The Structure-Activity Relationship in Barbiturates and Its Similarity to

The Studies on the Structure-Activity Relationship of Allyl Substituted Oxopyrimidines Searching for the Novel Antagonist or Agonist of Barbiturates t...
0 downloads 0 Views 933KB Size
Journal of Medicinal Chemistry 0 Copyright

1867 by the American Ctiemacal Society

VOLUME10, NUMBER 5

AVGVST 25, 1967

The Structure-Activity Relationship i n Barbiturates and Its Similarity to That i n Other Narcotics’ COH\YIX H A S S C HAXD ~ ~ SVSAA11. AXDERHOS?~’ I)c.pctrlrt~rnlOJ Chemzstrv, Pomona College, Clnrrnront, CnlifoiII 111 IZeceiced A p r i l 67, 1967 From a study of the strlictiire-activity relationship of barbitiirat>esiii foiir biochemical systems, inhitiition of =1rhnciu egg rell division, inhihitioil of rat braiii oxygeii consiimption, hypnotic>activity, mid iiihibitioti of S h l ) H oxidation, it is foiind that very good striictiire-activity corrdatioiis can be ohtilitled coiiqideriiig oiily the relative hydrophobic character of the varioiis derivatives. Steric aiid electi,oiiic effects play a miiior role. Comparison of the structiiral reqiiiremeiits for activity of the barbit.iirates Kith inhibitors of bacterial liimiiiescetice, paramecium mobility, liing oxygen coiisiiniptioii, giit contractility, histamiiie release, narcotic actioti on frog heart and muscle, and narcotic action oil tadpoles aiid C-mitosis iiidicates a very similar strurtiire-activity relationship for a great variety of molecules in a variety of processes. The value of a aiiigle mathematical reference system for comparison of experimental systems aiid results carried olit originally for different piirposes is stressed. It is suggested that, interference with electron transport may be a conimoii mechanism of action for the different compounds in the various systems. The reuiilts of our aiialysis also cotistitute further evideiice for the addit,iveconstitutive nature of octanol-water partition coefficients and illustrate the application of this principle to prohlems of interest to the medicinal chemist.

Recently3-6 we have been attempting to place the discussion of structure-activity relationships of biologically active compounds on a matheniat ical basis. To this end we have developed an expression for the cwrrelation of the change of biological response for a set of congeners with extrathermodyiianiic~~lly related6 substituent contants. -Assuming that an equivalent biological response for a series of drugi c:m be related to their effects on one rate-controlling reaction whose rate or equilibrium constant is represented by lix, we c‘ari write4 eq 1. In eq 1, relative biological response is L(I/C.x) = - 1 s k L

(1)

defined in terms of the applied molar coilcentration ( l / C biologic*alresponse) of drug X producing a standard response in ti coiistant time interval. A x is the probability of drug X reacahiiig the site of action during the test time. \I;e have assumed4 that A is a function of the logarithm of the partition coefficient of the drug (thus it is related to its free energy of transfer from phase to phase) and -4 has the form of a normal digtribution. In ey 3, a and b are ronstants and P -4s = a esp [ - (log PS - log Pos)’/b] (2)

is the partition coefficient. We have normally used 1-octaiiol-water to represent the aqueous and lipid phases of the cell. Substituting eq 2 into eq 1, taltirig logarithms. and collec~ting constants. we obtain, remembering that 1’0 is a constant for a given system log (l/CS) = -kilog Px)2

+ k’ log P., + log kx + k”

(3)

The first two terms on the right of ey 3 take into accmmt the differewes in drug activity due to dfierences in the random-walk process bj- which drugs find the active sites. They may not find the receptor sites duriiig the time of test because of being localized in lipophilic pools, niet:tbolir destructicni, or elimination. All of these processes appear t o be highly dependent on log P . The effert of structure variation on Xs can be treated by the techiiique of 1ine:ir combinatioii of free-energy-based substitueiit caoIistants.j*6 We have normally replared log 1;s in eq 3 using espressioiis such as eq 4. The constants a, b , c are different from those log kx = a log Px + pu + bE, + c (4) in the previous equations. In eq 4, a log P X acrouiits for the free-energy rh:tiige in the hydrophobic biiidiiig of the drug to a critical enzyme or protein.’ The usu;il Hanimetts significance is attached to pu and E, is the Taft steric parameter. I t is possible to formuhte a higher order approximation as suggested by llliller.g This can be done utilizing eq 23 of llliller’s paper. Recasting his equation in terms o f log k x , log P x , u, aiid E , instead of his parameters

(1) This work m-as siipported l)y Research G r a n t G l I 07492 from t h e X a tional Institutes of IIealth. ( 2 ) ( a ) J o h n Simon Giiegenlieim Fellou; (1,) Smith Kline and French research associate. ( 3 ) C . Hansch, R. hf, hluir, T. Fujita, P. P. Maloney, C. F. Geiger, and A I . J. Streich, J . A m . Chem. Soc., 86, 2817 (1963).

(4) C. Hansch and T. Fujita, ibid., 86, 1616 (1964). ( 5 ) C . Hansch and E. n’. Deutach, Biochzm. Biophys. Acta, 116, 117

(1966). (6) J. E.Leffler a n d E. Grunwald, “Rats and Equilibria of Organic Reactions,” J o h n Wiley and Sons, Inc., Kea- York, N. Y., 1963.

745

(7) C. H m s c h , E. \V. Deutsch, and R . N Smith, J . 4 m . Chem. Soc.. 87, 2 i 3 8 (1965). ( 8 ) H. H. Jaff6, Chem. Rev., 83, 191 (1953). (9) S. I. Miller, J . A m . Chem. Soc., 81, 101 (1959).

7 Ili

y, .c, z , and w , we ohtaiii eq log kx = u log

5. Substitution of eq 5

+ pu + bE, + c(log P X ) U+ d(log P x ) E , + euE, + .f(log P x ) u E , + g

Px

(5)

into eq 3 yields an expression (eq 6) which, if the delog (1/Cx) = -kilog Pa)' ki(log P X ) U h.4(log P s ) E s

+

+ kl log Px + p~ + kiE, +

+ k : ~ E s+ ko(10g Px)uE, +

h.7

(6)

peiidence of log kx on log P , g , and E , is strict1)- linear, should give a better mrrelation than our previousljformulated expression. The difficulty, of course, with rq (i is that one does not often have sufficient data to btatisticwlly validate the nine constants, k-k, arid p . However, often it is unnecessary to include E , and eq 6 then reduces t o eq 7. For the work in this paper we log (l/Cx)

=

-!€(log Px)Z

+

h.1

log Ps

+ pu + ks(log Px)u

+

k7

Methods 111 our preliniinary report on harhiturates15 we correlated substituent effects for a single series of harbiturates using T values for substituents arid log P for barbituric acid as our base of reference. Barbituric acid is not t,he best reference molecule to use to correlat'e 5,5disubstituted barbiturates. Disubst'itution shields hydrophobic. and hydrophilic interactions of the 5varbori atom and groups adjacent to it. I n this work we have used S,5-dieth~,~tsarbitLiric acid (log P = 0.65 =t 0.02) as our reference molecule. We have also determined log P (octanol-water) for the ethylphenylbarbituric acid (1.42 =t 0.01) arid barbituric acid ( - 1.47 0.03). The value of barbituric acid is more accurate than our previously reported value.'j The shielding around t>he5-carbon atom by two alkyl groups can be estimated as follows. The calculated value of diethylbarbituric acid, assuming simple additivity, is

*

+ log I.' (barhitiiric acid) = 2.00

-

1.47 = 0.53

The difference between this and the experimental value (0.65 - 0.53 = 0.12) represents the increase in partition (10) A . hlbert, "Selective Toxicity," 3rd ed, .John Wile? a n d Sons, Inc., New l a r k , N. T.,1965. i l l ) T. Fujita, J . .Wed. Chem., 9 , i 9 7 (1966). K . Keltch, a n d lf.E. Krahl, J. Phnrmacol. Ezptl. Therap., 68, 312 (1940). (13) .M. E. K r a h l , J . P h y s . Chem., 44, 149 (1940). (14) F. F. Rlicke a n d R . H. Cox. "Medicinal Chemistry," Val. IV, .John \Tiley a n d Sons, Inc., New T o r k , N. Y., 1959, p l i . 115) C . Hansch, A . R . Steward, a n d .I. Iwasa, .Wd. F'hw-mtccol., 1, 87 (1965).

OF

T.\BI.EIT7 YADH OXIDATIOK

BY B.iRBITURh.TES

~l (2.14) by (iimd 5, respectivel!-. gives 0.42 : t i i d 0.44. Subtrwc*tiiig log I' 0.29 for ethyl niethJ.1 ketone from log E ) 1.50 for methyl 2-i.yclopropyl ketoneL6 gives t h e value of 1.21 for cyclopropyl; dividing by :3 yields 0.40. The x value for ( Olohexyl can also be of

I)lieticis!.:ic,~,tic.

ciilculated from log P cycalohexariol (1.23) and the x value for OH ( - 1.lG). Dividing this value of 2.39 by (j yields 0.40 for the cyclic CH, unit. Thus t h p value of 0.41 seems like a reasonable compromise. 1;or thc benzyl function in 3 and 4 we have used log m. Chem. Soc.. 86, 5175 r l i ) (a) 'r. Fujits, J. Iaass, and (2. Hansch rlt)84\;

(171

(',

Hanacti and '3. 31. .\nderson, . I . O r g . C h e m . . 33, 2.583 f l 9 6 7 1 .

From ref 26. Tadpoles of various ages were risrtl i i i t he I I + I \. The esters were tested on 3-day tadpoles, the ketol~eso n 7-(i:1>, the group from chloroform to et her ( i n 6-day, the :d(whols oII l i day, and the (::irbsmates on LI-day. t, Caltbrilated il+ilig r ~ :(it inhibit i o i i of gut COLIti,actilit y2J (Jg ( 1 /c) = 1.060 l(Jg 1' 0.621 (i0.l)i) 1xk0.11)

+

nli*ohol>(Ic; iiihibitioii ( i f histaniiiie l,ele;l.e2~ og (1/C) = 0 . 2 1 I ( J P ~ 1.O.iO i *(i.l).j) f i0.06)

+

7 0 . !)97 0.0H:i

It is of interest to ronip:ire the :i,bove ecpatioiis tor c.losel!. related sets of c*origeliers with iiiore c ~ m p l e x mixed sets. 1~:yu:itioii 34 is derived from l;uh~iei*'s datn i i i Table 1-11o i i the iiihibitioii of frog heart actioii. eq i3.5 vonies from the data of Overtoii (111 the narcosis of tadpoles, cq c~)mesfrom the data iii Table TTIII o i i the iiaivosis of frog muscle. ey 3 i and :38 are derived froiii T:ihle IX for tadpoles, aiid e(1 39 (wiies from thc d:itn i i i Tahlv X oii the iiiductioii of' C-mito CJIliOllS.

The slopes for cy 34-39 arc very close t o those for cq S, 13, and ".i-;33 iritlic.:itiiig quniititativel~~t h r siiiiiI:irit>- of merh:iiiisni of :irtioii :is defined b!. t h c oc~aiiol-~v:iterreferelive system. E;quatioii 35 has hrwi rederived using improved log I-' values:32mid ii slightl!. better correlatioii thaii O U previous ~ oiie16 is obtaiiied. The average value aiid the standard deviation for the shpes for the :thovr 11 equations (ornittiiig ey 33 aiid 38) is 1.03 0.13. The guod currelat,ioiis and the misistent value of 1 for the slope in the 11 examples is evidence that the octanol-water model is a satisfac1 ory oiie t o re1)reseiit the :iqueous :tiid lipophilic, 1jioph:ises. The ineaiiiiig of the iiitcrcqt is hard t o iiiterprct.

*

is a surprisingly uniform process with the slopes of dependence on log P falling in the range of 0.5-0.7 for those systems so far studied. This is quite a different dependence on hydrophobic character than we find in eq 8, 13, 25, and 27-39 and suggests that the simple adsorption of narcotics onto proteins is not the cause of narcotic action. It does not eliminate the possibility that perturbation of a protein by hydrophobic bonding is the cause of narcosis. From the above studies as well as our previous ones, sufficient data have accumulated t o show that the octanol-water model suggested from the work of Collander is a very useful reference system with which to study the interaction of organic compounds with biochemical systems. The exact meaning of such correlatiom is riot clear. There are a t least two important aspects. S o doubt increasing lipophilic character, up to a point,*B facilitates movement of the narcotics through lipophilic biophases onto the sites of action. However, the ultimate measured response is most likely brought about by interaction with an enzyme or lipoprotein membrane which might or might not be supporting enzymes. In such material, hydrophobic interactions could easily produce conformational pert u r b a t i o n ~ ~which * could disrupt cellulur processes. The great variety of saturated arid unsaturated, aliphatic and aromatic molecules in Tables I-X have nothing in common except relative lipophilic character. It is difficult to imagine a common mechanism of action other than solution in a lipoidal membrane. The slopes of essentially 1 in the 14 above examples in which the nieasured responses are so extremely diff erent imply that a common rate-limiting step is being influenced. The fact that so many diff ererit processes have the same dependence on log P and, except for the slightly greater activity of barbiturates and carbamates, the same concentration of isolipophilic compounds produce the same response would indicate that the mechanism of interference in the biophase is exactly like the transfer of the compound from water to octanol. All of the processes being inhibited are energy-requiring processes arid we might illustrate the problem in the following diagrammatic way. Certainly all of the steps a --t 11 in the energetic processes shown cariiiot be the snnie. The observer watching egg cell division, DPSH (40) l3. Uelleau.

J. M e d . Chem., 7, 776 (1964).

energy reservoir1 -+ a1 -+ bl 4 cl

.. . , .

4

ill

-t

f

movement of frog muscle

energy reservoir:! + a2 4 b2 4 c2 4 . . , .

4

ti2 .-c

energy reservoirs -+ a; .-c b8 -,c8 -t . . .

4

n3 4 C-mitosis

f.

4

consumption of 02 by brain

common sites of narcotic action

oxidation, etc., categorizes each molecule's activity on the basis of an ultimate response very likely many steps removed from the site of inhibition. Considerable evidence has accumulated to indicate that narcotics arid C-mitotic agents inhibit oxidative metabolism through int'erference with electron transport.z0137841-44 This appears to us to be a most attractive working hypothesis for the mechanism of narcotic action. Such interference could easily be the result of slight conformatiorial changes in a membrane which would, to varying degrees, disrupt the flow of electrons. The lipoidal material would seem to be rather loosely structured and very similar in nature or else one would not always get the same change in response for a given A(1og P ) . To account for the difference in intrinsic activity of the barbiturates or the carbamates, one would assume that a more specific interaction of these functions with polar couiiterparts in a lipoprotein matrix would give leverage to the hydrophobic interactions so that a greater conformat'ional change would be produced by molecules with no more hydrophobic bonding power (as defined by log P ) . In summary, we can say that the use of a standard set of substituent constarit#sallows us to make precise coniparisoris between diff ererit drugs acting in the same system and the same drugs acting in different systems. As more such correlations are made, they should form the basis for a quantitat'ive approach to pharmacodynamicas.

Acknowledgment.-We wish to thank the Eli Lilly Company for a sample of ethylcyclohexenylbarbituric acid. (-11) J. H. Quastei a n d .\. H. 11. \Yheatly, Proc. Roy. SOC.(London), I l S B , 60 (1932). (42) .I. 11. Quaatel, I'hurmrml. R e v . . 17, 198 (1965). (43) \V. N . .\Idridge a n d V. H . Parker, Biochem. J . , 76, 4 7 (1960). (44) l3. Chance a n d G. Hollunger. J. B i d . Chem., 238, 419 (1963).