Structure of HsaD, a steroid-degrading hydrolase, from Mycobacterium tuberculosis

The structure of HsaD, a carbon–carbon bond serine hydrolase involved in steroid catabolism that is critical for the survival of M. tuberculosis inside human macrophages, has been solved by X-ray crystallography. Data were collected at the Diamond Light Source in Oxfordshire, England: this paper describes one of the first structures determined at the new synchrotron.


Introduction
HsaD is a member of the -hydrolase superfamily, which includes the meta-cleavage product (MCP) hydrolases . MCP hydrolases occur in the microbial pathways responsible for the aerobic catabolism of aromatic compounds, catalysing the hydrolytic cleavage of a carbon-carbon bond in the 2-hydroxy-6-oxo-dienoates that result from the meta cleavage of catechols by extradiol dioxygenases. HsaD from Mycobacterium tuberculosis H37Rv is a class I MCP hydrolase; this class includes enzymes involved in steroid and biphenyl catabolism . Recent work has demonstrated that HsaD has high specificity for the steroid MCP 4,5-9,10-diseco-3-hydroxy-5,9,17-trioxoandrosta-1(10),2-diene-4-oic acid (4, and is involved in cholesterol catabolism (Van der Geize et al., 2007). The gene encoding HsaD in M. tuberculosis is found in an operon which was predicted (Payton et al., 2001) and subsequently shown (Anderton et al., 2006) to consist of genes encoding HsaA, HsaD, HsaC, HsaB (Van der Geize et al., 2007), a hypothetical protein and an arylamine N-acetyltranferase (NAT), as shown in Fig. 1. The nat gene has been shown to be required for intracellular survival of the M. tuberculosis model organism M. bovis BCG inside macrophage cells (Bhakta et al., 2004). The phenotype following ablation of the nat gene in M. bovis BCG was mimicked by growing the wild-type organism in the presence of a putative substrate of HsaC to act as a competitive inhibitor of this enzyme (Anderton et al., 2006). Largescale transposon-mutagenesis studies have suggested that the genes encoding HsaA and HsaD are also essential for intracellular survival of M. tuberculosis in human macrophages (Rengarajan et al., 2005). Recent work has demonstrated that this operon is part of a larger regulon that is involved in lipid metabolism (Kendall et al., 2007). Therefore, the enzymes encoded by the genes in this operon are important for understanding the biology of M. tuberculosis and also represent potential therapeutic targets.
In this paper, we report the 2.35 Å resolution structure of HsaD from M. tuberculosis solved by X-ray crystallography at the Diamond Light Source synchrotron.

Production and purification of HsaD
The HsaD open reading frame was cloned from M. tuberculosis strain H37Rv into the expression vector pVLT31 with a 20-aminoacid N-terminal hexahistidine tag (amino-acid sequence MGSSH-HHHHHSSGLVPR). The expression host used was Pseudomonas putida KT4224. Heterologous expression in LB medium at 303 K gave a typical yield of 15 mg HsaD protein per litre of bacterial culture after purification by immobilized nickel-ion affinity chromatography. The N-terminal hexahistidine tag could not be removed from the recombinant protein by thrombin cleavage. Details of the cloning, expression and protein purification will be published elsewhere.

Crystallization and structure solution
Purified recombinant HsaD protein (in 100 mM sodium phosphate pH 7.4) was concentrated to 10 mg ml À1 with an Amicon ultracentrifugation concentrator (Millipore, Watford, Hertfordshire). The crystals described in this paper were grown at 292 K by the sittingdrop vapour-diffusion method. For crystallization, 150 nl concentrated HsaD solution was mixed with 150 nl precipitant [30%(w/v) PEG 3000, 0.1 M CHES pH 9.5] and the volume of precipitant in the reservoir was 100 ml. Crystals typically grew within 3-5 d.
Crystals were briefly transferred to a cryoprotectant solution [a 3:1(v:v) mixture of precipitant and glycerol] prior to flash-freezing in liquid nitrogen. Diffraction data were collected from one plateshaped crystal of approximate dimensions 50 Â 50 Â 10 mm at beamline IO4 at the Diamond Light Source, Oxfordshire, England. Data from 178 images (oscillation range 0.5 ) were indexed and integrated with MOSFLM (Leslie, 1992) and scaled with SCALA (Evans, 2006). Initial phases were determined by molecular repla-  cement with the program Phaser (Read, 2001) by using an ensemble of the MCP hydrolases from Burkholderia xenovorans LB400 (PDB code 2og1; Horsman et al., 2006) and Rhodococcus jostii RHA1 1 (PDB code 1c4x; Nandhagopal et al., 1997) as search templates, with nonconserved residues truncated to the C atom with the program CHAINSAW (Schwarzenbacher et al., 2004). Initial model building was performed with ARP/wARP (Morris et al., 2003) and was followed by iterative cycles of manual building with Coot , refinement with REFMAC (Murshudov et al., 1997) or phenix.refine (Adams et al., 2002) and model-quality checking with MOLPROBITY (Davis et al., 2007). The protein model was solvated and the waters checked with phenix.refine and Coot. The stereochemical quality of the final model was assessed with the programs MOLPROBITY and PROCHECK (Laskowski et al., 1993). Datacollection and refinement statistics are shown in Table 1.

Results and discussion
The recombinant M. tuberculosis HsaD protein was produced, purified and crystallized as described in x2. A summary of the datacollection and refinement statistics is presented in Table 1. The asymmetric unit of the M. tuberculosis HsaD crystal structure consists of two chains, each of 284 amino-acid residues (Leu7-Gly290). The electron density was not sufficiently well resolved to model the 26  N-terminal residues, which include the 20-amino-acid hexahistidine affinity tag used for protein purification, or the C-terminal arginine residue.
The M. tuberculosis HsaD enzyme shares only modest amino-acid sequence identity with other MCP hydrolases, including MhpC from Escherichia coli (Dunn et al., 2005) and the BphD enzymes from B. xenovorans LB400 (Horsman et al., 2006) and R. jostii RHA1 (Nandhagopal et al., 1997), as shown in Table 2. Despite these modest sequence identities, the M. tuberculosis HsaD protein adopts a threedimensional fold which is highly similar to those of these known MCP hydrolases. The root-mean-square deviations of the C backbone atoms between M. tuberculosis HsaD and each of these three MCP hydrolases are given in Table 2. Fig. 2 shows a sequence alignment of these four proteins; an overlay of their three-dimensional structures is shown in Fig. 3.
In all four crystal structures shown in Fig. 3 an -helical lid domain (5-5, amino acids 153-232) is encompassed by an /-domain. The /-domain is made up of a central -sheet consisting of three antiparallel strands followed by five parallel strands (from the N-terminus to the C-terminus), surrounded by five -helices. The active sites of MCP hydrolases are composed of a polar portion (P) and a nonpolar portion (NP) , with a well conserved central catalytic triad of residues, Ser114, His269 and Asp241 (M. tuberculosis HsaD numbering). Of these three residues, only the histidine is conserved throughout all known -hydrolases and recent studies have explored the mechanistic role of the active-site histidine residue of BphD from B. xenovorans . The NP subsite of M. tuberculosis HsaD appears to be the entrance to the active-site cleft, in contrast to the proposed active-site entrance (P subsite) of R. jostii RHA1 BphD (Nandhagopal et al., 2001). One key difference in the three-dimensional fold of M. tuberculosis HsaD relative to these other three MCP hydrolases is the location of aminoacid residues 213-224 (HsaD numbering), which stretches from the C-terminus of 8 to the end of 4 (Fig. 3) and forms one edge of the NP subsite. The corresponding regions in the other MCP hydrolases are closer to the catalytic triad, resulting in a significantly smaller NP subsite in these proteins relative to M. tuberculosis HsaD (Fig. 4). The calculated volume of the NP portion of the active-site cavity of M. tuberculosis HsaD ($2100 Å 3 ) was approximately twofold larger than the corresponding cavity in B. xenovorans BphD ($1200 Å 3 ) and around fourfold larger than the cavities in the other two MCP hydrolases shown in Fig. 4 ($500 Å 3 ), as determined from the PDB files using the program VOIDOO (Kleywegt & Jones, 1994). This large nonpolar portion of the active site is consistent with the role of M. tuberculosis HsaD in hydrolysis of the cholesterol MCP 4,9-DHSA ( Van der Geize et al., 2007).
The M. tuberculosis HsaD protein structure was found to adopt a tetrameric assembly in the protein crystal, as shown in Fig. 5. This tetrahedral structure may be described as a dimer of dimers in which the two monomers in each dimer (green/yellow and blue/magenta in Comparison of the nonpolar (NP) portion of the active-site surfaces of (a) M. tuberculosis HsaD (PDB code 2vf2), (b) BphD from R. jostii RHA1 (PDB code 1c4x), (c) MphC from E. coli (PDB code 1u2e) and (d) BphD from B. xenovorans (PDB code 2og1). The surfaces are coloured by electrostatic potential. The surfaces were calculated for the overlaid structures (Fig. 3) and the figure was prepared with CCP4mg. Fig. 5) are arranged such that an extended 16-strand -sheet is formed. This tetrahedral assembly was predicted to be stable in solution using the PISA web service at the European Bioinformatics Institute (Krissinel & Henrick, 2007). The active site of each monomer appears to be distinct and the entrance to each active site (e.g. Fig. 5a) is in close proximity to the monomer-monomer -sheet interface of the opposite dimer (e.g. B-B 0 ). It may therefore be possible for particularly large or long-chain substrates bound at the active site to also interact with residues on the opposite dimer. The equivalent multimeric assemblies of the MCP hydrolases from B. xenovorans (PDB code 2og1) and E. coli (PDB code 1u2e) have also been described (Dunn et al., 2005;Horsman et al., 2006).
In summary, the three-dimensional structure of the HsaD protein from M. tuberculosis has been determined by X-ray crystallography at the Diamond Light Source in Oxfordshire, England. The structure shows very high overall similarity to those of other known MCP hydrolases, with the notable exception of one region bounding the active-site pocket. The result of this difference is a larger active site relative to the previously described MCP hydrolases, which is consistent with the substrate specificity of HsaD from M. tuberculosis. The gene encoding HsaD has previously been shown to be essential for the survival of M. tuberculosis inside macrophages and therefore this structure provides a basis for rational ligand design, which will be important both for understanding the biology of M. tuberculosis and for the development of novel antituberculosis agents.

Figure 5
The predicted tetrahedral tetrameric assembly of M. tuberculosis HsaD in solution. The molecular assembly was predicted using the Protein Interfaces, Surfaces and Assemblies service (PISA) at the European Bioinformatics Institute (http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html; Krissinel & Henrick, 2007). The assembly may be described as a dimer of dimers in which the monomers coloured green and yellow form one dimer and the monomers coloured blue and magenta form the second dimer. The view in (b) was generated by rotating view (a) by 180 around the y axis. The figure was prepared with CCP4mg.