Structure of Cryptosporidium IMP dehydrogenase bound to an inhibitor with in vivo antiparasitic activity

Compound P131 displays antiparasitic activity in a mouse model of Cryptosporidium infection, validating IMP dehydrogenase (IMPDH) as a drug target. Here, the structure of the enzyme–substrate–P131 complex is reported at 2.05 Å resolution. The structure is an important step to further refine the design of IMPDH inhibitors.


Introduction
Cryptosporidium parasites are protozoan pathogens of increasingly recognized importance in both the developing and the developed world (Striepen, 2013). Cryptosporidiosis is a major cause of morbidity in infants and malnutrition in young children in Africa and Asia, and is a major cause of waterborne disease in North America and Europe. Infections are self-limiting in immunocompetent individuals, but can be chronic and fatal in immunocompromised patients (Checkley et al., 2014). In addition, Cryptosporidium oocysts are resistant to commonly used methods of water treatment and are readily available, making this organism a potential weapon for bioterrorism. The potential damage of such an incident was illustrated in the naturally occurring outbreak in Milwaukee in 1993, where 400 000 cases were reported and an estimated $96 million in damages were incurred.
Nitazoxanide is the only FDA-approved drug for the treatment of cryptosporidiosis (Checkley et al., 2014). It has limited efficacy in immunocompetent individuals and no efficacy in immunocompromised patients. High doses of paromomycin have anticryptosporidial activity in mice, where nitazoxanide is not effective (Tzipori et al., 1994;Baishanbo et al., 2006), but has not proved effective in humans (Hewitt et al., 2000). Thus, there is a critical need for new chemotherapy to treat Cryptosporidium infections.
Cryptosporidium parasites have a very streamlined purinebiosynthetic pathway that relies on the salvage of adenosine from the host . Adenosine is converted into guanine nucleotides in a process that requires IMP dehydrogenase (IMPDH) activity. Curiously, the Cryptosporidium gene for IMPDH appears to have been obtained from bacteria via lateral transfer, so that the parasite enzyme is very different from the host ortholog. IMPDH is a clinically validated target for immunosuppressive, anticancer and antiviral therapy (Hedstrom, 2009). The IMPDH reaction involves two chemical transformations (Hedstrom, 2009). The first step is nucleophilic attack of the active Cys219 at the C2 position of IMP, followed by hydride transfer to form the covalent inter-mediate E-XMP* and NADH. The reduced cofactor dissociates from the enzyme and the mobile active-site flap (residues 302-330) folds into the cofactor site, positioning the catalytic Arg315 to activate water for the hydrolysis of E-XMP* to yield the product xanthosine 5 0 -monophosphate (XMP). IMPDH is a homotetramer with square-planar symmetry. The four active sites are located near the subunit interfaces. The IMP site is contained within a monomer, and the residues that contact IMP are strongly conserved among all IMPDHs (Hedstrom, 2009). The nicotinamide portion of the cofactor site also lies within the same monomer as IMP, and is also strongly conserved, as expected given that this is the site of chemical transformation. However, the position of the  (a) Structure of P131 and other cpIMPDH inhibitors. The numbering scheme is shown in italics. (b) 2mF o À DF c electron-density map for P131 contoured at the 1 level. The inhibitor is shown in ball-and-stick representation. (c) Structure of the CpIMPDH tetramer. Subunit A is in tan, subunit B is in light blue, subunit C is in pink and subunit D is in light green. IMP and P131 are shown as spheres with the following color coding: carbon, cyan for IMP and magenta for P131; oxygen, red; nitrogen, blue; phosphorus, orange; chlorine, green. (d) Superposition of the individual monomers. Dashed lines indicate disordered regions. This figure was produced with UCSF Chimera (Pettersen et al., 2004). adenosine portion of the cofactor site varies widely among IMPDHs from different organisms. The adenosine site of eukaryotic IMPDHs is largely within the same monomer as the IMP site (Colby et al., 1999;Prosise & Luecke, 2003), but in prokaryotic IMPDHs the adenosine portion of NAD + binds in a pocket in the adjacent subunit (Makowska-Grzyska et al., 2015).
We have been engaged in a program to design selective inhibitors of C. parvum IMPDH (CpIMPDH) as potential anticryptosporidial agents Johnson et al., 2013;Kirubakaran et al., 2012;, Maurya et al., 2009Sharling et al., 2010;Sun et al., 2014;Umejiego et al., 2004Umejiego et al., , 2008. Compound P131 is the most promising compound to arise from this effort (Fig. 1a), with superior activity to nitazoxanide and paromomycin in a mouse model of acute disease .
The prokaryotic origin of CpIMPDH suggested that these inhibitors might also be effective against IMPDHs from pathogenic bacteria . The CpIMPDH inhibitor collection does contain many potent inhibitors of Bacillus anthracis IMPDH, but only a few of these compounds possess antibacterial activity (Mandapati et al., 2014). The CpIMPDH inhibitors are generally more hydrophobic, with less polar surface area, than antibiotics (O'Shea & Moser, 2008), which suggests that they do not efficiently enter bacteria. Enzyme-inhibitor structures provide important information for the further modification of CpIMPDH inhibitors both for antiparasitic and antibacterial activities.
Here, we report the structure of the enzyme-IMP-P131 complex. This information will facilitate the further optimization of CpIMPDH inhibitors for the treatment of cryptosporidiosis.

Materials and methods
Unless otherwise noted, all reagents and solvents were purchased from commercial sources and were used without further purification. Compound P131 was synthesized as described previously .

Protein production and crystallization
The coding sequence for CpIMPDH was amplified by PCR and cloned into pMCSG7 vector (Stols et al., 2002) as described in Gorla et al. (2013). The recombinant construct produced a fusion protein with an N-terminal His 6 tag and a TEV protease recognition site (ENLYFQ#S) ( Table 1). The fusion protein was expressed in Escherichia coli strain BL21(DE3) harboring the pMAGIC plasmid encoding one rare E. coli Arg tRNA (covering the codons AGG/AGA) in the presence of 100 mg ml À1 ampicillin and 30 mg ml À1 kanamycin (Kim et al., 2011). The cells were grown in enriched M9 medium at 37 C followed by overnight induction with 0.5 mM isopropyl -d-1-thiogalactopyranoside (IPTG) at 18 C. Cells were harvested, resuspended in lysis buffer (50 mM HEPES, pH 8.0, 500 mM KCl, 20 mM imidazole, 1.5 mM TCEP, 5% glycerol) and stored at À80 C. CpIMPDH protein was purified according to a standard protocol . The protocol included cell lysis by sonication and nickel-affinity chromatography on an Ä KTAxpress system (GE Healthcare Life Sciences) followed by His 6 -tag cleavage using recombinant TEV protease (Blommel & Fox, 2007) and additional Ni 2+ -affinity chromatography to remove the protease, the uncut protein and the affinity tag. In the final  Table 1 Protein-production information.

Data collection and structure determination
Diffraction data were collected at 100 K on the 19-ID beamline of the Structural Biology Center (SBC) at the Advanced Photon Source (APS), Argonne National Laboratory (Rosenbaum et al., 2006). Data were collected at a single wavelength of 0.97931 Å (12.6603 keV) to 2.05 Å resolution from a single crystal of CpIMPDH complexed with IMP and P131. The crystal was exposed for 3 s per 1.0 rotation in ! at a crystal-to-detector distance of 300 mm. The data were recorded on an ADSC Q315r CCD detector scanning 190 . The SBC-Collect program was used for data collection and visualization. Data collection, integration and scaling were performed with the HKL-3000 program package (Minor et al., 2006). A summary of the crystallographic data is presented in Table 3.
The structure was determined by molecular replacement using the structure of CpIMPDH (PDB entry 3ffs; MacPherson et al., 2010) as a search model with the HKL-3000 suite using data to 3.0 Å resolution (Bricogne et al., 2003;Minor et al., 2006). The initial model contained four protein chains of the search model, and there was extra electron density for additional protein residues that were not part of the search model. The presence of IMP and P131 in the active site was apparent from the initial electron-density map (F o ). Extensive manual model building with Coot (Emsley & Cowtan, 2004) and subsequent refinement using phenix.refine (Bricogne et al., 2003;Afonine et al., 2012) were performed against the full data set to 2.05 Å resolution until the structure converged to an R factor (R work ) of 0.187 and an R free of 0.   the structure was checked with PROCHECK (Laskowski et al., 1993) and a Ramachandran plot. Refinement statistics are shown in Table 4. Atomic coordinates and experimental structure factors of the structure have been deposited in the PDB as entry 4rv8.

Results and discussion
The structure of CpIMPDH in complex with IMP and P131 was solved at 2.05 Å resolution using molecular replacement with the structure of apo CpIMPDH (PDB entry 3ffs) as the search model (Table 4). Density for P131 is clearly visible in all four subunits (Fig. 1b). The unit cell contains a single tetramer with D4 symmetry (Fig. 1c). The four subunits are nearly identical, with r.m.s.d.s ranging between 0.15 Å (chain A versus chains B, C and D) and 0.16 Å (chain C versus chains B and D) (Fig. 1d). Residues 311-324, which are part of the active-site flap (see below), and residues 378-387 near the C-terminus are disordered in all four subunits. In addition, residues 310 and 325 are disordered in chain C.
The conformation and interactions of IMP in the P131 complex are essentially identical to those observed in the previously reported complexes of CpIMPDH with C64, Q21 and N109 ( Fig. 2a; MacPherson et al., 2010;Gorla et al., 2013;Sun et al., 2014), as well as those in the complexes of Bacillus anthracis, Campylobacter jejuni and Clostridium perfringens IMPDHs with various CpIMPDH inhibitors (Makowska-Grzyska et al., 2015).
As observed with other CpIMPDH inhibitors, P131 binds in the cofactor site, spanning both the nicotinamide and adenosine subsites (Fig. 2b). The left-side aromatic ring (Fig. 1a) binds in the nicotinamide subsite, which is a pocket formed by IMP and residues Ser164-Ala165, Asn191, Gly212-Gly214, the catalytic Cys219, Thr221, Met302-Gly303, Met308, Glu329 and Tyr358 0 (where a prime denotes a residue from the neighboring subunit). The electron density indicates that Cys219 has oxidized to the sulfenic acid, as often occurs with reactive cysteine residues. Similar oxidation of the catalytic Cys was observed in the structures of Tritrichomonas foetus IMPDH (Prosise & Luecke, 2003;Prosise et al., 2002). Despite this modification, the position of Cys219 is very similar to that observed in other CpIMPDH-inhibitor complexes (Fig. 3a). The oxime group of P131 lies across the hypoxanthine ring of IMP, while the N5 amine forms hydrogen bonds to the mainchain carbonyl O atom of Gly212 and N3 of IMP in all of the subunits (Fig. 3). The amine also interacts with water molecules to create a hydrogen-bonding network that extends to the 2 0 -OH of IMP, the sulfenic acid of Cys219, Ala165 N, Gly212 O, Gly214 N, Thr221 OG, Asp252 O and Tyr358 OH 0 . While the protein and inhibitor conformations are nearly identical in this region in all four subunits, the water occupancy varies slightly. WAT213, WAT282 and WAT316 (numbering in chains A, B and D, respectively) form hydrogen bonds to Tyr358 OH 0 and Ala165 O in chains A, B and D, but this water is absent in chain C (Fig. 2b). Curiously, despite these many interactions, the presence of the alkylamine decreases the inhibitor potency by a factor of $30 versus the corresponding oxime (IC 50 = 0.66 nM for P96 versus 20 nM for P131; Gorla et al., 2012).
Also as observed with other CpIMPDH inhibitors Gorla et al., 2013;Sun et al., 2014), P131 bends in the linker region so that the plane of the right-   Makowska-Grzyska et al., 2015). P131 is shown in stick representation with C atoms in magenta. NAD + is shown in ball-and-stick representation with C atoms in cyan. The residues that contact NAD + are shown, with the exceptions of IMP, Cys219 and Met302-Gly303, which were omitted for clarity (CpIMPDH numbering). This figure was produced with UCSF Chimera (Pettersen et al., 2004). side aromatic moiety is almost perpendicular to the left-side aromatic moiety (Figs. 1d, 3b and 3c). Both urea N atoms of P131 form hydrogen bonds to Glu329 OE2 in all four subunits ( Fig. 3b and 3c). Analogous interactions with Glu329 are observed with other CpIMPDH inhibitors. In addition, watermediated hydrogen bonds are found between the urea N1 atom and Val327 O as well as between the urea carbonyl O atom and Ala165 N.
The right-side aromatic ring binds in the adenosine subsite formed by residues Ser22 0 , Val24 0 -Leu25 0 -Pro26 0 , Ser164-Ala165-His166, Ser354 0 and Gly357 0 -Tyr358 0 (Makowska-Grzyska et al., 2015). As noted above, this subsite is highly divergent, accounting for the selectivity of the CpIMPDH inhibitors. The analogous residues in human IMPDH type 2 are Ile42 0 , Phe44 0 -Thr45 0 -Ala46 0 , Ser275-Ser276-Gln277, His466 and Gln469 0 -Asp470 0 . As observed with the primary amine, the NO 2 group binds in a hydrophilic pocket. In subunits A, C and D, the NO 2 group is roughly parallel to the aromatic ring and forms water-mediated hydrogen bonds to Ser22   Interactions of P131 with CpIMPDH. (a) Interactions of the oxime and alkylamine. P131 is displayed in ball-and-stick representation and the protein residues are in ribbon representation with side chains depicted as sticks. Hydrogen bonds are shown in cyan. The dimer formed by subunits A and D is shown in tan and the corresponding waters are shown in brown. The BA dimer is shown in light blue and the corresponding waters are shown in dark blue. The CB dimer is shown in pink and the corresponding waters are shown in magenta. The DC dimer is shown in light green and the waters are shown in dark green. Residues within 5 Å of P131 are shown, with the exceptions of Asn191, residues 327-329 and the side chains of His166, Ile213 and Asn252, which have been omitted for clarity. (b) Interactions of the urea and the right side of P131 in the AD, CB and DC dimers. The color scheme is as above. (c) Interactions of the urea and the right side of P131 in the BC dimer. The color scheme is as above. (d) Interaction of the P131 Cl atom with the carbonyl O atom of Gly357 0 . The color scheme is the same as above. Hydrogen bonds are shown in cyan. The halogen bond is shown as a black dashed line. This figure was produced with UCSF Chimera (Pettersen et al., 2004). Asn171 N and ND2. In contrast, the NO 2 group is perpendicular to the ring in subunit B and forms a hydrogen bond directly to Ser164 OG (Figs. 2c and 2d).
The water-mediated hydrogen-bonding networks of both the left-side alkylamine and the right-side NO 2 group offer many opportunities for further inhibitor optimization. In particular, the repurposing of CpIMPDH inhibitors as antibacterial agents requires the introduction of additional polarity (Mandapati et al., 2014), which might be attained by extending the inhibitor to replace water in these networks. Efforts in this area are ongoing.