Synthesis of new (-)-Bestatin-based inhibitor libraries reveals a novel binding mode in the S1 pocket of the essential malaria M1 metalloaminopeptidase

Abstract

The malarial PfA-M1 metallo-aminopeptidase is considered a putative drug target. The natural product dipeptide mimetic, bestatin, is a potent inhibitor of PfA-M1. Herein we present a new, efficient and high-yielding protocol for the synthesis of bestatin derivatives from natural and unnatural N-Boc-D-amino acids. A diverse library of bestatin derivatives was synthesized with variants at the sidechain of either the α-hydroxy-β-amino acid (P1) or the adjacent natural α-amino acid (P1’). Surprisingly, we found that extended aromatic sidechains at the P1 position resulted in potent inhibition against PfA-M1. To understand these data, we determined the X-ray co-crystal structures of PfA-M1 with two derivatives having either a Tyr(OMe) 15 or Tyr(OBzl) 16 at the P1 position and observed substantial inhibitor-induced rearrangement of the primary loop within the PfA-M1 pocket that interacts with the P1 sidechain. Our data provide important insights for the rational design of more potent and selective inhibitors of this enzyme that may eventually lead to new therapies for malaria.

Introduction

The human parasitic pathogen Plasmodium falciparum accounts for most of the 1-2 million malaria related deaths per year.¹ This parasite has a complex life cycle involving several stages of growth: including a mosquito vector stage and a mammalian stage. The human asexual erythrocytic phase (blood stage) is the cause of malaria-associated pathology. The blood stage is a recurring lytic cycle lasting 48 hours, wherein vast amounts of host hemoglobin are taken up by the parasite for mostly metabolic purposes.² The roles of many endopeptidases, such as the digestive vacuole plasmepsins and falcipains, have been well established, however these endopeptidases appear to be largely redundant.³ Interest has recently increased in understanding the roles of aminopeptidases, several of which are genetically essential, and may contribute to several biological processes, including the breakdown of hemoglobin and the removal of the terminal methionine from newly synthesized proteins.

Amongst the 100 peptidases in the P. falciparum genome, there are eight metallo-aminopeptidases (MAPs). Four of these are methionine aminopeptidases that presumably have a housekeeping role in the removal of the initiator methionine from newly synthesized polypeptides.^4,5 Of the remaining four peptidases, PfA-M1 (aminopeptidase N; M1 family) is thought to be an essential parasite enzyme and a potential therapeutic target.⁶ The P. falciparum PfA-M1 enzyme shares low percentage identity (26%) to its human ortholog, aminopeptidase N, highlighting that targeting this essential parasite peptidase over human enzymes may be possible. It is noteworthy that there are some conserved amino acids around the active site, especially those that comprise and surround the HEXXH Zn binding site motif, however, the amino acids comprising the binding pockets show significant diversity.

MAPs catalyze cleavage on the amino terminal of peptide chains. These enzymes are widely distributed in organisms from bacteria to human and play essential roles in protein maturation and regulation of the metabolism of bioactive peptides.^7-9 Although the MAP superfamily is quite large and divergent, most enzymes share a common mechanism, exploiting the coordination of one or two Zn atoms in the active site to activate water for nucleophilic attack of a peptide or protein substrate. Inhibitors of this superfamily therefore tend to incorporate a Zn-binding group into a peptide-like structure. There have been several inhibitor scaffolds used to target the MAP family including, most prominently, phosphinic acids,¹⁰ hydroxamic acids¹¹ and the bestatin family.^12,13 Both phosphinic acids and hydroxamic acids have the capacity to inhibit metallo-endopeptidases and peptide deformylase and therefore will be intrinsically more difficult to design in specificity for a target MAP. Thus, we focused on the bestatin scaffold since it is synthetically tractable, potent and specific for the MAP family.

(-)-Bestatin is a natural product of actinomycetes that potently inhibits multiple families of metalloaminopeptidases (MAPs) including the M1, M17 and M18 families.^14-17 Bestatin has been shown to modulate many biological pathways; importantly, bestatin has been shown to inhibit growth of P. falciparum parasites in culture.¹⁸ However, bestatin type inhibitors have not been exploited widely for construction of inhibitor libraries and have not been developed for use as activity–based probes. Moreover, few if any specific inhibitors have been designed for MAPs from any organism regardless of scaffold.¹⁹ We therefore aimed to diversify the main binding determinants of bestatin in order to expand the repertoire of tools with which to study MAPs. We specifically aimed to explore structure activity relationships with PfA-M1 as a first step toward optimizing the potency of bestatin-like inhibitors against this enzyme.

Bestatin resembles a Phe-Leu dipeptide substrate; however, the first residue contains a α-hydroxy-β-amino acid. Bestatin-based inhibitors coordinate the active zinc atom of MAPs and also form interactions using the side chains of both the α-hydroxy-β-amino acid and the adjacent alpha amino acid which bind into the S1 and S1’ active site pockets of the target enzyme, respectively. Furthermore, the amide between the two amino acids of bestatin forms hydrogen bonds with the enzyme backbone and in some cases one or more MAP glutamate residues coordinate the free amine of bestatin, as highlighted in a recent structure for E. coli aminopeptidase N.¹⁷ A recent co-crystal structure of bestatin bound to malaria PfA-M1 enzyme revealed a similar set of interactions between the enzyme and the inhibitor (Fig. 1).²⁰ Bestatin interacts with MAPs in a non-covalent, reversible manner, and in some cases is a slow-binding inhibitor, with initial formation of a low-affinity complex followed by a slow conformational change leading to a high affinity complex.^12,13

Fig. 1. Schematic of Bestatin binding to the active site of PfA-M1.

S1 and S1’ represent canonical peptidase binding pockets. Residues that comprise the S1 and S1’ pockets are listed alongside the half moon circle. PfA-M1 sidechains are represented by numbered amino acids. Solid lines=Metallo-bonds; Dashed lines=Hydrogen bonds; CA=alpha-carbon. Black spheres represent carbon atoms.

We previously developed a solid-phase strategy that allows for the diversification of the P1’ position of Bestatin.²¹ Herein we seek to expand on this previous work by the development of a facile and general method for the preparation of bestatin analogs that are diversified both at the leucine (P1’) and at the α-hydroxy-β-amino acid position (P1). We develop a small, directed inhibitor library to explore potency determinants at the S1 and the S1’ pocket 115 against the essential malaria M1-family aminopeptidase, PfA-M1, via kinetic and structural analysis and live parasite assays.

Results

Chemistry

Stemming from the previous work, we wanted to design a simple solid-phase strategy for the synthesis of (-)-Bestatin and its P1/P1’ derivatives for library generation (Schemes 1,3). We decided to utilize a Boc protection strategy at the amino group for cleavage from Wang resin, which allows the generation of the final product in one pot.

Scheme 1.

Synthesis of Bestatin P1’ derivatives.

Scheme 3.

Synthesis of Bestatin and its P1 derivatives

Construction of P1’ Bestatin Library

Our initial aim was to construct a diverse set of compounds by changing the amino acid at the P1’ position of bestatin. Accordingly, a library of P1’ bestatin derivatives were synthesized employing solid-phase peptide synthesis as depicted in Scheme 1. We chose to use Wang resin to obtain the product with a C-terminal carboxylate. Norleucine, Lysine(Z), p-benzyloxy phenyl alanyl, 2-naphthyl alanyl and biphenyl alanyl amino acids were coupled to the Wang resin using HATU, DIEA and catalytic DMAP. Fmoc cleavage was achieved using 2%DBU+2%DIEA in DMF for 20 minutes. Coupling of the α-hydroxy-β-amino acid required HATU in DIEA for 1 hr. Cleavage of the resin using 95%TFA: 2.5%TIS: 2.5%H₂O yielded the bestatin analogs 1-6 in 60 to 70% yields.

Synthetic Strategy for α-hydroxy-β-amino acids for P1 bestatin library

After successful completion of the P1’ library we were interested in exploring specificity determinants at the P1 amino acid sidechain, which requires the synthesis of α-hydroxy-β-amino acids. There have been several noteworthy routes reported for the synthesis of α-hydroxy-β-amino acids, all with limitations, including incompatibility with Boc solid-phase chemistry, which led us to develop a more practical approach. Previous routes include: i) chiron approaches starting from amino acids^22-25 or sugars,^26-29 ii) asymmetric catalysis,^30,31 iii) amino hydroxylation,^32,33 iv) selective reduction of ketones,³⁴ and v) chemoenzymolysis.^35,36 The Passerini approach³⁷ (a one pot, three component reaction) for the synthesis of (-)-Bestatin gave a 1.5:1 mixture of diastereomers at the hydroxy center, but is limited by difficulties in producing isocyanides in high yields and is severely limited in side chain options. Homologation of α-amino aldehydes via cyanohydrin is not feasible with a Boc protection at the amine.³⁸ Lastly, a majority of the methods reported utilized expensive chiral catalysts, had lengthy preparations due to many reaction steps, were low yielding, and were not feasible with a Boc protection on the amino group. Herein we report a simple, efficient and high yielding approach for the synthesis of the α-hydroxy-β-amino acid for bestatin and its P1 derivatives (Scheme 2).

Scheme 2.

Synthesis of alpha hydroxy beta amino acids

The key intermediate α-hydroxy-β-amino acid 13a-h was prepared in 5 steps from commercially available D-amino acid 7. The transformation began with the conversion of N-Boc-D-amino acids 7a-h into Weinreb amides 8a-h using N-methyl morpholine, EDAC.HCl and N,O-dimethyl hydroxyl amine hydrochloride in dichloromethane.³⁹ The use of a Weinreb amide is advantageous in that it readily gives an aldehyde upon treatment with a reducing agent without loss of optical purity. Reduction of the Weinreb amide 8a-h using lithium aluminium hydride in ether at -33°C yielded the aldehyde 9a-h.⁴⁰ Wittig methylenation of α-amino aldehydes using n-butyllithium and methyltriphenyl phosphonium bromide in tetrahydrofuran gave a complex mixture of products. We employed the olefination reaction under salt-free Wittig conditions using potassium bis(trimethylsilyl)amide (KHMDS) as a base to give the olefin 10a-h.⁴¹ Dihydroxylation of the alkene 10a-h with osmium tetroxide and N-methylmorpholine N-oxide gave a diastereomeric mixture of the diol 11a-h.⁴² We explored the usage of AD-mix-α and methane sulfonamide in t-BuOH:H₂O (1:1) for dihydroxylation reaction but it did not significantly enhance the selectivity. The diastereomeric ratios of the diols obtained (2S,3R and 2R,3R) when AD-mix-α was used as the dihydroxylating agent is reported in Supplementary Table I. The crude diol 11a-h was selectively oxidized at the primary hydroxyl functionality to the α-hydroxy-β-amino acid 12:13 using TEMPO, NaClO₂ and NaOCl in NaH₂PO₄.^43,44 Surprisingly, the secondary alcohol remained intact under the oxidation conditions (Scheme 2). Given prior published data that the (2S,3R) diastereomers of bestatin are much more inhibitory than the corresponding (2R,3R) diastereomer,^45,46 the active (2S,3R) diastereomers 13a-h were purified by reverse phase HPLC. The ¹H NMR spectra of the 13a-h (2S,3R) diastereomers had the −NH (doublet) and −Boc (singlet) protons shift at similar ppm in all the examples under study in comparison to the phenyl derivative which is well studied (except for the diphenyl and n-propyl, where one diastereomer was exclusively formed). A detailed study showing the shift of -NH and −Boc protons of the α-hydroxy-β-amino acids (individual diastereomers) is listed in Supplementary Table I. This synthesis is a straightforward method for the preparation of α-hydroxy-β-amino acids from α-D-amino acids with good yields.

Construction of P1 Bestatin Library

The α-hydroxy-β-amino acids 13a-h were converted to bestatin P1 derivatives via coupling with preloaded Fmoc-Leu Wang resin using HATU and DIEA. Cleavage of the Fmoc group was achieved using 20% piperidine in DMF. Global deprotection of the Boc group and cleavage from Wang resin was accomplished using 95%TFA:2.5%TIS:2.5%H₂O, which gave Bestatin 1 and its P1 derivatives 14-20 in good yields ranging from 60-70% (Scheme 3). Prior work demonstrated that the PfA-M1 MAP had broad specificity,⁴⁷ however published structural data with bestatin demonstrated that the S1 pocket was lined with hydrophobic residues packing well around the P1 Phe sidechain (Fig.1).⁴⁶ Using this information we synthesized Bestatin derivatives using a series of α-hydroxy-β-amino acids consisting of aliphatic and aromatic side chains in increasing size and rigidity. To make use of these new α-hydroxy-β-amino acids, we synthesized a library on preloaded Wang resin wherein the second position (P1’) was fixed to a leucine, the amino acid yielding the most potent activity against PfA-M1 at this position. Ultimately, we developed a small library to probe the structure-activity relationships of this P1 bestatin side chain with the S1 binding pocket of the target enzyme PfA-M1.

Biological Results

P1’ and P1 Bestatin Library Screening

To explore interactions between the sidechains at the P1 and P1’ sites on the bestatin scaffold with the enzyme, a series of focused inhibitor libraries were synthesized using natural and non-natural aliphatic and aromatic sidechains of differing length and rigidity (Schemes 1-3). The P1’ library was first tested in vitro against recombinant PfA-M1 and showed a trend that favored small hydrophobic residues at this site, with leucine displaying the most potent activity (Table I).

Table I.

Potency of Bestatin P1’ derivatives against recombinant PfA-M1.

	Ki (nM) PfA-M1
R’ =
isobutyl (bestatin) 1	190
n-butyl 2	900
4-benzyloxycarbonyl amino butyl 3	1450
p-benzyloxyphenyl alanyl 4 4	>30000
2-naphthyl alanyl 5	>30000
biphenyl alanyl 6	>30000

Each K_i value as determined from assays at 8 inhibitor concentrations spanning the range 0.3 30 μM; NI=No Inhibition

The P1 bestatin library was constructed with a P1’ leucine, which yielded the most potent activity against PfA-M1. All inhibitors were again tested in vitro against recombinant PfA-M1. From this data, it was observed that aliphatics, such as a leucine or norleucine, yielded poor inhibitors against the enzyme unlike peptide substrates (Table II).⁴⁷ Interestingly, some larger aromatic substitutions at the P1, such as Tyr(OBzl) (16) or Tyr(OMe) (15), were found to be potent. In fact, 15 was observed to be five-fold more potent than the parental bestatin with a K_i of 43nM. Larger rigid aromatic structures, such as naphthyl or biphenyl substituents, rendered inhibitors inactive, which may indicate that the lack of flexibility of these side chains caused an unavoidable steric clash with the target enzyme. However, more flexible large aromatic side chains, such as Tyr(OBzl), were quite potent against the enzyme, which is unexpected given that the structure of PfA-M1 and homologous enzymes suggested that the S1 pocket would be unlikely to accommodate side chains much larger than phenylalanine.²⁰ Lastly, basic substituents such as Lysine were also inactive (not shown), which is surprising given that P1 lysine-containing peptide substrates are processed by PfA-M1 since there is a glutamate at the top of the S1 pocket.²⁰ However, discrepancies between peptide substrates and bestatin analogs are not surprising, as bestatin contains a α-hydroxy-β-amino acid and thus presents the P1 side chain in a different manner than do peptide substrates.¹⁷

Table II.

Potency of Bestatin derivatives against recombinant PfA-M1 and P. falciparum

	Ki (nM) PfA-M1	IC50 (μM) live parasites
R =
p-methoxy phenyl 15	43	6.4
phenyl (bestatin) 1	190	3.2
p-benzyloxy phenyl 16	490	29
2-naphthyl 17	3000	>50
n-propyl 20	5300	>50
isopropyl 14	NI	>50
biphenyl 18	NI	>50
gem-diphenyl 19	NI	>50

Each K_i value was determined from assays conducted at four substrate concentrations, each with six concentrations of inhibitor. Duplicate determinations were carried out for compounds 1, 14 and 15 and in all cases the replicates were within 12% of the mean.

All P1 inhibitors were then tested against live cultures of P. falciparum. The most potent inhibitors found in the in vitro studies also showed moderate potency against P. falciparum in culture with IC₅₀ values in the micromolar range (Table II). The most potent compound Tyr(OMe) (15) against the enzyme was found to have an IC₅₀ around 6 micromolar against parasites, which was slightly less potent than bestatin perhaps due to the lowered permeability caused by the increased hydrophilicity with the introduction of the methoxy group. In addition another Plasmodium aminopeptidase, Pf-LAP, is known to be targeted by bestatin, which could account for the difference between IC₅₀ values and in vitro data.

Structural analysis of bestatin derivatives 15 and 16 bound to PfA-M1

The X-ray crystal structure of PfA-M1 bound to bestatin revealed that the S1 subsite in the enzyme accommodated a P1 phenylalanine well with tight packing of enzyme side chains around the phenylalanine side chain.²⁰ The S1 pocket is lined by hydrophobic residues (Val459, Tyr575, Met1034) and a single polar residue (Glu572). Residues 570–575 (including Glu572 and Tyr575) are located on a loop central to the active site of the enzyme. Comparison of the unbound versus the bestatin-bound PfA-M1 complex revealed that the sidechain of only one of the residues (Met1034) that lines the PfA-M1 S1 pocket, moved slightly upon ligand binding. It was thus somewhat surprising that substitution of the Bestatin P1 Phe position with larger groups, Tyr(OMe) 15 or Tyr(OBzl) 16, which we would predict would sterically clash with the enzyme, resulted in molecules with good inhibitory activity.

To address this question, we determined the 1.8 Å X-ray crystal structures of PfA-M1 bound to compounds 15 and 16 (Fig. 2). Comparison of these co-crystal structures with that of bestatin-bound PfA-M1 revealed that the S1 pocket of the enzyme is capable of undergoing a major conformational change in response to binding a larger P1 moiety (Fig. 2A). Remarkably, when the PfA-M1 enzyme was inhibited by Tyr(OBzl) 16, the Cα atom of Glu572 moved by ~5 Å, respectively. In fact, this loop switched to an entirely different conformation with the Cα atom of the neighboring residue, Phe574, moving by ~1.6 Å. With the smaller 15 compound (Tyr(OMe), this loop movement was less pronounced with the Cα atom of Glu572 moving only 1.2 Å to accommodate the inhibitor. Of note, however, in the latter structure no electron density was observed for the side chain of Glu572, suggesting a highly mobile side chain. Two of the surrounding residues of the loop (Met571 and Asn573) retain the same conformation as the bestatin-bound enzyme, however Phe574 also shifts ~ 1 Å. In contrast, the S1’ pocket and at the Zn²⁺ atom the interactions are unchanged from that seen in the bestatin bound structure, with the α-hydroxy group and neighboring carbonyl coordinating the catalytic Zn²⁺. Accessible surface area calculations revealed that inhibitor 16 also buried a larger surface area (83.7 Ų), in comparison to 15 (65.5 Ų) and bestatin (23.8 Ų, Fig. 2B).²⁰ These accessible surface area maps highlight the significant movement in the S1 pocket of the enzyme to accommodate the larger unnatural side chains.

Fig 2. X-ray crystal structures of PfA-M1 bound to Tyr(OMe) (15) and Tyr(OBzl) (16).

(A) The cartoon diagram of the carbon atoms of three X-ray crystal structures of S1 pocket of PfA-M1 bound to bestatin (grey); Tyr(OMe) (15) (magenta) and Tyr(OBzl) (16) (green). The comparative movement of the cα-atom of Glu572 is indicated by the dashed lines. (B) The electrostatic potential surface of S1 pocket of PfA-M1 with bestatin, Tyr(OMe) (15) and Tyr(OBzl) (16) bound in active site. The zinc ion is shown as black sphere, the carbon atoms of inhibitors in grey and the position of Glu572 is indicated in each panel. Surfaces were color coded according to electrostatic potential (calculated by the Poisson-Boltzmann solver within CCP4MG). Lys and Arg residues were assigned a single positive charge, and Asp and Glu residues were assigned a single negative charge; all other residues were considered neutral. The calculation was done assuming a uniform dielectric constant of 80 for the solvent and 2 for the protein interior. The ionic strength was set to zero. The colour of the surface represents the electrostatic potential at the protein surface, going from blue (potential of +10 kT/e) to red (potential of -10 kT/e), where T is temperature, e is the charge of an electron, and k is the Boltzmann constant. The probe radius used was 1.4 Å.

Discussion

We anticipate that the strategy presented herein for the synthesis of P1 derivatives of bestatin will provide a practical method for a mixed solution and solid-phase route to synthesize novel bestatin-based inhibitors, which will be useful for inhibition of variety of MAPs. During the synthesis, it was surprising that we did not observe better stereoselectivity using the Sharpless AD-mix-α reagent. Since the (2S,3R) diastereomer was required for good inhibition of PfA-M1, it would be useful to optimize this step in the future. The lack of AD-mix-α stereoselectivity may be due to the steric bulkiness of the R group and the fact that the olefin was terminal.

Screening of the inhibitor libraries revealed that the S1 pocket accommodated some large aromatic side chains well; the (OMe)Tyr derivative (15) was the most potent inhibitor, however, the enzyme was able to accommodate a (OBzl)Tyr side chain (16), although the K_i decreased, possibly caused by the energetic cost of such a large movement in the enzyme pocket. Given this information, future efforts should be aimed at exploiting the novel binding mode of the P1 α-hydroxy-β-amino acid side chain of bestatin against PfA-M1. Derivatization of the P1 side chain using a wider variety of long hydrophobic side chains between the lengths of compounds 15 and 16 would represent one strategy. In addition, the S1’ pocket favored smaller aliphatic side chains and thus exploration of a series of small non-natural amino acids at the P1’ position may further increase activity. Other possibilities to enhance the potency against the enzyme and in live cells or animals would entail changing the peptide backbone and, thus, reducing the peptide character of the bestatin pharmacophore. This can be accomplished in variety of ways, however, the central idea would entail the elimination of the central amide bond between the α-hydroxy-β-amino acid and the leucine to create either: i) carbapeptides ii) azapeptides, or iii) vinylogous peptides.

Structural analysis of PfA-M1 bound to bestatin, 15 and 16 indicate that the S1 pocket loop is mobile with high B-factors for residues 570-575. Interestingly, structural characterization of PfA-M1 homologs, to date, has not identified this movement in the S1 pocket, and do not indicate any enhanced mobility of this region. Taken together, our current SAR data revealed that there is unprecedented plasticity of the S1 binding pocket in the PfA-M1 enzyme that permits the enzyme to accommodate larger side chains and that there is considerably more binding energy to be gained by increasing the size of the original phenylalanyl side chain of the parental bestatin. These data contrast with previous studies on this enzyme family, which generally reveal little change between inhibitor bound and unbound forms.²⁰ This information is important for future inhibitor design efforts since it may open new routes to building selective inhibitors that target the malaria M1 enzyme over mammalian counterparts.

It is noteworthy that there are 13 M1 family metallo-aminopeptidase members in humans, which may complicate future specificity studies. However, none of the enzymes share much more than ~26% identity with PfA-M1. Studies on various mammalian M1 orthologs (pig, rat and human) show significant variation in inhibitor activity between the three closely related enzymes, giving an indication that achieving selectivity is plausible.⁴⁸ Aminopeptidase N, the human M1 ortholog most similar to PfA-M1 and shares only 26% identity with the Plasmodium enzyme with extensive variation at the N- and C-terminal domains, indicating that inhibitor access to the active site may be quite altered. Unfortunately, there is currently no structural information available for the human M1 enzyme. Ultimately designing inhibitors that are specific for the parasite enzyme over the human M1 family members would be ideal, and there is hope for this possibility considering the low identity between the parasite and human enzymes.

In conclusion, we believe that this data provides robust data to indicate that the bestatin scaffold provides an excellent pharmacophore for future antimalarial inhibitor development against the PfA-M1 enzyme. Potentially this scaffold can also be optimized against the other essential MAPs such as the M17 enzyme (Pf-LAP) to generate a single scaffold that potently inhibits multiple enzyme targets, which may reduce the ability of the parasite to generate resistance. Lastly, evidence that the S1 pocket of this M1 family peptidase is quite flexible upon inhibitor binding may be exploited to develop more potent or selective inhibitors for other members of this family, including other parasitic and human orthologs.

Experimental Procedures

Recombinant PfA-M1

Details of the production of P. falciparum PfA-M1 in E. coli and its purification will be described elsewhere (D. Ragheb and M. Klemba, manuscript in preparation).

Determination of inhibitor K_i and IC₅₀ values

K_i values were determined according to the method of Dixon⁴⁹ using the fluorogenic substrate L-leucyl-7-amido-4-methylcoumarin (Leu-AMC). Briefly, assays contained 50 mM HEPES pH 7.5, 100 mM NaCl, 0.1% Triton X-100, 0.9 nM PfA-M1 and Leu-AMC concentrations of 12.5, 16.7, 25 and 50 μM in a volume of 200 μL. At each substrate concentration, a range of inhibitor concentrations was assayed, typically ranging from no inhibitor to an inhibitor concentration corresponding to ~4K_i. Changes in fluorescence were measured in a 96-well plate using a Victor³ microplate fluorometer (PerkinElmer) equipped with an excitation filter of 380 nm and an emission filter of 460 nm. While bestatin can be a slow-binding inhibitor of some aminopeptidases, we found that a steady state had been attained in the PfA-M1 assays by 40 minutes. Data were plotted as 1/rate vs. inhibitor concentration for each substrate concentration and a linear fit was calculated by non-linear regression using Kaleidagraph 4.1. A secondary plot was then made with intercepts or slopes vs. 1/[S], where S is substrate. The K_i value was determined from the ratio of the slopes of the secondary plots (intercept:slope). Inhibitors were assessed for uncompetitive inhibition by constructing a/v vs. i plots.⁵⁰ All inhibitors tested exhibited purely competitive inhibition.

IC₅₀ values were determined by assaying inhibitors at various concentrations in the range 0.014 to 30 μM with a Leu-AMC concentration of 44 μM. Data were plotted as ν_i/ν_o vs inhibitor concentration, where ν_i is the inhibited rate and ν_o is the rate in the absence of inhibitor, and fit by non-linear regression to the expression ν_i/ν_o = 1/(1 + [I]/IC₅₀).

Antimalarial Activity Assay

Plasmodium falciparum strain 3D7 was cultivated in RPMI 1640 medium (with L-glutamine) containing 20 mM HEPES, 32 mM NaHCO₃, 10% Albumax II (Invitrogen) in erythrocytes at 4% hematocrit.⁵¹ Quantitative assessment of antimalarial activity was determined by measurement of SYBR Green fluorescence. Assays were performed in 96-well black, clear-bottom plates. Briefly, 75 mL of culture (in phenol red-free media RPMI 1640) was aliquoted to each well at a starting parasitemia of 0.5%. Next, serial dilutions of compounds in media were added to each well to bring the final volume to 150 mL. All drugs were assayed in triplicate. Cultures were incubated at 37 °C for 72 h. At the end of incubation 10x lysis buffer (20 mM Tris-HCl, 5 mM EDTA, 0.16% Saponin wt/vol, 1.6% Triton X vol/vol) with 10x SYBR Green I (Invitrogen; from 10,000 supply) was added to each well and incubated for 1 h at room temperature. Levels of SYBR Green fluorescence, indicating parasite growth, were measured on a Tristar LB 941 microplate reader (Berthold Technologies). IC₅₀ values were determined using GraphPad Prism.

X-ray Crystallography

rPf-M1 enzyme was purified and crystallized as previously described.²⁰ Crystals of the PfA-M1-15 and PfA-M1-16 complex were obtained by soaking PfA-M1 crystals overnight in mother liquor containing 1 mM ligand. Data were collected at 100K using synchrotron radiation at the Australian synchrotron micro crystallography beamline 3ID1. Diffraction images were processed using MOSFLM⁵² and SCALA⁵³ from the CCP4 suite.⁵⁴ 5 % of each dataset was flagged for calculation of R_Free ⁵⁵ with neither a sigma nor a low-resolution cut-off applied to the data. A summary of statistics is provided in Supplementary Table II. Subsequent crystallographic and structural analysis was performed using the CCP4i interface⁵⁶ to the CCP4 suite,⁵² unless stated otherwise. The inhibitor complex was initially solved and refined against the unbound PfA-M1 structure (protein atoms only) as described previously⁴⁵ and clearly showed unbiased features in the active site for both structures. Superpositions were performed using MUSTANG.⁵⁷ Pymol was used to produce structural representations (http://www.pymol.org).⁵⁸ Hydrogen bonds (excluding water-mediated bonds), were calculated using Lig_contact and CONTACT.⁵⁴ CCP4MG was used to produce electrostatic diagrams.⁴⁸ Stereo diagram of both inhibitors showing electron density is represented in supplementary Fig 1. The coordinates and structure factors will be available from the Protein Data Bank (3Q43.pdb and 3Q44.pdb). Raw data and images will be available from TARDIS.⁵⁹

Chemistry

Column chromatography was performed on silica gel, fisher grade (170-400 mesh). TLC plates were visualized with UV, in an iodine chamber, or with phosphomolybdic acid, ninhydrin unless otherwise noted. Melting points were recorded on a Mel-Temp apparatus and are uncorrected. ¹H NMR and ¹³C NMR spectra were recorded on a Bruker Advance DMX-360 spectrometer. The chemical shifts are reported in parts per million (ppm, δ). The coupling constants are reported in hertz (Hz). The purity of all final compounds was established to be >95% pure by a combination of TLC R_f values in several solvent systems and HPLC. Additionally the absence of any extraneous peaks in the proton NMR spectrum confirmed the high level of purity. The N-Boc-(2S,3R)-2-hydroxy-3-amino-5-methylhexanoic acid was purchased from CNH technologies.

Solid-phase peptide chemistry was conducted in polypropylene cartridges, with 2-way Nylon stopcocks (Biotage, VA). The cartridges were connected to a 20 port vacuum manifold (Biotage, VA) that was used to drain excess solvent and reagents from the cartridge. HRMS were measured at the University of California Riverside Mass Spectrometry Service Center. All solvents were reagent grade. Unless otherwise stated, all reagents were purchased from commercial sources and used without additional purification. Reverse phase HPLC was conducted on a C18 column using an Agilent 1200 HPLC.

Purifications were performed at room temperature and compounds were eluted with a concentration gradient 0-70% of acetonitrile (0.1% Formic acid). LC/MS data were acquired using N LC/MSD SL system (Agilent). All compounds were purified to greater than 95% purity as assessed by LC analysis (see Supplemental Materials for LCMS traces).

General procedure for the synthesis of Bestatin (1) and its P1’ derivatives (2-6)

Solid-phase peptide synthesis was performed on Wang resin. The first amino acid was coupled to the Wang resin using HATU, DIEA and cat. DMAP for 1 h. Coupling of the α-hydroxy-β-amino acid required HATU for 1 hr. Fmoc protecting group was deprotected with 20% piperidine/DMF (30 min). To cleave products from resin, a solution of 95%TFA:2.5%TIS:2.5%H₂O was added to the resin. After standing for 2 h, the cleavage mixture was collected, and the resin was washed with fresh cleavage solution. The combined fractions were evaporated to dryness and the product was purified by reverse phase-HPLC. Fractions containing product were pooled and lyophilized. The yields of bestatin (1) and its derivatives (2-6) ranged from 60-70% respectively after HPLC purification.

HRMS calcd for PheNleOH 2 C₁₆H₂₅N₂O₄ 309.1809 (M⁺+H), found 309.1801.

HRMS calcd for PheLys(Z)OH 3 C₂₄H₃₂N₃O₆ 458.2286 (M⁺+H), found 458.2278.

HRMS calcd for PheTyr(Bzl)OH 4 C₂₆H₂₉N₂O₅ 449.2071 (M⁺+H), found 449.2068.

HRMS calcd for PheNaphOH 5 C₂₃H₂₅N₂O₄ 393.1809 (M⁺+H), found 393.1816.

HRMS calcd for PheBipOH 6 C₂₅H₂₇N₂O₄ 419.1965 (M⁺+H), found 419.1959.

Synthesis of α-hydroxy-β-aminoacids

General Procedure for the Synthesis of N-Methoxy-N-methyl Acetamide (Weinreb amides)

In an oven-dried, 50-mL, two-neck, round-bottom flask equipped with a magnetic stirring bar and argon balloon was placed N-Boc-D-amino acid (1.997 g, 6.33 mmol) in dry CH₂Cl₂ (15 mL) and the solution was cooled to -15 °C. N,O-dimethylhydroxylamine hydrochloride (0.636 g, 6.52 mmol) was added, followed by N-methylmorpholine (0.72 mL, 6.55 mmol). After 5 min 1-(3-methylaminopropyl-3-ethylcarbodiimide hydrochloride (1.238 g, 6.46 mmol) was added in 5 portions over 30 min. The reaction was warmed to room temperature, and after 5 h H₂O (2 mL) was added and the solution was extracted with CH₂Cl₂ (3 × 5 mL). The combined organic phases were washed with brine (5 mL), dried (Na₂SO₄), and concentrated. The residue was purified by flash chromatography.

(R)-N-(tert-butoxycarbonyl)-2-amino-3-phenyl-N-methoxy-methylpropionamide (8b)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 5/1) to give 8b (96%) as an oily liquid; [α]_D²⁵ = -24.26 (c = 0.68, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.39 (s, 9H), 2.87-2.90 (m, 1H), 3.02-3.08 (m, 1H), 3.16 (s, 3H), 3.66 (s, 3H), 4.94 (br s, 1H), 5.15 (br s, 1H), 7.16-7.30 (m, 5H); ¹³C NMR (90 MHz, CDCl₃) δ 28.5, 32.2, 32.3, 39.1, 51.8, 61.7, 79.8, 127.0, 128.6, 129.7, 136.9, 155.4.

(R)-N-(tert-butoxycarbonyl)-2-amino-3-(p-methyloxyphenyl)-N-methoxy-N-methylpropionamide (8c)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 5/1) to give 8c (97%) as an oily liquid; [α]_D²⁵ = -22.12 (c = 1.08, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.39 (s, 9H), 2.82-2.86 (m, 1H), 2.97-3.02 (m, 1H), 3.16 (s, 3H), 3.66 (s, 3H), 3.77 (s, 3H), 4.90 (br s, 1H), 5.14 (br s, 1H), 6.81 (d, J = 8.3 Hz, 2H), 7.08 (d, J = 8.3 Hz, 2H); ¹³C NMR (90 MHz, CDCl₃) δ 28.5, 32.3, 38.2, 51.8, 55.4, 61.7, 79.7, 114.0, 128.8, 130.6, 155.4, 158.7; HRMS calcd for C₁₇H₂₇N₂O₅ (M⁺+H) = 339.1914, found = 339.1912.

(R)-N-(tert-butoxycarbonyl)-2-amino-3-(p-benzyloxyphenyl)-N-methoxy-N-methylpropionamide (8d)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 5/1) to give 8d (85%) as a colourless solid; m.pt.: 102 °C; [α]_D²⁵ = -14.54 (c = 0.55, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.39 (s, 9H), 2.80-2.85 (m, 1H), 2.96-3.01 (m, 1H), 3.15 (s, 3H), 3.64 (s, 3H), 4.89 (br s, 1H), 5.02 (s, 2H), 5.13 (s, 1H), 6.88 (d, 2H, J = 9.0 Hz), 7.07 (d, 2H, J = 9.0 Hz), 7.25-7.42 (m, 5H); ¹³C NMR (90 MHz, CDCl₃) δ 28.5, 32.3, 38.2, 38.2, 51.8, 61.7, 70.2, 79.8, 115.0, 127.6, 128.1, 128.7, 129.2, 130.7, 137.4, 158.0, 207.1; HRMS calcd for C₂₃H₃₀N₂O₅ (M⁺+H) = 415.2233, found = 415.2229.

(R)-N-(tert-butoxycarbonyl)-2-amino-3-(2-naphthyl)-N-methoxy-N-methylpropionamide (8e)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 5/1) to give 8e (93%) as a colourless solid; m.pt.: 103 °C; [α]_D²⁵ = -20.19 (c = 1.02, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.35 (s, 9H), 3.01-3.25 (m, 5H), 3.65 (s, 3H), 5.04 (br s, 1H), 5.20 (br s, 1H), 7.31 (d, J = 10.0 Hz, 1H), 7.39-7.46 (m, 2H), 7.62 (s, 1H), 7.75-7.80 (m, 3H); ¹³C NMR (90 MHz, CDCl₃) δ 28.2, 28.5, 32.3, 39.2, 51.7, 61.8, 79.8, 125.7, 126.2, 127.8, 127.8, 128.1, 128.3, 132.6, 133.7, 134.3, 172.6, 207.1.

(R)-N-(tert-butoxycarbonyl)-2-amino-3-(p-phenylphenyl)-N-methoxy-N-methylpropionamide (8f)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 5/1) to give 8f (87%) as a colourless solid; m.pt.: 106 °C; [α]_D²⁵ = -11.37 (c = 0.73, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.39 (s, 9H), 2.92-2.94 (m, 1H), 3.07-3.13 (m, 1H), 3.19 (s, 3H), 3.69 (s, 3H), 4.99 (br s, 1H), 5.19 (br s, 1H), 7.23-7.58 (m, 9H); ¹³C NMR (90 MHz, CDCl₃) δ 28.3, 28.5, 32.3, 32.3, 38.7, 38.9, 51.7, 61.8, 79.8, 118.9, 127.2, 127.3, 127.4, 127.6, 128.9,130.1, 135.9, 139.9, 141.2, 145.6, 155.4, 172.5; HRMS calcd for C₂₂H₂₉N₂O₄ (M⁺+H) = 385.2127, found = 385.2116.

(R)-N-(tert-butoxycarbonyl)-2-amino-3,3-diphenyl-N-methoxy-N-methylpropionamide (8g)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 4/1) to give 8g (88%) as a colourless solid; m.pt.: 83 °C; ¹H NMR (360 MHz, CDCl₃) δ 1.31 (s, 9H), 2.93 (s, 3H), 3.61 (s, 3H), 4.33 (d, 1H, J = 10.8 Hz, -NH), 5.01 (d, 1H, J = 10.1 Hz, -CH(Ph)₂), 5.68 (t, 1H, J = 10.1 Hz, NHCHCO), 7.15-7.36 (m, 10H); ¹³C NMR (90 MHz, CDCl₃) δ 28.4, 28.6, 31.9, 52.0, 54.7, 61.9, 79.8, 79.9, 126.9, 127.0, 128.4, 128.6, 128.8, 129.0, 129.4, 140.4, 140.6, 155.2, 172.6; HRMS calcd for C₂₂H₂₉N₂O₄ (M⁺+H) = 385.2127, found = 385.2119.

(R)-N-(tert-Butoxycarbonyl)-2-amino-3-propyl-N-methoxy-N-methylpropionamide (8h)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 4/1) to give 8h (90%) as an oily liquid; ¹H NMR (360 MHz, CDCl₃) δ 0.89 (t, 3H, J = 6.8 Hz), 1.26-1.55 (m, 14H), 1.66-1.72 (m, 1H), 3.21 (s, 3H), 3.77 (s, 3H), 1.09 (brs, 1H), 5.15 (d, 1H, J = 7.6 Hz).

General Procedure for the Wittig reaction

In an oven-dried, 50-mL, single-necked, round-bottom flask equipped with a magnetic stirring bar, rubber septum, and argon balloon was placed weinreb amides (1.860 g, 5.19 mmol) in anhydrous Et₂O (20 mL). The solution was cooled to -33 °C and LiAlH₄ (1 M in THF, 6.2 mL, 6.23 mmol) was added dropwise, and the reaction mixture was warmed to 0 °C. After 45 min, the solution was cooled to -33 °C and 10% aq. KHSO₄ (10 mL) was added to quench the reaction. After warming to room temperature 1 N HCl (10 mL) was added, and the organic phases were separated, the aqueous phase was extracted with EtOAc (2 × 10 mL), and the combined organic phases were dried (MgSO₄) and concentrated to yield the crude aldehyde. In a separate 50-mL, single-necked, round-bottom flask was placed methyltriphenylphosphonium bromide (3.65 g, 10.24 mmol) in THF (20 mL), and KHMDS (0.5 M in toluene, 19.5 mL, 9.76 mmol) was added. The resulting yellow suspension was stirred at room temperature for 2 h, then cooled to -78 °C and the crude aldehyde (1.75 g, 5.85 mmol) in THF (5 mL) was added dropwise. It was warmed to room temperature over 2 h, and quenched with MeOH (5 mL). The resulting mixture was poured into a mixture of saturated potassium sodium tartarate and water (1:1, 15 mL), extracted with Et₂O (2 × 50 mL), dried (MgSO₄) and concentrated. The residue was purified by flash chromatography to afford the respective alkenes.

(R)-N-(tert-Butoxycarbonyl)-3-amino-4-phenyl-1-butene (10b)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10b (71%) as a colorless solid; m. pt.: 62 °C; [α]_D²⁵ = -37.19 (c = 0.28, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.39 (s, 9H), 2.83 (d, J = 6.5 Hz, 1H), 4.40 (br s, 1H), 5.05-5.11 (m, 1H), 5.74-5.83 (m, 1H), 7.16-7.30 (m, 5H).

(R)-N-(tert-Butoxycarbonyl)-3-amino-4-(p-methyloxyphenyl)-1-butene (10c)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10c (67%) as a colorless solid; m.pt: 58 °C ; [α]_D²⁵ = -36.33; ¹H NMR (360 MHz, CDCl₃) δ 1.41 (s, 9H), 2.77 (d, J = 6.5 Hz, 2H), 3.78 (s, 3H), 4.36 (br s, 1H), 5.06-5.12 (m, 2H), 5.74-5.83 (m, 1H), 6.82 (d, J = 8.6 Hz, 2H), 7.09 (d, J = 8.6 Hz, 2H); ¹³C NMR (90 MHz, CDCl₃) δ 28.6, 40.8, 40.8, 53.8, 54.0, 55.5, 79.6, 114.0, 114.9, 129.6, 130.7, 138.4, 155.5, 158.5; HRMS calcd for C₁₆H₂₄NO₃ (M⁺+H) = 278.1751, found = 278.1749.

(R)-N-(tert-Butoxycarbonyl)-3-amino-4-(p-benzyloxyphenyl)-1-butene (10d)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10d (71%) as a colorless solid; m. pt: 87 °C; [α]_D²⁵ = -29.62 (c = 0.53, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.40 (s, 9H), 2.77 (d, J = 6.7 Hz, 2H), 4.40 (br s, 1H), 5.04 (s, 2H), 5.06-5.12 (m, 2H), 5.74-5.83 (m, 1H), 6.90 (d, J = 11.2 Hz, 2H), 7.08 (d, J = 11.2 Hz, 2H), 7.30-7.44 (m, 5H); ¹³C NMR (90 MHz, CDCl₃) δ 28.5, 40.7, 53.8, 70.2, 79.4, 114.8, 114.9, 127.6, 128.0, 128.7, 130.0, 130.7, 137.3, 138.4, 155.4, 157.7; HRMS calcd for C₂₂H₂₇NO₃ (M⁺+Na) = 376.1889, found 376.1879.

(R)-N-(tert-Butoxycarbonyl)-3-amino-4-(2-naphtyl)-1-butene (10e)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10e (80%) as a colorless solid; m. pt.: 104 °C; [α]_D²⁵ = -42.46 (c = 0.32); ¹H NMR (360 MHz, CDCl₃) δ 1.38 (s, 9H), 2.97-3.01 (m, 2H), 4.50 (br s, 1H), 5.07–5.14 (m, 2H), 5.79-5.87 (m, 1H), 7.33 (d, J = 9.7 Hz, 1H), 7.40-7.47 (m, 2H), 7.63 (s, 1H), 7.76-7.82 (m, 3H); ¹³C NMR (90 MHz, CDCl₃) δ 28.5, 41.9, 53.5, 115.1, 125.7, 126.2, 127.8, 127.8, 128.1, 128.2, 128.3, 132.5, 133.7, 135.2, 138.3, 155.4, tertiary tert-butyl carbon not observed; HRMS calcd for C₁₉H₂₃NO₂ (M⁺+Na) = 320.1624, found 320.1620.

(R)-N-(tert-Butoxycarbonyl)-3-amino-4-(p-phenylphenyl)-1-butene (10f)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10f (67%) as a colorless solid; ¹H NMR (360 MHz, CDCl₃) δ 1.40 (s, 9H), 2.88 (d, J = 5.8 Hz, 2H), 4.47 (br s, 2H), 5.09-5.16 (m, 2H), 5.79-5.88 (m, 1H), 7.24-7.59 (m, 9H); ¹³C NMR (90 MHz, CDCl₃) δ 28.6, 41.4, 115.0, 127.2, 127.3, 127.4, 129.0, 130.2, 136.8, 138.3, 139.6, 141.1, 155.4.

(R)-N-(tert-Butoxycarbonyl)-3-amino-5,5-diphenyl-1-pentene (10g)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10g (78%) as a colorless solid; m.pt.: 98 °C; [α]_D²⁵ = -56.89 (c = 0.58, CHCl₃); ¹H NMR (360 MHz, CDCl₃) δ 1.36 (s, 9H), 3.97 (d, J = 9.4 Hz, 2H), 4.39 (br s, 1H), 4.99 (br s, 1H), 5.03-5.16 (m, 2H), 5.71-5.80 (m, 1H), 7.15-7.30 (m, 10H); ¹³C NMR (90 MHz, CDCl₃) δ 28.5, 55.4, 57.0, 115.5, 126.8, 127.0, 128.7, 128.7, 128.8, 128.9, 137.7, 141.8, 155.5, tert-butyl carbon not observed; HRMS calcd for C₂₁H₂₅NO₂ (M⁺+Na) = 346.1783, found 346.1784.

(R)-N-(tert-Butoxycarbonyl)-3-amino-4-propyl-1-butene (10h)

Prepared according to the general procedure and was purified by flash chromatography (n-hexane/EtOAc = 20/1) to give 10h (65%) as an oily liquid; ¹H NMR (360 MHz, CDCl₃) δ 0.89 (t, 3H, J = 6.8 Hz), 1.27-1.33 (m, 4H), 1.42-1.51 (s, 11H), 4.07 (brs, 1H), 4.53 (brs, 1H), 5.05-5.16 (m, 2H), 5.70-5.77 (m, 1H).

General method for the synthesis of α-hydroxy-β-amino acid

In an oven-dried, 50-mL, single-necked, round-bottom flask equipped with a magnetic stirring bar, rubber septum, and argon balloon was placed the alkene (1.20 g, 4.33 mmol) in t-BuOH:H₂O (12 mL, 1:1) and methanesulfonamide (0.412 g, 4.33 mmol) was added. The reaction mixture was cooled to 0 °C and AD-mix-α (6.54 g) was added. The resulting mixture was stirred at 0 °C for 36 h and at room temperature for 8 h, then quenched with solid Na₂SO₃ (7.0 g) and extracted with ethyl acetate. The organic extracts were washed with NaOH (1M, 10 mL), water and brine, dried (Mg₂SO₄) and concentrated. The crude diol was used as such for the next step.

The residue was dissolved in phosphate buffer (NaH₂PO₄, pH 6.7, 6 mL) and MeCN (6 mL). To the solution was added 2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO, 0.169 g, 1.07 mmol) and sodium chlorite (0.697 g, 7.71 mmol). The reaction mixture was warmed to 35 °C, and 4% aqueous sodium hypochlorite (NaOCl) solution (0.008 g, 0.12 mmol) was added. The mixture was stirred at this temperature for 6.5 h, and TEMPO (0.053 g, 0.34 mmol) along with 4% aqueous NaOCl solution (0.001 g, 0.02 mmol) were added. After stirring at 35 °C for 24 h, H₂O (5 mL) was added, and the pH of the mixture was adjusted to 8.0 with 2.0 M aqueous NaOH solution. The reaction flask was cooled to 20 °C and 6% aq. sodium sulfite was added until the solution had a pH of 9. After warming to rt for 30 min, Et₂O (5 mL) was added and the organic phase was separated. To the aqueous layer 1N HCl was added until the pH of the solution was 3, and extracted with ethyl acetate, washed with brine dried and concentrated to yield the α-hydroxy-β-amino acids. The combined yields of the α-hydroxy-β-amino acids ranged from 65-70%. HPLC purification of the mixture provided the individual diastereomers.

(2R, 3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-phenyl butanoic acid (12b)

m.pt: 148 °C; ¹H NMR (360 MHz, DMSO-d₆) δ 1.27 (s, 9H), 2.66-2.68 (m, 2H), 3.92-4.00 (m, 2H), 6.66 (d, J = 9.0 Hz, 1H, NH), 7.16-7.26 (m, 5H); ¹³C NMR (90 MHz, DMSO-d₆) δ 28.1, 34.9, 54.5, 72.6, 77.5, 125.8, 127.9, 129.0, 139.2, 154.9, 173.9.

(2S, 3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-phenyl butanoic acid (13b)

m.pt: 126 °C; ₁H NMR (360 MHz, DMSO-d₆) δ 1.30 (s, 9H), 2.66-2.71 (m, 1H), 2.78-2.83 (m, 1H), 3.96-4.03 (m, 1H), 5.14 (br s, 1H), 6.37 (d, J = 9.4 Hz, 1H, NH), 7.17-7.30 (m, 5H), 12.5 (s, 1H); ¹³C NMR (90 MHz, DMSO-d₆) δ 27.7, 28.1, 37.5, 54.7, 70.4, 77.8, 126.0, 128.1, 129.1, 138.6, 154.8, 174.0.

(2R,3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(p-methyloxyphenyl)-butanoic acid (12c)

¹H NMR (360 MHz, DMSO-d₆) δ 1.28 (s, 9H), 2.60-2.62 (m, 2H), 3.70 (s, 3H), 3.36-3.88 (m, 1H), 3.99-4.0 (m, 1H), 6.53 (d, J = 8.6 Hz, 1H, NH), 6.81 (d, J = 8.6 Hz, 2H), 7.08 (d, J = 8.6 Hz, 2H); ¹³C NMR (90 MHz, DMSO-d₆) δ 28.1, 33.9, 54.7, 54.9, 72.5, 77.5, 107.8, 113.4, 129.9, 130.9, 154.9, 157.5, 174.0.

(2S,3R)-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(p-methyloxyphenyl)-butanoic acid (13c)

m.pt: 55 °C; ¹H NMR (360 MHz, DMSO-d₆) δ 1.33 (s, 9H), 2.632.65 (m, 1H), 2.71-2.73 (m, 1H), 3.72 (s, 3H), 3.84-3.94 (m, 2H), 6.32 (d, J = 9.36 Hz, 1H, NH), 6.84 (d, J = 8.6 Hz, 2H), 7.12 (d, J = 8.6 Hz, 2H); ¹³C NMR (90 MHz, DMSO-d₆) δ 27.8, 28.2, 36.6, 54.9, 55.0, 70.3, 77.8, 113.7, 130.1, 130.5, 154.9, 157.7, 174.1; HRMS calcd for C₁₆H₂₄NO₆ (M⁺+H) 326.1598, found 326.1599.

(2R,3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(p-benzyloxyphenyl)-butanoic acid (12d)

m.pt: 129 °C; ¹H NMR (360 MHz, DMSO-d₆) δ 1.27 (s, 9H), 2.60-2.61 (m, 2H), 3.85-3.86 (m, 1H), 3.96-3.99 (m, 1H), 5.04 (s, 2H), 6.61 (d, J = 8.6 Hz, 1H, NH), 6.88 (d, J = 8.3 Hz, 2H), 7.08 (d, J = 8.3 Hz, 2H), 7.31-7.43 (m, 5H); ¹³C NMR (90 MHz, DMSO-d₆) δ 28.1, 33.9, 54.7, 69.1, 72.5, 77.5, 114.3, 127.5, 127.7, 128.3, 130.0, 131.2, 137.3, 155.0, 156.6, 174.0.

(2S,3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(p-benzyloxyphenyl)-butanoic acid (13d)

m.pt: 52 °C; [α]_D²⁵ = +56.84 (c = 0.19, MeOH); ¹H NMR (360 MHz, DMSO-d₆) δ 1.31 (s, 9H), 2.62-2.66 (m, 1H), 2.72-2.74 (m, 1H), 3.86-3.96 (m, 2H), 5.05 (s, 2H), 6.33 (d, J = 9.4 Hz, 1H, NH), 6.93 (d, J = 8.6 Hz, 2H), 7.13 (d, J = 8.6 Hz, 2H), 7.31-7.44 (m, 5H); ¹³C NMR (90 MHz, DMSO-d₆) δ 28.1, 36.6, 54.9, 69.2, 70.3, 77.8, 114.6, 127.6, 127.7, 128.4, 130.2, 130.7, 137.2, 154.9, 156.9, 174.1; HRMS calcd for C₂₂H₂₇NO₆ (M⁺-H) = 400.1760, found = 400.1755.

(2R,3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(2-naphtyl)-butanoic acid (12e)

Yield: 77%; m.pt: 154-155 °C; ¹H NMR (360 MHz, DMSO-d₆) δ 1.20 (s, 9H), 2.84-2.86 (m, 2H), 4.01-4.04 (m, 2H), 6.72 (d, J = 8.6 Hz, 1H, NH), 7.34-7.47 (m, 3H), 7.66 (s, 1H), 7.79-7.86 (m, 3H).

(2S,3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(2-naphtyl)-butanoic acid (13e)

m.pt: 113 °C; [α]_D²⁵ = +55.18 (c = 0.54, MeOH); ¹H NMR (360 MHz, DMSO-d₆) δ 1.26 (s, 9H), 2.88-2.90 (m, 1H), 2.97-2.99 (m, 1H), 3.93 (s, 1H), 4.11-4.13 (m, 1H), 6.42 (d, J = 9.7 Hz, 1H, NH), 7.39-7.48 (m, 3H), 7.71 (s, 1H), 7.82-7.87 (m, 3H); ¹³C NMR (90 MHz, DMSO-d₆) δ 28.1, 37.7, 54.7, 70.6, 77.8, 125.3, 125.9, 127.3, 127.4, 127.4, 127.6, 127.8, 131.8, 133.1, 136.2, 154.9, 174.0; HRMS calcd for C₁₉H₂₃NO₅ (M⁺-H) = 344.1498, found = 344.1497.

(2R,3R)-N-(tert-Butoxycabonyl)-3-amino-2-hydroxy-4-(p-phenylphenyl)-butanoic acid (12f)

¹H NMR (360 MHz, DMSO-d₆) δ 1.27 (s, 9H), 2.73 (br s, 2H), 3.95-4.03 (m, 2H), 6.71 (d, J = 8.8 Hz, 1H, NH), 7.26-7.63 (m, 9H); ¹³C NMR (90 MHz, DMSO-d₆) δ 27.7, 28.1, 34.5, 54.6, 72.6, 77.5, 126.2, 126.4, 127.1, 128.8, 129.6, 137.7, 138.5, 140.1, 155.0, 174.0.

(2S,3R)-N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-(p-phenylphenyl)-butanoic acid (13f)

m.pt: 162-163 °C; ¹H NMR (360 MHz, DMSO-d₆) δ ; 1.31 (s, 9H), 2.73-2.76 (m, 1H), 2.83-2.89 (m, 1H), 3.92-4.04 (m, 2H), 6.39 (d, J = 9.4 Hz, 1H, NH), 7.29-7.64 (m, 9H); ¹³C NMR (90 MHz, DMSO-d₆) δ 27.7, 28.1, 37.1, 54.7, 70.5, 77.8, 126.5, 127.2, 128.9, 129.8, 129.9, 137.9, 138.0, 140.0, 154.9, 174.0; HRMS calcd for C₂₁H₂₅NO₅ (M⁺-H) = 370.1654, found = 370.1650.

N-(tert-Butoxycarbonyl)-3-amino-4,4-diphenyl-2-hydroxypentanoic acid (12g)

¹H NMR (360 MHz, DMSO-d₆) δ 1.18 (s, 9H), 3.68 (s, 1H), 4.21 (d, J = 11.8 Hz, 1H), 4.79 (t, J = 11.6 Hz, 1H), 6.24 (d, J = 10.04 Hz, 1H, NH), 7.08-7.37 (m, 10H); ¹³C NMR (90 MHz, DMSO-d₆) δ 27.7, 27.9, 53.1, 55.1, 69.6, 77.6, 125.9, 126.5, 127.9, 128.0, 128.4, 128.7, 142.2, 142.5, 154.8, 174.0; HRMS calcd for C₂₁H₂₅NO₅ (M⁺-H) = 370.1654; found = 370.1655.

N-(tert-Butoxycarbonyl)-3-amino-2-hydroxy-4-propylbutanoic acid (12h)

¹H NMR (360 MHz, DMSO-d₆) δ 0.83 (t, 3H, J = 6.8 Hz), 1.15-1.40 (m, 15H), 3.67-3.69 (m, 1H), 3.92 (d, 1H, J = 4.3 Hz), 6.44 (d, 1H, J = 9 Hz); ¹³C NMR (90 MHz, DMSO-d₆) δ 13.8, 21.9, 27.7, 28.2, 28.3, 52.6, 72.6, 77.5, 120.7, 155.3, 174.1; ESIMS calcd for C₁₂H₂₃NO₅ (M⁺+H) = 261.1; found = 284.1 (M⁺+23).

General procedure for the synthesis of Bestatin (1) and its P1 derivatives (14-20)

Standard solid-phase peptide synthesis was performed on LeuWang resin. Coupling of the α-hydroxy-β-amino acid required HATU for 1 hr. Fmoc protecting groups were removed with 20% piperidine/DMF (30 min). To cleave products from resin, a solution of 95%TFA:2.5%TIS:2.5%H₂O was added to the resin. After standing for 2 h, the cleavage mixture was collected, and the resin was washed with fresh cleavage solution. The combined fractions were evaporated to dryness and the product was purified by reverse phase-HPLC. Fractions containing product were pooled and lyophilized. The yields of bestatin (1) and its derivatives 14-20 ranged from 60-70% respectively.

HRMS calcd for LeuLeuOH 14 C₁₃H₂₇N₂O₄ (M⁺+H) 275.1965, found 275.1973.

HRMS calcd for Tyr(OMe)LeuOH 15 C₁₇H₂₇N₂O₅ (M⁺+H) 339.1914, found 339.1913.

HRMS calcd for Tyr(OBzl)LeuOH 16 C₂₃H₃₁N₂O₅ (M⁺+H) 415.2227, found 415.2221.

ESIMS calcd for NaphLeuOH 17 C₂₀H₂₇N₂O₄ (M⁺+H) 359.1, found 359.1, 330.2 (M⁺-29).

ESIMS calcd for BipLeuOH 18 C₂₂H₃₀N₂O₄ (M⁺+H) 385.2, found 385.1, 355.1 (M⁺-29).

ESIMS calcd for DipLeuOH 19 C₂₂H₃₀N₂O₄ (M⁺+H) 385.2, found 355.1 (M⁺-29).

ESIMS calcd for NleLeuOH 20 C₁₃H₂₆N₂O₄ (M⁺+H) 275.1965, found 275.1964.

Abbreviations

Pf-LAP

Plasmodium falciparum-leucine amino peptidase

Tyr(OMe)

p-methoxy phenyl

Tyr(OBzl)

p-benzyloxy phenyl

Boc

tert-butyloxy carbonyl

DIEA

N,N-Diisopropylethylamine

DMAP

4-Dimethylaminopyridine

HATU

(2-(7-Aza-1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate)

DBU

1,8-Diazabicyclo[5.4.0]undec-7-ene

Fmoc

fluorenylmethyloxycarbonyl

TFA

trifluoroacetic acid

TIS

triisopropyl silane

EDAC.HCl

N-(3-dimethylaminopropyl)-N’-ethylcarbodiimide hydrochloride

TEMPO

2,2,6,6-Tetramethylpiperidine-1-oxyl