Journal of Medicinal Chemistry @ Copyright 1964 b g the American Chemical Societu
VOLUME7, SUMBER 6
XOVEJIBER 6, 1964
The Use of Substituent Constants in the Analysis of the Structure--$ctivit:- Relationship in Penicillin Derivatives CORKIS
HAXSCH A S D A. RUTHSTEWARD
DPp'.titimt of Chemistry, Pomona College, Claremont, California
Receicecl JIaj/ 9, 196'4
h substituent-constant analysis has been made correlating the relative activities of a series of 22 penicillin derivatives by means of tn-o parameters .T and T . Correlations have been obtained for both in z'itro and in z'ii,o experiments with two strains of Staphylococcus aureus. From this analysis it is apparent that the primary effect of side-chain substituents 011 biological activity results from a modification of the lipophilic character of the penicillins. Electronic and negative steric effects appear to be of minor importance. 11-e llave recently found' * t h a t by using two substituent coiistants u and T the effect of substituents on the biological activity of a parent molecule can be rationalized This paper is concerned with the extension of this approach to the excellent work of Gourevitch, Hunt, arid Lein3 on substituted penicillins. These investigator? studied the activity of derivatives of I.
;H
,S, ICH.)~CHCOSHCH-C"
0
,CH C-CH
O(!-S-(!"COOH
view. With these assumptions and considering steric effects to be constant, we have developed1,* eq. 1. In log 1/c'
?.
I
In I. X represents one or more substituents on the phenoxy ring as indicated in Table I. The side chain was varitd ironi methyl ( n = 0) to n-butyl (72 = 3). It is asbunled that substituents S in I might modify the parent niolecule in three primary ways: steric, electronic. and lipophilic. The latter two effects can be estiiuated by means of the Hanuuet4 constant r and our recently developed' constant T . r is a measure of the electronic effect of X and T is defined as: T = log P s - log P H , where P s is the partition coefficient betm een l-octanol and water of the given derivative, and PH is the partition coefficient of the parent coiiipound. Tor analysis of the penicillins we have used values of T obtained froiii the phenoxyacetic acid.' We have asemned that the rate a t n hich biologically active iiiolecules iiiake their way to the sites of action in a particular cell in a given organism is highly dependent on T and that the reaction a t the site of action can be treated as pseudo first order froiii the kinetic point of ( 1 ) C Hansch R Nuir, T Fuiita, P P Maloney F Geiger and hl. 3treich J A m Clirm. S O C ,86, 2817 (1963) (2) C Hansch and T Fujita, zbzl, 86, 1616 (1964) (3) \ Goureiitch. G h H u n t and J Lein A i i f i b i o t Chemotherapy, 10, 121 (1960) (4) H H Jaff6 C l m n X e i , 63, 191 (1953).
-kT2
+ k'T + + k"
(1)
pU
eq. 1, C represents the concentration of drug giving an equivalent biological response. I n the present instance this is the CD,, n-ith mice and the minimuin inhibitory concentration with the in vitro tests on bacteria. Equations 2-4 were generated froin a least-squares fit of the data in Table I to eq. 1. The correlation coef-
+ 0 019.T + ,5 731 log 1 / C = 0 0 3 i T 2 - 0 . 6 1 3 ~+ 5.i56 log 1/C = -0.443s + j.673
log
0
=
1/c =
0.033T2
r
S
- 0.6lOT
0918
0 192 ( 2 )
0.918
0 . l 8 i (3)
0.909
0191
14)
ficient is represented by r and s is the standard deviation. ,411 of the points in Table I were used to derive the constants except two. Coinpounds 21 and 22 in Table I were oiiiitted in deriving the numerical coefficients in these equations. The 2-chloro-4-phenyl derivatives was omitted because a precise value was not given for its activity. The acetaiiiido derivative was omitted because it is quite susceptible to hydrolysis and was therefore not expected to give consistent results in the in viwo tests. The results of the regression analysis as highlighted by A log 1 'C in Table I are, on the whole, quite satisfying. Of the 22 derivatives, only the 4acetaiiiido is badly predicted, and this mas anticipated. The correlation coefficient of 0.91 for all of the coinpounds except 21 and 22 is quite good considering the uncertainties involved in the biological testing and the necessary assumptions made in deriving eq. 1. Comparison of eq. 2 with eq. 3 in which the U-teriii has been dropped shows the great advantage in the use of the substituent constants T and u to separate the electronic effect of substituents. I n this instance it is quite clear that the electronic effects of the groups attached to the phenoxy ring are not important except in so far as they affect the partition coefficient of the
691
Iiydrolysis oi both the aiiiide side chain aiid the Iactaiii ring. I t has been recognized for soiiie tiiiic that oiic of t h e iiiaiii wcdaesscs of the penicillins is their liability to iiiactiyatioii through the opening of' tlir lactaiii ring ty,. tlic penicilliriase which is produccd by the bactciG. 'L'lic iide chaiii, ('cHSOCH(CH3) considered by (io111 \.itch, ('1 ~ l . would , ~ have a a-vnluc. of about 2 .i. ' I ' l i t b atwvcx aliphatic cliaiii ivitli T I = 1 iiould have a valiics oi ahout 0.8 aiid a valur of 1.3 u-itli ) i = 2. Iletcrocyclir uch ah pyridine could also tw uwd to :dvaii~ o u l dseeiii .\vel1 woi tlin-hile to iiivc.:,tigatcs t I m e as n ~ las l other hydrophilic t'unctioiis. Thc derivatives in Table I w r c . also tested in t h i ottiw z n titi o tests in addition to tho i l l riuo iiiouw t+t I i i uuc', th(>Sniith strain of LSlaphylococc~rsu t i / ( u s n ah used and i i i a n o t l w thc iiior~I vsistant ELI< straiii of this organisin \vas eiiiployrd. r'siiig the constant. fro111'I'ablc 1 aiitl thcl data in 'I'ahle IT \\-e liave oljtaiiic.cI
.
iiiolecule in question. This would appear to iiieaii that the fuiiction of the phenoxy side chaiii is that of affording lipotropic character to the rest of the niolecule. It may also play an important role in sterically hindering hydrolysis of the lactaiii ring. Another iiiiportant consequence of the good correlation obtained with eq. 3 is that steric effects of substituents on the phenoxy ring are shown to be of minor iniportance, at least within the limits set by such substituents as phenyl, 4-t-aiiiy1, arid pentachloro. This also appears to be true for the alkyl groups (CH, through C4H,) on the side chain. Such a point is not always easy to differentiate from other substituent effects and again illustrates thv usefuliiess of thtl p-U-: analysis.? As we pointed out earlier,? wheii the eoiiipounds in a given biologically active series have S :-values which are rdatively far from the ideal value, one can expect a pseudo-linear relation between 2~ and the logarithm of tlic biological response. X coiiiparisoii of eq. 3 and eq. 4 shows that in this case we are operating rather fai, f i . o i i i the ideal 2:-value (the T ? terni in eq. 3 is not statistically significant). Thus, cq 1 call he taken as describing adequately the situation for t h i t i:, with this equation that thc result3 i i i Table I i~erc' ohtnii ed. T I I P iiicist interesting aspect of eq. 1 is tlic negativcl coefficient associated with X . Thi5 would indicate the i i , u c ~ l i 11ioi'e active derivatives could he obtained by usiiig suhstitucnts in I which have iirgative T-values. Uniortuiiately only one such function was tested, the 4acctaiiiido ( T for -OCH3 is too siiiall to be significant). Ot t h e groups onc iiiiglit consider to \IC>suitable for devi*easingthe lipotropic charactcr of I. a function of IOK iiictaiiolic susceptibility, is CHdS02-( X -0.3).' .Itlrcrcmc in lipophilic charactcr could also be accoiiiplihlicd hy oiiiitting the aromatic ring arid eiiiploying iiistc>adail aliphatic group nitli a highly branched side cliaiii. Such a brariched chain as CH30(CH2).C(CT13)2 iiiight h a w the added advantage of showing the
-
(
(~15 . and tj tor t h e Smith strain and eq. 7 aiid 8 t o t tlicx i.eaults n-ith the U L K strain. 111 ccl. 3-8. C' is the niiiiilop 1
,(
=
- 0 Ob37r:
+ i).,'3La + 1) 2iOc
log 1 / ( = 0 LJOu
log 1 /('
=
+0
li j 4 J
i
7.24
- 0 0 1 77r2 - 0 Oj\r
i
0.4.39
0 ,314
IlS47
O