Separation of electronic and hydrophobic effects for the papain hydrolysis of substituted N-benzoylglycine esters

43

Biochimica et Biophysica Acra. 1079( 1991143-52 ¢, 1991 Elsevier Science PublishersB.V. 01¢7-4838/9 /$03 St| ADONIS 01674838ql00251E

BBAPRO 339fi4

Separation of electronic and hydrophobic effects for the papain hydrolysis of substituted N-benzo~!glycine esters C e s a r M. C o m p a d r e ~, C o r w i n H a n s c h -', T e r i E. K l e i n 3, 3 o a n n a P e t r i d o u - F i s c h e r 2, C y n t h i a D i a s S e l a s s i e 2, R . N e l s o n S m i t h *, W a y n e S t e i n m e t z ", C h u n - Z h e n g Y a n g 2., a n d G u a n g - Z h o n g Y a n g * *'2 l Department ofBiopharmaceutical Sciem'es, Unicersity of Arkansas for Medical Sciences. Little Rock. AP, tU.S~A.L : Department of Chemisto', Pomona College, Claremont, CA (U.S.A. J and s Computer Graphics Laboratory. Department of Pharmace ~tical Chemistr)', Uni~'ersityof Cahfornia. San Francisco CA ¢U.S A )

IReceived 3 December 1990) (Re~,isedmanuscriptreceived 18 March 19ql)

Key words: Papain: OSAR: N-Benzo~lglveineester En~'mic hydrolysis:SAR: Glycinateester

The role of hydrophobic and electronic effects on the kinetic constants kcat and K m for the papain hydrolysis of a series of 22 substituted N-benzoylglycine pyridyl esters was investigated. The series studied comprises a wide variety of suhstituents on the N.henzoyl ring, with about a 300000.fold range in their hydrophobicities, and 2.1-fold range in their electronic Hammer constants (~r). It was found that the variation in the log k~t and log "1/K m constants could he explained by the following quantitative-structure activity relationships (QSAR): log ! / K , , = 0.40 ~4 + 4.40 and log l / k e B t = 0.45o" +0.18. The substituent constant, ~'4, is the hydruphobic parameter for the 4-N-benzoyl substituents. QSAR analysis of two smaller sets of Ifi~cme phenyl and methyl esters produced similar results. A clear separation of the substituent effects indicates Chat in the case of these particular esters, acylation appears to be the rate limiting catalytic step.

Introduction The development of hydrophobic substituent constants (,-r, log P) from octanol-water partition coefficients [1] has clearly spurred the application of physical organic chemislry to the study of enzyme structure-active relationships [2-6]. These advances have come from the realization that the interaction of a set of congeners with an enzyme necessitates the use of multivariate analysis to delineate electronic, hydrophobic and steric effects of substituents. Perforce this lead~ to more complex equations than have been ~raditionaily

* Deceased, December 23, 1983. * VisitingScientist at F';nona College from the Institute of ilemaIOIo~,, Chinese Academyof Medical ~'ience. Tianjin, China. ** Visiting Sc;".ntist at P~mona College from the In:,titutc of Materia Medica. Chinese Academ~,of Medical Science. B~ijing.China Correspondence: C. Hansch. Department ol Chemistry. Pomona College, Clatemont, (A 91711. U.S.A.

used in linear-free energy relationships, which may ~ta~ to concern about the validity of such expressions. With the rapid development of molecular graphics, an independent check on the correlations can be made in those instances where the X-ray crystalh ~raphic structure of the enzyme has been established. A number of studies now show that there is good agreement between the two methodologies [2,3]. Thus the structure-activity picture of the dynamic enzyme in action in solution obtained from kinetic constants is found to agree with the static view of the enzyme obtained from the X-ray ~'~alysis of the solid crystals. However, just how far one can vary structural changes and still correlate kinetic parameters (,such as K~, kc~t, K i) with the physicochemical properties of the structural changes, is far from clear. In trying to systematically extend the reach of OSAR iQuantitative Structure Activity Relationships) in enzymr, tic reactions, we have found papain to be an attractive target errs'me. Its crystallographic structure is established at a resolution of 1.65 A, its stability is good,

44 and its well delineated active site has enabled us to develop viablc QSAR by means of the following esters I and Ill [2]. X-C.,H.~Ot_'OCH 2 NHCC~C~H ~ I CH~OCOCHeNHCOC~H

4Y

X-C~H.,OCOCH zNHSO~CH II 3-X-pyridyI-OCOCH-NHCOC~H4-Y

Ill

IV C ~ H ~ O C O C H ~ N H C O C ~ H ~-Y V

Several studies of I and I1 on a variety of cysteine hydrolases have yielded consistent correlation equations so that the effect of moderate size X on the leaving groups in hydrolysis is relatively clear. Thus, one must utilize steric, electronic and hydrophobie properties of X as well as consider the two equivalent orientations of meta substituents to obtain satisfactory QSAR. A study of the hydrolysis of variations of 1I! yielded much simpler Eqns. 1 and 2 [7]. log 1 / K i n = 1.01or + 1.46

n ~ 16, r = 0 . 9 8 1 , x = 0 . 1 6 5

(l)

log k rat = - 0.50~" ~ - 0 . 6 9 r , + 0 . 5 4

n = 16. r = 0 . 9 5 5 . s = 0 . 3 5 7

(2)

In attempting to extend om study of analogs of I!I with a wider range of Y substituents, it was decided to use a leaving group which would be visible in the ultraviolet region to facilitate kinetic analysis of hydrolysis while at the same time maintaining reasonable water solubility so that more hydrophobie Y groups could be examined. Congeners IV were found to meet these criteria while at the same time maintaining the same leaving group geometry as that of I and 1I. Studies with I and 11 have shown that a nitrogen in the meta position of the pyridine ring would likely be held away from the enzyme surface in the surrounding aqueous phase. We have now established successful QSAR for 22 variations of IV.

Experimental procedures

few drops of triethylamine were added to the filtrate and the solution was stirred at room temperature for 6 h. Evaporation of the dichloromethane yielded a crude product which was recrystallized from the appropriate solvent. See Table !. In an alternative procedure a substituted N-benzoylglycine (0.1 M), dicyclohexylearbodiimide (0.12 M), 3-hydroxypyridine (0.1 M) and a few drops of triethylamine were stirred in dichloromethane; the reaction was monitored by thin-layer chromatography. Some of the reaction mixtures were refluxed. When the reaction appeared complete the resulting mixture was workedup in the same fashion as described above. Synthesis of substituted N-benzoylglycines. in a typical example a mixture of the appropriate acid chloride, an equivalent amount of glycine and two equivalents of sodium hydroxide (10%) were stirred at 0 ° C for 1 h and then at room temperature for 2 h then acidified with HCI. The crude product was recrystallized from an appropriate solvent (Table I). The acid chlorides that were not commercially available were prepared by refluxing the substituted benzoic acid with excess SOCI 2. N-(3-Aminobenzoyl)glycine. This compound was prepared by catalytic hydrogenation of N-(3-nitrobenzoyl) glycine in absolute alcohol using 5% palladium charcoal at 50 psi. N-(4-Dimethylaminobenzoyl)glycine. A solution of 4nitrobenzoic acid (0.01 M), aqueous formaldehyde (40%, 2 ml) in 95% ethanol (100 ml) and 5% palladium on charcoal (0.2 g) was hydrogenated on a Parr apparatus until absorption of hydrogen was complete. The catalyst and solvent were removed and the product was purified by recrystallization (Table 1). Melting points are uncorrected. Elemental analysis were performed by Galbraith Laboratories. Substituent constants. The hydrophobic parameters ~r are those from the benzamide system [7]. The electronic parameters cr were taken from the compilation of Han~h and Leo [!]. Kinetic measurements. Papain, 2 × crystallized, was obtained from United States Biochemical Corporation. The enzymatic hydrolysi, of the pyridyl and phenyl esters was carried out at 25°C and followed spectrophotometrically using our earlier reported procedure [8,9]. Hydrolysis of methyl esters was followed using a pH-stat as previously reported [10].

Synthesis of N-benzoylglycine pyridyl and phenyl ester. The pyridyl and phenyl esters were synthesized by a procedure similar to that previously reported [7]. In a typical experiment a substituted N-benzoylglycine (0.1 M) and dicyclohexylcarbodiimide (0.12 M)were stirred in diehloromethane at room temperature for 3-4 h and the resulting solid dicyclohexylurea was removed by filtration. 3-Hydroxypyridine (Aldrich Chemical) and a

Results When formulating QSAR for congeners in which substitution occurs at more than one position, we have found that the systematic study of gach position helps greatly in arriving at a solution. Accordingly, Eqn. 3

45 w a s d e r i v e d for p a r a s u b s t i t u e n t s of IV only. using the

In this e × p r e s s i o n , K,,, is t h e M i c h a e l i s c o n s t a n t . ,% is t h c h v d r o p h o b c c o n s t a n t [1], the figures in p a r e n t h e s i s

d a t a in T a b l e II1.

~re fl~r t h e c o n s t r u c t i o n o f t h e 9 5 % c o n f i d e n c e intervals, n r e p r e s e n t s t h e n u m b e r of d a t a p o i n t s u s e d , r is t h e c o r r e l a t i o n coefficient, s is t h e s t a n d a r d d e v i a t i o n a n d F is t h e statistic for the s i g n i f i c a n c e o f t h e r t e r m .

log I / K ~ = 0.40: ± (t.05) ra + 4 3 g{ y {~.llt} ) ,t = 13, r - 0.982. s = 0.125. Fl.i, }= 3(11)

13~

TABLE 1

Chemical properttes of the 3-pyridyl Y-hippurat¢* (IV) and pho~yl Y-&ppurates ~I0 Y

Formula

3-Py.ridyl Y-laippurates flV)

Recryslallization

Elemental Analyses

Solvent

Calcd C

H

C

II

65.61 66.65 69.21

4.72 5.22 6.45

65.29 ~.').29 69.43

4.86

2 4-CH 3 3 4-CICH 3)3

Ct4H t2N20:~ CtsHI4N,O~ CI~H 2¢1N203

4 4-C,~tl.~

('2t~tt ~. N,O-~

72.28

4.84

72 37

5.11~

5 4-SOzNH 2

CutIt~N~O~S

5014

3.9(I

49.99

4 IV,

6 7 8 9

Ctt, H 15N304 ClaH bNxO 3 C~,H uN~O ~

61 33 61.98 64.05 55.81

4.82 4.83 3.94 3.68

61.08 62.18 64.22 56.~

529 5.03 4.16 3.82

l0 4-N(CH ~12

C~,H :N30~

64.20

5.72

64.44

6.tt2

l1 4-CI 12 4-I

C~aH uClN,O 3 CI~H ulN20~

578~ 44.00

3.~,2 2.90

57.85 44.10

3~88 2.86

13 14 15 16 17 18 19 20

CtsH2,N2Oa C2,,H ,aN20 , C~2H28N204 CtsltttlN~O3 Ct~H HIN:O ~ CIsHlaN204 C~sHI4N,O ~ Ct4H uN30 5

05.84 ~7.40 ~.73 ¢'A.05 44.@1 t52.93 66.65 ~,5.8t

6.14 6.79 7,34 394 2.9~1 4.93 522 3~8

65.37 67.18 68.79 64.03 44.30 63.1~ 66 ~', 55.09

6.10 6.91 7.3[I 3.87 2.56 5.17 54(t 375

21 3-NH: 22 3,5-CI z 23 3,5-(NOz),

CI4|It3N30~ CI4ttmCI2N:O~ CI~HtoN~O ~

61.98 51 7I 4~.56

4.83 31tl 2.91

~. 11 ;185 ~52

494 3.21 305

24 3,5-(C11~12 25 4-O{CH2)3CH ~, 6-CH 3-pyridyl

CIoH~6N20~ CI,~tt2:N20.~

67.59 66.65

5.67 6.48

67.97 66,71)

6.fRI 6 52

be~c,;e benzene benzenehcxar : ethytacetate melhylenechloridc acetonemethylenechloride elhylacetate acetone ethylacetate methylenechloride acetoneethylacetat¢ benzene benzeneethylacctatc benzene ethylacelatc ethylacetalc t:thqacetate eth_',lacetale ethylacetale ethylacetaie methylenechloride benzene eth~lacetate meth',lene chloride ethylacctate benzene

I 4-Br 2 4-NO~ 3 4-NH 2

CtsHl:BrNO ~ CIsH L,N:O , Ctstl t.~N20 ~

53'41 60 !b.1 66,t~

3.62 4.03 5.22

54.1~, 60.13 66.82

3 8.6 401 5AO

benzene benzene acetone-

4 4-C|1 ~

C I. H l~NO3

71.36

5 4-CI 6 4-F

CtsHIzCINO~ CI~ H L,FNO ~

~2.19 65.91

5.61 4.18 4.a0

71,50 01.92 O6.?.0

5.89 4.32 4.33

benzene ~nzene eth~l acela:e

1 H

4-NHCOCH, 4-NH 2 4-CN 4-NO 2

4-CRCHa)~CH ~ 4-O(CH 2)oCH 3 4-O(CH 2)sCH ~ 3-CN 3-1 3-OCH-~ 3-Ctt ~ 3-NO,

C~sHuN~O ~

mp ( ° C~

Yield (%)

Obsd.

5.62 6.13

g', t~l ~2-83 72-73

64 72 71

171 - 173

48

Ibis- tt~

t7

169- t70 154-156 161-162 182-184

31 7(.~ 62 .50

16,4-16(,

~8

132- t33 153 155

65 In.

I I~ 121i I19-121i 119-121 1[.I- I i5 125-126 N~-8I 88-h9 IM-t36

61 ':~d 18 57 67 42 65 ,15

142 114 15{,+157 1:52 I~,1

32 31 M

t(t3-105 1 ~ - 147

.n~ 60

Phenyl ¥-hippurates iV) 169-1711 13'7-1~ 159-1Ni

hexane

INt-121 159M01 150-151

46 TABL.E II

3~'~ lr(zlcopic ~md kmem parameterv related Io the papaipl catah'zcd h1"droh ~i.~ of mter.~ IV and V Y

a

.le

•

kurd," !(I 3

Kin. 10 s

1"~d,'10 ~

k~.~ (rain I)

730 2.211 3.44 5,08 2.38 2.03 4.76 0.33 3.32 2.91 2,82 1.48 11.(17 13.00 0.99 5.02 3.47 1.73 1.61 1.18 11.93 2.24 1t.66 1,38

2.711 4.86 2.16 2,53 3.11tl 4.15 1.55 8,34 4.45 1.81 2.37 1.32 1.(14 3.1t6 1.311 3.79 5.25 9.t15 2.67 1.98 1.48 1t.96 11.74 1.115

1.94 3,50 1.55 1.82 2.16 2.99 1.12 6JK) 3.20 1.30 1,71 0.95 0.75 2.20 (I.04 2.73 3.78 t~.5i 1.92 1.42 1.1)6 11.69 11.53 11.76

(rain-i)

kuncaI- 10 -4 (rain - ~ I

0.11) 1.29 0.50 11.66 11.62 1.411 0.60 1.01

4.811 8.711 201 6.411 4.60 2.50 1.211 3.70

(rain II 3-1~'ridyl esters (I19 3-NIl 2 3-NO, 3-C113 3-OCH ~ 3-1 3-CN 3,5-{CH3)2 3,5-(NO 21, 3,5-CI 2 4-H 4-CH 3 4-C(C113)~ 4-C6 H ~ 4 4 0 2 NH 2 4-NHCOCIt 3 4-CN 4-NO, 4-N(CH ~)2 4-CI 4-1 4-O((H 2).~CH ~ 4-O(CH 215CH 3 4-OtCH 2)7f i t 3 4-O(CH 2t3CH ~. 6-CH 3

271.3 313.11 313.0 315.tl 313.0 313.11 313,5 313.0 313,5 313,5 313,0 313.1/ 312,0 313.5 313.0 313.5 310.[) 278.5 313.5 314.0 313.t) 313.0 313.0 320

21132 2 629 2 551 2112 2528 25211 2211) 261)4 2477 1929 2347 2263 1950 2 498 2442 2355 3218 3 818 2445 2 396 2268 2025 1546 2534

1761t 787 19 523 35 106 425 597 Ig 21]0 145 12q 64 77 38 18 2056 91~19 17 14 13 43 35 15

221 2.87 2.0 2.64 2.40 2.76 2.1f~ 3.49 2.99 2.1(J 1.33 2.118 I).98 3.27 2.92 3.33 4.44 4.59 2.47 1,77 2,81 2.87 1.80 26.811

Phenyl esters (10

4-N H 2 4-NO 2 tt 4-F 4-CH 3 4-C1 4-Br 4-CIt 2CI13

261 294 270 26q 269 269 268 269

2175 1190 1598 t 739 1686 1237 1653 937

236 5116 754 442 1 387 11011 2529 1 a00

21.41 8.04 9,11 42 7 7.47 2.63 3,54 1.69

1.56 2028 7.00 9.13 9.69 1932 953 14.111

Two derivatives, 4-phenyl and 4-acetylamino, were omitted in deriving this QSAR. Reasons for their enhanced aberrant activity are not clear even with the help of graphics analysis. Inspection of the meta derivatives in Table I11, indicates that meta congeners show little variatioa in activity. In merging all of the data points of Table Ill, a role for meta substituents cannot be discerned as indicated in Eqn. 4 which includes all meta and para substituents except 4-acetylamino and 4-phenyl derivatives.

Although the meta substituted congeners arc not quite as well fit as the para, the parameters of Eqn. 4 are essentially the same as those of Eqn. 3. Adding terms in MR (molar refractivity) for steric effects or ~r for dectronic effects did not improve the correlation. Eqn. 5 correlates the catalytic step in the hydrolysis of the :,:~me set of congeners.

log 1 / K m - 0 . 4 0 ( ± 0.06)~- 4 + 4.40( _+ 0 . 0 9 )

What is most interesting is that variation in k~:, depends only on the electronic properties of the substitucnts and K,, variation depends only on their hy-

n = 22, r = 0.946, s = 0,176, Fi,z() = 171

(4)

log k ,,t = 0.45 ( + {I.()8)o" + 0,17( + 0.04 ) n - 23. r = 0.933, s - 0.004, Fi.zj = 141

t5l

47 TABLE 111 Parameter.s used to der# e Eqns 3 - 6 ~br the hydrolysis of 3-)(-pyrulyl Y-kqrpurates ll"

x

Obsd. log

Calcd' log

ILK,,

IlK m

1H 2 4-CH 3 3 4-C(CH3) 3 4 -C~.H5 ~" 5 4-SO:NH 2 6 4-NItCOCH3 f 7 4-NH, 8 4-CN 9 4.NO, l0 4-N(CH 3)2 I l 4-CI 12 4-1 13 4-OICH 2)3CH3 14 4-O(CH:)gCH ~ 15 4-O(CH 217CH3 16 3-CN 17 3.1 18 3-OCH ~ 19 3-CtI~ 211 3-NO, 21 3-NH 2 22 35 (CI)2 23 3 , 5 (NO2) 2 24 3.5 -(CH3) ", 25 4-O(CH,)3CH ~, 6,CH 3 g

4.54 4.55 4.88 6.14 3.89 5.07 3.88 4.30 4.46 4.7fi 4.79 4.93 5.05 5.65 6.18 4,69 4,62 4.29 4.47 4.66 4.14 4.48 4.03 4.32 4.86

4.40 4.62 5.14 5.19 3.83 4.15 3.99 4.34 4.47 4.60 4.76 4.94 5.13 5.611 6.07 4.4(I 4.4(I 4.411 4.40 4.4(I 4.40 4.411 4.40 4.40 5.14

.1 1/ Km 0.14 -11.07 -0.26 0.95 0.0fi 11.92 -0.11 -0.04 -0,01 0.16 0.03 -0.01 --0,08 0.05 0.11 0.20 0.22 -0.11 007 0.26 -0.26 11,0~ -11.37 -0.1)8 -[).28

Obsd log

Calcd. h h)g

kc:,t

k~, t

0.11 [),23 -0.02 --(1.12 11.34 -0.03 -0.15 0.44 0.58 - 0.19 0.28 0.28 0.15 I).03 -0.16 0.48 tl.33 11.26 [119 11.54 tl.29 0.51 0.78 0.05 -I).12

0,17 0.09 1/,08 0.17 (I.43 0.17 -0.13 0.47 G.52 - 0.21 0,27 0.25 0.03 0.03 0.03 0.4" 0.33 0.22 0.14 0.49 0.10 1!.51 11.81 0.11 0.03

Ob~,J ' )g

Calcd.' log

k~t/K m

k¢~,t/'K ~

4.65 4.78 4,81 6.02 4.23 5.t14 3.73 .+.74 5,[14 4.57 5,07 5,21 5.2{) 5.68 6.02 5,17 405 4,55 466 5.211 4,43 4.99 4.81 4.37 4.74

4.~0 4.73 5.21 5.34 4.3[) 4.36 3.92 4.82 5.011 4.43 5.0,1 5.18 5.15 5.6[1 ~,.t~4.83 4.74 4,65 4.57 4.90 4.53 4.91 5.2(} ~.54 5.15

~r

:'r d 4

0.(R) - 0.17 -0.20 -0.0l 0.57 0.00 -0.66 0,66 0.78 - 0.83 0.23 0.18 -0.32 0,32 -0.32 0,56 0.35 0.12 -0.07 0.71 -0.16 0,74 1.12 -0,14 -032

0.O() 0.54 1.87 1.98 - 1.43 -0.63 1.05 -[I.16 11.18 1).50 0.ql 1.35 1.84 3.02 4.20 /I.00 0,(g) 0.00 0.00 0.00 0.00 0.011 0.011 0,00 1,84

Calculated ufing Eqn, 3. h Calculated using Eqn. 4. c Calculated using Eqn. 6. a From Ref. 7. ~' Not used in the derivation of Eqns. 3-6. Not used in the derivation of Eqns. 3, 4 and 6. Not used in the derivation of Eqns. 3-6. "lhc activity is similar to (hal of the un~ubsfituted compound [N~). 131. h scem~ lha, the 6 methyl on the pyridine ring does not contact the active site.

d r o p h o b i c c h a r a c t e r . Eqn. 5 shows t h a t the substituents are n o t in significant steric o r h y d r o p h o b i c c o , t a c t with t h e e n z y m e in the a c y l a t i o n a n d deacylation steps. T h e 4 - a c e t y l a m i n o derivative is i n c l u d e d in E q n . 5, but the 4 - p h e a y l derivative is not. T h u s the 4 - a c e t y l a m i n o g r o u p which binds 10-times m o r e tightly in the ES c o m p l e x b e h a v e s n o r m a l l y in the catalytic step a n d , t h e r e f o r e , must be out of c o n t a c t with the enzyme. This c o u l d be a t t r i b u t e d to a c o n f o r m a t i o n a l c h a n g e in the e n z y m e w h i c h releases the s u b s t i t u e n t f r o m those s t r o n g b i n d i n g forces t h a t m a i n t a i n e d it in the ES complex. T h e 4-phenyl c o n g e n e r is mispredieted by 3.5-times t h e s t a n d a r d deviation o f Eqn. 5. Its d i m i n i s h e d activity m a y arise f r o m the u n u s u a l l y s t r o n g h y d r o p h o b i c b i n d i n g in the ES complex f r o m w h i c h it m u s t pull a w a y like the o t h e r s u b s t i t u e n t s in the catalytic process. Its failure to a c c o m p l i s h this m a n o e u v r e , while the c v c n m o r e h y d r o p h o b i c O ( C H 2 ) v C H ~ succeeds, is an i n t r i g u i n g a n o m a l y . T h e fact t h a t it b i n d s

10-fimes m o r e effectively in the ES complex t h a n exp e c t e d from Eqn. 4 suggests a n unusually effective fit of ',he bulky a n d rigid phenyl moiety to the active site w h i c h does not o c c u r with the o t h e r substituents. T h e results from Eqns. 4 a n d 5 arc s h o w n in Fig. 1. C o r r e l a t i o n with the specificity c o n s t a n t ( k , , f f K m) is o u t l i n e d in Eqn. 6. log k,d ~ / K m ~ 0.38( ± 0.(),";)~4 ~(I.431 ± 0.18)o" + 4,60(+ 0.10) n = 22, r = (1.926, s = 0,193

16)

Eqn. 6 is w h a t o n e would expect f r o m the s u m m a t i o n o f Eqns. 4 a n d 5. T h u s we see a consistent p a t t e r n t h r o u g h o u t the r e a c t i o n series, for a set of esters having quite a diverse set o f substituents, in t e r m s of h y d r o p h o b i c a n d electronic c h a r a c t e r i s t i c s as well as different types o f h e t e r o a t o m s a n d g e o m e t r y . Substituents differing in h y d r o p h o b i c i t y from 4 - N H ~ (It = - 1.05l to 4 - O ( C H . , ) T C H a(~r = 4.201, a r a n g e o f 5.25 in

48

i

,

,

;

-

;

-°'~-L

'

Observed

oio

d~

,.o

Observed

Fig. 1. ('omparim)n of experimental and calculated va;ues tor log I / K = and log k++~ from Eqns 4 and 5

log units, are very weli correlated. Although the coeffi cient in Eqn. 3 is not large, this range of almost 300000 in hydrophobicity clearly establishes the nature of the hydrophobic contact which molecular graphics confirms. The range in o" is also much greater than that usually employed in biochemical QSAR (~r4NH,= -0.66; ~3,5-di(NO2)= 1.42). Thus we feel that the conclusions drawn from Eqns. 3-6 are well supported. However, the comparison of Eqns. 4 and 5 with 1 and 2 was shocking. Eqn. 1 and espec!ally Eqn. 2 would appear to have come from a completely different enzyme! Our first inclination was to presume that somehow the pyridyl moiety of IV was the root of the problem. To test this possibility, 8 variations of V were prepared and tested and from the data in Table IV, Eqn. 7 to 9 were developed log l / K n , = 0.44( ± 0 19)'n" + 4 . 0 8 ( ± 0 1 5 ) a ~ 8, r = t},Olg, ~ ~, 11,150, FI+ , = 32.7

(7)

log k+,,t = t}.6' ( + 1).54)¢r - 0 . 2 3 ( ± 0 2 1 )

log k~+t / K , , , - 0.79( + 0.36)~" + 0.35( +0.04)o" + 3.72(-t-0.26) n = 8. r = {I.944, s = 0.243

tg)

Again we find complete disagreement with Eqns. 1 and 2, but reasonable agreement with Eqns. 4 to 6. The confidence limits on the parameters of Eqn. 7 are narrow, but those on Eqns. 8 and 9 are rather wide. The correlation of Eqn. 8 is poor, although it is statistically significant and cannot be improved by adding hydrophobic or steric terms. The message from Eqn. 8 is the same as from Eqn. 5: electronic factors are of paramount importance in the catalytic step. The ¢ term in Eqn. 9 is not justified by the F test. The poor solubility of derivatives of V precluded the testing of a wide range of compounds and thus the range in ovalues is limited. At this point it was decidea to retest a number of variations of 111 which were still available. From the data in Table V, QSAR 10 to 12 have been derived. tog I / K m = 0.31( _+0.13).:7" + 1 . 6 4 ( + 0 , | 0 )

n = 8, r = 07~7, .s - 0,241. F t ~ = 9 7 5

[8)

n = 7. • =

0.938..*

+ 0,(~,

/7 t 5 = 3 6 . 6

(10)

T A B L E IV Pammewr~ 1~+'4+to der~ e Eq,o 7-r# [or the hydndy~t, o] # l e n f i . 4 . Y-hlppurate~ V

Y

1 2 3 4 5 6 7 8

4-NH, 4-NO 2 H 4-F 4-CH~ 4.O 4-Br 4-C2H ~

Obsd. log I/ K =

Calcd ~ log I/ K m

Obsd. log k~

Catcd. ~ h)g k ~,

Obsd. log k c+I / Kr~

Calcd ~ log k ~I / K m

o"

3.67 4.06 4.D4 4.37 4.13 4.58 4.45 4,77

3.62 4,16 4.08 4.20 4.32 4.4q 4.58 4.60

- 1.00 0+11 -030 -0,18 (I.21 0.15 .- 0.22 0,00

-0.69 0.30 -0.23 -019 - O35 - 0.08 - 0 0¢~ -034

2.67 4.17 3.74 4A9 3 ~2 4 43 423 4.76

2.76 4.13 3.72 33)4 4.04 4.44 4.59 4.49

-0.56 0.78 0,00 0.06 -0.17 0.23 0.Z~t -0,15

+ Calculated via Eqn. 7. Calculated via Eqn, ~. " Calculated Ha Eqn. 9,

- 1.05 0.18 0.00 0,27 0.54 0.91 1.12 1.17

49 TABLE V Parameters u~ed to dem e Eqns. 10- 12 [or the h~drol~,sls of meth)l htppurate~ HI pit 6.0, 25 ° C. with 10<; ('tt~CN

Y

1 2 3 4 5 6 7

Obsd It,8 I/K,. 1.40 1.58 1.70 1.71 1.87 2.21 1.70

4-NH2 H 4-OCH 3 4-EH~ 4-CI 4-I 4-NO~

Calcd "

Ohsd. h )g k~,

log

I / K~ 1.32 1.64 1.71 1.81 1.93 2.116 1.70

Calcd. ~ h~g I~, 0.07 0.45 0.30 11.35 0.58 0.55 0.89

0.02 0.53 0.32 0.52 0.59 0,28 0.05

Obsd. log

(alcd. '

k~., / K m

k j t/K m

1.42 2 11 2.02 2 23 2.46 2.49 2.65

1.47 2 12 2.111 2.15 2.48 2.55 2.64

~r

r:

-0.66 0 (M) -0.27 -0.17 I).23 0.18 0.78

- 1.05 0.00 0.22 0.54 0.91 1.35 0.18

h ,g

a Calculated using Eqn. I0. b Calculated using Eqn. 11. c Calculated using Eqn. 12. log k~, = 0.57( _+0.351~ + 0.45( ± 11.151 n = 7, r = 0.880, s =0.152, F1.s = 19

(I I)

log k,~, ~Kin = 0.24( _+0.10)'n, +0.61( +0.17),;' +2.12(±0.17) n = 7, r = 0.993, s = 0.059

( 121

Again we find Eqns. 10-12 in a g r e e m e n t with their c o u n t e r p a r t s obtained from the pyridyl and phenyl hippurates, i.e., the formation o f the ES complex strongly d e p e n d s on the hydrophobic character of Y while the catalytic step d e p e n d s only on the electronic c h a r a c t e r of Y. F u r t h e r testing of Eqn. 1 is possible with data obtained by Storer and Carey [i !] in T a b l e VI on methyl h i p p u r a t e s I l l which y;~lded Eqns. 13 and 14. log I / K m = 0.25(±0,77)*r + 1.431±038)

n = 5, r = 0.514, s = 0.175. F t ~ = 1.09

(13)

log k¢~,,= 0.67( + 0.291e + 0.55( + 0. I 1) n = 5, r : 0.974, s : 0.076. F 13 = 55.6

( 141

TOO few data points are available to deriv~ a two variable equation for log k ~ , l / K m. TABLE VI Parameters used to deri~ e Eqn. 13 and 14 for the hydrot~t~ o( meth)l hippurate* Ill at pH 65. 20~C and 20c..'~ CH~CN Data [rcv'n Stor*r and Care~ /71

Y

Ot~d. k~ I/Km

lH 1.63 2 4-OCH ~ 1.40 3 4-CH~ 1.60 4 4-CI 1.69 5 4-NO 2 lY0

Calcd" log I/Kr,

Obsd. tog kr~ ~

Calcd. ~' # log k~

1.43 1.49 1.57 166 1.41~

0.54 0.31 0.46 081 10.t

0.55 0.37 0r~ 070 1.07

Calculated using Eqn. 13. i, Calculated using Eqn. 14.

*

0.Ofl - 027 -0.17 023

0.(~ 0.~ 0.54 091

0.7~;

0 IX

Although Eqn. 13 is similar to the other log 1 / K , , equations, it is r o t statistically significant. The reason for this is probably the narrow range of rr values for the small n u m b e r of substituents studied. For the data in Table V, the -ange in rr is 2.40 while in Table VI, the range is only 0.91. Eqn. 14 is an excellent correlation confirming that only the electronic effect of Y is significant in the kca t step. Clearly the correlations of Eqns. 1 and 2 are spuriotis, but the reasons for this a n o m a l o u s behavior are not at all apparent. From the d a t a now available to us, it is not possible to d e t e r m i n e just where the d i s c r e p a n t ; lies. A s n o t e d above, it is hard to believe that the work was done with papain; perhaps some variation of papain, such as chymopapain, was mistakenly employed. Shorter and Carey noted that their kinetic p a r a m e t e r s for the methyl glycinates did not agree with o u r values on which Eqns. 1 and 2 were based. It was suggested that this might be due to the fact that they used 20% acetonitrilc and buffer as the solvent while we had used only 10% acetonitrile. However, the data in Table !11 is in reasonable agreemenl with their results even though 10% acetonitrile was employed in this case. The higher quality' correlations with E,qn. 4 to 6 stem from the wider range in substituent constants of the various conger, zrs which provides greater variance. Also. congeners IV are easy to work with because of their increased water solubility. A disadvantage of these esters is that over time they tend to hydfotyze unLm,s tho" are kept away' from moist air; the I~'ridox3' group is a very good leaving group.

Discussion

From our earlier graphics analyses, we er4>¢ct and, in fact, find. a clear d e p e n d e n c e of binding of 4-substituents on their hydrophobic character. A t t e m p t s to imprtr,,e Eqn. 4 using steric and electronic t e r t m were

50 unsucce,,~ful. There seems to be no ~attcrn for the compounds which have deviations above the standard deviation. While there are no points which are misfit by more than two-times the standard deviation, except 4-phenyl and 4-acetylamino, the 3,5-dinitro, 3-cyano, 3-nitro and 3-amino are :dl somewhat larger than the standard deviation. We assume that this is probably duc to unaccountable small steric and dipolar interactions. The 4-t-butyl congener is less active than expected ~nd this seems reasonable because of its bulk. A graphic analysis reveals that the -NHCOC~,H 4moiety common to congeners I, 111, IV and V falls into a narrow hydrophobic cleft where it would have little freedom to rotate [9]. At about the point where Y substituents would fall, the cleft widens out into a large, flat, mostly hydrophobic surface {Fig. 2). This surface can easily accomodate the largest substitucnts in Table III, [~-O(CH2)TCH 3] and, in fact, could take carbon chains at least one carbon unit longer. We find that as we have noted before [2] the coefficient of 0.4 with the most reliable of the QSAR is similar to the value of 0.5 which we have observed for several exam-

t:,.,'~ ¢~.~.. • ~:~.~:..

t~." ~.?;':'~..:... :.~:'*."

~i:::~;~........,.:....~..~

':ii~!~:~;~~. ,-!,.~!-: '~:~!, ~'~""~.~x ':~,

,.~

~,~,'-'-~ ~.~

,~ ~.~,:.,~

~"~'~ '4~-~-:' :" ~.,,

~ .~

.

Fig, 2. The black dots represent solvent accessible hydrophohic surface and the white open spaces ~epresent polar space (o~'gen or nitrogen), The ligand is C~,H~OCOCH_~NHCOC,,It~-4O~C]| 2)-~CH ~ positioning by the construction of the basic mtxtel has been pre~,iously discussed [O].

pies of ligand substituents binding Io a more or less flat hydrophobic surface [2]. It is clear from the graphics that there is little space b c ~ e e n the phenyl ring and the enzyme :,urface for substituc:lts to fit well and the phenyl ring fits the cleft so tightly thai it cannot relieve the strain by twisting. The only possibility for binding of congeners with a meta substituent is for the phenyl ring to orient so that the substituent is held away from the enzyme in aqueous space. The congeners of special interest are the 3,5-dichloro and the 3,5-dinitro. In these examples one meta substituent could orient away from the surface, but one could not. The smaller, hydrophobic ehloro analogue is well fit by Eqn. 4, but the larger hydrophilic nitro is not. Since no parameterization by ~ is made for either chloro, it appears that a small steric effect is counteracted by a hydrophobic interaction so that this congener is well fit. In the case of the di-nitro derivative there is no compensating hydrophobic interaction to offset the steric effect, so that this congener is not well fit. Graphics analysis shows that the 4-phenyl substituent of 1V can bind in the area of a hydrophohic wall formed by Val-132 and Val-133. This would result in desolvation of one side of the phenyl ring corresponding to one side of the long chains being desolrated on the flat surface. This does not explain the extra tight binding of the 4-phenyl congener. On the large rather flat hydrophobic area, X-ray crystallography was established the presence of two bound water molecules. 29 and 40, which seem to be anchored via Ser-131. Although these are too distant from thc 4-acetylamino group for direct hydrogen bonding, it is possible that another water molecule might serve as a bridge to form a complex which could anchor this congener more firmly in the ES complex than Eqn. 4 predicts. Such a complex might be broken up in the catalytic step so that the 4-NHCOCH 3 congener would be well fit by Eqn. 5. In Fig. 1, the black dots represent solvent accessible hydrophobic (carbon) surface as defined by moving a molecular probe water molecule over the solvent accessible enzyme surface [2]. The white open spaces are polar regions (oxygen or nitrogen). The 'wire' model of the ligand is that for the 4-O(CH2)TCH 3 of IV. Clearly there is sufficient hydrophobic surface for the long alkoxy group to be desolvated on one side. The graphics justifies the hydrophobic term in Eqns. 3, 7, 10 and 13. ]'he analyses of the four data sets in terms of log 1~Kin, log kejt and log kc~t/K m show that the present convention of doing mechanistic studies of enzymic reactions solely in terms of the specificity constant, k~,/K~, can overlook valuable insight into reaction mechanisms. Consideration of only Eqns. 6, 9,

51 or 12, would not have elucidated the fact that ES formation was dependent only on hydrophobic factors and that catalysis was dominated by electronic factors. It is currently accepted that ester hydrolysis by papain occurs by a mechanism comprising at least three steps: binding ( K~ = k _ n/k ~), acylation (k_,) and deacylation (k 3) defined as follows: l, k, k~ E+S~- E S ~ E S ' ~ E + P , Pi Where ES is the Michaelis complex, ES' is the acyl-enzyme, P~ is the alcohol moiety of the ester and P: is the acid fraction of the ester. Since K m and k ~ are related to the rate constants as follows: kca t = k ~k 3 / k : + k 3

and K m = k 3 ( k_ 1 + k , ) / k l ( k ," + k3) then if deacylation is the rate-determining step, k~,, = k.~andifk j~k,,then Km=k i/klkz=K,k~/k2. If acylation is the rate-determining step, kc, ~ = k, and if k _ u >> kz, then K m = k _ i / k l = K~. The sharp correlations of Eqns. 4 and 5 based on appropriate size data sets suggest that the kca t and K m steps can be separated and that acylation may possibly be the rate-limiting step. if deacylation were the rate-limiting step, then both k~t(= k3) and Km ( = K ~ k 3 / k : ) would be functions of k~ and it would not be possible to formulate independent correlations for both parameters. However, more experimental work is required before this mechanism can be justified. Another point supporting the lack of importance of the dea~lation step is that there is no ~r term in Eqn. 5 correlating k¢~t. If deacylation were rate limiting, a negative term in rr would be expected in this equation for the desorption of the acyl moiety from the hydrophobic surface. Such is the case for the emulsin hydrolysis of phenylglucosides [121. It is surprising that such a term does not appear. It c a n be argued that in the acylation step there is a positive dependence on ~.~ which is balanced by a negative dependence on 7r4 of the deacylation step. Why are electronic effects absent in the formation of ES, but important in the kca t step? This becomes clear when one considers ,o for log 1 / K ~ for the papain hydrolysis of congeners ! and il [2]. A mean value of p of 0.56 shows that electronic effects are small. In these examples, X is separated from the earbonyl carbon by a -C~H~O- moiety while in the ease of congeners IV, Y is separated by -C,H~CONHCH z-. Simply placing a -CH,- unit between the phenyl group and the reaction center has a strong effect on p. For example, the alkaline hydrolysis of methyl benzoates in 33% dioxane at 24.7~C has a P of 1.93 [13] and the

alkaline hydrolysis of t-butyl peroxybenzoates in 50~;~ethanol at 25 °C has a p of 2.18 [14], while the alkaline l~ydrolysis of ethyl phcnylacetates in 56'~,~ acetone at 25 ° has a p of 0.97 [15]. It should be noted that p for ester hydrolysis is not highly sensitive to solvent or even structural features of the parent congener. For example, O = 2.25 for the ~kaline hydrolysis of methyl benzoates in 85% methanol at 25 °C [16]. The hydrolysis o( methyl 4,5,-X-pyridine-2-carbox3'lates in 85~ methanol has a p of 2.1)5 [17]. The insertion of a CH, in the case of the phenylacetates has halved the p, compared to the benzoate esters, and from this alone we would expect a p of only 0.23 for congeners IV in the formation of ES. But, in addition to the CH 2, IV contains a - C O N H - moiety compared to a -O- for congeners I and I1. Hence a p of 0.1 to 0.2 might be expected for the correlation of log 1 / K m. The signal to noise ratio probably preve,lts detection of such a small effect. In reality one should not expect a large value of p in the fi~rmation of ES which is primarily an absorption process and indeed we have found p for log 1~Kin for a variety of hydrolases acting on two different types of esters to be about 0.6 [19]. It is with the chemical processes of acylation and deacylation that one would expect the electronic effect of substituents to be relatively large and indeed this is what we find. Attempts by different individuals in our laboratory to formulate a Hammett equation for the rate of buffer hydrolysis of IV were unsuccessful. No meaningful equation could be established. It is of interest that in this connection, Brown and Newsson could not obtain a satisfactory Hammett equation for the hydrolysis of a set of phenoxyacetic acid esters [20] despite the fact that we find p for the ionization of a set of phenoxyacetic [191 acids to be 0.30 i_+0.05). The hydrolysis of acids is normally less sensitive to substituent effects than the hydrolysis of the corresponding esters. Thus it may be that heteroatoms in the side chain assist hydrolysis anchimerically complicating the mechanism in a way not possible in the enzymatic hydrolysis. In summary, analysis of the four data sets in terms of log 1 / K m and log kca t brings to light aspects of the reaction mechanism in the examples of papain, emulsion and chymotrypsin [21] as well as other enzymes [2] which would have been missed had only log kcat/K m been considered. Bender and K~zdy's highly influential remark [221 that only k,:at/K m or k 3 are suitable parameters for mechanistic analysis has caused an ~ e r r e action Y,',hing is lost by examining log l / K m and log kc~, separately, and possibly, a great deal cen be gained in certain instances. No doubt there will be instances where K~ and k~, cannot be treated separately, but we feel this should not be an automatic a~umption. However, developing biological QSAR is vastly more

52

difficult than working with simple organic reactions in homogeneous solution. In undertaking biological QSAR, it is important to study a single position at a time on a parent congener by making a set of derivatives for which the structural changes are reasonably noncollinear with respect to steric, electronic and hydrophobic parameters. In the case of phenyl rings it it clear from the present study as well as others [23], that meta substituents, or other structures with similar elements of symmetry, are apt to present special problems. Considerable thought is necessary in parameterizing such congeners. It is apparent that deriving a biological QSAR is not a routine problem; each case seems to present special problems. Just how far QSAR analysis can be carried is extremely difficult to say at this early stage of development. Actually, it depends to a high degree on what kind of an answer one is willing to accept. To obtain an answer in which the standard deviation of each data point is within the limits of experimental error for a complex receptor reacting with a complex set of chemicals may be forever beyond our reach, at least by using only a small number of easily obtained descriptors. However, only a few years ago we would not have imagined it possible to obtain the results found in this report. In solving such complex problems, we must proceed by taking one small step after another. The results in this report taken with others [2-4] continue to bring out the importance of hydrophobic parameters (~, log P) and multiple regression analysis in opening up biochemical reactions to interpretation by the methods of physical organic chemistry. References I Hansch. C. and Leo, A. (1979)'Suhstitnted Constants Ior Correlation Anal:'sis in Chemistry and Biology'. Wiley-lnterscience, New York.

2 Hansch, C. and Klein. T. (1986) Ace. Chem. Res. 19. 392-400. 3 Blaney. J.M., Hansch, C., Silipo, C. and Vittoria. A. (1984) Chem. Rev. 84, 333-407. 4 Gupta, S.P. (1987) Chem. Rev. 87, 1183 t253. 5 Sakurai, A.L., Margolin, A.J., Russell, A.J. and Klibanov. AM. (19~8) J. Am, Chem. Soc. 110, 7236-7239. 6 Matsumura, M, Becktel, W.J. and Manhews. B.W. (1988) Nature 334, 4116-410. 7 Hansch, C., Smith, R.N., Rockoff, A., Calel. D.F,, Jow, P.Y.C, and Fukunaga, Y. (1977) AJL.h. Bi,~hem. Biophys. 183, 383-392. 8 Carotti. A., Smith, R.N., Won8, S., Hansch, C.. Blaney, JM. and Langridge, R. (1984) Arch. Biochem. Biophys. 229, 112-125. 9 Smith, RN., Han~h, C., Kim. KH.. Omiya, B., Fukumara. G., Selassie, C.D,, Jow, P.Y.C., Blaney. J,M. and Langridge, R. (1982) Arch. Biochera. BiolJhys. 215, 319-328. 10 Compadre, C.M.. Hanseh, C., Klein, T.E. and Langridge, R. (1990) Biochem. Biophys. Acta 1038, 158-163, 11 Storer. A.C. aad Carey. P.R. (1985) Biochemistry. 24. 6808-6818. 12 Hansch, C., Deutsch, E.W., Smith, R.N. (1965) J. Am. Chem. Soc. 87, 2738-2742. 13 Bender. M.L. and Thomas, R.J. (1%1) J. Am. Chem. Soc. 83, 4189-4193. 14 Antonoviskii, V.L., Forlova, Z.S and Romantsova (1969) Zh. Org. Khim. 5. 42-44EE. 15 Norman, R.O.C. and Ralph, P.D. 11963) J. Chem. Soc., 54315436. 16 Campbell. AD.. Chooi. S.Y., Deady, LW. and Shanks. R.A. (197(I) Aust. J. Chem., 23. 203-204. 17 Campbell, A.D., Chan. E. and Chooi, S.Y., Deady, S.Y. and Shanks, R.A. (1970) J. Chem. Soc. 1065-1067. 18 Brown, R.F., Newsom, H.C. (1%21J. Org. Chem. 27. 3010-3015. 19 Morgenstern. L.. Recanalini. M., Klein. T.E., Steinmetz, W., Yang, C.-Z., Langridge, R. and Hansch, C. (1987) J. Biol. Chem. 262, 10767-10772. 20 Hayes, N.V. and Branch, G.E.K. (1043) J. Am. Chem. Soe. 65, 1555-15f~4. 21 Hansch, C,, Grieco, C., Silip~), C. and Vittoria, A, (1977) J. Med. Chem. 20. 1420-1435. 22 Bender, M.L,, K~zdy, F.J. (19651Annu. Rev. Biochem. 34. 49-76. 23 Selassie. C.D., Fang, Z.-X.. Li, R.L.. Han~h. C., Debnath, G., Klein, T.E.. Langridge, R. and Kaufman, B.T. (1989) J. Med. Chem. 32, 18~5-1905.