TiPS - October 1990 [Vol. 11] 417
Structure-activity relationships of K ÷ channel openers Gillian Edwards and Arthur H. Weston Seven groups of synthetic agent, distinguished by a combination of their chemical and pharmacological characteristics exert some or all of their effects by opening plasmalemmal K + channels primarily in smooth muscle. Progress over the past two years now allows broad structure-activity relationships to be formulated within many of the individual groups of agent. Gillian Edwards and Arthur Weston reviezo the historical basis of these discoveries and comment on the significance of new developments. They focus on the search for tissue and channel selectivity, two factors likely to be important for the successful clinical deployment of these substances as antihypertensive and bronchodilator agents. K + channels comprise the most diverse group of ion channels so far investigated. To date, at least 13 major types of K + channel are recognized and, within each type, several sub-groups existL Since the topic was last reviewed in TiPS in 1988 (Ref. 2), major advances have been made in the K + channel field which have been aided by a marked increase in the n u m b e r and diversity of synthetic and naturally-occurring agents that modulate the opening and closing of these channels. K + channel openers are powerful smooth muscle relaxants with in vivo hypotensive and bronchodilator activity, originally typified by three substances - cromakalim, nicorandil and pinacidil. Over the past two years the n u m b e r of agents with K + channel opening properties has expanded greatly. They can be separated into seven distinct groups based on their chemical and pharmacological characteristics. A variety of tests for activity have been used to formulate structure-activity relationships. These range from the measurement of hypotensive effects in spontaneously hypertensive rats to the inhibition of agonist-induced contractions in isolated smooth muscle. Electrophysiological studies involving smooth and cardiac muscle in conjunction with an assessment of changes in K ÷ permeability have GiUianEdwardsis a ResearchAssociatein the Smooth Muscle ResearchGroup, Department of Physiological Sciences, University of Manchester, M13 9PT, UK, and Arthur II. Weston is the Leech Professor of Pharmacology in the same Department.
also been performed. Complete structure-activity (SAR) data are often not available and thus detailed assessment of SAR within a chemical series is not yet possible. So far, no l i g a n d - b i n d i n g studies have been successful and patchclamp experiments with these agents have proved difficult, especially in smooth muscle. Thus it is by no m e a n s certain that K + channel openers from all groups interact with the same K + channel or even with the same site on a given channel. In spite of these difficulties sufficient information now exists to allow the broad structure--activity features of most of the seven groups of K + channel opener to be described. Such data are of great importance in the ongoing search for tissue and channel-selective agents that will be clinically useful.
Benzopyran, guanidine, pyridine derivatives The development of the K + channel openers began with three chemically distinct molecules: the benzopyran cromakalim ~,...~J| ........ I 1 LI i Kline Beecham); the pyridine derivative nicorandil (Chugai) and the cyanoguanidine pinacidil (Leo). The benzopyrans The initial synthesis of these agents by John Evans and his group yielded the racemate BRL34915, subsequently given the generic name cromakalim. This was the first agent to be designated a K + channel opener or activator 3 and its structural features, which are typical of the
benzopyran group, are given in Fig. 1. With reference to the benzopyran nucleus, key features for biological activity include a gem dimethyl group at C2, a C4 cyclic amido substituent containing a 2-oxo function and an electron-withdrawing group (typically CN) at C6. Within the benzopyran nucleus both C3 and C4 can be chiral, giving rise to two pairs of enantiomers differing only by the relative positions of the substituent groups at C3 and C4. Cromakalim is a mixture of the trans enantiomers, with biological activity residing largely in the (-)3S,4R form4.5, designated BRL 38227, and now known by the generic name lemakalim (Fig. 1). The (+)-3R,4S enanfiomer, BRL38226 is approximately 100 times less potent 4.5. The cis racemate equivalent to cromakalim has been described, but published details of its biological activity are sparse 6. Recent studies 7 suggest that cromakalim does not adopt a rigid structure in solution, in contrast to earlier reports 8. Until very recently, cromakalim was the standard K + channel opener against which the activity of all others was compared. However, reports of cardiac papillary muscle lesions, a side-effect typical of peripheral vasodilators in certain experimental animals, prompted curtailment of human studies with cromakalim 9. Clinical trials are now proceeding with the active enantiomer, lemakalim; this development of a single enantiomer for clinical use is likely to be favoured by regulatory authorities l°. Little information is available concerning the biological activity of other single enantiomer benzopyrans. However, the activities of SDZPCO400 and WAY120491 (Fig. 1) have recently been briefly described !],~2. SDZPCO400 has a profile similar to cromakalim, whereas WAY120491 is distinguished by a very long half-life, at least in experimental animals. A recent surprise has been the report that the Hoechst compound S0121, a close relative of the relatively inactive BRL38226 (see Fig. 1) exhibits selective relaxant effects on ureteric smooth muscle, with little effect on blood pressure 13. Such a selective action is of great interest since it suggests that small chemical changes can confer tissue (channel?) selectivity.
Fb, e v i e r S c i e n c e Publi,,her,, I.td. (UK)
0105 - ~1471q01502.00
TiPS- October 1990 [Vol. 11]
Lemakalim (BRL 38227) (-)-3S,4R
." H3 ;
SDZ PCO 400
CH 3 CF3CONH
Optimum activity is seen with a 6-membered ring
I I which can be unsaturated I I The 2-oxo function is I . . . . . . , \ I important for hypotensive I J For nypotensive acdvizy, me I \ I activity I ] substituent at C4 is typically I ~ = i " i I cyclic amido; aliphatic amides I [ ~ ] / ~ I are also active. Substituted L I HI / I If c3 and C4 are chiral, I amines are antidepressant I~ ~ ~ | I trans substitution is I without hypotensive properties I ~ "N_~O I opt'mlal for blood pressure ...... '.... ..~.l_.__/~--~-J lowering. The C3-C4 / al~: ~ ~)r/ ~,a 2 I bond can Ioe double, , :'T"6" f " ~ l ~ ~ e l d i n g a non-chiral Electron-withdrawing I ~ -li ,,, ~ substituent at C6; for I ~ ~ 2 ~ ''~1L;113 ~ " optimal activity I / ~ ~0/"!~'CH,'~X CF3 = CN > C2Hs I/ ~ . . . . . . -~. . . . . : ~n3 IrP:;
Pir an n i ~~r~d:a~iiti!g!yaPt2~!e n t ' I
EMD 52692 (SR 44866)
Fig. 1. Structure-activity features of the benzopyran K *-channel openers with selected examples. Lemakalim is the prototype molecule. The agents form two broad groups depending on the presence or absence of chiral carbon atoms (*) at positions C3 and C4. If the molecule contains chiral carbons, biological activity is greatest in the transenantiomer with the absolute configuration 3S,4R. With the exception of SR46142A and S0121 (see text) all molecules exhibit hypotensive activity'in vivo. BRL, Beecham Research Laboratories; SDZ, Sandoz; WAY, Wyeth-Ayerst; Ro, Roche; EMD, E. Merck, Darmstadt; S, Hoechst; CGP, Ciba Geigy Pharmaceuticals; SR, Sanofi Research. Compiled from Refs 6, 11-15, 19 and from Paciorek, P. M. et al. (1990) J. Cardiovasc. Pharmacol. 15, 188-197.
The potential regulatory problems associated with the clinical introduction of racemates 10 has led to the synthesis of non-chiral benzopyrans. These are typified by EMD52692 (E. Merck) and Ro316930 (Roche) which differ further from lemakalim by the presence of a C3-C4 double bond and different substituents at C4 (Fig. 1). The development of EMD52692 has recently been de-
tailed (see Fig. 1) and this agent is currently under investigation as a coronary vasodilator. Ro316930 is the most potent benzopyran so far described in detail and its cardiovascular and bronchodilator profile are currently being evaluated. The hypotensive activity of more potent agents has been briefly described by Bergmann and Gericke 14 (e.g. EMD, undesignated, Fig. 1).
Very recently, benzopyran derivatives like SR46142A with substituted amine moieties at C4 have been described 15 (Fig. 1). These molecules are claimed not to lower blood pressure but to exert antidepressant activity in animal tests. Whether such effects result from an interaction with neuronal K + channels in the CNS is not clear, but these compounds do contain m a n y of the basic structural and stereochemical features of the benzopyran nucleus necessary for such an interaction. Possible effects of K + channel openers on transmitter release are discussed below. Guanidine/thiourea derivatives Pinacidil, a compound developed from thiourea derivatives with hypotensive activity in the early 1970s by H. J. Petersen, is a K + channel opener 16 whose basic structural feature is a guanidine nucleus substituted to yield a pyridylalkyl cyanoguanidine (Fig. 2). Structure-activity studies have shown that N-3-pyridyl substitution and a short, branched alkyl substituent on the N' nitrogen atom is usually optimal. Such a molecule is typified by the potent P1060 (Fig. 2). As with cromakalim, pinacidil is a racemic mixture. The molecule contains only a single chiral carbon atom, with biological activity largely residing in a single enantiomer 17 which exists in the (--)-R configuration (H. J. Petersen, pers. commun.). Eli Lilly have produced further structural modifications to pinacidil in an attempt to introduce tissue selectivity into the molecule (Fig. 2). Relatively simple modifications to the N'-substituent have produced a significant increase in vascular:cardiac selectivity. With pinacidil this ratio is approximately unity, but with LY222675 and LY211808 (Fig. 2) the ratios are increased to 6:1 and 1180:1, respectively 18A9. The Squibb compound EP-A0354553 replaces the N-pyridyl substituent with a phenyl moiety substituted with an electron-withdrawing g r o u p . Although no quantitative biological data concerning this compound have been released it is claimed to be a K + channel opener with an antifibrillatory action (Eur. Patent Appl. 0354553), a property also shared by cromakalim and pinacidil. While the
TiPS - October 1990 [Vol. 11] 419 shortening of action potential duration by N /C N K ÷ channel openers would theoretically be expected to be proCH= CH 3 CH3 fibrillatory, these LY222675 Pinacidil agents can produce (P1134) Squibb EP-A-0354553 either pro- or antiarrhythmic activity de'If R is pyridyl, a 3-pyridyl J Thiourea or guanidine ~O-timal h--otensi .... ' moiety is usually more active L nucleus j is~ . . . . . nyp ..~ .ve .act,vzzy I pending on the expersnort, |th:yr~h~lC~::eSP°v~dlng I ~'"~R"NH-C.N'HiRI-,,~---------I branched alkyl moieties [ imental circumstances ~,~ : r[ ; % I wn,cn may confer tissue I (see Ref. 20). Clearly i i \ I se,ec,,vity I this aspect of their If R is phenyl, :. . . . . . . . . . ": I f the " I..o. or4c. t| I action requires further J substitution is ~ contains a chiral C, I evaluation. loptimal ~ X can be NH, S or NCN. biological activity resides ' If X is S, the resulting in a single enantiomer | Some years ago it thiourea derivative is was reported that the Fig. 2. K+ channel openers less active than the based on the thiourea or guanequivalent cyanoguanidine adrenergic neuron (X=NCN) idine nucleus. The racemate blocking action of pinacidil is the prototype. If the guanethidine (a submolecule contains a chiral carstituted guanidine) bon atom (*), biological activity is greater in the enantiomer could be antagonized with the absolute configuration by K+ channel (R). The designation of the blockers 2z and more Squibb compound has not CH3 recently, evidence was been released; number refers to a European Patent Appliobtained that guanaP1060 LY211808 cation in which the shown benz, a close chemical structure was identified. LY, Eli Lilly; P, Leo, Denmark. Compiled from Refs 16-19 and from Petersen, relative of guanethiH. J. et al. (1978) J. Med.Chem.21, 773-781; Steinberg, 114.I. et al. (1989) FASEBJ. 3, A435; and dine, possessed K + Atwal, K. S. et al. (1990) Eur. PatentAppl. 0354553. channel opening properties 22. Neither pinacidil nor cromakalim has a detectable inhibitory effect Nicorandil, the prototype of this widely regarded as just another on noradrenaline release from pergroup was the first molecule nitrovasodilator. It is now clear ipheral sympathetic nerves23,24 alshown to possess K + channel that nicorandil is a hybrid molthough some effects on cholinergic opening properties 27. However, ecule with K + channel opening transmission in airways have subsequent experiments emproperties associated with the been detected 25. In the substantia phasized the ability of this agent pyridyl moiety and guanylate cynigra, lemakalim, nicorandil and to stimulate soluble guanylate cyclase activation generated by the pinacidil i n h i b i t GABA release, an clase activity in vascular smooth nitroxy side-chain. Structureeffect antagonized by oral hypomuscle 28 and nicorandil became activity data are scarce 29, but the glycaemic agents such as glibenclamide 26. Further work to clarify the effects of K + channel openers in different neuronal pathways is clearly required, but, by analogy, H~ONO, N~~H~oNO2 modulation of transmitter release in the CNS 26 may underlie the KRN 2391 N ~ CN Nicorandil O hehavioural effects of compounds ke SR46142A. •
Pyridine derivatives A separate classification for pyridine derivatives like nicorandil and for the substituted guanidines just described is a matter of some conjecture, since the latter can also be regarded as pyridine derivatives (compare Figs 2 and 3). However, in vivo, agents like nicorandil largely exhibit a coronary vasodilator profile, while the substituted guanidines exert a marked hypotensive effect. While we have chosen to classify the pyridines and guanidines separately, some overlap exists.
N~-.,~ ~'%.; ~.N.~ ~ "i;~ri-':,-ne " ~ ' . , J'l~- -'R1 f nucleus
N~ ' ~ N ~ o , " C ~ c H O
Nitroxyalkyl substituents I increase efficacy by activating l l soluble guanylate cyclase I
R can be O or NCN; I potency ratio NCN > O I
Fig. 3. SAR in K" channel openers with a pyridine nucieus, i4t..orandil is the prototype. The presence of the nitroxy group additionally confers sol,.ible guanylate cvclasestimulating activity. KRN, Kirin Brewery; SG, ChugaL Compiled from Refs 27-51.
TiPS - October 1990 [Vol. 11]
..CH3 H:CH~N-..CH 3
O\ / O "
established antihypertensive agents and smooth muscle relaxants, exert at least some of their in vitro vasorelaxar, activity by opening plasmalemmal K + channels32,33.
O\ / ° .
CH,=CH.CH, I1 /
R1can be NH=, but is typically a substituted primary or secondary amine or a N-containing saturated ring eg. piperidinyl
Pyrimidine K + channel openers These agents were in active development in the early 1960s and are typified by minoxidil. Early derivatives based either on the p y r i m i d i n e or the triazine nucleus (Fig. 4) had poor in vitro activity, and m a n y were found to be pro-drugs requiring in vivo sulphation to become active. It is not yet known whether SKF 11197 or the Upjohn undesignated molecules are also K + channel openers, as results derived from the appropriate electrophysiological and ion flux experiments have not been published.
.NH2 R ~ ; i" i "R3 ~ i~J"~ jL= ~i.~/"~ ~1 ! R~, "R 2
I When R3=O, the I molecule is a pro-drug. Sulphation yields the N-oxide sulphate, active I in v/vo and in v i t r o
Pynmidin'e"or triazine (R=N) nucleus
I R==NH2 or
O\ / O .
Fig. 4. Development of . . . ~ N ~. O "" S~O minoxidil (su;phate). Many early molecules in this series were triazmes, but eva/uation of these ceased in favour of those based on pyrimidine, SKF, SmithKline Minoxidil sulphate Minoxidil & French. Compiled from Refs 32 and 33, and from Freyburger, W. A. et al. (1965) Naunyn Schmiedeberg's Arch. Exp. Path. u. Pharmak. 251, 39-47; Ehrreich, S. J. et al. (1969) Arch. Int. Pharmacodyn. 179, 284-313; Zins, G. R. et al. (1~q68)J. PharmacoL Exp. Ther. 159, 194-205, and McCall, J. M. et al. (1983) J. Med. Chem. 26, 1791-1793.
structural features of nicorandil appear optimal for smooth muscle relaxant activity. In vitro, the n,Zloxy substituent is very important 29 (potency ratio: nicorandil > SG 209 > SG 103; see Fig. 3). In vivo, the cardiovascular effects of nicorandil are paradoxically completely antagonized by pretreatment with glibenclamide30 (which does not prevent guanylate cydase activation, A. H. Weston, unpublished data). The recent develop-
ment of KRN2391 (Fig. 3) represents a logical extension of the SAR. This molecule is more potent than nicorandi131, but the extent to which this is due to K + channel opening or guanylate cyclase activation has not yet been described. Re-discovered K ÷ channel openers Evidence is now available to show that both minoxidil (sulphate) and diazoxide, two well-
,yc.3 NyC.3 H
Electron withdrawing groups (typically CI, Br, CF3) at C6 and/or C7 enhance hypo. tensive activity. Substitution with H2NSO2at C7 confers diuretic properties and greatly reduces hypotensive action
C l a N
o// "~o Quinethazone
Diazoxide SG 95213
I I I I
Benzothiadiazines The observation that benzothiadiazines with ~aiuretic properties such as chloroth'azide (Fig. 5) were also antihypertensive agents prompted the search for ~tructures in which blood-pressure lowering activity was predominant. Structure-activity studies soon established that removal of the sulp h o n a m i d o substituent at C7 and its replacement (at C6 or C7) with an electron-withdrawing group enhanced antihypertensive activity. Diazoxide is the prototype substance; additional structural features required to optimize antihypertensive activity are shown in Fig. 5. In recent years, interest
I Methylation reduces potency I / benzothia~.~azine nucleus I Saturated, short chain ............... .~ / . -(n