TiPS - November
Willy Haefely, James R. Martin and Peter Schoch activities. Rt~~eoiiinzeyiuesi!i ciirzical use iznve R range of pharnzacological Sonze, e.g. sedatiorz, foleraizce azzd addictiozz, are rzot welcome. Undesirable side-e,tfects of drugs are often nmtrolfed by developizzg compounds that birzd zrzore selectively to ozze partirulc- receptor subtype. An alterzzative approach, discussetf here b!y Willy Haefely and colleagues, is tlze development of partial agoksts wlzich e.rploit regio,zal differerzces in receptor reserve to tease apart biological respotzses. Partial agonists for tlze benzodiazepine modulnfoy site ou the GA13AAconzplex have been developed azzd their pharmacological profilescan be interpreted to szlggest tlzat tzezdrons mediatirzg azzticonvulsant rzzd afzfi-a?zsiehJ effects do irzdeed have a higher receptor reserve than neurons mediating otizer zczzwanted effects.TIzis szlggesfs tlzat benzodiazepine receptor partialagonisfs nzay hare important therapeutic potential. Partial agonists or low-efficacy compounds induce smaller fractional responses in their target cells than do full agonists at the same fractional receptor Occupancy. Thus ligands with different intrinsic efficacies will produce different maximal responses in biological systems that differ in receptor density (spare receptor capacity) or efficiency of signal transduction’. Partial agonism has had important practical consequences both in terms of basic pharmacology and, more recently, in various areas of drug deve!opment. In a single target cell or a target organ with uniform receptor density (e.g. one type of neuron, a piece of myocardium or one segment of the vascular bed), low-efficacy agonism prevents overstimulation of a receptor population, thereby reducing overdose problems, desensitization responses and adaptation (tolerance, dependence). Even in tissues with a large receptor reserve where maximal effects can be achieved by partial agonists, dose-response curves are shallower than for full agon-
ists, making it easier to titrate optim& therapeutic doses. If the receptors involved are responsible for maintaining a physiological tone, partiai agonists stabilize receptor activation at a distinct submaximal set-point by activating the receptor in the absence of sufficient endogenous ligand and by reducing the overstimulation of receptors exposed to excessive concentrations of endogenous ligand. Partial agonists such as xamoterol act at PI-adrenoceptors to maintain a certain level of tonic (chronotropic or inotropic) stimulation of the heart but pr.zvent overstimulation of adrenoceptors. Partial agonists such as some ergot derivatives, which act at vasoconstrictive c+adrenoceptors, maintain a certain level of vascular tone but prevent excessive adrenergic vasoconstriction. Terguride and the amino-ergolines SDZ208911 and SDZ208912, which act as partial agonists at central dopamine receptors, may maintain a minimal tonic activation thereby reducing the risk of extrapyramidal side-effects2. However, when the target is broader and includes cells or tissues that differ in their receptor density, partial agonists will be differentially effective (‘tissue or system selectivity’). For example, segments within vascular beds differ in their a-adrenoceptor density; partial agonists at a given dose will therefore produce greater constriction in areas of higher receptor density than in
1990 [Vol. 1 Zi
areas of lower receptor density. The same receptor type may exist in the subsynaptic membrane as well as in the membrane (somadendritic, terminal) of the presynaptic neuron; if receptor reserve is higher in the presynaptic structures, partial agonists may be more effective here; this is assumed to be the case for example for dopamine, 5-HT and a[~autoreceptors. Various neuronal circuits of the CNS may contain cellular elements differing in the density of the same receptor type; hence partial agonists will differ from full agonists in their overall pharmacological profiles (e.g. opioids3) and can have reduced liabilities for acute and/or chronic side-effects. Of particular clinical interest is the possibility that partial agonists at the benzodiazepine receptor in anxiolytic and anticonvulsant doses may produce less sedation (with impairment of learning and attention), muscle relaxation (and ataxia), tolerance and physical dependence. GABAJBDZ receptor The benzodiazepine receptor is an integral part of the heterooligomeric GABA* receptor Clchannel compiex4 whose subunits (a, p, y, 6 in unknown stoichiometry) form a transmembrane anion channel gated by the primary iigand GABA and modulated by secondary (allosteric or heterotopic) ligands. Multiple subunit variants are expressed in brain4. The binding site for GABA is associated primarily with the p subunit while the LYsubunit contains the binding site for benzodiazepines; the presence of a y subunit appears to be a requirement for benzodiazepine receptormediated modulatory effects. The benzodiazepine receptor is an allosteric modulatory site capable of fine-tuning the affinity and/or availability of GABA-binding sites in either direction; it may also modulate the conformational transition that couples GABA receptor activation to Cl- channel opening5e6. The benzodiazepine receptor recognizes ligands with greatly varying chemical structures (benzodiazepines and nonbenzodiazepines) and different intrinsic efficacies5n6. At maximally effective full doses, agon-
TiPS - November 1990 /Vol.
2 21
ists, such as diazepam, produce a parallel, left-ward shift of the GABA dose-response curve for Cl- current induction; antagonists, such as flumazenil, have little if any effect on GABA-induced Cl- flux. Partial agonists produce efrects on the Cl- current, according to their individual intrinsic efficacies, between that of full agonists and antagonists. The allosteric modulatory benzodiazepine receptor also recognizes inverse agonists that have full or partial negative intrinsic efficacy or negative modulatory activity (reducing the channel-gating effect of GABA). Although partial inverse agonists are of potential therapeutic interest, they are not considered further here.
453
FN
,N'-O
Nf+N-I, N \
c (7-r
a2N
Cl
0
FG8205 clonazepam
l3r 0
Bretazenil Bretazenil (Ro166028; Fig. 1) has an approximately tenfold higher affinity for the benzodiazepine receptor than diazepam. Its profile of activity in relation to benzodiazepine receptor full agonists has been investigated. e Bretazenil produces anticonflict effects at much lower doses and over a much wider dose range than diazepam. Its anticonvulsant activity is characterized by potent, full protection from a maximaZy pentetrazole effective dose (quanta1 effect technique) with a dose-response curve shallower than that for diazepam, while the maximal increase of the convulsant threshold dose of pentetrazole (graded effect technique) is only about one quarter of that achieved with diazepam at receptor saturation (E. Brouillet, pers. commun.). 0 Mild degrees of sedation occur at doses much higher than those producing anticonvulsant and anticonflict effects, and several end-points of severe sedation readily achieved with full agonists are not obtained even at the highest doses= (Fig. 2). Potentiation of ethanol-induced sedation is much less pronounced with bretazenil than with diazepam’,‘“. 0 Bretazenil antagonizes the effects of full agonists to varying extents. The effects of diazepam are completely abolished in tests measuring severe sedation (where bretazenil is ineffective by itself) and partially abolished in tests
bretazenil (Ro166028) abecarnil (ZKf12119)
divaplon (RU32698) a!$dem Fig, 1. Slructural formulae of benzcdiazepine receptor partial agonists.
where it has only a smaller maximal effect than diazepam. Complete antagonism is obtained against diazepam-induced motor impairment, for example in the horizontal wire test. No clear anticonvulsant antagonism of effects has been demonstrated in the quanta1 effect paradigm. An intermediate type of interaction is found SNhen the cGMP content in the cerebellum is measured to give an indication of Purkinje cell activity: bretazenil is more potent than diazepam in reducing cGMP facilitation of content by GABAergic inhibition, but its effect reaches a plateau at about 50% of the diazepam maximum; if administered in increasing doses
together with a maximally effectivc dose of diazepam, bretazenil
dose-dependently reduces the effect of the full agonist to values
obtained
when bretazenil is given alone6. Bretazenil precipitates withdrawal symptoms in squirrel monkeys made physically dependent on diazepam, albeit to a milder degree than that precipitated by the antagonist flumazen’l or even by an inverse agonist’. CBRepeated
treatment
of
mice
with bretazenil did not result in tolerance to the anticonvulsant
effect, in contrast to the effect of full agonists”*‘“. Desensitization of GABAA receptors and reduction of the positive modulatory function of the benzodiazepine receptor was absent or clearly less marked with bretazenil than after full agonists r3*r4.Again unlike full agonists, physical dependence could not be induced in squirrel monkeys after repeated very high doses of bretazenil (40 mg kg-‘)
TiPS - November 1990[Vol. 111
rat -
open field
continuous monkey avoldance horizontal wrre rat rat -
ataxta ataxra sedation
dog -
monkey -
locomotor
PTZ seizures
1
10
100
1000
Dose (mg kg-‘, orally)
Fg. 2. Profiles of activity of the full agonisl dkwepam (0) and the partialagonisl bretarenil Is). Values are EO, values or minimal effective doses (anticonflict activity). Pli?. pentetrazbe (L9ptazol).
as assessed by challenge with flumazenil’“; moreover, monkeys would not self-administer (i.v.) bretazenil15. 0 Studies using in o&o labelling of cerebral beruodiazepine recepwith [3H]fkJmazenil or tor [“C]flumazenil (PET studies) showed bretazeni: to require higher fractional receptor occupancy than diazepam for the same magnitudes of pharmacological effects (Ref. 16 and E. Brouillet, pers. commnn.). 0 Facilitation of GABA-induced Cl- conductance measured electrophysiologicaily in single neuror?s was less marked with bretazenil than with full agonists”*ls. Can this profile of activity of bretazenil be explained by invoking partial agonism? By analogy witit other receptor systems it may be assumed that the neurosai systems that ere preferentially involved in the various effects of beruodiazepines differ in their GABAAlbenzodiazepine re\:eptor density and reserve (Fi?. 3). Neurons mediating anticonvulsant and antianxiety effects would thus be expected to have the highest receptor reserve; by contrast, a smaller reserve (or no reserve) wocld characterize neurons involved in, for example, arousal control. As a partial agonist,
bretazenil would be able to produce sufficient GABA potentiation in the former systems but insufficient poteatiation in the latter systems to reach the same maximal CNS depression as full agonists. Moreover, bretazenil antagonizes those effects of full agonists that require higher fractional receptor occupancy, such as ataxia and severe sedation. Clinical studies’0*‘9 have established that the therapeutic effect of bretazer;il is a tenth of the effective dose of diazepam in generalized anxiety disorder (double-blind, placebo and diazepam-controlled i study in 225 patients) and panic disorders (double-blind, placebo-controlled study in 24 patients); it has minimal sedative side-effects at this dose. In the latter study bretazenil was given at the first signs of an imminent panic attack. Surprisingly, the drug was found in open noncontrolled studies to be effective in am&orating psychotic sympt0m.s in 60% of 73 patients with schizophrenia according to DMS III criteria” - an effect that is also being considered as a future therapeutic indication. It will of course be important to confirm in humans the absence of reinforcing effects, the reduced tolerance, low physical dependence potential and minimal ethanol potentiation demonstrated in animals.
Clonazepam Clonazepam (Fig. 1) is a classical 7-nitrobenzodiazepine in use as an antiepileptic for over 15 years, which has recently been found to be thera~e~~~cally active in panic disorder. Electrophysiologica117,21 and behavioural studie@ have demonstrated that clonazepam has an int-insic efficacy slightly lower than that of diazepam. Its partial agonism is best illustrated by its incomplete (and biphasicf antagonism of diazepam-induced motor performance deficit*‘. The lower liability for tolerance to the anticonvulsant activity of clonazepam corn ared with a series of full agonists P’ may be related to its partial agonistic properties. FGS205 FG8205 (Fig. 1) is an imidazobenzodiazepine with about a fifth of the affinity of bretazenil for the benzodiazepine receptor ~PZ aitro2? It shows features characteristic of a benzodiazepine receptor partial agonist, and its intrinsic efficacy is probably similar to that of bretazenil. Its anxiolytic-like activity is similar in magnitude to that of diazepam in a rat conditioned emotional beha~ourai response test but it is active over a broader dose range, probably as a result of a reduced sedating effect. FG8205 is approximately equipotent and equieffective with hiazepam in taming aggressive cvnomolnus monkevs23. Its sedativelmus~le relaxing efficacy is much lower thati that of diazepam, the effect of which can be tota!ly or partially antagonized by FGS2OS in these paradigms. The relative efficacy of FG8205 in potentiating the depressant effect of the GABA, receptor agonist isoguvacine in hippocampal slice is about half that of full agonists and the GA3A-induced shift in the receptor affinity of FG8205 is about 75% of the GABA-induced shift for diazepam binding. Imidazo[l,Z-ulpyrimidines Some imidazo[l,2-nlpyrimidines are benzodiazepine receptor partial agonists24. For example, both divaplon (RU32698; Fig. i; Ref. 25) and RU32514 (Ref. 26) are effective in various models of anxiety in rats with a potency similar to that of chlordiazepoxide. Divaplon, but not RU32514, exhibits broad anticonvulsant activity; however, both
TiPS - November 1990 [Vol. 11 I show little or no activity in tests measuring sedation/motor performance impairment. All the c!ffects of these compounds were reversed by flumazenil. RU32514 ,antagonized the anticonvulsant activity of diazepam. Thus, divaplon and RU32514 appear to be benzodiazepine receptor partial agonists with intrinsic efficacies at the lower end of the spectrum. No results of clinical studies have been published. Abecamil Abecarnil (ZK112119; Fig. 1j27,28 is a fl-carboline derivative with high affinity for benzodiazepine receptors. Like bretazenil, it is effective in lower doses than diazepam in tests predictive of anxiolytic activity and in some models of
455 epilepsy, but is clearly less effective in tests measuring sedation and muscle relaxation and is weaker than diazepam in potentiating ethanol and hexobarbitone. Abecarnil antagonizes the depressant effect of diazepam on the righting reflex and performance in a forced motor task. The increase of abecamil binding by GABA (GABA shift) is clearly less marked than the effect of GABA on diazepam binding. Fractional receptor occupancy at comparable endpoints of pharmacological activity is markedly higher with abecamil than with diazepam. This profile of activity characterizes abecamil as a benzodiazepine receptor partial agonist of lower intrinsic efficacy than bretazeni!, with potential as an anxiolytic and antiepileptic. Ciinical
studies
are being
carried
out,
but
no data are yet available. Alpidem Alpidem (SL800342; Fig. 1) is an imidazopyridine compound reported to be anxio!ytic in 17 studies on over 1500 patients with situational anxiety, general anxiety disorder and anxiety in psychotic patients29*30. In animal experiments alpidem produces anxiolytic-like effects in some but not all models and anticonvulsant effects without sedation and muscle relaxation3’,3’. Unpublished results (E. P. Bonetti, J. R. Martin and L. Pieti) suggest that it has a very low efficacy in anticonflict and anticonvulsant paradigms; moreover, relatively high doses of alpidem clearly antagonize the sedative and muscle relaxant effects of diazepam.
Paradigms used to test effects of benzodiazepine receptor partial agonists A number of test paradigms used to characterize agonists.
in animals are commonly benzodiazepine receptor partial
The most frequently used paradigms for assessing potential therapeutic efficacy are anticonflict and anticonvulsant tests: e Anticozflict tests are considered to be predictive of the anxiolytic effects of benzodiazepine receptor ligands. Operant conflict paradigms often involve situations in which lever-press responding by hungry animals is simultaneously rewarded with food and punished by footshock; response rate is increased by both full and partial agonists at the benzodiazepine receptor. In an alternative approach (conditioned emotional response test), a conditioned aversive stimulus is presented during food-reinforced lever-pressing thereby inhibiting responding; benzodiazepine receptor full and partial agonists increase the response rate.
GABA function and so can be titrated
to induce convulsions; in this case benzodiazepine receptor fi?!l or partial agonists increase the convulsive threshold of leptazoi in individual animals (graded effect). Alternatively, leptazol can be given in a convulsive dose to a collective of animals, in which case the proportion of animals exhibiting convulsions is reduced by benzodiazepine receptor full or partial agonists (quuntul cflect). The latter technique does not necessarily permit an assessment of the maximal anticonvulsive effect of a test drug. Undesired effects for the therapeutic indication anxiety are produced by benzodiazepine receptor full agonists and, to a lesser extent, by partial agonists. e Motor impairment can be assessed after drug administration by observation of either the spontaneous behavior of animals or forced motor activity [e.g. walking on a rotating cylinder (rotarod);. In the horizontal wire test, motor impairment can be studied by allowing an animal to grip a wire strung between two posts and then determining if the hanging animal can caise a hindpaw to grasp the wire. - _ e Ethanol interactiotl can be eva!uated in a situation in which, for example, the threshold dose of ethanol required to induce a short loss of righting reflex is admin;stered to animals, followed by the test drug, and any resultant increase in duration over the control situation determined. o Phusical dependence liabilitv can be investigated in animals after ‘chronic administration of a benzodiaz=*nr nartial agonist by either iooking for epine receykv. 1 spontaneous withdrawal signs ;ftcr te~ination of treatment or those precipitated by the injection 05 a benzodiazepine receptor antagonist. e Self-administration of a test drug by an animal (e.g. lever-pressing to activate a pump system to inject a drug via a permanently implanted cannula) provides one indication for possible abuse liability. LI
e Anticonvulsant effects can be evaluated in experimental situations involving electrically or chemically induced convulsions. Leptazol (pentetrazole) reduces
TjpS -
90% receptor reserve
tOO%i
1
4 3
1
0 50% receptor reserve
in terms both of affinity and in degree or direction of activation. However, there is at present no pharmacological or func38 15 Martin, I. R. rtnl. Sot. Nerrrosci. A&r. fin press) 16 Potier, M-C., Prada de Carvalho, L., Verrault, I’., Chapouthier, G. and Rossier, J. (1988) Eur. I. PltantmcoI. 156, 169-172 17 Chart, C. Y. and Farb, D. H. (1985) J. Ne~rrosci. 5, 2365-2373 18 Yakushiji, T., Fukuda, T., Oyama, Y. and Akaike, N. (1989) Br. J. P~r~rt?r~co~. 98,735-740 19 Katschnig, N. ef ni. (1989) in ~ia~a~jcar Prycfriafry (Saletu. B., ed.), pp. 167-169, Ceorg Thieme 20 Merz, W. A. et nf. (1988) r’s~c~zu~/~a~~zffco&y Y6 (Suppl.), 237 21 Eonetti, E. P., Pole, I’., Laurent, J-P., Schoch, P. and Haefely, W. (1987) Nelrroscierrce 22 (Suppl.), 58.2 22 Gent, J. P., Feely, M. P. and Haigh, J. R. M. (1985) Life 5’1%37.849-856 23 Tricklebank, M. D. et RI. Br. J. ~~zffrt?f~co~.(in press) 24 Gardner, C. R. (1988) Drirg Dev. Res. 12, 1-28 25 Gardner, C. R. et ~1. (1987) Br. J. Pll~ri~i~rai. 92, 537P 26 Gardner, C. R., Deacon, R., Mann, G., Budhram, P. and Guy, A, P. (1987) Br. J. PI~armncot. 92, 655P 27 Stephens, D. N. et nl. (1990) J. Pltanrmcat. Exp. Tker. 253‘334-343 28 Turski, L. et al. (19%) f. Pf~a~J?aca~.E.ry. Ti!er. 253, 344-352 29 Morselli, P. L. (1990) P~iar~f~ffc~* psychintr!/ 23, 129-134 30 Morton, S. and Lader, M. (1990) PI~art~tncopsyc~i~atr!/ 23, 120-123 31 Langer, S. 2. et al. (1990) Phnnimcopsychiatry 23, 103-107 32 Zivkovic, 3. et nl. (1990) Pltarrttncopsychiatry 23, 108-113 RIJ32514: 2-benzoy!,.5-met~oxy-6,7,8,9tetrahydro-imidazo[1.2-njquinazoline SDZ208911: N-[(8~)-dimethylergoii~e-8~ yl]-2,2-dimethylpropanamide 502208912: N-[($oi)-2-chloro-f&methylergoline-8-yip-2,2-dimethy~propanam~de DSM III: Diagnosticand Statistical Manual of Mental Disorders, 3rd edition; a set of diagnostic criteria which conform to the Intemationai Classification of Diseases (US)
Willy Haefely, James R. Martin and Peter Schoch activities. Rt~~eoiiinzeyiuesi!i ciirzical use iznve R range of pharnzacological Sonze, e.g. sedatiorz, foleraizce azzd addictiozz, are rzot welcome. Undesirable side-e,tfects of drugs are often nmtrolfed by developizzg compounds that birzd zrzore selectively to ozze partirulc- receptor subtype. An alterzzative approach, discussetf here b!y Willy Haefely and colleagues, is tlze development of partial agoksts wlzich e.rploit regio,zal differerzces in receptor reserve to tease apart biological respotzses. Partial agonists for tlze benzodiazepine modulnfoy site ou the GA13AAconzplex have been developed azzd their pharmacological profilescan be interpreted to szlggest tlzat tzezdrons mediatirzg azzticonvulsant rzzd afzfi-a?zsiehJ effects do irzdeed have a higher receptor reserve than neurons mediating otizer zczzwanted effects.TIzis szlggesfs tlzat benzodiazepine receptor partialagonisfs nzay hare important therapeutic potential. Partial agonists or low-efficacy compounds induce smaller fractional responses in their target cells than do full agonists at the same fractional receptor Occupancy. Thus ligands with different intrinsic efficacies will produce different maximal responses in biological systems that differ in receptor density (spare receptor capacity) or efficiency of signal transduction’. Partial agonism has had important practical consequences both in terms of basic pharmacology and, more recently, in various areas of drug deve!opment. In a single target cell or a target organ with uniform receptor density (e.g. one type of neuron, a piece of myocardium or one segment of the vascular bed), low-efficacy agonism prevents overstimulation of a receptor population, thereby reducing overdose problems, desensitization responses and adaptation (tolerance, dependence). Even in tissues with a large receptor reserve where maximal effects can be achieved by partial agonists, dose-response curves are shallower than for full agon-
ists, making it easier to titrate optim& therapeutic doses. If the receptors involved are responsible for maintaining a physiological tone, partiai agonists stabilize receptor activation at a distinct submaximal set-point by activating the receptor in the absence of sufficient endogenous ligand and by reducing the overstimulation of receptors exposed to excessive concentrations of endogenous ligand. Partial agonists such as xamoterol act at PI-adrenoceptors to maintain a certain level of tonic (chronotropic or inotropic) stimulation of the heart but pr.zvent overstimulation of adrenoceptors. Partial agonists such as some ergot derivatives, which act at vasoconstrictive c+adrenoceptors, maintain a certain level of vascular tone but prevent excessive adrenergic vasoconstriction. Terguride and the amino-ergolines SDZ208911 and SDZ208912, which act as partial agonists at central dopamine receptors, may maintain a minimal tonic activation thereby reducing the risk of extrapyramidal side-effects2. However, when the target is broader and includes cells or tissues that differ in their receptor density, partial agonists will be differentially effective (‘tissue or system selectivity’). For example, segments within vascular beds differ in their a-adrenoceptor density; partial agonists at a given dose will therefore produce greater constriction in areas of higher receptor density than in
1990 [Vol. 1 Zi
areas of lower receptor density. The same receptor type may exist in the subsynaptic membrane as well as in the membrane (somadendritic, terminal) of the presynaptic neuron; if receptor reserve is higher in the presynaptic structures, partial agonists may be more effective here; this is assumed to be the case for example for dopamine, 5-HT and a[~autoreceptors. Various neuronal circuits of the CNS may contain cellular elements differing in the density of the same receptor type; hence partial agonists will differ from full agonists in their overall pharmacological profiles (e.g. opioids3) and can have reduced liabilities for acute and/or chronic side-effects. Of particular clinical interest is the possibility that partial agonists at the benzodiazepine receptor in anxiolytic and anticonvulsant doses may produce less sedation (with impairment of learning and attention), muscle relaxation (and ataxia), tolerance and physical dependence. GABAJBDZ receptor The benzodiazepine receptor is an integral part of the heterooligomeric GABA* receptor Clchannel compiex4 whose subunits (a, p, y, 6 in unknown stoichiometry) form a transmembrane anion channel gated by the primary iigand GABA and modulated by secondary (allosteric or heterotopic) ligands. Multiple subunit variants are expressed in brain4. The binding site for GABA is associated primarily with the p subunit while the LYsubunit contains the binding site for benzodiazepines; the presence of a y subunit appears to be a requirement for benzodiazepine receptormediated modulatory effects. The benzodiazepine receptor is an allosteric modulatory site capable of fine-tuning the affinity and/or availability of GABA-binding sites in either direction; it may also modulate the conformational transition that couples GABA receptor activation to Cl- channel opening5e6. The benzodiazepine receptor recognizes ligands with greatly varying chemical structures (benzodiazepines and nonbenzodiazepines) and different intrinsic efficacies5n6. At maximally effective full doses, agon-
TiPS - November 1990 /Vol.
2 21
ists, such as diazepam, produce a parallel, left-ward shift of the GABA dose-response curve for Cl- current induction; antagonists, such as flumazenil, have little if any effect on GABA-induced Cl- flux. Partial agonists produce efrects on the Cl- current, according to their individual intrinsic efficacies, between that of full agonists and antagonists. The allosteric modulatory benzodiazepine receptor also recognizes inverse agonists that have full or partial negative intrinsic efficacy or negative modulatory activity (reducing the channel-gating effect of GABA). Although partial inverse agonists are of potential therapeutic interest, they are not considered further here.
453
FN
,N'-O
Nf+N-I, N \
c (7-r
a2N
Cl
0
FG8205 clonazepam
l3r 0
Bretazenil Bretazenil (Ro166028; Fig. 1) has an approximately tenfold higher affinity for the benzodiazepine receptor than diazepam. Its profile of activity in relation to benzodiazepine receptor full agonists has been investigated. e Bretazenil produces anticonflict effects at much lower doses and over a much wider dose range than diazepam. Its anticonvulsant activity is characterized by potent, full protection from a maximaZy pentetrazole effective dose (quanta1 effect technique) with a dose-response curve shallower than that for diazepam, while the maximal increase of the convulsant threshold dose of pentetrazole (graded effect technique) is only about one quarter of that achieved with diazepam at receptor saturation (E. Brouillet, pers. commun.). 0 Mild degrees of sedation occur at doses much higher than those producing anticonvulsant and anticonflict effects, and several end-points of severe sedation readily achieved with full agonists are not obtained even at the highest doses= (Fig. 2). Potentiation of ethanol-induced sedation is much less pronounced with bretazenil than with diazepam’,‘“. 0 Bretazenil antagonizes the effects of full agonists to varying extents. The effects of diazepam are completely abolished in tests measuring severe sedation (where bretazenil is ineffective by itself) and partially abolished in tests
bretazenil (Ro166028) abecarnil (ZKf12119)
divaplon (RU32698) a!$dem Fig, 1. Slructural formulae of benzcdiazepine receptor partial agonists.
where it has only a smaller maximal effect than diazepam. Complete antagonism is obtained against diazepam-induced motor impairment, for example in the horizontal wire test. No clear anticonvulsant antagonism of effects has been demonstrated in the quanta1 effect paradigm. An intermediate type of interaction is found SNhen the cGMP content in the cerebellum is measured to give an indication of Purkinje cell activity: bretazenil is more potent than diazepam in reducing cGMP facilitation of content by GABAergic inhibition, but its effect reaches a plateau at about 50% of the diazepam maximum; if administered in increasing doses
together with a maximally effectivc dose of diazepam, bretazenil
dose-dependently reduces the effect of the full agonist to values
obtained
when bretazenil is given alone6. Bretazenil precipitates withdrawal symptoms in squirrel monkeys made physically dependent on diazepam, albeit to a milder degree than that precipitated by the antagonist flumazen’l or even by an inverse agonist’. CBRepeated
treatment
of
mice
with bretazenil did not result in tolerance to the anticonvulsant
effect, in contrast to the effect of full agonists”*‘“. Desensitization of GABAA receptors and reduction of the positive modulatory function of the benzodiazepine receptor was absent or clearly less marked with bretazenil than after full agonists r3*r4.Again unlike full agonists, physical dependence could not be induced in squirrel monkeys after repeated very high doses of bretazenil (40 mg kg-‘)
TiPS - November 1990[Vol. 111
rat -
open field
continuous monkey avoldance horizontal wrre rat rat -
ataxta ataxra sedation
dog -
monkey -
locomotor
PTZ seizures
1
10
100
1000
Dose (mg kg-‘, orally)
Fg. 2. Profiles of activity of the full agonisl dkwepam (0) and the partialagonisl bretarenil Is). Values are EO, values or minimal effective doses (anticonflict activity). Pli?. pentetrazbe (L9ptazol).
as assessed by challenge with flumazenil’“; moreover, monkeys would not self-administer (i.v.) bretazenil15. 0 Studies using in o&o labelling of cerebral beruodiazepine recepwith [3H]fkJmazenil or tor [“C]flumazenil (PET studies) showed bretazeni: to require higher fractional receptor occupancy than diazepam for the same magnitudes of pharmacological effects (Ref. 16 and E. Brouillet, pers. commnn.). 0 Facilitation of GABA-induced Cl- conductance measured electrophysiologicaily in single neuror?s was less marked with bretazenil than with full agonists”*ls. Can this profile of activity of bretazenil be explained by invoking partial agonism? By analogy witit other receptor systems it may be assumed that the neurosai systems that ere preferentially involved in the various effects of beruodiazepines differ in their GABAAlbenzodiazepine re\:eptor density and reserve (Fi?. 3). Neurons mediating anticonvulsant and antianxiety effects would thus be expected to have the highest receptor reserve; by contrast, a smaller reserve (or no reserve) wocld characterize neurons involved in, for example, arousal control. As a partial agonist,
bretazenil would be able to produce sufficient GABA potentiation in the former systems but insufficient poteatiation in the latter systems to reach the same maximal CNS depression as full agonists. Moreover, bretazenil antagonizes those effects of full agonists that require higher fractional receptor occupancy, such as ataxia and severe sedation. Clinical studies’0*‘9 have established that the therapeutic effect of bretazer;il is a tenth of the effective dose of diazepam in generalized anxiety disorder (double-blind, placebo and diazepam-controlled i study in 225 patients) and panic disorders (double-blind, placebo-controlled study in 24 patients); it has minimal sedative side-effects at this dose. In the latter study bretazenil was given at the first signs of an imminent panic attack. Surprisingly, the drug was found in open noncontrolled studies to be effective in am&orating psychotic sympt0m.s in 60% of 73 patients with schizophrenia according to DMS III criteria” - an effect that is also being considered as a future therapeutic indication. It will of course be important to confirm in humans the absence of reinforcing effects, the reduced tolerance, low physical dependence potential and minimal ethanol potentiation demonstrated in animals.
Clonazepam Clonazepam (Fig. 1) is a classical 7-nitrobenzodiazepine in use as an antiepileptic for over 15 years, which has recently been found to be thera~e~~~cally active in panic disorder. Electrophysiologica117,21 and behavioural studie@ have demonstrated that clonazepam has an int-insic efficacy slightly lower than that of diazepam. Its partial agonism is best illustrated by its incomplete (and biphasicf antagonism of diazepam-induced motor performance deficit*‘. The lower liability for tolerance to the anticonvulsant activity of clonazepam corn ared with a series of full agonists P’ may be related to its partial agonistic properties. FGS205 FG8205 (Fig. 1) is an imidazobenzodiazepine with about a fifth of the affinity of bretazenil for the benzodiazepine receptor ~PZ aitro2? It shows features characteristic of a benzodiazepine receptor partial agonist, and its intrinsic efficacy is probably similar to that of bretazenil. Its anxiolytic-like activity is similar in magnitude to that of diazepam in a rat conditioned emotional beha~ourai response test but it is active over a broader dose range, probably as a result of a reduced sedating effect. FG8205 is approximately equipotent and equieffective with hiazepam in taming aggressive cvnomolnus monkevs23. Its sedativelmus~le relaxing efficacy is much lower thati that of diazepam, the effect of which can be tota!ly or partially antagonized by FGS2OS in these paradigms. The relative efficacy of FG8205 in potentiating the depressant effect of the GABA, receptor agonist isoguvacine in hippocampal slice is about half that of full agonists and the GA3A-induced shift in the receptor affinity of FG8205 is about 75% of the GABA-induced shift for diazepam binding. Imidazo[l,Z-ulpyrimidines Some imidazo[l,2-nlpyrimidines are benzodiazepine receptor partial agonists24. For example, both divaplon (RU32698; Fig. i; Ref. 25) and RU32514 (Ref. 26) are effective in various models of anxiety in rats with a potency similar to that of chlordiazepoxide. Divaplon, but not RU32514, exhibits broad anticonvulsant activity; however, both
TiPS - November 1990 [Vol. 11 I show little or no activity in tests measuring sedation/motor performance impairment. All the c!ffects of these compounds were reversed by flumazenil. RU32514 ,antagonized the anticonvulsant activity of diazepam. Thus, divaplon and RU32514 appear to be benzodiazepine receptor partial agonists with intrinsic efficacies at the lower end of the spectrum. No results of clinical studies have been published. Abecamil Abecarnil (ZK112119; Fig. 1j27,28 is a fl-carboline derivative with high affinity for benzodiazepine receptors. Like bretazenil, it is effective in lower doses than diazepam in tests predictive of anxiolytic activity and in some models of
455 epilepsy, but is clearly less effective in tests measuring sedation and muscle relaxation and is weaker than diazepam in potentiating ethanol and hexobarbitone. Abecarnil antagonizes the depressant effect of diazepam on the righting reflex and performance in a forced motor task. The increase of abecamil binding by GABA (GABA shift) is clearly less marked than the effect of GABA on diazepam binding. Fractional receptor occupancy at comparable endpoints of pharmacological activity is markedly higher with abecamil than with diazepam. This profile of activity characterizes abecamil as a benzodiazepine receptor partial agonist of lower intrinsic efficacy than bretazeni!, with potential as an anxiolytic and antiepileptic. Ciinical
studies
are being
carried
out,
but
no data are yet available. Alpidem Alpidem (SL800342; Fig. 1) is an imidazopyridine compound reported to be anxio!ytic in 17 studies on over 1500 patients with situational anxiety, general anxiety disorder and anxiety in psychotic patients29*30. In animal experiments alpidem produces anxiolytic-like effects in some but not all models and anticonvulsant effects without sedation and muscle relaxation3’,3’. Unpublished results (E. P. Bonetti, J. R. Martin and L. Pieti) suggest that it has a very low efficacy in anticonflict and anticonvulsant paradigms; moreover, relatively high doses of alpidem clearly antagonize the sedative and muscle relaxant effects of diazepam.
Paradigms used to test effects of benzodiazepine receptor partial agonists A number of test paradigms used to characterize agonists.
in animals are commonly benzodiazepine receptor partial
The most frequently used paradigms for assessing potential therapeutic efficacy are anticonflict and anticonvulsant tests: e Anticozflict tests are considered to be predictive of the anxiolytic effects of benzodiazepine receptor ligands. Operant conflict paradigms often involve situations in which lever-press responding by hungry animals is simultaneously rewarded with food and punished by footshock; response rate is increased by both full and partial agonists at the benzodiazepine receptor. In an alternative approach (conditioned emotional response test), a conditioned aversive stimulus is presented during food-reinforced lever-pressing thereby inhibiting responding; benzodiazepine receptor full and partial agonists increase the response rate.
GABA function and so can be titrated
to induce convulsions; in this case benzodiazepine receptor fi?!l or partial agonists increase the convulsive threshold of leptazoi in individual animals (graded effect). Alternatively, leptazol can be given in a convulsive dose to a collective of animals, in which case the proportion of animals exhibiting convulsions is reduced by benzodiazepine receptor full or partial agonists (quuntul cflect). The latter technique does not necessarily permit an assessment of the maximal anticonvulsive effect of a test drug. Undesired effects for the therapeutic indication anxiety are produced by benzodiazepine receptor full agonists and, to a lesser extent, by partial agonists. e Motor impairment can be assessed after drug administration by observation of either the spontaneous behavior of animals or forced motor activity [e.g. walking on a rotating cylinder (rotarod);. In the horizontal wire test, motor impairment can be studied by allowing an animal to grip a wire strung between two posts and then determining if the hanging animal can caise a hindpaw to grasp the wire. - _ e Ethanol interactiotl can be eva!uated in a situation in which, for example, the threshold dose of ethanol required to induce a short loss of righting reflex is admin;stered to animals, followed by the test drug, and any resultant increase in duration over the control situation determined. o Phusical dependence liabilitv can be investigated in animals after ‘chronic administration of a benzodiaz=*nr nartial agonist by either iooking for epine receykv. 1 spontaneous withdrawal signs ;ftcr te~ination of treatment or those precipitated by the injection 05 a benzodiazepine receptor antagonist. e Self-administration of a test drug by an animal (e.g. lever-pressing to activate a pump system to inject a drug via a permanently implanted cannula) provides one indication for possible abuse liability. LI
e Anticonvulsant effects can be evaluated in experimental situations involving electrically or chemically induced convulsions. Leptazol (pentetrazole) reduces
TjpS -
90% receptor reserve
tOO%i
1
4 3
1
0 50% receptor reserve
in terms both of affinity and in degree or direction of activation. However, there is at present no pharmacological or func38 15 Martin, I. R. rtnl. Sot. Nerrrosci. A&r. fin press) 16 Potier, M-C., Prada de Carvalho, L., Verrault, I’., Chapouthier, G. and Rossier, J. (1988) Eur. I. PltantmcoI. 156, 169-172 17 Chart, C. Y. and Farb, D. H. (1985) J. Ne~rrosci. 5, 2365-2373 18 Yakushiji, T., Fukuda, T., Oyama, Y. and Akaike, N. (1989) Br. J. P~r~rt?r~co~. 98,735-740 19 Katschnig, N. ef ni. (1989) in ~ia~a~jcar Prycfriafry (Saletu. B., ed.), pp. 167-169, Ceorg Thieme 20 Merz, W. A. et nf. (1988) r’s~c~zu~/~a~~zffco&y Y6 (Suppl.), 237 21 Eonetti, E. P., Pole, I’., Laurent, J-P., Schoch, P. and Haefely, W. (1987) Nelrroscierrce 22 (Suppl.), 58.2 22 Gent, J. P., Feely, M. P. and Haigh, J. R. M. (1985) Life 5’1%37.849-856 23 Tricklebank, M. D. et RI. Br. J. ~~zffrt?f~co~.(in press) 24 Gardner, C. R. (1988) Drirg Dev. Res. 12, 1-28 25 Gardner, C. R. et ~1. (1987) Br. J. Pll~ri~i~rai. 92, 537P 26 Gardner, C. R., Deacon, R., Mann, G., Budhram, P. and Guy, A, P. (1987) Br. J. PI~armncot. 92, 655P 27 Stephens, D. N. et nl. (1990) J. Pltanrmcat. Exp. Tker. 253‘334-343 28 Turski, L. et al. (19%) f. Pf~a~J?aca~.E.ry. Ti!er. 253, 344-352 29 Morselli, P. L. (1990) P~iar~f~ffc~* psychintr!/ 23, 129-134 30 Morton, S. and Lader, M. (1990) PI~art~tncopsyc~i~atr!/ 23, 120-123 31 Langer, S. 2. et al. (1990) Phnnimcopsychiatry 23, 103-107 32 Zivkovic, 3. et nl. (1990) Pltarrttncopsychiatry 23, 108-113 RIJ32514: 2-benzoy!,.5-met~oxy-6,7,8,9tetrahydro-imidazo[1.2-njquinazoline SDZ208911: N-[(8~)-dimethylergoii~e-8~ yl]-2,2-dimethylpropanamide 502208912: N-[($oi)-2-chloro-f&methylergoline-8-yip-2,2-dimethy~propanam~de DSM III: Diagnosticand Statistical Manual of Mental Disorders, 3rd edition; a set of diagnostic criteria which conform to the Intemationai Classification of Diseases (US)
