We Utilized a Similar Approach with DEP-Chymotryp- Two of The

PROBING THE TOPOGRAPHY OF THE ACTIVE SITE OF a-CHYMOTRYPSIN BY BERNARD F. ERLANGER

DEPARTMENT OF MICROBIOLOGY, COLLEGE OF PHYSICIANS AND SURGEONS, COLUMBIA UNIVERSITY Communicated by Erwin Chargaff, May 16, 1967 This paper is a report of a study of the topography of the active center of a- chymotrypsin. In particular, it will deal with that portion involved in substrate binding and with its position in space relative to the nucleophilic serine residue that is believed to function in the catalytic mechanism.1 Since an enzyme must accommodate a substrate, the simplest approach to an understanding of the topography of the former might be a study of the conformation of the latter. However, this is not feasible with chymotrypsin. A typical substrate of chymotrypsin is acetyl L-phenylalanine ethyl ester.1 There is much evidence linking the susceptibility of this substrate to the aromatic ring of the amino acid (though aromaticity is not necessary since cyclic hexahydro derivatives are also substrates)2 and to the acylamido portion of the molecule. Benzoylamido may substitute for acetylamido and, in fact, yields a somewhat better substrate (higher keat)3 in the case of acyl L-tyrosine amide. However, since all of the above substrates are flexible molecules and therefore able to assume many conformations, there is no way of knowing their conformation when bound to the active center of a-chymotrypsin. In our early studies we chose to use an approach based upon the experiments of Wilson with acetylcholinesterase.4 Acetylcholinesterase and chymotrypsin belong to the class of enzymes called serine esterases.1 All members of this class contain an unusually reactive serine residue which acts as a nucleophile in the catalytic mechanism. They are inactivated by certain organophosphorus compounds by a reaction that can be understood within the context of the enzymatic mechanism of the serine esterases. Briefly, they participate in the acylation step of the mechanism but deacylate at an extremely slow rate unless nucleophiles considerably stronger than water are introduced, e.g., hydroxylamine, oximes, or hydroxamic acids. Wilson4 had shown that the specificity sites of diethylphosphoryl (DEP)-cholin- esterase were still available for binding of substrate-like molecules. It was possible for him to design reagents which were bound by the inactivated (phosphorylated) enzyme in such a way as to position their nucleophilic functional group one bond length away from the phosphorus atom, thus ideally situated for a nucleophilic displacement reaction. From a consideration of the structure of his best reactivators, he was able to map out some details of the topography of the active center of acetylcholinesterase. In our early experiments,' we utilized a similar approach with DEP-chymotrypsin. A large number of oximes and hydroxamic acids were tested as reactivators. Two of the compounds were found to be unusually reactive: N-phenylbenzohydroxamic acid and N-phenylnicotinohydroxamic acid. They were, respectively, 15 and 8 times faster reactivators than either benzohydroxamic acid or N-phenyl- 703 Downloaded by guest on September 26, 2021 704 BIOCHEMISTRY: B. F. ERLANGER PROC. N. A. S.

acetohydroxamic acid. Hence, it was tentatively concluded that DEP-chymotrypsin had two sites capable of binding aromatic rings and that when it did so with N-phenylbenzohydroxamic acid and N-phenylnicotinohydroxamic acid, the nucleophilic hydroxamic acid group was favorably situated for a displacement of the phosphate group. Supporting evidence was obtained from experiments with DEP-trypsin.6 Since the specificity of trypsin is entirely different from that of chymotrypsin, N-phenylbenzohydroxamic acid would not be expected to be more effective than benzohydroxamic acid or aliphatic hydroxamic acids. It was, in fact, somewhat less active. Thus, our early experiments allowed us to conclude that the active center of a- chymotrypsin contained two ring structure-binding sites that could be utilized by the enzyme in its catalytic mechanism. However, their positions in space relative to each other and to the active serine residue could not be deduced because of the structural flexibility of the reactivator molecules. Later findings enabled us to define the topography more exactly. In 1963, it was found that diphenylcarbamyl chloride (DPCC) was a specific inactivator of chymotrypsin.7 Like the organophosphates, it inactivated the enzyme by a process that made use of the catalytic mechanism of the latter. Subsequent studies by Metzger and Wilson8 showed that methyl phenylcarbamyl chloride was also an inactivator of a-chymotrypsin but that it was about 150 times less active than DPCC, indicating again the presence on the enzyme of two ring- binding sites which could orient the carbonyl carbon properly with respect to the reactive serine. Construction of a space-filling model of DPCC revealed considerable steric interference with the freedom of rotation of the two aromatic rings, but still enough freedom remained to make an unambiguous mapping of the active center unfeasi- ble. Encouraged by these findings, we set about to design a completely constrained inactivator. Among the compounds synthesized was phenothiazine-N-carbonyl chloride (PCC),9 the structure of which is given in Figure 1, along with several other reagents specific for chymotrypsin. Its resemblance to DPCC is apparent; however, the additional sulfur bridge restricts the movement of the two aromatic rings completely. Even rotation around the N-C bond is restricted severely by neigh- boring hydrogens on the beuzene rings. Hence, fixing onie aromatic ring determines the 1)ositioIl in space of the susceptible carbonyl chloride bond. iPCC is an effective and specific inactivator of chymotrypsin. The second-order rate constant for its inactivation of a-chymotrypsin (105 liter mole-' sec-) is about one fourth that of DPCC. However, as shown by its reaction with OH-, it is inherently less reactive than DPCC by a factor of about 20. Correction for this lower reactivity in nucleophilic reactions leads to the conclusion that lPCC is even a more specific reagent for a-chymotrypsin than is DPCC. Another interesting facet of the reaction of PCC with a-chymotrypsin is that spontaneous reactivation occurs at a slow but measurable rate (k4 = 25.3 X 10-4 min' at 370, pH 7.55). PCC, therefore, is more like a susceptible substrate. This would imply that less distor- tion of the active center occurs as a result of acylation of the enzyme by PCC than by DPCC. Downloaded by guest on September 26, 2021 VOL. 58, 1967 BIOCHEMISTRY: B. F. ERLANGER7705

Cl Cl C=O C=O

Diphenylcorbamyl Phenothiazine -N- Corbonyl Chloride Chloride

0 A tteC5OCHI H C-OCH3 Il"?H NH 0

0 iIi D-i-Keto-3-Corbomethoxy- o 0- l,2,3,4,-Tetrchydroieoquinoline IfOH Benzoyl L-Phenylalanine C1. .,HMethyl Ester

N -Phenylbenzohydroxomic Acid FIG. 1.-Chemrical structures of specific substrates, inactivators, and reactivators of chymotrypsin.

If we now construct models of all of the reagents shown in Figure 1, it becomes possible to map the topography of the active center of chymotrypsin. Among the reagents included is the substrate discovered by Hein et al.,'0 D-1-keto-3- carbomethoxy-1,2,3,4-tetrahydroisoquinoline (D-CDIC). D-CDIC can be looked upon as a D-phenylalanine analogue of benzoyl L-phenylalanine methyl ester with one aromatic ring serving a dual role as the acylamido group and as the benzene ring of phenylalanine. Surprisingly, the L-enantiomorph was a poor substrate. Wilson and ErlangerII provided an explanation for this unusual situation by postulat- ing that the aromatic ring was bound as if it were the benzoylamido group of benzoyl L-phenylalanine ethyl ester. This orientation will be utilized in the present discussion. An important aspect of this substrate is its highly constrained structure and hence its potential utility as a mapping tool. A model of benzoyl L-phenylalanine methyl ester was constructed in order to find a conformation that could be assumed by all of the constrained reagents. Only one could be found; it is shown in Figure 2. The molecule on the upper left is phenothiazine-N-carbonyl chloride; on the uppel right, beulzoyl L-phenylalanine methyl ester; on the lower right, D-CDIC. Asterisks indicate the target carbonyl carbons of the carboxylic acid derivatives; the leaving group, whether chloride or ester, is marked X. As noted above, the D-substrate is arranged so that its aromatic ring coincides with the benzoyl group (ring A) of benzoyl L- phenylalanine methyl ester. It should also be noted that the amide bonds coincide. The position of ring B could not be ascertained from models of benzoyl L-phenylalanine methyl ester and the D-substrate alone. The rigidity of PCC, however, allows us to position ring B in space. Its plane is at about 900 to that of ring A. The susceptible carbonyl carbons (asterisks) are attacked by the enzyme from below. As required, their positions in space coincide in the three models. Downloaded by guest on September 26, 2021 706J BIOCHEMISTRY: B. F. ERLANGER PROC. N. A. S.

FiG. 2.-Models of phenothiazine-N-carbonyl chlor- FIG. 3.-Models of N-phenyl- ide (upper left), benzoyl-Lphenylalanine methyl ester benzohydroxamic acid and benzoyl- (upper right), and D-l-keto-3-carbomethoxy-1,2,3,4- L-phenylalanine methyl ester (X tetrahydroisoquinoline (lower right). Leaving group serving to indicate ester leaving indicated by X; acylamido oxygen, OI; carboxyl group). Arrow indicates carboxyl oxygen, Oli. Enzyme makes attack from below on carbon which is in position that carboxyl carbon marked with asterisk. would be occupied by phosphorus atom of inactivated chymotrypsin. Asterisk designates nucleophilic oxygen of hydroxamic acid. Other designations similar to those in Fig. 2. In Figure 3 are models of benzoyl L-phenylalanine methyl ester (upper) and N phenylbenzohydroxamic acid (lower). The benzene ring of benzohydroxamic acid is aligned with that of the benzoyl function of the substrate, bringing the carbonyl groups in coincidence. Also, the N-phenyl moiety of the reactivator can be practically superimposed upon the benzene ring of the phenylalanine residue of the substrate. An asterisk indicates the nucleophilic oxygen of the hydroxamic acid. It is nearer to ring A than is the target carbon atom of the substrate (arrow) and is below the plane of the latter, placing it in an ideal position for a nucleophilic attack on the carbonyl (or phosphoryl) group.'2 It is interesting to note that the reactivator can also be turned around so that its N-phenyl group coincides with ring A as shown in Figure 4. Now the liucleo- philic oxygen falls in the same plane but one bond length to the right of the carboxyl carbon-again properly placed. 13 Figure 5 is a drawing of what is believed to be the conformation of benzoyl L- phenylalanine methyl ester when it is bound to the enzyme. It is the same as shown with the space-filling models in the previous figures, but somewhat exploded (and distorted) to make its details more obvious. This conformation is assumed by the molecule as a result of interactions with three sites on the enzyme: two which bind rings A and B, respectively, and one which interacts with the amide function of the substrate. Acetyl-L-phenylalanine ethyl ester, which lacks ring A, assumes the reactive conformation by interaction of ring B and the amido group Downloaded by guest on September 26, 2021 VOL. 58, 1967 BIOCHEMISTRY: B. F. ERLANGER 707

0 -' / en

A d C t

FIG. 4.-Same as Fig. 3, FIG. 5.-Sketch of conformation of benzoyl-Iphenylalanine except that positions of rings methyl ester when bound to surface of enzyme. E-OH A and B of N-phenylbenzohy- represents nucleophilic portion of enzyme making attack on droxamic acid are interchanged. susceptible carboxyl carbon.

with the active center of the enzyme. This is sufficient for proper orientation of the susceptible carbethoxy function. The nucleophilic portion of the enzyme is represented by E-OH. It is attacking 0 the carbonyl in a direction perpendicular to the plane of the C-X group. As noted, the carboxyl carbon is 0.5 A above the plane of ring A and 1.2 A to the left of plane B. We believe that this arrangement is in agreement with all of the experimental data on chymotrypsin. In any case, it arises from our inactivation-reactivation studies as well as from a consideration of the findings of others. There are several implications that follow from the suggested conformation: (a) The carbomethoxy group of D-CDIC is equatorial when it is attacked by chymotrypsin. This is contrary to the suggestions of Hein and Niemann"4 and of Awad, Neurath, and Hartley,15 both of whom require that the carbomethoxy group be axial to the ring. However, it is known16 that hydroxide ion preferentially hydrolyzes an equatorial carboxylic ester, and the recent work of Silver17 using 4-t-butyl-cyclohexane substrates of chymotrypsin indicates that enzymic hydrolysis of the equatorial ester group is also favored over that of an axial ester. Thus, the published experimental evidence favors an equatorial conformation. (b) The acylamido group plays an important part in the binding of benzoyl L-phenylalanine methyl ester to the enzyme and is in the cis configuration. The latter is a requirement; it is not possible to construct the benzoyl L-phenylalanine methyl ester model "properly" if a trans configuration is used. It is also likely, but not necessary for our hypothesis, that the amide function also plays an impor- Downloaded by guest on September 26, 2021 70.S0BIOCHEMISTRY: B. F. ERLANCGER PROC. N. A. S.

taut part ill the hiitdliig of D)-CDIC; ald, of course, if' this case it, caI oily be ill the cis coillfiguratioiL'8 Silver'7 co(nsiders this to he the major defect of our model aid, inl fact, lie dismisses it as a possibility for this otie reason, iniplying that flexible substrates must have a trans amide bond inl solution. This leaves only the aromatic ring as a participant in binding, a condition which, in our opinion, may not be adequate for the proper placement of the susceptible carbomethoxy group. There are a number of studies that have shown that amides can exist both in the cis and trans configurations in solution and can, in fact, be converted from one form to another rather easily. For example, Berger, Lowenstein, and M\eiboom'9 have shown by N1IR studies that N,N-dimethylacetamide has a considerable degree of free rotation around the amide bond in aqueous solution below pH 5.0. The same authors showed the same to be true for N-methylacetamide at pH's below and above pH 5.0. They proposed that this free rotation occurred because of the existence of a small but significant quantity of a protonated, freely rotating species in equilibrium with the unprotonated amide. Hanlon and Klotz20 have shown that polyamino acids can be protonated at the peptide bond in solutions containing extremely low concentrations of trifluoroacetic acid and that the result is an increase in free rotation around the peptide bond. Even formic acid has been shown to be able to protonate peptide bonds.21 There is no reason to assume that the same type of protonation, and hence free rotation, cannot take place at the surface of the enzyme under the influence of properly placed proton donors and that, subsequent to this, the cis conformation be favored by the geometry of the binding site. We believe that this is more likely than the suggestion that the acylamido groups are not important in the enzymic process. A recent paper by Cohen and Schultz22 also minimizes the importance of the amido linkage of D-CDIC, and, like Silver,'7 suggests that it does not take part in the binding interaction. The basis of their conclusion is that an ester analogue, D-methyl-3,4-dihydroisocoumarin-3-carboxylate, is an excellent substrate. This finding, however, does not rule out the possibility that it is the carbonyl group that is essential for the binding reaction. In a paper of Silver and Sone,23 naphthoic acid nitrophenyl esters were shown to be substrates of chymotrypsin. A particu- larly good one was the 1,2-dihydronaphthoic acid derivative in which the ester group was equatorial. Silver and Sone estimated that it was about as reactive as the nitrophenyl ester of the D-CDIC would be. (The latter has never actually been synthesized and studied.) This work is taken to support the minor importance of the amido group in the reactivity of the D-substrate. There are two major criticisms of this work. First of all, no comparisons were made of the relative reactivities of the various esters in the absence of enzyme. Therefore, no account was taken of the steric and electronic factors that influence their behavior in nucleophilic reactions. The second criticism concerns their extrapolation of the results with nitrophenyl esters to ethyl or methyl esters. It would be safer to wait until studies with the less active esters have been completed. It is not a necessary condition of our hypothesis that the amide function play a part in the binding and orientation of D-CDIC, although we think that it is a likely possibility. It must, however, participate in the binding of flexible substrates, such as benzoyl L-phenylalanine ethyl ester. According to the models of Silver'7' 23 and of Cohen and Schultz,22 it cannot participate in the binding of the D-CDIC Downloaded by guest on September 26, 2021 VOL. .59, l.967 BIOCHEMISTRY: B. F. ERLANGER 70on

because they require that the aromatic ring of the D-substrate be bound by the site of the enzyme reserved for the benzene ring of phenylalanine (ring B of Fig. 4). This orientation of the substrate is the major difference betwveen their models and the one described in. this paper. As far as we can determine using space-filling models, their hypotheses cannot explain the activity of PCC. Our suggested conformation also agrees with the finding by Hein and Niemann24 that indole was a competitive inhibitor of the hydrolysis of the D-substrate but not of the (slower) hydrolysis of its L-isomer, for which the inhibition was of the mixed type. This can be explained if we assume that the L-isomer is hydrolyzed when bound to the other ring-binding site.25 In order to explain this phenomenon, Hein and Niemann postulated'4 that binding of the D-isomer occurred in a plane that subtended two binding sites rather than at either site. In a later paper,24 however, the same authors stated that the experimental evidence suggested "that the pi locus is involved in binding of the (D) substrate." The P1 locus is their designation of the acyl binding site which in our case is defined as the site that binds the benzoyl group (ringA). Their own model, however, could not be made to fit this conclusion and they were led to conclude that "the attempt to incorporate these conforma- tionally constrained molecules into the theory has done more to demonstrate the striking differences between these compounds and the more conventional acylated a-amino acid derivatives than to clarify the theory." Cunningham' is critical of this conclusion and so are we. Finally, if one measures the dimensions of the space-filling model of benzoyl L- phenylalanine methyl ester, the dimensions of the active site "cavity" of chymotrypsin can be estimated. Excluding the space required for the leavinggroups, the dimensions of the cavity are 9.5X X11 X X 6.5 A, or approximately 680 A3. In summary, the inactivation-reactivation studies on chymotrypsin have allowed us to propose a conformation assumed by susceptible substrates when bound to the enzyme. The proposed conformation has been shown to be in agreement with experimental data originating in other laboratories as well as in our own. It re- mains now to test the suggested conformation by attempting to design new substrates of chymotrypsin. We wish to acknowledge the financial assistance of the Office of Naval Research, the National Science Foundation, and the National Institutes of Health. 1 Cunningham, L. W., in Comprehensive Biochemistry, ed. M. Florkin and E. H. Stotz (New York: Elsevier Publishing Company, 1965), vol. 16, p. 85. Jennings, R. R., and C. Niemann, J. Am. Chem. Soc., 75, 4687 (1953). 'Green, N. M., and H. Neurath, in The Proteins, ed. H. Neurath and K. Bailey (New York: Academic Press, 1954), vol. 2, pt. B, p. 1057. 4Wilson, I. B., Federation Proc., 18, 752 (1959). 6Cohen, W., and B. F. Erlanger, J. Am. Chem. Soc., 82, 3928 (1960). Cohen, W., M. Lache, and B. F. Erlanger, Biochemistry, 1, 686 (1962). 7Erlanger, B. F., and W. Cohen, J. Am. Chem. Soc., 85, 348 (1963). 8 Metzger, H. P., and I. B. Wilson, Biochemistry, 3, 926 (1964). Erlanger, B. F., S. M. Vratsanos, N. Wassermann, and A. G. Cooper, submitted for publica- tion. 10 Hein, G., R. B. McGriff, and C. Niemann, J. Am. Chem. Soc., 82, 1830 (1960). Wilson, I. B., and B. F. Erlanger, J. Am. Chem. Soc., 82, 6422 (1960). 12 The model of benzoylILphenylalanine methyl ester is used in Fig. 3 only to indicate the conformation that must be assumed by the reactivator. It is not intended to imply that N-phenyl- Downloaded by guest on September 26, 2021 710 BIOCHEMISTRY: B. F. ERLANGER PROC. N. A. S.

benzohydroxamic acid can displace an inactivator possessing the structure of the substrate. In fact, we have found that N-phenylbenzohydroxamic acid does not reactivate diphenylcarbamyl chymotrypsin, presumably because the binding sites of the enzyme are already occupied by the inactivator and unavailable to the reactivator. 13 It is interesting to note at this point that N-phenylphenylacetohydroxamic acid has been found to be a specific reactivator of DEP-chymotrypsin (Cohen, W., and B. F. Erlanger, un- published experiments). Its reactivation rate is about one half that of N-phenylbenzohydroxamic acid, possibly because of the fact that its conformation is correct only when bound in a manner analogous to that shown in Fig. 3. With its rings interchanged as in Fig. 4, its nucleophilic oxygen is displaced beyond the region occupied by the susceptible target group of the inactivated enzyme. 14 Hein, G., and C. Niemann, these PROCEEDINGS, 47, 1341 (1961). 15Awad, E. S., H. Neurath, and B. S. Hartley, J. Biol. Chem., 235, PC35 (1960). 16 Gould, E. S., Mechanism and Structure in Organic Chemistry (New York: Holt, Rinehart and Winston, Inc., 1959), p. 241. 17 Silver, M. S., J. Am. Chem. Soc., 88, 4247 (1966). 18 If we are correct in our assumption that all substrates of chymotrypsin when bound to the enzyme have their acylamido groups in the cis configuration, it is a distinct advantage for the D- substrate to have an acylamido group already fixed in that manner. This may be one of the reasons for the fact that it is a good substrate. Furthermore, since native proteins contain peptide bonds stabilized in the trans configuration, we can readily explain why denatured proteins are more rapidly hydrolyzed by enzymes: in the "random coil" the energy barrier between the cis and the trans configurations should be considerably lower than in the a-helix. After this manuscript was completed, the author's attention was called to a report of the presence of cis and trans peptide bonds in cyclo-(Gly-Phe-Leu-Gly-Phe-Leu) (D,L,L,L) and (L,L,L,L) (Bltha, K., J. Smol6kov4, and A. Vftek, Coll. Czech. Comm., 31, 4296 (1966)). The report was based on infrared studies. 19Berger, A., A. Lowenstein, and S. Meiboom, J. Am. Chem. Soc., 81, 62 (1959). 20 Hanlon, S., and I. M. Klotz, Biochemistry, 4, 37 (1965). 21 Klotz, I. M., S. F. Russo, Sue Hanlon, and M. A. Stake, J. Am. Chem. Soc., 86, 4774 (1964). 22 Cohen, S. G., and R. M. Schultz, these PROCEEDINGS, 57, 243 (1967). 23 Silver, M. S., and T. Sone, J. Am. Chem. Soc., 89, 457 (1967). 24 Hein, G. E., and C. Niemann, J. Am. Chem. Soc., 84, 4487, 4495 (1962). 25 Models of the L-isomer show that when bound in this way, it is necessary for the carbomethoxy group to be axial to the amide ring in order to have it coincide with the position of the carbomethoxy group of the D-substrate. This would explain, in part, why the L-isomer is a poor substrate. Downloaded by guest on September 26, 2021