Structural characterization of aspartate-semialdehyde dehydrogenase from Pseudomonas aeruginosa and Neisseria gonorrhoeae

Gonorrhoea infection rates and the risk of infection from opportunistic pathogens including P. aeruginosa have both risen globally, in part due to increasing broad-spectrum antibiotic resistance. Development of new antimicrobial drugs is necessary and urgent to counter infections from drug resistant bacteria. Aspartate-semialdehyde dehydrogenase (ASADH) is a key enzyme in the aspartate biosynthetic pathway, which is critical for amino acid and metabolite biosynthesis in most microorganisms including important human pathogens. Here we present the first structures of two ASADH proteins from N. gonorrhoeae and P. aeruginosa solved by X-ray crystallography. These high-resolution structures present an ideal platform for in silico drug design, offering potential targets for antimicrobial drug development as emerging multidrug resistant strains of bacteria become more prevalent.

the emergence of resistant strains of N. gonorrhoeae [13][14][15][16][17] . Of further concern, N. gonorrhoeae is a cofactor that can facilitate HIV-1 infection 18 . In 2015, in response to the threat of antimicrobial resistance the World Health Organization (WHO) adopted a global action plan. The plan briefly involves increased awareness and understanding, increased surveillance and research, increased preventative measures, increased investment for treatment, diagnosis, vaccination and other interventions, and to ensure that the plan is inclusive for all countries 19 . The design of new antimicrobial drugs is critical to counter emerging multidrug-resistant pathogenic microorganisms.
ASADH has been identified as a target for antimicrobial drugs in both pathogenic and non-pathogenic microorganisms including gram-negative bacteria 20,21 , gram-positive bacteria 22 , Mycobacterium tuberculosis 23 , and archael 24 and fungal species 2,25,26 . Humans and other mammals do not have a homolog of ASADH or require the aspartate biosynthetic pathway, making this a highly attractive pathway to target in microorganisms. Some organisms possess a second, redundant ASADH enzyme, which despite sharing a high structural similarity has less than 50% sequence identity to other functional ASADH proteins 27 . Despite ASADH sharing an overall high structural similarity between fungal and bacterial orthologs, secondary structural differences have been identified that offer the potential for the development of species-selective antimicrobial drugs 2 .
Several antibiotic resistance genes have been identified within the P. aeruginosa genome and antibiotic resistance through horizontal gene transfer has been demonstrated 8,28 . Therefore, new inhibitors are required to combat antibiotic resistance. For microorganisms that use the aspartate biosynthesis pathway for the production of essential amino acids and metabolites, inhibition of ASADH is fatal 29 . The asd gene encoding ASADH is one of the minimal genes essential for life in these microorganisms 7,[30][31][32][33] .
Currently, inhibitors are used in combination with various antibiotics on antimicrobial-resistant pathogens 34 and the use of antibiotic adjuvants can effectively reduce bacterial resistance to commonly used antibiotics 35,36 . Selective inhibition of key enzymes for biosynthesis pathways in microorganisms can be used to counter antimicrobial drug resistance of pathogenic species such as P. aeruginosa. To provide an enhanced understanding of the structural similarities of ASADH between important microorganisms, as well as to provide a platform for the design of new anti-microbials, here we solve and analyze the structures of two ASADH enzymes from N. gonorrhoeae and P. aeruginosa.

Materials and methods
ASADH proteins were purified using standardized protein expression and purification protocols as described previously 37 . Briefly, open reading frames encoding ASADH were amplified from genomic DNA and cloned into an expression vector encoding an N-terminal non-cleavable 6-His tag. The proteins were expressed in BL21(DE3) bacterial cells and then purified by nickel affinity and size-exclusion chromatography 37 . The proteins were concentrated to ~ 20 mg/mL and stored at − 80 °C until used in crystallography experiments. The sparse matrix screening method was used to produce crystals from purified ASADH proteins via the sitting drop vapor diffusion method. Crystals of P. aeruginosa ASADH (SSGCID ID: PsaeA.17885.a.B1.PS02341) were grown at 19.6 mg/ml with 4 mM NADP, and mixed 1:1 with Microlytic MCSG1 screen, well D9 (containing 0.2 M sodium chloride, 0.1 M TRIS:HCl pH 8.50, 25% PEG3350) at 289 K. Crystals harvested were cryo-protected with crystallant supplemented with 15% ethylene glycol and 5 mM NADP and flash-cooled in liquid nitrogen. X-ray diffraction data with a resolution of 2.1 Å was collected at a temperature of 100 K on a RAYONIX MX-225 CCD detector at a wavelength of 0.9787 Å on the Advanced Photon Source (APS, Argonne, IL) 21-ID-F beamline. Crystals from N. gonorrhoeae ASADH (SSGCID ID: NegoA.17885.a.B1.PW37925) were obtained in a condition from Microlytic MCSG1 screen (F11) containing 25% PEG 3350, 200 mM ammonium sulphate, HEPES free acid/NaOH buffer pH 7.5 at temperature 290 K. X-ray diffraction data with a resolution of 2.1 Å was collected at a temperature of 100 K on the RIGAKU SATURN 944 + CCD detector at a wavelength of 1.5418 Å. The diffraction data were reduced using the XDS/XSCALE package 38,39 . The structures of the P. aeruginosa and N. gonorrhoeae ASADH proteins were solved by molecular replacement using ASADH proteins as starting models from Vibrio cholerae (1MB4) in BALBES 40 and Burkholderia thailandensis (3UW3) in MoRDa respectively. Protein structures were refined using PHENIX 41 . Protein structures shown were modelled using PyMOL and SeeSAR version 9.2 (BioSolveIT GmbH) software 42 . SeeSAR was utilised to model docking of the SMCS molecule with ASADH proteins. SeeSAR calculates estimated affinities and allows the selection of best poses based on HYDE scoring and analysis of torsion. The default parameter settings were used. The model adopted a similar binding site for docking as defined by "find unoccupied pocket". From the molecular docking, parameters such as estimated affinity and torsion were acknowledged in selecting the best model. DALI 43 (heuristic PDB search) protein structure comparison by alignment of distance matrices were performed. The final protein structures were deposited into the Protein Data Bank (PDB) and assigned the codes 6BAC for N. gonorrhoeae ASADH and 5BNT for P. aeruginosa ASADH. www.nature.com/scientificreports/

Results and discussion
Structural analysis of ASADH proteins from N. gonorrhoeae and P. aeruginosa. The structures of ASADH proteins from N. gonorrhoeae and P. aeruginosa were determined using x-ray crystallography, deposited into the Protein Data Bank and assigned the PDB codes 6BAC and 5BNT, respectively ( Table 1). Crystals of both ASADH proteins produced x-ray diffraction data with a resolution of 2.1 Å. The crystallographic structure from N. gonorrhoeae ASADH was solved in the space group P 31 2 1, with unit cell dimensions of a = 109 Å, b = 109 Å, and c = 72.4 Å, and one ASADH protein was present in the asymmetric unit. Continuous electron density allowed modelling of all 371 residues and included two His residues of the non-cleavable N-terminal polyhistidine tag (MAHHHHHH) present in the structure at the N terminus of the protein. The enzyme structure adopted a well-described coenzyme binding domain at the N-terminus (residues 1-150) comprised of six α-helices and seven β-strands, and a catalytic and dimerization domain at the C-terminus (residues 152-351; Fig. 2A,B). The parallel β-sheet flanked by two alpha-helical bundles forms a Rossmann fold, which typically binds enzyme cofactors including β-nicotinamide adenine dinucleotide (NAD) and β-nicotinamide adenine dinucleotide phosphate (NADP) 44 . The highest structural homology to N. gonorrhoeae ASADH, was the ASADH enzyme from B. thailandensis (3UW3) 45 , with an RMSD of 0.7 Å (Table 2).
To determine whether N. gonorrhoeae ASADH was likely to form a multimer, we examined the crystal structure by expanding the asymmetric unit. The biological assembly for the ASADH enzyme (6BAC), predicted using Proteins, Interfaces, Structures and Assemblies (PISA), was a dimer with a large interface present between two protomers 46 (Fig. 2C). Overall, the protein interface exhibits 52 hydrogen bonds, 18 salt bridges and 3406 Å 2 of buried surface area. This biological interface was structurally equivalent to the ASADH protein dimer from B. thailandensis (PDB 3UW3) which exhibited 47 hydrogen bonds, 17 salt bridges and 3463 Å 2 of buried surface area.
The structure of the P. aeruginosa ASADH was solved in the space group P 41 21 2, with unit cell dimensions of a = 125.68, b = 125.68, and c = 199.44. The asymmetric unit contained four ASADH protomers arranged as two dimers. The electron density allowed for modelling of residues 1-370. Each monomer is comprised of an N-terminal coenzyme-binding domain (residues 1-133) and a C-terminal catalytic and dimerization domain (residues 135-351), with similar architecture and topology to the ASADH from N. gonorrhoeae (Fig. 3A,B). PISA analysis identified a dimer as the biological unit (Fig. 3C), with the protein interface comprised of 56 hydrogen bonds, 23 salt bridges and 3558 Å 2 of buried surface area 46 . This interface was comparable to ASADH enzymes exhibiting NADP binding including Haemophilus influenzae ASADH (1PQU) involving 52 hydrogen bonds, 18 salt bridges and 3415 Å 2 of buried surface area, and V. cholerae ASADH (3PZR) involving 60 hydrogen bonds, 24 salt bridges and 3451 Å 2 of buried surface area. In addition, there was a high structural similarity between P. www.nature.com/scientificreports/  NADP binding in P. aeruginosa ASADH. P. aeruginosa ASADH was co-crystallized with NADP. Each monomer of P. aeruginosa ASADH bound a single NADP molecule (Fig. 4A,B). To examine the conservation of NADP binding residues, structurally similar ASADH proteins exhibiting NADP binding were compared (Fig. 4C). We found many NADP binding residues to be strictly conserved between P. aeruginosa (5BNT) and two structurally similar ASADH proteins; H. influenzae (1PQU) and V. cholerae (3PZR) (Fig. 4D). Overall, the P. aeruginosa ASADH-NADP interface was mediated through 14 hydrogen bonds with a buried surface area of 582.6 Å 2 . Comparable ASADH-NADP binding was observed in H. influenzae (1PQU) and V. cholerae (3PZR) involving 13 hydrogen bonds, with buried surface areas of 505.8 Å 2 and 502.8 Å 2 respectively. ASADH protein is previously described as having two states, apoenzyme and the NADP-inhibitor complex (with SMCS) 20 . Changes in enzyme conformation are attributed to NADP binding and involve the repositioning of two surface loops adjacent to the NADP binding site. Here, we show that the NADP binding of ASADH in P. aeruginosa shows a different orientation of the nicotinamide ring relative to other ASADH structures (Fig. 4C). The position of the nicotinamide ring is constrained through hydrogen bonds (Fig. 4A). As binding at the active site is predicted to cause a conformational change in the conserved loops enclosing the bound NADP, the difference in positioning likely affects the ability of NADP to stabilise the ASADH protein dimer. However, as the active site is relatively unchanged with NADP binding, active site binding is likely not significantly affected. This should be further explored using modelling approaches of docking inhibitors.
Comparison to known structures. To determine the sequence similarity of the ASADH proteins to other ASADH proteins of known structures, Clustal Omega 49 sequence alignment tools were used to generate a phylogenetic tree (Fig. 5). The selection of ASADH protein sequences for comparison was dependent on the availability of ASADH crystallized structures, which represent primarily organisms of pathogenic interest. P. aeruginosa and N. gonorrhoeae ASADH protein sequences are most closely related to ASADH proteins from B. thailandensis and V. cholerae.
To determine the most structural similar ASADH proteins available from those with structures available on the PDB, a homology search using DALI (heuristic PDB search) was performed 48 using N. gonorrhoeae ASADH (6BAC) and P. aeruginosa (5BNT, chain B). Only one representative structure is listed for each organism ( Table 2). The protein with the most sequence similarity to ASADH proteins from P. aeruginosa and N. www.nature.com/scientificreports/  www.nature.com/scientificreports/ gonorrhoeae is B. thailandensis ASADH (3UW3) 45 . This protein had a sequence similarity of 73% and RMSD of 0.8 Å and a sequence similarity of 71% and RMSD of 0.7 Å for P. aeruginosa and N. gonorrhoeae ASADH proteins respectively. This was expected for P. aeruginosa ASADH as Burkholderia (previously a part of Pseudomonas) is a gram-negative pathogenic group of bacteria that are closely related to Pseudomonas 50 . Notably, when comparing the two structures presented here, the NADP-bound ASADH from P. aeruginosa (5BNT) had an RMSD 0.8 Å and 68% sequence similarity to N. gonorrhoeae ASADH (6BAC). The ASADH protein structures with the next most similar structural homology to the P. aeruginosa ASADH (5BNT) is V. cholerae ASADH. The V. cholerae ASADH protein was crystallized with NADP and product of S-carbamoyl-L-cysteine (3PZR), S-allyl-L-cysteine sulfoxide (3Q0E) and NADP and S-methyl-L-cysteine sulfoxide (1MB4) 6,20 . The crystal structures showed an RMSD 0.5 Å and sequence similarity 69% to P. aeruginosa ASADH (5BNT). Comparatively, ASADH protein from N. gonorrhoeae (6BAC) showed RMSD 0.8-0.9 Å and a sequence similarity of 66% to V. cholerae ASADH.
A multiple sequence alignment was generated using the ASADH proteins identified as having a high structural similarity from the DALI heuristic PDB search, highlighting highly conserved residues and binding motifs of ASADH proteins (Fig. 6). Conserved protein sequences include a G-G-G binding motif for nucleoside binding and a T-QA-SG-G binding motif predicted to have a role in active site communication 5 . ASADH proteins also contain two conserved active site residues; Cys135, which initiates catalysis and His274 (H-bond interaction) 51 . The STS binding motif previously reported in NADP binding of ASADH proteins is not conserved in the N. gonorrhoeae ASADH (6BAC) sequence, which may affect binding affinity.
Studies to investigate the binding of inhibitors to ASADH both experimentally and using virtual modelling have predicted that a successful ASADH inhibitor will bind through electrostatic interactions with two highly conserved active site arginine residues 4 . Arginine residues within the active site from S. pneumoniae ASADH (Arg99, Arg245) and V. cholerae ASADH (Arg101, Arg267) bind the phosphate and the carboxyl group moieties www.nature.com/scientificreports/ of the aspartate β-semialdehyde (ASA) substrate, respectively. The residue Arg99-103 in bacterial ASADH proteins corresponds to Arg112-116 in fungal and archaeal ASADH proteins (Fig. 6, green box).

SMCS binding to ASADH.
To determine whether the binding pocket for an active site inhibitor of ASADH, S-methyl-L-cysteine sulfoxide (SMCS), in a structurally similar ASADH protein is conserved in P. aeruginosa (5BNT) and N. gonorrhoeae (6BAC) ASADH proteins, we performed a structural alignment of the binding pockets. Binding of SMCS was modelled from the V. cholerae ASADH (1MB4) structure (Fig. 7A,B). Inactivation of ASADH by SMCS (Cys) is a result of disulfide bond formation at Cys134. The H-bond interaction at His274 is also highly conserved and critical for mediating SMCS interaction. Other H-bond interactions with SMCS in V. cholerae ASADH (1MB4) also occur at Asn133, Gln161, Glu240 and Arg267 (Fig. 7C,D). The docking results show binding site similarity between existing literature and support the prediction that the SMCS molecule could be used to target the P. aeruginosa and N. gonorrhoeae ASADH proteins. Several small molecule inhibitors of ASADH proteins with high binding efficiency have been identified (Table 3). Where some ASADH inhibitors are able to selectively inhibit ASADH in gram negative bacteria, other ASADH inhibitors have been shown to selectively inhibit fungal ASADH proteins 52 . There is an increasing need to design small molecule inhibitors that can selectively target pathogenic microorganisms that display antibiotic resistance.
P. aeruginosa has demonstrated antibiotic resistance to β-lactams, aminoglycosides, quinolones and other antibiotics 8 . Recently, whole genome sequencing of Pseudomonas spp. revealed antibiotic resistance genes to various antibiotics 28 . New antimicrobial drugs are required to treat P. aeruginosa infection and to combat antibiotic resistance 10 . New therapies that offer selective inhibition of key enzymes in pathogenic microorganisms may be an alternative or used in combination with current antibiotics 8 . Docking software can be used to predict structural and kinetic binding of small molecules. The ASADH protein from V. cholerae (1MB4) was used here to model SMCS binding within the newly solved structures of P. aeruginosa (5BNT) and N. gonorrhoeae (6BAC) ASADH proteins. This ASADH protein showed high structural and sequence similarity to the protein structures presented here, in particular, within conserved active site residues and binding site cavities. Both P. aeruginosa and N. gonorrhoeae ASADH proteins are potential targets for the small molecule inhibitor SMCS. Here, SMCS binding is modelled within ASADH, however, there are various other inhibitors, catalytic intermediates and substrate analogues including phosphonamidites analogues of the substrate aspartyl phosphate 53 that could be modelled to predict binding. Selective inhibitors of ASADH proteins have been identified using screening tools and a fragment library containing 378 compounds 52,54,55 , virtual screening of databases 23 and molecular docking approaches 4,56 . Due to structural similarities with other characterised ASADH proteins, we predict that various other ligands could be modelled within ASADH protein structures presented in this paper via screening of inhibitors shown to bind ASADH proteins.

Conclusion
Antimicrobial drug resistance is an urgent and ongoing medical threat. High-resolution structures of enzymes that are critical for microbial survival represent an important starting point for the development of new antibiotics. Here we present the crystallographic structures P. aeruginosa and N. gonorrhoeae ASADH proteins, that represent ideal platforms for potential in silico drug screening. www.nature.com/scientificreports/