Inhibition of human serine proteases by substituted 2-azetidinones

ARCHIVESOFBIOCHEMISTRYAND BIOPHYSICS Vol. 296, No. 2, August 1, pp. 704-708, 1992

COMMUNICATION Inhibition of Human Serine Proteases by Substituted 2-Azetidinones Wilson B. Knight,’ Renee Chabin, and Barbara Green Department of Enzymology, Merck, Sharp, and Dohme Research Laboratories, Building 8OY-150, P.O. Box 2000, Rahway, New Jersey 07065

Received February 10, 1992, and in revised form April 23,1992

trans-4-Ethoxycarbonyl-3-ethyl-1-(4-nitrophenylet al. sulfonyl)-azetidin-3-one described by Firestone (1990, Tetrahedron 46,2265) as an inhibitor of human leucocyte elastase (HLE) displayed potent, time-dependent inhibition of both HLE and human cathepsin G (CatG). The c&isomer was 7- and ISO-fold less active, respectively. The mechanism likely involves opening of the @-la&am ring by the active site serine to form an acylenzyme intermediate(s). This intermediate partitions with ratios of 4: 1 between turnover of the inhibitor and formation of relatively stable enzyme-inhibitor complexes from both enzymes. The final HLE-inhibitor complex reactivated with a half-life of 48 h at 25°C and was 16-fold more stable than the Cat-G-inhibitor complex. The stability of the acyl-enzymes supports a “double hit” chemical mechanism involving both serine acylation and alkylation of the histidine. These observations suggest that /3-lactams may be developed as a class of serine protease inhibitors. (D1992 academic press, IW.

Serine proteases have been implicated in the pathogenesis of a number of disease states involving destruction of structural proteins of connective tissues. Therefore, there is ongoing interest in designing inhibitors of these proteases. Human leucocyte elastase (HLE)’ is a potent serine protease found in human neutrophils. Several different classes of small heterocyclic compounds have been reported to inhibit HLE (1). These included bicyclic /?-la&am derivatives of cephalosporins (2). Recently, Knight et al. (1992) (3) provided evidence that this class 1To whom correspondence should be addressed. ’ Abbreviations used: Ala, alanine; Cat-G, cathepsin G, E, enzyme; I, inhibitor; HLE, human leucocyte elastase; DMSO, dimethyl sulfoxide; L-652,117, trans-4-ethoxycarbonyl-3-ethyl-1-(4-nitrophenylsulfonyl)azetidin-3-one; L-652,306, cb-4-ethoxycarbonyl-3-ethyl-1-(4-nitrophenylsulfonyl)-azetidin-3-one; PMN, polymorphonuclear neutrophils or leucocytes; PPE, porcine pancreatic elastase; Tes, 2-{ [2-hydroxy-l,lbis(hydroxymethyl)ethyl]amino}ethanesulfonic acid; Val, valine. 704

of inhibitors inactivates HLE according to the kinetic mechanism shown in Scheme 1. Recently, Firestone et al. (4) reported that two classes of simple /3-lactams, with different patterns of substitution, were inhibitors of HLE. These workers proposed mechanisms for the inhibition of HLE by these two classes, but did not present supporting data. We have examined the kinetic mechanism of inhibition of HLE by members of one class of these &lactams. In addition, due to the mechanistic similarity of serine proteases we suggest that p-lactams can be developed as a general class of serine proteases. In support of this premise, we have examined the inhibition of human cathepsin-G (Cat-G), a serine protease also found in human leucocytes, by j3-lactams. MATERIALS

AND

METHODS

MeO-AAPV-pNA: Succ-AAPA-pNA, and Succ-AAPF-pNA were purchased from Calbiochem Co., Chemical Dynamics Corp., and Bachem, Inc., respectively. Z-GLF-CMK was purchased from Sigma Chemical Co. Peptidep-nitroanilide stock solutions were prepared in DMSO. The stock concentrations were determined either from the absorbance at 315 nm (E = I5,OOO)or by complete enzymatic hydrolysis to free pnitroanilide (E = 9350 at 410 nm in buffer A (45 mM Tes at pH 7.5,450 mM NaCl, and 10% DMSO), Green et al. (5)). Buffers were purchased from Sigma Chemical Co. and were titrated to the appropriate pH with either NaOH or HCl prior to use. HLE and Cat-G were purchased from Elastin Products and Athens Research Products, respectively. L-652,117 and L-652,306 (Structure I) were obtained from the Merck Chemical Collection and were originally synthesized according to the method of Firestone et al. (4). All experiments were conducted at 25’C. HLE activity in the partitioning and reactivation experiments was determined in 1 ml of buffer A with 1 mM MeO-AAPV-pNA. The kinetic constants &,/[A) for the inhibition of 30 nM elastase by the inhibitors were determined in the presence of 0.2 mM Succ-AAPA-pNA in buffer A. Cat-G activity in the presence and absence of inhibitors was determined with 0.5 mM Succ-AAPF-pNA and 200 nM enzyme in buffer A. Substrate protection experiments. & for the inhibition of HLE by L-652,117 was determined in the presence of 0.2 and 1 mM SUCC-AAPA-

s Peptide-based substrates and inhibitors are abbreviated using the standard one-letter representation of the amino acids. Additional functionalities present were abbreviated as follows: MeO-, methoxysuccinyl; pNA, para-nitroanilide; Succ, succinyl; Z-, carbobenzoxy. ooo3-9861/92 $5.00

Copyright (D 1992 by Academic Press,Inc. All righta of reproductionin any form reserved.

INHIBITION

OF ELASTASE

BY 8-LACTAMS

ki

E+I

=

E.1

-

SLOW

-

E-l

705

(E-l)*

E+I Modified*

-

E + bhdified

SCHEME

pNA in buffer A. The concentrations flLM and 30 nM, respectively.

of inhibitor

and enzyme were 1.2

Partitioning between inhibitor turnover and inactivation. The minimum amount of L-652,117 required to yield complete inactivation was determined by preincubation of 4 pM HLE with from O-6 eq of inhibitor in buffer A. After 2 h the residual enzyme activity in lo-n1 aliquots was then determined. In preliminary experiments we determined that, after 2 h, there was no further loss of activity at all levels of inhibitor. Similar experiments were conducted with 2.1 pM Cat-G except preincubation was for either 2 or 3 h and from 0 to 17 eq of inhibitor was used. Reuctiuation rates. HLE (0.12 mM) was incubated with 6 eq of L652,117 at 25°C for 30 min (99% inactivated). The control lacked inhibitor. The solutions were then subjected to centrifugal gel filtration on 3 ml Sephadex A-25 columns equilibrated in buffer A to remove excess inhibitor or any metabolites generated. HLE activity was then monitored for a period of approximately 2 days at 25°C. The reactivation of L-652,117-inhibited Cat-G was monitored similarly except 3.9 pM enzyme was preincubated with 26 PM inhibitor for 30 min before filtration. Activity was monitored for 24 h. uv-visible spectroscopy was conducted on a Varian DMS-300 spectrophotometer. The hydrolysis of peptide p-nitroanilides in buffer A was monitored at 410 nm. The data were fit by linear least-squares regression to obtain the initial rates. The nonlinear progress curves observed in the presence of time-dependent inhibitors of HLE and CatG were fit to Eq. [l] to obtain the first-order rate constant, kobs(6). The second-order rate constants (k&a) were calculated from kO, and the inhibitor concentration. When the ratio of [S]/K,,, is much less than 1 and the concentration of inhibitor is small relative to K; then k&[ZJ is approximately equal to h-/K, according to Eq. [2]. Y = u.*x + ((u. - u,)(l - e(-ko*x))/k,J

ki.actlK = (k,t,,l[ll)

X (1 + WKJ

L-652,1 17 Trans) L-652,306 I Cis) STRUCTURE

I

+ A,,

111 121

1

RESULTS In Fig. 1 a typical progress curve for the hydrolysis of a peptide p-nitroanilide substrate by HLE in the presence of L-652,117 is shown. L-652,306 was also a time-dependent inhibitor of HLE (data not shown). Similar progress curves were obtained for the Cat-G-catalyzed hydrolysis of its substrate in the presence of the two inhibitors. Progress curves for substrate hydrolysis in the absence of inhibitors were linear for both enzymes over the time course of the experiments. From data such as Fig. 1 Iz,b. was determined and k,,.,JA calculated. These results are summarized in Table I. The first-order rate constants for the inhibition of HLE by L-652,117 in the presence of 0.2 and 1.0 mM Succ-AAPA-pNA were 0.007 + 0.001 and 0.0049 + 0.0002 s-l, respectively. These values are consistent with the dependence of kobaon the substrate concentration predicted by Eq. [2] (K, for this substrate is 2.7 mM (5)).4 In Figs. 2A and 2B the titrations of HLE and Cat-G activity with L-652,117 are presented. It required 5 eq of the compound to completely inactivate HLE. Due to the relative instability of the Cat-G-L-652,117 complex we can estimate

only

that

a maximum

of 5 eq are required

to

inactivate 1 eq of cathepsin G. The L-652,117-derived HLE and Cat-G enzyme-inhibitor complexes regained activity with time. Both processes followed first-order kinetics and the first-order rate constants and half-life for the return of HLE and Cat-G activity after inhibition by these compounds are summarized in Table I. DISCUSSION The structurally simple /3-lactams L-652,117 and L-652,306 were time-dependent inhibitors of HLE and Cat-G. Both enzymes were preferentially inhibited by the trun.s-isomer L652,117. Cat-G displayed a trunsck selectivity of 160 while HLE displayed only a 7-fold preference. The potency of L-652,117 versus HLE rivaled that displayed by the cephalosporin derivatives reported by Green et al. (3). The potency of L-652,117 as a serine protease inhibitor is more evident when compared to known inhibitors of Cat-G. The compound was at least a 400* Equation [2] predicts a ratio of kk at 1 mM to k,b at 0.2 mM of 0.78 with succ-AAPA-pNA. The experimentally determined ratio is 0.7. As pointed out during review the prediction of the amount of protection afforded by substrate may be complicated by partitioning of an E-I complex. Waley (17) suggests that if (1 + r) (E)/(Z) is much less than 1 (r, the partition ratio), the interpretation is straightforward. In this work this value is 0.125 and less than 10% of the substrate is consumed during the course of the reaction. The observation that the ratio of firstorder rate constants at the two substrate concentrations is within experimental error of the calculated value suggests that partitioning does not significantly affect the observed substrate protection.

706

KNIGHT,

CHABIN,

AND GREEN TABLE I Kinetic

Parameters for the Inhibition and Cat-G by j3-Lactams

Inhibitor HLE L-652,117 0

f

I

0

100

I

I

I

200

300

400

Time (s)

FIG. 1. Time-dependent inhibition of the HLE (30 n&catalyzed hydrolysis of 0.2 mM Succ-AAPA-pNA by 1.2 pM L-652,117.

L-652,306

Cat-G L-652,117 L-652,306

Z-GLF-CMK”

6000 + 840 *

20 90

4700 f 200 26+ 2

lo+

1

of HLE

Partition ratio *

kctb 0-l)

4.0 +

0.0145 f 0.005

48

nd

nd

nd

+ 0.06 nd

3 nd

nd

nd

G4 nd

nd

0.25

h/z (h)

’ The values are calculated from the first-order rate constant for inactivation and the inhibitor concentration in the assay. The errors are standard deviations from at least two determinations. bThe partition ratio was determined from titrations of enzyme activity with inhibitor as in Fig. 2. The reactivation rates were determined from the first-order return of enzymatic activity from the E-I complex after removal of excess inhibitor. The errors are standard errors. ’ This activity is fivefold lower than the value reported by Powers et al. (7) under somewhat different experimental conditions.

fold better inhibitor of Cat-G than the chloromethyl ketone previously reported (7). The compound was almost 50-fold more potent versus Cat-G than the most potent 3-halo-3-(l-haloalkyl)-1(3H)isobenzofuranones and dichloroisocoumarins reported by Hemmi et al. (8) and Harper et al. (9). The observations that complete inhibition of HLE and CatG required more than a single equivalent of inhibitor suggests that there is partitioning between turnover of inhibitor (lz,) and enzyme inactivation (ki). The partition ratios’ kdki for the inactivation of HLE and Cat-G by L-652,117 were 4 and <4, respectively. The latter value is an upper limit due to the instability of the Cat-G-inhibitor complex (vide infra). An optimally efficient inhibitor would not partition (kJki = 0). The partition ratio for the inactivation of HLE by L-652,117 is similar to the most efficient cephalosporin derivative reported by Knight et al. (3). The efficiency of L-652,117 is more evident when compared to the partition ratios of 115-7000 obtained during the inactivation of /3-lactamases by fl-lactams (10, 11). L-652,117 forms stable enzyme-inhibitor complexes with both HLE and Cat-G that slowly reactivate. The complex formed between the compound and HLE is 16-fold more stable than formed with Cat-G. The data presented above support the minimal kinetic mechanism shown in Scheme 1 for the inhibition of HLE and CatG by L-652,117. While we were unable to clearly demonstrate that kobsfor the inhibition of HLE or Cat-G was saturable due to the solubility of the compound it is reasonable to assume formation of a reversible Michaelis complex. That the inhibitor binds at the active site of HLE is supported by the observation of substrate protection from inhibition. The Michaelis complex converts to a complex that can then partition between turnover and formation of the more stable (E-I*) complex. The chemical identity of the individual complexes is more speculative. Firestone et al. (4) proposed a chemical mechanism similar to that in Fig. 3 for the inhibition of HLE by this class of &lactams (see Fig. 3). Given the mechanism of serine protease catalysis it is likely that the active site serine attacks the /3la&am ring, forming a tetrahedral intermediate (2) that collapses by opening the /%lactam ring (3). The stability of acyl-enzymes derived from cephalosporins and HLE has been explained by a

“double hit” mechanism involving both acylation and alkylation of the histidine (2, 3). These proposals were supported by the X-ray crystallographic structure observed after the inactivation of PPE with a cephalosporin derivative (12). In these reports an additional electrophile was generated during the course of the HLE-catalyzed opening of a @-lactam ring as the result of expulsion of a leaving group. The stability of the L-652,117derived complex is similar to that reported by Knight et cd. (3) for the cephalosporin derivatives. This is consistent with a double hit mechanism. To accommodate this requirement, Firestone et al. (4) proposed that HLE catalyzed the production of an imine (4) by P-elimination of nitrobenzosulfinate from the acyl enzyme (3). Addition of the histidine to the imine would produce the stable E-I complex (6). The identity of the complex that partitions could be any of the acyl-enzymes (3, 4, or 5) prior to alkylation of the histidine. The observation that L-652,117 inhibits both HLE and CatG was somewhat surprising considering their differences in substrate specificity toward the Pi and P’i sites6 (13,14). This could be explained if the inhibitor bound in different modes to the active site of the two enzymes. HLE prefers small aliphatic residues at Pi (the scissile bond) such as Ala or Val. Firestone et al. reported that p-lactams substituted with small alkyl groups at C-3 of the lactam ring were better inhibitors of HLE. This observation and the substrate specificity of HLE suggest that the ethyl group at C-3 of L-652,117 binds in the S1 subsite of HLE (see Fig. 4). Cat-G is a chymotrypsin-like serine protease and prefers large aromatic residues at the P1 site of substrates (13,14). McRae et al. reported that an N-acetylated hexapeptide with Phe at Pi yielded kc&Cm of 1000 Me1 s-l (14). Substitution of Val for Phe at P1 resulted in a loo-fold decrease in activity

5The partition ratio equals the number of equivalents required for complete inhibition minus 1.

8The enzyme subsites and the amino acid residues of substrates are numbered according to the nomenclature of Schecter and Berger (16).

INHIBITION

0.00

I

0

v

8

1

=

I

2

g

a

3

OF ELASTASE

0

4

707

BY fi-LACTAMS

0.001 0

2

4

6

6

10

[L-652,1 17]/ [Cat-G]

[L-652,1 17] / [HLE]

(A) Residual HLE activity as a function of the ratio of L-652,117 to HLE after preincubation with inhibitor for 2 h. (B) Residual CatG activity as a function of the ratio of L-652,117 to Cat-G after (0) preincubation with inhibitor for 2 h prior to assay and (0) preincubation with inhibitor for 3 h prior to assay.

FIG. 2.

versus the hexapeptide. A reasonable binding mode of L-652,117 in the active site of Cat-G would place the nitrophenyl group in the S, subsite (see Fig. 4). In fact, Hubbard and Kirsch proposed that the p-nitrophenyl group of p-nitrophenyl benzoates binds

to the hydrophobic S, subsite of chymotrypsin (15). The same study reported acylation of chymotrypsin by p-nitrophenyl-pnitrobenzoate. These data suggest that the active sites of chymotrypsin-like enzymes could accommodate the nitrophenyl

s

-so2 u-&

NO2

FIG. 3. Proposed chemical mechanism for the reaction of L-652,117 with serine proteases. The active site serine opens the B-la&am ring via a tetrahedral intermediate (2) to produce the acyl-enzyme (3). B-Elimination of p-nitrobenzosulfinate, catalyzed by proton transfer to the histidine,

produces the imine (4). Attack of histidine on the imine after proton transfer (6) produces the relatively stable acyl-enzyme (6).

708

KNIGHT,

CHABIN,

AND

GREEN

2. Doherty, s

J. B., Ashe, B. M., Argenbright, L. W., Barker, P. L., Bonney, R. J., Chandler, G. O., Dahlgren, M. E., Dorn, C. P., Finke, P. E., Firestone, R. A., Fletcher, D., Hagmann, W. K., Mumford, R., O’Grady, L., Maycock, A. L., Pisano, J. M., Shah, S. K., Thompson, K. R., and Zimmerman, M. (1986) Nature 322, 192-194. 3. Knight, W. B., Maycock, A. L., Green, B. G., Ashe, B. M., Gale, P., Weston, H., Finke, P., Hagmann, W., Shah, S., and Doherty, J.

HLE 1

(1992)Biochemistry

31.

4. Firestone,

/

R. A., Barker, P. L., Pisano, J. M., Ashe, B. M., and Dahlgren, M. E. (1990) Tetrahedron 46, 2255-2262. 5. Green, B. G., Weston, H., Ashe, B. M., Doherty, J., Finke, P., Hagmann, W., Lark, M., Mao, J., Maycock, A., Mumford, R., Walakovits, L., and Knight, W. B. (1991) Arch. Biochem. Biophys. 286, 284-

Cat-G Sl

292.

FIG. 4. Proposed schematic of the binding of L-652,117 in the active sites of HLE and Cat-G. The difference in substrate specificity between the enzymes and the potent inhibition of both enzymes by L-652,117 can be explained by recognition of different sites of the inhibitor by the specificity pockets (S,) of the two enzymes.

6. Morrison, 301.

J. F., and Walsh, C. T. (1988) Adu. Enzymol.

7. Powers, J. C., Gupton, B. F., Harley, A. D., Nishino, R. J. (1977) Biochim. Biophys. Acta 485, 156-166. 8. Hemmi, K., Harper, 24,1841-1848.

61,

201-

N., and Whitley,

J. W., and Powers, J. C. (1985) Biochemistry

9. Harper,

The existence of different binding modes would extend the utility of /3-lactams as serine protease inhibitors. Specificity could be fine-tuned by varying the substituents at either C-3 or N-l of the lactam ring. While these data do not address this issue, it may also be possible to adjust the specificity of this class of compounds by varying the group at C-4 of the lactam ring. group.

ACKNOWLEDGMENT We acknowledge

the aid of M. Driscoll

in the preparation

of this

paper. REFERENCES 1. Stein, R. L., Trainor, D. A., and Wildonger, R. A. (1985) in Annual Reports in Medicinal Chemistry (Bailey, D., Ed.), Vol. 20, pp. 237246, Academic Press, San Diego.

J. W., Hemmi, K., and Powers, J. C. (1985) Biochemistry 24,1831-1841. 10. Fisher, J., Charnas, R. L., and Knowles, J. R. (1978) Biochemistry 17,2180-2184.

11. Brenner,

D. G., and Knowles,

J. R. (1991) Biochemistry

20,3680-

3687. 12. Navia,

M. A., Springer, J. P., Lin, T. Y., Williams, H., Firestone, R. A., Pisano, J. M., Doherty, J. B., Finke, P. E., and Hoogsteen, K. (1987) Nature 327, 79-82. 13. Nakijima, K., Powers, J. C., Ashe, B. M., and Zimmerman, M. (1979) J. Biol. Chem. 254,4027-4032. 14. McRae, B., Nakajima, K., Travis, J., and Powers, J. C. (1980) Bio-

chemistry19,3973-3978. 15. Hubbard, 2493.

C. D., and Kirsch,

J. F. (1972) Biochemistry

11, 2483-

16. Schecter, I., and Berger, A. (1967) Biochem. Biophys. Res. Commun. 27,157-162.

17. Waley, S. G. (1985) Bzixhem. J. 227,843-849.