UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Unsaturated analogues of valproic acid : structure activity relationships and interaction with gaba metabolism Palaty, Jan 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1995-983361.pdf [ 3.7MB ]
JSON: 831-1.0088895.json
JSON-LD: 831-1.0088895-ld.json
RDF/XML (Pretty): 831-1.0088895-rdf.xml
RDF/JSON: 831-1.0088895-rdf.json
Turtle: 831-1.0088895-turtle.txt
N-Triples: 831-1.0088895-rdf-ntriples.txt
Original Record: 831-1.0088895-source.json
Full Text

Full Text

UNSATURATED ANALOGUES OF VALPROIC ACID: STRUCTURE ACTIVITYRELATIONSHIPS AND INTERACTION WITH GABA METABOLISMbyJAN PALATYB.Sc. (Chemistry), University of British Columbia, 1987M.Sc. (Chemistry), University of British Columbia, 1990A THESIS SUBMITTED iN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESFACULTY OF PHARMACEUTICAL SCIENCESDivision of Pharmaceutical ChemistryWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril, 1995© Jan Palaty, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.V (Signature)________________Department of /4/CE JC /EA’CESThe University of British ColumbiaVancouver, CanadaDate /,_DE.6 (2/88)iiABSTRACTVaiproic acid (VPA) is a versatile clinical antiepileptic drug which is alsocharacterized by rare but potentially fatal side effects such as hepatotoxicity andteratogenicity. Its principal metabolite, the aji-unsaturated acid 2-ene VPA, appears to sharemost of VPA’s therapeutic properties while lacking its toxicity and is thus a useful leadcompound for the development of safer antiepileptic drugs. The objectives of this thesis wereto shed some light on the anticonvulsant properties of 2-ene VPA analogues by aninvestigation of their influences on GABA metabolism and membrane fluidity.A group of x,3-unsaturated acids were synthesized by established methods or minormodifications thereof. The compounds were then evaluated for anticonvulsant activity inmice using the subcutaneous pentylenetetrazole test. Cyclooctylideneacetic acid (compound17) exhibited a potency markedly exceeding that of VPA itself with no more than modestlevels of sedation. Potency, as log(ED5O), was highly correlated with both volume andlipophilicity rather than with one of the shape parameters calculated by molecular modellingtechniques, arguing against the existence of a specific receptor site. These relationshipsremained essentially intact when ED5O was replaced with the brain concentration of the drug15 mm following an ED5O dose.Subsequent studies focused on the properties of nerve terminals from whole brainhomogenates prepared from mice administered an ED5O dose of each drug. GABA levelswere generally found to be elevated, supporting the central role of this neurotransmitter in theanticonvulsant properties of VPA and its analogues. Selectivity for regional or functionalpools of GABA was suggested as a cause for the variability. The activity of GABA’ssynthesizing enzyme glutamate decarboxylase was mostly unchanged but some drugs, notablycompound 17, showed a significant decrease in activity compared to the control. Theseinfluences on GAD activity were unrelated to the extent of binding of the enzyme’s co-factor,pyridoxal 5’-phosphate, following in vivo administration of VPA, 2-ene VPA and compoundiii17. Finally, the activity of the GAD was found to be inhibited in a non-competitive mannerby compound 17 with K =9 mM.ivTABLE OF CONTENTSAbstract 11Table of contents ivList of tables viiList of figures ixList of abbreviations xiiAcknowledgements xvii1. Introduction 11.1. Epilepsy 11.1.1. Clinical and therapeutic aspects 11.1.2. Animal models 21.2. Valproic acid 41.2.1. Clinical use and toxicology 41.2.2. Pharmacodynamics VPA and GABA The GABA hypothesis Glutamate decarboxylase VPA and post-synaptic effects Influence of VPA on membrane ion channels VPA and membrane fluidity 221.2.3. Structure-activity studies 251.3. Objectives 292. Experimental 312.1. Materials 312.2. Synthesis 322.2.1. (±)-4-( 1 -Methylethyl)-2-oxazolidinone (la) 322.2.2. Cyclopentylacetic acid 332.2.3. (±)-3-( 1 -Oxoalkyl)-4-( 1 -methylethyl)-2-oxazolidinone(2a/a-f) and (±)-3-( 1 -oxoalkyl)-2-oxazolidinone (2b/e-f) 33V2.2.4. Aldol addition: (±)-(eiythro)-3-[3-hydroxy-2-alkyl- 1-oxopentyl]-4-( 1 -methylethyl)-2-oxazolidinone (3a/a-f) and(±)-(erythro)-3-[3-hydroxy-2-alkyl- 1 -oxopentyl]-2-oxazo-lidinone (3b/e-f) 372.2.5. Methyl (±)-(E)-2-alkyl-2-pentenoate (5a-e) and (±)-(erythro)methyl 2-cyclopentyl-3-hydroxypentenoate (40 402.2.6. (E)-2-Alkyl-2-pentenoic acid (5-8) 432.2.7. 4(R)-Methoxycarbonyl)- 1 ,3-thiazolidine-2-thione (9x) 452.2.8. N-(3-Methylbutanoyl)-4(R)-(methoxycarbonyl)-1,3-thiazolidinethione (lOx) 462.2.9. (Erythro)-N-[3-hydroxy-2-( 1 -methylethyl)- 1 -oxopentylj-4(R)-(methoxycarbonyl)- 1 ,3-thiazolidine-2-thione (lix) 472.2.10. N-(2-(Methoxypropenoyl)-2-( 1 -methylethyl))-2-pentenamide (12x) 482.2.11. Attempted preparation of methyl (Z)-2-methyl-2-pentenoate (lSx) 482.2.12. 1-Cyclopentenyl-1-carboxylic acid (12) 492.2.13. 1-Cyclohexenyl-1-carboxylic acid (13) 512.2.14. (±)-Methyl bicyclo[2.2. 1 Jhept-3-oxo-2-carboxylate (17x) 512.2.15. (±) Bicyclo{2.2. ljhept-2-ene-2-carboxylic acid (ii) 522.2.16. Cyclohexylideneacetic acid (15) 532.2.17. Cyclopentylideneacetic acid (14) 532.2.18. Cycloheptylideneacetic acid (16) 542.2.19. Cyclooctylideneacetic acid (17) 552.2.20. Attempted preparation of 2-cyclooctylidene-propanoic acid (23x) 552.3. Physicochemical properties 572.4. Anticonvulsant potency 572.5. Mouse brain drug determination 582.6. Membrane fluidity 592.7. GABA and GAD assays 602.8. GAD inhibition assay 622.9. GAD saturation assay 62vi3. Results 643.1. Synthesis 643.2. Anticonvulsant activity evaluation 733.3. Brain concentrations at ED5O dose 823.4. Molecular modelling 883.5. Lipophilicity 953.6. Quantitative structure-activity relationships 1003.7. Membrane fluidity 1083.8. Synaptosome preparation 1133.9. GABA and GAD assays 1163.9.1. General 1163.9.2. Assay characteristics 1163.9.3. Application to synaptosomes 1183.9.4. Demonstration of linearity and saturation conditions 1203.9.5. Results from drug-treated mice 1233.9.6. Inhibition of GAD by acid 17 1263.10. GAD saturation 1284. Discussion 1324.1. Synthesis 1324.2. Anticonvulsant evaluation and correlation with physicochemicalproperties 1334.3. Effect of 2-ene VPA analogues on GABA levels 1414.4. Interaction of 2-ene VPA analogues with glutamate decarboxylase 1505. Conclusions 1576. References 159viiLIST OF TABLESTable 1. Properties, occurrence and therapy of the most common seizure types 1Table 2. Potencies of clinical anticonvulsants in rats using common screeningmethods 3Table 3. Anticonvulsant properties of VPA and its analogues 26Table 4. Anticonvulsant and lipophilicity properties of some 2-substitutedpentanoic acids (Elmazar et at., 1993) 28Table 5. Anticonvulsant and sedative properties and brain concentrations ofVPA and its analogues 82Table 6. Brain concentration assay parameters 85Table 7. Drug recovery as a function of amount per vial 86Table 8. Brain/body partition coefficient Q 88Table 9. Molecular modelling of VPA analogues as carboxylates inan aqueous medium: population-weighted mean dimensions and number ofconformers within 10 kJ of global minimum 92Table 10. Molecular modelling of VPA analogues as protonated acidsin an aqueous medium: number of conformers within 10 U of globalminimum and population-weighted mean Y 93Table 11. Molecular modelling of selected VPA analogues as Z isomers 93Table 12. Determination of lipophilicity (log P) by HPLC and CLOGP 98Table 13. Correlation matrix of biological and physicochemical propertiesof VPA analogues 103Table 14. The effect of VPA analogues (20 mM) on membrane fluidity ofhuman erythrocyte ghosts as measured by fluorescence anisotropy(r,x 10) 113Table 15. Comparison of synaptosome parameters with results ofLoscheretal. (1981) 114Table 16. Analysis of GABA and GAD in mouse whole brain synaptosomes 119Table 17. Interaction with synaptosomal GABA metabolism after an ED5O dose 123Table 18. Effect of high VPA doses on synaptosomal GABA 124Table 19. Full correlation matrix of biological and physicochemical propertiesof VPA analogues 125Table 20. Effect of acid 17 on activity of GAD in vitro 126Table 21. Interference with holoenzyme activity in post-gel filtration GAD assay 130viiiTable 22. Effect of ED5O doses of VPA, 2-ene VPA and compound 17 onsaturation of synaptosomal GAD 15 mm postdose 131ixLIST OF FIGURESFigure 1. Some metabolic pathways of vaiproic acid (VPA). 5Figure 2. Spiro[4.6jundecane-2-carboxylic acid 8Figure 3. The metabolism of GABA (adapted from Meidrum, 1985) 10Figure 4. The regulation of glutamate decarboxylase (GAD). Modifiedfrom Martin and Rimvall, 1993. 17Figure 5. Anticonvulsant 2-substituted pentanoic acids (Elmazar et at., 1993). 28Figure 6. Synthesis of 2-substituted 2-pentenoic acids. 65Figure 7. 300 MHz1H-NMR spectrum of adduct 3be. 66Figure 8. Envisaged alternative pathway to final acid product. 68Figure 9. Thiazolidinethione-based approach to 2-substituted 2-pentenoic acids 68Figure 10. Attempted preparation of methyl (Z)-2-methyl-2-pentenoate. 70Figure 11. Preparation of acid 11. 70Figure 12. Preparation of cycloalkylideneacetic acids 14-17. 71Figure 13. Attempted preparation of 2-cyclooctylidenepropanoic acid 23x. 71Figure 14. Compounds selected for evaluation. 72Figure 15. Determination of ED5O dose for acid 2. 74Figure 16. Determination of ED5O dose for acid 3. 74Figure 17. Determination of ED5O dose for acid 4. 75Figure 18. Determination of ED5O dose for acid 5. 75Figure 19. Determination of ED5O dose for acid 6. 76Figure 20. Determination of ED5O dose for acid 7. 76Figure 21. Determination of ED5O dose for acid 8. 77Figure 22. Determination of ED5O dose for acid 9. 77Figure 23. Determination of ED5O dose for acid 10. 78Figure 24. Determination of ED5O dose for acid 11. 78Figure 25. Determination of ED5O dose for acid 12. 79Figure 26. Determination of ED5O dose for acid 13. 79Figure 27. Determination of ED5O dose for acid 14. 80xFigure 28. Determination of ED5O dose for acid 15. 80Figure 29. Determination of ED5O dose for acid 16. 81Figure 30. Determination of ED5O dose for acid 17. 81Figure 31. Quantitation and recovery determination of drugs in brain homogenate. 83Figure 32. Derivatization of acids with MTBSTFA. 84Figure 33. SJM chromatograms of MTBSTFA-treated 2-ene VPA (a) and internalstandards octanoic acid (b) and cyclohexylcarboxylic acid (c). 84Figure 34. STERIMOL scheme illustrated for (E)-2-pentenoate. 91Figure 35. Conformations of (E)- and (Z)-2-ene VPA carboxylates in an aqueousmedium. The structures at right are as seen looking along they axis on to the xz plane. 94Figure 36. Log P vs. log (capacity factor) for some saturated carboxylic acids. 99Figure 37. Log P comparison: CLOGP vs. HPLC. 99Figure 38. Log(ED5O) vs. volume for acids 2-17. VPA (not used in equation) isshown as an empty circle. 106Figure 39. Log(ED5O) vs. lipophilicity for acids 2-17. VPA (not used inequation) is shown as an empty circle. 106Figure 40. Log[drug]j11vs. volume for acids 2-17. VPA (not used inequation) is shown as an empty circle. 107Figure 41. Log[drug]j vs. lipophilicity for acids 2-17. VPA (notused in equation) is shown as an empty circle. 107Figure 42. Structures of membrane probes DPH (R = H) and TMA-DPH(R = -N(CH3)j 109Figure 43. Emission spectra of erythrocyte ghosts with and withoutincorporated fluorescent probe TMA-DPH. Numbers in parenthesesrefer to recorder potentiation. 111Figure 44. Emission spectrum of erythrocyte ghosts containing fluorescentprobe DPH. 111Figure 45. Fluorescence polarization of DPH in erythrocyte ghosts. 112Figure 46. Fluorescence polarization of TMA-DPH in erythrocyte ghosts. 112Figure 47. Preparation of synaptosomes for GABA and GAD assays. 115Figure 48. Electron impact mass spectrum of GABA disilylated with MTBSTFA. 117Figure 49. SIM chromatograms for GABA-d6and GABA from synaptosomeGABA assay. 117Figure 50. Effect of glutamate concentration on GAD activity. 121xiFigure 51. Effect of PLP concentration on GAD activity. 121Figure 52. Time-dependence of GAD activity. 122Figure 53. Effect of protein concentration on GAD activity. 122Figure 54. In vitro inhibition of synaptosomal GAD by 17. Error barsindicate SD for one sample evaluated in triplicate. 127Figure 55. a) GAD saturation assay: approximate relative GAD activityvs. column fraction #; b) GAD saturation assay: proteinconcentration vs. column fraction #. 129xiiLIST OF ABBREVIATIONSAb where A = X, Y, Z and b = -, +: shape descriptorsA AngstromsAnal. elemental analysisapo-GAD apoenzyme form of GADaq. aqueousATP adenosine triphosphatebp boiling pointBu butyldegrees Celsiuscaic. calculatedCBZ carbamazepineCLZ clonazepamcm1 wavenumberCNS central nervous systemCSF cerebrospinal fluidd doubleto chemical shift (NMR)A heat or changeDBU 1 ,8-diazabicyclo[5 .4.O]undec-7-eneDPH diphenylhexatriene[drug]j concentration of drug in brain at 15 mm following an ED5O doseED5O dose producing given effect in 50% of animalsEDTA ethylenediaminetetraacetic acidEEG electroencephalogram2-ene VPA (E)-2-propyl-2-pentenoic acideq. equivalentsxiiiEt ethyletal. etaliaETH ethosuximideg gram(s)GABA ‘y-aminobutyric acidGABA-T GABA transaminaseGAD glutamate decarboxylaseGC-MS gas chromatography-mass spectrometryGHB y-hydroxybutyrateh hour(s)HMPA hexamethyiphosphoramide1H-NMR proton nuclear magnetic resonanceholo-GAD holoenzyme form of GADHPLC high pressure liquid chromatographyi.c.v. intracerebroventricularImIAc imidazole-acetate (buffer)i.p. intraperitoneali-Pr isopropylJR infraredintensity of fluorescence in plane of excitation‘vh intensity of fluorescence perpendicular to plane of excitationJ coupling constant°K degrees Kelvink’ capacity factor (chromatography)K1 inhibition constantkg kilogram(s)kJ kilojoulesKm Michaelis constantxivlog P logarithm of n-octanollwater partition coefficientm multipletM molarM+ molecular ionmA milliamperesMe methylMES maximal electroshockmg milligramMHz megahertzmm minute(s)rnL niillilitremlvi millimolarmm Hg millimetres mercury (pressure)mmol millimolesMsCI methanesulfonyl chlorideMSTFA N-trimethylsilyl-N-methyltrifluoroacetamideMTBSTFA N-tert-butyldimethylsilyl-N-methyltrifluoroacetamidemp melting pointmV millivoltsm/z mass-to-charge ration number of replicatesnmol nanomolesPB phenobarbitalPBS phosphate-buffered salinePHT phenytoinPLP pyridoxal 5-phosphatepmol picomolesppm parts per millionxvPTZ pentylenetetrazolePr n-propylq quartetQ brain-to-body partition coefficientQSAR quantitative structure-activity relationshipr fluorescence anisotropyr correlation coefficientR universal gas constantRPLC reverse phase liquid chromatography5 singlets standard error of the estimatescPTZ subcutaneous PTZSD standard deviationSIM selected ion monitoringSUCA spiro[4.6jundecane-2-carboxylic acidt tripletT temperature in Kelvintert-BDMS tert-butyldimethylsilylTHF tetrahydrofuranTHIP 4,5,6,7-tetrahydroisoxazolo[5,4-cjpyridin-3(2H)-oneTLC thin layer chromatographyTMA-DPH (trimethylamino)diphenylhexatrieneTMS trimethylsilyltr retention timeug microgram(s)Ui microlitreuM micromolarv solvent volumeV molecular volumeVPA vaiproic acidvs. versusw weightxvixviiACKNOWLEDGEMENTSI would like to express my gratitude to my supervisor, Dr. Frank Abbott, for hisguidance and support throughout these studies. Whether in the form of discussions of theproject or financial assistance to attend a conference, his help has been most invaluable. I amalso grateful to the other members of my committee, Professors Lawrence Weiler, StelvioBandiera, Peter Soja and Jack Diamond (chair), for their constructive suggestions. I wouldalso like to thank Drs. Weiler and Bandiera for the use of laboratory facilities.Thanks are also extended to Mr. John Jackson for assistance with the membranefluidity assay, to Ms. Slavica Beisher for help with the drug assays and to Dr. Eric Bigham atBurroughs Weilcome (North Carolina) who willingly performed the CLOGP calculations forsomeone he’d only met on the Internet. As with any project, one’s fellow lab inmates deserverecognition for many helpful comments and experimental assistance: in this case, my thanksgo to Drs. Anthony Borel and Wei Tang and Ms. Sashi Gopaul. I would especially like tothank Mr. Roland Burton for assistance with the GC-MS and computational work: his expertknowledge proved invaluable in response to my many frantic cries for help.I thank my wife Chrystal for tolerating my endless complaints about mice and thewashing of autosampler vials. Finally, I extend heartfelt gratitude to my parents for theirsupport throughout the years.Funding from the PMACIHRF in the form of a summer studentship is acknowledged.The work was supported by a grant from the Medical Research Councilof Canada.11. INTRODUCTION1.1 EPILEPSY1.1.1. CLINICAL AND THERAPEUTIC ASPECTSThe commonly used term “epilepsy” refers to “a group of CNS disorders having incommon the repeated occurrence of sudden and transitory episodes (seizures) of abnormalphenomena of motor (convulsion), sensory, autonomic or psychic origin” (Rail and Schleifer,1990). Epilepsy occurs in 0.5-1.0% of the population (Zielinsky, 1982), being particularlyprominent in children. These seizures can be classified into partial or generalized, dependingon whether or not the onset can be traced to a discrete region, as well as primary (oridiopathic, having no apparent cause) or secondary (symptomatic, apparently resulting from aspecific insult). Further descriptions are listed in Table 1.Table 1Properties, occurrence and therapy of the most common seizure typesType Symptoms and ictal Relative DrugsEEG activitya abundanceb of choicec(%)PartialSimple No loss of consciousness; 9.2 CBZ, PHT, (VPA), (PB)symptoms reflect affectedbrain region; usually nochange in EEGComplex Consciousness lost or 27.9 CBZ, PHT, (VPA), (PB)blunted; may becomegeneralized; automatismstypical; variable EEGGeneralizedPrimary Loss of consciousness; 28.1 VPA, PBtonic-clonic stiffening then jerkingof limbs; EEG: burst ofspikes followed by spikewave activitySecondary As above, but originates 30.8 CBZ, PHT, VPA, PBtonic-clonic in discrete region2Table 1 (continued)Properties, occurrence and therapy of the most common seizure typesType Symptoms and ictal Relative DrugsEEG activitya abundanceb of choicec(%)GeneralizedPrimary Blank stare ± brief 1.7 ETH, VPAabsence automatisms; characteristicgeneralized 3 Hz spike-wavedischarges in EEGMyoclonic Rapid jerks of face or 1.3 VPA, CLZextremities; multiplespikes or spike-waveEEG patternsa) Fukuzako and Izumi, 1991b) Keränen etal., 1988c) CBZ = carbamazepine; PHT = phenytoin; VPA valproic acid; PB = phenobarbital; ETH= ethosuximide; CLZ = clonazepam. Secondary agents are listed in parentheses. Adaptedfrom Fisher, 1991.As the table indicates, most seizures are caused by unusually high levels of brainactivity. For example, tonic-clonic seizures are controlled by drugs such as phenytoin andcarbamazepine that serve to limit the high-frequency neuronal discharges through voltage-dependent inhibition of sodium channels. Absence seizures, on the other hand, are effectivelycontrolled only by drugs such as ethosuximide, a blocker of the Ca2 T current. It isparticularly important to note that carbamazepine and phenytoin are as ineffective in dealingwith absence seizures as ethosuximide is with convulsive seizures. Nevertheless, valproicacid (VPA) has proven to be effective in both types of disorders, despite this apparentparadox. The means by which VPA exerts its antiepileptic action will be the subject of thisthesis.1.1.2. ANIMAL MODELSBefore delving deeper into the subject of antiepileptic drugs, it would be useful todiscuss some of the properties of actual human epilepsy and how these are modelled when3evaluating anticonvulsants. As mentioned, epilepsy refers to a tendency to suffer recurrentseizures, meaning that a seizure resulting from an acute situation such as an alcohol overdoseobviously does not represent epilepsy. Yet, the most common tests for antiepileptic activity,the subcutaneous pentylenetetrazole injection (scPTZ) and the maximal electroshock seizure(MES)(Swinyard et al., 1989), are both clearly acute seizure models. In the scPTZ test, arodent is administered 80-85 mg/kg PTZ into the loose skin at the back of the neck and thenobserved for tonic-clonic seizure activity. The MES test is similar except that electricalcurrent is administered via corneal or auricular electrodes. The scPTZ test acts primarily toscreen for anti-absence drugs because carbamazepine and phenytoin are ineffective here. Onthe other hand, this test also detects barbiturates that are ineffective for treating absenceseizures. The MES test, on the contrary, is quite specific for drugs targetted against tonic-clonic seizures.Table 2Potencies of clinical anticonvulsants in rats using common screening methods (mg/kg i.p.)Drug MESS scPTZb GHBC GEPR9d KindlingePhenytoin 29 inactive - 6 >100Carbamazepire 8 inactive 3 30VPA 490 180 + 150 120Ethosuximide >1200 54 + 230 inactiveClonazepam 186 0.06inactiveg 0082hDiazepam 15g 18ha) ED5O for abolition of tonic hindlimb extension upon administration of supramaximalelectroshock via corneal electrodes (50 mA: Swinyard et al., 1989)b) ED5O for abolition of tonic hindlimb extension upon administration of subcutaneouspentylenetetrazole (125 mg/kg: Swinyard et al., 1989)c) Ability to antagonize y-hydroxybutyrate-induced absence seizure (Snead and Bearden,1980; Vayeretal., 1987b)d) Dose for prevention of stimulated seizures (generalized tonic-clonic) in 50% of highlyseizure-prone rats (Dailey and Jobe, 1985)e) ED5O for prevention of stage 5 seizure following stimulus in rats kindled with cornealelectrodes (Edafiogho et al., 1992)f) Sato et al., 1990g) As in a), but 150 mA via auricular electrodes (Loscher and Nolting, 1991)h) As in b), but 90 mg/kg scPTZ (Loscher and Nolting, 1991)41.2. VALPROIC ACID1.2.1. CLINICAL USE AND TOXICOLOGYVPA, also known as 2-propylpentanoic acid, is a simple a-branched aliphaticcarboxylic acid whose anticonvulsant properties were discovered fortuitously in 1962 byMeunier et al.. Further work established its effectiveness in a wide range of seizure disorders(Davis et al., 1994) leading up to its approval for therapeutic use in the United States in 1978.Recently, the drug has also been used successfully for bipolar illnesses (Hopkins andGellenberg, 1994). VPA is well tolerated, only infrequently giving rise to gastrointestinalcomplaints and minor CNS effects such as sedation, ataxia and tremor (Rall and Schleifer,1990). However, the drug has been found to produce a potentially fatal hepatotoxicity,particularly in young children on polytherapy with other medical conditions (Kaneko et al.,1988; Omzigt et al., 1992). This toxicity has been linked to the metabolism of VPA to 4-eneVPA and 2,4-diene VPA (Figure 1) leading to hepatic steatosis presumably through covalentbinding to proteins (Prickett and Baillie, 1986; Porubek et al., 1989) and/or depletion ofglutathione via a Michael-type addition (Kassahun et al., 1991). Nevertheless, it should beemphasized that since the vulnerability of this group has been recognized, the occurrence ofVPA-associated fatal hepatotoxicity has declined to 1 in 50,000 (Dreifuss, 1989). Acutepancreatitis has also occurred occasionally, but although the symptoms are severe they arealso usually reversible upon withdrawal from VPA therapy. Finally, it has been shown thatchildren of epileptic mothers receiving VPA have a tendency to exhibit neural tube defectssuch as spina bifida although the occurrence is quite rare (Dreifuss, 1989). It has been shown(Nau et al., 1991; Collins et al., 1992) that unlike anticonvulsant activity, teratogenicity isstrongly dependent on the configuration of C(2) in asymmetric VPA analogues (Hauck et aL,1991), indicating a possible stereospecific drug-protein interaction.As a consequence of these potentially serious side effects of VPA, there has beensome interest recently in the use of 2-ene VPA, a major metabolite of VPA, as a replacementfor the parent drug (Loscher, 1992). This metabolite is equipotent with the parent drug butCO2HdieneVPAGlutathione conjugatesC02-GluVPA glucuronideCO2HVPA(7)5CO2FIVPA (1)CO2H4-ene VPA12,3-diene VPA (9/10)Figure 1. Some metabolic pathways of vaiproic acid (VPA).6does not exhibit apparent teratogenicity (Lewandowski et al., 1986; Vorhees et aL, 1991)The hepatotoxicity of this compound also appears to be negligible (Kesterson et al., 1984)despite its known conversion to the toxic 2,4-diene VPA (Kassahun and Baillie, 1993). Thisapparent contradiction is resolved by the fact that the formation of 2,4-diene VPA from 4-eneVPA occurs in the mitochondria while conversion from 2-ene VPA takes place in the cytosol.Consequently, 2-ene VPA shows promise as a less toxic successor to VPA.Unlike other antiepileptic drugs, VPA has an intrinsically low potency. Initial dailydoses are usually 15 mg/kg, increasing up to 60 mg/kg over several weeks (Rail and Schleifer,1990). The human pharmacokinetics of VPA are characterized by high bioavailability andplasma protein binding (90%) at therapeutic concentrations of 0.2 1-0.69 umol/mL (Levy andLai, 1982). Once inside the brain, the drug does not show any appreciable specificity asGoldberg and Todoroff (1980) have reported that VPA does not bind to rabbit brainhomogenates or subcellular fractions of brain tissue. In patients, the brain levels were 6.8-27.8% of total plasma levels (Vajda etal., 1981).Following i.p. injection in mice, peak drug brain concentrations were reached withinminutes (Hariton et al., 1984), with CSF concentrations being generally equal to the plasmaunbound fraction. The distribution of VPA and 2-ene VPA in the dog brain were studied indetail by Loscher and Nau (1983). They found 2-ene VPA to be the sole significantmetabolite of the parent following either acute or prolonged VPA administration.Interestingly, it can be detected in the brain longer than VPA and while an MESanticonvulsant effect is still present. This has also been observed in patients and monkeys(Lockard and Levy, 1976). With regard to distribution, these workers noted that VPA failedto show any significant accumulation in any given brain region 7 h after a 40 mg/kg bolus.When the drug was administered at 225 mg/kg/day via osmotic minipumps and the brainsanalyzed at 1, 3 and 14 days, VPA and 2-ene VPA showed some accumulation in thesubstantia nigra but VPA was otherwise quite evenly distributed. In addition, 2-ene VPAshowed elevated levels in the hippocampus, colliculi and medulla, which in some cases wereup to one quarter of the VPA concentrations. These findings are supported by7autoradiography studies in the rat with[14C]VPA that showed an even distribution throughoutthe brain, with some localization to the olfactory bulb (Schobben et al., 1980).The uptake of VPA into the brain has been assumed to occur via a saturablemechanism as the whole brain concentration was not found to increase proportionately withserum concentration 15 mm postdose (Pollack and Shen, 1985). The “monocarboxylic acid”carrier is probably not involved here as Cornford (1983) failed to observe saturation ofvalproate uptake at 100 ug/mL. Likewise, Frey and Loscher (1978) demonstrated theexistence of an active transport mechanism operating in the opposite direction across theblood-brain barrier because pretreatment with probenecid, which blocks carrier-mediatedtransport out of the brain, led to an increase in the CSF/serum concentration by 40% duringi.v. infusion in dogs. Following efflux of the drug from the brain, elimination from the bodyis by hepatic processes that account for >96% of the administered dose (Levy and Shen,1989).This carrier-dependent mechanism of blood-brain barrier crossing has been questionedrecently by Lucke et al. (1994) in light of data acquired using VPA-sensitive microelectrodessimultaneously sampling both blood and CSF in vivo. They found that the two VPAconcentrations were in an equal equilibrium by 5 minutes, suggesting that the drug is able torapidly and freely cross the barrier. It was argued that these results are in fact fully consistentwith those described above, as the latter were based on measurements of VPA in whole brainrather than the extracellular compartment where the bulk of the drug resides (Lucke et al.,1993). When this correction was made, a “good accord” was obtained with themicroelectrode data.An intriguing property of VPA that has defied explanation to date is why its potencyincreases with repeated dosing. For example, the protection of rats from amygdala-kindledseizures increased from 12% following a single dose at 200 mg/kg i.p. to 88% after sevensuch doses given three times per day (Loscher et al., 1988). In humans, this phenomenon ismanifested in the lack of a correlation between plasma VPA levels and seizure protection(Baruzzi et al., 1977; Guelen and van der Kleijn, 1978; Dickinson et al., 1979; Johannessen8and Henriksen, 1980; Nau et at., 1981; Nau and Loscher, 1982; Snead and Miles, 1985).Steady state conditions are reached within 2 days but the onset of a useful anticonvulsanteffect is several weeks. Similarly, there is a carryover effect lasting 2 weeks after cessation oftherapy (Schobben et at., 1980). The effect does not appear to be the result of anaccumulation of more potent metabolites, Loscher (1981a) having shown that none are morepotent than the parent, but bears an interesting resemblance to the drug’s previouslymentioned property of providing sustained seizure protection well after its plasma levels havefallen below the limit of detection and suggests an irreversible effect of some kind (Lockardand Levy, 1976).The idea that the action of VPA is dependent on the accumulation of activemetabolites was studied by Liu et al. (1990) using spiro[4.6Jundecane-2-carboxylic acid(SUCA, Figure 2), a potent (scPTZ ED5O = 0.42 mmol/kg in mice) anticonvulsant chemicallyrelated to VPA but lacking significant Phase I metabolism. Thus, the anticonvulsantproperties for this compound cannot be attributed to its metabolites. The study found thatSUCA, like VPA, failed to show a correlation between serum levels and the time-course ofthe anti-PTZ anticonvulsant effect. Therefore, it was concluded that VPA’s prolongedduration of action need not depend on the formation of active metabolites.yCO2HFigure 2. Spiro [4.6] undecane-2-carboxylic acidA final note of caution needs to be made with regard to the markedly differentpharmacokinetics of VPA in rodents compared to humans (reviewed by Loscher, 1985). Thisdifference is especially prominent in mice, where it takes the form of a short half-life (0.8 vs.9.3-18.4 h), reduced bioavailability (34-47 vs. 70-100%) and a lesser degree of proteinbinding (12 vs. 80-95%). Thus, there are limitations to using the mouse as an in vivo modelfor the behavior of antiepileptic drugs in humans.91.2.2. PHARMACODYNAMICS1.2.2.1. VPA AND GABA1. THE GABA HYPOTHESISy-Aminobutyric acid (GABA) is the principal inhibitory neurotransmitter in themammalian brain. It is a target for various anticonvulsants and sedatives-hypnotics, such asbenzodiazepines and barbiturates, most of which influence the GABA receptor rather thanGABA release. GABA’s life cycle is depicted in Figure 3. It is synthesized from glutamateby glutamate decarboxylase (GAD) as part of the “GABA shunt” from the tricarboxylic acid(TCA) cycle prior to release and uptake by neurons and glia. In neurons it then undergoestransamination by GABA transaminase (GABA-T) to succinic semialdehyde which is thenprimarily oxidized to succinate and leads back into the TCA cycle. A small fraction of thesemialdehyde is reduced to ‘y-hydroxybutyrate, a reaction that would be consideredinsignificant were it not for the fact that y-hydroxybutyrate is a known convulsant (Marcus etal., 1967) as well as being a neurotransmitter in its own right (Vayer et aL, 1987a). Thiscompound also has strong links to absence seizures as illustrated by the markedly enhancedbinding of[3H]’y-hydroxybutyrate in lateral thalamic nuclei of the genetic Wistar rat model ofabsence epilepsy, in comparison with control animals (Snead et al., 1990).The above depiction of GABA is somewhat complicated by the existence of at leasttwo distinct poois within the nerve terminal. This idea evolved from the experiments of Abeand Matsuda (1983) where [‘4C]glutamate and [3H]GABA were injectedintracerebroventricularly in mice and the relative content of[14C]GABA and[3HIGABA thenmonitored in synaptosomes on the assumption that the former represented newly synthesizedGABA and the latter newly taken up GABA, respectively. It was found that the GABA-Tinhibitor aminooxyacetic acid caused a three-fold increase in the levels of[3H]GABA with nochange in the[14C]GABA. Also, the release of[‘4CIGABA from synaptosomes was 50%higher than that of[3H]GABA. These results suggested that newly taken up GABA is simplydegraded by GABA-T and is not recycled significantly back into the pool responsible for the10TCA cycleGlutamatedecarboxylaseSuccinic semialdehydeSuccinicsemialdehydedehydrogenaseSuccinatecx-KetoglutarateGABAtransaminaseGlutamateAldehyde reductaseminoi.‘y-hydroxybutyrateTCA cyclea-Ketoglutaratedehydrogenasecomplexa-Ketoglutarate-O2CCO AspaateNH3 AOxaloacetateAspartatetransaminase-o2ccoGlutamateSuccinyl CoANH3-o2cGABAO2CHFigure 3. The metabolism of GABA (adapted from Meidrum, 1985).11GABAergic signal across the synapse. This scheme was further elaborated by Sihra andNicholls (1987) and Wood et at. (1988), who proposed the existence of an uptake pool thatequilibrates slowly with a neurotransmission pool of comparable size and, unlike the latter,releases GABA in aCa2+independent manner upon depolarization.Although the finding that VPA increases brain GABA levels has become generallyaccepted since it was first reported by Simler et at. (1968) and Godin et a!. (1969), it is stillunclear how this is accomplished and whether a simple increase in this inhibitoryneurotransmitter can account for the drug’s anticonvulsant properties. It has been found thatED5O doses of VPA and 2-ene VPA can significantly elevate GABA levels in whole brainsynaptosomes (Loscher et at., 1981), indicating an increase of this neurotransmitter in thenerve terminal compartment. Furthermore, VPA can also enhance the potassium-inducedrelease of GABA from both cortical slices (Ekwuru and Cunningham, 1990) and corticalneurons in culture (Gram et at., 1988). This was supported by the finding of Concas et at.(1991) that while the drug has no effect in vitro on the binding of[3H]GABA,[3H]flunitrazepam or[35S]TBPS (a competitive antagonist at the GABA recognition site),when administered intraperitoneally prior to the assay the binding of[35SJTBPS wasmarkedly inhibited, an effect greatly potentiated by the co-administration of very low doses ofdiazepam, a facilitator of GABA binding. As there was no effect on the binding affinity, thisresult was interpreted as an increase in GABA release.A temporal correlation between anticonvulsant effect and GABA levels wasdemonstrated in a more extensive study by Nau and Loscher (1982) who followed GABA,GAD and GABA-T in mouse brain homogenates, as well as electroconvulsive seizurethreshold, up to 16 h following a 200 mg/kg (1.4 mmollkg) i.p. dose of VPA. They observedparallel sharp rises in GABA, GAD and the seizure threshold that peaked at about 30 mmbefore commencing first a rapid decay ending at 2 h and then a much slower one over thecourse of the remainder of the experiment. GABA and GAD were significantly elevated upto at least 8 h and returned to control levels by 16 h. No change in GABA-T was notedthroughout the experiment. Similar results were obtained by Horton et al. (1977), Schechter12et at. (1978) and Chapman et at. (1984), also pointing to a distinct relationship between GAD,GABA and anticonvulsant activity.Further evidence for the central role of GABA in the actions of VPA was provided byKerwin et at. (1980) in a study that demonstrated altered GABAergic activity withoutmeasuring GABA itself. These workers showed that in rats the onset of full anticonvulsantefficacy (MES test) varied in a dose-dependent manner from 2 mm (400 mg/kg i.p. VPA) to10 mm (100 mg/kg VPA) but did not appear to coincide with a reduction in firing of GABAsensitive neurons in the pars reticulata of the substantia nigra, that appeared at a fairlyconstant latency of 4-6 minutes. Unfortunately, the number of nigral cells sampled neverexceeded five for any dose (100, 200 or 400 mg/kg) and none of these showed anunambiguous trend in activity. The situation for cortical cells was identical except for the 200mg/kg dose where the decrease in firing rate was unequivocal (n = 16, of which only 3showed no change) and coincided with the onset of seizure protection. Because these cellswere inhibited by microiontophoretic application of GABA, this positive temporal correlationprovides strong support for GABA as the prime effector of the anticonvulsant effect.The uptake of GABA was unaffected in rat brain homogenate, cortical brain slices, orsynaptosomes (Balcar and Mandel, 1976; Loscher, 1980; Ross and Craig, 1981; Slevin andFerrara, 1985). While Hackman et at. (1981) observed a nearly complete blockade of GABAuptake in frog lumbar spinal cord slices at 10 tM VPA, a similar effect was also observed forglycine uptake, casting doubts on the specificity of this inhibition as well as the validity ofthis model system. Inhibition of GABA uptake was also inferred by Klee et at. (1985) on thebasis that the chloride current elicited by microiontophoretically applied GABA in frog dorsalroot ganglion neurons was enhanced by 1-10 mM VPA much more effectively in 90 mMexternal sodium than in a sodium-free solution. Like other neurotransmitters, the uptake ofGABA is driven by the sodium potential gradient.The above observations are not consistent with a post-synaptic site of action resultingin feedback inhibition and an accumulation of GABA in the presynaptic nerve ending,implying that the GABA increase could only be effected by either an inhibition of GABA13degradation or an increase in GABA synthesis (Loscher, 1993 a). In contrast, both pre- andpost-synaptic mechanisms are apparently viable for 2-ene VPA because this drug exhibited a44 ± 6% displacement of[3HJGABA from its GABAA receptor at 1 mM in vitro, aconcentration at which VPA was ineffective (Nau and Loscher, 1984).The VPA/GABA hypothesis has been elaborated by the suggestion of ladarola andGale (1981), based on the abilities of drugs such as aminooxyacetic acid (AOAA) to greatlyincrease GABA levels without possessing significant anticonvulsant activity, that GABAexists in both metabolic and neurotransmitter pools with VPA acting specifically on the latter.This was demonstrated in an experiment where the substantia nigra of rats was transected onone side of the brain in order to destroy the nerve terminals and hence the neurotransmitterpooi of GABA. The substantia nigra on the transected side contained twice as much GABAfollowing a dose of AOAA than an equipotent one of VPA, where the level was comparableto that of the control. Thus, VPA was ineffective in enhancing GABA outside the nerveterminal. On the other hand, when the transected GABA levels were subtracted from the totalGARA levels of the intact side, the resultant (nerve terminal) concentration of GABAfollowing VPA was much higher than for AOAA, directly demonstrating the specificity ofVPA for the nerve terminal pool. This experimental model was further used to show a linearrelationship between the GABA level in the nerve terminal compartment and the degree ofprotection against MES seizures, indicating that this GABA pool is in fact the one involved inseizure protection. Finally, AOAA, unlike VPA, showed essentially constant increases inGABA across the brain regions. Because the density of GABAergic terminals variesappreciably, this once again showed VPA’s unique selectivity.The mechanism by which VPA elevates GABA concentrations is unclear. A decreasein GABA catabolism could be achieved initially by inhibition of GABA-T, the enzyme thatcatalyzes a transamination between GABA and ot-ketoglutarate. Because the bulk of GABAT activity is contained in glia and neuronal cell bodies, rather than the nerve terminals, theserepresent the site of most of the degradation of GABA (ladarola and Gale, 1981). However,VPA inhibits this enzyme significantly only at concentrations well above those normally14found in the brain (K = 9.5-40 mM, depending on source, for purified enzyme: Maitre et al.,1978, Whittle and Turner, 1978), although it is interesting to note that it it is markedly morepotent against neuronal GABA-T (IC50 = 0.63 mM for cultured cerebral neurons), than thepredominant glial GABA-T (IC50 = 1.2 mM for cultured astrocytes)(Larsson et al., 1986).This would explain why inhibition of GABA-T activity is observed in synaptosomes but notin whole brain homogenates (Loscher, 1993a). Interestingly, Nau and Loscher (1984) foundthat while 1 mM VPA had no effect on rat GABA-T activity in vitro, 1 mM 2-ene VPAcaused a 27 ± 1 % inhibition. The inhibition of the human enzyme by 2-ene VPA is evenmore pronounced as indicated by K = 4.5 mM (corresponding K1 for VPA was 40 mM:Maitre et al., 1978). Unfortunately, GABA-T activity has shown no temporal correlation withseizure protection in rats (Nau and Loscher, 1982).VPA has also been shown to inhibit two other enzymes in the GABA degradationpathway, succinic semialdehyde dehydrogenase (SSADH) and non-specific succinicsemialdehyde reductase (SSAR) with K1 values of 0.5-1.5 mM (Harvey et al., 1975; van derLaan et a!., 1979) and 38-85 uM (Whittle and Turner, 1978), respectively. In principle, theresulting accumulation of succinic semialdehyde could either promote its conversion back toGABA by GABA-T or merely inhibit the forward reaction of this enzyme. However, thenearly complete inhibition of SSADH by p-hydroxybenzaldehyde was found to have no effecton GABA levels (Simler et al, 1981). The situation is similar for SSAR, where variousflavonoids which are known potent inhibitors fail to show any anticonvulsant activity in theMES test (Whittle and Turner, 1981), except for the fact that the normal enzyme product, yhydroxybutyrate, produces absence-like epileptogenic effects (Snead et a!., 1980). This isparticularly intriguing because such seizures are normally promoted in animal absence modelsby direct or indirect GABA agonists (Vergnes et a!., 1985 and references cited within).Consequently, this feature of VPA could explain how it could potentiate GABAergicneurotransmission while still being an effective drug for absence seizures. Unfortunately forthis theory, it has been shown that VPA in vivo increases, rather than decreases brain levels of‘y-hydroxybutyrate (Snead et a!., 1980).15To complete the list of VPA’s inhibition of various enzymes involved in GABAmetabolism, there is a report by Luder et al. (1990) showing that both VPA and 2-ene VPAinhibit the x-ketoglutarate dehydrogenase complex as their CoA esters. Thus, inhibition ofthis enzyme system would decrease flux through the TCA cycle and thus divert OLketoglutarate into the GABA shunt. The potencies for both are very respectable but theirmechanisms appear different as VPA acted solely in a competitive manner (K = 2.6 uM)while 2-ene VPA behaved as a mixed-type inhibitor (K = 6.1 uM and = 1.0 uM). Thisfundamental difference was further demonstrated by the irreversible inactivation of theenzyme when incubated with 2 mM 2-ene VPA. The authors suggested this as theexplanation for the duration of the anticonvulsant effect after discontinuation of VPA therapy.No such effect was observed with up to 20 mM VPA.The evidence for a link betweem the anticonvulsant effects of VPA and enhancementof GABA synthesis is more encouraging. Injection of VPA (80 mg/kg i.p.) followingadministration of[14C1-glucose was found to result in a significant increase in the productionof labelled GABA (Taberner et al., 1980). This line of work was recently extended byBolanos and Medina (1993), who incubated neonatal brain slices with [2-14CJ, [3,4-1C] and[6-’4Cjglucose, respectively, and measured the amount of 14C02 produced. Recall thatglucose carbon atoms 3 and 4 are lost as CO2 during oxidation of pyruvate to acetyl CoAimmediately prior to entry into the TCA cycle, carbon atoms 1 and 6 are lost during theoxidation of the TCA intermediate isocitrate to o-ketoglutarate and carbon atoms 2 and 5 arelost during either the successive oxidation to succinate or the conversion of glutamate toGABA. They found the production of labelled CO2 from [3,4-1C]glucose and [[6-14C]glucose to be unaffected by VPA and 2-ene VPA, indicating that these drugs do notinfluence glucose utilization and the TCA cycle, respectively. On the other hand, these drugssubstantially enhanced the amount of ‘4C02 derived from [2-14C]glucose. The authorsestimated the activity of the GABA shunt pathway to be increased by 44 and 14 1%,respectively. In addition to supporting an enhancement of GABA synthesis as the cause ofthe elevated GABA levels, this result suggests that neither GABA-T nor o-ketoglutarate16dehydrogenase are likely targets of the two drugs as their inhibition would result in smalleramounts of‘4C02being detected.There is good evidence for this increased GABA shunt activity being caused byactivation of glutamate decarboxylase, whose time course parallels those of the GABAincrease and anticonvulsant effects (Nau and Loscher, 1982). Interestingly, this increase inactivity is generally observed only in brain homogenates, slices and synaptosomes followingin vivo VPA administration (Loscher, 1993a). Bolanos and Medina (1993) found no effect onenzyme activity when crude or “partially purified” enzyme was incubated with VPA, althoughTaylor et al. (1992) reported a marked enhancement of GAD activity using 0.25 mM VPAand >90% pure enzyme. Similarly, Nau and Loscher (1984) reported that 1 mM of eitherVPA or 2-ene VPA had no effect on GAD or GABA-T activity in vitro. The rapid onset ofthis activation could be explained by the conversion of the inactive apoenzyme to the activeholoenzyme (see below), although this has yet to be proven. In fact, the minimalinvestigation into the mechanism of this activation by ED5O doses of VPA and 2-ene VPA(Loscher et al., 1981) to date would make a summary of the properties of this enzymeappropriate at this point. GLUTAMATE DECARBOXYLASEGAD (reviewed by Martin and Rimvall, 1993) exists primarily as two isozymes ofmolecular weights 65.4 kDa (GAD65)and 66.6 kDa (GAD67)(Erlander et al., 1991). GAD65is localized primarily in neuronal terminals whereas the other isozyme is more evenlydistributed throughout the cell. However, most differences between the two forms aregenerally minor. A significant feature of the enzyme, from a regulatory standpoint, is the factthat it is found primarily in the apoenzyme form, without bound pyridoxal 5-phosphate (PLP)co-factor. Consequently, the cell has a large reserve capacity for synthesizing GABA when asudden demand arises. In synaptosomes, the bulk of this apoenzyme can be attributed to theGAD65 species. It is this apo-GAD/holo-GAD equilibrium that is primarily responsible for17-ATPE-PLP E + PMP + succinic semialdehyde/ E(PLP)E = apo-GADE(PLP) = GAD with reversibly associated PLPE-PLP = holo-GADE(PLP-Glu) = holo-GAD with bound glutamateEQ = quinoid intermediatepyridoxamine 5’-phosphateFigure 4. The regulation of glutamate decarboxylase (GAD).Modified from Martin and Rimvall, 1993.Co2E(PLP-Glu) EQ E(PLP) + GABAp43- PLPOHPLP =PMP =pyridoxal Y-phosphate18the regulation of the enzyme, as both substrate and co-factor are believed to be present atlevels well above their effective concentrations. For example, in brain slices K+stimulatedGABA release was unaffected by elevated concentrations of glutamate precursors glucose andglutamine, indicating an intracellular concentration of glutamate well above its Km value(Szerb and O’Regan, 1984, 1986).Normally the enzyme is present in the apoenzyme form and it combines tightly withPLP to form the holoenzyme that proceeds to form the quinoid intermediate EQ withglutamate as shown, a process promoted by phosphate through an allosteric mechanism(Figure 4). At this point, the species usually continues to form GABA and the regeneratedholoenzyme, but it may also dissociate into apoenzyme, pyridoxamine 5’-phosphate andsuccinic semialdehyde. The apoenzyme may then be converted back to the holoenzymeexcept in the presence of ATP which binds to the enzyme in competition with PLP (Sze et al.,1983; Porter and Martin, 1988) and stabilizes the protein against thermal decomposition, asthe apoenzyme is normally unstable at physiological temperatures (Porter and Martin, 1988).Evidence for the operation of this mechanism in vivo comes primarily from acomparison of how apo-/holo-GAD levels are influenced by the concentrations of glutamate,phosphate and ATP. For example, the normal concentration of phosphate found in cells issufficient to promote the decarboxylation reaction while the concentration of free ATP (about100 uM: Meeley and Martin, 1983) is well above the K value for inhibiting the binding ofPLP to apo-GAD. It should be noted that this role of ATP can also be played by other anionicspecies (Tursky, 1970; Martin and Martin, 1982) as synaptosomes, which have low levels ofATP, have high levels of apo-GAD. Similarly, postmortem holo-GAD levels rise rapidlywhen ATP levels drop and the phosphate concentration rises (Miller et at., 1977). Jn addition,the apo-/holo-GAD equilibria are influenced by factors such as depolarization, whichconverts apo-GAD to holo-GAD in synaptosomes (Miller and Walters, 1979). Finally, therehas not been any evidence to date that PLP plays a role in this regulatory process, although itwould seem to be a likely candidate. Unfortunately, this molecule is highly bound to protein19within the cell, making it difficult to measure changes in its free available concentration(Martin and Rimvall, 1993). YPA AND POST-SYNAPTIC EFFECTSA post-synaptic site of action for VPA, whereby it would potentiate the action ofGABA, has also been investigated. Macdonald and Bergey (1979) reported that VPA appliedmicroiontophoretically to cultured mouse spinal cord neurons voltage clamped at -90 mVaugmented the depolarizing response to GABA but not glycine or glutamate. Similarly,Hackman et at. (1981) observed an increase in the GABA-induced response of frog spinalcord in the presence of 10 uM VPA. There have been numerous reports since of a similareffect of VPA on the GABA response using microiontophoretic application to cortical(Schmutz et aL, 1979; Kerwin et at., 1980; Baldino and Geller, 1981) and nigral neurons(Kerwin et at., 1980), potentiation of the depressant effect of muscimol in spontaneouslyfiring locus coeruleus neurons (Olpe et at., 1988) and potentiation of the responses of bothGABA and muscimol in medullary reticular formation neurons (Gent and Phillips, 1980).However, there have also been a similar number of studies where no effect of VPA onthe GABA response was found. For example, McLean and Macdonald (1986) were unable todetect a significant enhancement of the response of cultured mouse spinal cord neurons toGABA when VPA was applied by microiontophoresis. Buchhalter and Dichter (1986) foundthat up to 3 mM VPA had no effect on the increased chloride conductance of dissociatedcortical neurons in response to GABA. Similarly, no potentiation of the GABA response wasobtained by Preisendorfer et al. (1987) with 5 mM VPA using hippocampal neurons, andGallagher et at. (1981), using 0.1-1 mM VPA applied to cat dorsal root ganglion cells. Thislatter result is significant because the preparation is inherently free of synapses. Thus, anypotentiation would have been the result of VPA acting at the level of the GABA receptorcomplex. It has even been reported that VPA promotes spontaneous firing of neurons in therat cortex and dorsal hippocampus (Blume et at., 1979), an effect that was found to be20antagonized by GABA. However, these results should be treated with caution as the animalswere anesthetized with pentobarbital, an indirectly acting GABA agonist. Finally, Harrisonand Simmonds (1982) found that while VPA did indeed enhance the muscimol-induceddepolarization of rat cuneate afferent fibres, as determined by extracellular recording, thiseffect was only apparent in the 3-10 mM range, well above the normal concentration ofunbound drug in the brain.A different approach to demonstrate the lack of an exclusive interaction between VPAand GABAergic synaptic transmission was provided by Agopyan et al. (1985), who studiedrhythmic synchronous bursts in hippocampal slices in a low-Ca2 and high-Mg2mediumwhere synaptic transmission is virtually eliminated. VPA (0.5 mM) markedly decreased thesebursts even in the additional presence of the GABA site antagonist bicuculline. It should benoted that this experiment merely showed that the effects of VPA are not exerted exclusivelyon the GABA receptor itself. INFLUENCE OF VPA ON MEMBRANE ION CHANNELSLike other antiepileptic drugs, VPA also influences the permeability of the cellmembrane to ions. Most studies to date have shown that sodium conductance is reduced andthe gating of its channel is slowed down (Zona and Avoli, 1990 and references cited below).For example, van Dongen et al. (1986) found that 0.5 mM VPA produced a 5-40% reductionin the peak sodium current obtained in the sciatic nerve of Xenopus laevis under voltageclamp conditions. The decreased excitability of the nerve was further demonstrated by thedecreased action potential amplitude and maximal rate of rise, increased threshold potentialand diminished ability to transmit high frequency signals. The latter finding is consistent withthe work of McLean and Macdonald (1986), where even 30 uM VPA, along with phenytoinand carbamazepine, inhibited both spontaneous and stimulated sustained repetitive firing ofdissociated mouse spinal cord neurons, and is clearly of particular relevance to the ability of adrug to prevent the spread of seizure activity. However, a common site of action is probably21not involved as VPA, unlike phenytoin and carbamazepine, is unable to reduce thebatrachotoxin-induced influx of 22Na+ into neuroblastoma cells and rat brain synaptosomes(Willow et al., 1984) and has a much longer latency to inhibition of repetitive firing in areaCAl of the rat hippocampal slice (Franceschetti et at., 1986). Van den Berg et at. (1993), intheir detailed investigation of the sodium channel gating effects of 1 mM VPA in cultured rathippocampal neurons, explained this property as the result of a 200% increase in the recoverytime constant for the sodium current. In addition, they found a marked reduction in the peakconductance and the current decay time constant and showed that VPA’s inhibition of Na+influx was both use- and voltage-dependent, being more pronounced at more positivepotentials. Again, this is consistent with the actions of a drug that acts to inhibit elevatedlevels of neuronal activity. Furthermore, the voltage-dependence indicates that VPA’smechanism did not involve simple channel blockade, as seen with the local anesthetics(Fohimeister et al., 1986).This promotion of sodium channel inactivation and delay of recovery has also beenreported for Myxicola giant axons (Schauf, 1987) and squid giant axons (Fohimeister et al.,1986). However, in both studies the drugs were applied internally and in the latter report noeffect was observed when up to 20 mM VPA was added to the perfusion medium.There is little agreement between studies on the effect of VPA on potassium channels.Some have seen little or no effect on potassium conductance or channel gating (Fohlmeister etat., 1986; Franceschetti et at., 1986 (indirectly, as these workers monitored extracellularpotassium levels); Schauf, 1987; Zona and Avoli, 1990). One early study (Slater andJohnston, 1978) even claimed to observe an increase in potassium conductance in Aplysianeurons. However, the concentrations of VPA employed were very high (5-30 mM) and therelevance of the model system to human epilepsy is questionable. On the other hand, it hasalso been shown (van Dongen et al, 1986; van Erp et at., 1990) that in the sciatic nerve of X.laevis 2.4 mM VPA had a biphasic, voltage-dependent effect on potassium conductance,increasing it at potentials more negative than -50 mV and reducing it at more positivepotentials. Although a reduced potassium conductance would hardly seem compatible with22the mode of action of an antiepileptic drug, the authors suggested that this may serve to limitthe concentration of extracellular potassium whose accumulation may promote sustainedrepetitive firing (Dichter and Ayala, 1987). This idea, however, is not supported by the workof Franceschetti et al. (1986), who found only a very minor reduction by 1 mM VPA of theextracellular potassium accumulation during spontaneous epileptic activity in rat hippocampalslices.There is good evidence that VPA reduces the low threshold (T) Ca2 current that isbelieved to play an important role in absence seizures. It has been suggested that it is the deinactivation of these channels by GABAB receptor-mediated hyperpolarization that is thecause of their aggravation of absence seizures (Hosford et at., 1992). Kelly et at. (1990)found that VPA reduced the total Ca2+ (T) current in dissociated rat nodose ganglion neuronsat approximately therapeutic concentrations (0.1-0.5 mM) but the reductions (10-27%) wereless than the values observed with the exclusively antiabsence drug ethosuximide in thethalamic relay neurons that are believed to play a critical role in generalized absence seizures(Coulter et at., 1989). Because ethosuximide was not tested by Kelly et at. (1990), onecannot say whether or not their results can explain VPA’s antiabsence properties. By contrast,Coulter et at. (1989) failed to observe an effect of even 1 mM VPA on the low threshold Ca2current in isolated ventrobasal complex (thalamic) neurons from rats, but found ethosuximideto be a most effective blocker. VPA AND MEMBRANE FLUIDITYEven if the GABA hypothesis of VPA’s mode of action is true, there still remains thequestion of the actual effector mechanism in the absence of any known binding sites. Anappealing candidate for this role is the plasma membrane itself, whose perturbation by theincorporation of VPA could lead to altered ionic conductances and thus altered synthesis andrelease of GABA.23Seeing that VPA has no known binding sites and distributes throughout the brain in anessentially homogeneous manner (Schobben et al., 1980), it is certainly relevant to ask how itexerts its effects on ion channels and GABA-metabolizing enzymes. One possibility is thatthe drug acts via the plasma membrane to increase its fluidity and hence the properties of theproteins embedded in it. In other words, the greater the degree to which the drug can disturbthe membrane, the greater its potency.The best illustration of such a mechanism is provided by the volatile anesthetics thatpossess no activity-shape dependence, exhibit a very high correlation between potency andlipophilicity and require quite high concentrations to be effective. For example, the minimumalveolar concentrations of most clinical anesthetics are in the 0.01-0.1 atm range whichcorresponds to 0.39-3.9 mM in the lung. At equilibrium, this will be the brain concentrationas well, owing to the anesthetics? high lipophilicity that allows them to readily cross theblood-brain barrier. Such properties, in fact, compare favourably with those observedfollowing therapeutic doses of VPA. The mechanism by which the anesthetic effect isexerted remains unclear, however. One explanation offered is based on the anesthetics’promotion of membrane disordering leading to a decrease in the gel-liquid crystallinetransition temperature. Because this transition is accompanied by a change in volume, theeffect of the anesthetics would be to decrease the magnitude of the resultant changes involume. Because integral membrane proteins such as ion channels would be expected to besensitive to such changes in their environment, the incorporation of a membrane disorderingagent could thus markedly influence their activity, though the exact effect would be open tospeculation (Kennedy and Longnecker, 1990).There has been remarkably little interest in exploring this particular facet of VPA’saction, even though Perlman and Goldstein (1984) have shown that when corrected for brainconcentrations, ED5O values of a wide range of linear and branched fatty acids show a veryhigh degree of correlation with their ability to increase the fluidity of synaptosomalmembranes measured using fluorescence polarization spectroscopy. A bothersome aspect ofthis study, however, was the very high concentration (>10 mM for VPA) of drugs needed to24detect a significant change in membrane fluidity. This may have been due to the fact that theprobe used in this study was embedded in the middle of the membrane bilayer and might thusnot be able to detect changes in membrane structure near the surface where such anionicdrugs might be expected to act. Alternately, it is possible that changes in fluidity that wouldexert a profound effect on an integral membrane protein might simply not be detected withthis assay, necessitating the use of higher drug concentrations. Nevertheless, there is solidsupport for a role of increased membrane fluidity in the anticonvulsant action of VPA. Luckeet al. (1993) used the diffusion properties of VPA and 2-ene VPA to conclude that the drugsare unlikely to significantly penetrate the plasma membrane and thus reside primarily in it orin the extracellular space. This was demonstrated by measuring the drugs’ diffusioncoefficients in agar (D, representing a strictly aqueous medium) and cortical tissue (D *)• Theresultant ratio 2 = D/D*, where is known as the tortuosity factor, then represents the relativeimpedance of the biological medium to the diffusion of the analyte. A substance that freelycrosses biological membranes will have ? approaching unity, whereas the value of a strictlyextracellular species such as the tetramethylammonium ion will be much larger. Lucke et a!.found that the tortuosity values for VPA (1.90) and 2-ene (1.70) were similar to thosereported for the tetramethylammonium ion, indicating a predominantly extracellulardistribution of these species. This finding also implies that determinations of brainconcentrations of VPA- and 2-ene VPA-related drugs using whole brain homogenate (such asthe one described in this thesis) will greatly underestimate the true effective concentrations, asthe extracellular space represents only about 20% of the total brain volume.Indirect evidence for such a mechanism has also been provided by Hauck et a!. (1991)in their study of the effect of configuration on two VPA analogues (2-(2’-propenyl)pentanoicand 2-(2’-propynyl)pentanoic acids). The anticonvulsant potencies of each compound’senantiomers were evaluated separately and found to be independent of configuration, arguingagainst the existence of an enantioselective receptor site. Instead, this role could be served bythe asymmetric medium of the interior of the plasma bilayer that would be sensitive to thedrug’s basic structure but not to its chirality.25A more clinically relevant example was offered by Tangorra et at. (1991), who foundthat the fluidity of erythrocyte membranes from epileptic patients was significantly affectedby therapeutic doses of various antiepileptic drugs including VPA. However, these workersemployed the cationic probe TMA-DPH which is tethered to the membrane surface andwhose fluorescence is thus a function of the fluidity of the environment near the surface of themembrane. They found that patients receiving antiepileptic drugs showed decreasedmembrane fluidity which appeared to correlate with the lipoperoxide content of themembrane. However, it is difficult to draw conclusions from the data as the authorspresumably used non-epileptic patients not receiving drug therapy as a control. Thus, onecannot distinguish the effects of epilepsy itself from those of the medication.1.2.3. STRUCTURE-ACTIVITY STUDIESEarly structure-activity studies (Carraz, 1967; Taillandier et at., 1975) demonstratedthat while many carboxylic acids and their derivatives possessed significant anticonvulsanteffects, VPA represented the optimum balance between seizure protection and sedation.Specifically, increasing size by the use of longer alkyl chains increased anticonvulsantpotency but at the cost of enhanced sedation.There have been few recent structure-activity studies of VPA and its analogues, whichis hardly surprising in view of the proven efficiency of the prototype drug. Loscher and Nau(1985) examined a fairly broad range of VPA analogues, focusing primarily on theiranticonvulsant potency and toxicity. They found that an alkyl substituent at the c€ carbonatom was virtually essential for activity, which also increased with the size of the substituent.Interestingly, there was no further significant increase in potency for ct-monosubstitutedaliphatic acids larger than VPA, although a-disubstituted acids were shown to be even morepotent than VPA, further emphasizing the requirement for a branched structure.Unfortunately, this latter group frequently caused delayed deaths after several days,demonstrating that VPA represented the optimum structure for this group of compounds in26terms of both anticonvulsant efficacy and toxicity. Qualitatively similar results were obtainedby Keane and Morre (1985).Table 3Anticonvulsant properties of VPA and its analoguesCompound i.p. ED5O i.c.v. ED5O(mmollkg) (umol)Loscher and Nau (l985)VPA 1.7(E)-2-ene VPA 1.62-Propyihexanoic acid 1.62-Butyihexanoic acid 1.5l-Methyl-1-cyclohexanoic acid 1.92-Ethyl-2-methylpentanoic acid 1.32,2-Dimethylpentanoic acid 1.5Keane etal. (l983)VPA 0.81Butyric acid inactive2-Ethylbutyric acid 2.232-Ethylpentanoic acid 1.212-Ethyl-hexanoic acid 1.14Pentanoic acid inactive2-Methylpentanoic acid 3.642-Butylpentanoic acid 0.752-Pentylpentanoic acid 0.6Hexanoic acid inactive2-Butyihexanoic acid 0.77Chapman etal.(1984)bVPA 1.25 6.02-Ethyihexanoic acid 0.66 10.22-Propylhexanoic acid 0.68 5.0Pentanoic acid 5.80 > 15a) Endpoint: prevention of scPTZ-induced seizures in miceb) Endpoint: prevention of audiogenic seizures in DBAI2 miceChapman et al. (1984) studied the potencies of VPA analogues using both i.p. andintracerebroventricular (i.c.v.) routes of injection. They found that the enhanced potencies ofthe larger acids were due to pharmacokinetic rather than pharmacodynamic effect because themost potent compound by the i.p. route was amongst the least effective when intrinsicpotency was evaluated by i.c.v. administration. This suggests that pharmacokinetic factors27make a significant contribution to the observed anticonvulsant effect. Similar results havebeen obtained by Penman and Goldstein (1984), who found that on the basis of brainconcentrations, both heptanoic and octanoic acids were up to seven times more potent thanVPA and thus emphasized the need for normalization of anticonvulsant data.More recently, Elmazar et at. (1993) looked at 2-substituted pentanoic acids as VPAanalogues. As shown, substitution with various .alkyl and alkenyl groups did not significantlyaffect potency. However, the ED5O dropped dramatically when a triple bond was introducedinto one of the branches. In addition, the protective index (TD5O/ED5O, using rotorod andscPTZ tests, respectively) was greatly improved. It is interesting to note that the calculatedlipophilicity of this compound was quite low, in contrast to the usual trend where potency is alinear function of log P.There have been even fewer quantitative structure-activity studies of this family ofcompounds. Abbott and Achaempong (1988) examined a broad array of carboxylic acids andtetrazoles for anti-scPTZ activity in mice, using Taft’s steric parameter, log P, (log P)2 andPKa as physicochemical descriptors in the Hansch linear free energy model. The Taftparameter was found to correlate with log P and was not used. The best correlation wasfound for log P and PKa, which indicated optimal activity with increasing lipophilicity anddecreasing acidity. Because the brain concentrations of these drugs were not measured, it isnot known if log P and PKa influenced primarily the interaction of drug with the effector orsimply the brain’s drug uptake ability.28CO2H CO2H2-(2’-Propynyl)- 2-Propyihexanoic acid 2-(Methylcyclopropyl)-pentanoic acid pentanoic acidCO2H ço22-Methyl-2-propyl- 2-(2’-Butynyl)- 2-(2’-Methyl-2’-pentanoic acid pentanoic acid propenylpentanoic acidFigure 5. Anticonvulsant 2-substituted pentanoic acids (Elmazar et al., 1993).Table 4Anticonvulsant and lipophilicity properties of some 2-substituted pentanoic acids (Elmazar etal., 1993)Compound log P ED5O NeurotoxicityaVPA . 2.72 0.71 332-(2’-Propynyl)pentanoic acid 1.31 l.78 204-Ene VPA 2.18 1.29 292-Methyl-2-propylpentanoic acid 3.12 0.40 802-Propyihexanoic acid 3.25 0.78 802-Ene VPA 2.59 0.68 602-(Methylcyclopropyl)pentanoic acid 2.63 0.96k 202-(2’-Butynyl)pentanoic acid 1.84 0.28 02-(2’-Methyl-2’-propenyl)pentanoic acid 2.57 0.84 0*) Significantly different (p <0.05) from VPA.a) Percentage of mice (n = 7) showing minimal neurological deficits on a rotating rod 15 mmfollowing i.p. administration of 1.5 mmol/kg drug.291.3. OBJECTIVESAt this point, we can summarize VPA as a clinically useful drug with two significantdrawbacks. The first is practical, for although VPA’s side effects are generally few, thosewhich may occur, such as its hepatotoxicity and teratogenicity, can be fatal. The incidence ofsuch events can be minimized by not prescribing the drug to children on polytherapy or topregnant women, but it clearly indicates the potential for improving the therapeutic utility ofVPA. The second problem with VPA is primarily academic in that its mechanism of actionstill contains unanswered questions. For example, no study has sought a correlation for aseries of VPA analogues between anticonvulsant activity and synaptosomal GABA levels orthe activities of enzymes in the GABA pathway. Working on the assumption of an identicalmechanism of action, such a correlation could not only prove that elevated GABA was thekey to the anticonvulsant properties, but would also indicate VPA’s biochemical target.Both problems could theoretically be addressed by studying the properties of 2-eneVPA and its analogues. This compound, formed by the 13-oxidation of VPA, exhibits noteratogenicity and minimal hepatoxicity and it may be that its analogues exhibit similarproperties. Furthermore, its centrally located double bond makes it ideal for QSAR work as itimparts a degree of rigidity lacking in VPA itself. Combined with the fact that it has many ofthe same biological properties as its parent, including equivalent anticonvulsant potency, 2-ene VPA would thus appear to be an ideal starting point to probe the mechanisms of VPA andperhaps develop more efficient anticonvulsant analogues.The objectives of this project were the following:1) prepare at least 10-15 analogues of 2-ene VPA: specifically, aliphatic a,13-unsaturatedacids with alkyl or alkenyl substituents2) for the above compounds, determine:a) anticonvulsant and sedative propertiesb) lipophilicity and solution conformation30c) membrane-disordering potencyd) effect of ED5O dose on nerve terminal GABA levels and GAD activities3) develop quantitative relationships between these physicochemical and biologicalparameters4) investigate the mechanism of GAD interaction by the effect of drugs ona) saturation of GAD with its cofactor PLPb) GAD activity in vitroIt should be stressed that the primary objective of this study was to shed light on themechanism of action of 2-ene VPA, and thus presumably VPA itself, rather than to search fora more potent anticonvulsant drug, which could be done more efficiently by a more broadlybased screening protocol.312. EXPERIMENTAL2.1. MATERIALSFlash chromatography was performed with silica gel 60 (Merck 9285, 230-400 mesh).Melting points were determined with a Thomas-Hoover melting point apparatus and areuncorrected. 1H-NMR spectra were recorded on Bruker 200 MHz or 300 MHz instruments atthe Chemistry Department, UBC. JR spectra were recorded on a Bomem MB-l00 instrument.Qualitative GC-MS analyses were performed on a Hewlett-Packard HP5700A gaschromatograph (packing: 3% Dexsil 300; oven: 50°C initial to 260°C @ 16°C/mm or32°C/mm) interfaced to a Finnegan MAT-ill mass spectrometer. Where noted, alcohols andacids were derivatized with MSTFA or MTBSTFA in ethyl acetate prior to analysis.Quantitative GC-MS analyses were performed using a Hewlett-Packard HP 5890 gaschromatograph interfaced to a HP 5989A mass spectrometer. GC: 34.5 kPa helium headpressure, Hewlett-Packard HP-i capillary column (12 rn x 0.2 mm ID x 0.33 urn film).Injection: 1 ul. MS: electron impact ionization potential 70 eV, single ion monitoring, dwelltime 75 msec, source 275°C, quad 100°C.Elemental analyses were performed by Mr. Peter Borda at the Department ofChemistry, UBC.Male CD-i mice were obtained from University of British Columbia Animal Servicesor Charles River, PQ: in the latter case, the animals were allowed one day of recovery uponarrival. GABA-d6 was purchased from MSD Isotopes (Montreal, PQ). TMA-DPH waspurchased from Molecular Probes (Eugene, OR). Linear regression calculations wereperformed using QSAR-PC (BIOSOFf, Cambridge, MA).322.2. SYNTHESISAcids 3, 9 and 10 were prepared as described by Lee (1991) and Lee et al. (1989).Acids 2 and 4 were obtained commercially (Aldrich).2.2.1. (±)-4-( 1 -Methylethyl)-2-oxazolidinone (la)HNyODL-Valinol (25.06 g, 243 mmol) was dissolved in diethyl carbonate (170 mL).Potassium tert-butoxide (6.95 g, 62 mmol) was added and the mixture was brought to refluxover 15 mm. Distillate was collected until the stillhead temperature rose to 126’C (totaldistillation time: 30 mm). The solution was allowed to cool and then diluted with diethylether (150 mL) and washed with saturated ammonium chloride (3 x 30 mL) and brine (30mL). The organic extract was dried over magnesium sulfate, filtered and evaporated invacuo. The solid residue was decanted with petroleum spirit (2 x 30 mL) and 1:9 and 1:4diethyl ether/petroleum spirit solutions (30 mL each). The resultant suspension was refluxedbriefly with a 2:3 solution (30 mL), cooled at -10’C, filtered and washed with 1:4 diethylether/petroleum spirit (2 x 15 mL) to afford the oxazolidinone (21 g, 68%) as a white powder.mp 69-72°C1H-NMR (200 MHz, acetone-d6)ö: 0.90 (6H, m, -CH3), 1.68 (1H, m, -Cjj(CH3)2,3.62 (1H,m, -CHNH), 4.08 (1H, dd, J = 9, 8 Hz, -CH2O-), 4.40 (1H, dd, J = 9, 9 Hz, -CH2O-), 6.90(1H, hr s, NH).IR (CHC13,cm1): 3244, 2950, 1749, 1474, 1405.GC-MS (iWz): 129 (Mj, 86, 42.332.2.2. Cyclopentylacetic acid[)—CH2COHCyclopentylacetic acid was prepared from diethyl malonate and cyclopentyl bromideby the general procedure of Furniss et al. (1987a). Fractional distillation afforded 2-cyclopentylacetic acid (9.26 g, 39%, bp 106-112°C / 6 mm Hg: lit. bp 226-230°C / 760 mmHg, bp 139-140°C /26 mm Hg (Buckingham, 1982)) as a clear oil.1HNMR (200 MHz, acetone-d6) ö: 1.05-1.26 (2H, m, cyclopentyl), 1.42-1.70 (4H, m,cyclopentyl), 1.70-1.91 (2H, m, cyclopentyl), 2.04 (1H, m, H(3)), 2.09-2.31 (2H, m, H(2)).IR (neat, cm-1): 3140-3000, 2950, 1709, 1427, 1292, 1217, 1136.GC-MS (m/z, as TMS derivative): 200 (Mj, 185, 132, 117, 73, (±)-3-( 1 -Oxoalkyl)-4-( 1 -methylethyl)-2-oxazolidinones (2a/a-f) and (±)-3-( 1-oxoalkyl)-2-oxazolidinones (2b/e-f)x 2aa: X = i-Pr; R = Me 2be: X - H; R = i-PrR 2ab: X = i-Pr; R = Et 2bf: X = H; R = cyclopentyl2ac: X = i-Pr; R = Pr2 N.,O 2ad: X = i-Pr; R = BuI] 2ae: X = i-Pr; R i-Pr0 0 2af: X = i-Pr; R cyclopentylCyclopentylacetyl chloride (bp 57-58°C I 6 mm Hg: lit. bp 55°C / 10 mm Hg(Seubold, 1954)) and pentanoyl chloride (bp 127-132°C: lit. 124-127°C (Furniss et al.,1987b)) were prepared from their respective acids by the procedure described by Furniss et al.(1 987b). The remaining acid chlorides were obtained commercially (Aldrich).The oxazolidinone (la/b: 46.5 mmol) was dissolved in dry THF (180 mL) under anitrogen atmosphere and the solution cooled to -78°C. Butyllithium (1.6 M in hexanes, 32.0mL, 51.2 mmol, 1.1 eq.) was added and the solution stirred for 5 mm. A solution of the acid34chloride (51.2 mniol, 1.1 eq.) was then added as a solution in THF (10 mL) and stirringcontinued for 10 mm. The reaction was quenched with saturated aqueous ammoniumchloride (100 mL) and the mixture warmed to room temperature. The layers were separatedand the aqueous phase extracted with diethyl ether (180 mL). The combined organic extractswere washed with saturated aqueous sodium bicarbonate (40 mL) and brine (40 mL) beforebeing dried over magnesium sulfate, filtered and evaporated in vacuo. The crude product wasthen purified by distillation using a 10 cm Vigreaux column.(±)-3-( 1 -Oxopropyl)-4-( 1 -methylethyl)-2-oxazolidinone (2aa)7.35 g (85%, bp 82-87°C / 0.1 mifi Hg).1H-NMR (200 MHz, acetone d6) & 0.85 (3H, d, J = 7.0 Hz, -CH(C113)2,0.91 (3H, d, J = 7Hz, -CH(CH3)2, 1.08 (3H, t, J = 7.5 Hz, H(3)), 2.20-2.43 (1H, m, -CH(CH3)2,2.70-3.05(2H, m, H(2)), 4.20-4.50 (3H, m, -CHCH2O-).JR (neat, cm1): 2956, 1778, 1704, 1487, 1463, 1382, 1224, 1072.GC-MS (m/z): 185 (Mj, 142, 130, 100, 85, 68, 57.(±)-3-(l-Oxobutyl)-4-(l -methylethyl)-2-oxazolidinone (2ab)7.62 g (82%, bp 83°C / 0.05 mm Hg).1H-NMR (200 MHz, acetone-d) & 0.80-1.00 (9H, m, 3 x -CH3), 1.65 (2H, m, H(3)), 2.31(1H, septet of d, J = 5, 4 Hz, -Ckj(CH3)2,2.67-3.02 (2H, m, 11(2)), 4.26-4.52 (3H, m, -CHCH2O-).JR (neat, cm1): 2953, 1777, 1701, 1486, 1463, 1382, 1216.GC-MS (m/z): 199 (Mj, 184, 171, 156, 130, 86, 71.(±)-3-( 1 -Oxopentyl)-4-( 1 -methylethyl)-2-oxazolidinone (2ac)8.77 g (89%, 97-105°C / 0.05 mm Hg).351H-NMR (200 MHz, acetone-d6)& 0.80-1.00 (9H, m, -CH(cfl3)2H(5)), 1.28-1.48 (2H, m,H(4)), 1.48-1.70 (2H, m, H(3)), 2.20-2.44 (1H, m, -Cll(CH3)2,2.70-3.09 (2H, m, H(2)),4.28-4.56 (3H, m, -CHCH2O-).JR (neat, cm-1): 2948, 1779, 1708, 1463, 1384, 1205.GC-MS (m/z): 213 (Mj, 198, 184, 171, 130, 85.(±)-3-( 1 -Oxohexyl)-4-( 1 -methylethyl)-2-oxazolidinone (2ad)7.49 g (64%, bp 104-6°C / 0.05 mm Hg).1H-NMR (200 MHz, acetone-d6)6: 0.68-0.80 (9H, m, -CH(Ck13)2H(6)), 1.09-1.23 (4H, m,H(5), H(4)), 1.38-1.60 (2H, m, H(3)), 2.02-2.29 (1H, m, -Ci(CH3)2,2.51-2.88 (2H, m,H(2)), 4.03-4.39 (3H, m, -CHCH2O-).JR (neat, cm): 2944, 1772, 1702, 1486, 1464, 1383, 1215.GC-MS (m/z): 228 (Mj, 198, 184, 171, 130, 99.(±)-3-( 1 -Oxo-3-methylbutyl)-4-( 1 -methylethyl)-2-oxazolidinone (2ae)6.78 g (9 1%, bp 94-98°C / 0.05 mm Hg) at 3/4 scale of above.‘H-NMR (200 MHz, acetone-d6)6: 0.80-1.00 (12H, m, -CH(C113)2,2.00-2.12, 2.12-2.41(1H, 1H, m, m, -CH(CH3)2H(3)), 2.60 (2H, dd, J = 7.0, 15.0 Hz, H(2)), 4.22-4.55 (3H, m, -CHCH2O-).JR (neat, cm4): 2954, 1778, 1700, 1466, 1388, 1305, 1210.GC-MS (m/z): 213 (Mj, 198, 171, 130, 85, 69.(±)-3-( 1 -Oxo-2-cyclopentylethyl)-4-( 1 -methylethyl)-2-oxazolidinone (2af)9.13 g (82%, bp 127-130°C / 0.2 mm Hg).mp 33-37°C1H-NMR (300 MHz, acetone-d6)6: 0.88, 0.95 (3H, 3H, d, d, J = 7 Hz each, -CH(CH3)2,1.10-1.32 (2H, m, cyclopentyl), 1.46-1.73 (4H, m, cyclopentyl), 1.73-1.92 (3H, m,36cyclopentyl, H(3)), 2.21-2.40 (1H, m, -Cfl(CH3)2,2.77 (1H, dd, 3 = 7, 16 Hz, H(2)), 3.04(1H, dd, J = 7, 16 Hz, H(2)), 4.28-4.52 (3H, m, -CHCH2O-).JR (CHC13,cm-1): 3030, 2957, 1777, 1699, 1486, 1383.GC-MS (m/z): 239 (Mj, 196, 171, 130, 111.(±)-3-( 1 -Oxo-3-methylbutyl)-2-oxazolidinone (2be)10.44 g (66%, bp 89-92°C / 0.05 mm Hg)1H-NMR (200 MHz, acetone-d6)6: 0.95 (6H, d, J = 7 Hz, (CH3)2CH-), 2.0-2.2 (2H, m,(CH3)2CH-, H(3)), 2.75 (2H, d, 3 7 Hz, H(2)), 4.00 (2H, t, J = 8 Hz, -NCHCHO-), 4.44(2H, d of t, J = 1, 8 Hz, -NCHCHO-)JR (CHC13,cm): 2964, 1781, 1699, 1482.GC-MS (m/z): 156 (M-15), 129, 114, 101.(±)-3-( 1 -Oxo-2-cyclopentylethyl)-2-oxazolidinone (2bf)mp 44-46°C1.91 g(58%,bp 118°C/0.05 mm Hg)1H-NMR (200 MHz, acetone-d6) 6: 1.05-1.29 (2H, m, cyclopentyl), 1.44-1.72 (4H, m,cyclopentyl), 1.72-1.94 (3H, m, cyclopentyl, H(3)), 2.27 (1H, hr septet, J = 7 Hz, Cfi(CH2),2.88 (2H, d, 3 = 7 Hz, H(2)), 4.00 (2H, hr t, J = 8 Hz, -NCHCHO-), 4.45 (2H, hr t, J = 8 Hz,-NCHCHO-).JR (CHC13,cm): 2960, 1781, 1702.GC-MS (m/z): 197 (Mj, 177, 168, 154, 129, 111, 101, 88, 83.372.2.4. Aldol addition: (±)-(erythro)-3-[3-hydroxy-2-alkyl- 1 -oxopentylj-4-( 1 -methylethyl)-2-oxazolidinone (3aJa-f) and (±)-(erythro)-3-[3-hydroxy-2-alkyl- 1 -oxopentylJ-2-oxazolidinone(3bIe-f)X 3aa:X=i-Pr;RMe 3be:X=H;R=i-PrR 3ab: X = i-Pr; R Et 3bf: X H; R = cyclopentyl4 1 3ac: X i-Pr; R = Pr3ad: X i-Pr; R BuI II II 3ae: X = i-Pr; R = i-PrOH 0 0 3af: X = i-Pr; R = cyclopentylTo a solution of the carboximide 2a/b in dry dichloromethane (0.2 M, 30-40 mmol) at-78°C was added dibutylboron triflate (1.0 M solution in dichloromethane, 1.1 eq.) followedby triethylamine (1.4 eq.) in dichloromethane (5 mL). The pale yellow solution was stirred at-78°C for 1 h and then at 0°C for 15 mm. The solution was re-cooled to -78°C and freshly-distilled propionaldehyde (1.5 eq.) in dichloromethane (5 mL) was added: the solution wasthen stirred at -78°C for 1 h and 0°C for 1 h before being partitioned between aqueous sodiumbisulfate (1 M, 150 mL) and ethyl acetate/petroleum spirit (300 mL, 1:1 v/v). The organiclayer was washed with brine (45 mL), evaporated in vacuo and combined with diethyl ether(225 mL), saturated aqueous ammonium chloride (75 mL) and hydrogen peroxide (30%, 7.5eq.) at 0°C. The mixture was stirred at 0°C for 1 h and then diluted with diethyl ether (150mL). The phases were separated and the organic extract was washed with saturated aqueoussodium bicarbonate (35 mL) and brine (2 x 35 mL), dried over magnesium sulfate andevaporated in vacuo. Samples of the adducts were further purified by flash chromatography.(±)-(Erythro)-3-[3-hydroxy-2-methyl- 1 -oxopentylj-4-( 1 -methylethyl)-2-oxazolidir.one (3aa)1H-NMR (200 MHz, acetone-d6)ö: 0.80-0.95 (9H, m, 3 x -CH3), 1.20 (3H, d, J = 7 Hz,H(3’)), 1.3 1-1.50 (2H, m, H(4)), 2.30 (1H, septet of d, J = 4,7 Hz, -CH(CH3)2,3.61 (1H, d, J= 5 Hz, -OH), 3.62-3.85 (2H, m, H(2), H(3)), 4.25-4.55 (3H, m, -CHCH2O-).IR (CHC13,cm1): 3538, 3033, 2964, 1779, 1684, 1382.GC-MS (m/z, TMS ether): 300 (M-15), 230, 170, 158, 143, 130.38(±)-(Erythro)-3-[3-hydroxy-2-ethyl- 1 -oxopentyl]-4-( 1 -methylethyl)-2-oxazolidinone (3ab)-H-NMR (200 MHz, acetone-d6)ö: 0.79-1.05 (12H, m, 4 x -CH3), 1.35-1.90 (4H, m, H(3’),H(4)), 2.20-2.43 (1H, m, -Cll(CH3)2,3.67 (1H, s, -OH), 3.78-3.88 (1H, m, H(3)), 3.90-4.05(IH, m, H(2)), 4.30-4.59 (3H, m, -CHCH2O-).JR (neat, cm-1): 3436, 2933, 1764, 1693, 1462, 1384, 1211.GC-MS (m/z, TMS ether): 314 (M-15), 270, 228, 204, 184, 143, 130.(±)-(Erythro)-3-[3-hydroxy-2-propyl- 1 -oxopentyl]-4-( 1 -methylethyl)-2-oxazolidinone (3ac)1H-NMR (200 MHz, acetone-d6)6: 0.85-0.98 (12H, m, 4 x -CH3), 1.22-1.55 (4H, m, H(4),H(4’)), 1.63-1.89 (2H, m, H(3’)), 2.31 (1H, septet of d, J = 7, 4 Hz, -CH(CH3)2,3.56-3.68(1H, m, H(3)), 3.72 (1H, d, J = 6 Hz, -OH), 4.00-4.12 (1H, m, H(2)), 4.30-4.60 (3H, -CHCH2O-).JR (neat, cm1): 3445, 2923, 1771, 1690, 1462, 1382, 1302, 1220, 1098.GC-MS (m/z, TMS ether): 328 (M-15), 284, 258, 198, 186, 158, 143, 125.(±)-(Erythro)-3- [3-hydroxy-2-butyl- 1 -oxopentyl]-4-( 1 -methylethyl)-2-oxazolidinone (3ad)1H-NMR (200 MHz, acetone-d6)6: 0.83-0.98 (12H, m, 4 x -CH3), 1.21-1.54 (6H, m, H(3),H(4’), H(5’)), 1.72-1.86 (2H, m, H(4)), 2.34 (1H, septet of d, J = 7, 4 Hz, -CH(CH3)2,3.55-3.70 (1H, m, H(3)), 3.73 (1H, d, J = 6 Hz, -OH), 3.99-4. 10 (1H, m, H(2)), 4.28-4.60 (3H, m, -CHCH2O-).JR (neat, cm): 3474, 2921, 1770, 1690, 1457, 1381, 1205.GC-MS (m/z, TMS ether): 342 (M-15), 298, 272, 212, 200, 158, 143.(±)-(Erythro)-3- [3-hydroxy-2-( 1 -methylethyl)- 1 -oxopentyl]-4-( 1 -methylethyl)-2-oxazolidinone (3ae)mp 88-93CC391H-NMR (200 MHz, acetone-cl6)& 0.85-1.04 (15H, m, 5 x -CH3), 1.20-1.60 (2H, m, H(4)),2. 14-2.43 (2H, m, 2 x -Cfl(CH3),3.71-3.87 (2H, m, H(3), -OH), 4.12 (1H, dd, J = 6, 8 Hz,H(2)), 4.28-4.58 (3H, m, -CHCH2O-).IR (CHC13,cm1): 3414, 2967, 1768, 1678.GC-MS (m/z): 213, 198, 171, 130, 111, 85. Extremely broad peak indicative ofdecomposition in instrument.GC-MS (m/z, tert-BDMS ether): 328 (M-57), 284, 242, 198, 125.(±)-(Erythro)-3-[3-hydroxy-2-cyclopentyl- 1 -oxopentyl]-4-( 1 -methylethyl)-2-oxazolidinone(3af)1H-NMR (200 MHz, acetone-d6) ö: 0.88-1.05 (9H, m, 3 x -CH3), 1.1-2.4 (1211, m,cyclopentyl,-CH(CH3)211(4)), 3.69-3.85 (2H, m, -OH, H(2)), 4.19 (JH, dcl, J = 6, 10 Hz, -CHCH2O-), 4.29-4.57 (3H, m, H(3), -CHCH2O-).JR (CHC13,cm1): 3511, 2961, 1766, 1692.GC-MS (m/z, TMS ether): 354 (M-15), 340, 311, 211.(±)-(Erythro)-3-[3-hydroxy-2-( 1 -methylethyl)- 1 -oxopentyl]-2-oxazolidinone (3be)1H-NMR (200 MHz, acetone-d6): ö: 0.84-1.00 (9H, m, CH3), 1.2-1.6 (2H, m, H(4)), 2.23(1H, septet of d, J = 7Hz, H(3’)), 3.69 (1H, d, J = 6Hz, -OH, disappears with D20), 3.73-3.88(1H, m, H(3)), 4.00-4.15 (3H, m, -CHCH2O-, 11(2)), 4.39-4.50 (2H, dcl, J = 9, 9 Hz, -CHCH2O-)JR (CHC13,cm1): 3525, 2968, 1776, 1684.GC-MS (m/z, TMS ether): 286 (M-15), 258, 242, 162, 144, 125.(±)-(Erythro)-3-[3-hydroxy-2-cyclopentyl- 1 -oxopentylj-2-oxazolidinone (3bf)mp 91-94°C401H-NMR (200 MHz, acetone-d6) & 0.92 (3H, t, J = 8 Hz, H(5)), 1.2-2.0 (1OH, m,cyclopentyl, H(4)), 2.13-2.38 (1H, m, H(3’)), 3.50-3.85 (2H, m, -OH, H(2)), 4.00-4.20 (3H, m,H(3), -CH2CO-), 4.38-4.52 (2H, m, -CH2CO-).JR (CHCI3,cm-1): 3535, 2954, 1772, 1694, 1383.GC-MS (m/z, TMS ether): 312 (M-l5), 284, 268, 162, 151, Methyl (±)-(E)-2-alkylpent-2-enoate (5a-e) and (±)-(erythro) methyl 2-cyclopentyl-3-hydroxypentanoate (40R 5a:R=Me5b:R=EtThe crude adduct 3a/b was dissolved in THF/water (400 mL, 3:1 v/v) and cooled to0°C. 3,5-Dibutyl-4-hydroxytoluene (80 mg, 0.4 mmol), hydrogen peroxide (30%, 16.4 mL,145 mmol) and lithium hydroxide monohydrate (3.06 g (73 mmol) in 5 mL water) were thenadded and the solution stirred for 3 h at 0°C. Aqueous sodium sulfite (1.5 M, 106 mL, 160mmol) and saturated aqueous sodium bicarbonate (50 mL) were then added and the THFevaporated in vacuo. The aqueous residue was extracted with chloroform (200 mL), acidifiedwith hydrochloric acid and extracted with ethyl acetate (2 x 200 mL). The ethyl acetateextract was washed with brine (40 mL), dried over magnesium sulfate, filtered and evaporatedin vacuo to afford the 3-hydroxyacid. The hydroxyacid was then dissolved in anhydrousdiethyl ether and treated with diazomethane using the standard procedure (Furniss et aL,1987c). The ethereal solution was evaporated and the ester redissolved in drydichioromethane (0.14 M) at 0°C. Triethylamine (2 eq.) and methanesulfonyl chloride (2 eq.)were added and the solution was stirred for 1 h and then evaporated in vacuo and filtered withTHF in a volume equal to that of the initial dichloromethane solution. DBU (2 eq.) was then41added and the solution refluxed for 1 h. The mixture was diluted with an equal volume ofpetroleum spirit and washed with 1 M hydrochloric acid, saturated aqueous sodiumbicarbonate and brine. The extract was dried over magnesium sulfate, filtered and evaporatedin vacuo to afford a clear oil which was then purified by fractional distillation.Methyl (E)-2-methyl-2-pentenoate (5a)2.18 g (55% from carboximide 2aa, bp 133-136°C, lit. 51°C Ill mm Hg (Lide, 1994))1H-NMR (200 MHz, acetone-d6)8: 0.90 (3H, t, J = 8 Hz, H(5)), 1.70 (3H, m, H(3t)), 2.08(2H, m, H(4)), 3.56 (3H, s, -OCR3), 6.60 (1H, t, J = 8 Hz, H(3)).JR (neat, cm-1): 2950, 1720, 1450, 1345.GC-MS (m/z): 128 (Mj, 113, 97.Methyl (E)-2-ethyl-2-pentenoate (5b)1.21 g (31% from carboximide 2ab, bp 112-117°C / 20mm Hg)1H-NMR (200 MHz, acetone-d6)& 0.90-1.10 (6H, m, H(4t), H(5)), 2. 10-2.30 (4H, m, H(3),H(4)), 3.68 (3R, s, -OCH3), 6.68 (1H, t, J = 15 Hz, H(3)).JR (neat, cm1): 2961, 2876, 1714, 1647, 1440, 1297, 1239, 1149.GC-MS (m/z): 142 (Mj, 127, 113, 111,95,83,67.Methyl (E)-2-propyl-2-pentenoate (5c)4.02 g (54% from carboximide 2ac, purified by flash chromagraphy using 1:19 diethylether/petroleum spirit, v/v)1H-NMR (200 MHz, acetone-d6)8: 0.88 (3H, t, J = 8 Hz, H(S)), 1.04 (3H, t, J = 8 Hz, H(5’)),1.25-1.51 (2H, m, H(4’)), 1.75-1.84 (2H, m, H(3’)), 2.17-2.34 (2H, m, H(4)), 3.69 (3H, s, -OCH3), 6.71 (1H, t, J = 8 Hz, H(3)).JR (neat, cm-1): 2951, 2874, 1715, 1646, 1451, 1354, 1279, 1223, 1148.GC-MS (m/z): 156 (Mj, 127, 113, 95, 67, 55.42Methyl (E)-2-butyl-2-pentenoate (5d)4.12 g (64% from carboximide 2ad: bp 64-90°C /8 mm Hg).1H-NMR (200 MHz, acetone-d6)ö: 0.80-1.00 (3H, m, H(6’)), 1.04 (3H, t, J = 8 Hz, H(5)),1.20-1.40 (4H, m, H(4’), H(5’)), 2.15-2.40 (4H, m, H(4), H(3’)), 3.68 (3H, s, -OCH3), 6.72(1H, t, J = 15 Hz, H(3)).IR(neat, cm1): 2946, 2869, 1716, 1646, 1458, 1439, 1261, 1205, 1148, 1112.GC-MS (m/z): 170 (Mj, 155, 141, 127, 109, 95, 81, 69.Methyl (if) 2-( 1 -methylethyl)-2-pentenoate (5e)Using 1.6 g of the hydroxyester 4e (obtained in 4.47 g (70%) yield from carboximide 2be)gave only 100 mg of the methyl ester 5e using the above procedure.1H-NMR (200 MHz, acetone-d6)ö: 1.02 (3H, t, J = 8 Hz, H(5)), 1.2-1.4 (6H, m, H(4’),11(4”)), 2.20 (211, q, J = 8 Hz, H(4)), 2.2-2.3 (1H, m, H(3’)), 3.65 (3H, s, -OCH3), 6.68 (1H, t,J=8Hz,H(3)).JR (CHC13,cnr1): 2960, 1708, 1449.GC-MS (m/z): 156 (Mj, 141, 127, 109, 95, 81.(±)-(Erythro) methyl 2-cyclopentyl-3-hydroxypentanoate (4f0.94 g (62% from carboximide 2b1)1H-NMR (200 MHz, acetone-d6)8: 1.0-1.9 (1OH, m, cyclopentyl, H(4)), 2.12-2.28 (1H, m, -CHCH(CH2),3.50-3.64 (5H, m, -OCH3H(2), -OH), 3.78-3.82 (1H, m, 11(3)).JR (CHC13,cm1): 3450, 2954, 1725, 1443.GC-MS (m/z): 142 (M-58), 111,99, 83, 74.432.2.6. (E)-2-Alkyl-2-pentenoic acid (5-8)R 5:R=MeThe ester 5a-d was dissolved in methanol (8.2 mL/mmol). Aqueous potassiumhydroxide (2 M, 8.2 mL/mmol) was added and the solution refluxed for 2 h before themethanol was removed in vacuo. The solution was acidified with 6 M hydrochloric acid andextracted with diethyl ether. The organic extract was washed with brine, dried overmagnesium sulfate and evaporated in vacuo. The residue was purified by distillation or flashchromatography on silica using diethyl ether/petroleum spirit as an eluent to afford the pureacid.(E)-2-Methyl-2-pentenoic acid (5)1.17 g crystalline solid (60%; bp 79-88°C / 0.7 mm Hg; lit. 106.5°C (Lucas and Prater,1937)); melted upon warming to room temperature.1H.NMR (200 MHz, acetone-d6)& 1.03 (3H, t, J = 8 Hz, H(5,E)), 1.17 (3H, t, J = 7 Hz,H(5,Z)), 1.64 (3H, d, J = 5 Hz, H(3’,Z)), 1.78 (3H, d, J = 1 Hz, H(3’)), 2.20 (2H, q of d, J = 8,8 Hz, H(4,E)), 2.46 (2H, m, H(4,Z)), 5.55 (1H, m, H(3,Z)), 6.75 (1H, t of d, J = 1, 8 Hz,H(3,E)). E.Z= 85:15.IR(neat,cm1):3200-3040, 2849, 1693, 1645, 1421, 1282, 1165.GC-MS (m/z, TMS ester): 186 (Mj, 171, 157, 127, 97, 73, 75.Anal. caic. forC6H1002.C, 63.14, H, 8.83, found. C, 63.15, H, 8.75.(E)-2-Ethyl-2-pentenoic acid (6)1.01 g clear oil (62%: bp 88°C / 0.5 mm Hg)1H-NMR (200 MHz, acetone-d6)& 0.89-1.08 (6H, m, H(4’), H(5)), 2. 10-2.38 (4H, m, H(3),H(4)), 6.70 (1H, t, J = 9 Hz, H(3)).44JR (neat, cm1): 3200-3040, 2964, 1688, 1642, 1457, 1418, 1288, 1255, 1166.GC-MS (m/z, tert-BDMS ester): 185 (M-57), 141, 111.Anal. caic. forC7H1202:C, 65.60; H, 9.44, found: C, 65.55; H, 9.38.(E)-2-Propyl-2-pentenoic acid (7)1.78 g crystalline solid (49%: purified by flash chromatography)mp 33-4CC1H-NMR (200 MHz, acetone-d6)ö: 0.90 (3H, t, J = 8 Hz, H(5’)), 1.04 (3H, t, J = 8 Hz, H(5)),1.3-1.4 (2H, m, H(4’)), 2.10-2.26 (4H, m, H(3’), H(4)), 6.75 (1H, t, J = 9 Hz, H(3)).JR (neat, cm4): 3200-3040, 2932, 1674, 1632, 1458, 1418, 1274, 1166.GC-MS (m/z, TMS ester): 214 (Mj, 199, 185, 169, 124.Anal. caic. forC8H1402:C, 67.57; H, 9.92, found: C, 67.74; H, 10.08.(E)-2-Butyl-2-pentenoic acid (8)2.74 g crystalline solid (76%: bp 135-139°C /4 mm Hg).1H-NMR (200 MHz, acetone-d6)ö: 0.90 (3H, t, J 7 Hz, H(6)), 1.05 (3H, t, J = 8 Hz, H(5)),1.20-1.47 (4H, m, H(4’), H(5’)), 2. 14-2.36 (4H, m, H(3’), H(4)), 6.75 (1H, t, J = 7.6 Hz, H(3)).JR (neat, cm4): 3200-3040, 2898, 1692, 1639, 1459, 1418, 1279, 1214, 1163.GC-MS (m/z, tert-BDMS ester): 213 (M-57), 179, 139, 105.Anal. caic. forC9H1602:C, 69.19; H, 10.32, found: C, 68.87; H, 10.33.452.2.7. 4(R)-(Methoxycarbonyl)- 1 ,3-thiazolidine-2-thione (9x)S?23—\FINySThe following esterification procedure was adapted from Brenner and Huber (1953)and the subsequent cyclization step from Hsiao et at. (1987).Thionyl chloride (5.0 g, 42 mmol) in chloroform (50 mL) was added slowly to drymethanol (80 mL) at room temperature. L-Cysteine (1.82 g, 15 mmol) was added graduallyand the solution then stirred for 20 mm before being evaporated in vacuo. The residue wasdissolved in methanol (15 mL) and evaporated again to afford crude methyl L-cysteinehydrochloride (2.79 g) as a sticky white foam. The ester was suspended in drydichioromethane (25 mL): triethylamine (1.87 g, 18.5 mmol) and carbon disulfide (1.20 g,15.8 mmol) were added, each as a solution in dichloromethane (1 mL), and the bright yellowsuspension stirred for 1 d at room temperature. The solvent was evaporated in vacuo and theresidue suspended in ethyl acetate (30 mL) and filtered through a 10 cm column of silica gel(60-120 mesh). The product was eluted with further portions of ethyl acetate (2 x 50 mL) andthe filtrate evaporated to afford a foul-smelling yellow oil (1.05 g) which was subjected toflash chromatography (1:1 ethyl acetate/petroleum spirit) to give pure 4(R)-methoxycarbonyl)-1,3-thiazolidine-2-thione (0.28 g, 11% overall) as a translucent oil.1H-NMR (200 MHz, acetone-d6)& 3.77 (1H, dd, J = 5, 11 Hz, CH2S), 3.80 (3H, s, -OCH3),4.01 (1H, dd, J = 10, 11 Hz, CH2S), 5.04 (1H, dd, I = 5, 10 Hz, NCH), 9.25 (1H, br s, NH).JR (neat, cm1): 3350-3114, 2954, 1710, 1453, 1264, 1200,1025.GC-MS (m/z): 124, 118, 78 (unidentified fragments). Distinct peak in total ion current scan.462.2.8. N-(3-Methylbutanoyl)-4(R)-(methoxycarbonyl)-. 1 ,3-thiazolidine-2-thione (lOx)To a solution of the thiazolidinethione 9x (1.00 g, 5.65 mmol) in dry dichioromethane(60 mL) at -78°C was added pyridine (0.50 g, 6.5 mmol), the cloudy suspension stirred 5 mmand then isovaleryl chloride (0.83 g, 6.8 mmol) added. The yellow suspension was stirred at-78°C for 0.75 h, following which the cold bath was removed and the suspension warmed toroom temperature over 2 h. The clear yellow solution was then diluted with dichioromethane(40 mL) and washed with water (10 mL), 5% aqueous oxalic acid (10 mL) and water (10 mL)before being dried over magnesium sulfate, filtered, evaporated in vacuo and purified by flashchromatography using diethyl ether/petroleum spirit (2:3, v/v) gave N-(3-methylbutanoyl)-4(R)-(methoxycarbonyl)- 1 ,3-thiazolidine-2-thione (1.13 g, 77%) as oily crystals.1H-NMR (200 MHz, acetone-d6)8: 0.96 (6H, d, J = 7 Hz, H(4), 11(4’)), 2.10-2.30 (IH, m,H(3)), 3.08 (1H, dd, J = 7, 16 Hz, H(2)), 3.25 (1H, dd, J = 7, 16 Hz, H(2)), 3.47 (1H, dd, J = 2,12 Hz, -NCHCH2S-), 3.80 (3H, s, -OCH3), 3.93 (1H, dd, J = 9, 12 Hz, -NCHCH2S-), 5.75(1H, dd, I = 2, 9 Hz, -NCHCH2S-).IR (CHC13,cm1): 2962, 1750, 1705, 1318, 1266, 1174.GC—MS: broad peak indicative of decomposition472.2.9. (Erythro)-N-[3-hydroxy-2-( I -methylethyl)- I -oxopentyl]-4(R)-(methoxycarbonyl)- 1,3-thiazolidine-2-thione (lix)To a solution of the imide lOx (1.00 g, 3.83 mmol) in dichioromethane (60 mL) at 0°Cwas added dibutylboron triflate (1.0 M in) dichloromethane, 4.45 mL, 4.45 minol). Afterstirring for 5 mm, triethylamine (0.49 g, 4.8 mmol) was added as a solution indichloromethane (1 mL) and stirring was continued for 30 mm at 0°C. The lime-greensolution was cooled to -78°C, propionaldehyde (0.31 g, 5.3 mmol) in dichloromethane (1 mL)was added and stirring continued for 30 mm at -78°C before the solution was allowed towarm to 0°C over 20 mm. Phosphate buffer (pH 7, 60 mL) was added and the mixture stirredvigorously for 3 mm prior to separation of the layers. The organic phase was separated, driedover magnesium sulfate, filtered and evaporated in vacuo. Flash chromatography (1:2-1:1diethyl ether/petroleum spirit) afforded starting material (0.35 g, 35%) and the adduct (0.65 g,53%) as a bright yellow oil.1H-NMR (200 MHz, acetone-d6)6: 0.96 (3H, t, J = 8 Hz, H(5)), 1.04, 1.05 (3H, 3H, d, d, J1 Hz, H(4’), H(4)), 1.10-1.60 (2H, m, H(4)), 2.21 (IH, septet of d, J = 7Hz, H(3’)), 3.36-3.39(1H, m, -NCHCH2S-), 3.40 (1H, d, J = 2 Hz, -OH), 3.50-3.65 (1H, m, H(3)), 3.80 (3H, s, -OCH3), 3.82-3.90 (1H, m, H(2)), 5.05 (1H, dd, J = 8, 8 Hz, -NCHCH2S-), 5.73 (1H, dd, I = 2,8 Hz, -NCHCH2S-).JR (CHC13,cm1): 3508, 2976, 1760, 1696, 1264, 1160.GC-MS: broad peak indicative of decomposition.482.2.10. N-(2-(Methoxypropenoyl))-2-( 1 -methylethyl)-2-pentenamide (12x)4,,CO2H3A solution of the adduct lix (480 mg, 1.5 mmol), triethylamine (0.17 g, 1.7 mrnol)and methanesulfonyl chloride (0.19 g, 1.7 mmol) in dry dichloromethane (12 mL) was stirred1.5 h at 0C before being evaporated, filtered with THF (12 mL) and stirred for 3 h at roomtemperature in the presence of DBU (0.91 g, 6.0 mmol). Petroleum spirit (12 mL) was addedand the mixture washed with 1 M hydrochloric acid (2 x 3 mL) and brine (3 mL). Thesolution was dried over magnesium sulfate, filtered and evaporated in vacuo prior topurification by flash chromatography using 1:1 diethyl ether/petroleum spirit (v/v) afforded240 mg (71%) of the amide 12x as a clear oil.1H-NMR (200 MHz, acetone-d6)ö: 0.95-1.04 (9H, m, -CH3), 1.63-1.92 (2H, m, H(4)), 2.05-2.11 (1H, m, 11(3’)), 3.88 (3H, s, -OCH3), 4.90-5.02 (1H, m, H(3)), 5.84, 6.50, (1H, 1H, s, s,C=CH2), 8.52 (1H, br s, NH).JR (CHC13,cm1): 3395, 2965, 1723, 1687, 1516, 1442.GC-MS (m/z): 225 (Mj, 210, 182, 154, 141, 125, Attempted preparation of methyl (Z)-2-methyl-2-pentenoate (15x)3,H34,5 34549To a solution of carboximide 2aa (0.60 g, 3.2 mmol) in dry dichloromethane (16 mL)at -78°C was added dibutylboron triflate (3.6 mL of 1 M solution in dichloromethane, 3.6mmol) and triethylamine (0.46 g, 4.5 mmol) and the resultant yellow solution stirred for 1 h at-78°C and 15 mm at 0°C before being re-cooled to -78°C and added through a double-tippedneedle to a solution of diethylaluminum chloride (9.7 mL of a 1 M solution indichloromethane, 9.7 mmol) and freshly-distilled propionaldehyde (0.28 g, 4.9 mmol) indichloromethane (12 mL) at -78°C. The resultant solution was stirred for 2 h at -78°C beforethe reaction was quenched by the addition of water (5 mL) and the solution allowed to warmto room temperature. The layers were separated and the aqueous phase re-extracted withdiethyl ether (40 mL). The combined organic extracts were washed with saturated sodiumcarbonate (8 mL) and brine (8 mL) prior to being dried over magnesium sulphate, filtered andevaporated in vacuo to afford the crude adduct (1.35 g). The hydrolysis and dehydration stepswere carried out as described above to afford, following flash chromatography using diethylether/petroleum spirit (1:40, v/v), the unsaturated ester 15x (200 mg, 49% from carboximide)as a clear oil.1H-NMR (200 MHz, acetone-d6)8: 0.92 (3H, t, J = 7 Hz, H(5)), 1.70 (3H, s, H(3’,E)), 1.75(3H, s, H(3’,Z)), 2.09 (2H, m, H(4,E)), 2.30 (2H, m, H(4,Z)), 3.56 (3H, s, -OCH3), 5.84 (1H, t,J = 7Hz, H(3,Z)), 6.60 (1H, t, J = 7Hz, H(3,E)). The ratio of E/Zsignals was 86: 1-Cyclopentenyl-1-carboxylic acid (12)CO2HTo a solution of ethyl (2-oxocyclopentyl)carboxylate (6.00 g, 38.4 mmol) in methanol(100 mL) at 0°C was added sodium borohydride (1.60 g, 42.3 mmol). The solution wasstirred for 1 h at 0°C, then water (30 mL) was added, followed by acidification with 6 Mhydrochloric acid. The solution was condensed in vacuo, extracted with diethyl ether (2 x50100 mL) and the extract washed with brine (20 mL), dried over magnesium sulfate, filteredand evaporated in vacuo to afford an orange oil, which was diluted with diethyl ether (80 mL)and washed with saturated aqueous sodium bicarbonate (2 x 10 mL) and water (10 mL)before being dried over magnesium sulfate, filtered and evaporated to afford a yellow oil(4.94 g). Owing to apparent decomposition during distillation in a previous experiment, thecrude material was used directly in the next (dehydration) step. Dry dichloromethane (100mL) was added, followed by triethylamine (3.48 g, 34.3 mmol) and methanesulfonyl chloride(3.94 g, 34.3 mmol), each as dichioromethane solutions (5 mL). The solution was stirred 1 hat room temperature, then evaporated in vacuo and filtered with THF (50 mL). DBU (5.22 g,34.3 mmol) in TI{F (10 mL) was added and the mixture stirred 1 d at room temperaturebefore being diluted with diethyl ether (100 mL), washed with 1 M hydrochloric acid (2 x 15mL) and brine (15 mL), dried over magnesium sulfate, filtered and evaporated in vacuo.Distillation afforded ethyl cyclopent-l-enyl-l-carboxylate (bp 82-105°C / 10 mm Hg; 3.02 g,56% overall) as a clear liquid.JR (CHC13,cm1): 2941, 1716, 1448.GC-MS (m/z): 140 (Mj, 112, 95, 67.Ethyl cyclopent-1-enyl-1-carboxylate (1.04 g, 7.43 mmol) was dissolved in 95%ethanol (30 mL). Aqueous potassium hydroxide (2M, 30 mL, 60 mmol) was added and thesolution refluxed 2 h before being cooled, acidified with hydrochloric acid and extracted withdiethyl ether (100 mL). The organic extract was washed with brine (2 x 15 mL), dried overmagnesium sulfate, filtered and evaporated to afford cyclopentenylcarboxylic acid as acrystalline white solid (0.80 g, 96%).mp 114-117°C (lit. 123-124°C (Philp and Robertson, 1978))1H-NMR (200 MHz, acetone-d6)& 1.82-2.00 (2H, m, CH2II), 2.40-2.56 (4H, m,CH2Ij),6.69-6.77 (1H, m, C=CH).51JR (CHC13,cm-1): 3040, 2950, 1691, 1621, 1428, 1288.GC-MS (rni’z, tert-BDMS ester): 169 (M-57), 125, 95.Anal. caic. forC6H802:C, 64.27; H, 7.19, found: C, 64.26; H, 1 -Cyclohexenyl- 1 -carboxylic acid (13)c2Xl-Cyclohexenyl-l-carboxylic acid was prepared by hydrolysis of methyl 1-cyclohexenyl-l-carboxylate: the crude product was purified by flash chromatography usingdiethyl ether/petroleum spirit (1:2, v/v) to afford the acid 1.98 g (87%) as a white crystallinesolid.mp 36-38°C (lit. 38°C (Boorman and Linstead, 1935))1H-NMR (200 MHz, acetone-d6)ö: 1.5-1.7 (4H, m, -CH7C2), 2.1-2.3 (4H, m, -CH2=),6.9-7.0 (1H, m, -CH=), 10.1 (1H, br s, -CO2H).IR (CHC13,cm): 3200-3000, 2936, 1686, 1644, 1425.GC-MS (m/z, tert-BDMS ester): 183 (M-57), 139, 109.Anal. caic. forC7H1002:C, 66.65; H, 7.99, found: C, 66.50; H, (±)-Methyl bicyclo [2.2.1 ]hept-3-oxo-2-carboxylate (17x)C02H352(±)-Methyl bicyclo[2.2. ljhept-3-oxo-2-carboxylate was prepared as described byMander and Sethi (1983) to afford the product 17x (5.39 g, 89%) as a yellowish oil.1H-NMR (200 MHz, acetone-d6)& 1.4-1.9 (6H, m, H(5), H(6), H(7)), 2.1-2.3, 2.4-2.6 (1H,1H, m, m, H(1), H(4)), 2.9 (1H, m, H(2)), 3.75 (3H, s, -OCH3).JR (CHCI3,cm-1): 3035, 2961, 1764, 1731, 1442, 1272.GC-MS: broad peak indicative of decomposition.2.2.15. (±)-Bicyclo[2.2. lJhept-2-ene-2-carboxylic acid (11)CO2HThe methyl ester 18x was prepared from 17x as described for ethylcyclopentenylcarboxylate. Distillation afforded a crude product (1.14 g, 36%, bp 93-123 °C/10 mm Hg) which was hydrolyzed in refluxing potassium hydroxide as describedabove to the acid (0.40 g, 14% from 17x, bp 132-135°C / 0.4 mm Hg) as a clear oil whichcrystallized upon standing.mp 30-32°C (lit. 21.5-22.5°C (Finnegan and McNees, 1964))1H-NMR (200 MHz, acetone-d6)8: 1.0-1.9 (6H, m, H(5), H(6), H(7)), 3.00 (1H, m, H(4)),3.20 (1H, m, H(1)), 6.92 (d, 1H, J = 3 Hz, H(3)).JR (CHC13,cm): 3038, 2948, 1691, 1597, 1411, 1283.GC-MS (m/z, tert-BDMS ester): 237 (M-15), 195, 167, 123.Anal. caic. forC8H1202:C, 69.55; H, 7.29, found: C, 69.74; H, 7.33.532.2.16. Cyclohexylidineacetic acid (15)The ester 20b was prepared according to Wadsworth and Emmons (1973) and thecrude product then hydrolyzed in refluxing potassium hydroxide as above to afford thecrystalline white acid (1.28 g, 46%) after decanting with petroleum spirit.mp 86-88°C (lit. 89°C (Wolinsky and Erickson, 1965))1HNMR (200 MHz, acetone-d6)6: 1.52-1.74 (6H, m, -CH2C), 2.2-2.3 (2H, m, -CH2=), 2.8-2.9 (2H, m, -CH2C=), 5.63 (1H, t, J = 1 Hz, -C=CH), 10.3 (1H, hr s, -CO2H).JR (CHCI3,cm1): 3200-3000, 2935, 2859, 1688, 1643.GC-MS (m/z, tert-BDMS ester): 197 (M-57), 153, 123.Anal. calc. forC8H1202:C, 68.55; H, 8.63, found: C, 68.45; H, Cyclopentylideneacetic acid (14)Cyclopentylideneacetic acid was prepared from cyclopentanone using the methoddescribed for cyclohexylideneacetic acid. The final product was purified by flashchromatography using 1:2 diethyl ether/petroleum spirit (v/v) to afford the acid (1.60 g, 84%)as a clear oil which crystallized upon standing.54mp 58-60°C (lit. 60-61°C (Weiland and Arens, 1956))1H-NMR (200 MHz, acetone-d6)& 1.5-1.7 (4H, m, -CH2C), 2.12-2.30 (4H, m, -CH2C=),6.95-7.05 (1H, m, -C=CH), 10.4 (1H, br s, -CO2H).JR (CHCI3,cm1): 3200-3000, 2911, 1688, 1644, 1424, 1283.GC-MS (m/z, tert-BDMS ester): 183 (M-57), 139, 109.Anal. caic. forC7H1002:C, 66.65; H, 7.99, found: C, 66.74; H, Cycloheptylideneacetic acid (16)Cycloheptylideneacetic acid was prepared from cycloheptanone using the methoddescribed for cyclohexylideneacetic acid. The final acid product was purified by flashchromatography using 1:2 diethyl ether/petroleum spirit (v/v) to afford the acid as a clear oilwhich crystallized upon standing.mp 50-2°C (lit. 54-55°C (Wolinsky and Erickson, 1965))‘H-NMR (200 MHz, CDC13)& 1.5-1.8 (8H, m, -(CH2)4), 2.40 (2H, hr t, J = 7 Hz, -CH2C=),2.85 (2H, hr t, J = 7 Hz, -CH2C=), 5.66 (1H, t, J = 1 Hz, C=CH-).JR (CHC13,cm1): 3200-3000, 2930, 1685, 1628, 1420.GC-MS (m/z, tert-BDMS ester) : 211 (M-57), 167, 137.Anal. calc. forC9H1402:C, 70.10; H, 9.15, found: C, 70.24; H, 9.18.552.2.19. Cyclooctylideneacetic acid (17)Cyclooctylideneacetic acid was prepared from cyclooctanone using the methoddescribed for cyclohexylideneacetic acid. The final product was purified by flashchromatography using 1:2 diethyl ether/petroleum spirit (v/v) to afford the acid as a clear oilwhich crystallized upon standing.mp 81-3°C (lit. 89-89.5°C (Wolinsky and Erickson, 1965))1H-NMR (200 MHz, CDC13)ö: 1.4-1.9 (1OH, m, -(CH2)5), 2.38 (2H, t, J = 7 Hz, -CH2C=),2.78 (2H, t, J = 7 Hz, -CH2C=), 5.78 (IH, s, C=CH-).JR (CHCI3,cm1): 3200-3000, 2913, 1687, 1626, 1450.GC-MS (m/z, tert-BDMS ester): 225 (M-57), 181, 151.Anal. caic. forC10H602:C, 71.39; H, 9.59, found: C, 71.33; H, Attempted preparation of cyclooctylidenepropanoic acid (23x)22x423x456Cyclooctylidenepropanoic acid was prepared as above except that triethyl 2-phosphonopropionate was used instead of the acetate. A sample of the intermediate ester 22was purified by flash chromatography using 1:24 diethyl ether/petroleum spirit (v/v).1H-NMR (200 MHz, acetone-d6)6: 1.0-1.2 (m, -CH2, -OCH2CkI3), 1.3, 1.5 (m, -CH2, -CHCjJ3), 1.74 (s, CH3=), 2.2, 2.3 (m, CH2=), 2.96 (q, J = 8 Hz, -CH-CH3), 3.8-4.2 (m, -OCH2CH3),5.40 (t, J = 8 Hz, C=CH). The and z34 isomers were present in a 3:1 ratio.GC-MS (m/z): two peaks, t, 5.25 mm (210 (Mj, 195, 182, 167, 153, 137, 109, 102, 95) and tr5.67 mm (210 (Mj, 195, 182, 165, 149, 136, 109) in approximately 1:3 ratio.The mixture of esters was hydrolyzed with refluxing potassium hydroxide as describedpreviously. The crude product was purified by flash chromatography using 1:2 diethylether/petroleum spirit (v/v) to afford a mixture of the acids 23x and 23y as a clear oil.1H-NMR (200 MHz, acetone-d6)6: 1.19 (3H, d, J = 7 Hz, -CH3 23y), 1.4-2.2 (m, -CH2),2.32, 2.60 (2H, 2H, m, m, CH2=, 23x), 3.12 (1H, q, J = 8 Hz, H(2), 23y), 5.52 (1H, t, J = 7Hz, H(4), 23y).IR (neat, cm1): 3050, 2920, 1691, 1605, 1453, 1409, 1280.GC-MS (m/z, tert-BDMS esters) : two peaks, tr 6.08 mm (239 (M-57), 195, 167, 153) and tr6.33 mm (239 (M-57), 165, 143, 95) in approximately 1:3 ratio.572.3. PHYSICOCHEMICAL PROPERTIESMolecular modelling was performed using MacroModel v. 3.5 on a Silicon GraphicsIRIS workstation. The input structure was minimized using the MM2* force field in anaqueous medium and the van der Waals volume calculated for this structure. A Monte Carloconformational search was then conducted for 500 conformations using a 10 kJ/mol energycutoff. The atomic Cartesian coordinate for each conformer were then transformed with acustom PASCAL program (written with the assistance of Mr. Roland Burton) so as to placethe carbonyl atom at the origin, C(2) on the x-axis and the C(2) substituent in the xy plane(Figure 34). The population-weighted mean internuclear dimensions, expressed as themaximal x, y or z values, were calculated based on a Boltzmann distribution at 37°C.The lipophilicity of most compounds were initially obtained by an HPLC method(Achaempong, 1985). Methanolic solutions of a series of compounds with previouslydetermined log P values (butyric, valeric, hexanoic, 2-ethylbutyric, 2-ethylhexanoic, VPA)were injected into an HP 1090 liquid chromatograph equipped with an Altex UltrasphereODS column and a 50 mM pH 3.5 phosphate/acetonitrile (54:46, v/v) mobile phase. Thecapacity factor for each was calculated and used to derive a calibration curve for the log Pvalues of 2-15.Lipophilicity was also calculated for the undissociated species using the CLOGPprogram (Daylight Chemical Information Systems, Irvine, CA, v. 3.54) by Dr. Eric Bigham atBurroughs Wellcome Co. (Research Triangle Park, NC). The program was not able todistinguish cis and trans isomers.2.4. ANTICONVULSANT POTENCYAnticonvulsant potency (Swinyard et al., 1989) was determined for each drug usingfive intraperitoneal doses (4 mL/kg) of its sodium salt and eight 6 week old mice per dose.Pentylenetetrazole (85 mg/kg, 10 mL/kg) was injected subcutaneously at 10 mm and the58animals observed for a further 30 mm. An animal was considered to be unprotected if itshowed a 5 s clonus with loss of balance. ED5O was determined from a graph of percentageprotection vs log(dose) following the method of Litchfield and Wilcoxon (1949), wherepercentage protection refers to the percent of animals in each dose group which wereprotected against seizures.2.5. MOUSE BRAIN DRUG DETERMINATIONThe frozen brain homogenate samples (500 ul: from homogenates used for GADassay) were thawed, mixed with sodium octanoate solution (500 ul, 22.4 nmol) and 1.5 Mhydrochloric acid (1000 ul) and 380 ul aliquots combined with ethyl acetate (2000 ul). Themixtures were vortexed thoroughly and then subjected to gentle rotation for 30 mm. Thevials were centrifuged (2060g, 10 mm) and a 1000 ul portion of the organic layer removedand dried over sodium sulfate for 10 mm. A fraction (800 ul) of the supernatant wasremoved, evaporated to approximately 100 ul under a flow of dry nitrogen in a 40°C waterbath and finally derivatized with 500 ul of an ethyl acetate solution containing 7%MTBSTFA, 0.07% tert-BDMSC1 (vlv) and 2.38 nmollmL cyclohexylcarboxylic acid for 1 hat 65°C. Upon cooling, the solutions were analyzed by GC-MS as described below.The corresponding calibration curve was prepared by combining 47.5 ul aliquots ofstock aqueous solutions of the sodium salts of each drug (0, 2.5, 12.5, 25, 50, 100, 200 and400 nmollmL) and octanoic acid (100 nmollmL) with control brain homogenate (95 ul) and1.2 M hydrochloric acid (238 ul). Ethyl acetate (2000 ul) was added and the samplesprocessed as described above.Recovery values were evaluated by running a second set of calibration standards todetermine the actual amount of drug in each of the above calibration samples. In this case,100 ul aliquots of solutions of each drug in ethyl acetate (0, 1.875, 3.75, 7.500, 15.00, 25.00and 50.00 nmol/mL) containing octanoic acid (17.02 nmollmL) were combined with the59MTBSTFA solution described above (500 ul) and heated for 1 h at 65°C prior to analysis byGC-MS.The acids in the above assays were analyzed by GC-MS as noted in Section 2.1 withthe following additional specifications. Oven: 50°C initial, increasing to 260°C @ 10°C/mmwith final hold time 5 mm. Equilibration time: 0.5 mm. MS: single ion monitoring at rn/z(M- 57) for tert-butyl(dimethylsilyl) esters of drug and the two internal standards. Astandard auto-tuning sequence was used.2.6. MEMBRANE FLUIDITYMembrane fluidity studies were initially performed using erythrocyte ghosts preparedby the general method of Dodge et al. (1963). Freshly collected human whole blood wascentrifuged (91 3g, 10 mm), decanted, the cells washed three times with phosphate-bufferedsaline (PBS: 10 mM sodium phosphate, 0.9% NaC1 (w/v), pH 7.4), lysed with hypotonicbuffer (10 mM sodium phosphate, pH 7.4), centrifuged (2060g. 20 mm), decanted, washedwith PBS and resuspended to afford a 10% ghost suspension. For the fluidity assay, the ghostsuspension was combined with PBS, membrane probe (DPH or TMA-DPH as THF-PBS orethanol-PBS solutions, respectively) and drug solution in PBS and incubated in the dark atroom temperature for 2 h. Final concentrations: ghosts 1.3%; probe 2 uM; drug 20 mM.Fluorescence was recorded at 20°C using a Shimadzu RF-540 polarizer-equippedspectrofluorophotometer. Excitation 355 nm (DPH) or 363 nm (TMA-DPH); emission 428nm; slit widths 10 nm. The parallel (I) and perpendicular (Ivh) fluorescence components foreach sample (five per drug) were measured three times without stirring and the fluorescenceanisotropy r calculated using corrections for scatter (4%) and grating polarization G (0.9757).Fluorescence anisotropy showed the expected temperature dependence in all preparations.Synaptic membranes were prepared as described by Jones and Matus (1974).602.7. GABA AND GAD ASSAYSSynaptosomes from mouse whole brain were prepared by a modification of themethod of Dodd et al. (1981). All steps were carried out at 0-4°C. For the assay of GABA,all sucrose solutions were 1 mM in sodium 3-mercaptopropionate (3-MP) to inhibit the largepost-mortem GABA increase (Loscher et al., 1981). Ten mice (five per assay) were injectedwith 4 mL/kg of either 0.9% saline or VPA as an aqueous solution of its sodium salt (0.83mmollkg), decapitated at 15 mm postdose into 0.9% saline, the brains were quickly removed,homogenized in 0.32 M sucrose (4 mL, pH 7.0), diluted with sucrose (2 mL) and centrifugedat 1,000g for 10 mm. A 500 ul sample was taken prior to centrifugation from each 3-MP-freehomogenate for subsequent drug analysis (Section 2.5). The supernatant was layered onto 1.2M sucrose (9.5 mL, pH 7.0) and then centrifuged (220,000g. &t = 1.6 x 1010 rad2/s): theinterface was collected by pipet, made up to 10 mL, layered onto 0.8 M sucrose (9.5 mL, pH7.0) and centrifuged as before. The supernatant was removed and the pellet resuspended ineither water (GAD assay, 5 mL) or buffer (GABA assay, 5 mL: 0.5 M KCI, 0.4 M sodiumphosphate, 10 mM EDTA, 0.5% Triton X-100, 47.4 nmol GABA-d6, pH 6.4, 5 mL)(Holdiness etal., 1981). The protein content of the synaptosomes was 19.6 ± 2.1 mg/g tissuefor the GABA assay (n = 10) and 16.3 ± 2.0 mg/g tissue for the GAD assay (n = 10), using amodified Lowry assay (Markwell et at., 1978). The GABA and protein assays wereconducted promptly, whereas the GAD assays were performed on samples stored at -78°C.In the GAD assay (Loscher, 1981), each synaptosome sample (50 ul) was incubated intriplicate for 1 h at 37°C with 50 ul of a solution containing glutamate (1 umol) and PLP (50nmol) in 0.1 M pH 6.4 sodium phosphate buffer. Final concentrations were 5 mM and 0.25mM for glutamate and PLP, respectively. For the blank (run in duplicate), the glutamatelPLPmixture was replaced with 50 ul of buffer. The reaction was quenched by the addition of1.7% trichioroacetic acid (53 ul) containing GABA-d6(660 pmol), the precipitated materialseparated by centrifugation (13,600g. 10 mm), the supernatant lyophilized at ambienttemperature and derivatized with an acetonitrile solution (300 ul, containing 30 ul MTBSTFA61and 0.3 ul tert-BDMSC1 (Mawhinney et al., 1986)) at 65°C for 1 h prior to GC-MS analysis.GAD activity was defined as the amount of GABA (per mg protein) in the sample incubatedwith glutamate and PLP less the amount obtained in the blank. In separate controlexperiments, the glutamate and/or PLP were omitted from the phosphate buffer solution. Ablank experiment (n = 4) was performed that retained the G1uIPLP mixture but replaced thesynaptosomes with an equivalent volume of water.Experiments designed to test the conditions of the above assays were also conductedusing the procedure described above except for the changes noted below. The samesynaptosome preparation was used throughout. The activity for each set of conditions wasmeasured in triplicate and then subtracted from an appropriate blank measured in duplicate.In the first set, the final concentration of PLP was fixed at 0.25 mM and final glutamateconcentrations varied as 10, 7.5, 5, 2.5, 1.25 and 0.5 mlvi. In the second, glutamate was fixedat 5 mM and PLP varied as 0.5, 0.375 and 0.125 mM. Finally, the assay was conducted usingthe standard glutamate and PLP concentrations for 5, 10, 20, 30, 45, 60, 90 and 120 mmtimepoints.The dependence of activity on protein concentration was determined from asynaptosome pellet which that been resuspended in 2.5 mL, rather than the usual 5 mL, ofwater. In addition to an undiluted sample, the GAD assay was then run with samples diluted1:0.5, 1:1, 1:2, 1:3, 1:4 and 1:5 (v/v) with water.In the brain GABA assay, run in triplicate for each synaptosome preparation,resuspended synaptosome samples (50 ul) were acidified with 1.7% trichioroacetic acid (60ul) and centrifuged (13,600g. 10 mm). An aliquot (80 ul) of the supernatant was thenremoved, lyophilized and derivatized as described above. A blank experiment (n = 5) wasconducted where the synaptosomes were replaced with the resuspension buffer.The GABA contents of the above assays were analyzed by capillary GC-MS as notedin Section 2.1 with the following additional specifications. Oven: 50°C initial, increasing to130°C @ 30°C/mm then to 200°C @ 10°C/mm and finally to 280°C @ 30°C/mm, with a finalhold time of 2 mm. Equilibration time: 1 mm. MS: single ion monitoring at m/z 27462(GABA) and 280 (GABA-d6). The tune was optimized for m/z 264 with the high-energydynode set at 10 kV.Individual standard curves were prepared for the GABA and GAD assays usingaqueous solutions of GABA and GABA-d6 containing the appropriate amounts of buffer andtrichioroacetic acid. These solutions were evaporated to dryness and derivatized as above.For the GAD blank and the GABA assay, a range of 400-1900 pmol GABA was used, while arange of 3-20 nmol was employed for the GAD assay.2.8. GAD INHIBITION ASSAYThe inhibition of GAD activity by 17 in vitro was studied by incubation of frozensynaptosomes from saline-treated mice with varying concentrations of the drug. An aliquotof synaptosomes (50 UI) was mixed with a substrate solution containing 17 (finalconcentrations: 0, 1, 5 mM), glutamate (final concentrations: 1, 3, 5, 10 mlvi) and PLP (finalconcentration: 0.25 mM) and incubated at 37°C for 10 mm prior to being quenched andprocessed as described in Section 2.7. For each set of conditions (n = 3), a blank (n = 2)containing synaptosomes and buffer only was also run and subtracted from the former toafford the enzyme activity. A double-reciprocal plot of activity vs. glutamate concentrationwas then constructed to obtain the various kinetic parameters.2.9. GAD SATURATION ASSAYThe GAD saturation assay was based on procedures described by Martin (1986) withthe modifications of Nathan et al. (1994). All operations were carried out at 0-4°C. Maturemale CD-i mice were given saline or an ED5O dose of VPA, 7 or 17 and decapitated at 15mm postdose into ice-cold 0.9% saline. The cranium was quickly exposed underwater andthe head cooled briefly. The brain was removed, homogenized and processed as described forsynaptosome preparation except that all sucrose solutions contained 5 mM ATP rather than 3-63MP. The final pellet was resuspended in a mixture of 5 mM ATP (1 mL, pH 6.5) and 80 mMIm/Ac buffer (1 mL, 80 mM 1:1 imidazole/acetic acid, 4 mlvi 2-aniinoethylisothiouroniumbromide, 2 mM EDTANa4,1% Triton X-100, pH 6.5 with concentrated NaOH). A 500 ulaliquot was applied to a column of Sephadex G-25 (medium, 4.0 g packing, 20 mL totalvolume) and eluted at 0.2 mL/min with 40 mM Tm/Ac buffer (a 1:1 (v/v) dilution of the 80mM buffer with water). The protein fractions were combined and 150 ul aliquots mixed witheither pyridoxal phosphate (25 ul, 0.8 mM, n =4) or buffer (n = 6). Following pre-incubationfor 5 mm at 37°C, glutamate (25 ul, 8 mM) was added to all except two control tubes, wherebuffer (25 ul) was added. The samples were incubated for a further 5 mm at 37°C and thenquenched (50 ul, 6% trichloroacetic acid, 8.78 uM GABA-d6)and centrifuged (1 3,600g, 10mm). The supematant (230 ul) was removed, lyophilized, treated with MTBSTFA solution(400 ul, 10% MTBSTFA and 0.1% tert-BDMSC1, v/v, in acetonitrile) at 65°C for 1 h andanalyzed by GC-MS as described before. Controls were also run using the pyridoxalphosphate and glutamate solutions individually and in combination (n = 3 each). Calibrationstandards were made up with the appropriate amounts of buffers, trichioroacetic acid andGABA-d6and treated identically as the samples.Synaptosome GAD activity was determined using samples frozen for several days at -78°C, with the assay being performed as described in Section 2.7 except that the 0.1 Mphosphate buffer was replaced by 40 mM Tm/Ac buffer containing 5 mM ATP. All proteinconcentrations were measured using the method of Markwell et al. (1978).643. RESULTS3.1. SYNTHESISThe initial choice of 2-ene VPA analogues for study was a series of (E}.2-alkyl-2-pentenoic acids (5-8). Their preparation was based on the highly stereoselective addition ofpropionaldehyde to dibutylboryl enolates derived from racemic N-acylisopropyloxazolidinones (Figure 6) using the procedure described by Evans et al. (1986). The carboximides2aIa-f were first prepared according to Evans et al. (1981) from the oxazolidinone lasynthesized by a modification of the method of Newman and Kutner (1951). The adducts 3awere then isolated and characterized by 1H-NMR and by GC-MS of their silyl ethers, thespectra being consistent with the presence of a pair of presumably erythro enantiomers. Notethat all structures in Figure 6 are racemic. This finding was supported by the fact that all ofthe subsequent unsaturated acids 5-8, where the double bond had been formed by astereospecific E2 elimination mechanism, were found to exist predominantly as their Eisomers, based on the chemical shift 6 of their H(3) 1H-NMR signals. Specifically, 6(H(3))for our preparation of acid 7 agreed well with not only the published value but also with thosefor the other acids 5, 6 and 8 (see Figure 14 for a full list of final structures).Hydrolysis of the adducts 3a with lithium hydroperoxide (Evans et al., 1987) followedby treatment with diazomethane gave high yields of the methyl hydroxyesters 4a-fHowever, the branched side-chain adducts 3ae and 3af hydrolyzed very slowly (isopropyl:29%; cyclopentyl: 15%) even when the reaction was conducted at room temperature. Theprobable explanation is the steric crowding in the region of the “amide” carbonyl by thesegroups. Assuming that the dibutylboryl enolate Z stereochemistry alone would be sufficientto direct the reaction to the erythro adduct (Evans et al., 1982), we synthesized the isopropylsubstituted adduct 3be derived from 2-oxazolidone lb itself. The stereoselectivity of theprocess had not been compromised, as indicated by1H-NMR spectroscopy (Figure 7), but theoverall conversion of carboximide 2be to hydroxyester 4e proceeded in 70% yield. A similaryield was obtained with the cyclopentyl analogue 2bf. This indicates that the stereochemical653aa: X = i-Pr; R = Me3ab: X = i-Pr; R = Et3ac: X i-Pr; R = Pr3ad: X = i-Pr; R = Bu3ae: X = i-Pr; R i-Pr3af: X = i-Pr; R = cyclopentyl3be: X = H; R = i-Pr3bf: X = H; R = cyclopentyl2aa: X = i-Pr; R = Me2ab: X = i-Pr; R = Et2ac: X = i-Pr; R = Pr2ad: X i-Pr; R = Bu2ae: X = i-Pr; R = i-Pr2af: X = i-Pr; R = cyclopentyl2be: X = H; R = i-Pr2bf: X = H; R cyclopentyl4a: R = Me4b: R=Et4c: R = Pr4d:R=Bu4e: R = i-Pr4f: R = cyclopentyl5a: R = Me5b: R=Et5c: R = Pr5d: R = Bu5e: R=i-Pr5: R=Me6: R=Et7: R = Pr8:R=Bux1) Bu2OTf, Et3N, -78°CxHN 01)BuLi2) RCOC1la: X=i-Prib: X = Hx*2) CH3HO3) 11202, 1120RLiOOHNyOTf-H20,0°COH 0 0(±)C112N(±)1) MsC1, Et3N2) DBU, AOCH3 KOHaq. CH3O , A0HFigure 6. Synthesis of 2-substituted 2-pentenoic acids.T1 C C N z 043IIIIIIjIIIIIIIILFJJTTjTT’TT210PPMtJ 91+a0 CD11 I’I JUL67control of the reaction is determined by the dibutylboryl enolate geometry rather than thestructure of the carboximide, allowing for the use of inexpensive 2-oxazolidinone rather thancostly 4-isopropyl-2-oxazolidinone.The subsequent dehydration” step proved to be surprisingly difficult with thesaturated analogues although many earlier reactions with the aldol adducts themselves as wellas with precursors of the bis-unsaturated acids 9 and 10 invariably proceeded in high yieldsusing 1.1 equivalents of all reagents (triethylamine, methanesulfonyl chloride and DBU) andstirring overnight at room temperature (Lee, 1991). The hydroxyester 4c, however, did notshow more than approximately 50% conversion (as judged by GC-MS) by this protocol, evenwhen excess DBU was added and the solution refluxed for several hours. It was found thatusing 4 equivalents of all reagents and conducting the elimination step under reflux for 1 hafforded reasonable yields of esters 5a-d, although only trace amounts were obtained for Se.This is consistent with the conditions reported elsewhere for the elimination of mesylatesleading to isolated double bonds (Williams and Maruyama, 1987). All products were isolatedas single isomers, indicating that the integrity of the E2 elimination was maintained at theelevated temperature. The final acid products were then obtained by basic hydrolysis assingle isomers with the exception of 5, whose1H-NMR spectrum indicated the presence oftwo isomers in an 85:15 (E:Z) ratio.Unfortunately, TLC and GC-MS analysis of the dehydration step for the cyclopentylester 4f indicated no conversion, probably due to excessive steric congestion. In view of thisresult and the poor yield obtained for 5e, an alternative approach seemeddesirable.The dehydration step could be performed on the adduct itself, which had previouslybeen shown to be a very facile process, and then hydrolyzing this species to our final product,the 2,3-unsaturated acid (Figure 8). However, such a hydrolysis reaction would involve aneven less electrophilic ‘amide” carbon atom, owing to its conjugation with the double bond,thus making the desired hydrolysis of the branched side-chain adducts even less favourable.One solution to this problem was the use of an oxazolidinone-related auxiliary group thatwould likewise allow formation of the dibutylboryl enolate and the aldol addition to68R RR1) MsCI, Et3NLY__________xaddition 2) DBU0 OH 00RHydrolysis0Figure 8. Envisaged alternative pathway to final acidproduct.CO2H3 CO2H31) Pyñdine, CH21__1) Bu2OTf, Et3N, -78°C2) (CH3CHCHO 12) CH3HOS 0 S9x lOx(9% from L-cysteine) (77%)C02C1131) MsC1, Et3N 12x (71%) 0 CO2H32) DBU, 20°C, 3 hOH 0 SCO2H3lixi2y 0 sFigure 9. Thiazolidinethione-based approach to2-substituted-2-pentenoic acids.69proceed with high selectivity and facilitate the formation of the 2,3-unsaturated adduct, butwould simultaneously be much more susceptible to removal by basic hydrolysis. Thethiazolidinethiones reported by Hsiao et at. (1987) appeared to meet these requirements,leading to the sequence of reactions described in Figure 9. Although the dehydration reactionproceeded quantitatively at room temperature, the major product was apparently the di-unsaturated species shown, indicative of a facile elimination of CS2. Further studies todetermine if the mesylate alone could be selectively eliminated were not performed.We briefly investigated the feasibility of the threo-selective aldol addition, describedby Walker and Heathcock (1991), using the 2-methyl carboximide 2aa. The aldol addition ofthe carboximide to diethylaluminum chloride-complexed propionaldehyde, along with thesubsequent steps (Figure 10), proceeded in reasonable yields but the1H-NMR spectrum of thefinal unsaturated ester 15x revealed an 86:14 mixture of isomers. Because the dehydrationwas stereospecific, as shown previously by the presence of only one set of signals in the 1H-NMR spectrum of the ester 4a prepared by the route described in Figure 6, this ratiocorresponded to the degree of stereoselectivity of the aldol addition itself.To further study the effects of chemical structure on anticonvulsant activity, we alsosynthesized the acids 11-13 shown in Figure 11. The norcamphor derivative was prepared bythe methoxycarbonylation of norcamphor (Mander and Sethi, 1983) followed by reductionwith sodium borohydride, dehydration and hydrolysis. A similar approach was used for 12and 13 except that the starting materials were the commercially available ethylcyclopentanonecarboxylate and ethyl cyclohexenylcarboxylate, respectively.Compounds 14-17, where the double bond is exocyclic to the ring, were prepared byreaction of the corresponding ketone with the sodium salt of triethyl phosphonoacetate(Wadsworth and Emmons, 1973) followed by hydrolysis of the resultant ester (Figure 12).We also attempted to prepare the acid 23x using an identical approach with triethylphosphonopropionate. Unfortunately, the 1H-NMR spectra of both the ester and the finalacid, even after purifications by flash chromatography, indicated inseparable 3:1 mixtures ofregioisomers 22x/22y and 23x123y, respectively (Figure 13).701) Bu2OTf, Et3N 2)CH3HO-EtAICICH21,-78°CNOHO0 0 B”°2aa /“ IBu Bu13xLiOOH CH2N 1) MsC1, Et3N2) DBUOH 0 0 015x14x E:Z = 86:14(49% from 2aa)Figure 10. Attempted preparation of methyl(Z)-2-methyl-2-pentenoate.1) LDA, THF, -78°C NaBHCH3O2) CH3OCOCN, HMPACO2H316x 17x(89%)1) MsC1, Et3N KOH2) DBU CO2H3 aq. CH3O CO2H18x 11(36%) (14%)Figure 11. Preparation of acid 11.71CO2Et CO2H06 (EtO)2P(O)CWCOtNa KOHC6H aq. EtOH, ‘I19a:n=1 20a:n=1 14:n=119b:n=2 20b:n=2 15:n=219c:n=3 20c:n=3 16:n=319d:n=4 20d:n=4 17:n4Figure 12. Preparation of cycloalkylideneacetic acids 14-17.rNa CO2Et CO2EtL OEt+ 22yc5 EtO—P C6H21xIKOH aq. EtOH,C0211 CO2H23x + Q 23y3:1Figure 13. Attempted preparation of 2-cyclooctylidenepropanoic acid 23x.72CO2H CO2HfCO2H1 2 31C02H CO2H CO2H4 5 6COH CO2H CO2H7 8 9CO2HCO2HCO2H1210 11CO2HçO2H1314 1516 17Figure 14. Compounds selected for evaluation.733.2. ANTICONVULSANT ACTIVITY EVALUATIONCompounds 2-17 were evaluated for anticonvulsant activity by subcutaneous injectionof 85 mg/kg PTZ 10 mm following Lp. administration of each drug as its sodium salt inaqueous solution. The criterion for a convulsion was a tonic-clonic seizure lasting severalseconds accompanied by a loss of balance. Administration of PTZ alone produced seizures in87% of a group of 48 mice, with a mean latency period of 7.2 ± 4.3 mm and a median periodof 6 miii. Using the method of Litchfield and Wilcoxon (1949), the percentage protection wasplotted against the log dose to obtain the dose corresponding to 50% protection (ED5O), asshown in Figures 15-30 and summarized in Table 5. These graphs show all data collected,with outlying points not used in the construction of the regression line depicted as emptycircles. Error was determined from the 95% confidence interval of each plot. Sedation wasevaluated qualitatively by the degree of activity and muscle tone present during handlingwhen the mice were administered PTZ. “No sedation” (-) implies normal activity anduncooperative behavior whereas moderate (+) to high (+i-) sedation describes states where theanimal lies increasingly limp and motionless (Swinyard et al., 1989). This parameter isobviously subjective and should be considered only in a most qualitative way. More reliabletests, such as the Rotorod, were not available.Examination of Table 5 reveals that potency increased with molecular weight.Although this trend was less evident with the most potent drugs, the highest molecular weightdrug (17) was also the most potent. The sedative potencies of the drugs also followed size butthe effect was much more noticeable for the acyclic drugs because, with the exception ofcompound 17, all cyclic compounds showed no sedative effects at the ED5O dose. Thisproperty was exploited in compounds 16 and 17, which preserved or even exceeded thepotency of 2-ene VPA while remaining free of substantial sedation.As has been noted earlier by Lee (1991), compound 9 ((2E,3’E)-diene VPA) induced aprofound neurotoxicity characterized primarily by rigidity. This was in marked contrast tothe ataxia and decreased muscle tone observed as the sedative effects of the other drugs.74100— II I75- /..• -i::0 I I I3.6 3.7 3.8 3.9 4.0 4.1 4.2log (dose [umo]IkgJ)Figure 15. Determination of ED5O dose for acid 2.100 II II ::fCO2H0 I I I3.75 3.80 3.85 3.90 3.95 4.00 4.05log (dose [umollkg])Figure 16. Determination of ED5O dose for acid 3.75100I75- ./• -.50- -t .-.•CO2Ho . 425 — ./o I3.5 3.6 3.7 3.8 3.9 4.0log (dose [umollkgj)Figure 17. Determination of ED5O dose for acid 4.1J0 — II I I —- .-‘::0 I .r I I II I3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4.0log (dose [umollkg})Figure 18. Determination of ED5O dose for acid 5.761001 S‘75 -‘ .. . /-CO2H0 1 I2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6log (dose [umoilkgj)Figure 19. Determination of ED5O dose for acid 6.100— II I I:75- .‘ .... -I::...............,.....CO2H0 I I 12.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4log (dose [umollkgj)Figure 20. Determination of ED5O dose for acid 7.77100 —I I I I C!?.75- ..;• •/ -50-0 .• • -I./// 8CO2H2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4log (dose [umollkgj)Figure 21. Determination of ED5O dose for acid 8.100I I I I I —75 - .. .. -50-0 - CO2H0: 925 -. -0 I I I: I II2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4log (dose [umollkgj)Figure 22. Determination of ED5O dose for acid 9.78—I I ...... I —75......../‘ :: - 10I I I2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3log (dose [umollkg])Figure 23. Determination of ED5O dose for acid 10.03.2 3.802Hlog (dose [umollkg})Figure 24. Determination of ED5O dose for acid 11.79100I CO2Hlog (dose [umollkgj)Figure 25. Determination of ED5O dose for acid 12.100I I I I75 - -.CO2HI:::0__• I2.9 3.0 3.1 3.2 3.3 3.4 3.5log (dose [umollkg])Figure 26. Determination of ED5O dose for acid 13.801cHI I I ISCO2H2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8log (dose [umollkg})Figure 27. Determination of ED5O dose for acid 14.100— I I I I•,.” I75- -/ CO2H152.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5log (dose [umollkg])Figure 28. Determination of ED5O dose for acid 15.81iE—I I I I I —75- ...../ -/ .. CO2H50-25- •..- 160 I :1 I I I I2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3log (dose [umollkgj)Figure 29. Determination of ED5O dose for acid 16.100—I I ••,..I’ —75- .... -/ ..../ 170•’• I I I2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2log (dose [umollkg])Figure 30. Determination of ED5O dose for acid 17.82Table 5Anticonvulsant and sedative properties and brain concentrations of VPA and its analoguesCompound ED5O 95% Brain Sedation’(mmol I kg) confidence interval concentration(mmol / kg) (nmol I g tissue)a1 0.83C 0.48- 1.14 617± 239 +2 6.0 5.2 - 6.5 2222 ± 943 -3 7.0 5.6 - 9.8 4460 ± 842 -4 5.2 3.5-6.1 27lO±594 -5 4.4 2.4-17 1520±468 -6 1.6 0.68-1.75 664±41 -7 0.84 0.44-1.15 261±71 +8 0.87 0.59 - 1.00 144±549 1.4 0.46 - 1.65 684±95e10 0.78 0.26 - 1.66 348 ± 44d-11 2.9 1.2-3.8 2l50±326d -12 4.7 2.3-5.5 49680 -13 2.2 1.6-2.5 1210±243 -14 2.3 1.3-3.7 455±6ld -15 1.4 0.77-1.9 320± 105 -16 0.96 0.73 - 1.3 159 ± 31d-17 0.73 0.65 - 0.80 94± 37 +a) [DrugJj: mean ± SD drug levels at 15 mm postdose in whole brain homogenates fromfive mice given an ED5O dose and sampled in triplicate (Section 3.3)b) -: no sedation; +: some loss of muscle tone; ++: significant loss of muscle tone;c) W. Tang, personal communicationd) As in a), but four brain samples only: a single deviant value was omitted if the resultantSD was decreased by more than 50%e) Animals exhibited severe ataxia and rigidity rather than sedation3.3. BRAIN CONCENTRATIONS AT ED5O DOSEThe concentration of each drug at 15 mm following an ED5O dose was determinedfrom samples of frozen whole brain homogenates (obtained during the assay of synaptosomalGAD activity: Section 3.9.5) as summarized in Figure 31. The final results are included inTable 5 in the previous section for ease of comparison with the ED5O data. Octanoic acidwas used as an internal quantitation standard at a concentration comparable to that of VPAitself. The final extracts were treated with MTBSTFA to produce the acids as tertbutyl(dimethylsilyl) esters (Figure 32: Abbott et al., 1986) characterized by strong M - 5783Control brain Brain homogenate from 5 drug-dosed micehomogenate1 (each sampled in triplicate)I + sodium octanoatespike with drug as Na salt 4’+ hydrochloric acidvortex+ ethyl acetatea) vortex b) gentle rotation (30 mm)4centrifugeIremove 1 mL of organic layerdry over sodium sulphateIevaporate under N2 to 100 ulSolutions of drugin ethyl acetate2 add 7% MTBSTFA in ethyl acetate365 °C/lhJrGC-MS analysis1) Standards for quantitation calibration curve2) Standards for recovery calibration curve3) Containing cyclohexylcarboxylic acidFigure 31. Quantitation and recovery determination of drugs in brain homogenate.840 CHCF3Ni—C H3)ICH3 00 CH3ROH EtOAc, 65°C, 1 hRH30—i—C(CH)CH3Figure 32. Derivatization of carboxylic acids with MTBSTFA.Abundance a Abundance b Abundance C—I —200000 -- 900000- 500000-180000-800000- -160000- 400000-700000-600000- -120000- 1 300000-500000-100000-400000-80000- : 200000300000-60000-200000-40000-20000- in I• 4 I --• c-,J II -.9,5 10.0 10.5 10.511.0 9.5 10.0Time (mm)Figure 33. SIM chromatograms of MTBSTFA-treated 2-ene VPA (a) and internal standardsoctanoic acid (b) and cyclohexylcarboxylic acid (c).85peaks, where the molecular ion had lost the labile tert-butyl substituent. Despite the similar oreven identical masses of the internal standards (octanoic acid and cyclohexylcarboxylic acid)to some drugs, there were no problems with interference or inadequate sensitivity. TypicalSIM chromatograms are shown in Figure 33.The brain concentrations of all drugs, with the exception of compounds 2 and 4, fellwithin the boundaries of the calibration curves. The brain concentrations of these drugs wereestimated by extrapolation from the standard calibration curve. However, the overall error islikely minimal in view of both the small extrapolations from highly linear standard plots(Table 6) as well as the fact that the primary source of variation for these or any other drugswas between the individual brain samples. The latter was the reason that the full set ofstandards for some drugs was not always used (Table 5). Values from a single sample wererejected whenever their omission produced a greater than 50% reduction in the overallstandard deviation.The number of data points used in the calibration curves is variable in order tominimize skewing. For example, calibration curves for compounds present in the brain athigh concentrations omit the lowest concentration standards and vice versa.Table 6Brain concentration assay parametersCompound Calibration curvea n r1 y = -0.0 1827 + 0.6528x 7 0.999612 y=0.01213+0.1094x 4 0.998983 y=0.01703+0.1055x 4 0.999424 y = 0.007725 + 0.1321x 4 0.99999S y = 0.0006771 + 0.1 842x 5 0.999736 y=-0.013828+0.3913x 6 0.999847 y = 0.001472 + 0.2369x 6 0.999548 y=0.03999+0.2007x 6 0.999799 y—_-0.01652+0.1356x 6 0.9999610 y=-0.001103+0.1457x 5 0.9999911 y=0.006322+0.07l5x 4 0.9996312 y=-0.006050+0.1467x 5 0.9998313 y=0.006177+0.2134 5 0.9995414 y=0.01829+0.1918x 4 0.9998315 y=0.008353+0.2028x 4 0.9987216 y=-0.01481+0.2441x 5 0.9996186Table 6 (continued)Brain concentration assay parametersCompound Calibration curvea17 y = -0.00956 1 + 0.3071x 6 0.99892a) Where: y = area ratio [(drug M - 57) / (octanoic acid M - 57)] and x = nmol drugAlthough the quantitation protocol clearly compensated for losses, it was stillimportant to estimate recovery in order to demonstrate the validity of the method. Thus, theactual amount of drug present in the final MTBSTFA solutions of the quantitation calibrationcurve standards was determined using a second set of drug standards directly dissolved inethyl acetate and treated with MTBSTFA. For each quantitation calibration curve standard,the drugts SJ..M chromatogram peak area, expressed relative to the area of the commoninternal standard cyclohexylcarboxylic acid, was divided by the corresponding ratio expectedbased on the recovery calibration curve. This quotient represents the recovery. Theindividual recovery values are presented in Table 7.Drug(pmol) 1 2 3 419,000 68.1 93.1 81.6 98.39500 72.9 89.0 94.6 93.04750 74.7 97.3 95.1 95.42375 66.5 94.1 96.3 94.81188 37.2 83.4 101.4 93.9594 60.9 113.9 89.0297 94.7 133.0 83.538,000 90.4 81.719,000 98.2 87.1 81.3 93.99,500 103.2 95.2 82.0 105.74,750 105.7 99.8 95.9 88.72,375 113.8 89.1 82.9 91.61,188 107.9 93.8 81.7 102.3594 113.7 87.9 78.3 114.3297 118.6 141.3n rTable 7Drug recovery as a function of amount per vialaRecovery (%)5 6 7 887Table 7 (continued)Drug recovery as a function of amount per vialaDrug Recovery (%)(pmol) 9 10 11 1219,000 111.5 87.9 90.2 82.69,500 112.7 73.9 89.1 81.74,750 107.9 84.9 99.1 85.32,375 109.2 86.3 98.6 78.41,188 112.6 84.0 88.7 93.3594 130.3 101.7 107.9 78.0297 156.3 137.5 134.1 111.013 14 15 1619,000 100.4 107.6 101.5 99.79,500 106.9 113.2 114.1 104.44,750 100.9 97.9 (240.3) 108.72,375 93.4 99.9 109.3 110.41,188 94.5 104.4 117.6 94.2594 115.8 116.5 170.4 104.7297 38.5 146.7 163.8 126.91719,000 122.99,500 115.44,750 132.92,375 117.71,188 119.4594 136.4297 159.9a) Approximately 8.7 ug of brain tissue (as homogenate in 0.32 M sucrose) was added toeach vial.As shown, the recovery values were predominantly 80% or better, but there are a fewcases where the values were consistently above 100%. Considering that the signal from theaqueous blank standard was invariably non-zero, this phenonomenon was likely the result ofinterference from endogenous compounds.Having both the ED5O values as well as the resultant brain concentrations allows foran estimate of the relative ability of the drugs to penetrate into the brain at their therapeuticdose (Table 8). This parameter Q can be considered as a distribution ratio between the brainand overall tissue compartments under the non-steady-state conditions of the experiment.88Q = Brain concentration (umol I g wet brain tissue) I ED5O (umol I g body weight) (1)Table 8Brain/body distribution ratio QCompound Q1 0.742 0.373 0.644 0.525 0.356 0.417 0.318 0.179 0.4910 0.4511 0.7512 0.1113 0.5514 0.2015 0.2316 0.1717 0.133.4. MOLECULAR MODELLINGThe objective of molecular modelling using the MacroModel program was todetermine the structures in an aqueous solution of not only the most stable conformation ofeach molecule but all stable conformations whose energies were within a certain range of theglobal minimum. Using the initial coordinates provided by the user, the relative energy forthe input structure was calculated using the MM2* force field (Mohamadi et al., 1990), whichis simply an equation expressing the potential energy of the molecule as a function of thepositions of its constituent atoms. It contains terms describing the vibrational, torsional,bending and van der Waals and Coulombic interaction potential energy contributions. Thesolvation energy of the molecule was determined by treating the surrounding medium as acontinuum, or shell, beginning near the van der Waals surface of the molecule. The solvationenergy was then calculated as the sum of the van der Waals solute-solvent interaction energyand the electrostatic polarization.89The computer program then rotated the specified bonds until it found the nearestpotential energy well, or the “local minimum”. This was done by calculating the energy ofthe current conformation, shifting the atomic coordinates in a systematic way and thenrecalculating the energy. By taking the first derivative of the potential energy with respect tothe coordinates, the program determined how to move the individual atoms so as to minimizethe energy. When this derivative equalled zero, a local minimum had been found.However, a typical molecule will have many such minima across its conformationalspace and there is no guarantee that this particular one is the “global minimum”. A moreexhaustive search of the entire conformational space is thus necessary in order to find thelowest possible energy. Furthermore, we still required knowledge of all other conformationswithin a reasonable separation from the global minimum because the free energy of a drug’sbinding to a receptor is greater than the free energy difference between similar conformers(Crippen, 1979). In other words, the actual pharmacophore’s energy can be well above that ofthe global minimum, while still retaining a favourable overall free energy change uponreceptor binding.If the inter-conformational energy and structural differences are relatively small, amolecule will possess many conformations within a reasonable energy limit of the globalminimum. Consequently, we chose to estimate effective solution structure dimensions bytaking a weighted mean of all conformers within 10 kJ/mol of the global minimum. Thisenergy range was chosen in order to retain at least 95% of the conformations of a givenmolecule while keeping their total number within a reasonable limit.The actual conformational search was conducted using the Monte Carlo routine (seeChang et al., 1989), which generates conformations in a fairly random manner and thenminimizes the energies of these structures as described above to evaluate the entire potentialenergy hypersurface. The Monte Carlo routine began with the structure corresponding to theinitial minimum. Several bonds capable of exhibiting independent torsions were thenselected. Because this number usually exceeded the limits of the computer, the programrandomly chose a set from this group for each iteration. Each of these chosen bonds was then90varied by some random amount between 0 and 180 degrees. The program checked that theresultant structure was plausible, in terms of non-bonded interactions and ring closures, andthen proceeded with the energy minimization routine to find the local minimum. Once theminimum had been reached, the new structure was checked to ensure that its energy,interatomic distances and torsional angles were within the specified limits, compared withstored structures from previous iterations and used for the next cycle. The number of thesecycles required will clearly vary with the size of the molecule and the number of possibleminima. As the process is a random one, a more extensive search will not only improve thechances of sampling the entire hypersurface but will also increase the number of times a givenconformation is found, thus ensuring that the conformation is in fact valid. In our searches, a500-iteration paradigm was used, with a structure being considered valid only if found at leastfive times. Each search typically produced at least one conformation that was found fewerthan five times.However, the Cartesian coordinates for these final structures described moleculesoriented randomly in space, making it necessary for them to be converted to a commonorientation in order for their dimensions to be easily calculated. We chose an orientation withthe carbon atom at the origin, C(2) on the x-axis and the C(2) substituent in the y plane, asshown in Figure 34. The transformation calculations were performed by a custom PASCALprogram. The various parameters for the individual conformations i were then simply readoff from this list of final coordinates as the maximum x value for X+(i), minimum x value forX(i) and so on. These parameters correspond to the STERIMOL scheme proposed byVerloop et al. (1976) as a means of quantifying shape as well as size in a Hansch-type linearfree energy relationship.To arrive at an average molecular size, it was assumed that the populations p ofconformations i would follow a Boltzmann distribution:[p1 / [p+1 = exp(E÷j - E) I RT) (2)91x+AzY+z+zAY=Y’+YAZ = Z + ZHFigure 34. STERIMOL scheme illustrated for(E)-2-pentenoate.92where R is the universal gas constant, T is temperature (37°C) and E is potential energy.From these ratios, the fraction P of the overall population corresponding to each conformationwas then calculated and used to determine the final parameters listed in Table 9 below. Forexample,X÷ = Z(P.X) (3)Table 9Molecular modelling of VPA analogues as carboxylates in an aqueous medium: population-weighted mean dimensions and number of conformers within 10 kJ of global minimumCompound Number of X Z Z Volumeconformers (A) (A) (A) (A) (A) (A3)2 4 5.42 0.99 2.66 1.36 1.36 94.03 5 3.73 0.99 4.45 1.25 1.37 93.94 2 4.27 1.01 3.02 0.90 0.90 94.25 3 5.18 1.97 3.19 1.40 1.17 110.06 2 5.18 2.86 2.90 1.33 2.18 127.77 6 5.19 3.46 2.85 2.46 1.99 144.78 26 5.21 4.22 2.89 3.31 2.32 161.99 10 5.29 3.76 2.69 1.93 1.93 138.010 9 5.31 2.77 3.16 2.27 2.91 139.111 2 4.99 2.20 2.14 1.86 2.31 123.112 2 4.58 1.99 2.14 1.29 1.30 99.813 2 5.60 2.11 2.08 1.56 1.33 117.114 2 5.22 0.98 4.35 1.59 1.27 117.415 2 5.56 1.01 4.31 1.56 1.55 133.316 13 5.41 0.99 4.31 1.56 1.55 150.517 14 5.56 1.01 4.27 2.45 2.50 166.6Calculations on the protonated forms of some acids were also performed for the caseswhere protonation might be expected to influence structure. As shown in Table 10, the maineffect was to increase the conformational lability, as indicated by the number of conformersbelow the energy cut-off, with little effect on the actual molecular shape as measured by +.As most remaining structures are either cyclic or possess very limited conformationalmobility regardless of the conditions, it is thus probably safe to assume that any drug’s shapeis relatively independent of its degree of dissociation.93Table 10Molecular modelling of VPA analogues as protonated acids in an aqueous medium: numberof conformers within 10 kJ of global minimum and population-weighted mean YCompound Number ofconformers (A)2 4 0.973 4 0.995 5 1.966 4 2.867 12 3.648 24 4.159 12 4.0710 7 2.81Finally, similar calculations were performed on selected Z isomers from the 2-substituted 2-pentenoate group (Table 11). Again, the compounds are as their carboxylates inan aqueous medium and the number of conformers refers to the number of stableconformations found within 10 kJ of the global minimum. As the data suggests, the Zisomers are inherently more conformationally labile than the E series. This is furtherillustrated in Figure 35.Table 11Molecular modelling of selected VPA analogues as Z isomersCompound Number of conformers(Z)-2-Pentenoate 5(Z)-2-Methyl-2-pentenoate 4(Z)-2-Ethyl-2-pentenoate 8(Z)-2-Propyl-2-pentenoate ((Z)-2-ene VPA) 1794(E)—2-ene VPA____*(z) -2 -ene VPA ,,,,NK / ‘.Figure 35. Conformations of (E)- and (Z)-2-ene VPA carboxylates in an aqueous medium.The structures at right are as seen looking along the y axis on to the xz plane.953.5 LIPOPHILICITYThe lipophilicity of a molecule is usually described in terms of its partition coefficient(log P), which in turn generally refers to the logarithm of the ratio of the neutral unionizedspecies in each phase of an n-octanollwater mixture (reviewed by Taylor, 1990; Kaliszan,1986). The standard approach to measuring log P is by the “shake flask” method, where thecompound is added to a mixture of n-octanol and an aqueous buffer solution of theappropriate pH, allowed to reach equilibrium and then assayed in both phases. Thedisadvantages of this method include the length of time required, the formation of emulsionsand the need for appreciable amounts of a potentially scarce compound, but also a rapidlyincreasing error as log P rises above zero. The latter is the result of the parameter being aratio, so that as log P increases, detection of the diminishing quantities of analyte in theaqueous phase will become increasingly uncertain. Although it may seem obvious to simplyincrease the overall amount of analyte in the mixture in these situations, the assay is alsounder the further constraint of having to keep concentrations in the organic (C0) and aqueousphases (C) less than 1 M and 10 mM respectively in order for the activity coefficient toremain at unity and the fundamental equation P = C0 I C, to remain valid.Consequently, much effort has been invested in the search for more efficient ways ofdetermining log P. Reverse-phase liquid chromatography (RPLC) is one such example, dueto the relationship between capacity factor k’ and log P:logP=alogk’+b (4)k’ = log [adjusted retention time I void time] (5)A further attraction of RPLC is the intuitive similarity of the analyte-stationary phaseinteractions with those that might be expected between the compound and a membranebilayer (Baker et al., 1977). In practice, this translates to using a series of standards of known96log P values to develop a calibration curve for the analytes of interest. The method is rapidand gives highly linear plots provided that the compounds share similar chemical structures.However, because the use of water as the mobile phase does not afford a log P / log k’correlation, a co-solvent (usually methanol) must be added, which clearly casts doubts onwhether or not the method genuinely reflects log P for a strictly aqueous medium. Thisobstacle is often avoided by measuring k’ as a function of the fraction of the organic solvent(em) in the mobile phase in order to obtain the “true” value (k):log k’ = log k - Bfm (6)A more empirical approach to log P was pioneered most notably by Nys and Rekker(1973) and since refined by the Pomona group (Hansch and Leo, 1979) and incorporated intotheir ongoing CLOGP program. In this method, a molecule ABC is divided into fragments A,B and C such that:log ABC = fA + fB + fc + E F1 (7)where the sum of the lipophilicity values of the fragments f, plus factors F to compensate fornon-additivity, equals log P for the molecule as a whole. The exact method whereby theCLOGP algorithm calculates these values has been described in detail recently by Leo (1990).An important aspect of any of the log P methods discussed above that is particularlyrelevant to VPA analogues is their measurement of log P for the unionized species eventhough the VPA analogues are 99.9% ionized at the pH of the biological medium. This isclearly not a problem if the resultant QSAR equation is used simply to predict the effective invivo activity (e.g. ED5O), but raises serious questions if used to probe actual mechanisms,such as an interaction with an enzyme. However, while such an equation would assume theneutral species to be the active form, which is possible but unlikely, the relative lipophilicitiesof the molecules should be constant regardless of their state of dissociation.97Finally, in the debate over the validity of various log P assays one should not losesight of the reason for using it in the first place. Namely, the need for an index oflipophilicity that can be correlated with biological results. If this need is fulfilled by log k’,for example, then it is irrelevant to ask whether or not log k is an accurate representation oflog P.In this study, lipophilicity was determined initially by the standard HPLC method asdescribed by Acheampong (1985) where, using a set of compounds of known log P values asstandards, lipophilicity was calculated from the capacity factor k. A plot of this calibrationcurve is shown in Figure 36. The log P values of compounds 2-15 were then calculated fromthis plot. Because the principal source of error was in the calibration curve itself, theuncertainty associated with log P was calculated from the corresponding 95% confidenceinterval.However, this approach suffered from the drawback of having to deal with a widerange of lipophilicity values such that HPLC conditions appropriate to the small C5 acidsproduced extremely long retention times for large analogues such as 8 (16 and 17 were notmeasured in this assay). Consequently, the CLOGP program was used instead to measure thelog P values for all compounds, which are used in all subsequent calculations. As shown inFigure 37 and Table 12, there was good agreement between the values obtained by the twomethods as the differences were generally similar to the 95% confidence interval for thevalues obtained by the HPLC method. Note that CLOGP does not distinguish cis and transdouble bonds.98Table 12Determination of lipophilicity (log P) by HPLC and CLOGPAcid mean tr’ log k’ log P(mm) Shake flask/HPLCa CLOGPVPA 14.0 0.483 2•75b 2.72Butyric 1.65 -0.403 0•98bValeric 3.22 -0.151Hexanoic 5.30 0.104 193b2-Ethylbutyric 5.59 0.0460 1 .68l2-Ethyl-Hexanoic 12.7 0.484 2.6412 2.24 -0.240 1.27±0.15 1.223 3.53 -0.183 1.39±0.14 1.224 3.18 -0.213 1.32±0.13 1.095 4.19 0.0328 1.79±0.11 1.536 6.82 0.244 2.19±0.11 2.067 10.8 0.446 2.59±0.14 2.598 18.8 0.675 3.04±0.24 3.119 10.5 0.329 2.37±0.13 2.2210 9.30 0.267 2.26±0.12 2.2211 5.52 0.0499 1.84±0.10 1.4212 3.43 -0.147 1.45±0.13 1.1013 5.89 0.0390 1.82±0.11 1.6614 5.75 0.0775 1.88 ±0.11 1.7215 7.68 0.483 2.68±0.16 2.2816 2.8417 3.40a) Expressed as 95% confidence interval determined from calibration curveb) Values from Keane et al. (1985)993. k’Figure 36. Log P vs. log (capacity factor) for some saturated carboxylic acids.CCc-.)4321HPLC log P4-0.4 -0.2 0.0 0.2 0.4 0.61 2 3Figure 37. Log P comparison: CLOGP vs. HPLC.1003.6 QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIPSModern QSAR theory has largely resulted from the seminal papers of Hansch and coworkers in the early 1960’s that showed how biological activity could be expressed as a sumof one or more physicochemical parameters related to free energy. The equations are usuallylinear as a function of at least one variable, and are thus collectively termed “linear freeenergy relationships” (LFERs). In view of its importance to our work, the derivation of thefundamental equation as described by Hansch and Fujita (1964) is shown below. Considerthe rate-controlling step for the expression of a drug’s biological activity as:Rate = A•C•k (8)where k is a rate constant and A•C, the product of the probability of a drug moleculereaching the target site and the extracellular concentration, is the effective concentration at thesite. Previous work (Collander, 1954, Hansch et al., 1963) having shown that the migrationof a drug with substituent X through a biological medium was dependent on lipophilicity (it)in a linear or, if log P was given a sufficiently broad range, second order manner with optimalactivity at it0, the probability A can be taken as a function of a drug’s lipophilicity. Assumingthis distribution to be normal,A = a.exp( —(it — it)2 / b) (9)it = log (P / ‘H) (10)where lowercase letters refer to constants and P and H to the n-octanollwater partitioncoefficients of the derivative and parent drugs, respectively. Substituting equation (9) for Aaffords:101Rate = a.exp( —(it — it0)2 / b)•C.k (11)Now, if the rate is taken as a constant biological endpoint, such as seizure protection, it willthus be equivalent to the concentration of drug required to reach that endpoint (e.g. ED5O).1/C = d.exp( —(it — it0)2 / b).k (12)Taking the logarithm then gives:log (1 / C) = -dit2 + 2dit•it0 - d.(ic0)2 + log k + e (13)The k term can also be expanded because of its relationship with the Hammett equationlog (k / kH) = pa x G (14)which states that the ratio of the rate constant k for a molecule with substituent X to that ofthe unsubstituted parent (e.g. X = H) is equal to the product of the two parameters p and ,related to the electronic properties of X and the type of reaction involved, respectively.Because the molecules under scrutiny are all derived from the same parent, kH and c will beconstant:log (1/ C) = -dit2 + 2ditit0 - d.(it0)2 + pa + f (15)At this point, the equation expresses biological potency of a series of related drugs in terms ofthe lipophilicity it and electronic inductive effect p. However, there are still two limitations.First, it refers to the lipophilicity of substituent X, which cannot be measured directly, ratherthan the derivative drug itself. By recognizing log H to be a constant, one obtains:102log (1/ C) = -g.(logP)2 + h•logP + p.o + i (16)The second problem arises from the lack of a stereochemical term describing shape.Based on the Hammett equation, one can derive a similar constant E such that:= log (k I kH) (17)Because the original k (equation (8)) is actually the product of all parameter-dependent rateconstants involved, such thatk = k(electromc) k,(steric) (18)the lipophilicity term already having been incorporated into the equation elsewhere as a freeenergy term in its own right, the final version can be written as:log (1 IC) = -g(logP)2 + hlogP + p.a + j.E + i (19)Because E simply represents size, it can be replaced with molecular volume or a so-called STERIMOL parameter, as described in Section 3.4. In this study, there was no role forelectronic parameters and examination of the data failed to show any sign of a second-orderdependence of anticonvulsant activity on lipophilicity. Thus, the first and third terms ofequation (19) were omitted to produce the following final version:log (1/ C) = h•logP + j.(steric descriptor) + i (20)Links between biological and physicochemical parameters were then investigated byanalyzing for all possible correlations as shown in Table 13. Not surprisingly, lipophilicitywas found to be significantly correlated with molecular volume V and most width parameters:103consequently, it is not possible to distinguish the roles of shape and lipophilicity in drugpotency which does not permit the use of both parameters simultaneously in equation (20).Table 13Correlation matrix of biological and physicochemical properties of VPA analogues aDoseb Cc logPd V X AZ ZC 0.871 1.000logP -0.920 -0.888 1.000V -0.955 -0.874 0.973 1.000X -0.636 -0.584 0.569 0.598 1.000AY -0.683 -0.548 0.688 0.692 0.155 1.000-0.186 -0.289 0.352 0.223 -0.055 0.252 1.000-0.453 -0.258 0.331 0.433 0.178 0.676 -0.543 1.000AZ -0.845 -0.703 0.816 0.882 0.442 0.680 0.014 0.581 1.0001 -0.731 -0.512 0.627 0.721 0.358 0.556 -0.066 0.533 0.908 1.000Z -0.750 -0.674 0.772 0.816 0.406 0.704 -0.037 0.786 0.918 0.736a) Correlation coefficient rb) Log(ED5O)c) Log([drugJj)The best correlations of log(ED5O) with these physicochemical properties foranalogues 2-17 are shown below, along with the relevant statistical parameters. Specifically,s is the standard error of estimate (square root of residual mean square) and r is the correlationcoefficient. Fcaic is used to test the null hypothesis that slope = 0 (e.g. no correlation) bycomparison with the corresponding table value of the F distribution for a 1-tailed test at a 1%significance level.log(ED5O) = (-0.0139±0.0012).V + (5.0520±0.149) (21)(n = 16; s = 0. 1069; r = 0.954; Fcaic = 142.29 Ftable = 8.86)log(ED5O) = (-0.4331 ± 0.049).log P + (4.1641 ± 0.104) (22)(n = 16; s = 0.1403; r = 0.920; Fcaic = 76.63 Ftable = 8.86)104Multiparameter equations are also quite possible. For the second parameter in thefollowing equations, + and Y- were chosen as they show the lowest correlation withvolume/log P.log(ED5O) = (-0.0136 ± 0.0013)•V + (-0.0153 ± 0.028).Y + (5.0437 ± 0.153) (23)(n = 16; s = 0.1097; r = 0.955; Fcaic = 67.70; Ftable = 6.7)log(ED5O) = (-0.0140±0.0012)•V + (0.0121 ±0.034).Y + (5.0254±0.171) (24)(n = 16; s = 0.1104; r = 0.955; Fcaic = 66.75; Ftable = 6.7)log(ED5O) = (-0.4076 ± 0.050).log P + (-0.05 14 ± 0.033).Y + (4.2162 ± 0.105) (25)(n = 16; s = 0.1336; r= 0.933; Fcaic = 43.47; Ftable = 6.7)log(ED5O) = (-0.4592 ± 0.0051)•log P + (0.0638 ± 0.044).Y + (4.0108 ± 0.145) (26)(n = 16; s = 0.1350; r 0.93 1; Fcaic = 42.47 Ftable = 6.7)Because the ED5O values in Table 5 are a measure of all the rate constants involvedbetween the injection of drug and its interaction at the final site, they are a poor measure ofthe actual pharmacodynamic properties of the drug. Hence, the initial absorption steps wereeliminated by replacing ED5O in the previous equations with the brain concentrations of thedrug following an ED5O dose.log([drug]j) = (0.0184 ± 0.0027)V + (5.101 ± 0.348) (27)(n = 16; s = 0.2496; r = 0.875; Fcaic = 45.56; Ftable = 8.86)log([drug]j) = (-0.0197 ± 0.0030).V + (0.0670 ± 0.064).Y + (5.1462 ± 0.348) (28)(n = 16; s = 0.2488; r = 0.885; Fcaic = 23.48 Ftable = 6.7)1051og([drug]j) = (-0.0179 ± 0.0028)•V + (-0.0577 ± 0.079)•Y + (5.2367 ± 0.394) (29)(n 16; s = 0.2539; r = 0.880; Fcaic = 22.29; Ftable = 6.7)log([drugjj) = (-0.6029 ± 0.0084)•log P + (3.997 ± 0.176) (30)(n = 16; s = 0.2370; r = 0.888; Fcaic = 52.09; Ftable = 8.86)However, these equations could be misleading because the molecular weight of thedrugs themselves is also highly variable. The question to ask is whether or not the differencesin potency could have been accounted for by different masses of drug being injected. Shouldthis have been the case, we would conclude that the anticonvulsant effect was the result of acompletely non-specific interaction. As ED5O is a fixed endpoint this would require potencyto increase in parallel with molecular weight in order to keep the mass of the injected drugconstant. However, this possibility of a completely non-specific mechanism can be dismissedby the fact that ED5O values (0.73-7.0 mrnollkg) vary much more than molecular weights(100 g/mol (2-4) to 168 g/mol (17)), indicating that protection against seizures was a functionof the drug itself rather than simply the weight of material administered. Further justificationfor the molar rather than mass basis was provided by the inferior correlations shown below.Log(ED5O) {mglkg) -(0.0109 ± 0.0012).V + (3.793 ± 0.148) (31)(n = 16; s=0.1066; r=0.929; Fcaic = 88.8 Ftable 8.9)Log([drug]jfl) {ug/g} = -(0.0154 ± 0.0027).V + (3.853 ± 0.347) (32)(n = 16; s = 0.2494; r = 0.835; Fcaic = 32.3 Ftable 8.9)1064.0 I3.8c.. -3.6 -N..-..3.0 -•-.2.6I I100 120 140 160 180Volume (A3)Figure 38. Log(ED5O) vs. volume for acids 2-17.VPA (not used in equation) is shown as an empty circle.4.0 I I I I3.8-•3.63.4-3.2-3.0-in •.Q\w. .• N •2.8 - .2.6 I I I1.0 1.5 2.0 2.5 3.0 3.5 4.0logPFigure 39. Log(ED5O) vs. lipophilicity for acids 2-17.VPA (not used in equation) is shown as an empty circle.107I I I I I I I3.8 --3.6-3.4•9-3.2 -Cj3.O--l028 o-J- •2.6-12 14 •o.— 2.4 -2.2-2.0-1.8 -1.6 I I90 100 110 120 130 140 150 160 170 180Volume (A3)Figure 40. Log[drug]f vs. volume for acids 2-17.VPA (not used in equation) is shown as an empty circle.4.0 I I I3.8 -—3.6 .- -2A—.j.r3.2 N.. -C3.0--2.8 - 0 -•1214..<-..-2.6-2.2--2.0-1.8 - .... -1.6 I I1.0 1.5 2.0 2.5 3.0 3.5 4.0log PFigure 41. Log[drugjjfl vs. lipophilicity for acids 2-17.VPA (not used in equation) is shown as an empty circle.1083.7 MEMBRANE FLUIDITYThe fluidity assay employed in this project (reviewed by Shinitzky and Barenholz,1978) is based on the degree of fluorescence depolarization when a hydrophobic fluorophoreembedded in a membrane is excited with polarized light. A more fluid membraneenvironment will allow the probe to rotate more easily during its excited state and diminishthe intensity of the emitted light polarized in the plane of the exciting beam (‘) by reemitting the energy in other planes, such as the one perpendicular to the exciting beam (L,,h).This can be expressed as the fluorescence anisotropy r:r = (I- GIh) / (Lv,, + 2GIVh) (33)where G is an instrumental factor to correct for instrumental polarization of the emitted beam,calculated by:G‘hh’’hv (34)when the excitation polarizer is set in the horizontal position. The parameter r is related to theactual membrane fluidity by the Perrin equation:r0/r = 1 + [C(r)•Tt/11] (35)= limiting anisotropy, or that value of r which would be observed in a rigid matrixC(r) = parameter relating to fluorophore shape and the location of its transition dipoles= lifetime of excited state= membrane viscosityT = temperature109Equation (35) suffers from several drawbacks. First, the C parameter will vary with rfor a non-spherical fluorophore. Second, we cannot readily determine the lifetime of theexcited state. Finally, as the viscosity will vary with the particular axis of symmetry becausethe membrane is an anisotropic medium, it needs to be replaced with the term“microviscosity” (), which is defined as the geometrical mean of the viscosities along thesymmetry axes, but which may thought of as the “effective viscosity”. If the Perrin equationis rewritten as([rIrJ- 1)_i = I{C(r)•T•tJ (36)we now effectively obtain([r0 / rj - 1)1= 0 / constant (37)because the variations of C(r) and t with temperature or between individual preparationslabelled with the same probe are considerably smaller than Ti. Because C(r) and t decreasewith increasing temperature, the result is that the denominator of equation (36) is essentiallyconstant that allows for the relation of anisotropy to membrane fluidity.The most commonly used membrane probe is diphenyihexatriene (DPH), a highlylipophilic compound with a simple spectrum that localizes in the centre of the membranebilayer (Figure 42). Furthermore, it exhibits fluorescence only when in a non-polar medium,which minimizes background interference from molecules not incorporated into a membrane.Figure 42. Structures of membrane probes DPH (R = -H) and TMA-DPH (P. = -N(CH3)j110The properties of the analogous compound trimethylammonium-diphenylhexatriene(TMA-DPH) are very similar to those of DPH with the important exception that the moleculeis anchored in the region immediately below the membrane surface by its cationic substituent(Prendergast et aL, 1981).Initial experiments using DPH with various membrane preparations (e.g. crudeerythrocyte ghosts, synaptosomal membranes and intact synaptosomes) indicated a barelysignificant difference between controls and samples incubated with 10 mM VPA, a result ingeneral agreement with the work of Penman and Goldstein (1984). It thus appeared that theproblem of low sensitivity was the result of the membrane-disordering potency of VPA beingeither very low or exerted in a region not sampled by the highly lipophilic probe DPH.Because VPA in its dominant dissociated state might be expected to interact primarily withthe extracellular surface rather than the interior of the plasma bilayer, we thus investigated theuse of the polar probe TMA-DPH which should be more responsive to changes in this region.The parameters for this fluidity assay were established as follows. The erythrocyteghosts were diluted to afford a 0D366 value of about 0.1, which gave a maximum of 4%scattering at the emission wavelength of 428 nm (Figures 43 and 44). The integrity of thepreparation was verified by Arrhenius plots of anisotropy for control and 20 mM VPApreparations (Figures 45 and 46). The negative slope indicates that anisotropy was decreasingwith increasing temperature, as expected. Thus, the degree of membrane fluidity increased asthe probe was more readily able to rotate during the period between absorption and emission.The results of this TMA-DPH assay showed little or no effects of 20 mMconcentrations of representative drugs on membrane fluidity, although a few exhibitedanisotropy shifts that reached statistical significance (Table 14). In contrast, all drugs showeda significant decrease in anisotropy when the assay was repeated with DPH, although therewere no apparent trends that could be correlated with any physicochemical properties of thecompounds. For comparison, Perlman and Goldstein obtained & = -0.0010 for DPH in a 11mM VPAlsynaptic membrane preparation at 25°C. Seeing that they found r to be .a linearfunction of concentration, this value agrees quite well with our own for 20 mM VPA.111C,)1)ci) 40ci)Wavelength (nm)500>-‘C.)>C.)Figure 44. Emission spectrumof erythrocyte ghosts containing fluorescent probe DPH.Figure 43. Emission spectra oferythrocyte ghosts with andwithout incorporated fluorescent probe TMA-DPH.Numbers in parentheses referto recorder potentiation.80-601 (+probe, 9)‘vh (+probe, 9)1008O— -60—:40—2O—4:0360 400 500Wavelength (nm)2001120.26 I I0.250.24-0.23->o -o0. .H I3.15 3.20 3.25 3.30 3.35 3.40(l/T)x 1000(1/K) VFigure 45. Fluorescence polarization of DPH in erythrocyte ghosts.:::0.240.23 I I3.20 3.25 3.30 3.35(l/T) x 1000 (1/K)Figure 46. Fluorescence polarization of TMA-DPH in erythrocyte ghosts.113Statistical significance was determined by an independent two-tailed Student’s t-testusing the pooled method for calculating the standard error of the difference and using asignificance level of 0.05.Table 14The effect of VPA analogues (20 mM) on membrane fluidity of human erythrocyte ghosts asmeasured by fluorescence anisotropy (r, x 10)TMA-DPH DPHCompound r z\r r &Control 279.2 ± 4.4 - 243.6 ± 5.9 -1 281.8±6.7 +2.6 226.9±8.6* 16.7*7 278.0±4.6 -1.2 229.6±4.1* 14.0*8 277.9 ± 8.3 -1.3 237.5 ± 7.0*9 274.9 ± 35* 43* 208.3 ± 5.6* 353*10 281.3±4.3 +2.1 208.4±9.7* 35.2*14 275.9±4.1* 33* 224.9±8.6* l8.7*15 278.5 ± 4.8 -0.7 163.2 ± 4.2* 80.4*16 281.4±7.3 +2.2 237.3±6.9*17 271.0±6.1**) Significantly different (p <0.05) from control.Unfortunately, we were subsequently unable to reproduce the results obtained withDPH using several different, but identically prepared, ghost membrane suspensions.Similarly, there was no difference when both the incubation and the measurements wereconducted at 37°C, or when the highly lipophilic drug tamoxifen (100 uM), a potentmembrane-ordering agent (Wiseman et al., 1993), replaced VPA. Thus, the membranefluidity studies were considered inconclusive.3.8 SYNAPTOSOME PREPARATIONThe synaptosomes from the brains of mice administered ED5O doses of each drug (asdescribed in Section 3.2) were prepared by an adaptation of the Loscher et al. (1981) versionof the method of Dodd et al. (1981), as summarized in Figure 47. Although there was noclear evidence, such as that provided by electron microscopy, that the preparation did in fact114consist of resuspended synaptosomes, the GABA content is quite similar to that obtained byLoscher et al.. Furthermore, because GAD activity is found predominantly in nerve terminals(Martin, 1986), it can double as a marker enzyme for this region. Again, the activitiescalculated are not significantly different from those of Loscher et at.. In both cases, enzymeactivity was determined by measuring the amount of GABA formed following incubation ofthe synaptosome suspension with glutamate and pyridoxal 5-phosphate. Some differencesare likely accounted for by the fact that these workers used a radioreceptor-type displacementassay for GABA and GAD. However, the synaptosomal protein content in the GAD assaywas determined by the same literature method (Markwell et at., 1978) and yet shows a largediscrepancy.Table 15Comparison of synaptosome parameters with results of Loscher et at. (1981)Experimental Loscher et at.Protein (GABA assay) 19.6± 2.1(mg I g wet tissue)Protein (GAD assay) 16.3 ± 2.Oa 5.29 ±0•65d(mg I g wet tissue)GABA concentration 20.0 ± 1.41) 16.6 ± 42d(nmol / mg protein)GAD activity 325 ± 23C 442 ± 134d(nmol I mg protein I h)a) Mean ± SD of two separate experiments using 5 mice: each individual synaptosomesuspension was sampled in duplicateb) As in a), but one experiment only and sampling in triplicatec) As in a), but sampling in triplicated) n=48115Dose mice with drug or saline1decapitate at 15 mmfreeze portionhomogenize brain in 0.32 M saline —* for braindrug assaycentrifuge: 1000g. 10 mmsupernatantlayer on to 1.2 M sucrose1centrifuge: 220,000g. o2t = 1.6 x 1010 rad2/s‘Srcollect interfacecentrifuge as abovewater KCI-Triton-GABA-d6freeze resuspendsynaptosome phosphate bufferpelletGAD assay GABA assaytrichioroacetic GABA-d6 trichioroaceticacid acidcentrifuge: 13,600g. 10 mmsupernatantlyophilize65°C/i h+ MTBSTFA in acetonitrile GC-MS analysisFigure 47. Preparation of synaptosomes for GABA and GAD assays..1163.9 GABA AND GAD ASSAYS3.9.1 GENERALWe initially employed the method of Bertilsson and Costa (1976) for thederivatization of GABA with pentafluoropropionic anhydride and hexafluoroisopropanolfollowed by GC-MS analysis. However, the peak shapes were poor and it was felt that thelength of the derivatization procedure and its acidic column-degrading by-products were notfully compatible with our requirements. Consequently, other methods were investigated. Ourinitial treatment of GABA with N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA)afforded the N, O-bis(trimethylsilyl) GABA derivative. The compound showed sharp peaksand good sensitivity with the m/z 232 ion, corresponding to a species having lost one methylgroup presumably from the amino substituent. When MSTFA was replaced with its tert-butylanalogue, MTBSTFA (containing 1% tert-butyldimethylsilyl chloride (tert-BDMSC1)),GABA was derivatized to the N, O-bis((tert-butyl)dimethylsilyl) compound. Sensitivityimproved approximately ten-fold using the analogous ion m/z 274 (molecular ion less a tertbutyl substituent) that appeared as the base peak (Figures 48 and 49).3.9.2 ASSAY CHARACTERISTICSOwing to differences in the buffer medium and concentrations of GAB A, separatecalibration curves were required for the synaptosomal GABA and GAD assays. Both curvesshowed good linearity with r2 > 0.99. Coefficients of inter-day variation for the GC-MSportion of the method were 7.7% and 7.9% for the GABA and GAD assays (n 7),respectively, calculated from their respective calibration curves. The coefficients of intra-dayvariation were 0.95% (five replicates of five samples) and 1.9% (three replicates of tensamples) for the GABA and GAD assays, respectively. The response was essentially linearfrom 0.1 to at least 20 nmol GABA.117Abundance HNS18000027417160000 / M - 5773140000120000100000800006000040000 I I20000 Ii [I 184 2160 WL JI1l! .1. 1 L.IhI ..d. . , .____________100 200 300Mass!ChargeFigure 48. Electron impact mass spectrum Of GABA disilylated with MTBSTFA.Abundance GABA-d6(m’280) Abundance GABA (m/z274)15000 500004000010000 3000020000500010000lii,II. III81 8.3 8.5 8.7 8.1 8.3 8.5 8.7Time (men) Time (mm)Figure 49. SIM chromatograms for GABA-d6and GABA from synaptosome GABA assay.118For a 3:1 signal/noise ratio, the detection limit was about 10 fmol of injected GABAderivative, corresponding to about 6 pmol GABA per 50 ul synaptosome aliquot. Thecalibration curve (y = 0.00424 + 0.8182x) for this range (6.7-107 fmol GABA per injection)was linear with r2 = 0.9998. This sensitivity was more than adequate for the GABA andGAD assays because the average synaptosome sample in either assay contained at least 1000pmol GABA.The derivatized samples were stable for at least several weeks at room temperature.3.9.3 APPLICATION TO SYNAPTOSOMESGABA levels and GAD activity for both fresh and frozen synaptosomes are shown inTable 16. As might be expected, GABA concentrations were relatively similar for both freshand frozen samples. Encouragingly, this was also the case for GAD activity. The noticeablyhigher GABA levels recorded in our assay, in comparison with Loscher et al. (1981), may bethe result of the detergent in the resuspension buffer, which may have permitted a moreefficient extraction of GABA from vesicle sites although this factor was not investigatedfurther. Note the similarity of the values for GABA and the GAD blank, in terms of nmollmgprotein, indicating that the latter (e.g. synaptosomes incubated with buffer only) does notcontain appreciable amounts of glutamate and PLP and thus represents primarily endogenousGABA.The question of GAD assay reproducibility was further examined by separatepreparations of synaptosomes. The first trial afforded a (fresh) activity of 335 ± 29nmol/mg/h, while the second gave 315 ± 9.0 nmol/mglh, representing a difference of about6%. These values also agree well with the control activity determined in the GAD saturationassay (Section 3.10).There were no significant interferences for either the GABA or GABA-d6peaks. Thiswas demonstrated by the blanks run using only the resuspension buffer (GABA assay) or theG1uIPLP/phosphate solution (GAD assay). For the former, the GABA/GABA-d6ratio was1190.68 ± 0.38 % of the mean value for the intact synaptosome assay. For the GAD assay, theblank gave a ratio that was about 0.081 ± 0.023 % of the value obtained when thesynaptosomes were incubated with the G1uJPLP solution.The recovery of GABA-d6 from the resuspended synaptosomal pellet in the GABAassay was 78 ± 3 % (n = 3), based on the ratio of the GABA-d6peak area obtained in theassay to the peak area recorded when this internal standard was incorporated into theMTBSTFA derivatization mixture rather than into the resuspension buffer. Although therecovery value is modest, the associated error is apt to be minimal owing to the chemicalsimilarity of GABA-d6 to GABA. It is unlikely that protein binding was the cause of theselosses as the recovery was still only 86.9 ± 4.3 % (n = 6) when GABA-d6 was addedimmediately prior to lyophilization rather than at the initial pellet resuspension. Thedetermination of the recovery of GABA itself using a spiking method might not be reliablebecause the endogenous compound is in vesicles rather than simply in solution.Table 16Analysis of GABA and GAD in mouse whole brain synaptosomesaConditions GABA GAD activityb(nmollmg protein) (nmol GABAImg protein/h)Fresh synaptosomes + Glu + PLP 20.0 ± 1.4 314.9 ± 9.0Frozen synaptosomes + Glu + PLP 18.9 ± 1.5 329 ± 18Frozen synaptosomes + PLP 7.9 ± 3.5Frozen synaptosomes + Glu 32.0 ± 4.4a) n = 5b) Corrected for a blank value of 25.9 nmol GABAImg protein/h.1203.9.4. DEMONSTRATION OF LINEARITY AND SATURATION CONDITIONSAlthough the GAD assay conditions used in the project were those of Loscher(1981b), evaluations of the dependence of GAD activity on time and concentrations ofsubstrate, co-factor and protein were also conducted, albeit subsequently to the resultsreported in Section 3.9.5.. The results are shown graphically in Figures 50-53: circled pointscorrespond to the standard assay conditions used subsequently.It is clear that while the samples contained saturating levels of PLP, the same couldnot be said for glutamate. This might be accounted for by the higher protein content of oursynaptosomes compared to those of Loscher et al.. Figure 52 indicates a gradual decrease inrate with time over the standard 60 mm incubation period, suggesting that a shorter duration,on the order of 10-20 mm, would have been more appropriate. Finally, Figure 53demonstrates that activity was in fact linear with respect to protein concentration.1214501 I I I I400— -350— -300- -0 - -200- -- 150- -t 50 -I I I I0 2 4 6 8 10[Glutamate] (mM)Figure 50. Effect of glutamate concentration on GAD activity.350 I300 —200 -150 -I0.1 0.2 0.3 0.4 0.5[PLP] (mM)Figure 51. Effect of PLP concentration on GAD activity.122I I I I500— -400- V -030O- -200- -4l00 -0 I I I I0 20 40 60 80 100 120 140Time (mm)Figure 52. Time-dependence of GAD activity.800 I I I700 -_600 - .. 500-400 -g200 -100-0 ‘• I I I0 20 40 60 80 100 120 140Protein (ug / sample)Figure 53. Effect of protein concentration on GAD activity.1233.9.5 RESULTS FROM DRUG-TREATED MICEUsing an ED5O dose of each drug, the levels of GABA and activities of GAD weredetermined in mouse brain synaptosomes as indicated. GABA was measured using freshlyprepared synaptosomes while GAD activity was determined from samples stored at -78°C.Table 17Interaction with synaptosomal GABA metabolism after an ED5O doseCompound GABA % GAD activitya %(nmol I mg change (nmol I mg changeprotein) protein / h)15 mmControl 20.0 ± 1.4 - 329 ± 18 -1 22.2±1.6* +11* 351±21 +6.72 22.4±3.7 +12 318±44 -3.33 22.2± 3.4 +11 310±12 -5.54 20.7± 1.9 +3.4 370±22* ÷12*5 17.7±1.1* *11* 269±38* *18*6 18.8±1.8* *6.2* 293±45 -117 22.1 ± 1.8* +10* 308±44 -6.48 20.6 ± 1.9 +2.9 297±27 -9.79 22.2±1.1* +11* 328± 16 -0.310 21.3±2.8 +6.7 278±10* -15k11 22.3±3.0 +11 299±23*12 19.6± 1.1 -2.1 276±64 -1613 20.3 ± 1.4 ÷1.6 337 ±7 +2.414 23.4±2.8* ÷17* 355±31 ÷7.915 21.4±2.6 ÷7.0 294±18*fl*16 19.1 ± 1.8 -4.5 338 ± 27 +2.717 16.4±2.2* 48* 273±15* *17*30 mm1 23.7±2.1* +18*7 22.7 ± 3.6* ÷13*8 19.8±2.6 -1.09 21.85±1.1* +9.2*10 20.8±2.1 ÷4.2120 mmControl 19.35± 1.35 -*) Significantly different from control (p <0.05) by one-tailed Students t-test.a) Mean±SD124Synaptosomal GABA levels were also assayed in mice administered higher doses ofVPA in order to demonstrate dose-dependence.GABA % change(nmollmg protein)15 3 2.08 24.2 ± 2.4* +21*30 3 2.08 25.4 ± 3.2* +27*15 1 4.17 26.72±0.27* 34*a) Mean±SDA correlation matrix analogous to the one shown in Table 13 was constructed with theadditional entries for changes in GABA levels and GAD activities (Table 19).Clearly, there were no significant correlations for GABA with physicochemicalparameters or GAD activity, although the relationship of GAD as a function of GABA and Z(mutually unrelated parameters as indicated in the Table 19) deserves mention. It isnoteworthy that where the changes in GABA and GAD were both significant, their directionswere usually identical, as indicated by the positive value of r. Some of the more interestingequations are shown below.GAD activity (nmollmg/h) = (6.54 ± 3.58)GABA + (172 ± 75) (38)(n = 16; s= 28.1; r=0.438; Fcaic = 3.33 8.9)GAD activity (nmollmg/h) -(26.4 ± 12.3)Z + (354 ± 23) (39)(n = 16; s = 27.1; r=0.498; Fcaic = 4.6; Ftable = 8.9)GAD activity (nmollmg/h) (5.60 ± 3.28)•GABA - (23.6 ± 1 1.7).Z + (233 ± 74) (40)(n= 16;s=25.5;r=0.621;Fcalc=4.1;Ftable=6.7)Table 18Effect of high VPA doses on synaptosomal GABATime n Dose(mm) (mmollkg)DosebC’logPDose1.00C0.871.00logP-0.92-0.901.00V-0.95-0.870.97X+-0.64-0.580.57AY-0.68-0.550.69-0.19-0.290.35-0.45-0.260.33AZ-0.84-0.700.82Z-0.73-0.510.63Z+-0.75-0.670.77GAD0.160.22-0.21GABA0.270.40-0.40a)Correlationcoefficientrb)Log(ED5O)c)Log([drug1j)d)GADactivity(nmol/mg/h)e)Nmol/mgTable18Fullcorrelationmatrixofbiologicalandphysicochemicalpropertiesof VPAanaloguesavx-’-yy-y+AZZ-Z-GADdGABAe1.000.601.000.690.151.000.22-’)1263.9.6. INifiBITION OF GAD BY ACID 17The simultaneous decrease in GABA levels and GAD activity for acid 17 prompted usto examine whether or not this compound could also inhibit the enzyme in vitro.Control synaptosomes were incubated with a saturating concentration of PLP andvariable amounts of glutamate and compound 17 (as its sodium salt) for 10 mm, a length oftime previously shown to afford a good estimate of the initial rate (Figure 50). Theconcentrations of glutamate (1-10 mM) spanned the range from saturating to those affordingGABA levels only about half those present in the blank. The results are expressed as adouble-reciprocal plot in Figure 54, which clearly shows the enzyme being inhibited at both 1and 5 mM concentrations of compound 17. Based on the apparently common x-intercept butdifferent y-intercept values for the three lines, we conclude the mode of inhibition to be non-competitive. The following equations were then used in conjunction with Figure 54 to obtainthe parameters listed in Table 20.x-intercept = -1 I Km (41)Vmax = Km I slope (42)Vma) = Vmax / (1 + [11/IS) (43)Vmjj = Vmax in presence of inhibitor I at concentration [I] (44)Table 20Effect of acid 17 on activity of GAD in vitro[17] Km (glutamate)(mM) (mM)VI K1max(nmol / mg protein) (mM)0 2.1 98 -1 2.0 83 5.55 2.4 70 12(mean: 9 mM)127ft040.03C0.020.01Figure 54. In vitro inhibition of synaptosomal GAD by 17.Error bars indicate SD for one sample evaluated in triplicate.1.0 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.81f[Gluj (ms’F’)1283.10. GAD SATURATIONFinally, it was of interest to determine the mechanism by which these drugs might leadto GAD activation or inhibition. Because GAD is only partially saturated with PLP cofactorunder normal physiological conditions, an intuitively obvious way to increase its activity is toincrease the extent of cofactor binding. This would be manifested as enhanced enzymeactivity when the preparation is assayed following removal of all unbound cofactor, in thiscase by gel filtration. Thus, our approach was to isolate synaptosomes by the standardmethod from mice given ED5O doses of VPA, 2-ene VPA and compound 17, removeunbound low-molecular weight species by passage through a Sephadex G-25 size exclusioncolumn and conduct a modified enzyme activity assay. This enzyme assay differed from thestandard version used above in its short reaction time (5 mm vs. 60 mm), low concentration ofglutamate (1 mM vs. 5 mM: Km = 1 mM (Martin and Rimvall, 1993)) and avoidance of aphosphate buffer in an attempt to minimize apo-GAD/holo-GAD interconversions. GABAproduction was then analyzed by GC-MS as before. However, it was found that therecommended concentration of imidazole-acetate buffer (100 mM: Martin, 1986) did notafford satisfactory or reproducible derivatization of GABA to its disilylated form, presumablybecause both imidazole and acetic acid, unlike phosphate, can serve as substrates forMTBSTFA. Increasing the amount of MTBSTFA was unsuccessful, but reducing the bufferconcentration gave a concentration-dependent improvement in the response.An initial trial run was conducted in order to test the effectiveness of the gel filtrationcolumn. Individual column fractions were assayed for protein concentration and approximaterelative GAD activity (expressed only as the GABAIGABA-d6area ratio) as shown in Figure55. Protein recovery was approximately 87% (fractions 7-11) and there was good agreementbetween the protein concentration and GAD activity curves. The level of GABA in the peakfraction was minimal (Table 21: compare with Table 16), demonstrating the integrity of thegel filtration protocol. Finally, the enzymatic origin of the produced GABA was proven bycomparing solutions containg only Glu andlor PLP as well as blanks where both were omitted8-7-. 6-U3-< 2-ct i0—600 -400 -300 -200 -100 -5Fraction #Fraction #129Figure 55a. GAD saturation assay: approximate relative GAD activity vs. column fraction #.500-I I I--EI5 6 7 8 9 10 11 12Figure 55b. GAD saturation assay: protein concentration vs. column fraction #.130(Table 21) with the complete mixture. These results show that the maximal interference withthe GAD activity was only about 3% and was apparently derived mainly from the bufferitself.Table 21Interference with holoenzyme activity in post-gel filtration GAD assay’Conditions GABA/GABA-d6 Interferencebarea ratio (%)1.20±0.81 -0.075 ± 0.044 6.2 ± 4.40.039 ± 0.010 3.2 ± 1.20.032 ± 0.0027 2.7 ± 0.40.0174 ± 0.0099 1.5 ± 0.9Eluate + Glu + PLPPLPGluPLP + GluEluate onlya) n = 3b) (GABAIGABA-d6area ratio in Conditions”) I (GABAIGABA-d6area ratio (eluate + Glu+ PLP))The full procedure was carried out using 3 mice per drug. In the initial brain isolationstep, decapitation into ice-cold saline was promptly followed by cutting of the scalp in orderto cool the brain as rapidly as possible prior to its equally rapid removal from the cranium. Ithas been shown that prolonged exposure to room temperature at this point will markedlyaffect the saturation of the enzyme (Miller et al., 1977). Synaptosomes were then prepared asdescribed and an aliquot from each was purified by the size exclusion column. The combinedprotein fractions were assayed for GABA production in the presence and absence of addedPLP. The degree of saturation of the enzyme with cofactor was then calculated as the ratio ofthese two values following correction for endogenous GABA (Table 22).The large uncertainty associated with the control value for synaptosome activity wasthe result of one apparently anomalously high result. All three synaptosomal (un-filtered)activity measurements from this sample were in agreement, indicating that the preparationitself was suspect. When this value was omitted, the activity for synaptosomes from VPAtreated mice reached statistical significance.131Table 22Effect of ED5O doses of VPA, 2-ene VPA and compound 17 on saturation of synaptosomalGAD 15 mm postdoseaTreatment Synaptosomal Gel-filtered GAD activity SaturationGAD activity (nmol GABAJmgI5 mm) (%)(nmol GABAImg/h) With PLP Without PLPIndividual resultsSaline 309 14.18 ± 0.37 6.06 ± 0.13 42.7327 12.17±0.15 5.34±0.10 43.9383 14.45±0.31 6.11±0.14 42.3VPA 373 14.39±0.16 6.189±0.059 43.0365 13.48±0.21 5.68±0.15 42.1347 11.36±0.39 4.77±0.17 42.02-EneVPA 309 11.96±0.41 4.53±0.10 37.9357 9.692±0.095 3.710±0.082 38.3336 11.94±0.18 5.057±0.064 42.317 343 11.87±0.85 5.10±0.33 43.0366 9.96±0.46 3.64±0.11 36.6396 10.00±0.41 3.51±0.13 35.1Mean resultsSaline 339 ± 38 13.60 ± 1.25 5.84 ± 0.43 42.97 ± 0.82VPA 361 ± 14 13.08 ± 1.55 5.55 ± 0.71 42.38 ± 0.552-EneVPA 334± 24 11.20± 1.30 4.10±0.41 39.50±2.4717 369±27 10.61± 1.09 4.08±0.88 38.21±4.19a) Mean ± SD for mice sampled in triplicate.1324. DISCUSSION4.1. SYNTHESISBecause the initial objective of this study was to focus on E and Z isomers of 2-substituted 2-pentenoic acids, most of the described synthetic efforts are aimed in thisdirection. First, in any study examining the effect of chemical substitution on biologicalactivity it is clearly essential that the compounds be available as single isomers. Furthermore,in our case, it was desirable that the method give consistent results with all types of substrateswith only minimal modification. For these reasons, we chose as a general approach thedehydration of 3-hydroxy carbonyl compounds, which can be prepared with high degrees ofdiastereoselectivity through the highly stereoselective aldol addition established by Evans etal. (1981) involving the addition of dibutylboryl enolates of N-acyloxazolidinones toaldehydes. This method affords the erythro adducts with consistently high yields andselectivity for a wide variety of substrates. When dehydrated by an E2 mechanism-basedprocess, these adducts should then lead to the E-unsaturated derivatives (Figure 6).Furthermore, it has recently been reported by Walker and Heathcock (1991) that the threoisomers are also available via minor modifications of this approach, thus allowing access tothe Z isomers as well.The hydrolysis of the aldol adducts to the free acid was reported to be highly efficientusing lithium hydroperoxide. However, it was doubtful that the same would apply to thecorresponding oç3-unsaturated adduct. Thus, it seemed advisable to perform the dehydrationstep following adduct hydrolysis. In order to prevent side-reactions such as ketene formationin the dehydration step, the carboxylic acid would then be protected as the methyl ester withdiazomethane. Subsequent dehydration via elimination of the corresponding mesylate andsaponification should then afford the free acid. Indeed, the results for compounds bearingunbranched substituents at C(2) demonstrate that this overall approach was successful for theE isomers. The only apparent failings of this method were in the case of compound 5, whichwas found to undergo E/Z isomerization upon basic hydrolysis, and the branched side-chain133compounds that could not be induced to undergo methanesulfonyl chloride-mediateddehydration in any but the lowest yields.The previously unreported syntheses of the other analogues are unremarkable, beingbased on either a similar approach starting from a 13-keto ester or a routine WadsworthEmmons reaction. As mentioned, acids 3, 9 and 10 were prepared as described in theliterature.4.2. ANTICONVULSANT EVALUATION AND CORRELATION WITHPHYSICOCHEMICAL PROPERTIESThe next step was to evaluate the anticonvulsant potencies of compounds 2-17 as theirsodium salts and compare these values with that of VPA. The test used was the standardsubcutaneous pentylenetetrazole injection administered 10 mm following the drug (Swinyardet a!., 1989). We found that the convulsant alone produced seizures in 87% of all animals,which was in agreement with the test’s objective of evaluating how the drugs influenceseizure threshold rather than seizure spread, as would be the case with the MES test (Pireddaet al, 1985). The evaluation of VPA and its analogues in vivo supported earlier findings(Tables 3-4) that increasing the size of the aliphatic group attached to a carboxylic acidfragment will also increase anticonvulsant potency. This was shown especially clearly in thecases of the 2-substituted 2-pentenoic acids (compounds 2 and 5-8) and thecycloalkylideneacetic acids (compounds 14-17). Quantitative support for this phenonomenonis provided by equation (21), depicted graphically in Figure 38, which shows a high degree ofcorrelation between log(ED5O) and volume. This relationship largely persists whenlog(ED5O) is replaced by the logarithm of the brain concentration (Figure 40) although thecorrelation parameters are clearly inferior. The situation could not be improved significantlyby the further introduction of shape parameters Y, Y and zY describing the C(2)substituent, chosen on the basis of their independence of volume/log P. This suggests that thepharmacodynamics of the drugs are governed to a lesser degree by shape-independent134properties such as lipophilicity than in the overall situation where the pharmacokinetics ofdrug access into the brain must be considered. Although this evidence is less than robust, itsupports the involvement of a receptor site with some minor shape requirements that mediatesthe anticonvulsant effect.These results showing good correlations of anticonvulsant potency with volume andlipophilicity are in agreement with the previously mentioned studies of Abbott andAcheampong (1988) which examined a series of VPA analogues. Interestingly, Lien et al.(1979) reported similar correlations (r = 0.72-0.76) with lipophilicity and molecular weightfor a group of 12 hydantoins and barbiturates. The equations were unsuccessful inapproximating the potencies of diazepam and clonazepam, however, and the use of ED5Orather than actual brain concentration indicates that the equation may represent nothing morethan the relative abilities of the drugs in reaching the brain, rather than their interaction withtheir effector sites.What was the influence of a ring system on anticonvulsant potency? Comparingcyclic and acyclic compounds with equal numbers of carbon atoms indicates that theintroduction of a ring closure caused a reduction in activity most readily explained by paralleldecreases in both volume and lipophilicity as revealed by Tables 9 and 12. Thus, structure isapparently less important than the simple bulk properties of volume and lipophilicity. Thisrelationship is demonstrated particularly clearly by the C8 acids 2-ene VPA (7), 15 and 11with their varying number of ring systems.The finding that the volume and log P dependence of potency persists largelyundiminished when actual ED5O brain concentrations are evaluated is quite significant. Muchof the work to date (Section 1.2.3) has focused strictly on ED5O itself as the biologicalindicator of potency that does not allow for reliable conclusions about mechanisms of actionas it incorporates pharmacokinetic as well as pharmacodynamic terms. Thus, this studyclearly shows that a drugs lipophilicity is important in its interaction with the still unknownreceptor site as well as in its passage from the periphery into the brain.135Considering the chemical similarity of the compounds tested, it is relevant to askwhether or not a QSAR study is justifiable in this case. The only physicochemical termsinvolved here are highly correlated parameters describing volume and lipophilicity becauseall compounds are identical in the electrostatic or inductive properties of their substituents onthe basic a,f3-unsaturated carboxylic acid unit, with the minor exception of compounds 9 and10. Thus, is there enough diversity in the set to obtain a balanced picture of the cause of theanticonvulsant effect? We believe the answer to be in the affirmative if the ranges oflipophilicities (Table 12) and potencies (Table 5) are considered. For example, thecompounds vary from those that are nearly water-miscible in the undissociated form(compounds 2 and 3) to one that is barely water-soluble even as its sodium salt (compound17). Similarly, ED5O potencies vary from 0.7 to 7.0 mmollkg, demonstrating that acids 2-17represent a diverse group of physical and biological, if not chemical, properties.More importantly, our primary objective here was to study the interaction of the drugswith the effector leading to seizure protection. For this assumption of a common mechanismto hold, it is evident that the tested analogues must not stray far from the fundamentalstructure of 2-ene VPA. The simplicity of this molecule clearly restricted allowedsubstituents on the oc,13-unsaturated carboxylic acid moiety to those of an aliphatic non-polarnature, although some liberties were taken by the use of cyclic, branched or unsaturatedsubstituents. Introduction of electrostatic charges or strongly electron-withdrawing species,for example, would profoundly alter the properties of the molecule that could lead to alteredpharmacokinetics (notably transfer across the blood-brain barrier) and quite. possibly differentpharmacodynamics as well. When combined with the fact that the scPTZ anticonvulsant testis fairly non-specific (recall that the mechanistically unrelated drugs ethosuximide anddiazepam are both effective here: Table 2), this could clearly lead to considerable ambiguityabout how a drug is exerting its anticonvulsant effect. Consequently, a correlation equationwith ED5O or [drug]j for a group of pharmacologically diverse compounds would nolonger carry implications about the mechanism through which seizure protection was beingachieved.136It should be noted that all tested compounds were achiral with the exception of acid 11which had been prepared from (±)-norcamphor and was thus a racemate. This clearly raisesthe question of the relative pharmacokinetics and phannacodynamics of the two enantiomers,but is likely largely academic in view of the findings of Hauck et al. (1991) that show a lackof enantioselectivity in the anticonvulsant properties of various asymmetric VPA analogues.A note of caution about the data presented in Table 5 concerns the reliability of theED5O values. It is important to realize that the experiment is subject to a high degree ofvariability, as is clearly demonstrated by the large ED5O confidence intervals. In some cases,in fact, the data collected from the group of 40 mice for a given drug gave a plot with aninfinite confidence interval, requiring that the entire experiment be repeated. This variabilitywas also the reason for routinely omitting data points, despite the already low initial number,in an effort to obtain meaningful results from an experiment. This editing was unavoidable insome experiments because the percent protection is obviously a linear function of log(dose)across only a limited range of doses. That is, there will be two continuums of dose rangesaffording zero and full protection, respectively. Elimination of data points will influence thereliability of the result, making it vital to consider the respective confidence intervalswhenever comparing potencies. Finally, this variability would be expected to affectsubsequent experiments involving an initial ED5O dose. For example, one would not expectto observe a small coefficient of variation for a synaptosomal GABA increase following anED5O dose with a large confidence interval, especially considering that ED5O by definition isan inherently uncertain quantity (e.g. only half of the animals receiving this dose will beprotected against seizures) made even more so when small values of n are involved. Were itnot for the comparable coefficients of variation for ED5O and brain levels, the fact that suchsmall variations were routinely found in the GABA and GAD assays would suggest that theprincipal cause of ED5O variation lies downstream from GABA production, perhaps in theform of individual differences in GABA receptor densities or neural architecture.Whereas VPA remained as one of the most effective overall anticonvulsant drugs asmeasured by its ED5O value, its intrinsic potency, that is inversely related to its brain137concentration at 15 mm following an ED5O dose (Table 5), was remarkably low incomparison with a large number of other analogues, including 2-ene VPA. This was shownby the compact group of analogues with ED5O values less than 1 mmollkg, whose intrinsicpotencies must be very good indeed if they could afford anticonvulsant protection equal tothat of VPA at essentially the same dose but at substantially lower effective brainconcentrations. The therapeutic advantage of VPA must therefore be derived mainly from itsability to be efficiently transported into the brain rather than from an exceptionally effectiveinteraction with its final neurochemical target. A truly significant reduction in ED5O can thusbe achieved only by dealing with the issue of blood-brain barrier transport as well as intrinsicpotency. Specifically, a rational step to further improve the effectiveness of the most potentcompound, acid 17, would not be to increase volume or log P, for example by increasing thesize of the ring or adding further substituents, but rather to incorporate the molecule into aprodrug such as a diacylglyceryl ester (Mergen et al., 1991), which has been shown to be anefficient carrier of VPA into the brain. The need to consider primarily pharmacokineticsrather than intrinsic potency is perhaps best illustrated by the considerable difficultiesencountered in preparing an aqueous solution of compound 17 as its sodium salt for an ED5Odose, which demonstrated that lipophilicity cannot be increased significantly beyond thispoint if aqueous solubility is to be retained.Dividing the ED5O brain concentration by the ED5O dose was taken as an index of thedrug’s ability to penetrate into the brain from the initial site of injection at its therapeutic dose(Q: Table 8). This value represents a drug’s instantaneous distribution ratio between the twocompartments because the system is not at equilibrium. Although the results appearrandomly distributed at first, some distinct trends do emerge. First, VPA clearly is veryefficient in reaching the brain, as noted above. This is likely due to the lack of a double bondconjugated with the carboxyl moiety, although it is less clear how this feature is expressedbiochemically. It may represent a lower degree of plasma protein binding due to alteredcharge densities on the carboxyl group or reduced dissociation leading to a greater freeconcentration and thus a potentially larger gradient with the CSF compartment. This notion is138undermined by the fact that the dienes 9 and 10 have above-average values of Q despite beingthe most acidic compounds in the table (PKa VPA = 4.95; PKa 7 = 4.36; PKa 9/10 = 4.02:Abbott and Acheampong, 1988). Also, the unsaturated analogues 3 and 11 have Q valuescomparable to that of VPA itself. Alternative explanations for the facile transport of VPAacross the blood-brain barrier are hard to conceive as these compounds have very similarpotencies and physicochemical properties (such as shape and lipophilicity).The remaining compounds appear to have similar Q values within their own subgroups with an inverse relationship with molecular weight that suggests increased difficultiesin penetrating the blood-brain barrier for the larger molecules. There is no apparentdifference between the 2-pentenoic acids (compounds 5-8) and the cyclic analogues(compounds 14-17) when one compares compounds with identical numbers of carbon atoms.Compounds 11-13 are unusual in exhibiting a wide range of Q values although they possessconsistently compact and rigid structures. For example, it is difficult to see how 12 and 13could behave so differently despite their structural similarity.A comparison of Tables 5 and 9 reveals an appreciable correlation between a drug’ssedative properties and the number of conformers occurring within 10 kJ of its globalminimum. Thus, the greater a molecule’s floppiness, the greater its accompanying sedation.This is shown clearly when one compares the 2-substituted-2-pentenoic acids, which showrapidly increasing sedation with size, with the cyclic compounds, which show noticeablesedation only at the level of the flexible cyclooctylidene ring. The literature generallysupports this trend but not without some exceptions. While both Loscher and Nau (1985) andLiu and Pollack (1994) reported that the neurotoxicity of l-methyl-1-cyclohexylcarboxylicacid was somewhat less than that of VPA, the latter workers noted that the closely-related 1-cyclohexylcarboxylic acid was only about one half as toxic as VPA. Furthermore, Scott et al.(1985) found that while the large rigid molecule spiro[4.6]undecane-2-carboxylic acid (Figure2) gave an scPTZ ED50 value of 0.42 mmollkg, its neurotoxicity was about twice that ofVPA. This suggests that sedation may simply be a function of volume/lipophilicity thatincreases in parallel with conformational lability. This idea would account for the seemingly139contradictory results presented in Table 9 that while 16 and 17 have an equally high numberof conformations, only the latter shows sedative effects. However, the acid 16 has a greatervolume/lipophilicity than 2-ene VPA and yet only the latter is noticeably sedating. Onepossible explanation is that it is the spatial range of the conformational interconversions thatmust be considered, rather than simply the volume for a given conformation. Specifically, thecoordinates of given carbon atoms in compound 16, unlike those in 2-ene VPA or compound17, do not show appreciable variation between conformations, resulting in a relatively smalltime-averaged volume. Therefore, the decreased toxicity of compound 16 may in fact stemfrom the corresponding reduction in volume, as noted previously.Nevertheless, it is clear that although both anticonvulsant potency and sedationincrease with volume, regardless of the structure, the presence of a ring system in themolecule can selectively diminish the dependence of sedation. Thus, a C10 homologue ofacid 8 would be profoundly sedating unless it was a cyclic molecule such as compound 17.As shown in Figures 38-41, the absence of outliers representing different structural classesindicates that there is no such relationship with anticonvulsant potency. This partialdissociation of sedation from molecular size was demonstrated above by 1-cyclohexylcarboxylic acid and VPA, the former showing half the neurotoxicity of the latterdespite their nearly identical molecular formulas.The molecular modelling calculations for compounds 5-8 also revealed that the Zisomers show much more conformational lability than the E isomers. Considering the strongdependence of the in vivo effects of these drugs on similar bulk properties such aslipophilicity, this may explain why the Z isomer of compound 7 has been reported to be lesspotent than the E form (Loscher, 1992). However, the potency difference might also besimply due to the shape resulting from the orientation of the double bond itself.Having shown that both anticonvulsant activity and sedation increase withsize/lipophilicity, it is pertinent to now ask what the actual effector site might be. Thepreviously mentioned studies of Perlman and Goldstein (1984) as well as the repeatedlydemonstrated effects of VPA on ion channel kinetics (Section justify proposing a role140for membrane disordering in the pharmacodynamics of VPA, analogous to the one for volatileanesthetics. Such a target site is supported by the work of Lucke et al. (1993) who found thatneither VPA nor 2-ene VPA readily cross the plasma membrane. Furthermore, it has beenreported that epileptic patients on VPA therapy exhibit markedly higher fluidity of theirerythrocyte membranes, showing the existence of an in vivo effect following normal doses(Tangorra et al., 1991).The specific advantage of the fluorescence polarization assay is that while it evaluatesthe lipophilicity of a drug, like the log P determinations, it does so in the anisotropicenvironment of a cell membrane. This is in marked contrast to the isotropic medium in ashake-flask or HPLC log P experiment. Thus, the fluorescence polarization method examinesthe actual biological target site rather than a synthetic model.In this project, erythrocyte ghosts rather than the synaptosomal membranes of Perlmanand Goldstein were employed for several reasons. One such issue was membrane purity:whereas the ghosts represent a single pure plasma membrane, synaptosomal membranesprobably contain contributions from the plasma membrane, mitochondria and endoplasmicreticulum. From a more practical viewpoint, erythrocyte ghosts are also preferable becausetheir preparation is more rapid and easily yields the amounts of membrane required for theassay. In contrast, our attempts to produce synaptic membranes by the laborious method ofJones and Matus (1974) consistently gave negligible amounts of the final product even withlarge amounts of starting material (16 g of brain from 8 rats).Unfortunately, the results of the membrane fluidity assays can only be described asinconclusive on account of the repeated failures to reproduce the initial encouraging data(Table 14) using the same conditions. We had hoped to be able to show a correlation betweenanticonvulsant potency and/or sedation but, even if one accepts the data in Table 14 as valid,it is evident that no such relationship exists.Results notwithstanding, there remain serious questions about the suitability of theapproach used. For example, there is the issue of the very high concentrations of drugs usedin the experiment due to the inherently low potency of the analogues in this assay. Second,141there is the issue of the similarity of erythrocyte membranes to those of a neuron. Humanerythrocyte membranes have a cholesterollphospholipid molar ratio of about 0.92 (Cunnane etat., 1989; Baldini et at., 1989; Sumikawa et at., 1993; Tangorra et at., 1991), compared withabout 0.54 for murine synaptic plasma membranes (Hitzemann and Johnson, 1983; North andFleischer, 1983; Koblin et at., 1980). Thus, it is possible that the erythrocytes’ membranefluidity was too high for a wealdy membrane-disordering agent such as VPA to exert anappreciable effect, whereas this influence might have been readily apparent in a morestructured system such as a synaptic plasma membrane. While it might therefore beinappropriate to consider erythrocyte ghosts as behaving identically to synaptic plasmamembranes, this difference in membrane composition would not be expected to significantlyinfluence the relative membrane disordering potencies of the drugs. Finally, one might askwhether a membrane protein could be sufficiently affected by fluidity changes far too subtleto be detected by the present method, especially in light of its high degree of associated error.For example, the standard error of the anisotropy alone in Table 14 is about ± 0.002. Thisresult contrasts sharply with those of Perlman and Goldstein, who reported standard errors foranisotropy differences on the order of ± 0.0005 or less using a similar number of samples.4.3. EFFECT OF 2-ENE VPA ANALOGUES ON GABA LEVELSThe best-proven mechanism for VPA’s anticonvulsant effects to date remains itsenhancement of brain GABA. Prior to studying the effects of our drugs on GABAmetabolism, however, the development of a simple but sensitive GC-MS-based method wasrequired to evaluate how therapeutic doses of these drugs would affect the neurotransmitterpool of GABA located in the nerve terminals. Monitoring drug-induced changes in GABAlevels in nerve terminals, rather than whole tissue, has the advantage of reducing interferencefrom metabolic poois of GABA that do not participate in seizure protection and thus providesa better reflection of the pharmacodynamics of the drug (Gale, 1992).142There are numerous reports in the literature describing the analysis of GABA by GCMS (for example, see Schaaf et al., 1985 and references cited within), the most cormrion ofwhich is based on the derivatization of GABA with pentafluoropropionic anhydride andhexafluoroisopropanol developed by Bertilsson and Costa (1976) to assay for GABA andglutamate in rat cerebellum. Although its sensitivity is good (detection limit about 24 fmolinjected GABA derivative: Schaaf et al.), this method employs fairly low molecular mass (m/z204) ions for selective ion monitoring in electron ionization mode that increases the risk ofinterference from other species. Furthermore, the length of the derivatization procedure, thelow stability of the product (Singh and Asbraf, 1988) and the formation of acidic by-products,that may degrade the column’s stationary phase, make this approach less than ideal for largenumbers of samples.An alternative derivatization procedure is silylation of GABA to N,N, 0-tris(trimethylsilyl)-GABA (Cattabeni et at., 1976). This derivatization and associated GC-MSassay overcomes the drawbacks of the Bertilsson and Costa method although the detectionlimit (60 fmol) is inferior. The silylation protocol has subsequently been modified by the useof the versatile reagent MTBSTFA to afford N, O-bis((tert-butyl)dimethylsilyl)-GABA(Mawhinney et at., 1986; Kapetanovic et at., 1990, 1993). However, these procedures usedeither aqueous standards or tissue homogenates and did not fully evaluate the recovery,variability and detection limit.The reliability of the GC-MS portion of the method was demonstrated by thefavourable coefficients of variation both for intra-day and inter-day experiments. Similarly,there was good agreement for the GAD activities between completely separate assays. Thiswas accompanied by a reasonable recovery of internal standard (78%) for the GABA assay.The value for the GAD assay, where the GABA-d6is added at the quenching step rather thanduring resuspension of the synaptosomal pellet, would likely be higher still. Finally, it isworth re-emphasizing the value of this method in a project where many hundreds of samplesneed to be analyzed in a time-efficient manner.143Using this method, the GABA concentration in a control synaptosomal preparationwas found to be 20.0 nmollmg protein, which compares favourably with published values(Table 15) and, in conjunction with the 19.4 nmollmg obtained in a separate experiment for 2h postdose, demonstrates the validity and reliability of the synaptosome preparation protocoland subsequent analysis. An ED50 dose of VPA elicited a significant increase insynaptosomal GABA 15 mm postdose, demonstrating that the drug affects GABAergicneurotransmission at a therapeutic dose. Our values for VPA and 2-ene VPA comparefavourably with the respective data of 22.1 ± 2.3 nmollmg and 21.6 ± 2.2 nmo]Jmg for 30 mmpostdose reported by Loscher et al. (1981) for VPA (1.18 mmollkg) and 2-ene VPA (1.41mmollkg), although the magnitude of our observed increase is significantly smaller (VPA:11% vs. 33%; 2-ene VPA: 10% vs 30%). The increase in GABA levels was temporallyconsistent with the onset of seizure protection. As the median time to the first seizure in thePTZ test was also about 15 mm, this finding implies that the drug had reached therapeuticallyeffective levels in the brain by this time and had begun to exert its anticonvulsant effect.While this does not prove a causal relationship between the two, it does at least provide betterproof than the 30 mm postdose data of Loscher et al.In order to investigate the time course of this GABA increase for a few drugs ofparticular interest, as well as to provide a further check on the reliability of the data, thesynaptosomal GABA levels for VPA and compounds 7-10 at 30 mm following an ED5O dosewere also examined (Table 17). The results indicate a sustained elevation of GABA levelsconsistent with the demonstrated continuing seizure protection at this time point. As a furthertest of reliability it was also shown that the GABA increase is dose-dependent (Table 18).On the assumption that drugs 1-17 all exert their anticonvulsant effect by an identicalGABAergic mechanism, we expected to find that equipotent (i.e. ED5O) doses of thesecompounds produced identical increases in GABA. Although at 15 mm postdose both VPAand 2-ene VPA showed significantly elevated GABA concentrations, the overall values werehighly variable (20.7 ± 1.9 nmollmg protein), with many compounds showing no significantchange and several others showing a marked decrease. This apparent lack of consistency in144the GABA data could also be explained more simply by the low sample number and the factthat the individual experiments were conducted separately without a simultaneous control.Furthermore, because subtle changes in GABA concentrations were being sought followingadministration of a drug dose (e.g. ED5O) incorporating a very large uncertainty in itsbiological endpoint, there was no guarantee that a drug would actually convey seizureprotection to a given mouse. This situation could have been improved by using higher doses(up to 4.2 mmol/kg for mice are routinely employed in the literature) but at the cost ofphysiological relevance. Despite the variability, though, the data appears to be sound asdemonstrated, albeit indirectly, by the 30 mm post-dose GABA assays that are all fullyconsistent with their 15 mm counterparts.The resultant question of how anticonvulsants expected to exert their effect throughpromotion of GABAergic transmission can do so while decreasing levels of thisneurotransmitter presents a strong, though not definitive, argument against our centralassumption. Furthermore, it is valid to ask if GABA is responsible for the anticonvulsantproperties of VPA itself, given the diversity of effects on the GABA system produced by suchclosely-related compounds.The hypothesis of GABA’s dominant role in the anticonvulsant effect of VPA hasbeen questioned occassionally. Perry and Hansen (1978) founi no effect on GABAconcentrations when VPA was administered orally to rats for 7 d at 315-365 mg/kg/d,although this could be due to their analysis of whole brain homogenate rather than morespecific preparations. However, this is not in agreement with other studies using rodent brainhomogenate (Simler et al., 1968; Iaclarola et at., 1979). In fact, some workers have evenfound increases in GABA in the CSF (Zimmer et a!., 1980; Loscher and Siemes, 1984, 1985)and plasma (Loscher and Schmidt, 1980, 1981) of patients on VPA therapy. Second, it hasbeen reported that VPA exerts clear anticonvulsant effects at i.p. doses that fail to influenceGABA levels (Anlezark et al., 1976). Most recently, Wolf and Tscherne (1994) found that0.56 mM VPA applied via a push-pull cannula both outside and inside the pars reticulata ofthe substantia nigra in rats significantly decreased GABA release into the perfusate.145However, this decrease was judged to be unaccompanied by modified GABAergictransmission as the animals failed to show the circling behavior characteristic of unilateralGABA transmission in the extrapyramidal motor system. The authors thus concluded that theaction of VPA was to simultaneously suppress presynaptic nigral GABA release and tosynergistically enhance the post-synaptic response to released GABA.Hackman et at. (1981) also failed to observe any enhancement of GABA releasedfrom frog lumbar spinal cord slices, as did Farrant and Webster (1989) using a push-pullcannula embedded in the substantia nigra of VPA-dosed rats. A similar conclusion regardingGABA release was also reached by de Boer et at. (1982) using rat cortical slices, as well asby Abdul-Ghani et at. (1980) who examined superfused sensorimotor cerebral cortex.Unfortunately, the latter study was marred by these workers’ inability to detect GABA in thecontrol superfusion fluid and their very low readings even for rats pretreated with the GABAT inhibitors y-vinyl GABA and ‘y-acetylenic GABA that increase GABA levels four- to fivefold at the concentrations used (Schechter et at., 1977). Wolf et at. (1988), on the other hand,reported that perfusion of the preoptic area of rats with 0.3-1.4 mM VPA produced a highlysignificant decrease in GABA release. The magnitude of this effect decreased with increasingVPA concentration until, at 11 mM, enhancement of GABA release was observed. Apotentiation of the post-synaptic GABA response leading to decrease of GABA release bynegative feedback was suggested.To further cloud the issue, Biggs et at. (1992) recently reported that VPA has abiphasic dose-dependent effect on extracellular GABA levels measured by in vivomicrodialysis. They found that while 400 mg/kg i.p. VPA doubled basal GABA levels, 200mg/kg had no effect and 100 mg/kg actually cut the level in half.Finally, no discussion on VPA and GABA would be complete without examining therole of VPA in treating absence seizures. While VPA is an effective drug for absenceseizures both in humans and genetic rodent models, there is good evidence that seizures in thelatter are in fact promoted by GABAergic agonists. Vergnes et at. (1985) reported that VPApotentiated seizure enhancement by the GABAA agonist THIP in the genetic Wistar rat model146despite VPA being an effective anticonvulsant alone. Similar observations were made bySmith and Bierkamper (1990) using a different rat model wherein the animals werechronically treated with the cholesterol biosynthesis inhibitor AY-9944. They found thatGABA agonists induced seizures, whereas GABA antagonists such as bicuculine andpicrotoxin, ethosuximide and the clinical anti-absence benzodiazepines diazepam andclonazepam had protective effects. VPA was an anticonvulsant at 30 mg/kg i.p. but enhancedseizure occurence at 60 and 120 mg/kg. In an attempt to explain this biphasic effect of VPAas well as the behavior of the benzodiazepines, whose fundamental mechanism of action iswell-established, the authors proposed the existence of two distinct pools of GABA withdifferent sensitivity to such anticonvulsants: a synaptosomal one concerned withanticonvulsant effects, and a larger pool that promotes seizures. Nevertheless, they did notrule out the possibility that the anticonvulsant mechanism of VPA in absence seizures isdistinct from its GAl3Aergic properties.Concluding that the acids 1-17 could not be considered as a single homogeneousgroup of GABA-elevating agents, we sought to at least define some distinct trends in the dataof Table 17. First, the data as a whole was examined using a correlation matrix (Table 19),but no links to any of the previous physicochemical parameters were revealed. Next, welooked at the individual sub-groups. The 2-pentenoic acids (compounds 2, 5-10) show afairly consistent GABA increase although there are some irregularities with compounds 5 and6. Interestingly, the highly neurotoxic acid 9 elevated GABA to a greater extent than itsbenign isomer 10 at both 15 and 30 minutes, but it is doubtful that this difference was trulysignificant if one compares the absolute GABA concentrations. Thus, an unusually high levelof this neurotransmitter is unlikely to be the cause of the effects seen with compound 9although one cannot dismiss the possibility of the drug’s selective action on GABAergictransmission a particular brain region. In fact, Lee (1991) has shown that compound 9 isfound in particularly high concentrations in the substantia nigra, which is a major source ofGABAergic fibers. Thus, if the observed GABA increase with this drug treatment is confinedto this region, the local augmentation might be sufficiently large to produce the noted147neurotoxicity. However, this does not agree with the previously discussed finding of Wolfand Tscherne (1994) that VPA applied directly into the substantia nigra had no apparenteffect on behavior.The other principal sub-group, the cyclic acids 14-17, shows a noticeable, thoughpossibly coincidental, trend in GABA levels. Specifically, GABA levels dropped evenlyfrom an initial marked increase (compound 14) to a profound decrease (compound 17),suggesting that while the drugs may be interacting with the effector site in a structure-dependent manner, the effect could simply be the result of their declining brain concentration.This may also represent influences on a non-neurotransmitter pooi of GABA that is notdirectly involved in seizure protection (Gale, 1992). In fact, such a result is not inconsistentwith the literature as Smith and Bierkamper (1990) have shown that VPA itself, whichnormally targets the neurotransmitter pool, exerts a biphasic effect in a rat absence epilepsysuggestive of both a GABA agonist and antagonist properties. Similarly, one can consider thecase of the benzodiazepines diazepam and clonazepam. Although both would presumably actexclusively at the GABA receptor, the former is used solely for status epilepticus and thelatter only for absence seizures (Rall and Schleifer, 1990), illustrating the diversephysiological roles of GABA. The difference between these two drug types is furtherdemonstrated in Table 2.This subdivision of GABA may be regional as well as functional. Although neitherLee (1991) nor Loscher and Nau (1983) found significant localization of VPA or 2-ene VPAin rat brain, Loscher and Vetter (1985) observed widely varying degrees of GABA elevationin a study examining this property in the synaptosomal fraction of 11 different brain regionsin rats administered 200 mg/kg i.p. VPA. Even at 5 mm postdose, they noted increases from60-80% in the hippocampus, hypothalamus, and tectum down to essentially none for thestriatum, medulla and, oddly enough, substantia nigra. This variation did not appear to bederived from different control levels as these were all quite similar, indicating the existence offunctionally distinct (e.g. VPA-sensitive vs. -insensitive) pools of GABA.148The influence of 2-ene VPA on whole-tissue GABA levels in various brain regionswas markedly different in the study of Weissman et al. (1978). Specifically, these workersfound a broad range of increases from 10% in the hippocampus to 80% in the substantianigra. In view of the even distribution of this compound, it can be concluded that themagnitudes of the GABA increases appear to be the result of specificity for certain GABApools rather than the drug concentration in a particular region.This heterogeneity of drug-GABA interactions can be used to explain the apparentdecrease of the levels of this neurotransmitter obtained in the experiments with acids 5, 6, 16and 17. First, a change in nerve terminal GABA does not carry any specific implications forthe traffic across the synapse (e.g. GABA release), which would be the ideal measure of theinfluence of VPA and its analogues. Instead, this experiment considered the sum of newly-synthesized GABA, GABA taken up from the synapse and GABA present in non-transmitterpools (Jadarola and Gale, 1981; Wood et al., 1988), in addition to GABA from contaminatingglial and neuronal cell body fragments. Drug inhibition of GABA re-uptake, for example,would result in an enhanced trans-synaptic signal that would be recorded as a GABA decreasein our procedure. Alternately, the overall reduction in GABA release might be very real butmight be accompanied by increases in specific brain regions that could still discourage seizureactivity if appropriately localized, as suggested by the results of Weissman et al. (1978). Itcould even be argued that seizure protection was the result of a suitably located decrease inGABAergic output. Such mechanisms for the action of the aforementioned drugs are ofcourse hypothetical but at least provide an explanation of how a decrease in GABA need notbe incompatible with an anticonvulsant effect.While our GABA measurements alone cannot be used to infer a specificanticonvulsant mechanism for drugs 1-17, it is at least clear that they do not all function in thesame manner. They may all still act via GABA but the variable change in the levels of thisneurotransmitter implies there is no single unique mode of interaction.The central assumption of GABA playing the critical role in the expression of ananticonvulsant effect is worth examining: could not other neurotransmitters also be involved?149The scattered studies performed to date on the possible role of such alternateneurotransmitters have been generally negative. Zeise et al. (1991) reported that neuronalresponses to glutamate were unaffected by VPA in vitro but transient depolarizations evokedby N-methyl-D-aspartate were suppressed by 0.1-1 mM VPA. Acute VPA also promotes adecrease in glutamate levels in the hippocampus and striatum but an increase in the neocortexand cerebellum (Chapman et at., 1982). Likewise, it is known that VPA decreases the levelsof the excitatory amino acid aspartate in rats and mice (Chapman et at., 1982; Sarhan andSeiler, 1979), albeit at high (200-400 mg/kg) doses, but the fact that this change is confined tothe non-synaptosomal compartment implies interaction with a metabolic rather thanneurotransmitter pool of aspartate. Chronic administration of VPA (100 mg/kg i.p. for 3weeks) failed to influence the uptake or binding of aspartate or glutamate (Slevin and Ferrara,1985).There is evidence for an increase in the inhibitory neurotransmitter glycine (MartinGallard et al., 1985; Similae et at., 1979). There have also been clear demonstrations of theenhancing effects of the drug on serotonin and dopamine levels by Horton et al. (1977) andothers (Hwang and van Woert, 1979; Whitton and Fowler, 1991; Biggs et at., 1991; Zimmeret at., 1980; Nagao et at., 1979). However, this study showed the anticonvulsant effect to benot only temporally uncorrelated with the concentrations of these neurotransmitters but alsoretained following administration of their respective biosynthesis inhibitors.However, these studies invariably sought to find a connection betweenneurotransmitter levels and anticonvulsant rather than sedative activity. While evidence forthe influence of these neurotransmitters on the former is weak, there is good support for theircontribution to the behavioral effects. For example, Horton et at. (1977) found a vaguetemporal correlation between behavior (as judged by degree of sedation) and brain levels ofthe serotonin metabolite 5-hydroxyindoleacetic acid. This is particularly interesting in viewof the propensity of modest doses of VPA (200 mg/kg), but not 2-ene VPA, to induce wet-dogshakes, that have been repeatedly linked to elevated serotonin concentrations (Pagliusi andLoscher, 1985). The relevance of these results to our data lies in both the clear dissociation of150sedative properties with GABA levels and the exceptional neurotoxicity of the diene 9,leading to the question of whether these discrepancies might be explained by the drugs’influence on serotonin levels.Alternately, seizure protection might not be the result of synaptic mechanisms at all,but rather an influence on the kinetics of membrane ion channels. This explanation isappealing on the grounds that such effects have been shown to occur at very lowconcentrations of VPA and that this is also the mechanism of several other anticonvulsants,namely phenytoin, carbamazepine and ethosuximide. Nevertheless, this idea does not explainthe proven increase in GABA (although this might be a purely secondary effect) andencounters some difficulties in explaining how one drug combines the channel-blockingproperties of both phenytoin (Na) and ethosuximide (Ca2j.4.4. INTERACTION OF 2-ENE VPA ANALOGUES WITH GLUTAMATEDECARBOXYLASEThus, GABA remains the most likely source of the anticonvulsant effect of VPA and2-ene VPA, although the situation is clearly more ambiguous for some other analogues. Themost obvious means of controlling synaptosomal GABA levels is through the activity of itssynthesizing enzyme GAD. Unfortunately, the regulation of this enzyme, beyond the apoIholo-GAD equilibria discussed earlier, is poorly characterized to say the least. For example,there is no evidence for direct and rapid regulation by membrane depolarization or any secondmessenger systems and no group has yet addressed the question of how VPA would interactwith GAD seeing that the Vmax and Km values of GAD activity are unaffected in vitro.There is an elevation of holoenzyme activity in synaptosomes upon incubation in a highK+medium, but the effect takes at least 20 mm to become apparent (Miller and Walters, 1979).Similarly, aCa2-dependent enhancement of GAD activity by K-induced depolarization ofstriatal slices was observed by Gold and Roth, 1979) but only after a pre-incubation of at least10 mm in the depolarizing medium.151A solution to this question of control may be the long-known property of GAD to bindirreversibly to membranes in the presence of calcium, albeit at the unphysiologicalconcentration of 1 mM. Nathan et at. (1994) have recently demonstrated that GAD occurs inboth soluble (SGAD) and membrane (MGAD-I, MGAD-ll, MGAD-ffl) forms in the porcinebrain. MGAD contributed 53% of the total brain homogenate GAD activity, with each formrepresenting about one third of this value. Of particular interest was their fmding thatMGAD-ffl binds to membranes in a reversible calcium-dependent manner, rising to a plateauat 10 uM calcium, indicating that MGAD-Ill, unlike MGAD-I and MGAD-ll, is a peripheralrather than an integral membrane protein. Nathan et at. thus proposed that because GAD isapparently activated by dephosphorylation by a Ca2-dependent phosphatase (Bao et at.,1993) and a subpopulation of GAD is associated with synaptic vesicles (McLaughlin et at.,1975), the Ca2 influx accompanying the arrival of the action potential at the nerve terminalmay simultaneously activate the enzyme and favor its association with the delivery site of itsproduct. All this is, of course, highly speculative because the MGAD proteins have not beensequenced and it is uncertain how, or even if, they correspond to the soluble GAD65 andGAD67 isoforms mentioned above. Nevertheless, a GAD enzyme sensitive to the calciumlevels expected to follow depolarization of the nerve terminal membrane clearly offers severalpotential points of enzyme regulation.To retain the validity of our initial assumption of a common mechanism, such as theelevation of GABA as a result of GAD activity enhancement being the key to anticonvulsantactivity, we would have expected to observe a fixed constant increase for both parametersbecause the ultimate endpoint (ED5O) was also being kept constant. As Tables 17 and 19show, however, the changes in GAD activity were neither constant nor did they correlate withthe changes in GABA levels (r = 0.44), although the two generally coincided in direction ifnot magnitude. Still, it is worrisome that while most of the significant GABA decreases wereaccompanied by a decrease in GAD activity, there is no case showing a significant increase inboth indices. This indicates that while GABA elevation might be the principal mechanism ofthe anticonvulsant effect, although this is by no means proven by our data, it is certainly not152achieved exclusively through the enhancement of GAD activity. Rather, it may be the resultof the drugs’ interacting with one or more additional enzymes or sites as well. One likelytarget site would be the GABA-degrading enzyme GABA transaminase (Figure 3), which hasbeen shown by Loscher (1993b) to be inhibited by VPA (200 mg/kg i.p.) in synaptosomalpreparations from numerous brain regions by up to 25% (substantia nigra). This was inmarked contrast to whole tissue samples, that were unchanged compared to the controls, andwas taken to prove that VPA was specific for the neuronal, rather than the extraneuronal,form of GABA-T. Thus, it may be that the resultant GABA level is the product of theinteractions of the drugs in Table 17 with multiple enzymes, such as GAD and GABA-T. Thepossibility that the drugs may be selectively acting on particular subspecies of GAD, such asthose distinguished by sequence or location, should also be considered. It is conceivable thatthe measured production of GABA involved in seizure protection is the domain of only asmall GAD sub-group whose change in activity cannot be detected when the entire populationis assayed. The possibility of selective localization of the drugs themselves is unlikely in thelight of the finding of Lee (1991), who found that at the time of peak brain concentration (30mm), a 150 mg/kg i.p. dose of either VPA or 2-ene VPA to rats gave essentially constant (±10%) levels of the drug in the ten brain regions studied. However, this does not rule out somekind of subcellular selectivity that might target the drugs to a specific pool of GAD in the cell.The potential for such selectivity is shown by the fact that GAD65 binds PLP more weaklythan GAD67. Evidence favouring such distinct forms of GAD was provided by Phillips andFowler (1982) who found that homogenates of five different regions of rat brain had verysimilar GAD activities but varied greatly in their response to 400 mg/kg i.p. VPA. Themedulla/pons, cerebellum and midbrain showed an approximately 22% increase in GADactivity but the striatum and cerebral cortex showed no significant change.How might the drugs be interacting with the enzyme? Do the most active compoundsshare any common structural features that might point to a specific binding site on theenzyme? Or could it be that they do not directly interact with the enzyme at all, but ratherexert their influence at some point further upstream in the pathway that translates the initial153action potential signal into an increased synthesis of GABA? These questions are difficult toaddress for several reasons. First, only a handful of the enzyme activities reach statisticalsignificance presumably because of the modest, but therapeutically relevant, doses used in theexperiment. Second, a trustworthy structure-activity relationship can be devised only if eitherGAD activity or drug concentration is kept fixed, which is clearly not the case in thisexperiment where the constant parameter was chosen to be anticonvulsant activity.Consequently, the individual changes in GAD activity might be as much the result of variabledrug concentrations as due to a particular physicochemical parameter.Having verified that therapeutic doses of some drugs influenced the activity of GAD,albeit in a variable manner, the project was concluded with investigations of this mechanism.In synaptosomes (Miller and Walters, 1979), depolarization with a high potassium mediumrequired at least 20 mm to influence the apo-/holo-GAD ratio, suggesting that a directmembrane effect is probably not responsible. This is unfortunate, as such a mechanismwould have the advantage of not requiring the drug to enter the intracellular compartmentwhich, as Lucke et al. (1993) have shown, is not a favourable process. A more promisingapproach would be alteration of PLP binding. Because about half of GAD exists in theapoenzyme form, this would represent an easy means of modifying the activity in eitherdirection.This idea of VPA analogues interacting directly with GAD receives only mixedsupport from the literature. While Bolanos and Medina (1993) reported that 1 mM VPA doesnot affect the kinetics (Vmax and Km) of crude or “partially purified” enzyme, Taylor et at.(1992) found that even 0.25 mM VPA enhanced the activity of a >90% pure enzymepreparation by 19% through an allosteric site. In either case, this still leaves the possibility ofan allosteric mechanism operating in vivo whose effects might be partially retained in vitro.For example, the drug could inhibit ATP from promoting the conversion of the quinoidintermediate (Figure 4) to apo-GAD, leading to an effective increase in the overall saturationof the enzyme population. Arguing in favor of such a possibility are the fact that this role canalso be performed by nucleotides other than ATP, demonstrating a lack of absolute154specificity, and that both VPA and ATP are anionic. Such an interaction would alter thedegree of saturation of the enzyme in the pre-assay environment that could persist into theassay. It is important to note that even though the standard assay used in Table 17 allegedlymeasures the total GAD activity, where the initial degree of saturation should be irrelevant,the combination of enzyme with substrate is by no means instantaneous. For example, Milleret al. (1977) showed that the total activity decreased upon modification of the initial brainhomogenate preparation conditions to those inhibiting conversion of apo-GAD to holo-GAD,thus producing a lower degree of PLP saturation in the fmal enzyme assay. Conversely, thissuggests that a greater initial degree of saturation would ultimately be manifested as a greaterenzyme activity.The results of the saturation assay suggested that the observed changes in GADactivity elicited by VPA, 2-ene VPA and compound 17 cannot be explained by alterations inthe binding of endogenous PLP. First, we found that the GAD activities in the control andVPA- and 2-ene VPA-treated frozen synaptosomes were in good agreement with ourpreviously obtained values (Tables 16-17), although the same could not be said for thoseobtained from mice dosed with compound 17. This raises the question of whether or not thedrug-enzyme-PLP interactions in this study were truly comparable to those observed in thestandard GAD assay (Table 17).The values for the saturation of GAD with PLP for the drug-treated samples were notsignificantly different from the control, possibly on account of the low value of n (Table 22).The obvious solution to this problem would have been to use more samples but this wouldhave resulted in highly variable processing times for individual samples. Furthermore, theexperiment is very laborious so repetition did not seem worthwhile in light of actualmagnitude of the change in saturation, if any.The numbers calculated for the ratio of holoenzyme to total GAD compare well withthose obtained by Miller et al. (1977), who found 43 ± 3% saturation (total GAD: 7.3 ± 0.5nmollmg/5 mm; holo-GAD: 3.2 ± 0.4 nmollmg/5 mm) for whole brain GAD from ratsdecapitated into liquid nitrogen. As our numbers indicate, there was no apparent effect on155PLP binding in vivo by VPA, but there was a suggestion of decreased PLP binding by 2-eneVPA and compound 17. Interestingly, both total and holoenzyme activities were alsodecreased relative to the control for the latter two compounds, which is intriguing in view oftheir previously observed inhibition of intact synaptosomal GAD activity and would appear tosupport our initial hypothesis of a mechanism involving altered PLP binding.At this point it was evident that our initial assumption of a common mechanism ofaction of drugs 1-17 was likely false. Consequently, even though VPA had been shown tohave no effect on GAD activity in vitro, this no longer implied that other drugs would actidentically and thus rationalized a study of the effects of compound 17 on the GAD activity invitro. This compound was chosen on account of its high anticonvulsant activity despiteprofound depression of GABA levels and GAD activity. Such behavior was not onlyparadoxical for an antiepileptic drug expected to promote GABAergic activity, but wassuggestive of a direct inhibition of the enzyme. Thus, we sought to ascertain whether or notthere was an effect on Vmax or Km. Accordingly, a double-reciprocal plot (Figure 54) wasconstructed using the standard saturating level of co-factor, variable glutamate concentrations,a short reaction time (in order to estimate the initial rate) and two concentrations ofcompound 17 (1 and 5 mM). The slopes yield a mean Km for glutamate of about 2.1 mM,that is not inconsistent with the approximately 1 mM cited by Martin and Rimvall (1993) anddemonstrates the GAD enzyme in this assay to be showing expected behavior. The graphindicates a concentration-dependent inhibition of GABA production by compound 17 thatwas noticeable even at 1 mM. This inhibition was likely non-competitive based on theconvergence of the three lines at the x-axis rather than the y-axis. The two K valuesobtained (Table 20) are in as good agreement with each other as might be expected with theminimal data obtained. Seeing that the mean value of about 9 mM is well above the effectivebrain concentration found following an ED5O dose for this compound (94 nmollg wet tissue,e.g. about 0.094 mM), one might question the relevance of this finding to the ex vivo results.These numbers could be reconciled by a selectivity of compound 17 for certain brain regionsor specific sub-populations of GAD. The former idea does not find much support in156previously mentioned studies showing an essentially homogeneous distribution of VPA and2-ene VPA throughout the brain. However, while the 94 nmoLIg refers to the concentration ofthe drug across all compartments (e.g. intracellular, extracellular and membrane), Lucke et at.(1993) have shown that VPA and 2-ene VPA do not readily cross membranes into the 80% ofbrain volume contained in intracellular spaces, a property that may be shared by compound17. Consequently, localized extracellular concentrations may be much higher. Furthermore,because the therapeutic effects of the drug likely involve only minor inhibition of thisimportant enzyme, rather than the 50% produced by the K1 concentration, it is evident that theconcentrations required for effective inhibition and those obtained following a normal dosemay in fact be quite similar.The alternative explanation invoking enzyme specificity is compatible with reportsthat VPA does not affect GABA levels (Loscher and Vetter, 1985) or GAD activities (Phillipsand Fowler, 1982) equally in all brain regions. Furthermore, it is now clear that GAD is not auniversally cytosolic enzyme with little functional diversity between the individual isoforms,as the reports of Nathan et at. (1994) and Christgau et at. (1992) have emphasized. Thus, itmight be that compound 17 selectively inhibits a GAD subspecies involved with a convulsant,rather than anticonvulsant, GABA pool as has been postulated for the seizure-promotingproperties of GABA agonists in rat models of absence epilepsy (Vergnes et at., 1985).A most interesting feature of Figure 54 is its implication of a non-competitive mode ofGAD inhibition. Combined with the apparent, though not statistically significant, decrease inthe enzyme’s saturation with PLP, this provides indirect evidence for a mechanism by whichbinding of co-factor to enzyme is inhibited. This might take the form of competition for thePLP site or compound 17 acting as an inverse agonist at the ATP site. The finding isnoteworthy in light of the reported interaction of VPA and various (R,S)-3-alkylglutamicacids at an allosteric site of the enzyme leading to increased Vmax without affecting Km ofglutamate (Taylor et at., 1992).1575. CONCLUSIONSThe structure-activity results are consistent with previous reports of increasinganticonvulsant potency with increasing molecular size/lipophilicity for saturated aliphaticcarboxylic acids. High degrees of correlation were obtained between activity (ED5O) andeither volume or log P, with the introduction of further parameters failing to significantlyimprove the correlation parameters. As it was not possible to find shape parametersindependent of volume/log P that also showed a linear relationship with activity, we wereunable to determine whether or not a distinct receptor site is involved. However, the fact thatthe correlation with volume/log P was largely undiminished when ED5O was replaced withthe actual brain concentration following an ED5O dose strongly argues against this receptorsite possessing distinct stereochemical requirements. If the drugs therefore interact with thissite in a shape-independent manner, we might tentatively identify the site as the plasmamembrane and the mode of action as one analogous to that of the volatile anesthetics. Wethus sought to find a correlation between a dmg’s potency and its ability to alter plasmamembrane fluidity as measured by fluorescence polarization spectroscopy, but were unable toobtain reproducible results.One analogue, cyclooctylideneacetic acid (compound 17), was significantly morepotent than the lead compound VPA while exhibiting only moderate sedative effects. Thispotency difference was even more pronounced when actual brain levels were compared,where compound 17 was found to be more than six times as potent as VPA. However, VPAretained a high affinity for brain tissue as indicated by the brain/body distribution ratio Q,suggesting that the key to further increase anticonvulsant potency is enhancement of drugdelivery into the brain as well as increased intrinsic potency. This was emphasized by thedifficulty in preparing aqueous solutions of compound 17, even as its Na salt, indicating thata further increase in drug lipophilicity would likely come at the expense of sufficient aqueoussolubility.158With regard to the means by which the anticonvulsant effect is exerted, there was onlymixed support for a link to increased levels of GABA following therapeutic doses. Asexpected, VPA and 2-ene VPA showed significant increases in synaptosomal GABA but themost potent analogue (compound 17), amongst others, showed a prominent decrease. Thissuggests either participation by other neurotransmitters or the drugs’ selectivity forsynaptosomal GABA in specific poois dedicated to various types of neurotransmitter actions.Such pools might be distinguished by their regional distribution or functions. GABA levelsfailed to show a correlation with sedative effects, indicating that their cause must be soughtelsewhere. Attempts to explain sedation as the outcome of altered cell membrane fluiditywere unsuccessful for reasons mentioned earlier.There were similar inconsistencies in the effects of the drugs on synaptosomal GADactivity, ranging from inhibition to activation and with many showing no significant effect atall. Enzyme activity failed to correlate significantly with either GABA level or anyphysicochemical parameter, although a positive correlation coefficient was at least obtainedin the former case. Again, these discrepancies might be accounted for by regional andlorfunctional pools of GAD, a subgroup of which might promote seizure protection but whosechange in activity might be masked by the overall value from the nerve terminals as a whole.There was weak evidence that the drugs interacted with GAD by influencingpyridoxal 5’-phosphate binding. Administration of ED5O doses of VPA, 2-ene VPA andcompound 17 failed to give significant changes in the extent of PLP binding, although therewere slight decreases for 2-ene VPA and compound 17.In vitro studies showed a clear inhibition of GAD with 1 mM of compound 17 and aK1 value of about 9 mM. The non-competitive nature of the kinetics suggests the PLP or ATPbinding sites as possible targets of the drug. Further studies would be useful in order todetermine the exact region as well as to compare compounds such as acid 17 with the knownGAD-activating 3-alkyl-4-aminobutanoic and 3-alkyl-4-glutamic acids (Taylor et at., 1992).1596. REFERENCESAbbott, F.S., Kassam, J., Acheampong, A.A., Ferguson, S., Panesar, S., Burton, R., Farrell,K., Orr, J. (1986) Capillary gas chromatography-mass spectrometry of vaiproic acidmetabolites in serum and urine using tert. -butyldimethylsilyl derivatives. J. Chrom.375: 285-298.Abbott, F.S., Acheampong, A.A. (1988) Quantitative structure-anticonvulsant activityrelationships of vaiproic acid, related carboxylic acids and tetrazoles. Neuropharmac.27: 287-294.Abdul-Ghani, A.-S., Coutinho-Netto, J., Druce, D., Bradford, H.F. (1980) Effects ofanticonvulsants on the in vivo and in vitro release of GABA. Biochem. Pharmac. 30:363-368.Abe, M., Matsuda, M. (1983) On the existence of two GABA pools associated with newlysynthesized GABA and with newly taken up GABA in nerve terminals. Neurochem.Res. 8: 563-573.Acheampong, A.A. (1985) Quantitative structure-anticonvulsant activity of valproic acidanalogues. Ph.D. thesis, University of British Columbia.Agopyan, N., Avoli, M., Rieb, L, Tancredi, V. (1985) Depression of hippocampal lowcalcium field bursts by the antiepileptic drug vaiproic acid. Neurosci. Lett. 60: 57-62.Anlezark, G.M., Horton, R.W., Meldrum, B.S., Stephenson, J.D. (1976) Proceedings: yaminobutyric metabolism and the anticonvulsant action of ethanolamine-O-sulphateand di-n-propylacetate. Brit. 3. Pharmacol. 56: 383P-384P.Balcar, V.J., Mandel, P. (1976) Inhibition of high-affinity uptake of GABA by branched fattyacids. Experientia 32: 904-905.Baldini, P., Incerpi, S., Lambert-Gardini, S., Spinedi, A., Luly, P. (1989) Membrane lipidalterations in erythrocytes from IDDM and NIDDM subjects. Diabetes 38: 825-831.Baldino, F., Jr., Geller, H.M. (1981) Sodium vaiproate enhancement of y-aminobutyric acid(GABA) inhibition: electrophysiological evidence for anticonvulsant activity. 3.Pharmac. Exp. Ther. 217: 445-450.Baker, J.K., Rauls, D.O., Borne, R.F. (1979) Correlation of biological activity and high-pressure liquid chromatographic retention index for a series of propranolol, barbiturateand anthranilic acid analogues. 3. Med. Chem. 22: 1302-1306.Baruzzi, B., Bondo, B., Bossi, L., Castelli, D., Gerna, M., Tognoni, G., Zagnoni, P. (1977)Plasma levels of di-n-propylacetate and clonazepam in epileptic patients. Int. J. Clin.Pharmac. 15: 403-408.Bao, J., Nathan, B., Wu, J.-Y. (1993) Regulation of brain L-glutamate decarboxylase byprotein phosphorylation and de-phosphorylation. Soc. Neurosci. Abstr. 19: 1385.Bertilsson, L., Costa, E. (1976) Mass fragmentographic quantitation of glutamic acid and yaminobutyric acid in cerebellar nuclei and sympathetic ganglia of rats. 3. Chrom. 118:395-402.160Biggs, C.S., Fowler, L.J., Pearce, B.R., Whitton, P.S. (1991) The effect of sodium vaiproateon dopamine and 3,4-dihydroxyphenylacetic acid levels in rat hippocampal dialysatesin vivo. Brit. J. Pharmacol. 104: 62P.Biggs, C.S., Pearce, B.R., Fowler, L.J., Whitton, P.S. (1992) The effect of sodium valproateon extracellular GABA and other amino acids in the rat ventral hippocampus: an invivo microdialysis study. Brain Res. 594: 138-142.Blume, H.W., Lamour, Y., Arnauld, E., Layton, B.S., Renaud, L.P. (1979) Sodium di-npropylacetate (vaiproate) action on single neurons in rat cerebral cortex andhippocampus. Brain Res. 171: 182-185.Bolanos, J.P., Medina, J.M. (1993) Evidence of stimulation of the y-aminobutyric acid shuntby valproate andE-z2-valproate in neonatal rat brain. Mol. Pharmac. 43: 487-490.Boorman, E.J., Linstead, R.P. (1935) Investigations of the olefinic acids. Part XVI. Additivereactions and tautomeric changes of cyclic unsaturated acids and analogousobservations on -methylpentenoic acids. J. Chem. Soc. 258-267.Brenner, M., Huber, W. (1953) Synthesis of tx-amino acid esters. Helv. Chim. Acta 36: 1109-1115.Buchhalter, J.R., Dichter, M.A. (1986) Effects of vaiproic acid in cultured mammalianneurons. Neurology 36: 259-262.Buckingham, J., (Ed.) (1982) Dictionary of Organic Compounds, Vol. 2, 5th ed., Chapmanand Hall, New York, p. 1403.Carraz, G. (1967) Theory of the activity of the di-n-propylacetic acid structure. Agressologie8: 13-20.Cattabeni, F., Galli, C.L., Eros, T. (1976) A simple and highly sensitive massfragmentographic procedure for y-aminobutyric acid determinations. Anal. Biochem.72: 1-7.Chapman, A.G., Riley, K., Evans, M.C., Meidrum, B.S. (1982) Acute effects of sodiumvaiproate and y-vinyl GABA on regional amino acid metabolism in rat brain:incorporation of 2-[’C]glucose into amino acids. Neurochem. Res. 7: 1089-1105.Chapman, A.G., Croucher, M.J., Meldrum, B.S. (1984) Anticonvulsant activity ofintracerebroventricularly administered valproate and valproate analogues. A dose-dependent correlation with changes in brain aspartate and GABA levels in DBAI2mice. Biochem. Pharmac. 33: 1459-1463.Chang, G., Guida, W.C, Still, W.C. (1989) An internal coordinate Monte Carlo method forsearching conformational space. J. Am. Chem. Soc. 111: 4379-4386.Christgau, S., Aanstoot, H.-J., Schierbeck, H., Tullin, S., Hejnaes, K., Baekkeskov, S. (1992)Membrane anchoring of the autoantigen GAD65 to microvesicles in pancreatic 13-cellsby palmitoylation in theNH2-terminal domain. I. Cell Biol. 118: 309-320.Collander, R. (1954) The permeability of Nitella cells to non-electrolytes. Physiol. Plantarum7: 420-445.Collins, M.D., Scott, W.J., Miller, S.J., Evans, D.A., Nau, H. (1992) Murine teratology andpharmacokinetics of the enantiomers of sodium 2-ethylhexanoate. Toxicol. Appi.Pharmac. 112: 257-265.161Concas, A., Mascia, M.P, Sanna, E., Santoro, G., Serra, M., Biggio, G. (1991) fn vivo”administration of valproate decreases -[35Slbutylbicyclophosphorothionate binding inthe rat brain. Naunyn-Schiniedeberg’s Arch. Pharmac. 343: 296-300.Cornford, E.M. (1983) Blood-brain barrier permeability to valproic acid: preliminaryindications of the independence of lactate and valproate transport. Epilepsia 24: 250.Coulter, D.A., Huguenard, J.R., Prince, D.A. (1989) Characterization of ethosuximidereduction of low-threshold calcium current in thalamic neurons. Ann. Neurol. 25:582-593.Crippen, G. (1979) Distance geometry approach to rationalizing binding data. J. Med. Chem.22: 988-997.Cunnane, S.C., Ho, S.-Y., Dore-Duffy, P., Ells, K.R. (1989) Essential fatty acid and lipidprofiles in plasma and erythrocytes in patients with multiple sclerosis. Am. J. Clin.Nutr. 50: 801-806.Dailey, J.W., Jobe, P.C. (1985) Anticonvulsant drugs and the genetically epilepsy-prone rat.Fed. Proc. 44: 2640-2644.Davis, R., Peters, D.H., McTavish, D. (1994) Vaiproic acid. A reappraisal of itspharmacological properties and and clinical efficacy in epilepsy. Drugs 47: 332-372.De Boer, T., Stoof, J.C., van Duijn, H. (1982) The effects of convulsant and anticonvulsantdrugs on the release of radiolabelled GABA, glutamate, noradrenaline, serotonin andacetylcholine from rat neocortical slices. Brain Res. 253: 153-160.Dichter, M.A., Ayala, G.F. (1987) Cellular mechanisms of epilepsy: a status report. Science237: 157-164.Dickinson, R.G., Harland, R.C., Lynn, R.K., Smith, W.B., Gerber, N. (1979) Transmission ofvalproic acid (Depakene) across the placenta: half-life of the drug in mother and baby.J. Pediatr. 94: 832-835.Dodd, P.R., Hardy, J.A., Oakley, A.E., Edwardson, J.A., Perry, E.K., Delaunoy, J.-P. (1981)A rapid method for preparing synaptosomes: comparison with alternative procedures.Brain Res. 226: 107-118.Dodge, J.T., Mitchell, C., Hanahan, D.J. (1963) The preparation and chemical characteristicsof hemoglobin-free ghosts of human erythrocytes. Arch. Biochem. Biophys. 100:119-130.Dreifuss, F.W. (1989) Valproate. Toxicity. In: Antiepileptic Drugs, 3rd ed. (Levy, R.H.,Mattson, R., Meldrum, B.S., Penry, J.K., Dreifuss, F.E., Eds.), pp. 643-652. Raven,New York.Edafiogho, 1.0., Hinko, C.N., Chang, H., Moore, J.A., Mulzac, D., Nicholson, J.M., Scott,K.R. (1992) Synthesis and anticonvulsant activity of enaminones. J. Med. Chem. 35:2798-2805.Ekwuru, M.O., Cunningham, J.R. (1990) Phaclofen increases GABA release from valproatetreated rats. Brit. J. Pharmacol. 99(Suppl.): 25 1P.Elmazar, M.M., Hauck, R.-S., Nau, H. (1993) Anticonvulsant and neurotoxic activities oftwelve analogues of vaiproic acid. J. Phar. Sci. 82: 1255-1258.162Erlander, M.G., Tillakaratne, N.J., Feldblum, S., Patel, N., Tobin, A.J. (1991) Two genesencode distinct glutamate decarboxylases. Neuron 7: 9 1-100.Evans, D.A., Bartroli, J., Shih, T.L. (1981) Enantioselective aldol condensations. 2. Erythroselective chiral aldol condensations via boron enolates. J. Am. Chem. Soc. 103: 2 127-2129.Evans, D.A., Nelson, J.V., Taber, T.R. (1982) Stereoselective aldol condensations. In: Topicsin Stereochemistry, Vol. 13 (Allinger, N.L., Eliel, E.L., Wilen, S.H., Eds.), Wiley,New York, pp. 1-116.Evans, D.A., Sjogren, E.B., Bartroli, J., Dow, R.L. (1986) Aldol addition reactions of chiralcrotonate imides. Tet. Lett. 27: 4957-4960.Evans, D.A., Britton, T.C., Eliman, J.A. (1987) Contrasteric carboximide hydrolysis withlithium hydroperoxide. Tet. Lett. 28: 6141-6144.Farrant, M., Webster, R.A. (1989) Neuronal activity, amino acid concentration and aminoacid release in the substantia nigra of the rat after sodium vaiproate. Brain Res. 504:49-56.Finnegan, R.A., McNees, R.S. (1964) Organometallic chemistry. VII. The reactions ofnorbornene, endo-5-hydroxymethylnorbornene, nortricyclene and norbornadiene. J.Org. Chem. 29: 3234-3241.Fisher, R.S. (1991) Clinical overview of epilepsy. In: Neurotransmitters and Epilepsy (Fisher,R.S., Coyle, J.T., Eds.), Wiley-Liss, New York, pp. 1-16.Fohlmeister, J.F., Adelman, W.J., Jr., Brennan, J.J. (1984) Excitable channel currents andgating times in the presence of anticonvulsants ethosuximide and valproate. J.Pharmacol. Exp. Ther. 230: 75-8 1.Franceschetti, S., Hamon, B., Heinemann, U. (1986) The action of vaiproate on spontaneousepileptiform activity in the absence of synaptic transmission and on evoked charges in[Ca2J0and [K]0 in the hippocampal slice. Brain Res. 386: 1-11.Frey, H.-H., Loscher, W. (1978) Distribution of valproate across the interface between bloodand cerebrospinal fluid. Neuropharmac. 17: 637-642.Fukuzako, H., Izumi, K. (1991) Clinical aspects of the epilepsies. In: GABA Mechanisms inEpilepsy (Tunnicliff, G., Raess, B.U., Eds.), Wiley-Liss, New York, pp. 1-30.Furniss, B.S., Hannaford, A.J., Rogers, V., Smith, P.W., Tatchell, A.R. (1987a) Vogel’sTextbook ofPractical Organic Chemistry, Wiley, New York, pp. 49 1-492.Furniss, B.S., Hannaford, A.J., Rogers, V., Smith, P.W., Tatchell, A.R. (1987b) Vogel’sTextbook ofPractical Organic Chemistry, Wiley, New York, p. 498.Furniss, B.S., Hannaford, A.J., Rogers, V., Smith, P.W., Tatchell, A.R. (1987c) Vogel’sTextbook ofPractical Organic Chemistry, Wiley, New York, pp. 289-292.Gale, K. (1992) GABA and epilepsy: basic concepts from preclinical research. Epilepsia 33(Supp. 5): S3-S12.Gallagher, J.P., Inokuchi, H., Nakamura, 3., Shinnick-Gallagher, P. (1981) Effects ofanticonvulsants on excitability and GABA sensitivity of cat dorsal root ganglion cells.Neuropharmac. 20: 427-433.163Gent, J.P., Phillips, N.J. (1980) Sodium di-n-propylacetate (valproate) potentiates responses toGABA and muscimol on single central neurones. Brain Res. 197: 275-278.Godin, Y., Heiner, L., Mark, J., Mandel, P. (1969) Effects of di-n-propylacetate, ananticonvulsive compound, on GABA metabolism. 3. Neurochem. 16: 869-873.Gold, B.I., Roth, R.H. (1979) Glutamate decarboxylase activity in striatal slices:characterization of the increase following depolarization. J. Neurochem. 32: 883-888.Goldberg, M.A., Todoroff, T. (1980) Brain binding of anticonvulsants: carbamazepine andvalproic acid. Neurology 30: 826-831.Gram, L., Larsson, O.M., Johnsen, A.H., Schousboe, A. (1988) Effects of vaiproate,vigabatrin, and aminooxyacetic acid on release of endogenous and exogenous GABAfrom cultured neurons. Epilepsy Res. 2: 87-95.Guelen, P.J., van der Kleijn, E. (1978) Anti-epileptic drugs in saliva and cerebrospinal fluid.In: Rational Anti-Epileptic Drug Therapy, pp. 28-36. Elsevier, New York.Hackman, J.C., Grayson, V., Davidoff, R.A. (1981) The presynaptic effects of vaiproic acidin the isolated frog spinal cord. Brain Res. 220: 269-285.Hansch, C., Muir, R.M., Fujita, T., Maloney, P.P., Geiger, F., Streich, M. (1963) Thecorrelation of biological activity of plant growth regulators and chloromycetinderivatives with Hammett constants and partitition coefficients. J. Am. Chem. Soc.85: 28 17-2824.Hansch, C., Fujita, T. (1964) p—a—ic Analysis. A method for the correlation of biologicalactivity and chemical structure. 3. Am. Chem. Soc. 86: 1616-1626.Hansch, C., Leo, A.J. (1979) Substituent constants for correlation analysis in chemistry andbiology, Wiley, New York.Hariton, C., Ciesielski, L., Simler, S., Valli, M., Jadot, G., Gobaille, S., Mesdjian, E., Mandel,P. (1984) Distribution of sodium vaiproate and GABA metabolism in CNS of the rat.Biopharm. Drug Disp. 5: 409-4 14.Harrison, N.L., Simmonds, M.A. (1982) Sodium valproate enhances responses to GABAreceptor activation only at high concentrations. Brain Res. 250: 20 1-204.Harvey, P.K., Bradford, H.F., Davison, A.N. (1975) The inhibitory effect of sodium ndipropylacetate on the degradative enzymes of the GABA shunt. FEBS Lett. 52: 251-254.Hauck, R.-S., Nau, H., Elmazar, M.M. (1991) On the development of alternative antiepilepticdrugs. Naturwiss. 78: 272-274.Hitzemann, R.J., Johnson, P.A. (1983) Developmental changes in synaptic membrane lipidcomposition and fluidity. Neurochem. Res. 8: 121-13 1.Holdiness, M.R., Justice, J.B., Salamone, J.D., Neill, D.B. (1981) Gas chromatographic-massspectrometric determination of glutamic acid decarboxylase activity in subregions ofthe rat brain. 3. Chrom. Biomed. Appl. 225: 283-290.Hopkins, H.S., Gellenberg, A.J. (1994) Treatment of bipolar disorder: how far have we come?Psychopharmac. Bull. 30: 27-38.164Horton, R.W., Anlezark, G.M., Sawaya, M.C., Meldrum, B.S. (1977) Monoamine and GABAmetabolism and the anticonvulsant action of di-n-propylacetate and ethanolamine-Osulphate. Eur. J. Pharmacol. 41: 387-397.Hosford, D.A., Clark, S., Cao, Z., Wilson, W.A., Lin, F., Morrisett, R.A., Hum, A. (1992) Therole of GABAB receptor activation in absence seizures of lethargic (lMh) mice.Science 257: 398-400.Hsiao, C.-N., Liu, L., Miller, M.J. (1987) The use of cysteine- and serine-derivedthiazolidinethiones and oxazolidinethiones as efficient chiral auxiliaries in aldolcondesations. J. Org. Chem. 52: 220 1-2206.Hwang, E.C., van Woert, M.H. (1979) Effect of vaiproic acid on serotonin metabolism.Neuropharmacol. 18: 391-397.ladarola, M.J., Raines, A., Gale, K. (1979) Differential effects of n-dipropylacetate andamino-oxyacetic acid on y-aminobutyric acid levels in discrete areas of rat brain. J.Neurochem. 33: 1119-1123.ladarola, M.J., Gale, K. (1981) Cellular compartments of GABA in brain and theirrelationship to anticonvulsant activity. Mol. Cell. Biochem. 39: 305-330.Johannessen, S.I., Henriksen, 0. (1980) Pharmacokinetic observations of dipropylacetate inchildren. In: Xth International Symposium on Epilepsy (Wada, J.A., Penry, J.K.,Eds.), Raven, New York, p.353.Jones, D.H., Matus, A.I. (1974) Isolation of synaptic plasma membrane from brain bycombined flotation-sedimentation density gradient centrifugation. Biochim. Biophys.Acta 356: 276-287.Kaliszan, R. (1986) Quantitative relationships between molecular structure andchromatographic retention. Implications in physical, analytical and medicinalchemistry. CRC Crit. Rev. Anal. Chem. 16: 323-388.Kaneko, S., Otani, K., Fukushima, Y., Ogawa, Y., Nomura, Y., Ono, T., Nakane, Y.,Teranishi, T., Goto, M. (1988) Teratogenicity of antiepileptic drugs: analysis ofpossible risk factors. Epilepsia 29: 459-467.Kapetanovic, I.M., Yonekawa, W.D., Kupferberg, H.J. (1990) Use of stable isotopes and gaschromatography-mass spectrometry in the study of different pools of neurotransmitteramino acids in brain slices. J. Chrom. 500: 387-394.Kapetanovic, I.M., Yonekawa, W.D., Kupferberg, H.J. (1993) Time-related loss of glutaminefrom hippocampal slices and concomitant changes in neurotransmitter amino acids. J.Neurochem. 61: 865-872.Kassahun, K., Farrell, K., Abbott, F.S. (1991) Identification and characterization of theglutathione and N-acetylcysteine conjugates of (E)-2-propyl-2,4-pentadienoic acid, atoxic metabolite of vaiproic acid, in rats and humans. Drug Metab. Disp. 19: 525-535.Kassahun, K., Baillie, T.A. (1993) Cytochrome P-450-mediated dehydrogenation of 2-n-propyl-2(E)-pentenoic acid, a pharmacologically active metabolite of valproic acid, inrat liver microsomal preparations. Drug Metab. Disp. 21: 242-248.Keane, P.E., Simiand, J., Mendes, A., Santucci, K., Morre, M. (1983) The effects ofanalogues of valproic acid on seizures induced by pentylenetetrazole and GABAcontent in brain of mice. Neuropharmacol. 22: 875-879.165Keane, P.E., Morre, M. (1985) Effect of vaiproate on brain GABA: comparison with variousmedium chain fatty acids. Pharmacol. Res. Commun. 17: 547-555.Kelly, K.M., Gross, R.A., Macdonald, R.L. (1990) Vaiproic acid selectively reduces the low-threshold (T) calcium current in rat nodose neurons. Neurosci. Lett. 116: 233-238.Kennedy, S.K., Longnecker, D.E. (1990) History and principles of anesthesiology. Iv: ThePharmacological Basis of Therapeutics, 8th ed. (Gilman, A.G., Rail, T.W., Nies, A.S.,Taylor, P., Eds.), Pergamon, New York, pp. 285-3 10.Keränen, T., Sillanpaa, Riekkinen, P.J. (1988) Distribution of seizure types in an epilepticpopulation. Epilepsia 29: 1-7.Kerwin, R.W., Olpe, H.-R., Schmutz, M. (1980) The effect of sodium-n-dipropyl acetate on yaminobutyric acid-dependent inhibition in the rat cortex and substantia nigra inrelation to its anticonvulsant activity. Brit. J. Pharmacol. 71: 545-551.Kesterson, J.W., Granneman, G.R., Machinist, J.M. (1984) The hepatotoxicity of vaiproicacid and its metabolites in rats. I. Toxicologic, biochemical and histopathologicstudies. Hepatology 4: 1143-1152.Klee, M.R., Akaike, N., Oomura, Y., Maruyama, T. (1985) Valproate reduces Na+dependentGABA uptake. Pflugers Arch. Eur. I. Physiol. 405 (Supp. 2): R47.Koblin, D.D., Dong, D.E., Deady, J.E., Eger, E.I. (1980) Selective breeding alters murineresistance to nitrous oxide without alteration in synaptic membrane lipid composition.Anesthesiology 52: 401-407Larsson, O.M., Gram, L., Schousboe, I., Schousboe, A. (1986) Differential effects of y-vinylGABA and vaiproate on GABA-transaminase from cultured neurons and astrocytes.Neuropharmacol. 25: 617-625.Lee, R.D. (1991) Pharmacokinetics, tissue distribution and pharmacodynamics of vaiproicacid and its unsaturated metabolites. Ph.D. thesis, University of British Columbia.Lee, R.D., Kassahun, K., Abbott, F.S. (1989) Stereoselective synthesis of the diunsaturatedmetabolites of vaiproic acid. 3. Phar. Sci. 78: 667-671.Leo, A.J. (1990) Methods of calculating partition coefficients. In: Comprehensive MedicinalChemistry, Vol. 4 (Hansch, C., Sammes, P.G., Taylor, J.B., Eds.), Pergamon, Oxford,pp. 295-3 19.Levy, R.H., Lai, A.A. (1982) Valproate: absorption, distribution and excretion. In:Antiepileptic Drugs, 2nd ed. (Woodbury, D.M., Penry, J.K., Pippenger, C.E., Eds.),pp. 555-565. Raven, New York.Levy, R.H., Shen, D.D. (1989) Vaiproate. Absorption, distribution and excretion. In:Antiepileptic Drugs, 3rd ed. (Levy, R.H., Mattson, R., Meldrum, B.S., Penry, J.K.,Dreifuss, RE., Eds.), pp. 583-599. Raven, New York.Lewandowski, C., Kiug, S., Nau, H., Neubert, D. (1986) Pharmacokinetic aspects of drugeffects in vitro: effects of serum protein binding on concentration and teratogenicity ofvaiproic acid and 2-en-valproic acid in whole embryos in culture. Arch. Toxicol. 58:239-242.Lide, D. R. (Ed.)(1994) CRC Handbook of Chemistry and Physics, 75th ed., CRC Press, BocaRaton, p. 3-249.166Lien, E.J., Liao, R.C., Shinouda, H.G. (1979) Quantitative structure-activity relationships anddipole moments of anticonvulsants and CNS depressants. J. Phar. Sci. 68: 463-465.Litchfield, J.T., Wilcoxon, F. (1949) A simplified method of dose-effect experiments. J.Phar. Exp. Ther. 96: 99-113.Liu, M., Scott, K.R., Pollack, G.M. (1990) Pharmacokinetics and pharmacodynamics ofvaiproate analogues in rats. I. Spiro[4.6jundecane-2-carboxylic acid. Epilepsia 31:465-473.Liu, M., Pollack, G.M. (1994) Pharmacokinetics of vaiproate analogues in rats. IV.Anticonvulsant action and neurotoxicity of octanoic acid, cyclohexanecarboxylic acidand 1-methyl-1-cyclohexanecarboxylic acid. Epilepsia 35: 234-243.Lockard, J.S., Levy, R.H. (1976) Vaiproic acid: reversibly acting drug? Epilepsia 17: 477-479.Loscher, W. (1980) Effects of inhibitors of GABA transaminase on the synthesis, binding,uptake and metabolism of GABA. J. Neurochem. 34: 1603-1608.Loscher, W. (1981 a) Anticonvulsant activity of metabolites of valproic acid. Arch. Tnt.Pharmacodyn. Ther. 249: 158-163.Loscher, W. (1981 b) Effect of inhibitors of GABA aminotransferase on the metabolism ofGABA in brain tissue and synaptosomal fractions. J. Neurochem. 36: 1521-1527.Loscher, W. (1985) Valproic acid. In: Antiepileptic Drugs (Frey, H.H., Janz, D., Eds.),Springer, Heidelberg, pp. 507-536.Loscher, W. (1992) Pharmacological, toxicological and neurochemical effects of z2(E)valproate in animals. Pharm. Weekblad. Sci. 14: 139-143.Loscher, W. (1993 a) Effects of the antiepileptic drug valproate on metabolism and function ofinhibitory and excitatory amino acids in the brain. Neurochem. Res. 18: 485-502.Loscher, W. (1993b) In vivo administration of valproate reduces the nerve terminal(synaptosomal) activity of GABA aminotransferase in discrete brain areas of rats.Neurosci. Lett. 160: 177-180.Loscher, W., Bohme, G., Schafer, H., Kochen, W. (1981) Effect of metabolites of valproicacid on the metabolism of GABA in brain and brain nerve endings. Neuropharmac.20: 1187-1192.Loscher, W., Fisher, J.E., Nau, H., Honack, D. (1988) Marked increase in anticonvulsantactivity but decrease in wet-dog shake behavior during short-term treatment ofamygdala-kindled rats with valproic acid. Eur. 3. Pharmacol. 150: 22 1-232.Loscher, W., Nau, H. (1983) Distribution of vaiproic acid and its metabolites in various brainareas of dogs and rats after acute and prolonged treatment. J.Phar. Exp. Ther. 226:845-854.Loscher, W., Nau, H. (1985) Pharmacological evaluation of various metabolites andanalogues of valproic acid. Anticonvulsant and toxic potencies in mice.Neuropharmac. 24: 427-435.167Loscher, W., Nolting, B. (1991) The role of technical, biological and phannacological factorsin the laboratory evaluation of anticonvulsant drugs. IV. Protective indices. EpilepsyRes. 9: 1-10.Loscher, W., Schmidt, D. (1980) Increase of human plasma GABA by sodium vaiproate.Epilepsia 21: 611-615.Loscher, W., Schmidt, D. (1981) Plasma GABA levels in neurological patients undertreatment with valproic acid. Life Sci. 28: 2383-2388.Loscher, W., Siemes, H. (1984) Valproic acid increases y-aminobutyric acid in CSF ofepileptic children. Lancet II: 225.Loscher, W., Siemes, H. (1985) Cerebrospinal fluid y-aminobutyric acid levels in childrenwith different types of epilepsy: effect of anticonvulsant treatment. Epilepsia 26: 314-319.Loscher, W., Vetter, M. (1985) In vivo effects of aminooxyacetic acid and vaiproic acid onthe nerve terminal (synaptosomal) GABA levels in discrete brain areas of the rat.Correlation to pharmacological activities. Biochem. Pharmacol. 34: 1747-1756.Lucas, H.J., Prater, A.N. (1937) The isomeric 2-pentenes. J. Am. Chem. Soc. 59: 1682-1686.Lucke, A., Lehmenkuhler, A., Altrup, U., Nicholson, C., Reith, H., Schmitz-Moormann, C.,Speckmann, E.-J. (1993) Diffusion analysis of vaiproate and trans-2-en-valproate inagar and in cerebral cortex in rat. Brain Res. 631: 187-190.Lucke, A., Mayer, T., Altrup, U., Lehmenkuhler, A., Dusing, R., Speckman, E.-J. (1994)Simultaneous and continuous measurement of free concentration of vaiproate in bloodand extracellular space of rat cerebral cortex. Epilepsia 35: 922-926.Luder, A.S., Parks, J.K., Freeman, F., Parker, W.D., Jr. (1990) Inactivation of beef brain aketoglutarate dehydrogenase complex by vaiproic acid and vaiproic acid metabolites.Possible mechanism of anticonvulsant and toxic actions. J. Clin. Invest. 86: 1574-158 1.Macdonald, R.L., Bergey, G.K. (1979) Valproic acid augments GABA-mediated post-synaptic inhibition in cultured mammalian neurons. Brain Res. 170: 558-562.Maitre, M., Ciesielski, L., Cash, C., Mandel, P. (1978) Comparison of the structuralcharacteristics of the 4-aminobutyrate:2-oxoglutarate transaminases from rat andhuman brain, and of their affmities for certain inhibitors. Biochim. Biophys. Acta522: 385-399.Mander, L.N, Sethi, S.P. (1983) Regioselective synthesis of 13-ketoesters from lithiumenolates and methyl cyanoformate. Tet. Lett. 24: 5425-5428.Marcus, R.J., Winters, W.D., Mori, K., Spooner, C.E. (1967) EEG and behavioral comparisonof the effects of y-hydroxybutyrate, y-butyrolacetone and short chain fatty acids in therat. Tnt. J. Neuropharmac. 6: 175-185.Markwell, M.A., Haas, S.M., Bieber, L.L., Tolbert, N.L. (1978) A modification of the Lowryprocedure to simplify protein determination in membrane and lipoprotein samples.Anal. Biochem. 87: 206-210.168Martin, D.L. (1986) Brain glutamate decarboxylase. In: Neurotransmitter Enzymes,Neuromethods, Vol. 5 (Boulton, A.A., Baker, G.B., Yu, P.H., Eds.), Humana, Clifton,NJ, pp. 361-386.Martin, D.L., Rimvall, K. (1993) Regulation of y-aminobutyric acid synthesis in the brain. J.Neurochem. 60: 395-407.Martin, D.L., Martin, S.B. (1982) Effect of nucleotides and other inhibitors on theinactivation of glutamate decarboxylase. J. Neurochem. 39: 1001-1008.Martin-Gallard, A., Rodriguez, P., Lopez, M., Benavides, J., Ugarte, M. (1985) Effects ofdipropylacetate on the glycine cleavage enzyme system and glycine levels. Biochem.Pharmacol. 34: 2877-2882.Mawhinney, T.P., Robinett, R.S., Atalay, A., Madson, M.A. (1986) Anaiysis of amino acidsas their tert.-butyldimethylsilyl derivatives by gas-liquid chromatography and massspectrometry. J. Chrom. 358: 23 1-242.McLean, M.J., Macdonald, R.L. (1986) Sodium vaiproate, but not ethosuximide, producesuse- and voltage-dependent limitation of high frequency repetitive firing of actionpotentials of mouse central neurons in cell culture. J. Pharmacol. Exp. Ther. 237:1001-1011.McLaughlin, B.J., Wood, J.G., Saito, K., Roberts, E., Wu, J.-Y. (1975) The fine structurallocalization of glutamate decarboxylase in developing axonal processes andpresynaptic terminals of rodent cerebellum. Brain Res. 85: 388-371.Meeley, M.P., Martin, D.L. (1983) Inactivation of brain glutamate decarboxylase and theeffects of adenosine 5’-triphosphate and inorganic phosphate. Cell. Mol. Neurobiol. 3:39-54.Meldrum, B.S. (1985) GABA and other amino acids. In: Antiepileptic Drugs (Frey, H.H.,Janz, D., Eds.), Springer, Heidelberg, pp. 153-188.Mergen, F., Lambert, D.M., Saraiva, J.C., Poupaert, J.H., Dumont, P. (1991) Antiepilepticactivity of 1,3-dihexadecanoylamino-2-valproyl-propan-2-ol, a prodrug of vaiproicacid endowed with a tropism for the central nervous system. J. Pharm. Pharmacol. 43:8 15-8 16.Meunier, H., Carraz, G., Meunier, Y., Eymard, P., Aimard, M. (1963) Pharmacodynamicproperties of n-dipropylacetic acid. I. Antiepileptic properties. Therapie 18: 435-438.Miller, L.P., Walters, J.R. (1979) Effects of depolarization on cofactor regulation of glutamicacid decarboxylase in substantia nigra synaptosomes. J. Neurochem. 33: 533-539.Miller, L.P., Walters, J.R., Martin, D.L. (1977) Post-mortem changes implicate adeninenucleotides and pyridoxal-5’-phosphate in regulation of brain glutamatedecarboxylase. Nature 266: 847-848.Miller, L.P., Walters, J.R., Eng, N., Martin, D.L. (1980) Glutamate holodecarboxylase levelsand the regulation of GABA synthesis. Brain Res. Bull. 5 (Supp. 2): 89-94.Mohamadi, F., Richards, N.G., Guida, W.C., Liskamp, R., Caufield, C., Chang, G.,Hendrickson, T., Still, W.C. (1990) MacroModel V. 3.5X. J. Comput. Chem. 11: 440.Moody, W.J., Futamachi, K.J., Prince, D.A. (1974) Extracellular potassium activity duringepileptogenesis. Exp. Neurol. 42: 248-263.169Nagao, T., Oshimo, T., Mitsunobu, K., Sato, M., Otsuld, S. (1979) Cerebrospinal fluidmonoamine metabolites and cyclic nucleotides in chronic schizophrenic patients withtardive dyskinesia or drug-induced tremor. Biol. Psychiatr. 14: 509-523.Nathan, R., Hsu, C.-C., Bao, J., Wu, R., Wu, J.-Y. (1994) Purification and characterization ofa novel form of brain L-glutamate decarboxylase. J. Biol. Chem. 269: 7249-7254.Nau, H., Rating, D., Koch, S., Hauser, I., Helge, H. (1981) Vaiproic acid and its metabolites:placental transfer, neonatal pharmacokinetics, transfer via mother’s milk and clinicalstatus in neonates of epileptic mothers. 3. Pharmacol. Exp. Ther. 219: 768-777.Nau, H., Loscher, W. (1982) Vaiproic acid: brain and plasma levels of the drug and itsmetabolites, anticonvulsant effects and y-aminobutyric acid (GABA) metabolism inthe mouse. J. Pharmacol. Exp. Ther. 220: 654-659.Nau, H., Loscher, W. (1984) Vaiproic acid and metabolites: pharmacological andtoxicological studies. Epilepsia 25 (Supp. 1): S l4-S22.Nau, H., Hauck, R.-S., Ehiers, K. (1991) Vaiproic acid-induced neural tube defects in mouseand human: aspects of chirality, alternative drug development, pharmacokinetics andpossible mechanisms. Phannacol. Toxicol. 69: 3 10-321.Newman, M.S., Kutner, A. (1951) Reactions involving alkaline treatment of 3-nitroso-2-oxazolidones. 3. Am. Chem. Soc. 73: 4 199-4204.North, P., Fleischer, S. (1983) Alteration of synaptic membrane cholesterol/phospholipid ratiousing a lipid transfer protein. 3. Biol. Chem. 258: 1242-1253.Nys, G.G., Rekker, R.F. (1973) Statistical analysis of a series of partition coefficients withspecial reference to the predictability of folding of drug molecules. The introductionof hydrophobic fragmental constants (f values). Chim. Ther. 5: 521-535.Olpe, H.-R., Steinman, M.W., Pozza, M.F., Brugger, F., Schmutz, M. (1988) Vaiproateenhances GABA-A mediated inhibition of locus coeruleus neurons in vitro. NaunynSchmiedeberg’s Arch. Pharmacol. 338: 655-657.Omzigt, J.G., Los, F.J., Grobbee, D.E., Pijpers, L., Jahoda, M.G. (1992) The risk of spinabifida aperta after first-trimester exposure to vaiproate in a prenatal cohort. Neurology42: 119-125.Pagliusi, S.R., Loscher, W. (1985) Development of tolerance to the wet-dog shake behaviorbut not the increase in seizure threshold induced by L-5-hydroxytryptophan duringcontinued treatment in rats. Psychopharmacol. 86: 118-125.Penman, B.J., Goldstein, D.B. (1984) Membrane-disordering potency and anticonvulsantaction of vaiproic acid and other short-chain fatty acids. Mol. Pharmacol. 26: 83-89.Perry, T.L., Hansen, 5. (1978) Biochemical effects in man and rat of three drugs which canincrease GABA content. 3. Neurochem. 30: 679-684.Phillips, N.I., Fowler, L.J. (1982) The effects of sodium valproate on y-aminobutyratemetabolism and behavior in naive and ethanolarnine-O-sulphate pretreated rats andmice. Biochem. Pharmacol. 31: 2257-2261.Philp, R.P., Robertson, A.V. (1978) Cyclic hydroxycarboxylic acids. II. Lactonizationreactions on 2-hydroxycyclopentanecarboxylic acids and 3-hydroxyprolines. Aust. 3.Chem. 30: 123-130.170Piredda, S.G., Woodhead, J.H., Swinyard, E.A. (1985) Effect of stimulus intensity on theprofile of anticonvulsant activity of phenytoin, ethosuximide and vaiproate. J.Pharmacol. Exp. Ther. 232: 741-745.Porter, T.G., Martin, D.L. (1988) Stability and activation of glutamate apodecarboxylase frompig brain. J. Neurochem. 51: 1886-1891.Porubek, D.J., Grub, M.P., Baillie, T.A. (1989) The covalent binding to protein of vaiproicacid and its hepatotoxic metabolite, 2-n-propyl-4-pentenoic acid, in rats and in isolatedrat hepatocytes. Drug Metab. Disp. 17: 123-130.Prendergast, F.G., Haugland, R.P., Callahan, P.J. (1981) 1-[4-(Trimethylamino)phenylj-6-phenylhexa- 1 ,3,5-triene: synthesis, fluorescence properties, and use as a fluorescenceprobe of lipid bilayers. Biochemistry 20: 7333-7338.Preisendorfer, U., Zeise, M.L., Klee, M.R. (1987) Valproate enhances inhibitory postsynapticpotentials in hippocampal neurons in vitro. Brain Res. 435: 213-219.Prickett, K.S., Baillie, T.A. (1986) Metabolism of unsaturated derivatives of vaiproic acid inrat liver microsomes and destruction of cytochrome P-450. Drug Metab. Disp. 14:221-229.Rall, T.W., Schleifer, L.S. (1990) Drugs effective in the therapy of the epilepsies. In: ThePharmacological Basis of Therapeutics, 8th ed. (Gilman, A.G., Rall, T.W., Nies, A.S.,Taylor, P., Eds.), Pergamon, New York, pp. 436-462.Ross, S.M., Craig, C.R. (1981) Studies on y-aminobutyric acid transport in cobaltexperimental epilepsy in the rat. J. Neurochem. 36: 1006-lOll.Sarhan, S., Seiler, N. (1979) Metabolic inhibitors and subcellular distribution of GABA. J.Neurosci. Res. 4: 399-421.Sato, M., Racine, R.J., McIntyre, D.C. (1990) Kindling: basic mechanisms and clinicalvalidity. Electroencephalogr. Clin. Neurophysiol. 76: 459-472.Schaaf, J.M., Teelken, A.W., Muskiet, F.A., Nagel, G., Woithers, B.G. (1985) A specific andsensitive determination of 7-aminobutyric acid in CSF and brain tissue by gaschromatography-mass spectrometry. J. Neurosci. Meth. 13: 257-265.Schauf, C.L. (1987) Bepridil and valproate retard Na reactivation in Myxicola. Eur. J.Pharmacol. 138: 89-93.Schechter, P.J., Tranier, Y., Grove, J. (1978) Effect of n-dipropylacetate on amino acidconcentrations in mouse brain: correlations with anticonvulsant activity. J.Neurochem. 31: 1325-1327.Schechter, P.J., Tranier, Y., Jung, M.J., Bohlen, P. (1977) Audiogenic seizure protection byelevated brain GABA concentration in mice: effects of ‘y-acetylenic GABA and yvinyl GABA, two irreversible GABA-T inhibitors. Eur. J. Pharmacol. 45: 319-328.Schmutz, M., Olpe, H.-R., Koella, W.P. (1979) Central actions of valproate sodium. J. Phar.Pharmacol. 31: 413-414.Schobben, F., van der Kleijn, E., Vree, T.B. (1980) Therapeutic monitoring of valproic acid.Ther. Drug Monit. 2: 61-7 1.171Scott, K.R., Moore, J.A., Zalucky, T.B., Nicholson, J.M., Lee, J.M., Hinko, C.N. (1985)Spiro[4.5J and spiro[4.61 carboxylic acids: cyclic analogues of vaiproic acid.Synthesis and anticonvulsant evaluation. J. Med. Chem. 28: 413-417.Seubold, F.H. (1954) The cyclohexyl and cyclopentylmethyl radicals. 3. Am. Chem. Soc. 76:3732-3733.Shinitzky, M., Barenholz, Y. (1978) Fluidity parameters of lipid regions determined byfluorescence polarization. Biochim. Biophys. Acta 515: 367-394.Sibra, T.S., Nicholls, D.G. (1987) 4-Aminobutyrate can be released exocytotically fromguinea-pig cerebral cortical synaptosomes. J. Neurochem. 49: 26 1-267.Similae, S., von Wendt, L., Linna, S.L., Saukkonen, A.L., Huhtaniemi, I. (1979)Dipropylacetate and hyperglycemia. Neuropaedatrie 1: 158-160.Simler, S., Randrianarisoa, H., Lehmann, A., Mandel, P. (1968) Effects of di-n-propylacetateon audiogenic seizures in mice. J. de Physiologic 60 (Supp. 2): S47.Simler, S., Ciesielski, L., Klein, M., Gobaille, S., Mandel, P. (1981) On the mechanism of theaction of the anticonvulsant sodium dipropylacetate. C.R. Soc. Biol. (Paris) 175: 114-119.Singh, A.K., Ashraf, M. (1988) Analysis of amino acids in brain and plasma samples bysensitive gas chromatography-mass spectrometry. J. Chrom. 425: 245-255.Slevin, J.T., Ferrara, L.P. (1985) Chronic vaiproic acid therapy and synaptic markers ofamino acid neurotransmission. Neurology 35: 728-731.Slater, G.E., Johnston, D. (1978) Sodium vaiproate increases potassium conductance inAplysia neurons. Epilepsia 19: 379-385.Smith, K.A., Bierkamper, G.G. (1990) Paradoxical role of GABA in a chronic model of petitmal (absence)-like epilepsy in the rat. Eur. 3. Pharmacol. 176: 45-55.Snead, ifi, O.C. (1980) y-Hydroxybutyric acid in subcellular fraction of rat brain. J.Neurochem. 48: 196-201.Snead, III, O.C., Bearden, L.J., Pegram, V. (1980) Effect of acute and chronic anticonvulsantadministration on endogenous y-hydroxybutyrate in rat brain. Neuropharmac. 19: 47-52.Snead III, O.C., Miles, M.V. (1985) Treatment of status epilepticus in children with rectalsodium valproate. J. Pediatr. 106: 323-325.Snead III, O.C., Hechier, V., Verges, M., Marescaux, C., Maitre, M. (1990) Increased yhydroxybutyric acid receptors in thalamus of a genetic animal model of petit malepilepsy. Epilepsy Res. 7: 121-128.Sumikawa, K., Mu, Z., Inoue, T., Okochi, T., Takayoshi, Y., Adachi, K. (1993) Changes inerythrocyte membrane phospholipid composition induced by physical training andphysical exercise. Eur. J. Appl. Physiol. 67: 132-137.Swinyard, E.A., Woodhead, J.H., White, H.S., Franklin, M.R. (1989) Experimental selection,quantification and evaluation of anticonvulsants. In: Antiepileptic Drugs, 3rd ed.(Levy, R.H., Dreifuss, F.E., Mattson, R.H., Meldrum, B.S., Penry, J.K., Eds.), pp. 85-102. Raven, New York.172Sze, P.Y., Sullivan, P., Alderson, R.F., Towle, A.C. (1983) ATP binding to brain L-glutamatedecarboxylase: a study by affinity chromatography. Neurochem. mt. 5: 51-56.Szerb, J.C., O’Regan, P.A. (1984) Glutamine enhances glutamate release in preference to yaminobutyrate release in hippocampal slices. Can 3. Physiol. Pharmacol. 62: 9 19-923.Szerb, J.C., O’Regan, P.A. (1986) Possible reasons for the failure of glutamine to influenceGABA release in rat hippocampal slices; effect of nipecotic acid and methioninesulfoximine. Neurochem. Tnt. 8: 389-395.Taberner, P.V., Charington, C.B., Unwin, J.W. (1980) Effects of GAD and GABA-Tinhibitors on GABA metabolism in vivo. Brain Res. Bull. 5(Supp. 2): 62 1-625.Taillandier, G., Benoit-Guyod, J.C., Boucherle, A., Broll, M., Eymard, P. (1975)Dipropylacetic series. XII. Anticonvulsant branched aliphatic acids and alcohols. Eur.3. Med. Chem.-Chim. Ther. 10: 453-462.Tangorra, A., Curatola, G., Bertoli, E. (1991) Evaluation of antiepileptic drug effect onmembrane fluidity. Exp. Mol. Pathology 55: 180-189.Taylor, P.J. (1990) Hydrophobic properties of drugs. In: Comprehensive MedicinalChemistry, Vol. 4 (Hansch, C., Sammes, P.G., Taylor, J.B., Eds.), Pergamon, Oxford,pp. 24 1-294.Taylor, C.P., Vartanian, M.G., Andruszkiewicz, R., Silverman, R.B. (1992) 3-Alkyl GABAand 3-alkyiglutamic acid analogues: two new classes of anticonvulsant agents.Epilepsy Res. 11: 103-110.Tursky, T. (1970) Inhibition of brain glutamate decarboxylase by adenosine triphosphate.Eur. J. Biochem. 12: 544-549.Vajda, F.J., Donnan, G.A., Phillips, 3., Bladin, P.F. (1981) Human brain, plasma andcerebrospinal fluid concentration of sodium vaiproate after 72 h of therapy.Neurology 31: 486-487.Van Dongen, A.M., van Erp, M.G., Voskuyl, R.A. (1986) Valproate reduces excitability byblockage of sodium and potassium conductance. Epilepsia 27: 177-182.Van Erp, M.G., van Dongen, A.M., van den Berg, R.J. (1990) Voltage-dependent action ofvaiproate on potassium channels in frog node of Ranvier. Eur. J. Pharmacol. 184:151-161.Van den Berg, R.J., Kok, P., Voskuyl, R.A. (1993) Valproate and sodium currents in culturedhippocampal neurons. Exp. Brain Res. 93: 279-287.Van der Laan, J.W., de Boer, T., Bruinvels, J. (1979) Di-n-propylacetate and GABAdegradation. Preferential inhibition of succinic semialdehyde dehydrogenase andindirect inhibition of GABA-transaminase. J. Neurochem. 32: 1769-1780.Vayer, P., Mandel, P., Maitre, M. (1 987a) y-Hydroxybutyrate, a possible neurotransmitter.Life Sci. 41: 1547-1558.Vayer, P., Gobaille, S., Mandel, P., Maitre, M. (1987b) 3’,5’-cyclic guanosine monophosphateincrease in rat brain hippocampus after y-hydroxybutyrate administration. Preventionby vaiproate and naloxone. Life Sci. 41: 605-610.173Vergnes, M., Marescaux, C., Micheletti, G., Rumbach, L., Warter, J.M. (1985) Blockade of“antiabsence’ activity of sodium vaiproate by THIP in rats with petit-mal-likeseizures: comparison with ethosuximide. 3. Neur. Trans. 63: 133-141.Verloop, A., Hoogenstraaten, W., Tipker, J. (1976) Development and application of newsteric substituent parameters in drug design. In: Drug Design, Vol. 7 (Ariens, E.J.,Ed.), pp. 165-208. Academic, New York.Vorhees, C.V., Acuff-Smith, K.D., Weisenburger, W.P., Minck, D.R., Berry, J.S., Setchell,K.D., Nau, H. (1991) Lack of teratogenicity of trans-2-ene-valproic acid compared tovaiproic acid in rats. Teratology 43: 583-590.Wadsworth, W.S., Emmons, W.D. (1973) Ethyl cyclohexylideneacetate. In: OrganicSyntheses, Coil. Vol. V (Baumgarten, H.E., Ed.), pp. 547-549. Wiley, New York.Walker, M.A., Heathcock, C.H. (1991) Extending the scope of the Evans asymmetric aldolreaction: preparation of anti and t1non-Evans” syn aldols. J. Org. Chem. 56: 5747-5750.Weiland, J.H., Arens, J.F. (1956) Chemistry of acetylenic ethers. XVII. The reactions of someethoxyethynylcarbinols with mercaptans and thio acids. Rec. tray. chim. 75: 1358-1368.Weissman, D., Simler, S., Ciesielski, L., Mandel, P. (1978) The effect of 2-propyl-2-pentenoic acid on GABA levels in specific mouse brain regions. C. R. Soc. Biol.(Paris) 172: 707-712.Whitton, P.S., Fowler, L.J. (1991) The effect of valproic acid on 5-hydroxytryptamine and 5-hydroxyindoleacetic acid concentration in hippocampal dialysates in vivo. Eur. 3.Pharmacol. 200: 167-169.Whittle, S.R., Turner, A.J. (1978) Effects of the anticonvulsant sodium valproate onaminobutyrate and aldehyde metabolism in ox brain. J. Neurochem. 31: 1453-1459.Whittle, S .R., Turner, A.3. (1981) Anticonvulsants and brain aldehyde metabolism. Biochem.Pharmacol. 30: 1191-1196.Williams, R.M., Maruyama, L. K. (1987) Synthesis of functionalized bicyclicdioxopiperazines via intramolecular epoxide opening. J. Org. Chem. 52: 4044-4047.Willow, M., Kuenzel, E.A., Catterall, W.A. (1984) Inhibition of voltage-sensitive sodiumchannels in neuroblastoma cells and synaptosomes by the anticonvulsant drugsdiphenyl-hydantoin and carbamazepine. Mol. Pharmacol. 25: 228-234.Wiseman, H., Quinn, P., Halliwell, B. (1993) Tamoxifen and related compounds decreasemembrane fluidity in liposomes. FEBS Lett. 330: 53-56.Wolf, R., Tscherne, U. (1994) Valproate effect on y-aminobutyric acid release in parsreticulata of substantia nigra: combination of push-pull perfusion and fluorescencehistochemistry. Epilepsia 35: 226-233.Wolf, R., Tscherne, U., Emrich, H.M. (1988) Suppression of preoptic GABA release causedby push-pull-perfusion with sodium valproate. Naunyn-Schmiedeberg’s Arch.Pharmacol. 338: 658-663.Wolinsky, J., Erickson, K.L. (1965) Bromomethylenecycloalkanes. J. Org. Chem. 30: 2208-2211.174Wood, J.D., Kurylo, E., Lane, R. (1988) y-Aminobutyric acid release from synaptosomesprepared from rats treated with isonicotinic acid hydrazide and gabaculine. 3.Neurochem. 50: 1839-1843.Zeise, M.L., Kasparow, S., Zieglgansberger, W. (1991) Vaiproate suppresses N-methyl-Daspartate evoked transient depolarizations in the rat neocortex in vitro. Brain Res.544: 345-348.Zielinsky, 3.3. (1982) Epidemiology. In: A Textbook of Epilepsy, (Laidlaw, 3., Richens, A.,Eds.), pp. 16-33. Churchill Livingstone, New York.Zimmer, R., Teelken, A.W., Gundurewa, M., Ruther, F., Cramer, H. (1980) Effect of sodiumvaiproate on CSF GABA, cAMP, cGMP and homovanillic acid levels in men. BrainRes. Bull.(Supp. 2): 585-588.Zona, C., Avoli, M. (1990) Effects induced by the antiepileptic drug vaiproic acid upon theionic currents recorded in rat neocortical neurons in cell culture. Exp. Brain Res. 81:3 13-3 17.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items