGP27492, CGP35024, TACA, CACA and isoguvacine using poly (A) + RNA from rat cortex expressed in Xenopus oocytes are 107 uM, 938 uM, inactive at 1 mM, 133 uM, inactive at 5 mM, and 305 uM respectively (Woodward et al, 1993). These values are different to the values obtained from radio-ligand binding assays.

Data from measurement of the frequency of spontaneous discharges in rate neocortical slices using the grease-gap recording system.

Example 3 Specificity of GABAC Antagonists for GABAC Receptor Subtypes We have demonstrated that the benchmark GABAC antagonist, TPMPA, is an order of magnitude less potent at blocking human homo-oligomeric rho-2 receptors than rho-1 GABAC receptors.

Of particular interest is the dihydro derivative of TPMPA, piperidine-4- (methyl) phosphinic acid (PMPA): PMPA (Piperid-4-yl) methylphosphinate (PMPA) was synthesized by reduction of a precursor of TMPA and subsequent hydrolysis, as follows: Platinum oxide (PtO2. H20) (50 mg) was added to a solution of recrystallised Troc-precursor (isopropyl [1- 5,6-tetrahydropyridin- 4-yl] methylphosphinate, A) (1.50 g, 3.96 mmol) in methanol (25 mL) and the mixture was shaken with H2 (5 atm.) at room temperature for 24 h. The catalyst was filtered off through Celite, and the residue concentrated under reduced pressure to afford a viscous colourless oil. A n. m. r. examination of the crude reduction product indicated complete reduction of the olefinic bond together with significant concomitant reduction and partial deprotection of the Troc group.

A mixture of the residue from above, 48% aq. HBr (40 mL) and glacial acetic acid (40 mL) was refluxed for 60 h. The reaction mixture was concentrated under reduced pressure (water pump) and the final traces of HBr/AcOH were removed by the sequential addition of H2O and concentration

(several cycles). The final residue (HBr salt) was dissolved in a small volume of H20 and applied to a Dowex AG 50 (H+) column. After initial elution with H20 until the eluant was neutral, the eluting agent was changed to 1M aq. pyridine. Ninhydrin-positive fractions were combined and concentrated under reduced pressure (water pump).

Final traces of pyridine were removed by the sequential addition of H20 and concentration under reduced pressure (several cycles) to afford a quantitative yield of crude (piperid-4-yl) methylphosphinic acid (PMPA) (C) as an off- white solid (ca. 645 mg, air dried) which was recrystallised from EtOH/H2O (450 mg, 70%): m. p. 289-291°; H NMR (300 MHz, D20, Ref: DOH = 8 4.8) 8 1.19 (3H, d, J = 13.2 Hz, PCH3), 1.56-1.82 (3H, 2 x overlapping m, 2 x NCH2CHB and PCH), 2.00-2.08 (2H, m, 2 X NCH2CHA), 2.96 (2H, (apparent?) dt, J = 3.0,12.8 Hz, 2 x NCHBCH2), 3.43-3.51 (2H, m, 2 X NCHACH2); 13C NMR (D2O, 75.64 MHz, Ref: (internal) dioxane = 8 67.4) 8 13.6 (d, J = 91.5 Hz), 23.1, 36.0 (d, J = 96 Hz), 44.8 (d, J = 14.2 Hz). The chemical synthesis is shown in Figure 5.

We have found that PMPA is a much more potent antagonist than TPMPA against rho-2 receptors, and less potent than TPMPA against rho-1 receptors, as indicated in Table 4.

Table 4 Binding Affinity for rho-1 and rho-2 Receptors KB (RM) human rho-1 receptor human rho-2 receptors TPMPA 2.0 0.4 15.6 1.6 PMPA 6.0 1.2 4.2 0.2 PMPA and TPMPA show similar weak activity against GABAA and GABAB receptors.

The finding that TPMPA and PMPA show differing selectivity between rho-1 and rho-2 subtypes of GABAC receptors was quite unexpected.

Although the possibility that PMPA might have activity as a competitive antagonist of GABAC receptors is mentioned in U. S. Patent No. 5,627,169:"Selective Antagonists for GABArho Receptor"and in a paper by Woodward et al (1993), it appears that neither this compound nor its analogues was actually synthesised and tested. Consequently these prior disclosures are merely speculative paper examples.

CGP36742 was shown to be a moderately potent antagonist at GABAB receptors using a [3H]-CGP27492 binding assay (ICso = 35 uM) (Bittiger et al, 1992; Olpe et al, 1993; Froestl et al, 1995a). It had weak effects at GABAA receptors (ICso = 500 uM) (Bittiger et al, 1992), and had no effect at other receptor types, including NMDA, benzodiazepine, quisqualate, kainate, muscarinic cholinergic, adrenergic, serotoninergic and histaminergic receptors (1 mM) (Bittiger et al, 1992; Froestl et al, 1995b). However, we have now found that CGP36742 showed moderate antagonist activity at GABAC receptors (ICso = 62 uM), and that its apparent selectivity for GABAB compared to GABAC receptors was approximately 2-fold. This compound has shown promising therapeutic potential in the treatment of cognitive deficits, petit mal epilepsy and depression (Bittiger et al, 1992). Therefore it is possible that antagonism of GABAC receptors contributes to the cognitive enhancement potentiation by CGP36742, such enhancement is not shown by other orally-active GABAB receptor antagonists (Froestl et al, 1995b).

TPMPA was recently synthesised and tested at GABAA, GABAB, and GABAC receptors (Murata et al, 1996).

It is a conformationally-restricted analogue of CGP44530, and is the methylphosphinic analogue of isoguvacine. It was found to be more than 100-fold more selective as an antagonist for GABAC receptors than for GABAB receptors, and is 500-fold more selective at GABAC receptors than at GABAA receptors (Murata et al, 1996; Ragozzino et al, 1996).

Example 4 Effect of TPMPA on Memory The effects of TPMPA in memory consolidation in chicks This procedure trains each chick on anthranilate-coated red beads, which have a bitter taste. In the test, 120 min. after the initial exposure to the red bead each chick is presented with a blue and a red bead, and normally will avoid pecking at the red bead; the discrimination ratio measures how well it remembers to do this. Chicks trained on 100% anthranilate-coated beads produce a discrimination ratio better than 0.9, and drug-induced memory deficits can be detected in this group. However, chicks trained on 20% anthranilate-coated beads produce a discrimination ratio of around 0.6, and this group can be used to detect drug-induced memory enhancement. Drugs are delivered by two bilateral intracranial injections (10pL each).

Figure 6 shows that TPMPA at a dose of 30 RM enhances memory with the group trained with 20% anthranilate performing as well as the 100% anthranilate group. Figure 7 shows the dose-response relationship for the effects of TPMPA on discrimination ratio, with an ECso between 1 and 10 RM. Figure 8 shows the dependence of this effect on time of injection, with an optimum effect produced by injecting in the 2.5 minutes after training.

The effects of TPMPA on the plus-maze memory test In this assay mice are trained by placing them at the end of the open arm of the plus maze and allowing them to find the shelter of the closed arms. The time taken is measured as the latency'. Immediately after the trial the mice are injected with the test drug or with a saline control. Two and ten days later the test is repeated. An agent which enhances memory consolidation will significantly reduce the latency time relative to the

saline controls. The results are summarized in Figure 9, and show that TPMPA at 200 mg/Kg, but not at 50 mg/Kg, significantly reduces the latency in the 14 day test. This is consistent with enhancement of memory consolidation.

When we repeated the experiment, this time testing only after 14 days and using Swiss mice from a different source, we found that TPMPA at 50 mg/Kg but not 200 mg/Kg significantly reduced latency. The experiments overall are therefore positive but inconclusive, since the different origin of the mice may be a contributing factor.

It will be apparent to the person skilled in the art that while the invention has been described in some detail for the purposes of clarity and understanding, various modifications and alterations to the embodiments and methods described herein may be made without departing from the scope of the inventive concept disclosed in this specification.

References cited herein are listed on the following pages, and are incorporated herein by this reference.

REFERENCES Arunlakshana, O. and Schild, H. O.

Br. J. Pharmacol., 1959 14 48-58 Bittiger, H., Bernasconi, R., Froestl, W., Hall, R., Jaekel, J., Klebs, K., Krueger, L., Mickel, S. J., Mondadori, C., Olpe, H. R., Pfannkuch, F., Pozza, M., Probst, A., Van Riezen, H., Schmutz, M., Schuetz, H., Steinmann, M. W., Vassout, A., Waldmeier, P.

Pharmacol. Com., 1992 2 70-74 Bittiger, H., Froestl, R. W., Mickel, S. J., Olpe, H. R.,

1993 Trends Pharmacol. Sci., 1993 14 391-393 Bormann, J. and Feigenspan, A.

Trends Neurosci., 1995 18 515-519 Cutting, G. R., Lu, L., O'Hara, B., Kasch, L. M., Donovan, D., Shimada, S., Antonarakis, S. E., Guggino, W. B.

Uhl, G. R. and Kazazian H. H.

Proc. Natl. Acad. Sci. USA., 1991 88 2673-2677 Cutting, G. R., Curristin, S., Zoghbi, H., O'Hara, B., Seldin, M. F. and Uhl G. R.

Genomics, 1992 12 801-806 Dong ; C-J., Picaud, S. A. and Weblin, F. S.

J. Neurosci., 1994 14 2648-2658 Enz, R., Brandstatter, J. H., Hartveit, E., Wassle, H. and Bormann, J.

Eur. J. Neurosci., 1995 7 1495-1501 Froestl, W., Mickel, S. J., von Sprecher, G., Bittiger, H. and Olpe H-R.

Pharmacol. Com., 1992 2 52-56 Froestl, W., Mickel, S. J., Hall, R. G., von Sprecher, G., Strub, D., Baumann, P. A., Brugger, F., Gentsch, C., Jaekel, J., Olpe, H-R., Rihs, G., Vassout, A., Waldmeier, P. C. and Bittiger H.

J. Med. Chem., 1995a 38 3297-3312 Froestl, W., Mickel, S. J., von Sprecher, G., Diel, P. J., Hall, R. G., Maier, L., Strub, D., Melillo, V., Baumann, P. A., Bernasconi, R., Gentsch, C., Hauser, K., Jaekel, J., Karlsson, G., Klebs, K., Maitre, L., Marescaux, C., Pozza, M. F., Schmutz, M., Steinmann, M. W.,

van Riezen, H., Vassout, A., Monadori, C., Olpe, H-R., Waldmeier, P. C. and Bittiger H.

J. Med. Chem., 1995b 38 3313-3331 Johnston, G. A. R., Curtis, D. R., Beart, P. M., Game, C. J. A., McCulloch, R. M. and Twitchin, B.

J. Neurochem., 1975 24 157-160 Johnston, G. A. R., Allan, R. D., Kennedy, S. M. E. and Twitchin, B.

1978"GABA-Neurotransmitters", Alfred Benzon Symposium 12, Munksgaard, p 149-164.

Johnston, G. A. R.

Pharmacol. Ther., 1996a 69 173-198 Johnston, G. A. R.

Trends Pharmacol. Sci., 1996b 17 319-323 Kerr, D. I. B. and Ong, J.

Pharmacol. Ther., 1995 67 187-246 Kusama, T., Spivak, C. E., Whiting, P., Dawson, V. L., Schaeffer, J. C. and Uhl G. R.

Br. J. Pharmacol., 1993a 109 200-206 Kusama, T., Wang, T-L., Guggino, W. B., Cutting, G. R., Uhl, G. R.

1993b Eur. J. Pharmacol.-Mol. Pharmacol. Sect., 1993b 245 83-84 Lukasiewicz, P. D., Maple, B. R. and Weblin, F. S.

J. Neurosci., 1994 14 1202-1212 Murata, Y., Woodward, R. M., Miledi, R. and Overman, L. E.

Bioorg. and Med. Chem. Lett., 1996 6 2071-2076

Olpe, H-R., Karlsson, G., Pozza, M. F., Brugger, F., Steinmann, M. W., Van Riezen, H., Fagg, G., Hall, R. G., Froestl, W. and Bittiger, H.

Eur. J. Pharmacol., 1990 187 27-38 Olpe, H-R., Steinmann, M. W., Ferrat, T., Pozza, M. F., Greiner, K., Brugger, F., Froestl, W., Mickel, S. J. and Bittiger, H.

Eur. J. Pharmacol., 1993 233 179-186 Ogurusu, T. and Shingai, R.

Biochimica et Biophysica Acta., 1996 1305 15-18 Polenzani, L., Woodward, R. M., Miledi, R.

Proc. Natl. Acad. Sci. USA, 1991 88 4318-4322 Qian, H. and Dowling, J. E.

Nature, 1993 361 162-164 Ragozzino, D., Woodward, R. M., Murata, F., Eusebi, F., Overman, L. E. and Miledi, R.

Mol. Pharmacol., 1996 50 1024-1030 Shimada, S., Cutting, G. and Uhl, G. R.

Mol. Pharmacol., 1992 41 683-687 Wang, T-L., Guggino, W. B. and Cutting, G. R.

J. Neurosci., 1994 14 6524-6531 Woodward, R. M., Polenzani, L. and Miledi, R.

Mol. Pharmacol., 1993 43 609-625