|Home | About | Journals | Submit | Contact Us | Français|
Microsporidia are protists that have been reported to cause infections in both vertebrates and invertebrates. They have emerged as human pathogens particularly in patients that are immunnosuppressed and cases of gastrointestinal infection, encephalitis, keratitis, sinusitis, myositis and disseminated infection are well described in the literature. While benzimidazoles are active against many species of Microsporidia, these drugs do not have significant activity against Enterocytozoon bieneusi. Fumagillin and its analogues have been demonstrated to have activity in vitro and in animal models of microsporidiosis and human infections due to E. bieneusi. Fumagillin and its analogues inhibit methionine aminopeptidase type 2. Encephalitozoon cuniculi MetAP2 (EcMetAP2) was cloned and expressed as an active enzyme using a baculovirus system. The crystal structure of EcMetAP2 was determined with and without the bound inhibitors fumagillin and TNP470. This structure classifies EcMetAP2 as a member of the MetAP2c family. The EcMetAP2 structure was used to generate a homology model of the E. bieneusi MetAP2. Comparison of microsporidian MetAP2 structures with human MetAP2 provide insights into the design of inhibitors that might exhibit specificity for microsporidian MetAP2.
Microsporidia are a group of obligate, intracellular, parasites containing over 1200 species in at least 150 genera [1,2]. These organisms are significant pathogens with impacts on medicine, agriculture and aquaculture. The Microsporidia are eukaryotes containing a nucleus with a nuclear envelope, an intracytoplasmic membrane system, chromosome separation on mitotic spindles, a Golgi apparatus  and cryptic mitochondria . While the initial studies on small subunit rRNA genes led investigators to conclude that the Microsporidia were “primitive” protozoa; it is now appreciated that the Microsporidia are degenerate protists related to fungi [5,6].
These organisms have been identified as parasites in all classes of vertebrates (including most mammals) as well as in most invertebrates . Infections have been described in the gastrointestinal, reproductive, respiratory, muscular, excretory and nervous systems of their hosts . Serosurveys have demonstrated a high prevalence of antibodies to microsporidia suggesting asymptomatic infection may be common . Infections in humans have been described with the genera Nosema and Vittaforma , Pleistophora , Encephalitozoon , Enterocytozoon , Septata (now Encephalitozoon) , Trachipleistophora  and Anncaliia (previously Brachiola) [14-16].
Enterocytozoon bieneusi is the most common cause of microsporidiosis in humans causing malabsorption, diarrhea and cholangitis . The average prevalence of E. bieneusi infection in patients with AIDS and chronic diarrhea was 30% prior to the widespread use of highly active antiretroviral therapy  and similar prevalence rates have been reported in both children and patients with AIDS in the developing world . E. bieneusi infection also occurs as a complication of organ transplantation  and has been reported in the elderly, travelers and immunocompetent hosts [19,20]. Nosema, Vittaforma and Microsporidium have been reported in immunocompetent patients with stromal keratitis associated with trauma . Pleistophora, Anncaliia and Trachipleistophora have been reported in cases of myositis [9,13,15]. The Encephalitozoonidae (E. cuniculi, E. hellem, and E. intestinalis) have been reported in cases of keratoconjunctivitis, sinusitis, respiratory disease, prostatitis, hepatitis, encephalitis, diarrhea, superficial keratoconjunctivitis and disseminated disease [2,10,12].
Two groups of drugs have been identified as effective in the treatment of the various species of Microsporidia which infect humans . The first group is the tubulin binding benzimidazoles which are effective for microsporidiosis due to the Encephalitozoonidae . These drugs, however, are not effective for the treatment of microsporidiosis due to E. bieneusi . The second group of compounds is fumagillin and its derivatives. Fumagillin and its derivatives have activity against several groups of parasites including the Microsporidia [22,23]. It is used for the treatment of nosematosis, a microsporidiosis of honeybees, and was utilized to treat Entamoeba histolytica prior to the development of more effective amebacides . Fumagillin has in vitro and in vivo activity against both Encephalitozoonidae and Vittaforma corneae [2,21,22]. Fumagillin has been demonstrated to have efficacy for the treatment of diarrhea due to E. bieneusi in AIDS patients . TNP-470, a fumagillin derivative developed as an anti-angiogenesis drug, is active both in vitro and in vivo against several microsporidia including Enterocytozoonidae [22,23].
Fumagillin, TNP-470 and ovalicin bind to and irreversibly inhibit methionine aminopeptidase type 2 (MetAP2)1 [26,27]. The crystal structures of free and inhibited human MetAP2 demonstrated that a covalent bond is formed between a reactive epoxide of fumagillin and His231 (human MetAP2 numbering) in the active site of MetAP2 . In yeast and higher eukaryotes two isoforms (type 1 and type 2) of MetAP exist. Fumagillin and its analogues do not bind or inhibit the activity of methionine aminopeptidase type 1 (MetAP1) or other aminopeptidases. Saccharomyces cervisiae deficient in MetAP1 (i.e. map1 yeast dependent on MetAP2) are killed by ovalicin, but yeast deficient in MetAP2 (map2 yeast dependent on MetAP1) are not [26,27]. Deletion of both MetAP1 and MetAP2 is lethal. These results confirm that fumagillin selectively targets MetAP2 and not MetAP1. The microsporidium Encephalitozoon cuniculi lacks MetAP1 based on the published E. cuniculi genome ; therefore, inhibition of MetAP2 by fumagillin is the most likely cause of cell death, analogous to the situation in map1 yeast. It is likely, that other Microsporidia also lack MetAP1.
Our data  as well as that of other groups  indicate that the cellular target for fumagillin and its analogs in the Microsporidia is a MetAP2 ortholog and that this is an essential enzyme for these organisms. The crystal structure of human MetAP2 demonstrated extensive hydrophobic and water-mediated interactions that provide a mechanism for the observed selectivity of fumagillin for MetAP2 over MetAP1 . Studies have also demonstrated that fumagillin derivatives display tissue and species selectivity in their ability to inhibit MetAP2 , suggesting that it should be possible to design selective MetAP2 inhibitors. The experiments described in this manuscript were, therefore, undertaken to obtain the crystal structure of a microsporidian MetAP2 in order to provide data that could facilitate the development of more selective MetAP2 inhibitors for the treatment of microsporidiosis.
Encephalitozoon cuniculi methionine aminopeptidase type 2 (EcMetAP2) was cloned as previously described . The domain of MetAP2 encompassing residues 2-358 (numbered relative to reference sequence NP_586190) was amplified by PCR and TOPO cloned into a custom TOPO adapted pFastBac (KF) vector (Invitrogen Corporation). Expression in this vector generated MetAP2 residues 2-358 fused to a TEV protease-cleavable N-terminal 6X histidine tag. Following cleavage, the protein generated is GSL–MetAP2 (2-358), where the GSL residues originated from the vector sequence. Standard baculovirus expression using the Bac-to-Bac system (Invitrogen) was used to generate recombinant virus. A 6 liter 48hr fermentation in Sf9 cells was carried out and cells harvested by centrifugation and pellets were stored at −80ºC for purification.
For purification, the Sf9 cell pellet from a six liter fermentation was lysed in 240 ml of lysis buffer (0.05 M Tris-HCl pH 8.0, 0.25 M NaCl). HIS-tagged aminopeptidase was purified by Ni-NTA affinity chromatography in batch mode. After cleavage of the N-terminal his-tag by TEV protease, the aminopeptidase was further purified by a second Ni-NTA step. This material was concentrated and applied to a Sephadex S200 gel filtration column equilibrated with 0.01 M Hepes pH 7.5, 0.15 M NaCl, 0.01 M L-methionine, 10% glycerol, and 5 mM DTT. Fractions containing the aminopeptidase were pooled and concentrated to 2 mg/ml. As previously published, purification of recombinant EcMetAP2 yields an approximately 95% pure enzyme as judged by SDS-PAGE . Protein was stored at -80°C for long term storage.
Recombinant human MetAP2 was purchased from Mediomics (St Louis, MO).
Methionine aminopeptidase activity was measured spectrophotometrically by monitoring the free L-methionine formation after peptidolysis, using coupled enzyme assay comprised of L-amino acid oxidase (AAO) and horseradish peroxidase (HRP) . Oxidation of o-dianisidine was monitored as an increase in absorbance at 450nm. This assay was adapted to 96-well assay format with a reaction volume of 50μl. To determine the optimum temperature the assay was performed at different temperatures in HEPES buffer pH 7.5 and for optimum pH the assay was performed at different pH (using MOPS, HEPES, Tris and CAPS buffers). After determining the optimum temperature and pH of activity to be at 37°C and pH 8.5 (Tris buffer) for EcMetAP2 and pH 7.5 (HEPES) for HsMetAP2 all assays were done under these conditions. A typical assay contained 50mM of buffer (either Tris pH 8.5 or HEPES pH 7.5), 100mM NaCl, 100μM CoCl2, 1mM o-dianisidine, 0.5U HRP, 0.02U L-AAO, 0.5μg MetAP2, BSA at 0.1mg/ml and glycerol at 2% final concentration. The assay was initiated by the addition of substrate (2mM Met-Ala-Ser). The reaction was carried out at 37°C for 45min and absorbance at 450nm was measured at 5min intervals. Standard concentrations of o-dianisidine (oxidized by 0.6% H2O2 and 1U HRP) were used to calibrate the assay and determine the stoichiometry of the MetAP2 reaction. The initial rate was corrected for the background rate determined in the absence of MetAP2 enzyme but with all the other components of the assay. Assays were done in triplicates. To determine the optimum substrate concentration, Km and Vmax, assays were carried out at optimal conditions ie. pH 8.5 for EcMetAP2 and pH 7.5 for HsMetAP2. Assays were performed with the substrate (Met-Ala-Ser) concentration ranging from 0.1mM to 8 mM. Analysis of the kinetic data was performed by using KaleidaGraph, fitting the Michelis-Menton double reciprocal plot.
To examine the inhibition of MetAP2 enzyme activity by Fumagillin and TNP470, fresh solutions of Fumagillin and TNP470 in 100% DMSO were prepared for the assay. Human MetAP2 or EcMetAP2 (0.5μg) were incubated on ice for 30min with or without fumagillin or TNP470 in enzyme dilution buffer (20mM HEPES pH7.5, 40mM NaCl, 100μg/ml BSA, 2%DMSO). Reaction mixtures (50μl) contained 50mM assay buffer, 100μM CoCl2, 4mM Met-Ala-Ser substrate, 0.2U L-AAO, 0.5U HRP and 1mM o-dianisidine. The reaction was started by adding the enzyme to the assay components, increase in absorbance at 450nm was measured every 5min at 37°C. To determine the IC50 values of the inhibitors, 0.5μg of EcMetAP2 or HsMetAP2 enzyme was incubated for 30 min on ice with different concentrations of inhibitors diluted in enzyme dilution buffer having 2% DMSO.
The metal content of purified EcMetAP2 was determined by X-ray absorption spectroscopy at the X3B beamline at the National Synchrotron Light Source (NSLS) . The X-ray energy was set to 10 keV and the available instrumentation allowed for the detection of Mn, Fe, Co, Cu, Ni, and Zn. The appropriate volume of aqueous protein solution containing 0.1-0.15 mg of a protein was loaded in a sample well and dried overnight. Each sample was screened for the selected metals with sixty one-second-long scans using a 13-element Germanium X-ray fluorescence detector. Metal-to-protein stoichiometry was inferred by the comparison of the detected counts for a specific metal to an appropriately chosen set of standards for each metal under the same experimental conditions.
Diffraction quality crystals were grown by sitting-drop vapor diffusion at 20°C by mixing EcMetAP2 (2mg/mL) and 30%(w/v) polyethylene glycol 4,000, 0.2 M ammonium sulfate. Crystals formed in seven to ten days with approximate dimensions of 0.1mm x 0.05mm x 0.01mm. Prior to data collection, crystals were cryoprotected by brief transfer to mother liquor supplemented with 10% (v/v) 2-methyl-2,4-pentanediol) and flash cooled in liquid nitrogen.
Crystals of EcMetAP2 bound with fumagillin and TNP-470 were grown by co-crystallization. Prior to co-crystallization, EcMetAP2 protein (2mg/mL) was dialyzed against one liter of (10 mM Hepes, 150 mM NaCl, 10% glycerol, 5 mM DTT, pH 7.5) for three hours with a change to fresh buffer every hour in order to remove any L-methionine from the original protein buffer. For co-crystallization, 100 mM stock solutions of fumagillin or TNP-470 (in 100% ethanol) were diluted to a final concentration of 0.5 mM in (27% PEG 4,000, 0.2 M ammonium sulfate). Co-crystals of EcMetAP2 with either fumagillin or TNP-470 were grown as described for the free enzyme.
X-ray diffraction data (Table 1) were collected to resolutions of 2.18, 2.50, and 2.89 Å for the apo, fumagillin and TNP-470 bound crystals, respectively. All data were collected at beamline X29A at the National Synchrotron Light Source, Brookhaven National Laboratory (BNL). Data integration and scaling was carried out using HKL2000 . Diffraction from these crystals is consistent with the monoclinic P21 space group with two molecules in the asymmetric unit.
The apo structure of EcMetAP2 was determined by molecular replacement with the program MOLREP , using the structure coordinates of the human methionine aminopeptidase type 2 (HsMetAP2b) (PDB I.D.: 1YW9) as the search model (45% sequence identity). Residues not conserved between EcMetAP2 and HsMetAP2b were pruned to the last common atom using CHAINSAW  to generate the search model. Two monomers are present in the asymmetric unit with a Vm of 2.4. The molecular replacement solutions were refined using rigid-body followed by restrained refinement using the program REFMAC5 . In order to reduce any model bias, automated model building and refinement was carried out using the initial refined model in the program ARP/warp . After the last cycle of refinement and model building, water molecules were added using ARP/wARP. At this stage of refinement two large spherical features (7σ and 9σ) present in difference Fouriers calculated with Fo-Fc coefficients were modeled as iron ions. A final round of restrained refinement using all atoms was carried out. Final refinement statistics are summarized in Table 1.
Structure determination of EcMetAP2 with either fumagillin or TNP-470 in the active site was determined by difference Fourier methods using the coordinates of the native EcMetAP2 structure . After the last cycle of refinement and model building, water molecules were added using ARP/wARP and iron ions were modeled manually. Significant features in difference Fourier maps were readily modeled as fumagillin and TNP-470. Atoms for which no electron density was observed in fumagillin and TNP-470 include (C44, C45, C46, C47, C48, C49, C4A, O4B, O4C) and (O4B, C44, CL4), respectively. Refinement statistics are summarized in Table 1. For all structures of EcMetAP2, electron density for residues 3 to 358 was clearly visible.
Model building was carried using the program Coot . The stereochemical quality of the final refined models was evaluated using Procheck  and MolProbity . Figures of X-ray structures were produced using Pymol . The program ESPript  was used to produce sequence alignment figure 3.
Purified rEcMetAP2 had enzymatic activity similar to that exhibited by Human MetAP2 (HsMetAP2) (Table 2). The highest activity of rEcMetAP2 was exhibited using CoCl2 (11.7μmoles/min/mg protein), decreasing with MnCl2 (7.9 μmoles/min/mg protein) and the lowest with FeCl2 (2.9 μmoles/min/mg protein). rEcMetAP2 was inhibited by both Fumagillin and TNP470 (Figure 1). Using the assay described in materials and methods the IC50 values of fumagillin and TNP470 exhibited by rEcMetAP2 was 11.1 nM and 10.6 nM, respectively; while recombinant HsMetAP2 demonstrated significantly higher IC50 values for Fumagillin (188.5 nM) and TNP470 (95.0 nM).
We have determined the X-ray crystal structure of rEcMetAP2 in its apo, fumagillin, and TNP-470 bound states at resolutions of 2.2, 2.5, and 2.9 Å, respectively. E. cuniculi MetAP2 exhibits a classic ‘pita-bread’ fold present in other methionine aminopeptidases (MetAP) from E. coli, S. aureus, M. tuberculosis, T. maritima, P. furiosus, and H. sapiens [28,45-51]. The ‘pita-bread’ fold of EcMetAP2 is composed of a central beta-sheet with anti-parallel beta-strands β3, β5, β6, β9, β10, β15 and flanking alpha-helices α1, α2, α3, and α4 (Figure 2). A dinuclear metal center and the active site are located on the concave face of the central beta-sheet. Unlike the type-1 MetAP's which possess a ‘pita-bread’ fold with pseudo-two fold symmetry, this symmetry is broken in EcMetAP2 by the insertion of a sixty-one residue subdomain (β12, α5, α6, α7, β13) found in type-2 MetAP's and a two-stranded anti-parallel β-sheet (β7, β8) which is present only in the eukaryotic MetAP2 family members (Figures (Figures22 and and3).3). As predicted by sequence alignment and confirmed by this crystallographic analysis, the insertion of the helical subdomain into the ‘pita-bread’ fold categorizes EcMetAP as a type 2 methionine aminopeptidase (EcMetAP2). Superposition of the α-carbon positions of EcMetAP2 with Human MetAP2 (HsMetAP2b) result in a root-mean-squared deviation of 1.35Å over 355 α-carbon atoms, and further supports the inclusion of EcMetAP in the methionine aminopeptidase type 2 family (Figure 4). The members of the MetAP family are further differentiated based on the presence of an N-terminal extension of the catalytic domain. MetAP's possessing only the catalytic domain are designated 1a (E. coli) or 2a (P. furiosus), those with an approximately 100 residue N-terminal extension composed of two zinc finger domains followed by a linker are 1b (yeast/human) or composed of polybasic/acid residues are 2b (human), and those having relatively short extensions of approximately 40 residues have been designated 1c (M. tuberculosis) [45,47,52-55]. To date, X-ray structures for three MetAP type 2 enzymes have been determined from Pyrococcus furiosus, Homo sapiens, and in this work from Encephalitozoon cuniculi with each enzyme differing at the N-terminus of the catalytic domain. Like HsMetAP2b, EcMetAP2 possess an N-terminal extension of the catalytic domain, albeit shorter (approx. forty residues) and with no polybasic/acidic character, while P. furiosus has no extension (Figure 3). In addition, structural sequence alignment of a homology model of a second microsporidian MetAP2 from Enterocytozoon bieneusi [Genebank accession number EED44036], having 48% sequence identity with EcMetAP2, suggests that the E. bieneusi enzyme would also have an N-terminal extension that is shorter than EcMetAP2 (Figure Sequence alignment). With the addition of the E. cuniculi MetAP2 structure into the MetAP2 family and the differences at the N-terminus of the catalytic domains, a further subdivision of type 2 MetAP's should follow suit with E. cuniculi MetAP2 being designated 2c as its type 1c counterpart. Therefore, E. cuniculi MetAP2 will be designated as EcMetAP2c.
During the process of model building and refinement two large spherical features (7σ and 9σ) present in difference Fouriers, calculated with Fo-Fc coefficients, were observed amongst the putative metal coordinating residues suggesting the presence of two bound metal ions. In addition, these assignments are consistent with anomalous difference Fouriers, which revealed two large peaks of electron density (6σ and 8σ) amongst the putative metal coordinating residues (Supplementary Figure 1). As described in the experimental procedures, EcMetAP2c was overexpressed in Sf9 cells infected with recombinant baculovirus and purified from cell lysate. At no time during expression, purification and crystallization was exogenous divalent metal added. Therefore, the metal bound in the active site could have been acquired from contaminating metal in the purification buffers since trace metals were not chelated from the buffers or alternatively metal could have been bound in the cytoplasm of the Sf9 cells during expression and carried through during purification. This same phenomenon of acquiring divalent metals from Sf9 cells during baculovirus expression and coordination in the active site has been observed by two separate groups overexpressing human MetAP2b (HsMetAP2b) [33,56]. A consensus for the physiologically relevant divalent metal ion for MetAP's has yet to be reached and both Mn2+/Co2+ have been suggested for human MetAP2b, Fe2+ for Pyrococcus furiosus MetAP2a, Zn2+ for yeast MetAP-Ib, and Fe2+ for E. coli MetAP-Ia [33,56-60]. Metal analysis of purified recombinant EcMetAP2c based on the method of Shi et. al.  determined the metal-to-protein ratios for Fe, Zn and Co to be 0.6, 0.2 and 0.1, respectively. Since Fe is the most abundant metal present in purified recombinant EcMetAP2c and enzymatic activity is observed using this metal ion, although four fold less activity than with Co2+, we have modeled Fe2+ into the Fo-Fc difference Fourier maps. These data do not identify or attempt to identify the physiologically relevant divalent metal ion for EcMetAP2c as Fe2+, but support the existence of two metal ions bound by the putative metal coordinating residues in the active site and suggests that Fe2+ could be modeled into the corresponding electron density. As with other members of the MetAP family, the active site of EcMetAP2c resides on the concave face of the central beta-sheet, in a deep pocket with two metal ions bound and a hydrophobic side pocket that presumably provides specificity for binding of the N-terminal methionine of natural substrates. All the residues coordinating the Fe2+ ions are absolutely conserved in all MetAP sequences and in EcMetAP2c, including D130, D141, H210, E243, and E339, which are located on β-strands β5, β6, β9, β10, β15 (Figure 5). Metal coordination is similar to other MetAP's with D141 and E339 coordinating both metal ions in a bidentate fashion, D130 is bidentate with metal 2, and H210 and E243 are monodentate with metal 1, and a single water molecule bridges both metal ions.
The EcMetAP2c-apo structure was determined without L-methionine bound. However, superposition of the EcMetAP2c-apo structure with the product bound human HsMetAP2b-L-methionine structure (PDB I.D. 1KQ9) suggests the location of the methionine-binding site in the active site of EcMetAP2c. The methionine side chain is bound in a hydrophobic pocket that lies above the dinuclear metal center (Figure 6). Loops located on the periphery of the concave face of the central beta-sheet and from the subdomain insert provide the residues making up the S1subsite hydrophobic methionine-binding pocket. The residues making up the methionine-binding pocket include F97, P98, H109, I217, H218 from the peripheral loops of the central beta-sheet and H261, V263, P292, and Y324 from the subdomain insert. Only two residues (H109 and H218) are absolutely conserved amongst the entire MetAP family while the remaining residues are exclusive to type 2 family members (Figure 3). The methionine side chain is bound in the hydrophobic pocket with H109 and H218 positioned behind the side-chain. Residues H109 and H218 are likely involved in catalysis as mutation of the structural counterparts in HsMetAP2b (H231 and H339) and E. coli MetAP1a (H79 and H178) result in abrogation or significant reduction in activity of the recombinant enzymes [53,61-63]. The structures of E. coli MetAP1a with bound transition state analogs, product, and product analogs and comparison with other ‘pita-bread’ enzymes such as aminopeptidase P has led to the proposal that H178 (H218 in EcMetAP2c and H339 in HsMetAP2b) stabilizes a tetrahedral intermediate of the transition state by hydrogen bonding to the carbonyl oxygen of methionine, while H79 (H109 in EcMetAP2c and H231 in HsMetAP2b) hydrogen bonds to the amide proton of the scissile bond . In all structures of MetAP's with L-methionine bound in the active site, H178 of E. coli MetAP1a and its structurally equivalent residue in other MetAP's form a hydrogen bond with the carbonyl oxygen of L-met. In EcMetAP2c H218, H178 in E. coli MetAP1, is just outside of hydrogen bonding distance to the carbonyl oxygen of the superimposed L-met from HsMetAP2b. In addition, in EcMetAP2c metal coordinating interactions between the carboxylate and amino group of the superimposed L-met from HsMetAP2b are observed and include N-terminal coordination with iron 2 and a bridging coordination between a carboxyl oxygen of L-Met and both metal ions; these interactions are conserved in all L-Met bound MetAP structures. The structurally conserved positions of EcMetAP2c H109 and H218, deleterious effects on hydrolysis by mutation of corresponding residues in E. coli MetAP1a and HsMetAP2b, and predicted contacts to L-Met strongly suggest that H109 and H218 are involved in N-terminal methionine hydrolysis.
In an effort to understand how fumagillin and/or TNP-470 (Figure 7A) could be modified to generate specificity for microsporidian MetAP2 over human HsMetAP2b we have solved X-ray structures of EcMetAP2c with fumagillin and TNP-470 bound in the active site to a resolution of 2.5 Å and 2.9 Å, respectively (Table 1). Electron density for fumagillin and TNP-470 was clearly visible within the active site, with 2Fo-Fc density visible for the cyclohexane ring and epoxide bearing isoprenyl substituent (Figure 7B and 7C). However, electron density for both the unsaturated decanoic acid side chain on fumagillin and the O-(Chloroacetylcarbamoyl) group on TNP-470 was visible up to carbon-43. Methionine aminopeptidases, specifically type 2, have been shown to be irreversibly inhibited by fumagillin and its derivatives through covalent modification of an active site histidine residue with the carbon atom of the spirocyclic epoxide at position C3 on the cyclohexane ring of fumagillin [27,28,46,61,65]. In the X-ray structures of EcMetAP2c-fumagillin and EcMetAP2c-TNP-470 electron density is clearly visible for a covalent bond between the Nε2 imidazole nitrogen atom of, the absolutely conserved residue, H109 and carbon 2 of the cleaved spirocyclic epoxide. Cleavage of the spirocyclic epoxide also results in the liberation of the oxygen atom and coordination to iron 1 (3.25 Å). The epoxide bearing isoprenyl sidechain of fumagillin/TNP-470 is bound in the methionine-binding pocket and makes hydrophobic contacts with F97, P98, I217, H261, V263, P292, and Y324 (Figure 7B and 7C). No significant movement of the active-site residues are observed upon binding fumagillin or TNP-470 with the exception of H218 which adopts a side-chain rotamer that is positioned away from the inhibitor in order to avoid close contact (Figure 7D).
The ordered portion of the unsaturated decanoic acid side chain of fumagillin and the O-(Chloroacetylcarbamoyl) group of TNP-470 protrudes from the active site and make hydrogen bonding contacts with the backbone amide and side-chain Nδ1atom of H208. The O-(Chloroacetylcarbamoyl) group of TNP-470 is completely ordered in the human HsMetAP2b-TNP-470 structure (PDB ID. 1B6A) making hydrogen bonding contacts to ordered water molecules. In our EcMetAP2c-TNP-470 structure no electron density is observed for the terminal three atoms of the O-(Chloroacetylcarbamoyl) group, which is consistent with the absence of any additional potential interactions that might stabilize a single conformation. The unsaturated decanoic acid side chain of fumagillin in the human HsMetAP2b-fumagillin structure (PDB ID. 1BOA) is positioned in a single conformation by hydrophobic contacts with conserved residues L328, V374, and L447 and a single hydrogen bond with the main-chain amide of D376. Superposition of the HsMetAP2b-fumagillin structure (PDB ID. 1BOA) onto the EcMetAP2c-fumagillin structure suggests that the terminal carboxylate group of the decanoic acid side chain of fumagillin would be repelled by the negative charge from the side chains of D254 and D255 in EcMetAP2c (Figure 8A) and prevent the stabilization of a single conformation; this is consistent with a lack of electron density for a majority of the fumagillin decanoic acid side chain in our structure.
Sequence and structural comparison of EcMetAP2c and HsMetAP2b show that the active sites are virtually identical with very few changes in the residues making up the S1 and S1' subsites (Figure 8A). Comparison of a homology model of a MetAP2 from the microsporidia Enterocytozoon bieneusi with the HsMetAP2b structure, 43% sequence identity, shows that the active site residues are even more similar, than in E. cuniculi MetAP2c, to those in the human enzyme (Figure 8B). The microsporidian MetAP2 active sites are not, however; without amino acid differences compared to the human enzyme. Within the S1 hydrophobic methionine-binding subsite, E. cuniculi and E. bieneusi possess V263 and A245, respectively, while a methionine (M384 and M207) is present in the same position in the human and P. furiosus MetAP2 enzymes (Figure 8A and 8B). These changes to smaller amino acid residues in the S1 subsite of the microsporidian enzymes might provide some opportunities for the generation of fumagillin analogues that are specific for microsporidia MetAP2. Fumagillin analogues, active against human MetAP2b, have been synthesized with ring substitutions in place of the epoxide bearing isoprenyl group and with the isoprenyl group lengthened by one or more carbon atoms . These existing fumagillin analogues could take advantage of the differences within the S1 subsite in human and microsporidian MetAP2 by changing the register of the reactive spirocyclic epoxide on fumagillin to prevent covalent modification of the active site histidine residue on HsMetAP2b (H231) and retain inhibition of the microsporidial enzymes. The importance of the amino acid residue differences found within the S1 hydrophobic methionine-binding subsite between the microsporidian, E. cuniculi (V263) and E. bieneusi (A245), and human (M384) MetAP2 enzymes is supported by fumagillin and TNP-470 resistant single site mutants of human MetAP2b. It has been suggested that the fumagillin and TNP-470 specificity determinants lie in the S1 hydrophobic methionine-binding subsite as shown by single site mutations (H382Y and Y444C) in the S1 subsite of human MetAP2b that result in inhibitor resistance .
Within the S1′ subsite, where the unsaturated decanoic acid side chain of fumagillin lies, only modest sequence variation is observed, with E. cuniculi MetAP2c possessing D206, H208, and D254, while in human HsMetAP2b the structurally equivalent residues are N327, N329, and H375, respectively (Figure 8A). Superposition of the HsMetAP2b-fumagillin structure onto the EcMetAP2c-fumagillin structure suggests that the decanoic acid side chain of fumagillin could adopt a bound conformation in EcMetAP2c, as observed in the human enzyme, if the terminal carboxylate was modified to have functional groups that could make interactions with residues D254 and D255 in E. cuniculi MetAP2c. Modification of the terminal carboxylate group on the decanoic acid side chain of fumagillin to have a basic group (e.g., amino or guanidinium group) in place of a carboxyl oxygen atom might allow interactions with D254 on E. cuniculi and lock the fumagillin side chain into a single conformation in order to generate specificity for EcMetAP2c. Superposition of the homology model of E. bieneusi MetAP2 onto the HsMetAP2b-fumagillin structure suggest that the only amino acid variation in the S1′ subsite is a I235 and K236 in E. bieneusi and V374 and H375 in the structurally equivalent positions in the human enzyme (Figure 8B). For E. bieneusi MetAP2, specificity would be better sought by directing attention to changing or modifying the epoxide bearing isoprenyl group on fumagillin and taking advantage of the S1 subsite amino acid differences where E. bieneusi has a much smaller residue (A245) than human MetAP2b (M384).
The methionine aminopeptidase type 2 from Encephalitozoon cuniculi has been shown to cleave the N-terminal methionine from substrates with kinetic values that are comparable to human MetAP2b. As with human MetAP2b, E. cuniculi MetAP2c is inhibited by the MetAP2 specific inhibitors, fumagillin and TNP-470. The X-ray structures of E. cuniculi MetAP2c and metal analysis of the purified recombinant enzyme have confirmed the presence of a dinuclear metal center occupied with two iron ions in the active site. Comparison of the E. cuniculi MetAP2c and human MetAP2b structures has revealed amino acid differences in the active sites that may be exploited to design fumagillin analogues that are specific for microsporidian MetAP2. A homology model of the microsporidian MetAP2 from Enterocytozoon bieneusi has been constructed, using the EcMetAP2c structure, which has provided a picture of the active site for design of specific fumagillin analogues. With the X-ray structures of the E. cuniculi MetAP2c enzyme, structure based drug design can now be carried out in order to develop a fumagillin class of inhibitor that has increased specificity for microsporidian MetAP2.
Anomalous difference electron density (4σ contour) is shown as blue mesh. Residues coordinating two iron ions are shown as sticks with black dashes representing coordination bonds. Iron ions and water molecules are represented as silver and red spheres, respectively.
This work was supported by NIH grants AI31788 and AI069953. The New York Structural Genomics Research Consortium (NYSGXRC) is supported by NIH Grant U54 GM074945. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. E. cuniculi MetAP2c is identified with the target I.D. NYSGXRC-9201A.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Atomic coordinates and structure factors for the E. cuniculi MetAP2c-apo, MetAP2c-fumagillin, and MetAP2c-TNP470 structures have been deposited in the RCSB protein data bank with the PDB entries 3FM3, 3FMQ, and 3FMR, respectively.
1Abbreviations: MetAP2, methionine aminopeptidase type 2; Tris-HCl, Tris(hydroxymethyl)aminomethane hydrochloride; HEPES, 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid, N-(2-Hydroxyethyl)piperazine-N'-(2-ethanesulfonic acid); DTT, DL-Dithiothreitol; AAO, L-amino acid oxidase; HRP, horseradish peroxidase; MOPS, 4-Morpholinepropanesulfonic acid; CAPS, 3-(Cyclohexylamino)-1-propanesulfonic acid; BSA, bovine serum albumin; DMSO, dimethyl sulfoxide; PEG, polyethylene glycol.