Steric Parameters in Drug Design. Monoamine Oxidase Inhibitors and

AND CORWIN HANSCH. Department of Chemistry, Pomona College, Claremont, California 91 711. Received March 27, 1969. Taft's E, parameter ie employed ...
0 downloads 0 Views 721KB Size
July 1969

STERICPARAMETERS IN DRUG DESIGN

647

Steric Parameters i n Drug Design. Monoamine Oxidase Inhibitors and Antihistamines1 EBERHARD KUTTER~ A N D CORWIN HANSCH Department of Chemistry, Pomona College, Claremont, California

91 711

Received March 27, 1969 Taft's E, parameter ie employed t o correlate structure-activity relationships in phenoxyethylcyclopropylamine monoamine oxidase inhibitors and diphenhydramine antihistamines. New E, values for the halogens and certain other functions have been calculated from van der Waals radii using an extension of the approach suggested by Charton.

The usefulness3 of thermodynamically derived substituent constants for computer-based assaults on biochemical structure-activity problems continues to receive more at'tent'ion. While considerable experience has accrued in t'he use of Hammett, consta'nts ( u , u-, u+) from homogeneous organic reaction^,^^^ the use of hydrophobic parameters (log P , 7 r ) 3 , 6 ~ 7has been less thoroughly studied. Still less underst'ood are paramet,ers for steric effects. Taft's E , parameter4 and t,he modified form, E,", suggested by Hancock, et a1.,8 although not extensively studied in homogeneous organic reactions, are beginning t.0 prove of useg in biochemical systems quite different from that in which t'hey were derived. How far E , corist,ant,s and other st'eric parameters such as Exner's molar volume values (JIV) may be of use in medicinal chemical studies remains to be seen. Our initial successesg with E, have prompted this further study. E, constants have been defined by Taft, using the hydrolysis of aliphatic esters as the model reaction or the hydrolysis of ortho-substitut'ed benzoic esters (ESo)for ortho substituents in aromatic systems. The two groups have been related through the methyl group of value 0.00. Recently, Chartonlo has reexamined E, and shown that Taft's observat,ion t,hat, E , parallels group radii can be expressed in quantitative terms. Charton pointed out that for a symmetrical top-t,ype function such as CF3, one can use either a maximum (r,(max)) or a minimum (rv(min))van der Waals radius to estimate the st'eric action of the F atoms on neighboring atoms. The value of r,(min) refers to the junction point of the two F atoms. In his correlations he used y,(min). We have used an average (r,(av)) of the two values given by Chartonlo to calculate E, values for functions not available from Taft's work. This has been done by using the symmetrical functions in Table I for which E , is known and r,(av) can be calculated. From these data we have derived eq 1. In eq 1, the figures in

E,

=

[-1.839 (*0.22)]r,(av)

+ 3.484 (*0.55)

n r S 6 0.996 0.132 (1)

(1) This work was supported by Grant C.% 11110 from the National Institutes of Health. (2) Visiting Scientist from Dr. Karl Thomae GmbH, Biberach/Riss, Germany. (3) C. Hansch, Ann. Rept. M e d . Chem., 2967, 348 (1968). (4) J. E. Leffler a n d E. Grunwald. "Rates and Equilibria of Organic Reactions," John R i l e y a n d Sons, Inc., New York, X. Y., 1963. ( 5 ) C. D. Ritchie and W.F. Sager, Progr. P h y s . Org. Chem., 2, 323 (1964). (6) C. Hansch a n d S. M. Anderson, J . Org. Chem.. 82, 2583 (1967). ( 7 ) T. Fujita, J. Iwasa, a n d C. Hansch, J . Am. Chem. Soc.. 86, 5175

parentheses are the 95% confidence intervals, n is the number of data points employed, r is the correlation coefficient, and s is the standard deviation from regression. Using eq 1, the E , values listed in Table I1 have been calculated. The reason for taking r,(av) instead of r,(min) or r,(max) deserves consideration. If r,(max) is employed, we obtain a calculated value of E, for Br of 0.345 and, if r,(min) is used, we obtain a value of -0.33. From a study of the steric effects of Br, these values seemed too far from the standard value of 0.00 for methyl. There are many instances where Are and Br appear to have about t,he same steric influence; in fact, even their molar volumesll are quite close: Br = 26.19, methyl = 31.48. I n Taft's E," constants12 (from hydrolysis of o-benzoates), Br and RIe have the same Esovalue of 0.00. Although Chartonlohas shown that electronic effects are involved, the net effect is that RIe and Br behave in a very similar fashion. This similarity can also be seen in the AH of the trans + gauche conformational change13 of liquid butane (770 & 90 cal/mol) and liquid 1,Zdibromoethane (730 50 cal/mol). Here again electronic factors are involved, but for our purposes we assume these can be neglected. We have also observed that using r,(av) with biological data gives better correlations than r,(max) or r,(min) in certain examples where we believe the data to be of better than average precision. Of greatest use to us are the values of halogens in Table I1 which cannot be obtained by Taft's original method. I n the following two case studies, wherever possible, we have used Taft's E, values obtained from the hydrolysis of aliphatic esters. Where such were not available, we have used the calculated values of Table 11. Monoamine Oxidase Inhibition.-In a very interesting application of the extrathermodynamic approach to a biochemical structure-activity problem, Fuller, et al., correlated the inhibition of two types of monoamine oxidases by S-(phenoxyethy1)cyclopropylamines.14 From an inspection of the data (Table 111) it was apparent to Fuller, et al., that the same substituent in the meta and para positions showed rather

(1964). ( 8 ) C. K. Hancock, E. A. Meyers, and B. J. Yager, ibid.. 88, 4211 (1961). (9) (a) C. Hansch. E. W. Deutsch, a n d R. N. Smith, ibid., 87, 2738 (1965); (b) C . Hansch, F a r m a e o , 28, 293 (1968); (c) C. Hansch, J . Med. Chem., 11, 920 (1968); (d) C. Hansch and E. W. Deutsch, B i o c h i m . B i o p h y s . Acta, 126, 117 (1Y66). (10) M. Charton, J . Am. Chem. Soe., 91, 615 (1969).

(11) 0. Exner, Collect Czech. Chem. C o m m u n . . 32, 1 (1967). (12) R. W. Taft, Jr., "Steric Effects in Organic Chemistry," M. S. Newman, Ed., John Wiley and Sons, Inc., Ken York, s.Y . . 1956, p 556. (13) W.G. Dauben and K. S Pitzer, J . Phys. C h e m , 68, 53 (1964) (14) R. W.Fuller, b2. M. Marsh, and J hIllls, J . M e d . C h e m , 11, 397 (1968).

*

S

T.IBLI~: 11

van der \Vaal.< E,

I'uiiclion

('m)

radii

1 , .X'L

0.(i!) 0 ih

I .47 l),27 1 .i.i 13I' 0 . ox 1 .Xi I - 1 ) . I(; 1.98 NO? -1.28 2 . .5Bb so, 0 ,2 9 e 1.iiC - - . *I8 - r, (:JIj 3.80d CJIb 0, 1.'Tic a Calculated using oxygen radius only. Fnnctioii coplanar 1(1 iwiction renter. The value of 2.59 is taken from the work of Charton.10 Half-thickness of CBHBused.10 Estimated from I3ondi values: A. Bondi, J . Phys. Chem., 68,441 (1964). e Function perpendicular to reaction center. ( (1

luge differences in activity. They attributed this detrimental influence of meta substitution on inhibitory activity to be due to steric effects. They chose to compensate for this by assigning an arbitrary steric parameter, y, three different values: 1.3 for a single ineta substituent, 1.0 for a meta and para substituent, :ind 2.0 for 3,5 substitution. With these assumptions wc have formulated ey 2 from their data. Equation 3

+

+

1 ) l j o = [0.928 (*0.27)]7 [l..iX5 ( * 0 . 5 2 ) ] ~ 10.283 (*0.29)]7r 5.924 (bO.32)

+

I/

1Y

I'

S

0.940 0.312 ( 2 )

is comparable to eq 2 in every way except that E, has hecri used instead of y. The correlation with eq 3 i h pla1 = [&TO2 (*0,20)]I:',

+

+

11.640 (*0.50)]u [0.198 ( + 0 2 7 ) ] a 4.153 (+0.42)

+

ri

1s

I'

0.9 level (8' test), i t is 4gnificaiit at this lelcl 111 eq 4 A closer study of the three poorly fit drugs might yicld quite useful structure-activity information. Fuller, et a/., also studied the inhibition of h u m n i l nionoamine oxidase. Equations 5 and 6 arise from

s

0.945 0.330

(3)

p150

=

[1.305 (+0.71) Iy

+-

+

[ O X N (hl.(iO)]~ [ O . i 3 I ( * 1 . 0 3 ) ] ~ 6.

+

il

9

1

\

0 9 1 3 Otj91 (*>)

pId"= [1.0:30 (*0.:39)]1% + [l.OhY ( b l . ' 2 ) ] u

+

+

(k0.76)]7r 1.541 ( + O !I 0 935 0 43.5 (ti)

their data in Table 11. Again, gives :t better coirelation than y. The use of molar s-olume instwd of E , or y gave poorer correlation ineta substituents may be involved i n :in intm- or :it1 iiit,ermolecular steric repulsion. It is difficult to w ' how substitueiith iii the mela position could interact strongly with the side chain; therefore, it wemr mo'1 likely that the meta substituents iii some way hinder binding of the n'-phenoxyethylcyclopropyla~i~inesb> 11160 = [O.T66 (+0.15)]E8 [l.T32 ( * 0 . 4 0 ) ] ~ lO.180 (bO.18) 1. 3.996 (=J=0.30) the enzymes. 'The orily meta substituent under coilsideration which i5 ~ r o tof the iyininetric:il top cln+h i. I/ / S NOL. Since t l t i y I ' u i i ( - t i o 1 t i y he-t fit by the 15, ~ ~ I I I I S ~ : ~ I I I 1.i 0 976 0.20:3 (I) derived from its thickness, it would indicate o b t : t i i i ;L ~)ciorerequation having I' = 0.966 and s = effects are due to :t liiud of fit to n surface rxlhcr th:Lii 0.243. \Vhilc the 7~ term ill ccl 3 i5 iiot significant at ciigulfmerit of' t l i ~~ u b ~ t i t u e nby t eiizyinc~. It I Z , e\-

w r y slightly better. It is of course a satisfaction that the theoretically derived E , constants give as good a correlation as the three strictly empirical y values chosen for the purpose of making a good fit. This correlation also supports Charton's idea that E, values (sall be based 011 van der Waals radii. Three of the 18 dcrivatives (3-1Ie-4-C1, 3,5-lIe2, 3,4,5-RIeg) are poorly fit; leaving these aside, the high correlation of ('9 1 is obtained. Using y in eq 4 instead of E,, we

+

+

+

STEX~C P A ~ Z ~ ~ MIN ET L>I~UG E I ~ UESIGN Y

July 1969

citing to think that experimental E, values or those directly calculated from van der Waals radii may be of some general use for enzymic interactions. Antihistamines.-A second example in which we wish to consider the use of steric parameters is that of antihistamine activity. While an enormous amount of work has been carried out in the search for effective antihistamines, a very small amount of data are available on sets of congeners in which molecular modification nas conducted in a systematic fashion amenable to substituent constant analysis. Two exceptional studies are those of Harms and Nauta'j and Ensor, et a1.I6 The former was an in vitro study and the latter an in vivo analysis. It is these two studies on aryl-substituted diphenhydramines of structure I with which ive shall be concerned. Several different mechanisms

R

KY I

have been proposed t'o explain the influence of substituents of the phenyl rings on the biological activity in the di- and mephenhydramine series. Harms and Nauta suggested that in the case of the ortho derivat,ives, intramolecular interaction of t'he ortho substituents wit'h the flexible side chain occurs, preventing a curling up of t'he molecule. AriensI7 has pointed out that electronic eff ect,s, especially hyperconjugation, are important. Other authors have also discussed steric effect's of ortho substituents18 and electronic effects.Ig We have analyzed the problem using regression analysis and subst'ituent constant,s with the objective of disentangling steric, electronic, and hydrophobic influences of the ring substituents. From the data in Table IV on the in vitro activity (guinea pig ileum) of diphenhydramine derivatives we have derived eq 7-15. I n eq 7-9 are compared t,he log BR

log BR log BR

=

=

=

[0.440 (f0.09)]E,"lm - 2.204 (*0.31) n r S 30 0.886 0.307

(7)

[-0.433 ( * 0 . 2 5 ) ] ~- 0.142 (50.43) 30 0.550 0.555

(8)

[2.814 ( f l . 4 ) ] ~- 0.223 (*0.33) 30 0.629 0.519

(9)

[0.492 (f0.14)]E,"8" [0.585 (+1.23)]cr - 2.445 (f0.64) 30 0.895 0.303 (10)

log UIt

=

log U R

= [0.474

+

(+0.12)]E,"," [0.079 (*0.20)]a - 2.429 (*0.64) 30 0.889 0.301 (11)

(15) .'. F. Harms and W. T. Kauta. J . M e d . Cliem.. 2, 57 (1960). (16) ('. K. Ensor, 1). Kusw11, and U . (:lien, J . I'hurniucol. Exptl. Y l t c r n p , , 112,,318 (1954). (17) E. J. Ariens, "hlolecular Pliarmaculogy," .%cademic Press, Inc., New T o r k , N. Y., 1964, p 230. (18) B. Idson, Chem. Reo.. 47, 307 (1950). (19) R . 13. Barlow, "Introduction t o Chemical Pliarmacolupy," John n'iley and Sons, Inc., S e w York, K. Y., 1955, p 266.

ti49

log BR

=

[0.102 (*0.19)]a2 [0.828 (*0.76) ]a 0.132 (50.68) 30 0.578 0.552 (12)

log BR

=

[0.370 (AO.ll)]E,o~m LO.222 ( f0.20) ]ESP- 1.770 ( f0.49) 30 0.905 0.28s (13)

log BR

=

+

[O.3% (+O.lO)]E,o," - 0.264(EsP)' 0.173EsP - 1,325 (A0.53) 30 0.928 0.237 (14) ideal ESP = -0.33(-1.7 - 0.02)

log BR = [0.326 (f0.09)]E,D," - 0.346(ESP)' O.lSSE,P [0.563 (+0.43)]ESp'- 1.878 ( f O . G 3 ) 30 0.945 0.231 (15) idealESP = -0.27(-0.80 - 0.03)

+

single variables E,",", a, and U . The variable Eso>"refers to the sum of the E , values for ortho and meta substituents. The para position is ignored. E$sm also refers only to substituents on the most highly substituted ring. Substituents on the other ring are ignored. The above restrictions mere introduced into the analysis after a perusal of the data and some preliminary calculations. By far the best of the single variable equations is that of eq 7 , employing the steric parameter. The positive coefficient with E , indicates that the larger the substituent, the lower the biological response. A most important point is that the steric effects from the ortho and nzeta positions are so similar that they can be treated together in one term. This strongly argues against an intramolecular action and suggests an intermolecular effect of these groups. The selection of substituents employed in this study does not allow us to make as clean a separation between the roles of a and E, as one would like. However, it is quite clear that activity does not parallel a nearly SO well as it parallels Es0ym. Moreover, the coefficient with T is negative. This negative dependence can be interpreted in either of two ways. The first and most likely is that a steric effect is implied. For the set of substituents in hand, the size of the substituent very roughly sets its hydrophobic character. Hence it would seem that a is telling us the same story that E , relates. Another interpretation of the negative coefficient is that for the set of drugs under consideration, only compounds with superoptimal lipophilic character are in the set. In other words, the set falls on the "linear" portion (having a negative slope) of the normally expected parabola connecting log BR and a. Equation 12 indicates that this is quite unlikely. In the normal parabolic relationship between log BR and a, one expects and finds a negative coefficient with the exponential term. A positive coefficient is meaningless since it implies that as a approaches or - infinity, so does biological response. The linear combinations of E, with a or u (eq 10 and 11) do not result in improved correlations. This again downgrades roles for electronic and hydrophobic effects of substituents. In eq 13 and 14 we have given special consideration to mono-para substituents. That is, ESP applies only to the mono-para substituent. Equation 13 is only a slight improvement over eq 7 ; however, it is interesting to note the negative sign of

+

630

4 !)ti 4 96

4 96 4 . !Xi 4.06 1 !I6 3.72

l~rl~llbstltllted 4-F 44'1

4-fir 4-1 4-.\Ie 4-1