Journal of Medicinal Chemistry, 1970, vol. 12,
THYMIDINE PHOSPHORYLASE INHIBITORS
6 913
Structure-Activity Relationships i n Thymidine Phosphorylase Inhibitors.' A Correlation Using Substituent Constants and Regression Analysis EUGENE COATS,W. R. GLAVE,AND CORWINHANSCH Departrncnt of Chemistry, Pomona College, Claremont, Calafornia 91Y l l Received M a y 7, 1970 The structureactivity relationship of Baker's study of several types of thymidine phosphorylase inhibitors has been made using substituent constants and regression analysis. The quantitative relationships formulated support Baker's qualitative conclusions. Activity values of 6-anilinouracils calculated from a regression equation based on 19 derivatives yield values in rather good agreement with those estimated by Baker. This substantiates the premise that additivity principles can be used in the design of enzymic inhibitors.
Attempts at relating chemical structure to enzymatic binding and reaction rates in quantitative terms have frequently proved fruitless when using a single-parameter approach. Recently, several meaningful relationships have evolved2! utilizing a multiparameter approach in conjunction with computerized regression analysis. Three of the more useful parameters for such studies are Hammett's Q constant4 and analogous parameters5 for electronic effects of substituents, Taft's E,6 parameter and its modifications3,' for steric effects of substituents, and ir8which is related to the lipophilic character of a substituent. Thus it has been possible in many cases to delineate the relative importance of a certain parameter with regard to a specific type of biological action and relate this result to structureactivity correlations. The potential significance of such studies with regard to cancer chemotherapy along with successes in correlating the binding of molecules to enzymes with the physicochemical parameters for their substituents has prompted an examination of the factors involved in Baker's studies of the binding of pyrmidines to thymidine phosphorylase using regression analysis. Thymidine phosphorylase (EC 2.4.2.4.) catalyzes the reversible phosphorolytic cleavage of thymidine and other pyrimidine deoxynucleosides. Thymidine phosphorylase activity has been demonstrated in a variety of sources including mouse tissue,g horse liver,'" rat 1iver,l1 human spleen, l1 Escherichia coZi,12 Ehrlich ascites tumor,13 and others.14 Although the enzymatic reaction is reversible, the major function in human tissues appears to be catabolic and thus it has been shown to be responsible for the rapid
cleavage of the potent antimetab0lite,'59'~ 5-fluoro-2'deoxyuridine to 5-fluorouracil, leading to a lowered therapeutic effectiveness of the former compound and a much greater toxicity due to the latter compound. Inhibition of this reaction should therefore increase the efficiency of this and other nucleoside antimetabolites. Since thymidine phosphorylase appears to bind specifically to the deoxyribosyl portion'7918 of the nucleoside, and in view of the product inhibition seen with thymine, recent studies have concentrated on the modification of the pyrimidine moiety. 19-28 The extensive investigations conducted by Baker and his coworkers20-28have provided the data for this analysis.
(1) This work was supported by Grant CA 11110 from the S a t i o n a l Institutes of Health. (2) C. Hansch, J . M e d . Chem., 11, 920 (1968). (3) E. Kutter and C. Hansch, ibid., 12, 647 (1969). (4) L. P. Hammett, "Physical Organic Chemistry," RlcGraw-Hill, New T o r k , N. Y . , 1940. ( 5 ) .J. E. Leffler a n d E . Grunwald, "Rates and Equilibria of Organic Reactions," Wiley, N e w York, N. Y., 1968. (6) K. W. Taft, Jr., "Steric Effects in Organic Cliemistry," AI. 9. S e w man, Ed., \Vile>-, New York, N. Y., 1956, pp 548-604. (7) bl. Charton, J . Amer. Chem. Soc., 91, 615 (196Y). (8) T. Fujita, J. Iwasa. and C. Hansch, i b i d . , 86, 5175 (1964). (9) T. A . Krenitsky, M.Barclay, and .J. A. Jarr~iiez.J . B i d . Chem., 239, R05 (1964). (IO) M. Friedkin a n d D. Roberts, i b i d . , 207, 245 (1954). (11) bl. Zimmerman, ihid., 239, 2622 (1964). (12) W.E. Kazzell and H . G. Khorana, Biochim. Biophys. A c f o , 28, 662
(16) C. Heidelberger, L. Greishach, 0. Cruz, K. J. Sohnitzer, a n d E. Grunberg, Proc. Soc. E z p . B i d . .!-fed., 97, 470 (1958). (16) G. D. Birnie, H. Kroeger, and C . Heidelberger. Biochemistry, 2, 567
(1958).
(13) H. Pontis, G . Degerstedt, a n d P . Reichard, ibid., 61, 138 (1961). (14) T. A . Krenitsky, J. IT. Llellors, a n d R. K. Barclay, J . B i d . Chem., 240, 1281 (1965).
Method The inhibitory activities of the pyrimidine molecules have been reported in terms of the ratio of inhibitor to substrate affording 50% inhibition of the enzyme [I/S]o.j. The substrate in all cases studied by Baker was 5-fluor0-2'-deoxyuridine. The relationship between this ratio and the Michaelis-Menten constant is represented by eq 1where K i and Kmare the enzyme-
inhibitor dissociation constant and the enzyme-substrate dissociation constant, respectively. I is the concentration of inhibitor and S is the concentration of substrate. It has been shown that [1T/S]0,5for two inhibitors is related to the differences in free energy of the two molecules.29 As a first approximation, the free energy (AGBR') in a standard biological response can be factored as follows:
(1963). (17) P. Langen and G. Etzold, Biochem. Z . , 339, 190 (1963). (18) B. R. Baker, J . N e d . Chem., 10, 297 (1967). (19) P. Langen, G . Etzold, D. Barwolff, and 1:. Preussel, Uiochrtn. Pharmncol., 16, 1833 (1967). (20) 13. R . Baker and AI. Kawaau. ./. A/&. Cham., 10, 302 (1967). (21) 13. R . I3aker. X I . K a w a ~ u ,I).V . Santi, and T.T. Aclivan, i h i d . , 10, 811 (1Y67). (22) B. R . Baker and M. Kawazu, ibid., 10, 811 (1967). (23) B. R . Baker and M. Kawazu, i h i d . , 10, 313 (1967). (24) B. R. Baker and M , Kawazu, ibid., 10, 316 (1967). (25) B. R. Baker, M. Kawazu. a n d J. D. McClure, J . Phorm. Sci., 66, 1081 (1967). (2ti) 15. It. Baker and A I . ICa\\a&u,ibid., 66, 1086 (1967). Chem.. 10, 1109 (1967). (27) B. R. Baker and IT. Rzeszotarski%J . (28) €3. R . Baker and Vi-. Rzeszotarski, ibid., 11, 639 (1968). (29) I3. R . Baker, "Design of Active-Site-Directed Irreversible Enzyme Inhibitors," Wiley, Piew York, Pi. Y., 1967, pp 202-204.
T , has a lower standard deviation and higher correlation than eq 5. Equation 7, with the electronic parameter u*, shows that electronic factors alone are not important for effective binding by the 1 substituents. Combinations of 7, PE, and u* did not result in significant improvements over eq 6 as demonstrated by the F test ( a 5 0.10). The T values for the phenylpropyl, phenylbutyl, and phenylpentyl derivatives mere d j u s t e d to account for f ~ l d i n gthat ~ , ~can ~ occur with these compounds. The implication here is that when the connecting alkyl chain between the uracil ring : i d the P h ring reaches a certain length, the two T tems are allowed to interact, thus lowering their T valued. If this adjustment were not considered and “normal” T values were used, eq S, of considerably
n log
1 -
c
=
0.421(+0.13)T -
S
1’
11 0.920 0.213 (S)
2.266( *0.40)
poorer correlation, would result. The need for an anion at S1is mentioned in the section on .%substituted uracils. When the H-1 is replaced by small alkyl or aralkyl groups, binding ability is lost due to the loss of the anion at this position. Binding and, hence, inhibition can be regained by bridging this polar area with alkyl or aralkyl groups sufficiently long enough to reach a hydrophobic area. Four compounds were omitted from the correlation. The 2-hydroxyethyl and the j-hydroxypentyl deviated markedly from their observed activities even though suitable substituent constants were available. The other two, CH2C6H4COXH2-pand CHzC6H4SHAc-p, were omitted because of the lack of suitable substituent constants. Of the 11 compounds correlated, T accounts for most of the variance and indicates that binding in this area depends on the hydrophobic charmter of the substituent. Severtheless, polarizability should be kept in mind when considering substituents a t the 1 position. 5-Substituted Uracils.-Table I1 contains the subst itueiit coiihtntits for s-substituted uracils including pK, for the SIH. Equations 9-11 show that T alone 11
log
c
=
0.%19(+0.50)T-
I’
S
s 0.401 0.494
(9)
0.279( +0.46) log
1 c‘ ~
=
1.40(+0.92)u1 -
S 0.h34 0.297 (10)
0.34S( +0.34) log
1 c‘ ~
=
-0.273(f0.15)pKa
+ 2.178(
S OS79 0.257 (11)
f 1.31)
is not very significant, but as Bakerz3has pointed out, the degree of ionization a t iY,is quite important. It is intereqting to note that uI gives almost as good a correlation as pKa. This illustrates the versatility of u constants which in this instance gave good results in a complex heterocyclic system far removed structurdly from tliat system in which they were dcrivecl. ( 3 3 ) C. lransch arid S A I . Andersori, J. O r g . Chern., 33, 25Sd (146TJ.
TABLEI1 INHIUITOKS OF THYMIDINE PHOSPHORYL MI.: tj-SURSTITUTED URACILS
n
Log l/C-O~isdf CaIcdU
7--
Rs
7rh
pK,C
CIe
1s Log
I~CI
5,s 0.76 0.33 0.66 0.72 0.06 0.89 8.0 0.45 0.3.5 0 . 1 1 0.24 0.25 8.0 0..72 -0.11 -0.02 0.09 9.9 -0.08 0.50 -0.28 -0.48 CHI 0.20 9.gd 0.08 CsH: 2.13 -0.30 -0.15 0.15 COCHs -0.55 9.3d 0.18 -0.32 -0.54 0.02 H 0.00 9.5 0 . 0 0 -0.59 -0.48 0.11 KITS -0.84 10.Od 0.05 - 0 , 8 5 -0.79 0.06 Omitted X2+ and COzH. * From o-phenols where possible. pK,for N, H, enzyme assay at pH 5.9; see ref 23. Estimated pk’,. e From R . W.Taft, Jr., Elt’on Price, Irwin R. Fox, Irwin C. Lewis, K. K. Andersen, and George T. Davis, J. Amrr. Chem. SOC.,85, 709 (1963). ’From ref 22 and 23. g Calculated using eq 13. NO? Br F
Addition of a term in T to eq 10 and 11 yields eq 12 and 13. An F test indicates that ey 12 is not a signifin
1 logC
=
+
1.
S
S 0.906 0.249
1.368(+0.S2)~1
(12)
0.194(*0.27)7 - 0.606(*0.31) 1 log -
c
=
-0.269(+0.11)pKa
S 0.956 0.174
(13)
+ 2.080(+0.93)
+0.206(+0.18)~
cant improvement over eq 11; however, the F test demonstrated that eq 13 is a significant improvement over eq 11 (K,j = 8.18). The negative coefficient associated with pK, means that the lower the pK,, the better the inhibitor. The coefficients in eq 13 suggest that an ideal group would possess strong electron withdrawal while being lipophilic enough to take advantage of the positive coefficient associated with 7 in eq 13. Such a group is the SOzCF3group proposed by Baker.23 The Hammett u for SOzCF3is 0.93 compared with KO2 which is 0.78, while T for SO2CF3 is 0.93 (from phenoxyacetic acid system) compared with NO2 which is 0.33 (from o-nitrophenol system). By assuming 90% ionization for 5-trifluoromethylsulfonyluracil (based on p H 5.9 for the enzyme assayz3), the predicted activity calculated from eq 13 mould be log 1/C = 0.95 or [I/S]L.5= 0.11 which compares to [I/S]o 5 = 0.22 for the YO2 group. The dependence on the pK, term in eq 13 would seem to indicate that an anion at N1 is important for binding. Since 1methyluracil shows a 50-fold loss in bindingz3compared with uracil, an anion a t appears to be important for binding. 6-Substituted Benzyluraci1s.-Table I11 cont ains several benzyluracils along with the substituerit constants used for correlating their biological activity. The terms E,“ and ESPrefer to Taft’s steric parameter6 or a modificatioti of the steric purninetcr’~’for the meta and para positions on the benzyl moiety, respectively. Equation 14, with a term for E,“ and
/
/(
log
log
1
c 1
('
I!J)h',"
=
-O.h>(+O
=
0.ill(=k2.12)Tm
\
i 0.07.j 0 1!L5
+ i 0.32.i 0 . i l i 0.40'3(+ 1.t51)7rp+ 1.236(*0.h0)
( I 1I
(13)
accounts for 1ie:irly all of the variance ( r = O.Yi3), while the addition of terms in K or do riot bignificantly improve the corre1:itionb. Ttir negative sign asociated with the steric parameters i l l t ' q I4 buggebty that irihibiting ability 13 rnhuncrd b~ incrr.:i5ing stcric bulk. l'hr dsffererice I I I the coefficit~iiti for the n j e h :mcl pm n steric p:iranieter:, I > indicatis c that bteric bull; playb n siightly largt'r rolrl 111 iiicrtJa5ing the b i d i n g :it thc mptn position thiui i t ( h ~ :it s thri paia position. 'I'hcrtl :ire two coriforniatioiic i n n-liich t h w e compounctc c i m exiyt. The first conformation would place thc p~ rimidirie ring :1nd ph(~11yIring 111 :I V witli the. mcthylerie bridgr a t the :ipt'x while the two ringh :ire iuughly "parallel." 'l'hc. ,-.~coiid conform:ition \\ ould p h c e o i i v of the riiigy perpt~ridiculnr t o the o t h r . Variouc rotonicrc of t h c v coiiiorniatioiis such :L5 the folloli irig :ire posiblc: J3bp,
0
tlie viies oi beiiq 1 ckrivativw \tudictl, it is i i o t po\sible to driiw :in\ iirm coiiclusions as to which of ire inipoitnnt in binding to thc e n z n~i 1 ~ I h? IlbP of rt'gr(Wlor1 nn:II!, u ith rigid :m:iIogh nwulci bc necessary to shed light on this problem. S t ~ v ~ r i niolrculrr ere omitted form the correlatioii, \I-, of then1 duc to the lack of 5uitable iubstitwiit cotiit:irity The >eventh compound, (i-(:~-nitro-4-fliior.o)beiizl liirxi17 for which substituetit constmts \\'crt' av:til:~bl~~, deviated m:rrltedly from the other coinpooid. s t u d i d ,>
A 0
0
0.391 0.344
0 I1
n I1 i o 1 1 1 : L I I iiisl)(sclioii ol lcihw liirsclifcldthr 'l':iylor hpacc' filling niodels, thc conforniationr 13 arid D wcre
ruled unlikely bccause the metu substituents project into the vicinity of the hydrophobic area around the S1 position. The conformations A and C are indistinguishable by this analyii, hydrophobic p a r a n w t w iiidicatc that increasing t h e lipophilic character. of :i wbstituent 111creases binding in the order no > K , , ~> xi,. S o sig1iific:znt improvement i t i eq 23 was obtained !b the addition of v:irious electrotiic parameter.. J t i ot-tltlr to t+t for thv ricc o n i p o i i t ~ ~ i t('(1 , 20 ti
log
1
=
tY
+ 0.712( * O . T O ) x , + O . i S . ? ( = k O . 2 2 ) ~ , ,+ ~ 0.4142(*0.20)7r, +
-0.44.j(*0..j4)I;Q"
I
l!) 0 !W 0 214
(26)
l.(ilh( +=O.fil) \+:i> tleviird 15
hich include> the 1;; term with term*
for x,, xn,. and xp. The coefficient> for x, and x,,! :ire t>sseritiall;\.the 5ame arid suggest that grouping. :it these positiorih bind to tht. bame hydrophobic site l'h(. differencr. i i i t h e coeficirtits for x o in c y 23 an(1 20 may be the result of intramolecular interactiorih of the ot flu) subhtituerit reprevnted by the L',"term.
('otifornxitiom th:it the (i-atiiliriouracil~can .
:II,('
E
F
H
that a hydrophobic area exists beyond the XI position and at the 3 position, and that compounds with an acidic hydrogen a t the 1 position make good inhibitors. Since the 6-anilinouracils show good inhibition, incorporation of this moiety would be advantageous. The following compounds would be predicted to be good inhibitors of thymidine phosphorylase. The long alkyl chain in each compound would be expected to bridge to the hydrophobic area beyond the XI position. The r)-SOXF3 group would provide hydrophobic character as well as lowering the pKa of the S 1 hydrogen. The dichlorophenyl moiety n-ould provide binding in the area where the 6-anilinouracils bind. This study does
0
c1
0
I
I1
support in a quantitative way the qualitative findings of Baker and coworkers. I t also offers ideas for the development of more effective inhibitors.
Mixed Bifunctionality. 111. Antitumor Activity of Sesame Oil Solutions of Simple Alkylating Derivatives of Polynuclear Hydrocarbons’ 1IICHAIiI)
11. P E C K
AND h X . 4
1’.
o’COSSELL
The Institute for Cancer Research, Philadelphia, Pennsylvan iu 19111
Received February 18, 1970 Aiititumor activity of chloromethyl aromatic hydrocarbons is enhanced by administ,ration in sesame oil solution compared with saline suspension. Microgram amount,s of the most active compounds are curative in the Ehrlich mouse ascites tumor. Struct,ural variation of the polycyclic aromatic radical has been related to antitumor activity. These relationships only partially correspond with t,hose when mustard grortps rather than the chloromethyl group furnish the alkylat,ing function. The previously noted high activity of chloromethyl aromatic hydrocarbon us. t,he mustard-resistant S-37 tumor has been st,udied in detail.
We have previously reported the discovery that antitumor activity is conferred on a monofunctional PI; on S half m ~ s t a r dand , ~ on a simpler alkylating function5 by the presence of a polynuclear moiety in the same molecule. Several simple chloromethyl aromatic hydrocarbons were among the most potent compounds. This discovery was surprising since these are hydrophobic, insoluble chemicals which were given as fine suspensions in saline to tumor-bearing mice. In an effort to determine whether greater in situ availability would affect the antitumor activity of such compounds as 1-111, they were injected intraperitoneally as solutions in sesame oil into mice bearing ascites tumors.*-j This mode of administration in fact
CH:, I
I1
I11
markedly increased both the activity and the toxicity of 1-111 compared with these properties when 1-111 were given in suspension. I n view of this enhancement
of potency, further structural variation of the aromatic group was studied (see Table I and section on Biological Results). I n addition, the previously noted efficacy of some of these compounds against the mustard-resistant S-37 tumor6 has been examined (see Table 11). To obtain the previously unreported compounds in Table I, direct chloromethylation was not attempted, since it had been found that the impurity from even a small amount of excess chloromethylation can give a false enhancement of a ~ t i v i t y . ~ Where possible, the aldehyde was the preferred intermediate, followed by (1) reduction either with LiBH4 or S a B H ? , and ( 2 ) action of dry HCl. Several of the required aldehydes are known, and formylation of 2,9-dimethylanthracene gave the 10-carboxyaldehyde. However, 1,g-dimethylanthracene gave an intractable tar. The only method found to obtain this and other hydroxymethylanthracenes bearing alkyl substituents in the outer rings was via the ICH, derivatives available from the anthraquinone.6 Reaction of these iodo compounds with moist ,4g,O gave variable yields of the HOCH, compound. Dry HC1 yielded the ClCH, compound in every case except the same 1,9-dimethyl derivative. In one case, 7 in Table I, PCl, in C6H6was employed.’ Table I11 lists the intermediate HOCH, compounds riot previously reported.
( 1 ) Supported b y Research Grants Ch-02975, C.l-06927, and FR-05539 f r o m t h e Kational Institutes of Health, U. S. Public Health Service and by a n appropriation from t h e Commona-ealth of Pennsyl\.ania. (2) R. 11.Peck, R . I
