Leishmania major: Molecular Modeling of Cysteine Proteases andPrediction of New Nonpeptide Inhibitors

Paul M. Selzer,*,2 Xiaowu Chen,†,‡ Victor J. Chan,* Maosheng Cheng,‡ George L. Kenyon,†I. D. Kuntz,†,‡ Judy A. Sakanari,* Fred E. Cohen,†,‡,§,\ and James H. McKerrow*,†,‡,1

Departments of Pathology,* Cellular and Molecular Pharmacology,† Pharmaceutical Chemistry,‡ Biochemistry andBiophysics,§ and Medicine,\ University of California, San Francisco, California 94143, U.S.A.

Selzer, P. M., Chen, X., Chan, V. J., Cheng, M., Kenyon, G. L., Kuntz, I. D., Sakanari, J. A.,Cohen, F. E., and McKerrow, J. H. 1997.Leishmania major:Molecular modeling of cysteine pro-teases and prediction of new nonpeptide inhibitors.Experimental Parasitology87, 212–221. Thecrystal structures of papain, cruzain, and human liver cathepsin B were used to build homology-basedenzyme models of a cathepsin L-like cysteine protease (cpL) and a cathepsin B-like cysteine protease(cpB) from the protozoan parasiteLeishmania major.Although structurally a member of the cathep-sin B subfamily, theL. major cpB is not able to cleave synthetic substrates having an arginine inposition P2. This biochemical property correlates with the prediction of a glycine instead of aglutamic acid at position 205 (papain numbering). The modeled active sites of theL. major cpB andcpL were used to screen the Available Chemicals Directory (a database of about 150,000 commer-cially available compounds) for potential cysteine protease inhibitors, using DOCK3.5. Based on bothsteric and force field considerations, 69 compounds were selected. Of these, 18 showed IC50’sbetween 50 and 100mM and 3 had IC50’s below 50 mM. A secondary library of compounds,originally derived from a structural screen against the homologous protease ofPlasmodium falcipa-rum (falcipain), and subsequently expanded by combinatorial chemistry, was also screened. Threeinhibitors were identified which were not only effective against theL. major protease but alsoinhibited parasite growth at 5–50mM. © 1997 Academic Press

Index Descriptors and Abbreviations:Leishmania major;cysteine proteases; homology modeling;drug design; docking; reversible inhibitors; cpL, cathepsin L-like cysteine protease; cpB, cathepsinB-like cysteine protease; Z, benzoyloxycarbonyl; AMC, 7-amino-4-methyl coumarin.

INTRODUCTION

Parasitic diseases are major worldwide healthproblems now exacerbated by the emergence ofdrug-resistant organisms (Moran and Bernard1989). Leishmaniasis, a spectrum of diseasesproduced by protozoan parasites of the genus

Leishmania,affects more than 12 millionpeople. Current therapy for leishmaniasis issuboptimal due to toxicity of available thera-peutic agents and the emergence of drug resis-tance (Groglet al. 1992). Compounding theseproblems is the fact that many countries andregions where the disease is endemic are eco-nomically poor. As a result, major pharmaceu-tical companies have historically had little in-terest in anti-leishmanial drug development.

To facilitate the cost-effective developmentof new antiparasitic chemotherapy, we havebeen exploring the application of structure-based drug design, utilizing computationalscreens of available chemical databases as aninexpensive shortcut to identify potential che-motherapeutic leads. One appealing target in theLeishmaniaparasites is a family of cysteine pro-teases required for parasite replication and viru-lence (Coombs and Baxter 1984; McKerrowet

1 To whom correspondence should be addressed at: Vet-erans Affairs Medical Center, 4150 Clement Street 113B,San Francisco, CA 94121. Phone: 415-476-2940. Fax: 415-750-6947. E-mail: jmck@cgl.ucsf.edu.

2 To whom correspondence should be addressed at pres-ent addresses: I. Frauenklinik, Ludwig Maximilians Univer-sitat Munchen, Labor fu¨r Biochemie, Maistraße 11, D-80337 Munchen, Germany. Phone: +49-89-5160-4256. Fax:+49-89-5160-4916. E-mail: Paul.Selzer@med.uni-muenchen.de; and Boehringer/Mannheim GmbH, Nonnen-wald 2, D-82377 Penzberg, Germany. Fax: +49-8856-60-2003. E-mail: PaulSelzer@bmg.boehringer-mannheim.com.

EXPERIMENTAL PARASITOLOGY 87, 212–221 (1997)ARTICLE NO. PR974220

2120014-4894/97 $25.00Copyright © 1997 by Academic PressAll rights of reproduction in any form reserved.

al. 1993; Sakanariet al. 1995; Mottramet al.1996). Because of the relationship of these tar-get enzymes to cysteine proteases of knownstructure and catalytic mechanism, it is feasibleto utilize molecular modeling techniques to vi-sualize the active site of the enzyme and com-putationally design or screen for inhibitors.While molecular modeling is not a substitute forcrystallographic structure determination, it hasproven useful in identifying or designing inhibi-tors for two other parasitic infections, schisto-somiasis and malaria (Ringet al. 1993). TheLeishmaniacysteine proteases present a similaropportunity in that there are sufficient data fromstructural analysis of closely related enzymes toallow a reasonable and useful model to be built.

In this initial work, we have produced struc-tural models of the two major cysteine proteasesof L. major,a cathepsin L-like cysteine protease(cpL) and a cathepsin B-like cysteine protease(cpB). These models were used to develop hy-potheses on the structural basis of substratespecificity. The predicted active site binding re-gions of both theL. major cpL and cpB werethen used to screen a public domain database ofsmall molecular weight compounds for poten-tial chemotherapeutic leads. In addition a li-brary of compounds derived from a lead foundby a similar approach using the homologousmalaria cysteine protease as a target (Ringet al.1993; Li et al. 1994, 1995, 1996) was alsoscreened. Because reagent quantities of theL.major cpB were available it was used for con-firmatory screens of DOCK3.5-derived leads.The most promising compounds were investi-gated both for their ability to inhibit parasitegrowth and for any toxicity against host cells.

MATERIALS AND METHODS

Homology Modeling and Docking of Inhibitors

The mature protein sequences ofL. major cpL (216 aa,GenBank locus U43706) andL. major cpB (243 aa, Gen-Bank locus 43705) (Sakanariet al. 1997) were used tosearch the Brookhaven Protein Databank of three-dimensional structures (Bernsteinet al. 1978). Both se-quences show high homology with the papain family pro-teases. Based on the crystal structures of papain (Kamphuiset al.1984), cruzain (McGrathet al.1995), and human livercathepsin B (Musilet al. 1991) homology-based enzyme

models were built (Fig. 1) using the programs InsightII(Biosym Technologies, San Diego, CA) and Midas Plus(Computer Graphics Laboratory, University of CaliforniaSan Francisco) (Ferrinet al. 1988; Huanget al. 1991)(http://www.cgl.ucsf.edu/midasplus.html). ForL. majorcpLthe crystal structures of cruzain and papain (59 and 41%sequence identity, respectively) were used, whereas forL.major cpB, the three-dimensional structure of human livercathepsin B served as reference protein (54% sequenceidentity). Identities in the structural conserved regions(SCR) and especially within the active site cleft of the en-zymes reached values of up to 80%. Comparing knowncrystal structures with models built by homology, modelingrevealed that sequence identities higher than 70% lead tohighly accurate structural predictions (Mosimannet al.1995).

In order to find structurally conserved regions, sequencesof the reference proteins were manually aligned based ontheir secondary structure. The sequence of the correspond-ing parasite protease was then aligned to the reference pro-teases and the coordinates were assigned within the SCRregions. Loops or variable regions (VR), which are locatedbetween SCRs, were found exclusively at the surface of theproteases, not interacting with the active site. Coordinatesfor the VRs were either directly generated or assigned fromknown crystal structures. The conformation of side chainswere retained in conserved positions and the statisticallymost likely rotamer (rotational position of side chain basedon analysis of all known protein structures containing thatamino acid) was chosen when no conformational informa-tion was available. The final structures were refined byenergy minimization using the AMBER potential function.The quality of the models was validated with QPACK (Gre-goret and Cohen 1990), VADAR (University of Alberta,Protein Engineering Network of Centres of Excellence), andthe 3D profile method (Luthyet al. 1992).

All color figures shown are generated using Midas Plus.Primary sequence alignments were performed using thesoftware package GCG (Genetics Computer Group Inc.,Madison, WI). Three-dimensional structures of inhibitorswere generated with Sybyl and the CONCORD algorithm(Tripos Inc., St. Louis, MO). Partial charges were calculatedusing the Gasteiger–Masili method within Sybyl. Searchingthe Available Chemicals Directory (ACD) for potential pro-tease inhibitor leads was carried out using DOCK3.5 inSEARCH mode (contact and force field scores). Com-pounds of high interest were ‘‘redocked’’ using theSINGLE mode function. (For further details of the DOCKapproach and the program see Kuntz 1992; Kuntzet al.1994; and http://www.cmpharm.ucsf.edu/kuntz/dock.html).All computer-assisted modeling and docking was performedon Silicon Graphics Workstation (IRIS4D/35 or Indigo2).

Enzyme Assays

The nativeL. major cpB was a generous gift of Dr.Jacques Bouvier (Ciba–Geigy, CH-1566 St. Aubin, Swit-zerland). Its purity was confirmed by silver-stained SDS-

L. major: MOLECULAR MODELING AND DRUG DESIGN 213

PAGE. Papain [EC 3.4.22.2] and mammalian cathepsin B(bovine spleen) [EC 3.4.22.1] were purchased from Sigma.Recombinant cruzain was produced as previously described(Eakin et al. 1992, 1993).

All proteases were assayed at 25°C using an automatedmicrotiter plate spectrofluorometer (Labsystem FluoroscanII). Activity was detected by the liberation of 7-amino-4-methyl coumarin (AMC) (Knight 1995a) (excitation wave-length 4 355 nm and emission wavelength4 460 nm)from the synthetic peptide substrate Z-Phe-Arg-AMC orZ-Arg-Arg-AMC (Z 4 benzoyloxycarbonyl; Phe4 phe-nylalanine; Arg4 arginine) (Enzyme Systems Products,Livermore, CA). The enzyme concentrations were deter-mined by active site titration (Barrettet al. 1982; Knight1995b). Inhibitors (20-mM stock solutions, dissolved in di-methyl sulfoxide (DMSO), stored at −20°C) at various con-centrations were preincubated with the respective enzymefor 5 min before the reaction was started by adding thesubstrate. Enzyme activities were expressed as a percentageof residual activity compared to an uninhibited control andplotted versus increasing inhibitor concentrations in order tocalculate the IC50 values.

Assay Conditions

L. major cpB.100 mM Na acetate, pH 5.5, 10 mM di-thiothreitol (DTT), 1 mM EDTA, 0.1% Triton X-100, 50mM Z-Phe-Arg-AMC final concentration (from a 10-mMstock solution in DMSO)Km 4 7 mM.

Papain and mammalian Cathepsin B.100 mM Na ac-etate, pH 5.5, 10 mM DTT, 100mM Z-Phe-Arg-AMC finalconcentration,Km 4 50 and 110mM, respectively.

Cruzain.The assay conditions were the same as for pa-pain except the substrate concentration was 20mM, Km 4

1 mM. Km values were determined by nonlinear regressionusing the software Ultrafit (Biosoft Inc., Ferguson, MO).

Cell Culture

L. major promastigotes LV39(MRHO/SU/59/P) weregrown at 27°C in RPMI-1640 containing 10% (v/v) heat-

inactivated fetal bovine serum and 20% Brain Heart Infu-sion Tryptose. Cell growth was determined by counting theparasites with a neubauer hemocytometer. Inhibitors dis-solved in DMSO were added from a 20-mM stock solution.DMSO concentrations up to 0.5% showed no effect on theparasite.

RESULTS AND DISCUSSION

The structure of the active site clefts of cys-teine proteases from the papain family arehighly conserved. The cleavage site or catalytictriad of all cysteine proteases consists of a nu-cleophilic cysteine, a histidine, and an aspara-gine. However, the S-subsites show striking dif-ferences (Table I). These sites are responsiblefor the substrate specificity (Schechter andBerger 1967; Barrett and Kirschke 1981; Storerand Menard 1994, 1996). The S9-subsites, whileconserved, show an obvious difference in cpLsand cpBs. Cathepsin B-like proteases have anoccluding loop which contains the two histi-dines (His110 and His111 in cathepsin B humanliver; His101 and His 102 inL. major cpB,shown in Table I) that are required for the exo-peptidase activity (dipeptidyl carboxypeptidaseactivity) of cathepsin B (Aronson and Barrett1978; Turket al. 1995; Illy et al. 1997). Thepolypeptide chain of cysteine proteases fromthe papain family is folded into two domains,between which the V-shaped active-site cleft isformed (Musilet al. 1991). Although the inter-domain interface has a similar shape in papainand in cathepsin B, the amino acid residues in-volved are different. Specific residues of the

TABLE IStructural Alignment of Active Site Residues

Protease

S Subsites

Catalytic Triad

S9 Subsites

S2 S1 S19 S29

Papain W69 S205 F207 Y67 P68 V133 A160 C25 H159 N175 Q19 W177 — —Cruzain N69 E208 S210 L67 M68 A138 G163 C25 H162 N182 Q19 W184 — —Cathepsin B

(human) A77 E245 V247 Y75 P76 A173 A200 C29 H199 N219 Q23 W221 H110 H111

L. major cpL L69 Y209 V211 L67 M68 A139 G164 C25 H163 N183 Q19 W185 — —L. major cpB T75 G234 V236 I73 P74 T162 A189 C29 H188 N208 Q23 W210 H101 H102

Note.Key residues of the active site clefts of papain family cysteine proteases are shown. The top three proteases servedas reference proteins to build the homology models of the twoLeishmaniaproteases. The catalytic triad and the S9 subsitesare highly conserved whereas the S subsites show differences responsible for differences in substrate specificity. Note thetwo histidines in the S29 subsite, appearing only in cathepsin B andL. major cpB.

SELZER ET AL.214

domain interface of human liver cathepsin B areGlu36, Ser39, Glu171, Arg202, and Ala218(Musil et al. 1991). The occluding loop and theinterface residues are structural features ofmembers of the cathepsin B subfamily. An oc-cluding loop with the two histidines appears inL. major cpB and the residues within the inter-face are identical to human liver cathepsin Bwith only one minor exchange of Arg202 toLys. With these structural features theL. majorcpB is clearly a member of the cathepsin Bsubfamily. Consistent with this conclusion isthe fact that the primary sequence ofL. majorcpB has high overall identity (54%) to humanliver cathepsin B but a much lower overall iden-tity to papain (30%) or other cathepsin L-likeproteases. The sequence identity of theL. majorcpB (Sakanariet al. 1997) to theL. mexicanacpB, (Bartet al. 1995) is 80% consistent withthe close relationship of these two species.

A biochemical property of cathepsin B-likeproteases is their ability to cleave synthetic sub-strates with Arg in position P2 (e.g., Z-Arg-Arg-AMC). Cathepsin L-like proteases have astricter preference for substrates with a Phe inposition P2 (e.g., Z-Phe-Arg-AMC) (Khourietal. 1991; Storer and Menard 1996). In the assaysystem used in this study bovine spleen cathep-sin B cleaves both substrates at about the samerate, whereas papain has about 10% activity to-ward Z-Arg-Arg-AMC compared to 100% to-ward Z-Phe-Arg-AMC (data not shown). This isconsistent with previous data indicating that ca-thepsin B prefers Phe over Arg in position P2

only 3.6-fold while papain favors it by a factorof 904 (Storer and Menard 1996). Glutamic acid205 (papain numbering, Table I) is responsiblefor the ability to cleave substrates with an Argin position P2. In a double mutant of papain(Val133 → Ala/Ser205 → Glu), exchanging ser-ine 205 for glutamic acid by site-directed mu-tagenesis enhanced the ability of papain tocleave Z-Arg-Arg-AMC substrates by about100-fold (Khouri et al. 1991). Recently thecrystal structure of cathepsin B and a peptideinhibitor with an Arg in position P2 finallyproved this theory (Jiaet al.1995). Position 205is also a glutamic acid in cruzain which has the

ability to cleave Z-Arg-Arg-AMC but still pre-fers Z-Phe-Arg-AMC at acidic pH (data notshown).L. majorcpB shows no activity towardZ-Arg-Arg-AMC but is active against Z-Phe-Arg-AMC. Its counterpart fromL. mexicanaprefers Phe over Arg in Position P2 but stillshows 10% activity toward Z-Arg-Arg-AMC(Robertson and Coombs 1993). This differencein the substrate specificity is reflected in posi-tion 205. TheL. mexicanacpB has a serine, likepapain, at this position, whereasL. major cpBhas a glycine (Table I). Replacement of glu-tamic acid 205 by a glycine inL. major cpBresults in the S2 site having a much larger andmore hydrophobic pocket (Fig. 1, Fig. 2). Thus,while L. major cpB belongs to the cathepsin Bsubfamily by virtue of its structural homology,its activity is more ‘‘cathepsin L-like’’. Thisobservation confirms that there is a need foranalyzing sequence, structure, and enzymaticdata before concluding an enzyme is a memberof a specific protease subfamily. It also under-scores the potential functional diversity of thecathepsin B proteases as evidenced by the abil-ity to significantly alter substrate specificity bya single residue change in S2.

The structural models of theL. major cpBand cpL were used to search the ACD for po-tential protease inhibitors. ACD is a database ofabout 150,000 commercially available com-pounds formerly known as the Fine ChemicalsDirectory distributed by Molecular Design Lim-ited Information System (San Leandro, CA).This approach was performed with the softwareDOCK3.5 (University of California, San Fran-cisco) (Kuntz 1992; Kuntzet al. 1994). Thecomputational software DOCK3.5 is a suite ofprograms for locating feasible binding orienta-tions, given the structure of a ligand moleculeand a receptor molecule. In SEARCH mode,orientations are generated for each of the scor-ing molecules in a database, then the best-scoring orientation of each molecule is saved,and the best-scoring molecules are saved to afile. Two scoring lists were generated, a contactscore and a force field interaction energy score.The top 3% for each scoring method was savedand visually examined for size, packing, and

L. major: MOLECULAR MODELING AND DRUG DESIGN 215

interactions within the active site. In order tofind new lead inhibitors, the most promisingcompounds (about 0.05%) were selected to betestedin vitro. Out of 150,000 compounds 4500were saved for each scoring method (force fieldand contact score). Following visual examina-tion of fit, 69 compounds were finally selectedfor testing. Of these 69 compounds, 43 werefrom the contact list, 26 were from the forcefield list, and 15 compounds appeared on bothlists. Forty-five of the 69 cpB compounds alsoappeared within the top 3% lists (contact andforce field score) from theL. majorcpL screen.

Because theL. major cpB is available in re-agent quantities it was chosen for biochemicalscreens. Eighteen of the chosen compoundsshowed an IC50 between 50 and 100mM and 3(PS44, reactive orange, 16, 2-((4-(7-acetamido-1-hydroxy-3-sulfo-2-naphthylazo)phenyl)sulfo-nyl)-ethyl sulfate, contact list; PS50, succin-imidyl 4-(p-maleimidophenyl)butyrate, bothlists; PS28, 3,5 dichlorofolic acid,L-glutamicacid, N-[4[[(2-amino-1,4-dihydro-4-oxo-6-pteridinyl])methyl]amino]-3,5-dichlorobenzoyl,

both lists) showed an IC50 below 50mM (TableII, Fig. 3). Thirteen of the 18 compounds in-cluding PS28, PS44, and PS50 were also foundon the lists for theL. major cpL. PS50 is likelyan irreversible inactivator of the protease. Itsmaleimide group binds specifically to free sul-fur groups (Smythet al.1964). InL. majorcpBthe highly reactive sulfur in the cysteine of thecatalytic triad is a likely target. This would leadto a covalent bond between the maleimidegroup and theL. major cpB active site. Indeedinhibition by PS50 could not be competed byexcess amounts of substrate consistent with ir-reversible inhibition.

PS44 contains three sulfate groups whichmake it a highly negatively charged compoundand difficult to derivatize. Most likely these hy-drophilic groups are responsible for the inhibi-tory activity. However, since the S-subsites ofthe target proteases are rather hydrophobic,PS44 was not further considered as a lead com-pound.

The most promising lead compound is PS28.PS28 is a derivative of folic acid and therefore

FIG. 1. Homology-based space filling protein models of theL. majorcpL (a) and cpB (b). The active site cleftis formed between the two main domains. The S subsites (S) and the S9 subsites (S9) are shown in cyan, thecatalytic triad (T) in yellow. The two histidines of the occluding loop are shown in red.

SELZER ET AL.216

belongs to a well-studied group of compounds,some of which are drugs. Dihydrofolic acid is asubstrate of the dihydrofolic acid reductase(DHFR), itself a drug target (Schweitzeret al.

1990). Methotrexate and aminopterin, used inanticancer treatment, are highly active inhibi-tors of DHFR (Schweitzeret al. 1990). Theselatter 2 compounds were among the 18 com-pounds showing IC50’s between 50 and 100mM. While PS28 was a byproduct of anticancerdrug development (Martinelli and Chaykovsky1980), until now no enzyme inhibition activity,or biological activity, had been reported for thiscompound.

The three-dimensional structure of PS28 wasmodeled with Sybyl and docked into the activesite ofL. majorcpB by using the SINGLE modefunction of DOCK3.5. In SINGLE modeDOCK3.5 generates many orientations of oneligand for both contact scoring and force fieldscoring. These orientations can be examined forinteractions of the ligand and its receptor to de-termine the most likely orientation. A high per-centage of the orientations (contact and forcefield score) generated by DOCK3.5 for PS28andL. major cpB showed the pteroic acid half

TABLE IIInhibition of Cysteine Proteases

Compound

IC50 mM

L. majorcpB Cruzain Papain

MammalianCat B

PS28 10 >50 20 20PS44 10 >50 20 20PS50 5 >50 20 40

ZLIII115A 10 10 >50 20ZLIII43A 2 5 10 10ZLIII133A 0.5 0.6 50 20

Note. Inhibitory activity of first (PS)- and second-generation compounds (ZL) toward different cysteine pro-teases. Note differences in activity of parasite versus non-parasite proteases from 2- to 100-fold. IC50 determinedfrom plot of five or more assays at specific inhibitor con-centrations.

FIG. 2. The main residues of the active site cleft of theL. major cpB protease and a putative bindingorientation generated by DOCK3.5 of the lead compound PS28. Numbers refer to theL. major cpB maturesequence. Residues are colored: S, subsites in magenta, catalytic triad in yellow; S9, subsites in cyan andbackbone in gray. The ligand is colored: carbons in green, nitrogens in blue, oxygens in red, and chlorines ingray. Note the position of the hydrophilic glutamic acid group of the ligand close to the hydrophobic S subsites.

L. major: MOLECULAR MODELING AND DRUG DESIGN 217

of the compound close to the catalytic triad andthe S9-site and at least one of the chlorideswould interact with the catalytic triad (Fig. 2).The hydrophilic glutamic acid was predicted aspointing into either the hydrophobic S1-site orthe wide open hydrophobic S2-site, both unfa-vored interactions. This suggests that com-pounds having a more hydrophobic moiety maybe better inhibitors. The synthesis of suchchemical derivatives is now a goal of our proj-ect. The carboxylic acids of the glutamic acidwill be eliminated, and the glutamic acid will bereplaced by phenylalanine or homophenylala-nine.

A second approach to inhibitor discovery anddesign for theL. major cysteine proteases wasto test a library of compounds, synthesized bycombinatorial chemistry, and based on an origi-nal lead compound [oxalic bis(2-hydroxy-1-naphthylmethylene)hydrazide] found by DOCKusing the malaria cysteine protease falcipain asa target (Ringet al.1993). Of special note is thefact that the hydrazide lead was also among thecompounds selected by computer search ofACD for the L. major cpB and cpL. From thislead a second generation of compounds hadbeen synthesized (ZL-compounds, Fig. 3) andtested against falcipain as well as against the

malaria parasite in cell culture. Some of thesecompounds proved to be potent inhibitors ofboth the protease and parasite growth in culture(Li et al.1994, 1995, 1996). Although the over-all sequence identity of falcipain and theL. ma-jor cpB is only 31%, key regions of substratebinding are conserved and several of these in-hibitors had IC50’s in the nanomolar range ver-sus theL. major cpB (Table II). The enhancedinhibitory activity of these second-generationcompounds is the result of specific syntheticmodifications based on computer predictionsmade with DOCK. For theL. major cpB, theseresults again emphasize that while its overallhomology is to cathepsin B, it shares substrate-binding similarities with cathepsin L-like en-zymes like falcipain.

The three hydrazides (ZLIII43A, ZLIII115A,and ZLIII133A) were also tested inL. majorcell cultures. Inhibitors were added to replicat-ing promastigotes (106 cells ml−1) as a singledose and cell growth was monitored over 3days. All three compounds showed very similareffects on the replication rate of the parasite.Concentrations of 5mM led to about half maxi-mal growth whereas 20 and 50mM totally in-hibited cell growth (Fig. 4). Exchanging the me-dia every day for a total of 3 days, thereby keep-

FIG. 3. Chemical structures of first (PS28)- and second-generation (ZL’s) compounds.

SELZER ET AL.218

ing inhibitor concentrations stable (20 and 50mM), led to death of all the parasites. After thefourth day the media was exchanged with freshmedia without inhibitor and the flasks were keptunder culture conditions. Even after 10 days noparasites could be detected, indicating a totalcure of theLeishmaniaculture by the cysteineprotease inhibitors. ZLIII43A and ZLIII115A at40 mM had no effect on the growth or appear-ance of J774 cells, a mammalian macrophagecell line. Evaluation of these compounds on in-tracellular L. major amastigotes and in micewill be reported in another paper (Selzer, Pin-gel, Hsieh, Chan, Engel, Sakanari, and McKer-row, submitted for publication).

The results of these cell culture assays sug-gest that the cysteine proteases ofL. major arecrucial to the parasite. Mottram and co-workersreported that a cpL null mutant ofL. mexicanashowed an 80% decrease in virulence but wasstill able to grow (Mottramet al.1996).L. mexi-canahas multiple cysteine proteases of both thecpL and cpB type (Robertsonet al.1996). Sincethe inhibitors we tested inhibit both cpLs and

cpBs, they may target both types of proteaseswithin the parasite, overcoming the redundancyin activity suggested by the null mutant studies.Because the inhibitors are reversible, it was notpossible to identify specific protease targets byinhibitor labeling. The lack of toxicity to mam-malian cells at the inhibitor concentrationswhich kill parasites may reflect a greater prote-ase redundancy in mammalian lysosomes (Ma-son 1991; Kirschkeet al. 1995) or differentialuptake of inhibitor by parasites versus host cells(McGrathet al. 1995).

ACKNOWLEDGMENTS

The authors thank Christopher Franklin for excellenttechnical assistance. This investigation received financialsupport from the UNDP/WORLD BANK/WHO SpecialProgramme for Research and Training in Tropical Diseases(TDR) (Grant No. T21/181/29 to J.A.S. and No.M20/181/232 to F.E.C.) and the National Institutes ofHealth (AI35707-1) to J.H.M. The Midas Plus programfrom the Computer Graphics Laboratory, University ofCalifornia, San Francisco, was supported by the NationalInstitutes of Health (RR-01081). James H. McKerrow issupported by a Burroughs Wellcome Molecular Parasitol-ogy Scholar Award. Paul M. Selzer is supported by a fel-lowship of the Deutsche Forschungsgemeinschaft (Se762/1-1).

REFERENCES

Aronson, N. N., and Barrett, A. J. 1978. The specificity ofcathepsin b. Hydrolysis of glucagon at the C-terminus bya peptidyldipeptidase mechanism.Biochemical Journal171,759–765.

Barrett, A. J., Kembhavi, A. A., Brown, M. A., Kirschke,H., Knight, G. C., Tamai, M., and Hanada, K. 1982. L-trans-epoxysuccinyl-leucylamido(4-guanidino)butane (E-64) and its analogues as inhibitors of cysteine proteinasesincluding cathepsins B, H and L.Biochemical Journal201,189–198.

Barrett, A. J., and Kirschke, H. 1981. Cathepsin B, cathep-sin H, and cathepsin L.Methods in Enzymology80,535–561.

Bart, G., Cooms, G. H., and Mottram, J. C. 1995. Isolationof lmcpc, a gene encoding aLeishmania mexicanaca-thepsin B-like cysteine proteinase.Molecular and Bio-chemical Parasitology73, 271–274.

Bernstein, F. C., Kotzle, T. F., Williams, G. J., Meyer, E. F.,Brice, M. D., Rodgers, J. R., Kennard, O., Shimanouchi,T., and Tasumi, M. 1978. The protein data bank: A com-puter-based archival file for macromolecular structures.Archives of Biochemistry and Biophysics185,584–591.

Coombs, G. H., and Baxter, J. 1984. Inhibition ofLeish-

FIG. 4. Effect of ZLIII43A on the growth ofL. majorpromastigotes. The compound was added at time point 0and cell growth was monitored over 3 days. ZLIII115A andZLIII33A showed very similar inhibition profiles. (d) con-trol; (n), 5 mM; (s), 20mM; (L), 50mM. Points are meansof three independent experiments.

L. major: MOLECULAR MODELING AND DRUG DESIGN 219

maniaamastigote growth by antipain and leupeptin.An-nals of Tropical Medicine and Parasitology78, 21–24.

Eakin, A. E., McGrath, M. E., McKerrow, J. H., Fletterick,R. J., and Craik, C. S. 1993. Production of crystallizablecruzain, the major cysteine protease fromTrypanosomacruzi. Journal of Biological Chemistry268,6115–6118.

Eakin, A. E., Mills, A. A., Harth, G., McKerrow, J. H., andCraik, C. S. 1992. The sequence, organization and ex-pression of the major cysteine protease (cruzain) fromTrypanosoma cruzi. Journal of Biological Chemistry267,7411–7420.

Ferrin, T. E., Huang, L. E., Jarvis, L. E., and Langridge, R.1988. The MIDAS display system.Journal of MolecularGraphics6, 13–27.

Gregoret, L. M., and Cohen, F. E. 1990. Novel method forthe rapid evaluation of packing in protein structures.Journal of Molecular Biology211,959–974.

Grogl, M., Thomason, T. N., and Franke, E. D. 1992. Drugresistance in leishmaniasis: its implication in systemicchemotherapy of cutaneous and mucocutaneous diseases.American Journal of Tropical Medicine and Hygiene47,117–126.

Huang, C. C., Petterson, T. E., Klein, T. E., Ferrin, T. E.,and Langridge, R. 1991. Conic: A fast renderer for space-filling molecules with shadows.Journal of MolecularGraphics9, 230–236.

Illy, C., Quraishi, O., Wang, J., Purisima, E., and Vernet, T.1997. Role of the occluding loop in cathepsin B activity.The Journal of the Biological Chemistry272,1197–1202.

Jia, Z., Hasnain, S., Hirama, T., Lee, X., Mort, J. S., To, R.,and Huber, C. J. 1995. Crystal structures of recombinantrat cathepsin B and a cathepsin B-inhibitor complex.Journal of Biological Chemistry270,5527–5533.

Kamphuis, I. G., Kalk, K. H., Swarte, M. B. A., and Drenth,J. 1984. Structure of papain refined at 1.65 angstroms.Journal of Molecular Biology179,233–256.

Khouri, H. E., Vernet, T., Menard, R., Parlati, F.,Laflamme, P., Tessier, D. C., Gour-Salin, B., Thomas,D. Y., and Storer, A. C. 1991. Engineering of papain:selective alteration of substrate specificity by site-directed mutagenesis.Biochemistry30, 8929–8936.

Kirschke, H., Barrett, A. J., and Rawlings, N. D. 1995. Pro-teinases 1: lysosomal cysteine proteinases.Protein Pro-file 2, 1581–1643.

Knight, C. G. 1995a. Fluorimetric assays of proteolytic en-zymes.Methods in Enzymology248,18–34.

Knight, C. G. 1995b. Active-site titration of peptidases.Methods in Enzymology248,85–101.

Kuntz, I. D. 1992. Structure-based strategies for drug designand discovery.Science257,1078–1082.

Kuntz, I. D., Meng, E. C., and Shoichet, B. K. 1994. Struc-ture-based molecular design.Accounts of Chemical Re-search27, 117–123.

Li, R., Chen, X., Gong, B., Selzer, P. M., Li, Z., Davidson,E., Kurzbahn, G., Miller, R. E., Nuzum, E. O., McKer-row, J. H., Fletterick, R. J., Gillmore, S. A., Craik, C. S.,

Kuntz, I. D., Cohen, F. E., and Kenyon, I. D. 1996. Struc-ture-based design of parasitic protease inhibitors.Bioor-ganic & Medicinal Chemistry4, 1421–1427.

Li, R., Kenyon, L., Cohen, F. E., Chen, X., Gong, B.,Dominguez, J. N., Davidson, E., Kurzbahn, G., Miller,R. E., Nuzum, E. O., Rosenthal, P. J., and McKerrow,J. H. 1995. In vitro antimalaria activity of chalcones andtheir derivatives.Journal of Medicinal Chemistry38,5031–5037.

Li, Z., Chen, X., Davidson, E., Zwang, O., Mendis, C.,Ring, C. S., Roush, W. R., Fegley, G., Li, R., Rosenthal,P. J., Lee, G. K., Kenyon, G. L., Kuntz, I. D., and Cohen,F. E. 1994. Anti-malarial drug development using modelsof enzyme structure.Chemistry & Biology1, 31–37.

Luthy, R., Bowie, J. U., and Eisenberg, D. 1992. Assess-ment of protein models with three-dimensional profiles.Nature356,83–85.

Martinelli, J. E., and Chaykovsky, M. 1980. Aromatic chlo-rination of p-aminobenzoic acid derivatives. Improvedsynthesis of mono- and dichloromethotrexate.Journal ofOrganic Chemistry45, 527–529.

Mason, R. W. 1991. Proteinases of mammals: an overview.In ‘‘Biochemical Protozoology’’ (G. H. Coombs andM. J. North, Eds.), pp. 168–179. Taylor & Francis, Lon-don, New York, Philadelphia.

McGrath, M. E., Eakin, A. E., Engel, J. C., McKerrow,J. H., Craik, C. S., and Fletterick, R. J. 1995. The crystalstructure of cruzain: a therapeutic target for Chagas’ dis-ease.Journal of Molecular Biology247,251–259.

McKerrow, J. H., Sun, E., Rosenthal, P. J., and Bouvier, J.1993. The proteases and pathogenicity of parasitic pro-tozoa.Annual Review of Microbiology47, 821–853.

Moran, J. S., and Bernard, K. W. 1989. The spread of chlo-roquine-resistant malaria in Africa.Journal of the Ameri-can Association of Medicine262,245–248.

Mosimann, S., Meleshko, R., and James, N. G. 1995. Acritical assessment of comparative molecular modeling oftertiary structures of proteins.Proteins: Structure, Func-tions and Genetics23, 301–317.

Mottram, J. C., Souza, A. S., Hutchinson, J. E., Carter, R.,Frame, M. J., and Coombs, G. H. 1996. Evidence fromdisruption of thelmcpbgene array ofLeishmania mexi-canathat cysteine proteinases are virulence factors.Pro-ceedings of the Academy of Science USA93,6008–6013.

Musil, D., Zucic, D., Turk, D., Engh, R. A., Mayr, I., Huber,R., Popovic, T., Turk, V., Towatari, T., Katunuma, N.,and Bode, W. 1991. The refined 2.15 angstroms x-raycrystal structure of human liver cathepsin B: the struc-tural basis for its specificity.EMBO Journal10, 2321–2330.

Ring, C. S., Sun, E., McKerrow, J. H., Lee, G. K., Rosen-thal, P. J., Kuntz, I. D., and Cohen, F. E. 1993. Structure-based inhibitor design using protein models for the de-velopment of anti-parasitic agents.Proceedings of theNational Academy of Sciences of the United States ofAmerica90, 3583–3587.

SELZER ET AL.220

Robertson, C. D., and Coombs, G. H. 1993. Cathepsin B-like cysteine proteases ofLeishmania mexicana. Molecu-lar and Biochemical Parasitology62, 271–280.

Robertson, C. D., Coombs, G. H., North, M. J., and Mot-tram, J. C. 1996. Parasite cysteine proteases.Perspectivesin Drug Discovery and Design6, 99–118.

Sakanari, J. A., Nadler, S. A., Chan, V. J., Engel, J. C.,Leptak, C., and Bouvier, J. 1997.Leishmania major:Comparison of the cathepsin L- and B-like cysteine pro-tease genes with those of other trypanosomatides.Experi-mental Parasitology85, 63–76.

Sakanari, J. A., Selzer, P. M., Chan, V., Engel, J., and Bou-vier, J. 1995. Parasite proteases as targets for the struc-ture-based approach of drug design.Acta Medica Philip-pina 31, 188–197.

Schechter, I., and Berger, A. 1967. On the size of the activesite in proteases. I. Papain.Biochemical and BiophysicalResearch Communications27, 157–162.

Schweitzer, B. I., Dicker, A. P., and Bertino, J. R. 1990.

Dihydrofolate reductase as a therapeutic target.TheFASEB Journal4, 2441–2452.

Smyth, D. G., Blumenfeld, O. O., and Konigsberg, W.1964. Reaction of N-ethylmaleimide with peptides andamino acids.Biochemical Journal91, 589–595.

Storer, A. C., and Menard, R. 1994. Catalytic mechanism inpapain family of cysteine peptidases.Methods in Enzy-mology244,486–501.

Storer, A. C., and Menard, R. 1996. Recent insights intocysteine protease specificity: lessons for drug design.Perspectives in Drug Discovery and Design6, 33–46.

Turk, D., Podobnik, T., Popovic, T., Katunuma, N., Bode,W., Huber, R., and Turk, V. 1995. Crystal structure ofcathepsin B inhibited with CA030 at 2.0-A resolution: Abasis for the design of specific epoxysuccinyl inhibitors.Biochemistry34, 4791–4797.

Received 28 March 1997; accepted with revision 18 August1997

L. major: MOLECULAR MODELING AND DRUG DESIGN 221