J. Med. Chem. 1993,36, 3113-3119
3113
A Series of Penicillin-Derived C 2 - S y m m e t r i c Inhibitors of HIV- 1 Proteinase: Structural and Modeling Studies Alan Wonacott3 Robert Cooke3 Fiona R. Hayes,#Michael M. Hann,s Harren Jhoti,t Peter McMeekin,ll Ani1 Mistry,*Peter Murray-Rust,? Onkar M. P. Singh,t and Malcolm P. Weir’J Protein Structure Group, Computational Chemistry, and Departments of Protein Biochemistry and Medicinal Chemistry, Glaxo Group Research Limited, Greenford, Middlesex UB6 OHE, United Kingdom Received April 6, 1993.
The binding modes of a series of penicillin-derived Cz symmetric dimer inhibitors of HIV-1 proteinase were investigated by NMR, protein crystallography, and molecular modeling. The compounds were found to bind in a symmetrical fashion, tracing an S-shaped course through the active site, with good hydrophobic interactions in the S I I S ~and SdS2‘ pockets and hydrogen bonding of inhibitor amide groups. Interactions with the catalytic aspartates appeared poor and the protein conformation was very similar to that seen in complexes with peptidomimetics, in spite of the major differences in ligand structure.
Introduction Since its identification as an essential protein in the human immunodeficiency virus (HIV) replication cyc1e.l HIV-1proteinase (HIVP) has been the subject of intensive investigationas a target for chemotherapy against acquired immunodeficiencysyndrome (AIDS). A variety of potent and selective peptidomimetic inhibitors have been synthesized, including pseudo-Cz-symmetric compounds,2 designed to exploit the inherent CZsymmetry of the HIVP homodimer.3 The discovery of Cz-symmetric penicillinderived inhibitors has been previously reported4 and this series is further described in an accompanyingpaper? Here we report the X-ray structure determination and modeling studies of HIVP/inhibitor complexes which have led to a greater understanding of the binding mode of these inhibitors and to the synthesis of novel compounds.
Results and Discussion Initial molecular modeling of the interaction of the lead compound 1 (Table I) with the protein was based on the structure of HIVP complexed with a peptide-derived inhibitor.6 Three plausible binding modes emerged (Figure 1):two “symmetric”modes (A and B)in which the CZ axis of 1 is coincident with the virtual/pseudo-CZaxis of the proteinase dimer and an “asymmetric” mode (C).In mode A the thiazolidine rings occupy the SZ and Si pocketslo and the phenyl rings the SI and SI’ pockets (Figure l), with the functional group interactions reversed for the alternative symmetric mode (B).It was postulated that in both symmetric modes A and B the amide carbonyls of the central ethylenediamide “linker” would form hydrogen bonds to the buried water molecule which bridges between the flaps and the peptide-based inhibitor? with the amide protons from the linker hydrogen bonding to the carbonyls of residue Gly 27/27’ (Figure 1). In the asymmetric mode (C)the NH of one thiazolidine ring interacts with the catalytic aspartyl residues (Asp 25/25‘). The issue of symmetric versus asymmetric was resolved using solution-state NMR. The l9F NMR spectrum of
* Corresponding author.
Protein Structure Group. Protein Biochemistry Department. Computational Chemistry. I Department of Medicinal Chemistry. Abstract published in Advance ACS Abstracts, September 1,1993. t
Table I. Structure-Activity Relationships of Penicilli-Derived Inhibitors 0 R , J L H
1,
HiKR1 0 1 la 2
2a 3
3a 3b
CH2CHz CHzCH2 CH2CH(OH)CHz CH~CH~CHS CH2CHz CH2CHz CHzCHz
benzyl benzyl benzyl benzyl 2-phenylphenyl phenyl Me
ethyl CHzCFa benzyl benzyl ethyl ethyl ethyl
3.0 4.4 5.0 540
9.9 390 5.4
Ph
HIVP with compound la, containing two symmetrically disposed trifluoro groups, displayedtwo resonances (Figure 2). Based on relative linewidths and integrals of the peaks and the fact that their relative intensities changed upon addition of further compound, the downfield resonance was assigned to free inhibitor, and the upfield resonance to the complexed form. From the presence of only one resonance for the complex,given the strong chemical shift sensitivity of ‘9F to its environment: symmetric binding was strongly suggested. A proteinase complex with compound 1in which the linker was labeled with 15Nwas also studied by lH NMR using a 15Nfilter; analysis of the data (not shown) gave similar results. The crystal structure determination of 1complexed with HIVP established the predicted mode B symmetricbinding as the actual mode (Figure 3) and revealed an overall protein conformation similar to that seen in other published structures of HIVP which have peptide inhibitors bound, with the “flaps” closed down on the inhibitor and bridged by the water molecule W301.%6fi*g Crystallographic data for this and other structures are summarized in Table 11. Figure 3 shows an overlay of the protein Ca backbone
0022-2623/93/1836-3113$04.00/00 1993 American Chemical Society
3114 Journal of Medicinal Chemistry, 1993, Vol. 36, No.21 Symmetric mode (A)
=2
S1'
i s2
Symmetric mode (5)
s3
Asymmetric mode (C)
% s2 Asp25
. . Asp25'
s3
Figure 1. Schematic representation of proposed binding modes of inhibitor 1to HIV-1 protease derived by molecular modeling. NMR and crystallographyshowed symmetric mode B to be the correct overall mode.
/---
Figure2. W-NMRspectrumof inhibitor la complexed to HNP. The single bound resonance, broadened and shifted -0.35 ppm relative to unbound inhibitor,is consistentwith symmetricrather than asymmetric binding.
of the HIVPIl complex with the hydroxy isostere (JG365) complex;" the root mean-square deviation for all 99 Car
Wonacott et al.
atoms is only 0.55 A, indicative of a very similar overall conformation for the protein. The hydrogen-bonded contacts of inhibitor 1 (Figure 4) are with the same set of residues seen in HNP/peptidomimetic s t ~ c t u r e s ; ~ 1the ~3*~ phenylacetate and thiazolidine groups of the inhibitor fit into SdS2' and S&', respectively,1°with no major sidechain adjustments. In spite of the S-shaped course traced by 1 through the active site compared to an extended conformation for peptide mimetics (Figure 3b), HIVP apparantly binds such disparate molecules with no significant 3D structural alteration of the enzyme. The thiazolidine of 1 is close-packed in S1/S1' (Figure 5 ) with good van der Waals contacts and burial of the hydrophobic surface;the gem-dimethyls occupy the same space as the terminal nor-Leu atoms of the reduced amide isostere MVT-1016 although they approach from a different direction. The position of the thiazolidine ring is essentially the same in allthree compounds described here (Figure 6), suggesting it is a dominant interaction. The phenylacetyl side chain of 1 bends back into SdS2' and is close-packed into this pocket, consistent with known structure-activity relationships: which show reduced binding for para-substituted phenylacetyl and cyclohexylacetyl 6-substituents and poor binding for small substituents (none,acetyl) which do not fill the binding pocket well. The ethylenediamide "linker" of 1 apparently makes only weak electrostatic and van der Waals interactions with the protein (see Figure 4) although W301 is welllocated with near-perfect tetrahedral geometry;this water is present and superimposable in all three structures. Hydrogen bonding andlor electrostatic interactions of the NH groups of the ethylenediamide must be energetically weak as methylation results in only a small loss of potency." The lack of any explicit interactions of 1to the catalytic aspartah encouraged us to make the hydroxypropylamide compound 2, which was found to bind with comparable affmity to 1 but 100-fold more tightly than the cognate propylamide analogue 28, indicating that the hydroxyl/ Asp 251Asp 25/ interaction is worth 2-3 kcal/mol. The crystal structure of 2 shows it to bind in a similar fashion to 1 in spite of the longer linker (Figure 6); the hydroxypropyl linker twists to allow inward rotation of the linker amide carbonyls, which therefore maintain good hydrogen bonding to W301. The catalytic Asp 251Asp 25' pair twist further out of plane as the hydroxyl of 2 interacts with only one carboxyl group (heavy atom distances of 2.731 2.97 A to the oxygens of one Asp and 3.96/3.18 A to the other) while the central carbon of the propyl linker sits on the local COaxis. This contrasts with hydroxy isostere structures in which the hydroxyl causes the Asp 251Asp 25/pair to become coplanar.8 When comparing the binding of compounds 1 and 2a the extended linker compound binds 100-fold weaker; an equivalent affinity is regained on introduction of the hydroxy into the propylamide. The implication of this is that the S1/S1' and SdS2' occupancy is not as optimal with the longer linker but can be compensated for by the hydroxy group, giving good polar interactions to the aspartates. Cellular potency of them and related compounds against HIV-1 was found to be highly variable, although there is a weak correlation of IC60 in the enzyme assay with lipophilicity. An empirically diverse set of compounds was therefore synthesized, variation of the penicillin-
Penicillin-Deriued Inhibitors of HIV-I Proteinase. 1
Journal of Medicinal Chemistry, 159.3, Vol. SS, No.21 3115
Figure 3. (a, top) Superposed Col backbone and ligand conformations of H I W complexes with inhibitor 1 (red) and with the peptidomimetic JG3135~(blue). 1 traces an S-shaped course through the active site compared to the extended peptide conformation. The protein conformations are very similar. (h, bottom) Close-up of ligand overlays from complexes shown in part a. Table 11. CrystallographicData for HIV-1 Protease Complexes compd space group unit cell dimemiom (A) 1 2 3
p61
m122 P21212
= 63.4, b = 83.2 (I = 63.2, b = 83.3 (I = 59.4, b = 01.5, c = 46.1 (I
derived side chain being employed to this end.5 The benzoyl derivative 3a showed poor activity, and modeling suggested this was because the ring could not reach into SdSz’. However the oxacillin-derived compound 3b showed good activity against free enzyme, although it had
resolution 2.6 2.8 2.5
,,R 0.127 0.093
0.112
Rmyl
0.216 0.216 0.110
poor whole-cell activity. Molecular modeling indicated that the phenyl substituent of the isoxazole would fit into SdSz’, and this observation led to the design of the orthophenylderivative 3 in which Sp/S~’washypothesized to be occupied in a similar fashion. This compound was
3116
Wonacott et ai.
Journal ofMedicina1 Chemistry, 2993, Vol. 36, No.21
s2
..^,_
IB 50'
s1
S2'
\
Figure4. Diagram of nonhonded interactionsof 1mmpleaed to HIVP taken from 2.5-Acrystal structure. Hydrogen bonds are shown as dotted lines with heavy atom distances marked (A). It is evident that SI and Se suhsites are filled with the thiazolidine and phenylacetate, respectively.
Figure 5. Connolly surface of protein SI and S, pockets showing close-packingof 1 into subsites, with the thiazolidine occupying SI
and the penylacetate side chain in S2.
found to be a potent inhibitor of HIVP and was also 100fold more active than 3b in its whole-cell activity, which may be a consequence of its increased lipophilicity. When the crystal structure of 3 was determined, the proposed mode of binding was confirmed. The additional "backbone" carbon atom of 3 causes some displacement of the substituted phenyl ring, although the contacts to the protein and space occupied are much the same as in 1and 2;the crucial carbonyl hydrogen bond to the NH of Asp 29 is also conserved concomitant with a change in the amide plane (Figures 4 and 6). The only significant change in the protein S d S i pocket is displacement associated with the change in torsion angle of the Asp 30/30' side chain. Figwellshows theelectrondensityoftheinhibitor; the relative rotation of 60° of the o-phenyl ring attached to the benzoyl ring is clearly evident. This series of penicillin-derived dimers is relatively tolerant of the @-lactamopening substituent;6 indeed, the ammonia-opened compound has an ICw of 3 nM, much more potent than some compounds with larger substituents, indicating that interactions at the edges of the active Evidence for this is seen in the site can be deleteriou~.~
bound-crystal structure of compound 2, in which the bulky phenyl ring displaces the guanidinium group of Arg 8 and therefore disrupts the conserved salt bridge between Arg 8 and Asp 29'. In compounds 1 and 3, the smaller ethyl group packs against the Arg 8 side chain, which is moved only slightly. Compounds 1 and 3 are Cz-symmetric and there is no evidence for asymmetry in the core region of protein or ligand in their complexes with HIVP (Figure 6). Compound 2 is pseudo-C2-symmetric and significant a s p metry is seen in the complex, possibly due to the influence of the achiral hydroxyl (Figure 6). These results contrast withrecentdataonsymmetricdiolinhibitorsBbwhichbind asymmetrically to HIVP due to dominant hydrogen bonding with Asp 25/25'. There is noevidence for differentconformations of active site side chains in different halves of the dimer in the complexes of 1, 2, or 3, unlike the acetyl-pepstatin complex? possibly due to the high symmetry of the compounds. Significant differencesare seenat the protein surface, due mainly to crystal packing, in the complex with3, which crystallizes in the orthorhombic space group
Penicillin-Deriued Inhibitors of HIV-1 F'roteimse. 1
Journal of Medicinal Chemistry, 1993, Vol. 36,No. 21 3117
Figure 6. Overlays of ligand conformations taken from superposed HNP complexes. Inhibitor 1 in green; 3 in red; 2 in light blue (stippled)and dark blue, representing the two alternative refined conformationsof this pseudo-Cr-symmetriccompound. The Q&' conformations of the three ligands are similar, although half of 2 is rotated relative to the other half, pushing the phenyl ring deeper into Sa and rotating the linker amide. Hydrogen bonds to W301 are shown.
p21212 and shows a root mean square deviation of 0.45 A for one monomer Ca atoms compared to ita partner. This compareswith0.33Afor 1, whichisinthemoresymmetric P61 (pseud0-P6~22)space group. The Gly 51-NH hydrogen bond to Ile 50' CO across the flaps is inherently asymmetric" and is in a unique orientation in complexes with 1 and 3; the 2 complex is in the true P6122 space group with the monomer in the asymmetric unit and therefore this hydrogen bond cannot be present in an ordered form. Conclusions The Cz-symmetricpenicillin-deriveddimers5wereshown by crystallography to bind symmetrically to HIV-1 proteinase in one of the symmetric modes outlined by molecular modeling, confuming the NMR observations. These compounds trace a novel S-shaped path through the active site, with poor contacts at the catalytic aspartates, but excellent occupancy of the S1IS1' and SdSz' subsites. The compounds are extensively hydrogen bonded and maintain the feature of water 301 that bridges between the flaps and peptide-based analogues. Good hydrophobic packing in SdSl'and SdSz'andmaintenance of the hydrogen-bonding network are dominant forces in the binding of these compounds, and bulky groups at the edges of the active site can be deleterious to binding. The structural information enabled u8 to rationalize the observed SAR for the series of compounds described in this and the accompanying paper," higher affinity compounds being designed from the details provided. The
unique binding mode of these novel HIVP inhibitors is remarkable in that all the same protein residues are involved as are observed in the binding of more substratelike compounds.2.6aa Further analysis of these observations may lead to improved design of inhibitors to this enzyme and to greater understanding of molecular recognition in general. Experimental Section Protein PuriIieationandCrystallization. Theenzymewas purified'* and immediately inhibited with the appropriate compound. The compounds were dissolved in 250 pL dimethyl sulfoxideand added to the protein to achieve molar ratio of 101 with respect to the enzyme dimer. The complex was dialyzed for 3 h against 50 mm MES buffer,pH 5.0, containing 1mMEDTA, 5 mM DIT, and 1mM sodium azide at 4 O C . After equivalent addition of the inhibitor, the mixture was coneentrated to near 4 mg/mL using Amicon YM-10 membranes (MilliporeLimited). Proteincrystallizationsweresetupusingthehanging-dropvapordiffusion method. The well solution contained 50 mM MES buffer. pH range 5.Cb6.5,or 50 mM sodium acetate buffer, pH
range 4.5-6.0with3W700 mM ammonium sulfateasprecipitant. Thecrystalsgrewwithinaweekatboth 10and20°C. Chemicals were from BDH,Poole, UK. Crystal Characterization. The crystals of complexes with 1and2wherehexagonalcryetalswereohtained(TahleI1)showed high mosaicity and gave maximal resolution of 2.6 and 2.8 A, respectively. Forthe complex with 1thediffractionshowedclear deviationsfrom true mm symmetry in the hkO zone, particularly
athighresolution,fromwhichthespacegr0upP6~wasconcluded while for 2 it was apparent the crystals behaved as if they had
the higher symmetry of P6'22. The asymmetric unit of the symmetric compound 1 is the complete dimer
crystals for the
Wonacott et a1
3118 Jourml of Medicinal Chemistry, 1993, Vol. 36,No. 21
4
Figure View of electron density of final atomic model of 3 in HIVP/3 complex at 2.5-A resolution. The ~o-Fclac electron nsity map was calculated from a model containing all atoms of the protein with the ligand omitted and is contoured to 2.7. The planes of the phenyl rings in Sa are seen to be rotated by a relative angle of 6 P . whereas for compound 2 it is the monomer. For the p u p x e s of refinement, calculations for crystals of the complex with 2 were The orthorhombic crystals carried out in space group B1. obtained in complexes with compound 3 (Table 11)were of the same form as described for acetylpepstatinw with the proteinase dimer as the asymmetric unit of the crystal and diffracted to at least 2.5-A resolution. Data Collection. All diffraction data were recorded from single crystalsusing a FAST area detector: for compounds 1and 2, use was made of beamline 9.6 at SERC Daresbury Laboratory
withawavelengthof0.89Awhileforcompound3CuKolradiation from a rotating anode source was employed. In order to achieve as complete a data set as possible, two mans were collected about different axes in the crystal. The images were processed with the MADNES software package" and then scaled and merged with the CCP4 suite of programs." The resulting merging R factors for symmetry-related reflections are given in Table 11. Structure Determination. The structure of the complex with 1was solved by molecular replacement with the mrdinates of the HIV protease dimer derived from the MVT-101 complex6 using the faat rotation function (Crowther) with data from 10 to 3A. Thecorrectsolution wasthehighestpeak,andwasconaistent with the protease dimer axis being aligned closely with the a-axis of the unit cell. The position of the dimer was established by a 1-D translation search calculating R factors at 0.6-A intervals; this gave a dear minimum consistent with good packing in the B1cell. Complexes with 2 and 3 were closely homologous to other known structures and required no separate solution step. Refinement andModeling. AUstmctureswererefdusing X-PL0R"with initial rigid-body refinement. Water molecules were omitted from the starting modela and the ligand structure modeled into electron density map using FR0DO"after refinement of all protein atomic positions was complete. For the complex with 1 a cycle of simulated annealing at 3000 OC with a slow cooling protocol was carried out after inclusion of the
ligand. Cycles of refinement were alternated with map fitting in which corrections were made to the protein model and solvent molecules added. The resulting structures have R factor values R m given in Table I1 for all reflexions greater than 2.7 and have typiddeviationsfrom idealgeometryof0.02Aforbonddistanm and 3.5O for bond angles. NMR Spectroscopy. Samples for NMR experiments were inhibited and concentrated as above, with a final dialysis into 10 mM acetate@ in DlO or 90% Hz0/10% Dz0,pH 5.0. Protein concentrations ranged from 2 to 4 mg/mL. 1H NMR spectra were recorded on a Bruker AM 500 MHz spectrometer. 'SF spectra were recorded on a Varian VXR 400-MHz spectrometer. AU spectra were recorded at 298 K. Molecular Modeling. The molecular models demrihed were constructed with the aid of the INSIGHTIIL' modeling system, using, where possible, units from within the fragment library andthecrystalstructureof HIVpr0teinase/JG365.~Geometry
optimizationwasperformedusingDISCOvERutilizingtheCVFF force field.18 The results were visualized using INSIGHT11 running on a Silicon Graphics 4D35TG.
Acknowledgment. We thank Dr. Richard Hall and Dr. Dev Baines for t h e contribution to protein scale-up, Dr. Anne Cleasby for crystallographic assistance, and Dr. Alex Wlodawer (NCI, Frederick, MD) for provision of
HIVP coordinates. References (1)
(2)
Kohl,N. E.; Emini, E. A.: Schleif, W. A,; Davies, L. J.; Heimbach, J. C.; Dixon, R. A. F.:Scholnick,E. M.;Siial, I. S. Active Human ImmunaleficiencyV h Protease is Required for Viral Infectivity. Fbc. Notl. Acad. Sei. U.S.A.1988,85,4686-4890. (a) Eriekson, J.; Neidhm3.D. J.;VanDrie,Kempf, D. J.; Wmg.X. C.;Norbeek,D. W.PLattner,D.J.;Rittenhowe, J. W.;Turon,M.; WidebrYg,N.:Kohlbrenner,W.E.;Simmer,R.;Helfrieh,R.;Paul, D. A,; Knigge, M. h i p , Activity and 2.8 A Crystal Structureof
Penicillin-Derived Inhibitors of HIV-1 Proteinase. 1
Journal of Medicinal Chemistry, 1993, Vol. 36, No. 21 8119
R. L.; Haasell, A. M.; Meek, T. D.; Tomaszek, T. A.; Lewis, M. A symmetric inhibitor bmde HIV-1 protease asymmetrically. Biochemistry 1993,32,937-947. (9) (a) Fitzgerald, P. M. P.; McKeever, B. M.; Van Middleeeworth, J. F.; Springer, J. P.; Heinbach, J. C.; Len, C. T.; Herber, W. K.; Dixon,R. A. F.; Darke, P. L. CrystallographicAnalysisof a Complex T.;Heimbach,J.C.;Herber,W.K.;Sigal,I.S.;Darke,P.L.;Springer, between Human Immunodeficiency V i Type 1 Protease and J. P. Three Dimensional Structure of the Aspartyl Protease from Acetyl-Pepstatin a t 2.0 A Resolution. J. Biol. Chem. ISSO, 266, the AIDS V i HIV-1. Nature (London) 1989,337,616-20. (b) 14209-14219. (b) Mirthy, K. H. M.; Winborne, E. L.; Minnich, M. Wlodawer, A,; Miller, M. A.; Miller, M.; Jaskoleki, M.; SathyaD.; Culp, J. S.; Debouck, C. The Crystal Structures of 2.2 A narayana, B. K.; Baldwin, E.; Weber, I. T.; Selk, L. M.; Clawon, Resolution of Hydroxyethylene-baaed Inhibitors Bound to HIV-1 L.; Schneider, L.; Kent, S. B. H. Conserved Folding in Retroviral Protease Show that the Inhibitors are Present in Two Distinct Proteases: Crystal Structure of a Synthetic HIV-1 Protease. Orientations. J. Biol. Chem. 1992,267, 22770-22778. Science 1989,245,616-21. (c)Lapatto,R.;Blundell,T.;Hemming8, (10) The residue (P)and pocket (S)notation is employed cf Schechter, A.;Overington,J.; Wilderspin, A.; Wood, S.; Merson, J. R.; Whittle, I.; Berger, A. On the size of the Active Site in Proteases. I Papain. P. J.; Danley, D. E.; Geoghegan, K. F.; Hawrylik, S. J.; Lee, S. E.; Biochem. Biophys. Res. Commun. 1967,27, 157-162. Scheld, K. G.; Hobart, P. M. X-ray Analysis of HIV-1 Protinase (11) Holmes,D.S.;Clemens,I.R.;Cobley,K.N.;Humber,D.C.;Kitchin, at 2.7 A Resolution Confirms Structural Homologyamong Retoviral J.; Orr, D. C.; Patel, B.; Paternoeter, I. L.; Storer, R. Novel Dimeric Enzymes. Nature (London) 1989,342, 299-302. Penicilli DerivedInhibitors of HIV-1 Proteinaee: Interaction with (4) Humber, D. C.; Cammack, N.; Coates, J. A. V.; Cobley, K. N.; Orr, the Catalytic Aspartates. Biorg. Med. Chem. Lett. In press. D. C.; Storer, R.; Weingarten, G. G.; Weir, M. P. PenicillinDerived (12) Singh, 0. M. P.; Baines, B. S.; Hall, R. M.; Gray, N. M.; Weir, M. CrSymmetric Inhibitorsof HIV-1Proteinase. J.Med. Chem. 1992, P. Large Scale Expression and Purification of HIV-1 Proteinase. 35,3080-3081. J. Biotechnol. 1991,21, 127-136. ( 5 ) Humber,D.C.;Bamford,M. J.;Bethell,R.C.;Cammack,N.;Cobley, (13) Messerschmidt, A.; Pflugrath, J. W. Crystal orientation and x-ray K.; Evans, D. N.; Gray, N. M.; Hann, M. M.; Orr,D. C.; Saunders, pattern prediction routines for area detector diffractometer system J.; Shenoy, B. E. V.; Storer, R.; Weingarten, G. G.; Wyatt, P. G. in macromolecular crystallography. J. Appl. Crystollogr. 1987, A series of Penicillin Derived Crsymmetric Inhibitors of HIV-1 20,306-315. Proteinase: Synthesis, Mode of Interaction and Structure-Activity (14) CCP4. The SERC Collaborative computing project No. 4; A suite Relationships. J. Med. Chem. following paper in this issue. of programs for Protein Crystallography, distributed from Dares(6) Miller, M.; Schneider, J.; Sathyanarayana, B. K.; Toth, M. V.; bury Laboratory, UK. Marshall, G. R.; Clawson, L.; Selk, L.; Kent, S. B. H.; Wlodawer, (15) Brunger, A. T.; Kuriyan, J.; Karplua, M. Crystallographic R-factor A. Structure of a Complex of a S thetic HIV-1 Protease with a refinement by molecular dynamics. Science 1987,235,468-460. Substrate-Based Inhibitor a t 2.3 fiesolution. Science 1989,246, (16) Jones, A. T. Interactive computer graphics: FRODO. Methods 1149-1152. Enzymol. 1985,115,157-171. (7) Gerig, J. T. Fluorine Nuclear Magnetic Resonance of Fluorinated (17) Insight I1 User Guide; Biosym TeChnOlOgieS: San Diego, 1992. Ligands. Methods Enzymol. 1989,177, 3-23. (18) Dauber-Osguthorpe, P.; Roberts,V. A.; Osguthorpe, D. J.; Wolff, (8) (a) Swain, A. L.; Miller, M. M.; Green, J.; Rich, D. H.; Kent, S. B. J.; Genest, M.; Hagler, A. T. Structure and Energetics of Ligand H.; Wlodawer, A.; X-ray crystallographic structure of a complex Binding to Proteins: E. coli Dihydrofolate Reductase-Trimethobetween a synthetic protease of HIV-1 and a substrate-based prim,aDrugReceptorsSystem. Proteine: StrucLhtnct. Genetics hydroxyethylamine inhibitor. Prod. Natl. Acad. Sci. U.S.A. 1990, 1988,4,31-47. 87,8805. (b) Dreyer, G. B.; Boehm, J. C.; Chenera, B.; Des Jarlais,
a Cz Symmetric Inhibitor Complexed to HIV-1 Protease. Science 1990,249, 527-533. (b) Bone, R.; Vacca, J. P.; Anderson, P. S.; Holloway, M. K. X-ray Crystal Structure of the HIV Protease Complex with L-700,417, an Inhibitor with Pseudo Cz Symmetry. J. Am. Chem. SOC.1991,113,9382-9384. (3) (a) Navia, M. A.; Fitzgerald, P. M. D.; McKeever, B. M.; Leu, C.
