Correspondence: Gurdyal S. Besra, School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. Tel.: +44 0 121 4158125; fax: +44 0 121 414 5925; e-mail: firstname.lastname@example.org
Because of an increased emergence of resistance to current antitubercular drugs, there is a need for new antitubercular agents directed against novel targets. Diaminopimelic acid (DAP) biosynthetic enzymes are unique to bacteria and are absent in mammals and provide a rich source of essential targets for antitubercular chemotherapy. Herein, we review the structure and function of the mycobacterial DAP biosynthetic enzymes.
Tuberculosis (TB) is the second most common infectious cause of adult mortality after human immunodeficiency virus (HIV) and is ranked tenth of all causes of loss of healthy life worldwide (Corbett & Raviglione, 2005; Mathema et al., 2006). The incidence of TB cases is estimated to be 8 million, with 2 million deaths per annum (Corbett & Raviglione, 2005). HIV infection accounts for the increase in the global tuberculosis burden (Frieden et al., 2003). In addition, the emergence of multidrug-resistant (MDR) strains and extensively drug-resistant (XDR) strain has caused the increase in tuberculosis cases (Dorman & Chaisson, 2007; Harper, 2007).
There is a need for new drugs for the treatment of TB that exploit novel targets. meso-DAP biosynthesis exists only in bacteria and is absent in mammals (Cox et al., 2000; Diaper et al., 2005; Hudson et al., 2005). meso-DAP is synthesized in mycobacteria from aspartate in eight steps via l-2,3,4,5-tetrahydrodipicolinate (THDP) (Cirillo et al., 1994a; Pavelka & Jacobs, 1996) (Fig. 1). l-lysine is obtained from meso-DAP by a single decarboxylation step (Born & Blanchard, 1999) (Fig. 1). Several of the enzymes of DAP synthesis have been identified in Mycobacterium tuberculosis, disruption of which leads to cell death, because of the instability of peptidoglycan (Cirillo et al., 1994a; Born et al., 1998; Wheeler & Blanchard, 2005). The knockouts of genes in this pathway have been shown to be essential for mycobacterial growth (Pavelka & Jacobs, 1996; Wheeler & Blanchard, 2005), except for Mt-dapB that has been classified as a slow growth mutant by transposon mutagenesis (Sassetti et al., 2001, 2003). Based on this observation, an in-frame Mt-dapB deletion mutant needs to be constructed to address whether Mt-DapB is an essential enzyme. This review gives an overview of the structure and function of the mycobacterial DAP biosynthetic enzymes that have been characterized to date. N-succinyl-l,l-diaminopimelic acid desuccinylase is the only uncharacterized mycobacterial DAP biosynthetic enzyme, and as such, an overview of the enzyme from other bacteria is included.
meso-DAP biosynthetic pathway enzymes
Aspartokinase (ask) and aspartate semialdehyde dehydrogenase (asd)
The first two enzymatic steps, catalysed by aspartokinase and aspartate semialdehyde dehydrogenase, are encoded by the adjacent ask and asd genes (Rv 3709c and Rv3708c, respectively) (Cirillo et al., 1994b; Wheeler & Blanchard, 2005): the aspartokinase reaction involving the phosphorylation of l-aspartate by ATP, with the subsequent conversion of β-aspartyl phosphate to l-aspartic-β-semialdehyde by the aspartate semialdehyde dehydrogenase (Asd) (Pavelka & Jacobs, 1996). Unlike other bacteria that have multiple aspartokinase genes that encode enzymes that are differentially regulated by the end products of these amino acid pathways, there is only one mycobacterial ask gene (Wheeler & Blanchard, 2005). In Mycobacterium smegmatis, ask expression yields three differentially regulated aspartokinase isoenzymes (Sritharan et al., 1989; Pavelka & Jacobs, 1996; Pavelka, 2000).
The cloning and sequencing of the ask–asd operon of M. smegmatis has been reported (Cirillo et al., 1994b). There is no structural representative of Rv3709c in the Protein Data Bank, although a recent crystallization report for the β subunit has been published (Schuldt et al., 2011), but it shares ~70% identity with the Corynebacterium glutamicum Ask, whose structure reveals a unique α2β2 heterotetramer distinct from other aspartokinase structures: the larger α subunit is the translated product of the entire open reading frame, while the smaller β subunit is a shorter, in-frame translation product from the same gene (Cirillo et al., 1994b). The amino terminus of the mycobacterial Ask protein sequence is highly conserved across species, particularly between positions 198 through to 207, suggesting that these residues are catalytically important (Cirillo et al., 1994b). The relatively less conserved carboxy-terminal region is thought to be involved in maintaining the aspartokinase tertiary structure but is catalytically dispensible (Cirillo et al., 1994b). The aspartate pathway is essential in M. smegmatis (Pavelka & Jacobs, 1996). The first mycobacterial DAP auxotrophic mutant generated in M. smegmatis with a disruption in the ask gene causing lysis upon meso-DAP deprivation could be complemented with the wild-type ask gene (Pavelka & Jacobs, 1996; Pavelka et al., 1997).
Asd from M. tuberculosis has been cloned, expressed in Escherichia coli, purified and characterized (Shafiani et al., 2005; Vyas et al., 2008). Asd has a molecular weight of 38 kDa and is a homodimer (Vyas et al., 2008). The purified Mt-Asd is functionally active where the Kcat is 8.49 s−1. The Km and Vmax values in the direction reverse to DAP synthesis for all three substrates l-aspartate semialdehyde, NADP+ and Pi have been determined (Shafiani et al., 2005). A crystallization report for Mt-Asd exists, with data to 2.18 Å (Fig. 2) (Vyas et al., 2008), the associated as yet unpublished structure sharing structural homology to an Asd from Streptococcus pneumoniae (Singh et al., 2008). Mt-Asd has an N-terminal NADP-binding domain and a dimerization domain (Shafiani et al., 2005). Cys 130 is the active site catalytic residue in Mt-Asd, a residue conserved in all bacterial Asd proteins, forming an acyl intermediate during the conversion of β-aspartyl phosphate to l-β-aspartate semialdehyde (Singh et al., 2008). Two other residues, Gln 157 and His 256, located in the active site cleft are essential for catalysis (Singh et al., 2008). From homology modelling studies, Cys 130 and His 256 have been proposed as two important residues for selective inhibitor development against Mt-Asd (Singh et al., 2008).
Dihydrodipicolinate synthase (dapA)
Mt-DapA (Rv 2753c) catalyses an aldol condensation between l-aspartate-β-semialdehyde and pyruvate to form 2,3-dihydrodipicolinic acid (Kefala et al., 2008). Mt-dapA has been expressed in E. coli, purified, crystallized and solved to 2.28 Å (Kefala & Weiss, 2006; Kefala et al., 2008). A ribbon model of Mt-DapA is shown in Fig. 2. The protein structure reveals a classical α8β8 ‘TIM barrel’, with an active site architecture similar to homologues from other bacteria (Kefala et al., 2008). DapA exists as a tetramer with an apparent molecular weight of approximately 120 kDa, with two independent tetramers in the asymmetric unit (Kefala & Weiss, 2006). A recent study with a A204R variant (obligate dimer) revealed tetramerization to be nonessential for activity (Evans et al., 2011).
Dihydrodipicolinate reductase (dapB)
Mt-DapB (Rv2773c) reduces the α,β-unsaturated cyclic imine 2,3-dihydrodipicolinic acid to yield 2,3,4,5-tetrahydrodipicolinic acid using NADH or NADPH with nearly equal efficiency with Km values of 3.2 ± 0.4 and 11.8 ± 1.5 μM, respectively (Cirilli et al., 2003). Mt-DapB occurs as a 100-kDa homotetramer (Kefala et al., 2005). The first reported structures for Mt-DapB were ternary complexes with NADH/NADPH and the inhibitor pyridine-2,6-dicarboxylic acid (2,6-PDC) (Cirilli et al., 2003). In both structures, the enzyme was observed in a proposed closed conformation (Cirilli et al., 2003). Subsequent structures of Mt-DapB have been solved in an apo form and also as a binary complex with its cofactor NADH (Janowski et al., 2009). The fold of Mt-DapB consists of an N-terminal Rossmann-like catalytic domain and C-terminal αβ sandwich tetramerization domain, which exhibit significant interdomain flexibility (Kefala et al., 2005; Janowski et al., 2009). A ribbon model of Mt-DapB is depicted in Fig. 2.
Inhibitors of DapB have been identified by molecular modelling as well as from a conventional screening of a Merck library and screened against the Mt-DapB enzyme (Paiva et al., 2001). A number of sulphonamide inhibitors of DapB were identified by the molecular modelling approach. The Ki values of the inhibitors ranged from 7 to 48 μM, and the compounds inhibited competitively with respect to the substrate 2,3-dihydrodipicolinic acid; however, the sulphonamide compounds lacked good antimicrobial activity (Paiva et al., 2001). Compared to the E. coli enzyme, Mt-DapB has a larger substrate or inhibitor binding site because of differences in the shape of the pocket at the N-terminal end of β8 (β9 in E. coli enzyme) and the nearby hinge region (Cirilli et al., 2003).
Mt-DapD (Rv1201c) catalyses the fifth step of the DAP pathway, the conversion of the cyclic tetrahydrodipicolinate (THDP) into the acyclic compound N-succinyl l-2-amino-6-oxopimelate using succinyl-CoA (Schuldt et al., 2009). Various structures of Mt-DapD have been obtained, both in native form and in complex with succinyl-CoA (Schuldt et al., 2008, 2009). A ribbon model of Mt-DapD is shown in Fig. 2. Mt-DapD forms a biologically relevant homotrimer, and each monomer is composed of three distinct domains – an N-terminal α/β-globular domain, a left- handed parallel β helix and a small C-terminal domain (Schuldt et al., 2008, 2009). The amino acid residues Glu 199 and Gly 222 of Mt-DapD are important for enzymatic activity. Mt-DapD is activated by Mg2+, Ca2+ and Mn2+ and inhibited by Co2+ and Zn2+ (Schuldt et al., 2009).
N-succinyldiaminopimelate aminotransferase (dapC or dap-AT)
The sixth step in this pathway is catalysed by Mt-DapC (Rv0858c), which transfers an amino group from l-glutamate and converts the substrate N-succinyl-2-amino-6-ketopimelate to N-succinyl diaminopimelate by the use of a pyridoxal phosphate (PLP) cofactor (Weyand et al., 2006, 2007). Mt-DapC belongs to the aminotransferase family of class I PLP-binding proteins. Mt-dapC has been heterologously expressed, purified and crystallized in two related crystal forms that arise from a pH difference between the crystallization conditions (Weyand et al., 2006). In the tetragonal crystal form, a monomer was present in the asymmetric unit, whereas in the orthorhombic crystal form, a dimer was present in the asymmetric unit (Weyand et al., 2006). Because of the presence of PLP in the crystal, both crystal forms appeared as pale yellow (Weyand et al., 2006). The three-dimensional structure of Mt-DapC was refined to a resolution of 2.0 Å (Weyand et al., 2007) and displayed the characteristic S-shape of class I PLP-binding proteins. Distinct from other class I PLP structures, Mt-DapC has an eighth β-strand inserted between strands three and four (Weyand et al., 2007). A ribbon diagram of Mt-DapC is shown in Fig. 2.
Mt-dapE (Rv1202) encodes the N-succinyl-l,l-diaminopimelic acid desuccinylase. DapE catalyses the hydrolysis of N-succinyl-l,l-diaminopimelic acid (SDAP) to l,l-diaminopimelic acid and succinate (Born et al., 1998; Davis et al., 2006). The enzyme is a metal-dependent peptidase (MEROPS family M28) catalysing the hydrolysis of substrate by water with the help of one or two metal ions located in the active site (Born et al., 1998; Nocek et al., 2010).
DapEs have been over-expressed and purified from Helicobacter pylori, E. coli, Haemophilus influenzae and Neisseria meningitidis (Bouvier et al., 1992; Karita et al., 1997; Born et al., 1998; Bienvenue et al., 2003; Badger et al., 2005). DapEs from E. coli and H. influenzae are small proteins (approximately 42 kDa) requiring two Zn2+ ions per mole of polypeptide for their activity (Bouvier et al., 1992; Born & Blanchard, 1999; Bienvenue et al., 2003). DapE proteins are homodimeric with each monomer containing a dimerization domain and a catalytic domain with a dinuclear zinc active site (Uda & Creus, 2011). H67 and H349 act as active site Zn2+ ligands in the H. influenzae DapE (Gillner et al., 2009b), with E134 shown to function as both a general acid and a general base during catalysis (Bzymek & Holz, 2004). DapE is activated by several divalent metal ions, including Zn2+, Co2+, Cd2+ and Mn2+ (Born et al., 1998; Bienvenue et al., 2003). In the presence of Mn2+, Salmonella typhi DapE functions as an aspartyl dipeptidase (Broder & Miller, 2003). DapE proteins are known to bind two divalent cations: one with high affinity (Zn2+) and the other with lower affinity (Mn2+) (Broder & Miller, 2003). DapEs exhibit a strict specificity for the l,l-isoform of SDAP (Bienvenue et al., 2003; Nocek et al., 2010).
Recently, the crystal structures of H. influenzae DapE with one or two zinc ions bound in the active site have been solved to 2 and 2.3 Ǻ resolution, respectively (Nocek et al., 2010). Previous to this, the 1.9 Å structure of the apo form of DapE from N. meningitidis containing no metal ions was reported (Badger et al., 2005; Gillner et al., 2009b). Neisseria meningitidis DapE has a catalytic domain (residues 1–179 and 295–381) interrupted by a dimerization domain (180–294), and residues His68, Asp101, Glu136, Glu164 and His350 are involved in binding the two zinc atoms (Badger et al., 2005). Zn K-edge-extended X-ray absorption fine structure (EXAFS) spectra of H. influenzae DapE enzyme provided structural information on the active site and also helped establish the binding modes of phosphonate- and thiolate-containing inhibitors (Cosper et al., 2003).
Two known competitive inhibitors of DapE are 2-carboxyethylphosphonic acid (CEPA) and 5-mercaptopentanoic acid (MSPA). The thiol group of MSPA binds to one or more of the Zn2+ ions in the active site of H. influenzae DapE (Cosper et al., 2003). Additionally, both l,l- and d,l-diaminopimelic acids are competitive inhibitors with respect to substrate (Born et al., 1998). A number of micromolar inhibitors of H. influenzae DapE were obtained by screening compounds containing zinc-binding groups which included thiols, carboxylic acids, boronic acids, phosphonates and hydroxamates (Gillner et al., 2009a).
The dapE deletion mutants generated in H. pylori and M. smegmatis were lethal and confirmed that dapE is essential for bacterial cell growth and proliferation (Pavelka & Jacobs, 1996; Karita et al., 1997; Davis et al., 2006). The H. pylori dapE deletion mutant was unable to grow in spite of the addition of lysine to the growth medium (Karita et al., 1997; Gillner et al., 2009b).
Diaminopimelic acid epimerase (dapF)
The racemization of amino acids provides meso-DAP which gets incorporated into bacterial PG (Koo & Blanchard, 1999) (Fig. 3). DAP epimerase (DapF) is a unique member of the family of pyridoxal phosphate–independent amino acid racemases, and its substrates (ll-DAP and meso-DAP) contain two stereocentres (Pillai et al., 2006). DapF discriminates between the two stereocentres, such that it interconverts the l,l and d,l but not the d,ddiastereoisomers of DAP by epimerization (Antia et al., 1957; Girodeau et al., 1986; Lloyd et al., 2004).
Attempts to express the Mt-dapF (Rv 2726c) in E. coli failed, in spite of the highly efficient T7 promoter in the pET28 vector. It was reasoned out that the lack of dapF expression was related to poor translation (Usha et al., 2006). Mt-dapF was subsequently cloned and over-expressed using a novel codon alteration strategy and the purified recombinant enzyme functionally characterized (Usha et al., 2006). The Km for meso-DAP was determined to be 1217 μM. Mt-DapF exists as a monomer. Dithiothreitol is required for Mt-DapF activity, consistent with its requirement for two reduced active site thiols (Usha et al., 2006).
Mt-DapF activity is inactivated in the presence of nanomolar concentrations of the three different thiol-specific alkylating agents (Usha et al., 2008). Site-directed mutagenesis confirmed that the two conserved Cys87 and Cys226 residues were involved in catalysis (Usha et al., 2008). The crystal structure of the unliganded form of Mt-DapF has been refined to 2.6 Ǻ resolution. Mt-DapF is made up of two pseudosymmetrical α/β domains (Usha et al., 2009). The active site is located in the cleft between domains I and II. The ribbon model of Mt-DapF is shown in Fig. 2. Tyr76 is unique to suborder Corynebacterineae DapF, suggesting a route to the design of a species-specific inhibitor (Usha et al., 2009).
Role of meso-DAP in mycobacterial cell wall peptidoglycan
In mycobacteria, and most Gram-negative bacteria, the third residue in the peptidoglycan (PG) pentapeptide is d,l (meso)-diaminopimelic acid (Schleifer & Kandler, 1972). During exponential phase, mycobacteria cross-link the third (meso-DAP) residue and the fourth (d-Ala) residue of adjacent stem peptides (Schleifer & Kandler, 1972; Wietzerbin et al., 1974). On entering stationary phase, mycobacteria incorporate increasing amounts of meso-DAP→meso-DAP linkages, which results in an unusually high DAP content (Wietzerbin et al., 1974; Cirillo et al., 1994a). meso-DAP is essential for both types of mycobacterial PG cross-linking. The percentage of cross-linking is very high (70–80%) in Mycobacterium species compared to E. coli (20–30%) (Cirillo et al., 1994b; Matsuhashi, 1994). meso-DAP is introduced into the PG network as part of the cross-linking moiety between the polysaccharide fibres (Ghuysen, 1980) (Fig. 3). In addition, the synthesis of meso-DAP is required for protein synthesis, because after decarboxylation, it yields l-lysine.
Orthologues in M. tuberculosis of most of the DAP biosynthesis enzymes have been stably expressed in soluble form and functionally characterized. The crystal structures of most of the DAP biosynthesis enzymes have been solved and the chemical mechanisms studied. Although the inhibitory concentrations of the best inhibitors of Mt-DapB remain in a low micromolar range, these results provide the basis for further optimization and development of more potent inhibitors. The M. tuberculosis DAP biosynthesis genes have been demonstrated to be essential for in vitro growth and are therefore attractive targets for the development of novel antitubercular drugs.