Structures of the Peptidoglycan N-Acetylglucosamine Deacetylase

University of Crete, Vasilika Vouton, 70013 Heraklion, Crete, Greece. Biochemistry , Article ASAP. DOI: 10.1021/acs.biochem.7b00919. Publication D...
0 downloads 6 Views 2MB Size
Subscriber access provided by READING UNIV

Article

Structures of the peptidoglycan N-acetylglucosamine deacetylase Bc1974 and its complexes with zinc metalloenzyme inhibitors Petros Giastas, Athena Andreou, Athanasios Papakyriakou, Dimitris Koutsioulis, Stavroula Balomenou, Socrates J. Tzartos, Vassilis BOURIOTIS, and Elias Eliopoulos Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.7b00919 • Publication Date (Web): 19 Dec 2017 Downloaded from http://pubs.acs.org on December 20, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Biochemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Structures of the peptidoglycan N-acetylglucosamine deacetylase Bc1974 and its complexes with zinc metalloenzyme inhibitors Petros Giastas1,2*, Athena Andreou1, Athanasios Papakyriakou1,3, Dimitris Koutsioulis4, Stavroula Balomenou4,5, Socrates J. Tzartos2, Vassilis Bouriotis4,5, and Elias E. Eliopoulos1* 1

Department of Biotechnology, Laboratory of Genetics, Agricultural University of Athens,

IeraOdos 75, 11855 Athens, Greece 2

Department of Neurobiology, Hellenic Pasteur Institute, Vasilissis Sofias 127, 11521 Athens,

Athens, Greece 3

Institute of Biosciences and Applications, NCSR "Demokritos", 15310 Aghia Paraskevi,

Athens, Greece 4

Institute of Molecular Biology and Biotechnology, FORTH, 70013 Heraklion, Crete, Greece,

5

Department of Biology, Enzyme Biotechnology Group, University of Crete, Vasilika Vouton,

70013 Heraklion, Crete, Greece. KEYWORDS: crystal structure, N-acetylglucosamine deacetylase, substrate modeling, zinc binding ligand, hydroxamic acid, sulfonamide, Bc1974, Bacillus cereus, Bacillus anthracis

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 37

ABSTRACT

The cell wall peptidoglycan is recognized as a primary target of the innate immune system and usually its disintegration results in bacterial lysis. Bacillus cereus, a close relative of the highly virulent Bacillus anthracis, contains 10 polysaccharide deacetylases. Among these, the peptidoglycan N-acetylglucosamine deacetylase Bc1974 is the highest homolog to the Bacillus anthracis Ba1977 that is required for full virulence and is involved in resistance to the host’s lysozyme. These metalloenzymes belong to the carbohydrate esterase family 4 (CE4) and are attractive targets for the development of new anti-infective agents. Herein we report the first Xray crystal structures of the NodB domain of Bc1974, the conserved catalytic core of CE4s, in the unliganded form and in complex with four known metalloenzyme inhibitors and two amino acid hydroxamates that target the active site metal. These structures revealed the presence of two conformational states of a catalytic loop known as motif-4 (MT4), which were not observed previously for peptidoglycan deacetylases, but were recently shown in the structure of a Vibrio clolerae chitin deacetylase. By employing molecular docking of a substrate model we describe a catalytic mechanism that probably involves initial binding of the substrate in a receptive, more open state of MT4 and optimal catalytic activity in the closed state of MT4, consistent with the previous observations. The ligand-bound structures presented here, in addition to the five Bc1974 inhibitors identified, provide a valuable basis for the design of antibacterial agents that target the peptidoglycan deacetylase Ba1977.

ACS Paragon Plus Environment

2

Page 3 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Introduction There is an urgent need to explore new targets and develop novel therapeutics due to the emergence of multi-drug resistant bacteria (1). Bacillus cereus and Bacillus anthracis are Grampositive spore forming bacteria with similar genomes (2). Some strains of B. cereus are harmful to humans and cause foodborne illness, while B. anthracis spores can lead to anthrax disease in mammalian hosts (2,3). Sequencing of the B. cereus and B. anthracis genomes revealed in each a multiplicity (10 and 11 open reading frames, respectively) of putative polysaccharide deacetylases (PDAs) (4). According to the CAZy database (www.cazy.org) (5) PDAs belong to the carbohydrate esterase family 4 (CE4), which includes acetylxylanesterases, xylanases, chitin deacetylases, rhizobial NodB chitooligosaccharide deacetylases and peptidoglycan (PG) deacetylases (6,7) All CE4 esterases contain a conserved catalytic core termed the NodB homology domain (6) and catalyze the hydrolysis of either N-linked acetyl group of Nacetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) residues (chitin deacetylases, NodB, and PG deacetylases), or O-linked acetyl groups from O-acetylxylose residues (acetylxylanesterases, xylanases) (6–8). Evading the host lysozyme, that is usually secreted in high levels at infectious sites, represents one major strategy employed by the majority of pathogens. As a result, direct killing of the bacteria is prevented and simultaneously the immune system of the host is not stimulated (1). The predominant mechanism for lysozyme resistance employed by a large number of Grampositive and Gram-negative organisms is molecular camouflage: the modification of PG backbone (4). Three modifications which confer lysozyme resistance in bacteria have been described extensively and include N-glycolylation and O-acetylation of MurNAc and Ndeacetylation of GlcNAc of PG (9).

ACS Paragon Plus Environment

3

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 37

N-deacetylation of C-2 position of GlcNAc residues changes the bacteria surface to become more cationic and most likely impedes the binding of the also cationic lysozyme electrostatically (10). N-deacetylation of the PG layer was first observed several years ago in the naturally lysozyme resistant Streptococcus pneumoniae and the first gene encoding for GlcNAc deacetylase was termed PgdA(peptidoglycan deacetylase A) (11), while the structure of the corresponding enzyme was later determined (12). Mutant strains of PG GlcNAc deacetylases from various pathogens exhibit, similarly to B. anthracis ∆ba1977 mutant strain (13), reduced virulence and therefore they can be considered as validated antibiotic targets (12). PG deacetylases studied so far have been demonstrated to be metalloenzymes(12). Targeting bacterial metalloenzymes is an attractive approach for the development of new antiinfective drugs, as it has been demonstrated that metalloenzyme inhibitors can be quite selective for their targets and can be used either alone, or in association with known antibiotics to reduce drug resistant mechanisms (12,14). PG GlcNAc deacetylases exhibit favorable characteristics as antibacterial targets, since due to their localization the need for membrane permeable drugs is eliminated. Therefore, they are accessible to small molecule inhibitors, yet out of reach of multidrug resistance transporters, which protect the possible cytoplasmic targets. Bc1974 is a metal-dependent PG GlcNAc deacetylase with a rather promiscuous substrate specificity, as it is active for a wide range of substrates (4,13) and which is highly homologous to the B. anthracis Ba1977 (98.5% identity for the NodB domain). The latter has been shown to be involved in the resistance of B. anthracis to host lysozyme, it is localized at the cell membrane and its expression is required for full virulence of B. anthracis (13). Similarly to Ba1977, the LocateP database (http://www.cmbi.ru.nl/locatep-db/cgi-bin/locatepdb.py) predicted Bc1974 to be an N-terminally anchored membrane protein that lacks a cleavage site and is

ACS Paragon Plus Environment

4

Page 5 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

secreted through the Sec-(SPI) pathway. Transmembrane domain prediction servers (http://www.cbs.dtu.dk and http://www.ch.embnet.org) denote that its N-terminal transmembrane domain comprises residues 7–27. Most likely, Bc1974 is targeted to the secretory pathway by this N-terminal transmembrane domain that corresponds to an uncleavable signal peptide. We have determined the X-ray crystal structure of Bc1974 from B. cereus in its unliganded form and in complex with two amino acid-based hydroxamic acids, two histone deacetylase-8 inhibitors, a carbonic anhydrase inhibitor and a human O-GlcNAcase inhibitor, all of which bind to the catalytic zinc. To our knowledge, these are the first crystal structures of a peptidoglycan GlcNAc deacetylase complex with small-molecular weight ligands, one of which is also a Bc1974 inhibitor. Interestingly, the Bc1974 structures reveal that an active site loop (known as motif-4 or MT4) can adopt at least two distinct conformations, indicating an intrinsic flexibility of this region that was previously observed only in the substrate bound structures of the chitin deacetylase (CDA) of Vibrio cholera (15). By employing molecular docking of a model substrate, we propose a catalytic mechanism for Bc1974 and the putative role of the two distinct states in facilitating the binding of the PG substrate and in promoting the catalytic reaction. Both suggestions are in accordance with those proposed for the CDA of V. cholerae, which were based on the experimental crystallographic structures (15). From a preliminary screening of known zinc-binding ligands and metalloenzyme inhibitors, we identified five low micromolar inhibitors of Bc1974. The results presented here provide valuable structural insights that could pave the way for the development of antibacterial agents that target Bc1974 and the highly homologous Ba1977, a well characterized virulence factor of B. anthracis (13).

ACS Paragon Plus Environment

5

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 37

Materials and Methods Materials. All chemicals were purchased from Sigma-Aldrich at a minimum purity of 95%, unless indicated otherwise, and were used without any further purification. Compounds 1–6 and 15 (Table S1) were kindly provided by Prof. S. M. Ulrich (Ithaca College, New York, USA). For the screening assays, 96-well plates (COSTAR 3915) from Corning Incorporated (USA) were used. Cloning, expression and purification of Bc1974. Cloning, expression and purification of Bc1974 was performed as previously described in (13). Briefly, the gene was amplified from B. cereus ATCC 14579 genome using primers that incorporate a blunt end at the start and an XhoI site at the end of bc1974 gene. The amplified gene was digested with the suitable enzymes and ligated to pRSETa vector in order to produce a non His6-tag construct under the transcriptional control of the T7 lac promoter. 20 ml of saturated BL21 DE3 E. coli cells transformed with the pRSETa-bc1974 construct were inoculated into 1.0 L of LB medium containing 100 µg/mL ampicillin and incubated at 37oC until A600 of 0.6. Bc1974 E. coli culture was incubated at 30°C after addition of 0.2 mM CoCl2 and 0.5 mM IPTG. Cells were lysed using 25 mM Mes-NaOH (pH 6.5), 200 mM NaCl, 1.0 mM dithiothreitol and 0.3 mg/mL lysozyme for 150 min at 4°C.Bc1974 was purified in two chromatographic steps (S Sepharose Fast Flow chromatography column, previously equilibrated with 25 mM Mes-NaOH (pH 6.5) using a linear gradient of 0– 1.0 M NaCl and size exclusion Sephacryl S200 HR column, previously equilibrated with 25 mM Mes-NaOH (pH 6.5) and 200 mM NaCl. It should be noted that while the whole Bc1974 protein comprises 273 amino acid residues, the expressed construct did not include the first 25 residues, which belong to the predicted N-terminus α-helix. Moreover, the crystal structures revealed the

ACS Paragon Plus Environment

6

Page 7 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

structure of the region 68-273, meaning that we could not identify the electron density for the 42 N-terminus residues of the expressed construct. Crystallization, data collection and processing.Bc1974 crystallization screening was set up using an Oryx4 crystallization robot (Douglas Instruments Ltd.) with the vapor diffusion method in sitting drops. Protein crystals appeared only in one out of approximately 1400 conditions tested, which was subsequently optimized to produce crystals suitable for diffraction experiments. The best crystals were grown after mixing equal volumes of protein solution (concentrated at 22 mg/mL) and mother liquor consisting of 100 mM Na citrate pH 5.6, 30% PEG 4000, 10% ethanol. The crystals appeared within 7–10 days reaching their final size in 2 weeks at 16 °C. Crystals of the Bc1974–ligand complexes were grown under the same crystallization conditions using the following protocol in all cases. Ligands were all dissolved in DMSO to produce 100 mM stock solutions. Prior setting the crystallization drops, protein and ligand solutions were mixed to produce a ten-fold ligand molar excess solution and the mixture was incubated for 1–2 h at 4 °C. Crystals grew when equal volumes of protein/ligand solution and mother liquor were mixed decreasing simultaneously the final concentration of DMSO to 5% in the crystallization drops. Crystals were soaked in a cryoprotectant solution containing the mother liquor supplemented with 20% v/v ethylene glycol for 5–10 s prior their vitrification with liquid nitrogen. Diffraction measurements were carried out on beamline X06DA at the Swiss Light Source (Villigen, Switzerland), and single wavelength datasets were collected to resolutions of 1.45–3.05 Å (Table 1). Images were indexed, integrated, and scaled using XDS (16) in the space group that was determined with POINTLESS (17) using unmerged data. The high resolution limit was determined based on the CC1/2 criterion (>50%) and on the refinement statistics. First, the unliganded Bc1974 structure was solved with molecular replacement in

ACS Paragon Plus Environment

7

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 37

PHASER (18) using as a model the structure of the CE4 domain of SpPgdA (PDB ID: 2C1G) (12) and was subsequently refined in PHENIX (19). The asymmetric unit contained four molecules of Bc1974, which superimpose almost perfectly with an average rms deviation of 0.284 Å for the main chain atoms. Regular inspection of the electron density maps 2Fo−Fc and Fo−Fc and refitting of the model where necessary, were performed with COOT (20). The refinement statistics for the converged final model are given in Table S1. Density for a single heavy atom within the active site of the protein was observed after the initial inspection of the difference map, which was determined to be zinc by the fluorescent scan at zinc K-edge (Figure S1). Ligands were included when were unambiguously defined by the initial Fo-Fc and 2Fo-Fc maps and their occupancy was unrestrained refined with phenix-refine, while the binding motif of 6–10 (Scheme 1) was restrained to optimum distances. Omit map calculations were carried out at the final stages of refinement. Ligand topologies and coordinates were generated with either PRODRG (21) or Jligand (22) and were optimized with eLbow (23). Structure figures in the Supporting Information were generated using PyMOL Molecular Graphics System, Schrödinger LLC. Computational methods. Coordinates of the protein and zinc atoms only were taken from chains A and B of the ligand-free Bc1974 crystal structure (PDB ID: 5N1J) and were used without any further modification. To simulate the octahedral geometry of the substrate-bound catalytic zinc, one of the oxygen atoms (OXT) of the co-crystallized acetate was replaced by a water molecule. Polar hydrogen atoms were added and Gasteiger charges were applied using AutoDock Tools 1.5.6 (24). The initial conformations of (GlcNAc)3 pseudosubstrate and ligands were generated from SMILES representations using the program Omega 2.4 (25) and then Gasteiger charges were applied. The search space was defined by a grid box centered at zinc and

ACS Paragon Plus Environment

8

Page 9 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

comprised 71×61×61 grid points of 0.375 Å spacing. For each complex, 100 docking rounds were calculated with AutoDock 4.2 using the Lamarckian genetic algorithm with the default parameters of AutoDock 3 (26). The maximum number of energy evaluations was set to 10×106 and the docked conformations were clustered using a tolerance of 2.0 Å. Visual examination of the protein complexes and rendering of the figures was performed using VMD 1.9 (27). Calculations were carried out on Intel Xeon workstations running Linux 2.6.32 kernels. Inhibitor screening. Screening of the 25 compounds was performed using two different enzymatic assays. Initially, all compounds were tested using a fluorimetric assay that is based on fluorogenic labeling with fluorescamine, and which labels the free amines generated by the enzymatic deacetylation of (GlcNAc)5. This allowed miniaturization of the assay to 50 µL volumes, suitable for a 96-well format. In addition we employed a radiometric assay that is based on the estimation of the deacetylation reaction using O-hydroxy ethylated chitin (glycol chitin) as substrate, which is 3H labeled in N-acetyl groups (28). Both inhibition screenings were performed in 0.2 mM CoCl2, 25 mM MES (pH 6.8) at 25 oC with 15 min incubation time (fluorimetric assay) or 60 min (radiometric assay), using 20 µΜ of (GlcNAc)5 (fluorimetric assay), or 5 µL (1 mg/mL) radiolabeled glycol chitin (radiometric assay) as substrate, both at 0.2 µM enzyme concentration (Bc1974 at 0.5 mg/mL). Compounds were dissolved in DMSO, unless indicated otherwise. The final concentrations of the ligands were 31–1000 µΜ and the total volume was 50 µL with a final concentration of 10% DMSO. Reactions were monitored using a BMG LabtechFluoStar Optima microtiter plate reader, at excitation and emission wavelengths of 355 and 460 nm, respectively (fluorimetric assay) or measured by scintillation counter (radiometric assay). Experiments were run in quadruplicate and standard deviation did not exceed 5%.

ACS Paragon Plus Environment

9

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 37

All calorimetric titrations were performed on a VP-ITC microcalorimeter (Microcal). Protein samples were extensively dialyzed against the ITC buffer containing 25 mM MES (pH 6.8), 200 mM NaCl. We prepared the ligand solution by dissolving peptide in the flow through of the last buffer exchange. All studied ligands were first dissolved in 100% DMSO and then transferred stepwise to the ITC buffer at a final concentration of 10% DMSO. We also added DMSO at the same concentration to the protein solution to match the buffer composition of the ligand. In the titration experiments, the ligand was injected into a solution of the enzyme. Ligands’ concentrations were 200µM and the enzyme was at 20µM. Titrations were carried out with a stirring speed of 300 rpm and 240 s intervals between 15 µL injections. The first injection for each sample was excluded from data fitting. The experimental data were fitted to a theoretical titration curve using the Origin software package (version 7.0, Microcal) to afford values of K (binding constant in M-1), and ∆H (heat change in cal mole-1). A fourth parameter, ∆S (entropy change in cal mole-1 deg-1), is calculated from ∆H and K and displayed after fitting. The affinity of the ligands for the protein is given as the dissociation constant (Kd= K-1). ITC experiments were run in duplicate, analyzed independently, and the thermodynamic values were averaged.

Results and Discussion Crystal structure of Bc1974. The Bc1974 structure reveals a single domain (residues 68– 273), which adopts a fold reminiscent of an (α/β)8 topology, characteristic of the NodB homology domain present in all CE4 esterases (Figure 1A). Although the expressed construct contained 42 additional residues at the N-terminus, no electron density was observed for those residues, denoting the unstructured nature of the specific region (details provided in Materials and Methods). The presence of a metal ion in the putative binding site at the bottom of a well-

ACS Paragon Plus Environment

1

Page 11 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

formed groove coordinated by the highly conserved Asp-His-His triad was revealed by the electron density map (20-σ in difference map, average B-factor 25 Å2) and was confirmed to be zinc by x-ray fluorescent scan (Supporting Figure S1). Sequence alignment of Bc1974 with homologous enzymes suggests that most of the catalytic and zinc-binding residues are conserved and distributed over five catalytic motifs of the enzymatically and structurally characterized CE4 esterases (MT1–5), including chitin deacetylase ClCDA, PG deacetylases SpPgdA, BsPdaA, Bc1960, putative PDA Bc0361, and acetylxylan esterase SlCE4 (Supporting Figure S2). It is noteworthy that the NodB homology domain of the B. anthracis homologue Ba1977 differs only in two surface exposed residues and Arg199of the Bc1974 MT4, which is replaced by Lys199 (Figure 1A). In the structures presented here (Table 1), wherever zinc was not coordinated by a ligand molecule, an acetate ion was clearly determined in the electron density maps (Figure 1B). While acetate is one of the deacetylation products, its presence in the unliganded Bc1974 structures is likely due to a contaminant of the PEG crystallization media (12). The octahedral coordination sphere of the catalytic zinc is filled by a water molecule. The acetate ion was well ordered (present up to 8-σ in the Fo-Fc map), with an average B-factor of 29 Å2 that is close to the Bfactor of the metal ion. The acetate interacts with residues of the active site in a similar fashion in all unliganded Bc1974 structures (Figure 1B). One of its oxygen atoms interacts with the invariant Asp76 that is tethered with Arg164, a highly conserved residue except for the fibronectin domain-containing deacetylases (12,29). The same O interacts also with the structurally conserved His230, which in turn interacts directly with the adjacent Asp197of MT4. The methyl group of acetate occupies a small hydrophobic patch formed by the indole group of Trp191 and the side chain of Leu228. In all the structures resolved herein, there was no electron density

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 37

around the Cα of the conserved Pro166 to indicate the existence of a hydroxyl group (Supporting Figure S3), a feature observed previously in other characterized or putative peptidoglycan deacetylases, such as Bd3279 (30), Ba0330 (29) and Bc0361 (31). Finally, despite a significant differentiation observed at the N- and C-terminal domains of Bc1974 with respect to the corresponding domains of the homologous Bc1960 and SpPgDA (Supporting Figure S4 and remarks therein), the active site of Bc1974 is characteristic of PDAs.

Figure 1. (A) Overall fold of the NodB domain of Bc1974 with labeled secondary structured elements and the catalytic zinc shown as a green sphere. The three residues indicated are those that differ between Bc1974 and Ba1977 (the residue of the latter in parenthesis). (B) Close-up view of the Bc1974 active site. The metal-binding residues are colored with green C atoms and the interacting catalytic residues with cyan C; all other atoms are colored with blue for N and red for O. The zinc-bound acetate (ACT) and water (WAT) molecules are shown with spheres, whereas dashed lines indicate hydrogen bonding interactions. (C) Crystallographic dimer of acetate-bound Bc1974 molecules illustrating the stacking interaction between Trp198 of molecule A (orange carbons) and His269 of molecule B (cyan carbons).

Table 1. Data collection statistics and refinement

ACS Paragon Plus Environment

1

Page 13 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Data collection

Bc1974

Bc1974 – 1

Bc1974 – 6

Bc1974 – 7

Bc1974 – 8

Bc1974 – 9

Bc1974 – 10

Space group

P21

C2

P21

P21

P21

P21

P21

a, b, c (Å)

50.28, 117.67, 99.77

196.10, 44.35, 99.13

49.36 118.01 98.62

49.77 117.83 99.06

50.65 117.44 99.49

49.39 118.21 98.27

49.44 117.97 98.20

α, β, γ (ο)

90.00, 102.75, 90.00

90.00, 98.83, 90.00

90.00 102.28 90.00

90.00 102.38 90.00

90.00 102.92 90.00

90.00 102.09 90.00

90.00 102.10 90.00

1.000

1.000

1.000

1.000

1.000

1.000

Cell dimensions

Wavelength (Å) 1.000 Resolution (Å)

48.65 – 1.80 (1.87 – 1.80)

48.45 – 1.45 (1.5 – 1.45)

47.35 – 2.80 (2.92 - 2.80)

48.61 – 2.44 (2.53 – 2.44)

48.48 – 2.45 (2.53 – 2.45)

48.29 – 3.05 (3.17 – 3.05)

48.34 – 2.73 (2.83 – 2.73)

Protein molecules / ASU

4

4

4

4

4

4

4

Unique reflections

104462 (10419)

149678 (14328)

27111 (2632)

41046 (3720)

41648 (4057)

20589 (1717)

28074 (2256)

Rmeas

0.082 (0.890)

0.069 (0.910)

0.093 (0.923)

0.097 (0.735)

0.117 (0.750)

0.074 (0.603)

0.103 (0.963)

/

13.1 (2.1)

15.2 (1.4)

6.6 (1.9)

9.0 (1.9)

7.2 (1.9)

11.1 (1.8)

9.6 (1.9)

Completeness (%)

99.6 (99.7)

99.5 (96.0)

99.5 (95.7)

98.8 (90.8)

99.5 (97.3)

98.3 (83.5)

96.3 (85.4)

Redundancy

6.7 (6.5)

5.1 (4.9)

3.9 (3.8)

4.5 (4.3)

4.5 (4.3)

3.5 (3.3)

2.9 (2.8)

Reflections, work / test set

99165 / 5219

78443 / 3721

25796 / 1315

39170 / 1870

39625 / 2022

19594 / 996

26792 / 1280

Rwork / Rfree

0.1733 / 0.2023

0.1613 / 0.1798

0.2157 / 0.2788

0.2220 / 0.2860

0.2256 / 0.2885

0.1792 / 0.2549

0.2051 / 0.2663

6617 / 52 / 701

6628 / 76 / 622

6587 / 58 / 88

6591 / 65 /91

6611 / 38 / 99

6542 / 48 / 19

6562 / 71 / 59

22.30

49.21

45.71

41.84

60.66

54.91

Refinement

No. of atoms: Protein / ligands / water

Wilson B-factor 24.84 (Å2)

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 37

Average B factor (Å2) Protein / ligands / water

31.20 / 37.40 / 37.00

30.20 / 46.30 / 41.40

57.11 / 72.00 / 37.60

61.70 / 65.90 / 51.30

56.10 / 73.30 / 47.00

73.50 / 83.40 / 49.60

69.60 / 82.00 / 51.40

Bond lengths rmsds (Å)

0.007

0.006

0.012

0.009

0.008

0.010

0.009

Bond angles rmsds (o)

1.024

1.100

1.491

1.220

1.080

1.360

1.190

Most favored

97%

97%

97%

96%

95%

96%

95%

Outliers

0.61%

0.49%

0.25%

0%

0.61%

0%

0.12%

PDB ID

5N1J

5N1P

5NC6

5NC9

5NCD

5NEK

5NEL

Bc1974 was crystallized in the space group P21 with four molecules in the asymmetric unit. The major deviations between the molecules of the asymmetric unit were observed in the region 197–205, which comprises the residues of motif 4 (MT4). While the protein molecules of molecules A and C adopt an almost identical conformation in the specific region of MT4, they differ substantially from the corresponding regions of molecules B and D. Superposition of the chains A and B (or C and D) results in a root-mean-square (rms) deviation of 0.28 Å for main chain non-MT4 atoms, whereas the Cα atoms of the residues Trp198 and Arg199 deviate by 1.0 Å and 1.2 Å between the two chains, respectively. Downstream, the loops diverge further, reaching a peak of 4.2 Å for the Cα atoms of Lys202, which however do not point towards the catalytic cavity. This is not the first time that a catalytic loop of a CE4 enzyme is observed in two distinct conformations. Recently, Andres et al reported the crystal structures of VcCDA bound with substrate molecules, which showed large conformational movements of loop 4 (equivalent to MT4 herein) compared to the substrate-free enzyme (15). The equivalent structural differentiation between chains A, C and B, D in the present study may be the effect of the packing contacts between Bc1974 molecules. Specifically, His269 of chains B and D penetrate to

ACS Paragon Plus Environment

1

Page 15 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

the catalytic groove of chains C and A respectively (Figure 1C) and is stabilized there via π–π stacking interactions with the indole group of Trp198. As a result the whole MT4 region recedes from the binding site, which induces a significant increase of the catalytic cavity volume by ~7% (Supporting Figure S5). Given that the two distinct conformations of MT4 in the present structures were not mediated by ligand binding, these are suggestive of an intrinsic flexibility of MT4, which could in turn play some role in the enzymatic activity of Bc1974. Substrate interactions. With the aim to propose a putative catalytic mechanism for Bc1974, we calculated first the bound conformations of the substrate model (GlcNAc)3 into the catalytic groove. Both molecules A and B were employed in the docking calculations, in order to account for the different conformations of the MT4 catalytic loop and AutoDock 4 was used for flexible docking of (GlcNAc)3. In representative conformations of the substrate model bound to the two Bc1974 molecules A and B (Figure 2A,B), the N-acetyl group of the central glycan unit (labelled as 0) is bound to the catalytic zinc via its carbonyl group. The flanking sugars (labeled as –1 and +1) occupy a groove running over the face of the protein and are oriented so that the terminal O3 atoms point out of the substrate-binding groove, giving access to the large peptidoglycan. The predicted conformations of the (0) and (–1) moieties superimpose quite well in both molecules. The coordinated carbonyl group of (0) is stabilized by a hydrogen bond with the backbone amide of Tyr167, whereas the amide of its N-acetyl group is proximal to the imidazolium ring of His230 (Figure 2C). Additionally, GlcNAc (0) displayed a number of C–H aromatic interactions with the indole group of Trp198. The only difference in the interactions of GlcNAc (–1) with the two molecules was exhibited by the acetate group; in molecule A, the carbonyl group is hydrogen-bonded to the side chain amide of Asn201 (Figure 2A), whereas in molecule B with the imidazole Nε2 of His233 (Figure 2B). This difference is an effect of the more

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 37

compact conformation of MT4 observed in molecule B, in which Trp198 and Arg199 cap off the (+1) subsite whereas Asn201 shifts away from the substrate binding groove. Therefore, the bound conformations of GlcNAc (+1) deviate significantly between the Bc1974 molecules A and B, adopting nearly vertical orientations (Figure 2A,B). In molecule A, the only polar interaction is predicted between the acetate group and Asp194 (not shown), but in B, the (+1) sugar is stabilized further by two hydrogen bonds between O6 and the indole-NH of Trp198, and between the NAc carbonyl and the guanidinium group of Arg199. These possible configurations of the substrate model (GlcNAc)3 prompt us to hypothesize that molecules A and C of the Bc1974 crystal structure are representative receptive states that facilitate initial binding events of the large PG substrate and allow the proper organization of the central glycan unit within the active site. The more compact conformation of MT4 towards the (+1) subsite, as displayed in molecules B and D, would then stabilize the substrate further so as to promote the enzymatic activity. To evaluate our molecular docking models and our hypothesis about the possible functional roles of the distinct MT4 conformations, we performed structural superposition of the (GlcNAc)3–Bc1974 monomers with the triacetylchitotriose-bound VcCDA (DP3–VcCDA) (15). We observed remarkable similarities at the active sites of the two enzymes, particularly when considering the zinc-bound residues, albeit the significant differences in the catalytic loops with regard to their trajectories (Figure S6A and B) and length (Figure S7). By comparing the closed MT4 conformation of Bc1974 with the DP3–VcCDA structure, we found that the indole groups of the GlcNAc(0)-capping tryptophan residues (W198 in Bc1974 and W238 in VcCDA) coincide spatially and contribute to the stabilization of the sugar occupying the (+1) subsite via a H-bonding interaction (Figure S6B). As a result, the predicted conformation of (GlcNAc)3 in the closed Bc1974 superimposes almost perfectly with the DP3

ACS Paragon Plus Environment

1

Page 17 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

substrate of VcCDA at subsites (–1) and (0), whereas at subsite (+1) the sugars deviate slightly by 1.2–2.1 Å (Figure S6B). In contrast, the open MT4 conformation of Bc1974 where W198 remains distal from the substrate allows for higher flexibility at the (+1) subsite, as suggested by the perpendicular orientation of our docked model with respect to the VcCDA-bound substrate (Figure S6A). At this point it should be noted that the active site organization of Bc1974 (Figures 1B, 2C) is marginally affected by the different configurations of the MT4 residues, therefore, both distinct conformations should be representative of an active Bc1974 NodB domain. Taken together, the two crystallographically observed MT4 states indicate an intrinsic flexibility of this loop with plausible biological relevance, supported by the similarities of Bc1974 with the substrates-bound structures of VcCDA, and therefore should be taken into consideration in inhibitor design efforts.

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 37

Figure 2. (A, B) Molecular models of Bc1974–(GlcNAc)3 complex illustrating the more open substrate binding groove in molecule A with respect to B. GlcNAc is colored with light orange C and Bc1974 residues are shown with cyan C; all other colors are as in Figure 1. (C) Close-up view in the active site of Bc1974–(GlcNAc)3 model. The interactions between the middle sugar (0) of the substrate, a zinc-bound water molecule and the active site residues are indicated with dashed lines. (D) Proposed catalytic mechanism of Bc1974, initiated by the nucleophilic attack of a water (or HO–) to the C=O carbon atom of N-acetyl group.

Proposed catalytic mechanism. The widely accepted catalytic mechanism of zinc-depended N-deacetylases is based on the nucleophilic attack of a water molecule to the N-acetyl group of the central (0) glycan moiety. Within the active site of the (GlcNAc)3-bound Bc1974 model, GlcNAc (0) is bound to the catalytic Zn(II) in an octahedral arrangement that comprises Asp77, His126, His130, a water molecule, and the NAc oxygen and O3 of the substrate (Figure 2C). The NAc carbonyl and O3 hydroxyl groups are stabilized via hydrogen bonds with the backbone amide of Tyr167 and the carboxylate moiety of Asp77, respectively. The zinc-bound water that was modeled in the position occupied by the co-crystalized acetate molecule is within hydrogen bonding distance from Asp76 and His230. The latter are engaged in electrostatic interactions with Arg164 and a hydrogen bond with Asp197, respectively. This organization of the active site allows the proposal of a catalytic mechanism for Bc1974 (Figure 2C,D), which is in accordance with the mechanism proposed for Streptococcus pneumoniae peptidoglycan GlcNAc deacetylase (SpPgdA) (12) and for VcCDA (15). The polarized solvent molecule is activated by a conserved acidic residue (Asp76) that is properly oriented close to the catalytic zinc, and which acts as a general base that stabilizes the oxyanion intermediate and its flanking transition states (32). Asp76, which upon mutation to Asn results in

ACS Paragon Plus Environment

1

Page 19 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

a complete loss of Bc1974 enzymatic activity (33), is stabilized through a salt bridge interaction with the guanidinium group of Arg164 that is also hydrogen bonded to the side chain of Thr74. By analogy to the catalytic mechanism of thermolysin(34), the carboxylate group of Asp197 acts as a hydrogen bond acceptor from the positively charged His230, an interaction that ensures the proper orientation of its imidazolium side chain so as to act as a general acid and, most probably, serves as proton donor to the leaving amino group (Figure 2D). Previous work on the SpPgdA demonstrated that the analogous interaction between His417 (His230 in Bc1974) and Asp391 (Asp197 in Bc1974) is tuning the histidine’s pKa and is essential for the catalytic mechanism (12).

Scheme 1.Structures of the hydroxamic acid-containing inhibitors 1–5 and the co-crystallized Bc1974 ligands 1, 6–10 (shown in blue). The zinc-binding groups are indicated in bold and values in parentheses are the dissociation constants (Kd) of the Bc1974 inhibitors as determined by isothermal titration calorimetry.

Structures of ligand-bound Bc1974.To set a basis for structure-based discovery of Bc1974 and Ba1977 inhibitors we carried out a preliminary screening of 25 small-molecules, most of

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 37

which carry a hydroxamic acid moiety as zinc-binding group (Supporting Table S1). The selected compounds span a wide range of size (MW=120–558), hydrophobicity (calculated octanol/water partition coefficient, cLogP: –4.4 to 6.1), and consist of a diverse set of aryl- and amino acid-substituted hydroxamic acids (compounds 1–8 in Scheme 1 and 11–15 in Table S1), some known matrix metalloproteinase (MMP) inhibitors (18, 19, 21–23 in Table S1), antibiotic and anti-bacterial agents (20, 24 in Table S1). In particular, the hydroxamic acids 1–6 (Scheme 1), 15 and Vorinostat (16, SAHA in Table S1) are potent histone deacetylase (HDAC) inhibitors (35, 36), acetazolamide (9) is a sulfonamide-based carbonic anhydrase inhibitor, Thiamet-G (10) is a very potent human O-GlcNAcase inhibitor (37), and PUGNAc (25) is an inhibitor of β-Nacetylglucosaminidases (38). A fluorimetric assay based on labeling with fluorescamine was employed (12) (Materials and Methods) and revealed that the HDAC8 inhibitors 1–5 are also potent inhibitors of Bc1974, displaying low (