The structure of alanine racemase from Acinetobacter baumannii

The crystal structure of A. baumannii alanine racemase (AlrAba) from the highly antibiotic resistant NCTC13302 strain has been solved to 1.9 Å resolution. Comparison of (AlrAba) with alanine racemases from closely related bacteria demonstrates a conserved overall fold.


Introduction
Acinetobacter baumannii is a small nonmotile Gram-negative bacterium that is capable of replicating under a wide range of environmental conditions (Peleg et al., 2008). In vulnerable patients, A. baumannii can cause severe infections such as pneumonia, septicaemia, urinary-tract infections and meningitis (Lee et al., 2007). There are numerous strains of A. baumannii that are responsible for endemics and epidemics in hospitals (Scott et al., 2007;Eliopoulos et al., 2008) and many of these strains are antibiotic resistant (Zavascki et al., 2010). In fact, multidrug-resistant A. baumannii is considered to be among the most difficult Gram-negative bacteria to control and treat (Dijkshoorn et al., 2007). Therefore, new drug therapies are vital in combating these infections (Joshi et al., 2003;Gootz & Marra, 2008).
Alanine racemase (EC 5.1.1.1) is a pyridoxal 5 0 -phosphate (PLP)dependent enzyme which catalyses racemization between l-alanine and d-alanine (Walsh, 1989). d-Alanine is an essential building block of peptidoglycan, a critical component of the bacterial cell wall (Hols et al., 1997). Alanine racemase (Alr) is the sole source of d-alanine in many bacteria and its inhibition is lethal, making the enzyme an attractive antibiotic drug target (Lambert & Neuhaus, 1972). Kinetic studies of Alr have indicated that racemization occurs via a stepwise two-base mechanism, in which a highly conserved lysine and a tyrosine residue act as acid and base catalysts. These residues either accept or donate the -hydrogen of the alanine substrate, depending on the direction of the reaction (Watanabe et al., 2002;Spies & Toney, 2003).
Alanine racemase has been structurally characterized from numerous bacteria, including Geobacillus (formerly Bacillus) stearothermophilus (Shaw et al., 1997), Pseudomonas aeruginosa (LeMagueres et al., 2003), Streptomyces lavendulae (Noda et al., 2004), Mycobacterium tuberculosis (LeMagueres et al., 2005), Escherichia coli (Wu et al., 2008), Bacillus anthracis (Couñ ago et al., 2009), Enterococcus faecalis (Priyadarshi et al., 2009), Bartonella henselae (Abendroth et al., 2011), Streptococcus pneumoniae (Im et al., 2011), Staphylococcus aureus (Scaletti et al., 2012) and Oenococcus oeni (Palani et al., 2013). In these structures, the enzyme is a homodimer in which the two monomers interact in a head-to-tail fashion, generating two active sites. Each Alr monomer has two distinct domains: an /-barrel at the N-terminus and a predominantly -stranded domain at the C-terminus. The PLP cofactor forms an internal aldimine linkage with the highly conserved catalytic lysine residue in the /-barrel domain (Shaw et al., 1997). Here, we report the structure and partial kinetic characterization of A. baumannii alanine racemase (Alr Aba ) from the multidrug-resistant NCTC13302 strain. A crystallization report concerning alanine racemase from this bacterium has appeared in this journal (Nguyen et al., 2013), but no structure is available and it was not consulted for this work. The structure presented here will provide a template for future structurebased drug-design studies targeting this important enzyme.

Gene cloning, expression and protein purification
Prior to cloning, genomic DNA was extracted from A. baumannii strain NCTC13302, after which the alanine racemase alr gene was amplified by PCR using the following primers: 5 0 -ATGATCTACA-TATGCGTCAAGCAACAGTT-3 0 (forward) and 5 0 -TATCTCGAG-TTAAGTACCCTGACGGAC-3 0 (reverse). Full-length A. baumannii alr (residues 1-356) was cloned into pET-26b, which was transformed into E. coli BL21 (DE3) cells. Cells grown overnight at 310 K were used to inoculate the main culture, which was induced with 0.5 mM IPTG at an OD 600 of 0.5. After expression for 16 h, the cell pellet was collected by centrifugation and the cells were lysed via sonication. Following ammonium sulfate cuts of 20 and 60%, the pellet fraction was resuspended in 20 mM Tris pH 8.0 and further purified by hydrophobic interaction, anion-exchange and size-exclusion chromatography. Fractions containing Alr Aba were greater than 95% pure as indicated by SDS-PAGE.

Crystallization
Purified Alr Aba was concentrated to 6 mg ml À1 using a Vivaspin 20 (10 000 Da molecular-weight cutoff; GE Healthcare) and sitting drops were set up versus 100 mM 2-(N-morpholino)ethanesulfonic acid (MES) pH 6.0, 200 mM CaCl 2 , 20% polyethylene glycol 6000 (PEG 6000). Single yellow crystals grew within 4 d and were cryoprotected by soaking in mother liquor containing 20% glycerol. Crystals were flash-cooled in liquid nitrogen prior to data collection.

Data collection and processing
A native Alr Aba data set was collected at the Australian Synchrotron (Melbourne, Australia) on the MX1 beamline equipped with an ADSC Quantum 315r image-plate detector. Data were collected at 13.000 keV using an oscillation angle of 0.5 and an exposure time of 2 s per image. Diffraction images were processed with iMosflm (Battye et al., 2011), POINTLESS (Grosse-Kunstleve et al., 2002 and SCALA (Evans, 2006) within the CCP4 suite . The crystals of Alr Aba belonged to the monoclinic space group P2 1 , with unit-cell parameters a = 47.0, b = 83.0, c = 93.3 Å , = 90, = 97.1, = 90 . Data were collected and processed to 1.9 Å resolution with statistics as presented in Table 1.

Structure determination and refinement
The structure of Alr Aba was solved via molecular replacement with Phaser (McCoy et al., 2007) using the monomer of P. aeruginosa alanine racemase (with ligands and waters removed), to which it has a sequence identity of 41%, as the search model. This was performed assuming the presence of two monomers in the asymmetric unit, as suggested by the Matthews coefficient V M of 2.07 Å 3 Da À1 (Matthews, 1968). PHENIX (Adams et al., 2010) was used to build the initial model and to change the amino-acid sequence to that of Alr Aba . After several rounds of model building and refinement using Coot (Emsley & Cowtan, 2004) and refinement using REFMAC5 (Murshudov et al., 2011), the electron density improved and waters and the cofactor PLP were built into the structure. The final structure had an R factor of 19.7% and an R free of 23.4%. The r.m.s. deviations of bond lengths and angles were 0.016 Å and 1.75 , respectively. Intermonomer interactions in the structure were analysed using the Protein Interfaces, Surfaces and Assemblies service (PISA) at the European Bioinformatics Institute (http://www.ebi.ac.uk/pdbe/ prot_int/pistart.html; Krissinel & Henrick, 2007). Comparisons to other structures were made with the respective PDB files as deposited. The final model was validated using PROCHECK (Laskowski et al., 1993), with the resulting Ramachandran plot indicating that 98.2% of the residues were in the most favoured regions, with 1.6% in additionally allowed regions. Additional structure-determination and refinement statistics are presented in Table 1.

Enzyme kinetics
The d-alanine to l-alanine direction of the racemization catalysed by Alr Aba was characterized using a coupled-enzyme assay based on Esaki & Walsh (1986) and as described in previous studies (Strych et al., 2000(Strych et al., , 2001. This assay was carried out at 30 C and begins with the conversion of d-alanine to l-alanine by alanine racemase. The l-alanine is then converted to pyruvate and ammonia by l-alanine dehydrogenase, producing NADH, which is tracked by following the increase in absorbance at 340 nm. Concentrations of d-alanine ranging from 0.1 to 10 mM were assayed in triplicate using 200 ng Alr Aba per reaction. The production of NADH was monitored for  Table 1 Data-collection and refinement statistics.
Values in parentheses are for the highest resolution shell.
10 min with the slope calculated during the last 2 min of the assay. The kinetic constants K m and V max were determined using nonlinear regression fitting carried out within GraphPad Prism v.6 (GraphPad Software, La Jolla, California, USA).

Results and discussion
3.1. Overall structure of A. baumannii alanine racemase The tertiary structure of A. baumannii alanine racemase (Alr Aba ) is a homodimer in which the two monomers interact in a head-to-tail manner. This results in two active sites per enzyme, each comprised of residues from the N-terminal domain of one monomer and the C-terminal domain of the second monomer (Fig. 1a). Each Alr Aba monomer contains two distinct domains (Fig. 1b). The N-terminal domain corresponds to residues 1-230 in the structure and predominantly consists of an eight-stranded /-barrel. The C-terminal domain corresponds to residues 231-356 in the structure and mainly contains -strands (two -helices and eight -strands). The individual Alr Aba monomers are crystallographically distinct and form a dimer in the asymmetric unit. Following refinement, they have a low r.m.s difference of 0.41 Å after C -atom superposition. Overall, the Alr Aba structure is lacking clear density for residues 247-263 and 355-356 of monomer B. Missing density for these regions has previously been observed in alanine racemases from other species, including M. tuberculosis (LeMagueres et al., 2005), B. henselae (Abendroth et al., 2011), S. aureus (Scaletti et al., 2012) and O. oeni (Palani et al., 2013).
The essential PLP cofactor is covalently bound to Lys34 via an internal aldimine linkage and extends towards the centre of the /barrel. The pyridine N 1 of the PLP ring is stabilized by hydrogen bonding to Arg209, which is further stabilized by interactions with His159. The phosphate tail of PLP is stabilized by several residues from one monomer. The OP 1 of the phosphate group hydrogen bonds to Ile212 and Tyr38, OP 2 hydrogen bonds to Try341 and OP 3 hydrogen bonds to Ile212 and Ser190 (Fig. 2). Arg132 usually interacts with the phenolic O atom (O3 0 ) of PLP, aiding in maintaining the position of the cofactor (Shaw et al., 1997). However, Arg132 was not within hydrogen-bonding distance of PLP in the Alr Aba structure. There was no additional density in the Alr Aba structure consistent with the presence of any additional ligands besides PLP.
3.2. Structural and biochemical comparison with closely related alanine racemases 3.2.1. Overall topology, individual domains and inter-monomer hinge angle. The multiple sequence alignment presented in Fig. 3 indicates a high level of sequence similarity between Alr Aba and the closely related alanine racemases from P. aeruginosa (DadX Pao ), B. henselae (Alr Bhe ) and E. coli (Alr Eco ). The PLP-binding motif containing the catalytic Lys34 (sequence SMVKANAYGHG) is largely conserved between the various enzymes. In addition, residues which contribute to the inner and middle layers of the active-site entryway of the enzymes are also strongly conserved (Fig. 3). As noted previously, Alr Aba lacks electron density in the region containing Tyr254 (in monomer B only), which previous structural and kinetic studies have indicated to be part of the inner layer of the active-site entryway and the second catalytic residue involved in the reaction mechanism (Shaw et al., 1997;Spies & Toney, 2003).
Superposition of the C atoms of the Alr Aba monomer with alanine racemases from Gram-negative bacteria (DadX Pao , Alr Bhe and Alr Eco ) indicates a high level of structural similarity between the enzymes ( Table 2). The Alr Aba monomer is most similar to Alr Eco and DadX Pao , with which it shares the lowest r.m.s differences (1.30 Å ) and the highest sequence similarity (41%). Alr Aba is less similar to Alr Bhe , with which it has a higher r.  domains and active site of Alr Aba with the other alanine racemases also indicates high structural similarity ( Table 2).
The N-terminal domain of Alr Aba superimposed best with those from Alr Eco and DadX Pao , with which it shares the highest sequence identity (40 and 39%, respectively) and the lowest r.m.s differences (1.32 and 1.30 Å , respectively). Consistent with the superpositions involving whole monomers, the N-terminal domain was less similar to that of Alr Bhe (Fig. 4a). In the C-terminal domain, Alr Aba superimposed equally well with Alr Bhe , Alr Eco and DadX Pao (1.07, 1.02 and 1.08 Å , respectively; Fig. 4b). Analysis of the active sites indicates that this region superimposed the best, having much lower r.m.s differences than whole monomers or individual domains (Alr Bhe , 0.91 Å ; Alr Eco , 0.65 Å ; DadX Pao , 0.56 Å ). This indicates that the active site is highly conserved between the alanine racemase structures in spite of structural deviations between their domains.
Previous studies of alanine racemases have indicated that the angle between the N-terminal and C-terminal domains of the individual monomers (the inter-monomer hinge angle) differs between enzymes Structure-based sequence alignment of alanine racemases from A. baumannii, E. coli, B. henselae and P. aeruginosa. Identical residues are shaded black, while grey shading indicates amino acids with conserved physicochemical properties. An asterisk marks the highly conserved PLP-bound lysine and a black diamond marks the location of the catalytic tyrosine, while a black circle indicates the location of a residue which is often carbamylated in alanine racemases that have a lysine at this position. The purple box encloses the conserved PLP-binding motif containing the catalytic lysine. The red box indicates the region of electron density missing from monomer B of Alr Aba . I and M represent residues which form the inner and middle layers of the active-site entryway.    32-38, 54-58, 75-79, 95-99, 121-134, 154-162, 188-195, 206-213 and 338-345 from monomer B and 252-255, 272-276 and 299-304 from monomer A for Alr Eco and DadX Pao and vice versa for Alr Bhe . correlation between sequence identity and inter-monomer hinge angle. To date, the best explanation for hinge-angle differences involves the hydrogen-bond interactions formed between residues from the N-and C-terminal tails of opposite monomers (LeMagueres et al., 2003(LeMagueres et al., , 2005Couñ ago et al., 2009). Alr Bhe and alanine racemases from G. stearothermophilus (Shaw et al., 1997), B. anthracis (Couñ ago et al., 2009) and S. aureus (Scaletti et al., 2012) have additional residues in these areas capable of forming these hydrogen bonds, resulting in similar hinge angles. Shorter alanine racemases such as Alr Eco , DadX Pao and Alr Aba do not have these additional residues (Fig. 3), resulting in differing hinge angles.

Enzyme kinetics.
Kinetic characterization of Alr Aba revealed that the enzyme has a V max of 11.3 U mg À1 and a K m of 0.56 mM for the racemization of d-alanine to l-alanine (Table 3). The other direction of the racemization was not characterized in this study. This is very similar to the kinetic values reported for Alr Eco (V max = 8.8 U mg À1 , K m = 0.31 mM), with which it shares high structural similarity (Table 2) and a similar hinge angle. However, despite Alr Aba sharing a high structural similarity and hinge angle with DadX Pao , the kinetic parameters differed significantly between these enzymes (V max = 134 U mg À1 and K m = 1.40 mM for DadX Pao ). This indicates that despite high levels of sequence identity, similar activesite structures and hinge angles, the kinetics parameters of even closely related alanine racemases can differ. This is in agreement with what has been observed in previous studies (Couñ ago et al., 2009;Scaletti et al., 2012). No comparison could be made with Alr Bhe as no kinetic information is available.

Dimer interface, substrate entryway and active site.
Previous studies have found that alanine racemase is dependent on dimerization for enzyme activity, and that a number of residues involved in the dimer interface are highly conserved (Strych & Benedik, 2002;Im et al., 2011;Scaletti et al., 2012). This makes the dimer interface a potential target for structure-aided drug design. The area of the dimer interface of Alr Aba is 2360 Å 2 and is formed by roughly equal contributions from both monomers (73 and 70 residues from monomer A and monomer B, respectively). This interface contains five salt bridges and 22 hydrogen-bond interactions. The dimer interface of Alr Aba is most similar to the interface calculated  Table 3 Kinetic parameters for the racemization between l-and d-alanine by alanine racemases.
for Alr Bhe (2606 Å 2 ), which has similar numbers of residues involved in dimerization (71 and 75 in monomers A and B, respectively). The interface is less similar to the interfaces calculated for DadX Pao (1918 Å 2 ) and Alr Eco (2816 Å 2 ), both of which have a larger number of residues involved in dimerization (DadX Pao , 82 and 82 in monomers A and B, respectively; Alr Eco , 81 and 83 in monomers A and B, respectively).
The substrate entryway of Alr Aba is comprised of residues from both monomers and consists of an outer, a middle and an inner layer (with the inner layer being in closest proximity to the PLP cofactor and the outer layer being on the surface of the protein). Comparison of Alr Aba with the other enzymes indicates that the inner (Ala163, Tyr254 0 , Tyr273 0 and Tyr341) and middle (Asp164, Arg279, Arg299 and Ile339) layers of the entryway are highly conserved (Fig. 3), consistent with previous studies (LeMagueres et al., 2005;Im et al., 2011). The highly conserved nature of the alanine racemase substrate entryway makes it a potential target for the development of enzymespecific inhibitors.
The active-site residues of Alr Aba and other alanine racemases superimpose particularly well, with low r.m.s differences ( Table 2). The active-site structure of Alr Aba is most similar to that of DadX Pao (0.56 Å ; 58% sequence identity). Alr Aba also shares high similarity to Alr Eco (0.65 Å ; 60% sequence identity), but diverges most from Alr Bhe (0.91 Å ; 48% sequence identity). This indicates a positive correlation between sequence identity and structural similarity in this region. As mentioned in x3.1, the active site of Alr Aba is composed of residues from both monomers, several of which are involved in a hydrogen-bond network with the PLP cofactor (Figs. 2 and 5a) and are conserved between alanine racemases (Fig. 5b).
Alr Aba is lacking density for Tyr254 0 in one active site and previous studies have shown that this residue is important in substrate binding and catalysis (Shaw et al., 1997;Morollo et al., 1999;Watanabe et al., 2002). As noted above, missing density for this residue has previously been observed in the active sites of several alanine racemases (LeMagueres et al., 2005;Abendroth et al., 2011;Scaletti et al., 2012;Palani et al., 2013). His159 can be identified in the same active site and its side chain forms a hydrogen bond to Arg209, as well as a parallel displaced -stacking interaction with the PLP ring. The closest contact distance for this interaction is $3 Å , which is consistent with other reports in proteins (McGaughey et al., 1998).
In the second active site Tyr254 0 is present but the His159 side chain is difficult to localize, with little density beyond the -carbon. A hybrid view with these residues included from opposite active sites (Fig. 5b) reveals that these two residues are located in their usual positions when compared with the active sites from other bacterial alanine racemase structures.
Differences between the Alr Aba active site and the other alanine racemase structures included a lack of density consistent with Lys125 carbamylation (Figs. 3 and 5b). This modification is present in both Alr Eco and DadX Pao ; it is proposed to aid in the positioning of Arg132 within the active site (Morollo et al., 1999;LeMagueres et al., 2003;Wu et al., 2008) and is a process that is dependent on a relatively high pH (>8.0; Golemi et al., 2001). It is possible that the observed lack of carbamylation in the Alr Aba structure is owing to the low pH (6.0) at which the enzyme was crystallized. Furthermore, Alr Aba did not contain any extra density consistent with other molecules in its active site. Previous alanine racemase structures have reported the presence of additional molecules in the active site such as sulfate (Scaletti et al.,  Active site of A. baumannii alanine racemase. (a) 2F o À F c electron-density map of the active site contoured at 1.0 with isomesh map shown (1.6 Å carve). The main-chain and side-chain atoms of the Alr Aba active-site residues are depicted as sticks. C atoms are green, O atoms red, N atoms blue, S atoms yellow and phosphates orange. The PLP cofactor is depicted as a ball-and-stick model in which C atoms are coloured black. (b) Superposition of the active-site residues of alanine racemases from A. baumannii (green), B. henselae (blue), P. aeruginosa (red) and E. coli (orange). For A. baumannii, a hybrid view is depicted with residues from monomer B, except for the side chain of His159 which is included from monomer A. The PLP cofactors from each structure are depicted as ball-and-stick models. Primes denote residues contributed by the second monomer. The superposition was performed using the residue ranges stated in Table 2. This figure was produced in PyMOL (DeLano, 2002). 2012), chloride (Couñ ago et al., 2009) and acetate (Shaw et al., 1997;Couñ ago et al., 2009;Scaletti et al., 2012). The lack of substrate-like molecules in the active site could also be a contributory factor to the unusual position of Arg132 owing to this residue playing an integral role in substrate binding.

Conclusions
We have reported the crystal structure of A. baumannii alanine racemase from the highly antibiotic-resistant NCTC13302 strain to 1.9 Å resolution. The overall structure is very similar to closely related alanine racemases from Gram-negative bacteria. Interestingly, the partial kinetic analysis generated in this study indicated an inconsistent relationship between structural similarity and racemization rate. The substrate entryway and active sites of these alanine racemases were shown to be highly conserved and are thus possible targets for the development of enzyme-specific inhibitors. The structural determination and biochemical analysis presented here could provide a template for future structure-based drug-design studies targeting Alr Aba .