- Email: [email protected]
The integration of genomic and structural information in the development of high affinity plasmepsin inhibitors
International Journal for Parasitology 32 (2002) 1669–1676 www.parasitology-online.com
Invited review
The integration of genomic and structural information in the development of high affinity plasmepsin inhibitors Azin Nezami, Ernesto Freire* Department of Biology and The Johns Hopkins Malaria Research Institute, The Johns Hopkins University, Baltimore, MD 21218, USA Received 31 January 2002; received in revised form 5 August 2002; accepted 13 August 2002
Abstract The plasmepsins are key enzymes in the life cycle of the Plasmodium parasites responsible for malaria. Since plasmepsin inhibition leads to parasite death, these enzymes have been acknowledged to be important targets for the development of new antimalarial drugs. The development of effective plasmepsin inhibitors, however, is compounded by their genomic diversity which gives rise not to a unique target for drug development but to a family of closely related targets. Successful drugs will have to inhibit not one but several related enzymes with high affinity. Structure-based drug design against heterogeneous targets requires a departure from the classic ‘lock-and-key’ paradigm that leads to the development of conformationally constrained molecules aimed at a single target. Drug molecules designed along those principles are usually rigid and unable to adapt to target variations arising from naturally occurring genetic polymorphisms or drug-induced resistant mutations. Heterogeneous targets need adaptive drug molecules, characterised by the presence of flexible elements at specific locations that sustain a viable binding affinity against existing or expected polymorphisms. Adaptive ligands have characteristic thermodynamic signatures that distinguish them from their rigid counterparts. This realisation has led to the development of rigorous thermodynamic design guidelines that take advantage of correlations between the structure of lead compounds and the enthalpic and entropic components of the binding affinity. In this paper, we discuss the application of the thermodynamic approach to the development of high affinity (Ki , pM) plasmepsin inhibitors. In particular, a family of allophenylnorstatine-based compounds is evaluated for their potential to inhibit a wide spectrum of plasmepsins. q 2002 Published by Elsevier Science Ltd. on behalf of Australian Society for Parasitology Inc. Keywords: Malaria; Plasmepsins; Genomics; Protease inhibitors; Adaptive inhibitors; Amino acid polymorphisms; Plasmodium
1. Introduction The rapid spread of drug resistance underscores the need for new therapies and the identification of novel targets for drug development against malaria (Wyler, 1993). The Plasmodium parasite, responsible for malaria infection, invades red blood cells and consumes up to 75% of their haemoglobin content (Goldberg, 1993). Three classes of enzymes that digest haemoglobin have been identified in the food vacuole, one cysteine protease (falcipain) (Rosenthal et al., 1988), a metalloprotease (falcilysin) (Eggleson et al., 1999) and a family of aspartic proteases, the plasmepsins. Until recently only plasmepsin I and plasmepsin II (Francis et al., 1997a,b) were known. HAP, a histo-aspartic protease, was identified by Berry et al. (1999). The recently completed genome of Plasmodium falciparum has revealed a total of at least 10 different plasmepsins (Coombs et al., 2001). Plasmepsin I, II, HAP and IV are highly homologous * Corresponding author. Tel.: 11-410-516-7743; fax: 11-410-516-6469. E-mail address: [email protected] (E. Freire).
with 60–70% amino acid identity while the other ones are more distantly related. Plasmepsin I, II, HAP and IV have been localised in the food vacuole of P. falciparum (Banerjee et al., 2002). The inhibition of these enzymes leads to the starvation of the parasite and has been proposed as a viable strategy for drug development. Plasmepsins I and II are synthesised and processed to mature form soon after the parasite invades the red blood cell (Francis et al., 1997a,b; Banerjee et al., 2002). HAP has been termed a histo-aspartic protease due to the replacement of one of the catalytic aspartates by a histidine (Berry et al., 1999; Banerjee et al., 2002). The expression and production of active recombinant plasmepsin I has been shown to be difficult, yielding a truncated protein that lacks the kinetic properties of the native enzyme (Luker et al., 1996). Plasmepsin II, on the other hand, has been successfully expressed, the recombinant protein behaves similarly to the protein isolated from the parasite and its high resolution structure has been determined by X-ray crystallography in complex with the inhibitor pepstatin A (Luker et al., 1996; Silva et al., 1996) and H6680 (Isovaleryl-Val-Val-Sta-OEt),
0020-7519/02/$20.00 q 2002 Published by Elsevier Science Ltd. on behalf of Australian Society for Parasitology Inc. PII: S 0020-751 9(02)00196-0
1670
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
a low molecular weight statine-based inhibitor (Nezami et al., unpublished observation). For these reasons, plasmepsin II has been the primary target for structure-based design (Carrol et al., 1998; Haque et al., 1999; Jiang et al., 2001; Nezami et al., 2002). However, highly specific plasmepsin II inhibitors might not be effective against other members of the plasmepsin family and fail to completely suppress the parasite. Ideally, plasmepsin inhibitors must have extremely high activity against the primary target and maintain high affinity against the remaining plasmepsins by adapting to the existing differences in their binding sites. Inhibitors of this type are known as adaptive inhibitors (Freire, 2002; Velazquez-Campoy and Freire, 2001) and their design requires consideration of genomic diversity and accurate procedures for translating sequence information into three-dimensional space. The first step is to develop a detailed map of the target that includes information about conserved regions, regions of different geometry, regions of different structural stability and regions of different chemical characteristics. Adaptive
inhibitors establish their strongest interactions against conserved regions of the target, and contain flexible elements and asymmetrical functional groups that allow them to accommodate to variable regions within the target family. In this paper, we present a generalised description of the plasmepsin binding site and its application to the design of high affinity inhibitors based upon the allophenylnorstatine scaffold.
2. Mapping genomic diversity into plasmepsin structure Among the 10 genomically identified plasmepsins, plasmepsin I, II, HAP and IV are not only highly homologous but also share the same location inside the food vacuole of the Plasmodium parasite (Banerjee et al., 2002). In addition, the cluster of four genes in chromosome 14 that encode for those proteases are more similar to each other than to those encoding plasmepsins V–X (Coombs et al., 2001). Due to
Fig. 1. The amino acid sequences of the four most closely related plasmepsins (plasmepsin I, plasmepsin II, HAP and plasmepsin IV). Shown in yellow are the amino acids that define the binding site.
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
their high degree of sequence homology, it can be assumed that they share the same overall three-dimensional structure. In fact superposition of the crystallographic structure of plasmepsin II (pdb file 1sme) and the recently released structure of plasmepsin IV (pdb file 1ls5) reveals that their ˚ . Accordbackbones deviate from each other by only 1.05A ingly, the structure of plasmepsin II or IV can be used as a scaffold to map the structural distribution of the amino acid polymorphisms found in all plasmepsins. The amino acid sequences of plasmepsins I, II, HAP and IV are shown in Fig. 1, where the residues that define the binding sites are highlighted in yellow. These residues were identified by their loss in solvent accessibility in the bound structures of plasmepsin II with pepstatin A and H6680. It is evident that the binding site is not defined by a contiguous region in sequence even though they are contiguous in space once the protein adopts its three-dimensional structure. These four plasmepsins share a high degree of sequence homology. Taking plasmepsin II as reference, plasmepsin I shows 73% sequence identity, plasmepsin IV 69% identity and HAP 60% identity. These identities extend to the binding site region. In this case, plasmepsin I shows 84% identity, plasmepsin IV 68% and HAP 39%. HAP, the histo-aspartic protease, has the lower degree of identity; however, even in this case most amino acid polymorphisms within the binding site are rather conservative (55% similarity). The variability at each position in the amino acid sequence can be quantitatively described in terms of an P entropy function Sj ¼ 2 i Pj;i ln Pj;i calculated on the basis of the probability Pj,i that an amino acid of type i
1671
(any one of the 20 amino acids) occupies the j position in the sequence. The entropy function is zero if a position is absolutely conserved and assumes a positive value proportional to the degree of variability if the position is not conserved. In Fig. 2, the sequence entropy in the binding site has been mapped into the structure of plasmepsin II. This figure has been coloured using a rainbow scale in which red denotes the most variable regions and blue the most conserved ones. It is clear that the amino acid polymorphisms within the binding site are not randomly located. The catalytic dyad (except for the histidine in HAP) as well as the base of the binding cavity are highly conserved. Most of the variations occur in the flap and in one edge of the binding site. Conserved and variable regions need to be treated differently in the design of adaptive inhibitors (Freire, 2002; Velazquez-Campoy and Freire, 2001). Fig. 3 shows the expected variations in the binding site of the four plasmepsins. In this figure the binding site of each plasmepsin was constructed using as a starting point the three-dimensional scaffold of plasmepsin II, mutating the necessary side chains and minimising their conformation using methods described by Luque et al. (1998). As expected, plasmepsins I, II and IV exhibit a high degree of three-dimensional homology, while HAP, the most distant in sequence, shows the largest three-dimensional deviation from the other plasmepsins. The difference is more noticeable at the tip of the flap, which is occupied by an amino acid with a longer side chain in HAP (Lys 78) than in the other plasmepsins (Val or Gly), contributing to the more closed appearance of the HAP binding cavity.
Fig. 2. The sequence diversity in the binding site of plasmepsins (I, II, HAP and IV) has been mapped into the structure of plasmepsin II. The amino acids that define the binding site are colour coded according to their variability using a rainbow scale (residues in red exhibit the highest diversity and blue the lowest). Amino acids outside the binding site are depicted in blue independently of their sequence variability. The yellow arrows indicate the catalytic dyad.
1672
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
Fig. 3. Predicted structure of active sites of plasmepsin II, plasmepsin I, HAP and plasmepsin IV based upon the crystallographic structure of plasmepsin II and amino acid polymorphisms in the remaining plasmepsins. Changes in binding site geometry correlate with amino acid differences as discussed in the text.
While adaptive ligands are designed to bind to several members of a target family, they cannot be expected to bind with the same affinity to all members of the family. Targets within a family need to be ranked in terms of their importance to the life cycle of the infectious agent for structure-based design. Inhibitors need to be designed such that they achieve extremely high affinity against the primary target (e.g. pM levels) and are only mildly affected and remain viable (i.e. a loss in binding affinity by one order of magnitude or less) when confronted with other targets in the family.
3. Mapping structural stability into plasmepsin structure In addition to sequence heterogeneity, binding sites also exhibit conformational heterogeneity. It has been shown before that the structural stability of proteins is not uniformly distributed within their three-dimensional structure. In particular, binding sites have been shown to exhibit regions of low and high stability (Luque and Freire, 2000) which define them as dynamic rather than static structures. Furthermore, there appears to be a higher probability for amino acid polymorphisms in regions of low stability. Since the distribution of structural stability can be estimated
reasonably well from the structure (Freire, 1999; Hilser and Freire, 1996), this information can be used to differentiate between easily targeted, highly structured regions from regions displaying conformational and dynamic diversity. Fig. 4 shows the exposed molecular surface of plasmepsin II coloured using a rainbow scale according to the structural stability of the underlying regions (blue ¼ most stable to red ¼ least stable). The structural stability was calculated by using the COREX algorithm as described before (Freire, 1999; Hilser and Freire, 1996). This algorithm generates a large ensemble of partially folded conformations (.10 5) and calculates the probability of each conformation under a given set of conditions. The structural stability of any given residue is calculated from the probability of finding that residue in the native conformation in the entire ensemble. As illustrated in Fig. 4, the core of plasmepsin II is highly stable (blue to green colour) whereas the flap and other loops around the binding site (yellow to red) exhibit low structural stability. In particular, the binding cavity is characterised by a region of high stability, corresponding to the catalytic dyad. Other areas around the binding site but most notably the flap (residues 70–81) and the loops (residues 275–285 and 236–244) surrounding the binding cavity exhibit low structural stability. As observed for other proteins, a significant proportion of the observed poly-
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
1673
Fig. 4. Surface representation of plasmepsin II coloured according to its structural stability. The surface is coloured using a rainbow scale in which red denotes the lowest stability and blue the highest. This figure indicates that the binding cavity is defined by regions of high stability, in particular the catalytic dyad, and regions of low stability, in particular the flap and other loops (indicated by sequence number) that surround the binding site.
morphisms between plasmepsins I, II, IV and HAP are located in regions of low structural stability. Since residues in high stability regions are required to establish strong and specific interactions among themselves, their probability to mutate is lower than that of residues that are not required to establish those interactions. From the point of view of drug design, it is more advantageous to engineer the most critical drug interactions against structurally stable rather than unstable residues. From a thermodynamic point of view, an interaction with a low stability region needs to pay the additional energy penalty of stabilising that region into a unique conformation.
4. General characteristics of adaptive ligands Traditionally, structure-based drug design against protein targets has been predicated on the idea of the lock-and-key hypothesis. This approach assumes a unique and structurally stable protein binding site, and is aimed at developing drug molecules that are conformationally constrained and preshaped to the geometry of the selected target. The unwanted effect of this strategy is the inability of those molecules to adapt to changes in the target (e.g. drug-resistant mutations) or to be directed against an array of closely related targets, a situation that has become evident with the proliferation of genomic information for infectious microorganisms. Adaptive ligands should contain flexible elements that permit accommodation to variable regions in the target family while establishing strong interactions with conserved structural elements as shown in Fig. 2. Adaptive ligands need to be capable of presenting different interacting faces to variable regions in the binding site, a
property that can be achieved by the presence of asymmetric functional groups and flexible elements (VelazquezCampoy and Freire, 2001; Velazquez-Campoy et al., 2001a,b). However, the introduction of flexibility lowers the binding affinity, since flexible ligands lose more conformational entropy than their conformationally constrained counterparts. One rotatable bond that becomes immobilised upon binding carries a Gibbs energy penalty close to 0.5 kcal/mol due to the loss of conformational entropy. For that reason, flexibility needs to be selectively introduced at those critical points identified by a diversity analysis of the target. Also, the introduction of flexibility needs to be compensated by other favourable interactions that will maintain high affinity and specificity. Since the binding energy is determined by enthalpic and entropic interactions (DG ¼ DH 2 TDS), one way to compensate for weaker entropic interactions is by a more favourable binding enthalpy. High affinity ligands that contain a certain degree of flexibility need to establish strong enthalpic interactions with the target, since their entropic contribution is reduced when compared with that of their conformationally constrained counterparts. A clear example is given by the evolution of HIV-1 protease inhibitors used in the treatment of AIDS. The binding affinity of the first generation of protease inhibitors (Indinavir, Nelfinavir, Ritonavir, Saquinavir) was characterised by unfavourable or only slightly favourable binding enthalpies (Todd et al., 2000). The binding entropy (achieved by a combination of hydrophobicity and conformational constraints) had to be extremely high in order to overcome the unfavourable enthalpy and still achieve high binding affinity. On the other hand, secondgeneration inhibitors that are able to accommodate and maintain efficiency against common resistant mutations
1674
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
are characterised by highly favourable enthalpies and smaller binding entropies, i.e. they can contain flexible elements and simultaneously exhibit a binding affinity much higher than that of first-generation inhibitors (Velazquez-Campoy and Freire, 2001; Velazquez-Campoy et al., 2001a,b). The experience with HIV-1 protease inhibitors indicates that if the enthalpy change is favourable, the entropy change does not have to be maximised in order to achieve high binding affinity. Therefore, it seems reasonable to identify lead compounds on the basis of affinity and enthalpy rather than affinity alone. Enthalpically favourable lead compounds can be optimised to extremely high binding affinities without constraining them to the point of being unable to adapt to target variations. Enthalpically favourable lead compounds can be identified experimentally or computationally using a structure-based thermodynamic screening (Velazquez-Campoy and Freire, 2001; Velazquez-Campoy et al., 2001a,b).
5. Allophenylnorstatine-based plasmepsin inhibitors The allophenylnorstatine scaffold was chosen as a prospective plasmepsin inhibitor because of its close similarity to the primary cleavage site of plasmepsin II, the peptide bond between Phe 33 and Leu 34 in the a chain of haemoglobin, and the potential for the introduction of different types of functional groups for lead optimisation (Nezami et al., 2002) (Fig. 5). The allophenylnorstatine
scaffold contains four different positions (labelled R1–R4 in the structure in Fig. 5) where different chemical functional groups can be introduced in order to improve binding affinity and selectivity. Using the standard enzymatic nomenclature, the allophenylnorstatine moiety in these compounds corresponds to the P1 position, R1 corresponds to the P2 position, the thioproline group together with R2 and R3 correspond to the P1 0 position and R4 corresponds to the P2 0 position. We have measured the binding energetics of a diverse library of allophenylnorstatine-based compounds (KNI library) (Nezami et al., 2002). Some of the compounds in this library were shown to bind plasmepsin II with a binding affinity as low as 20 nM. The binding enthalpy of a subset of these compounds was determined by high sensitivity isothermal titration calorimetry. All the samples were found to bind plasmepsin II with a favourable binding enthalpy averaging 23.0 kcal/mol and ranging from 21.5 to 27.4 kcal/mol, strongly suggesting that the KNI core itself (the only constant in the nine compounds measured by microcalorimetry) is responsible for this strong interaction. After the first screening (Nezami et al., 2002), KNI-727 (Fig. 5) was selected for further optimisation because of its good thermodynamic profile. This compound binds plasmepsin II with a Ki of 70 nM and a binding enthalpy of 24.4 kcal/mol, suggesting a role as a lead compound for an adaptive inhibitor. It was also noticed that substitution of the symmetric t-butylamide group at position R4 by an asymmetric functional group linked to the allophenylnorsta-
Fig. 5. Chemical structures of Phe-Leu (A), the cleavage motif of plasmepsin II, the allophenylnorstatine scaffold (B), indicating the places where different chemical functional groups can be introduced, and KNI-727 (C), the compound selected in the first screening (Nezami et al., 2002).
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
tine scaffold by an additional rotatable bond resulted in better plasmepsin II binding affinity and improved inhibitory effect (Nezami et al., 2002).
6. Structure-based thermodynamic modelling of KNI inhibitors A good candidate for an adaptive inhibitor must establish its strongest and enthalpically favourable interactions against the conserved regions of the plasmepsin’s binding site. Since all KNI compounds examined by isothermal titration calorimetry yielded favourable binding enthalpies, we decided to investigate whether the allophenylnorstatine nucleus itself (the common element in all KNI compounds) establishes strong enthalpic interactions with plasmepsin II. Towards that end, the allophenylnorstatine scaffold was docked into the plasmepsin II structure using procedures reported before (Nezami et al., 2002). The results are shown in Fig. 6. In the resulting structure, the hydroxyl group in the allophenylnorstatine scaffold is within hydrogen bond distance from the two catalytic aspartates, adopt-
1675
ing a very similar structure to that of H6680. The rest of the scaffold establishes strong interactions with eight residues: Tyr 192, Asp 303, Gly 36, Asp 34, Thr 217, Asp 214, Met 15 and Gly 216 at the bottom of the active site. Of these residues, Asp 214, Asp 303 and Thr 217 are conserved in the four plasmepsins. Tyr 192, Gly 36, Asp 34, Met 15, and Gly 216 are conserved in three plasmepsins (I, II and IV). In HAP these five positions are occupied by Met 192, Ala 36, His 34, Leu 15 and Ala 216. The allophenylnorstatine scaffold also interacts strongly with the tip of the flap (residue 78) which is occupied by Val in plasmepsins I and II, Gly in plasmepsin IV and Lys in HAP. Even though this region is variable among the plasmepsins, most of this interaction involves backbone atoms and has a polar character resulting in a hydrogen bond between the carbonyl oxygen between P2 and P1 in allophenylnorstatine and the side chain of Ser 79. Ser 79 is conserved in plasmepsins I, II and IV and is replaced by Ala in HAP. Asn 76 and Tyr 77 which also participate in the interaction are highly conserved among the plasmepsins. Structure-based thermodynamic analysis (Leavitt and Freire, 2001; Luque et al., 1998; Velazquez-Campoy et al., 2001a,b) indicates that the allophenylnorstatine scaffold by itself binds to plasmepsin II with a favourable binding enthalpy of 26.7 kcal/mol. The favourable binding enthalpy originates from a significant number of polar interactions. ˚ 2 of surface The allophenylnorstatine complex buries 790A 2 ˚ (41%) have a polar character and area of which 326A establishes three hydrogen bonds with the enzyme. Thus, from a structural and thermodynamic standpoint the allophenylnorstatine scaffold appears to satisfy the requirements for the core of an adaptive inhibitor since it interacts with the most conserved regions of the target in an enthalpically favourable fashion. From a structural standpoint, this scaffold provides the necessary anchoring points to develop inhibitor molecules that will exhibit very high affinity against at least three of the plasmepsins (plasmepsins I, II and IV) and a reasonably high affinity against HAP. Acknowledgements
Fig. 6. Docking of allophenylnorstatine scaffold into the plasmepsin II structure. Protein residues have been coloured according to their predicted interaction with the inhibitor using a rainbow scale (blue ¼ strongest interaction, red ¼ no interaction). The minimised allophenylnorstatine scaffold was docked into the binding site of plasmepsin II using the structure of the complex of plasmepsin II with H6680 as a template. The hydroxyl group of the allophenylnorstatine scaffold was initially placed between Asp 34 and Asp 214 at the same position of the hydroxyl group in the statine-based inhibitor. The docking procedure was carried out using the Affinity module of Insight II (MSI, San Diego, CA) and the CVFF force field for docking and scoring. The ligand was confined in the binding site using a flatbottomed energy well in order to minimise repulsion. Several independent docking procedures were carried out with two minimisation iterations between docking steps. The structure with the lowest energy was then energy minimised using the Discover module of Insight II by relaxing the inhibitor and the binding site residues.
The authors thank Professor Y. Kiso from Kyoto Pharmaceutical University for helpful discussions and for supplying us with the KNI compounds. This study was supported by grants from the National Institutes of Health GM 57144 and the National Science Foundation MCB9816661. References Banerjee, R., Liu, J., Beatty, W., Pelosof, L., Klemba, M., Goldberg, D.E., 2002. Four plasmepsins are active in the Plasmodium falciparum food vacuole, including a protease with an active-site histidine. Proc. Natl. Acad. Sci. USA 99, 990–5. Berry, C., Humphreys, J., Matharu, P., Granger, R., Horrocks, P., Moon, R.P., Certa, U., Ridley, R.G., Bur, D., Kay, J., 1999. A distinct member
1676
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
of the aspartic proteinase gene family from the human malaria parasite Plasmodium falciparum. FEBS Lett. 447, 149–54. Carrol, C.D., Patel, H., Johnson, T.O., Guo, T., Orlowski, M., He, Z.M., Cavallaro, C.L., Guo, J., Oksman, A., Gluzman, I.Y., Connelly, J., Chelsky, D., Goldberg, D.E., Dolle, R.E., 1998. Identification of potent inhibitors of Plasmodium falciparum plasmepsin II from an encoded statine combinatorial library. Bioorg. Med. Chem. Lett. 8, 2315–20. Coombs, G.H., Goldberg, D.E., Klemba, M., Berry, C., Kay, J., Mottram, J.C., 2001. Aspartic proteases of Plasmodium falciparum and other parasitic protozoa as drug targets. Trends Parasitol. 17, 532–7. Eggleson, K.K., Duffin, K.L., Goldberg, D.E., 1999. Identification and characterization of falcilysin, a metallopeptidase involved in hemoglobin catabolism within the malaria parasite Plasmodium falciparum. J. Biol. Chem. 274, 32411–7. Francis, S.E., Banerjee, R., Goldberg, D.E., 1997a. Biosynthesis and maturation of the malaria aspartic hemoglobinases plasmepsins I and II. J. Biol. Chem. 272, 14961–8. Francis, S.E., Sullivan, D.J., Goldberg, D.E., 1997b. Hemoglobin metabolism in the malaria parasite Plasmodium falciparum. Annu. Rev. Microbiol. 51, 97–123. Freire, E., 1999. The propagation of binding interactions to remote sites in proteins. Analysis of the binding of the monoclonal antibody D1.3 to lysozyme. Proc. Natl. Acad. Sci. USA 96, 10118–22. Freire, E., 2002. Designing drugs against heterogeneous targets. Nat. Biotech. 20, 15–16. Goldberg, D.E., 1993. Hemoglobin degradation in infected red blood cells. Semin. Cell Biol. 4, 355–61. Haque, T.S., Skillman, G., Lee, C.E., Habashita, H., Gluzman, I.Y., Ewing, T.J.A., Goldberg, D.E., Kuntz, I.D., Ellman, J.A., 1999. Potent, lowmolecular-weight non-peptide inhibitors of malarial aspartyl protease plasmepsin II. J. Med. Chem. 42, 1428–40. Hilser, V.J., Freire, E., 1996. Structure based calculation of the equilibrium folding pathway of proteins. Correlation with hydrogen exchange protection factors. J. Mol. Biol. 262, 756–72. Jiang, S., Prigge, S.T., Wei, L., Gao, Y.E., Hudson, T.H., Gerena, L., Dame, J.B., Kyle, D.E., 2001. New class of small nonpeptidyl compounds blocks Plasmodium falciparum development in vitro by inhibiting plasmepsins. Antimicrob. Agents Chemother. 45, 2577–84.
Leavitt, S., Freire, E., 2001. Direct measurement of protein binding energetics by isothermal titration calorimetry. Curr. Opin. Struct. Biol. 11, 560–6. Luker, K.E., Francis, S.E., Gluzman, I.Y., Goldberg, D.E., 1996. Kinetic analysis of plasmepsins I and II, aspartic proteases of the Plasmodium falciparum digestive vacuole. Mol. Biochem. Parasitol. 79, 71–78. Luque, I., Freire, E., 2000. The structural stability of binding sites. Consequences for binding affinity and cooperativity. Proteins 4, 63–71. Luque, I., Gomez, J., Semo, N., Freire, E., 1998. Structure-based thermodynamic design of peptide ligands. Application to peptide inhibitors of the aspartic protease endothiapepsin. Proteins 30, 74–85. Nezami, A., Luque, I., Kimura, T., Kiso, Y., Freire, E., 2002. Identification and characterization of allophenylnorstatine-based inhibitors of plasmepsin II, an anti-malarial target. Biochemistry 41, 2273–80. Rosenthal, P.J., McKerrow, J.H., Aikawa, M., Nagasawa, H., Leech, J.H., 1988. A malarial cysteine proteinase is necessary for hemoglobin degradation by Plasmodium falciparum. J. Clin. Invest. 82, 1560–6. Silva, A.M., Lee, A.Y., Gulnik, S.V., Majer, P., Collins, J., Bhat, T.N., Collins, P.J., Cachau, R.E., Luker, K.E., Gluzman, I.Y., Francis, S.E., Oksman, A., Goldberg, D.E., Erickson, J.W., 1996. Structure and inhibition of plasmepsin II, a hemoglobin-degrading enzyme from Plasmodium falciparum. Proc. Natl. Acad. Sci. USA 93, 10034–9. Todd, M.J., Luque, I., Velazquez-Campoy, A., Freire, E., 2000. The thermodynamic basis of resistance to HIV-1 protease inhibition. Calorimetric analysis of the V82F/I84V active site resistant mutant. Biochemistry 39, 11876–83. Velazquez-Campoy, A., Freire, E., 2001. Incorporating target heterogeneity in drug design. J. Cell. Biochem. S37, 82–88. Velazquez-Campoy, A., Kiso, Y., Freire, E., 2001a. The binding energetics of first and second generation HIV-1 protease inhibitors: implications for drug design. Arch. Biochim. Biophys. 390, 169–75. Velazquez-Campoy, A., Luque, I., Freire, E., 2001b. The application of thermodynamic methods in drug design. Thermochim. Acta 380, 217–27. Wyler, D.J., 1993. Malaria overview and update. Clin. Infect. Dis. 16, 449– 58.
Invited review
The integration of genomic and structural information in the development of high affinity plasmepsin inhibitors Azin Nezami, Ernesto Freire* Department of Biology and The Johns Hopkins Malaria Research Institute, The Johns Hopkins University, Baltimore, MD 21218, USA Received 31 January 2002; received in revised form 5 August 2002; accepted 13 August 2002
Abstract The plasmepsins are key enzymes in the life cycle of the Plasmodium parasites responsible for malaria. Since plasmepsin inhibition leads to parasite death, these enzymes have been acknowledged to be important targets for the development of new antimalarial drugs. The development of effective plasmepsin inhibitors, however, is compounded by their genomic diversity which gives rise not to a unique target for drug development but to a family of closely related targets. Successful drugs will have to inhibit not one but several related enzymes with high affinity. Structure-based drug design against heterogeneous targets requires a departure from the classic ‘lock-and-key’ paradigm that leads to the development of conformationally constrained molecules aimed at a single target. Drug molecules designed along those principles are usually rigid and unable to adapt to target variations arising from naturally occurring genetic polymorphisms or drug-induced resistant mutations. Heterogeneous targets need adaptive drug molecules, characterised by the presence of flexible elements at specific locations that sustain a viable binding affinity against existing or expected polymorphisms. Adaptive ligands have characteristic thermodynamic signatures that distinguish them from their rigid counterparts. This realisation has led to the development of rigorous thermodynamic design guidelines that take advantage of correlations between the structure of lead compounds and the enthalpic and entropic components of the binding affinity. In this paper, we discuss the application of the thermodynamic approach to the development of high affinity (Ki , pM) plasmepsin inhibitors. In particular, a family of allophenylnorstatine-based compounds is evaluated for their potential to inhibit a wide spectrum of plasmepsins. q 2002 Published by Elsevier Science Ltd. on behalf of Australian Society for Parasitology Inc. Keywords: Malaria; Plasmepsins; Genomics; Protease inhibitors; Adaptive inhibitors; Amino acid polymorphisms; Plasmodium
1. Introduction The rapid spread of drug resistance underscores the need for new therapies and the identification of novel targets for drug development against malaria (Wyler, 1993). The Plasmodium parasite, responsible for malaria infection, invades red blood cells and consumes up to 75% of their haemoglobin content (Goldberg, 1993). Three classes of enzymes that digest haemoglobin have been identified in the food vacuole, one cysteine protease (falcipain) (Rosenthal et al., 1988), a metalloprotease (falcilysin) (Eggleson et al., 1999) and a family of aspartic proteases, the plasmepsins. Until recently only plasmepsin I and plasmepsin II (Francis et al., 1997a,b) were known. HAP, a histo-aspartic protease, was identified by Berry et al. (1999). The recently completed genome of Plasmodium falciparum has revealed a total of at least 10 different plasmepsins (Coombs et al., 2001). Plasmepsin I, II, HAP and IV are highly homologous * Corresponding author. Tel.: 11-410-516-7743; fax: 11-410-516-6469. E-mail address: [email protected] (E. Freire).
with 60–70% amino acid identity while the other ones are more distantly related. Plasmepsin I, II, HAP and IV have been localised in the food vacuole of P. falciparum (Banerjee et al., 2002). The inhibition of these enzymes leads to the starvation of the parasite and has been proposed as a viable strategy for drug development. Plasmepsins I and II are synthesised and processed to mature form soon after the parasite invades the red blood cell (Francis et al., 1997a,b; Banerjee et al., 2002). HAP has been termed a histo-aspartic protease due to the replacement of one of the catalytic aspartates by a histidine (Berry et al., 1999; Banerjee et al., 2002). The expression and production of active recombinant plasmepsin I has been shown to be difficult, yielding a truncated protein that lacks the kinetic properties of the native enzyme (Luker et al., 1996). Plasmepsin II, on the other hand, has been successfully expressed, the recombinant protein behaves similarly to the protein isolated from the parasite and its high resolution structure has been determined by X-ray crystallography in complex with the inhibitor pepstatin A (Luker et al., 1996; Silva et al., 1996) and H6680 (Isovaleryl-Val-Val-Sta-OEt),
0020-7519/02/$20.00 q 2002 Published by Elsevier Science Ltd. on behalf of Australian Society for Parasitology Inc. PII: S 0020-751 9(02)00196-0
1670
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
a low molecular weight statine-based inhibitor (Nezami et al., unpublished observation). For these reasons, plasmepsin II has been the primary target for structure-based design (Carrol et al., 1998; Haque et al., 1999; Jiang et al., 2001; Nezami et al., 2002). However, highly specific plasmepsin II inhibitors might not be effective against other members of the plasmepsin family and fail to completely suppress the parasite. Ideally, plasmepsin inhibitors must have extremely high activity against the primary target and maintain high affinity against the remaining plasmepsins by adapting to the existing differences in their binding sites. Inhibitors of this type are known as adaptive inhibitors (Freire, 2002; Velazquez-Campoy and Freire, 2001) and their design requires consideration of genomic diversity and accurate procedures for translating sequence information into three-dimensional space. The first step is to develop a detailed map of the target that includes information about conserved regions, regions of different geometry, regions of different structural stability and regions of different chemical characteristics. Adaptive
inhibitors establish their strongest interactions against conserved regions of the target, and contain flexible elements and asymmetrical functional groups that allow them to accommodate to variable regions within the target family. In this paper, we present a generalised description of the plasmepsin binding site and its application to the design of high affinity inhibitors based upon the allophenylnorstatine scaffold.
2. Mapping genomic diversity into plasmepsin structure Among the 10 genomically identified plasmepsins, plasmepsin I, II, HAP and IV are not only highly homologous but also share the same location inside the food vacuole of the Plasmodium parasite (Banerjee et al., 2002). In addition, the cluster of four genes in chromosome 14 that encode for those proteases are more similar to each other than to those encoding plasmepsins V–X (Coombs et al., 2001). Due to
Fig. 1. The amino acid sequences of the four most closely related plasmepsins (plasmepsin I, plasmepsin II, HAP and plasmepsin IV). Shown in yellow are the amino acids that define the binding site.
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
their high degree of sequence homology, it can be assumed that they share the same overall three-dimensional structure. In fact superposition of the crystallographic structure of plasmepsin II (pdb file 1sme) and the recently released structure of plasmepsin IV (pdb file 1ls5) reveals that their ˚ . Accordbackbones deviate from each other by only 1.05A ingly, the structure of plasmepsin II or IV can be used as a scaffold to map the structural distribution of the amino acid polymorphisms found in all plasmepsins. The amino acid sequences of plasmepsins I, II, HAP and IV are shown in Fig. 1, where the residues that define the binding sites are highlighted in yellow. These residues were identified by their loss in solvent accessibility in the bound structures of plasmepsin II with pepstatin A and H6680. It is evident that the binding site is not defined by a contiguous region in sequence even though they are contiguous in space once the protein adopts its three-dimensional structure. These four plasmepsins share a high degree of sequence homology. Taking plasmepsin II as reference, plasmepsin I shows 73% sequence identity, plasmepsin IV 69% identity and HAP 60% identity. These identities extend to the binding site region. In this case, plasmepsin I shows 84% identity, plasmepsin IV 68% and HAP 39%. HAP, the histo-aspartic protease, has the lower degree of identity; however, even in this case most amino acid polymorphisms within the binding site are rather conservative (55% similarity). The variability at each position in the amino acid sequence can be quantitatively described in terms of an P entropy function Sj ¼ 2 i Pj;i ln Pj;i calculated on the basis of the probability Pj,i that an amino acid of type i
1671
(any one of the 20 amino acids) occupies the j position in the sequence. The entropy function is zero if a position is absolutely conserved and assumes a positive value proportional to the degree of variability if the position is not conserved. In Fig. 2, the sequence entropy in the binding site has been mapped into the structure of plasmepsin II. This figure has been coloured using a rainbow scale in which red denotes the most variable regions and blue the most conserved ones. It is clear that the amino acid polymorphisms within the binding site are not randomly located. The catalytic dyad (except for the histidine in HAP) as well as the base of the binding cavity are highly conserved. Most of the variations occur in the flap and in one edge of the binding site. Conserved and variable regions need to be treated differently in the design of adaptive inhibitors (Freire, 2002; Velazquez-Campoy and Freire, 2001). Fig. 3 shows the expected variations in the binding site of the four plasmepsins. In this figure the binding site of each plasmepsin was constructed using as a starting point the three-dimensional scaffold of plasmepsin II, mutating the necessary side chains and minimising their conformation using methods described by Luque et al. (1998). As expected, plasmepsins I, II and IV exhibit a high degree of three-dimensional homology, while HAP, the most distant in sequence, shows the largest three-dimensional deviation from the other plasmepsins. The difference is more noticeable at the tip of the flap, which is occupied by an amino acid with a longer side chain in HAP (Lys 78) than in the other plasmepsins (Val or Gly), contributing to the more closed appearance of the HAP binding cavity.
Fig. 2. The sequence diversity in the binding site of plasmepsins (I, II, HAP and IV) has been mapped into the structure of plasmepsin II. The amino acids that define the binding site are colour coded according to their variability using a rainbow scale (residues in red exhibit the highest diversity and blue the lowest). Amino acids outside the binding site are depicted in blue independently of their sequence variability. The yellow arrows indicate the catalytic dyad.
1672
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
Fig. 3. Predicted structure of active sites of plasmepsin II, plasmepsin I, HAP and plasmepsin IV based upon the crystallographic structure of plasmepsin II and amino acid polymorphisms in the remaining plasmepsins. Changes in binding site geometry correlate with amino acid differences as discussed in the text.
While adaptive ligands are designed to bind to several members of a target family, they cannot be expected to bind with the same affinity to all members of the family. Targets within a family need to be ranked in terms of their importance to the life cycle of the infectious agent for structure-based design. Inhibitors need to be designed such that they achieve extremely high affinity against the primary target (e.g. pM levels) and are only mildly affected and remain viable (i.e. a loss in binding affinity by one order of magnitude or less) when confronted with other targets in the family.
3. Mapping structural stability into plasmepsin structure In addition to sequence heterogeneity, binding sites also exhibit conformational heterogeneity. It has been shown before that the structural stability of proteins is not uniformly distributed within their three-dimensional structure. In particular, binding sites have been shown to exhibit regions of low and high stability (Luque and Freire, 2000) which define them as dynamic rather than static structures. Furthermore, there appears to be a higher probability for amino acid polymorphisms in regions of low stability. Since the distribution of structural stability can be estimated
reasonably well from the structure (Freire, 1999; Hilser and Freire, 1996), this information can be used to differentiate between easily targeted, highly structured regions from regions displaying conformational and dynamic diversity. Fig. 4 shows the exposed molecular surface of plasmepsin II coloured using a rainbow scale according to the structural stability of the underlying regions (blue ¼ most stable to red ¼ least stable). The structural stability was calculated by using the COREX algorithm as described before (Freire, 1999; Hilser and Freire, 1996). This algorithm generates a large ensemble of partially folded conformations (.10 5) and calculates the probability of each conformation under a given set of conditions. The structural stability of any given residue is calculated from the probability of finding that residue in the native conformation in the entire ensemble. As illustrated in Fig. 4, the core of plasmepsin II is highly stable (blue to green colour) whereas the flap and other loops around the binding site (yellow to red) exhibit low structural stability. In particular, the binding cavity is characterised by a region of high stability, corresponding to the catalytic dyad. Other areas around the binding site but most notably the flap (residues 70–81) and the loops (residues 275–285 and 236–244) surrounding the binding cavity exhibit low structural stability. As observed for other proteins, a significant proportion of the observed poly-
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
1673
Fig. 4. Surface representation of plasmepsin II coloured according to its structural stability. The surface is coloured using a rainbow scale in which red denotes the lowest stability and blue the highest. This figure indicates that the binding cavity is defined by regions of high stability, in particular the catalytic dyad, and regions of low stability, in particular the flap and other loops (indicated by sequence number) that surround the binding site.
morphisms between plasmepsins I, II, IV and HAP are located in regions of low structural stability. Since residues in high stability regions are required to establish strong and specific interactions among themselves, their probability to mutate is lower than that of residues that are not required to establish those interactions. From the point of view of drug design, it is more advantageous to engineer the most critical drug interactions against structurally stable rather than unstable residues. From a thermodynamic point of view, an interaction with a low stability region needs to pay the additional energy penalty of stabilising that region into a unique conformation.
4. General characteristics of adaptive ligands Traditionally, structure-based drug design against protein targets has been predicated on the idea of the lock-and-key hypothesis. This approach assumes a unique and structurally stable protein binding site, and is aimed at developing drug molecules that are conformationally constrained and preshaped to the geometry of the selected target. The unwanted effect of this strategy is the inability of those molecules to adapt to changes in the target (e.g. drug-resistant mutations) or to be directed against an array of closely related targets, a situation that has become evident with the proliferation of genomic information for infectious microorganisms. Adaptive ligands should contain flexible elements that permit accommodation to variable regions in the target family while establishing strong interactions with conserved structural elements as shown in Fig. 2. Adaptive ligands need to be capable of presenting different interacting faces to variable regions in the binding site, a
property that can be achieved by the presence of asymmetric functional groups and flexible elements (VelazquezCampoy and Freire, 2001; Velazquez-Campoy et al., 2001a,b). However, the introduction of flexibility lowers the binding affinity, since flexible ligands lose more conformational entropy than their conformationally constrained counterparts. One rotatable bond that becomes immobilised upon binding carries a Gibbs energy penalty close to 0.5 kcal/mol due to the loss of conformational entropy. For that reason, flexibility needs to be selectively introduced at those critical points identified by a diversity analysis of the target. Also, the introduction of flexibility needs to be compensated by other favourable interactions that will maintain high affinity and specificity. Since the binding energy is determined by enthalpic and entropic interactions (DG ¼ DH 2 TDS), one way to compensate for weaker entropic interactions is by a more favourable binding enthalpy. High affinity ligands that contain a certain degree of flexibility need to establish strong enthalpic interactions with the target, since their entropic contribution is reduced when compared with that of their conformationally constrained counterparts. A clear example is given by the evolution of HIV-1 protease inhibitors used in the treatment of AIDS. The binding affinity of the first generation of protease inhibitors (Indinavir, Nelfinavir, Ritonavir, Saquinavir) was characterised by unfavourable or only slightly favourable binding enthalpies (Todd et al., 2000). The binding entropy (achieved by a combination of hydrophobicity and conformational constraints) had to be extremely high in order to overcome the unfavourable enthalpy and still achieve high binding affinity. On the other hand, secondgeneration inhibitors that are able to accommodate and maintain efficiency against common resistant mutations
1674
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
are characterised by highly favourable enthalpies and smaller binding entropies, i.e. they can contain flexible elements and simultaneously exhibit a binding affinity much higher than that of first-generation inhibitors (Velazquez-Campoy and Freire, 2001; Velazquez-Campoy et al., 2001a,b). The experience with HIV-1 protease inhibitors indicates that if the enthalpy change is favourable, the entropy change does not have to be maximised in order to achieve high binding affinity. Therefore, it seems reasonable to identify lead compounds on the basis of affinity and enthalpy rather than affinity alone. Enthalpically favourable lead compounds can be optimised to extremely high binding affinities without constraining them to the point of being unable to adapt to target variations. Enthalpically favourable lead compounds can be identified experimentally or computationally using a structure-based thermodynamic screening (Velazquez-Campoy and Freire, 2001; Velazquez-Campoy et al., 2001a,b).
5. Allophenylnorstatine-based plasmepsin inhibitors The allophenylnorstatine scaffold was chosen as a prospective plasmepsin inhibitor because of its close similarity to the primary cleavage site of plasmepsin II, the peptide bond between Phe 33 and Leu 34 in the a chain of haemoglobin, and the potential for the introduction of different types of functional groups for lead optimisation (Nezami et al., 2002) (Fig. 5). The allophenylnorstatine
scaffold contains four different positions (labelled R1–R4 in the structure in Fig. 5) where different chemical functional groups can be introduced in order to improve binding affinity and selectivity. Using the standard enzymatic nomenclature, the allophenylnorstatine moiety in these compounds corresponds to the P1 position, R1 corresponds to the P2 position, the thioproline group together with R2 and R3 correspond to the P1 0 position and R4 corresponds to the P2 0 position. We have measured the binding energetics of a diverse library of allophenylnorstatine-based compounds (KNI library) (Nezami et al., 2002). Some of the compounds in this library were shown to bind plasmepsin II with a binding affinity as low as 20 nM. The binding enthalpy of a subset of these compounds was determined by high sensitivity isothermal titration calorimetry. All the samples were found to bind plasmepsin II with a favourable binding enthalpy averaging 23.0 kcal/mol and ranging from 21.5 to 27.4 kcal/mol, strongly suggesting that the KNI core itself (the only constant in the nine compounds measured by microcalorimetry) is responsible for this strong interaction. After the first screening (Nezami et al., 2002), KNI-727 (Fig. 5) was selected for further optimisation because of its good thermodynamic profile. This compound binds plasmepsin II with a Ki of 70 nM and a binding enthalpy of 24.4 kcal/mol, suggesting a role as a lead compound for an adaptive inhibitor. It was also noticed that substitution of the symmetric t-butylamide group at position R4 by an asymmetric functional group linked to the allophenylnorsta-
Fig. 5. Chemical structures of Phe-Leu (A), the cleavage motif of plasmepsin II, the allophenylnorstatine scaffold (B), indicating the places where different chemical functional groups can be introduced, and KNI-727 (C), the compound selected in the first screening (Nezami et al., 2002).
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
tine scaffold by an additional rotatable bond resulted in better plasmepsin II binding affinity and improved inhibitory effect (Nezami et al., 2002).
6. Structure-based thermodynamic modelling of KNI inhibitors A good candidate for an adaptive inhibitor must establish its strongest and enthalpically favourable interactions against the conserved regions of the plasmepsin’s binding site. Since all KNI compounds examined by isothermal titration calorimetry yielded favourable binding enthalpies, we decided to investigate whether the allophenylnorstatine nucleus itself (the common element in all KNI compounds) establishes strong enthalpic interactions with plasmepsin II. Towards that end, the allophenylnorstatine scaffold was docked into the plasmepsin II structure using procedures reported before (Nezami et al., 2002). The results are shown in Fig. 6. In the resulting structure, the hydroxyl group in the allophenylnorstatine scaffold is within hydrogen bond distance from the two catalytic aspartates, adopt-
1675
ing a very similar structure to that of H6680. The rest of the scaffold establishes strong interactions with eight residues: Tyr 192, Asp 303, Gly 36, Asp 34, Thr 217, Asp 214, Met 15 and Gly 216 at the bottom of the active site. Of these residues, Asp 214, Asp 303 and Thr 217 are conserved in the four plasmepsins. Tyr 192, Gly 36, Asp 34, Met 15, and Gly 216 are conserved in three plasmepsins (I, II and IV). In HAP these five positions are occupied by Met 192, Ala 36, His 34, Leu 15 and Ala 216. The allophenylnorstatine scaffold also interacts strongly with the tip of the flap (residue 78) which is occupied by Val in plasmepsins I and II, Gly in plasmepsin IV and Lys in HAP. Even though this region is variable among the plasmepsins, most of this interaction involves backbone atoms and has a polar character resulting in a hydrogen bond between the carbonyl oxygen between P2 and P1 in allophenylnorstatine and the side chain of Ser 79. Ser 79 is conserved in plasmepsins I, II and IV and is replaced by Ala in HAP. Asn 76 and Tyr 77 which also participate in the interaction are highly conserved among the plasmepsins. Structure-based thermodynamic analysis (Leavitt and Freire, 2001; Luque et al., 1998; Velazquez-Campoy et al., 2001a,b) indicates that the allophenylnorstatine scaffold by itself binds to plasmepsin II with a favourable binding enthalpy of 26.7 kcal/mol. The favourable binding enthalpy originates from a significant number of polar interactions. ˚ 2 of surface The allophenylnorstatine complex buries 790A 2 ˚ (41%) have a polar character and area of which 326A establishes three hydrogen bonds with the enzyme. Thus, from a structural and thermodynamic standpoint the allophenylnorstatine scaffold appears to satisfy the requirements for the core of an adaptive inhibitor since it interacts with the most conserved regions of the target in an enthalpically favourable fashion. From a structural standpoint, this scaffold provides the necessary anchoring points to develop inhibitor molecules that will exhibit very high affinity against at least three of the plasmepsins (plasmepsins I, II and IV) and a reasonably high affinity against HAP. Acknowledgements
Fig. 6. Docking of allophenylnorstatine scaffold into the plasmepsin II structure. Protein residues have been coloured according to their predicted interaction with the inhibitor using a rainbow scale (blue ¼ strongest interaction, red ¼ no interaction). The minimised allophenylnorstatine scaffold was docked into the binding site of plasmepsin II using the structure of the complex of plasmepsin II with H6680 as a template. The hydroxyl group of the allophenylnorstatine scaffold was initially placed between Asp 34 and Asp 214 at the same position of the hydroxyl group in the statine-based inhibitor. The docking procedure was carried out using the Affinity module of Insight II (MSI, San Diego, CA) and the CVFF force field for docking and scoring. The ligand was confined in the binding site using a flatbottomed energy well in order to minimise repulsion. Several independent docking procedures were carried out with two minimisation iterations between docking steps. The structure with the lowest energy was then energy minimised using the Discover module of Insight II by relaxing the inhibitor and the binding site residues.
The authors thank Professor Y. Kiso from Kyoto Pharmaceutical University for helpful discussions and for supplying us with the KNI compounds. This study was supported by grants from the National Institutes of Health GM 57144 and the National Science Foundation MCB9816661. References Banerjee, R., Liu, J., Beatty, W., Pelosof, L., Klemba, M., Goldberg, D.E., 2002. Four plasmepsins are active in the Plasmodium falciparum food vacuole, including a protease with an active-site histidine. Proc. Natl. Acad. Sci. USA 99, 990–5. Berry, C., Humphreys, J., Matharu, P., Granger, R., Horrocks, P., Moon, R.P., Certa, U., Ridley, R.G., Bur, D., Kay, J., 1999. A distinct member
1676
A. Nezami, E. Freire / International Journal for Parasitology 32 (2002) 1669–1676
of the aspartic proteinase gene family from the human malaria parasite Plasmodium falciparum. FEBS Lett. 447, 149–54. Carrol, C.D., Patel, H., Johnson, T.O., Guo, T., Orlowski, M., He, Z.M., Cavallaro, C.L., Guo, J., Oksman, A., Gluzman, I.Y., Connelly, J., Chelsky, D., Goldberg, D.E., Dolle, R.E., 1998. Identification of potent inhibitors of Plasmodium falciparum plasmepsin II from an encoded statine combinatorial library. Bioorg. Med. Chem. Lett. 8, 2315–20. Coombs, G.H., Goldberg, D.E., Klemba, M., Berry, C., Kay, J., Mottram, J.C., 2001. Aspartic proteases of Plasmodium falciparum and other parasitic protozoa as drug targets. Trends Parasitol. 17, 532–7. Eggleson, K.K., Duffin, K.L., Goldberg, D.E., 1999. Identification and characterization of falcilysin, a metallopeptidase involved in hemoglobin catabolism within the malaria parasite Plasmodium falciparum. J. Biol. Chem. 274, 32411–7. Francis, S.E., Banerjee, R., Goldberg, D.E., 1997a. Biosynthesis and maturation of the malaria aspartic hemoglobinases plasmepsins I and II. J. Biol. Chem. 272, 14961–8. Francis, S.E., Sullivan, D.J., Goldberg, D.E., 1997b. Hemoglobin metabolism in the malaria parasite Plasmodium falciparum. Annu. Rev. Microbiol. 51, 97–123. Freire, E., 1999. The propagation of binding interactions to remote sites in proteins. Analysis of the binding of the monoclonal antibody D1.3 to lysozyme. Proc. Natl. Acad. Sci. USA 96, 10118–22. Freire, E., 2002. Designing drugs against heterogeneous targets. Nat. Biotech. 20, 15–16. Goldberg, D.E., 1993. Hemoglobin degradation in infected red blood cells. Semin. Cell Biol. 4, 355–61. Haque, T.S., Skillman, G., Lee, C.E., Habashita, H., Gluzman, I.Y., Ewing, T.J.A., Goldberg, D.E., Kuntz, I.D., Ellman, J.A., 1999. Potent, lowmolecular-weight non-peptide inhibitors of malarial aspartyl protease plasmepsin II. J. Med. Chem. 42, 1428–40. Hilser, V.J., Freire, E., 1996. Structure based calculation of the equilibrium folding pathway of proteins. Correlation with hydrogen exchange protection factors. J. Mol. Biol. 262, 756–72. Jiang, S., Prigge, S.T., Wei, L., Gao, Y.E., Hudson, T.H., Gerena, L., Dame, J.B., Kyle, D.E., 2001. New class of small nonpeptidyl compounds blocks Plasmodium falciparum development in vitro by inhibiting plasmepsins. Antimicrob. Agents Chemother. 45, 2577–84.
Leavitt, S., Freire, E., 2001. Direct measurement of protein binding energetics by isothermal titration calorimetry. Curr. Opin. Struct. Biol. 11, 560–6. Luker, K.E., Francis, S.E., Gluzman, I.Y., Goldberg, D.E., 1996. Kinetic analysis of plasmepsins I and II, aspartic proteases of the Plasmodium falciparum digestive vacuole. Mol. Biochem. Parasitol. 79, 71–78. Luque, I., Freire, E., 2000. The structural stability of binding sites. Consequences for binding affinity and cooperativity. Proteins 4, 63–71. Luque, I., Gomez, J., Semo, N., Freire, E., 1998. Structure-based thermodynamic design of peptide ligands. Application to peptide inhibitors of the aspartic protease endothiapepsin. Proteins 30, 74–85. Nezami, A., Luque, I., Kimura, T., Kiso, Y., Freire, E., 2002. Identification and characterization of allophenylnorstatine-based inhibitors of plasmepsin II, an anti-malarial target. Biochemistry 41, 2273–80. Rosenthal, P.J., McKerrow, J.H., Aikawa, M., Nagasawa, H., Leech, J.H., 1988. A malarial cysteine proteinase is necessary for hemoglobin degradation by Plasmodium falciparum. J. Clin. Invest. 82, 1560–6. Silva, A.M., Lee, A.Y., Gulnik, S.V., Majer, P., Collins, J., Bhat, T.N., Collins, P.J., Cachau, R.E., Luker, K.E., Gluzman, I.Y., Francis, S.E., Oksman, A., Goldberg, D.E., Erickson, J.W., 1996. Structure and inhibition of plasmepsin II, a hemoglobin-degrading enzyme from Plasmodium falciparum. Proc. Natl. Acad. Sci. USA 93, 10034–9. Todd, M.J., Luque, I., Velazquez-Campoy, A., Freire, E., 2000. The thermodynamic basis of resistance to HIV-1 protease inhibition. Calorimetric analysis of the V82F/I84V active site resistant mutant. Biochemistry 39, 11876–83. Velazquez-Campoy, A., Freire, E., 2001. Incorporating target heterogeneity in drug design. J. Cell. Biochem. S37, 82–88. Velazquez-Campoy, A., Kiso, Y., Freire, E., 2001a. The binding energetics of first and second generation HIV-1 protease inhibitors: implications for drug design. Arch. Biochim. Biophys. 390, 169–75. Velazquez-Campoy, A., Luque, I., Freire, E., 2001b. The application of thermodynamic methods in drug design. Thermochim. Acta 380, 217–27. Wyler, D.J., 1993. Malaria overview and update. Clin. Infect. Dis. 16, 449– 58.