JBC Papers in Press. Published on March 28, 2018 as Manuscript RA

Transcription

1 JBC Papers in Press. Published on March 28, 2018 as Manuscript RA The latest version is at Active site remodeling during the catalytic cycle in metal-dependent fructose-1,6- bisphosphate aldolases Benoit Jacques 1, Mathieu Coinçon 1 and Jurgen Sygusch* From the Department of Biochemistry and Molecular Medicine, Université de Montréal, Montréal, Québec, Canada H3C 3J7 1 Both authors contributed equally to this work. * to whom correspondence should be addressed: Biochimie et Médecine moléculaire, Université de Montréal, CP 6128, Station Centre Ville, Montréal, QC, H3C 3J7, Canada, Tel: , Fax: , Jurgen.Sygusch@umontreal.ca Running title: Reaction intermediates in Helicobacter pylori FBP aldolase Keywords: Enzyme mechanism, Crystal structure, Helicobacter pylori, Metalloenzyme, Fructose-1,6- bisphosphate aldolase, Zinc, Active site remodeling, Conformational change ABSTRACT: Crystal structures of two bacterial metal (Zn) dependent D-fructose 1,6-bisphosphate (FBP) aldolases in complex with substrate, analogues, and triose-p reaction products were determined to Å resolution. The ligand complexes cryotrapped in native or mutant H. pylori aldolase crystals enabled a novel mechanistic description of FBP C 3-C 4 bond cleavage. The reaction mechanism uses active site remodelling during the catalytic cycle implicating relocation of the Zn cofactor that is mediated by conformational changes of active site loops. Substrate binding initiates conformational changes, triggered upon P 1-phosphate binding, which liberates the Zn chelating His180, allowing it to act as a general base for the proton abstraction at the FBP C 4-hydroxyl group. A second zinc chelating His83 hydrogen bonds the substrate C 4- hydroxyl group and assists cleavage by stabilizing the developing negative charge during proton abstraction. Cleavage is concerted with relocation of the metal cofactor from an interior to a surface exposed site, thereby stabilizing the nascent enediolate form. Conserved residue Glu142 is essential for protonation of the enediolate form, prior to product release. A D- tagatose 1,6-bisphosphate enzymatic complex reveals how His180 mediated proton abstraction controls stereospecificity of the cleavage reaction. Recognition and discrimination of the reaction products, dihydroxyacetone-p and D-glyceraldehyde-3-P, occurs via charged hydrogen bonds between hydroxyl groups of the triose-ps and conserved residues, Asp82 and Asp255, respectively, and are crucial aspects of the enzyme s role in gluconeogenesis. Conformational changes in mobile loops β5-α7 and β6-α8 (containing catalytic residues Glu142 and His180, respectively) drive active site remodelling enabling the relocation of the metal cofactor. INTRODUCTION D-Fructose 1,6-bisphosphate aldolases (Fba; E.C ) are ubiquitous enzymes that catalyze the reversible transformation of D-fructose 1,6- bisphosphate (FBP) into dihydroxyacetone phosphate (DHAP) and D-glyceraldehyde 3-phosphate (G3P), thus playing a major role during glycolysis, gluconeogenesis and Calvin cycle. A common feature of class II aldolases is that they use a transition metal ion to facilitate aldol/retroaldol reactions (1). These metalloenzymes are evolutionarily and mechanistically unrelated to class I aldolases which employ an active site lysine in Schiff base formation (2). Metalloaldolases are found predominantly in microorganisms including human pathogens such as, but not limited to, Mycobacterium tuberculosis, Pseudomona aeruginosa, Helicobacter pylori, Yersinia pestis, Clostridium difficile, Candida albicans and Giardia lamblia, as well as in plant pathogens, e.g. Magnaporthe grisea. The enzyme is essential for survival of E. coli and Streptomyces species (3 6) as well as protozoa such as Giardi lamblia (7) making it a promising target for the development of novel antimicrobial drugs. H. pylori is a human pathogen that colonizes the gastric mucosa, resulting in an acute inflammatory response and damage to epithelial cells and progressing to a number of disease states, including gastritis, peptic ulceration, and gastric cancer (8 13). In H. pylori, enzymes coding for either glycolytic pathway or pentose phosphate shunt are incomplete (14, 15), the Entner-Doudoroff pathway being the only major route of sugar catabolism (16). Glucose catabolism, resulting in intracellular acid production, would be a potential burden on ph maintenance in H. pylori, which has to survive an external ph of 2 3 (17). By contrast, gluconeogenesis, which converts lactate and pyruvate into sugars required for nucleic acid and peptidoglycan

2 Schema 1. Catalytic mechanism and reaction intermediates. biosynthesis, removes H + from the cytoplasm, and is active in H. pylori (18) as enzymes of this pathway are fully present (14). The primary function of H. pylori aldolase is thus aldol reaction during gluconeogenesis. As H. pylori produces an organ-specific infection which is not normally complicated by coinfection with other pathogens, it may be well suited to treatment with a narrow-spectrum therapeutic agent which would not promote resistance in endogenous flora (19). Hence, it is of considerable interest to develop novel therapies which are more effective and specific for H. pylori. The metalloenzyme may thus serve as a promising target for prophylaxis of H. pylori infections. The main steps of the reversible aldol reaction catalyzed by class II aldolases are depicted in Schema 1. Class II enzymes catalyze aldol reaction by use of Zn 2+ or Co 2+ metal ion that is thought to polarize the DHAP carbonyl, 1, thereby facilitating stereospecific proton exchange at the α-carbon, which produces the enediolate intermediate. The oxyanion form undergoes electronic redistribution, 2, followed by nucleophilic addition of si face carbanion form unto the si face of D- G3P, 3, ultimately producing D-FBP, 4. Step 3 4 could also occur via an alkoxide intermediate instead of a concerted electronic rearrangement and proton exchange (20). Site-directed mutagenesis of conserved active residues identified one residue, Asp109 in E. coli aldolase (21), and the equivalent Asp83 in G. lamblia (7) (Asp82 in H. pylori), whose mutation reduced catalysis by several orders of magnitude. This conserved Asp was assumed to have a role in the proton transfer at step 3 4. Support from structural studies, however, were inconclusive (22, 23). Evidence for an essential catalytic role by Zn 2+ metal ion in the aldolase reaction mechanism comes from 13 C NMR study of E. coli FBP aldolase showing Zn ion coordination promoted formation of the enediolate-2 intermediate from FBP (24). The intermediate formation was 50% of maximum at ph 8.0, with maximum at ph 9.0, and represents an extraordinarily large shift in pka of the DHAP C3 α- carbon. Model compounds indeed corroborated that Zn 2+ stabilizes the enolate by shifting the pka of the C3 α-carbon by 10 units to a pka ~ 8.4 (25). The Zn 2+ metal ion in class II aldolases coordinates the imidazole groups of three histidine residues and undergoes a conformational transition upon ligand binding in the active site (26). The rotational isomerization of histidine side chains allows the Zn ion to migrate ~ 4 Å from a buried binding site to a surface exposed site bringing it within coordination distance of the C2 carbonyl and the C3 hydroxyl group of C1 phosphorylated ligands (7, 22, 26). The migration by the Zn ion is concomitant with conformational changes in two to three loop regions that decrease their mobility and serve to stabilize attachment at the P1 phosphate binding locus (7, 22, 23, 26, 27). The loop β6-α8 has a Zn cofactor chelating His residue whose displacement enables it to retain its coordination with the divalent metal cation when relocated to a surface exposed site, this residue has been identified as H226 (E. coli), H178 (G. lamblia), H180 (H. pylori) and H212 (M. tuberculosis) (7, 22, 26, 27). Biochemical characterization of an alanine mutant of the conserved residue Glu182, situated on the highly mobile loop β5-α7 in E. coli aldolase, suggested a role by the residue in stereospecific proton exchange at the DHAP C3 α-carbon (28), reaction step 1 2 as depicted in Scheme 1. However, such a role by Glu182 could not be immediately reconciled with the structural data of an enzymatic complex formed with E. coli aldolase by the potent DHAP analogue, phosphoglycolohydroxamate (PGH) (29, 30). In a later study with M. tuberculosis aldolase using the same inhibitor PGH, the equivalent Glu169 was found to interact with the exchangeable proton via hydrogen bonding to an intervening water molecule (31). In most class II aldolase structures including liganded ones, loop β5-α7 cannot be traced presumably due to its positional disorder and as such complicates the interpretation of the role of this conserved catalytic residue (Glu142 in H. pylori aldolase). In this study, we focus on the determinants of substrate recognition of FBP and triose-ps and on the elementary rate step in retroaldol/aldol reaction, namely C-C bond cleavage / formation catalyzed by class II 2

3 aldolases. To elucidate the molecular details of how the enzyme performs proton abstraction of the FBP C4 hydroxyl, we conducted crystallographic studies using flash cooled crystals of native and mutant H. pylori aldolases previously soaked with saturating concentrations of ligands. The structures revealed a conformational driven exchange of the Zn 2+ cofactor between mutually exclusive states in the presence of its obligate triose-p, DHAP. The Zn metal cofactor is either coordinated with the triose-p or was sequestered in the active site interior, making no interaction with the DHAP molecule. Zn independent ligand recognition was corroborated in E. coli aldolase crystals soaked with Zn depleted E. coli aldolase with hexitol 1,6- bisphosphates (HBP), alditol analogues of FBP. From structural and site mutagenesis studies, His180 adopts two conformational states: one in which it chelates the Zn cation sequestered in the active site interior, while in the second, it is no longer coordinated to the metal cofactor and acts as the general base responsible for proton abstraction at the C4 hydroxyl group, thereby initiating C-C bond cleavage. The cleavage mechanism explains the inhibitory role of the (4S)-D-tagatose 1,6- bisphosphate stereoisomer (TBP). We further corroborate the catalytic role attributed to conserved residue Glu142 (H. pylori) of loop β5-α7 in enediolate protonation following cleavage. By use of isosteric mutation D82N, we exclude a catalytic role for the conserved Asp82, its role being limited to substrate binding and maintaining active site integrity. RESULTS H. pylori aldolase native structure. H. pylori FBP aldolase polypeptide exhibited a (βα) 8-barrel (or TIM barrel) fold, as described previously (26). The secondary structure elements were assigned using DSSP (32, 33) and consisted of twelve α-helices and eight β-strands. The native enzyme crystallizes as a homodimer in the asymmetric unit cell, whose subunits have identical tertiary structure (RMSD 0.19Å). The dimer exhibits two-fold non-crystallographic symmetry with the two-fold rotation axis situated at the interface between subunits. Size exclusion chromatography corroborated the dimeric quaternary structure of H. pylori aldolase, which elutes at relative molecular weight of 70 kda. The catalytic zinc ion is found in a shallow pocket at the center of β-barrel near its C- terminal side, hereby referred to as site I or buried site, and is coordinated by residues His83 (Nε), His180 (Nε), His210 (Nδ), Glu134 and a water molecule. This coordination is reinforced by the interaction of His83 (Nδ) with Asp 82 and His210 (Nε) with Glu132. Additionally, His210 and Glu134 interact with a monovalent cation (Na + ) as do Gln47, Met102, Lys251 and Asn253. Weak electron density was associated with parts of loops β6-α8 and α9-α10 (residues and respectively), indicating conformational flexibility. No electron density was discernable for most of the loop β5-α7 (residues ). DHAP-bound structures. The P1-phosphate binding locus in H. pylori aldolase bound with DHAP, portrayed in Figure 1A, comprises backbone interactions with the amides of Gly181, Ser213, Asp255, Thr256 and side chain interactions with Lys184, Ser213 and Thr256 in which each phosphate oxygen participates in at least two hydrogen bonds. The DHAP molecule is further positioned by hydrogen bonds between Gly211 and its C2 carbonyl oxygen and a bidentate hydrogen bond between its C3 hydroxyl group and Asp82 carboxylate. A spine of three water molecules borders the solvent exposed side of DHAP where they interact with one another through hydrogen bonding. This spine of water molecules is anchored by hydrogen bonds to the phosphate oxyanion, and Asp82, while the central water molecule interacts with Asp255. The notable difference when comparing ligand binding in the DHAP-complexed structures of H. pylori and E. coli aldolase is the interaction of the P1- phosphate with Lys184 in the loop β6-α8 of H. pylori aldolase while in E. coli aldolase the interaction is made with a monovalent Na ion that interacts with the backbone carbonyls of Val225 and Gly227 of the equivalent loop β6-α8. This interaction is crucial as it triggers the closure of the loop β6-α8 and/or stabilizes it upon phosphate oxyanion binding (Figure 1B). This variation in the ligand binding is not unheard of: P1- phosphate binding locus in G. lamblia aldolase implicates Lys182 (23), whereas an Na + ion is involved in the structures of T. aquaticus aldolase (34) and M. tuberculosis aldolase (22). A continuous anisotropic electron density in DHAP-bound H. pylori aldolase was interpreted as the Zn 2+ metal ion binding site, and is represented by an ellipsoid in Figure 1A. The elongated electron density corresponding to Zn 2+ binding site was corroborated by a difference electron density map calculated based on the anomalous Zn signal and whose electron density showed the same elongated form and justifying anisotropic refinement of the Zn ion position at the exposed site II. Residual continuous density not coincident with, and extending beyond the Zn binding site, was interpreted as a tightly bound water molecule, W2, based on the associated B-factor at full occupancy. The associated water molecule is well positioned to coordinate with the Zn metal ion at site II. No evidence for Zn binding at the site I was found in the electron density difference maps. The Zn ion chelating histidine residues His83 and His210 could be satisfactorily positioned with respect to the Zn ion independently of whether the refinement of the Zn ion position at site II was anisotropic or in terms of two overlapping yet mutually exclusive isotropic positions. The latter interpretation of the active site Zn ion explains the 3

4 residual electron density observed at the extremities of the ellipsoid (Figure 1A). Different from the other chelating Zn ligands, electron density for the side chain of the chelating histidine residue 180 was weak at best and independent of the mode of Zn ion refinement and it could not be reconciled with an optimal Zn bonding geometry. Superposition of the DHAP-bound H. pylori aldolase structure with structures of DHAP-bound E. coli aldolase herein, DHAP-bound M. tuberculosis aldolase and TBP-bound G. lamblia aldolase (RMSD 0.98 Å, 0.94 Å and 0.58 Å, respectively) superposes the position of the Zn ion in these structures with exterior site II in the H. pylori aldolase structure (22, 23). Different from H. pylori aldolase, the Zn ion cofactor in E. coli aldolase was refined using an isotropic B-factor occupying two distinct clearly distinguishable yet partially populated sites (even though DHAP binding site is fully occupied). The two Zn binding sites overlap with site II in H. pylori aldolase and site I which is the buried site of the native enzyme. Zn ion occupancies at site I and II are respectively of 0.35 and 0.65 in one subunit, and 0.50 for both in the other subunit. The absence of direct interaction by DHAP with the Zn ion at site I suggests that Zn is not likely required for substrate binding. At site II, the Zn 2+ metal cofactor in both E. coli and H. pylori aldolase interacted with DHAP contacting the C2 carbonyl group (2.68±0.23Å a ) and coordinating the C3 hydroxyl group (2.35±0.11Å a ), as was reported for M. tuberculosis aldolase in complex with DHAP (22). This finding is consistent with FT infrared spectroscopy on class II aldolase from yeast and E. coli that indicated no polarization of the DHAP C2 carbonyl group by the intrinsic zinc ion from the enzyme (20, 21). At the current resolution, it is not possible to differentiate native DHAP unambiguously from its enediolate form. Coplanarity of DHAP C1, C2, O2, C3 and O3 atoms, a requisite structural feature in the enediolate intermediate, would nevertheless lend support to identification of the intermediate as the enediolate when the Zn metal ion occupies site II. The metal cofactor is, however, significantly out of the plane of the putative enediolate and in apparent contradiction to the expected coplanarity by the Zn ion with the enediolate (35). Hydrogen bonding interaction by the DHAP carbonyl with the conserved Gly211 amide in H. pylori aldolase (Gly265 in E. coli aldolase) of 2.83 ± 0.09 Å b, shown in Figure 1, would facilitate stabilization of the enediolate and could preclude the necessity of inner shell coordination with the Zn ion to polarize the DHAP carbonyl and hence obviate coplanarity by the Zn ion with the enediolate plane. However, the anisotropy of the Zn ion position at site II entails positional plasticity and allowing the Zn ion at a Average taken over all subunits binding DHAP closest approach along the major axis of its ellipsoid to interact strongly with the bound DHAP. Coplanarity by Zn at this proximal position necessitates only a slight conformational displacement by the bound DHAP that is not hindered in the active site. DHAP-bound mutant structures in H. pylori aldolase. Structures of variants H180Q, E142A and E149A bound with DHAP showed an identical polypeptide fold as the DHAP bound native aldolase (RMSD of 0.32, 0.34 and 0.22 Å, respectively). DHAP bound in H180Q variant is shown in Figure 1C, while the other two variants are shown in Figure S1. In H180Q and E142A structures, the Zn ion occupies mutually exclusive positions at site I and II and we observe a water molecule replacing His180 in the Zn ion coordination sphere of the H180Q structure. Anomalous difference omit map supported the presence of Zn 2+ cofactor at both sites I and II. As for the variant E149A structure, Zn ion refinement suggested two overlapping isotropic sites spanning site II, although anisotropic Zn ion refinement at site II could not be ruled out due to the resolution of the E149A data set, and attesting to facile Zn ion exchange between site I and II. Substrate turnover in the crystalline state. Electron density maps derived from substrate soaking experiments with H. pylori aldolase crystals indicate the trapping of an active enzyme conformer in the crystal lattice. Cleavage activity is consistent with the observed substantial decrease in the electron density between putative FBP C3 and C4 atoms in the electron density difference omit map of the native H. pylori aldolase, shown in Figure 2A. The observed electron density map corresponded to a day-long incubation at saturating concentrations of FBP that revealed only triose-ps trapped in the active site. This interpretation was supported by occupancy refinement of DHAP and G3P to respective values of 0.7 and 0.6 and by B-factor similarity of the triose-ps with contacting residues. We also obtained structures from shorter FBP soaking time in crystals of variants E142A and E149A (Figure S2). In these structures calculated electron density maps showed a mixture of substrate and triose- Ps that was not easily teased apart by refinement due the significant overlap between triose-ps and substrate electron densities and that was exacerbated by the less than full active site occupancy of the combined species. The predominant species in one subunit of the E142A variant was FBP (Figure S2A), with cleavage products in the other subunit. In the E149A variant, electron density maps were interpreted as cleavage products in one subunit (Figure S2B), while the other subunit appeared void of ligands. b Average taken over all subunits binding DHAP and FBP 4

5 G3P was bound as a free aldehyde in the active site although G3P exists preponderantly in solution as the hydrated gem-diol form (36). Interpretation as the gemdiol form was not consistent with the electron density shown in Figure 2A. A notable feature of the aldehyde is the C2 hydroxyl which makes a critical hydrogen bond with Asp255. An equivalent binding mode by DHAP would result in charged repulsion between the DHAP ketone carbonyl and Asp255 and indicates a role by Asp255, a conserved active site residue, in discriminating triose-p binding. The relative disposition in the active site of the refined triose-p positions was consistent with the si face of the D-G3P aldehyde being subjected to a nucleophilic attack by the si face of enediolate yielding the expected (3S,4R) diastereoisomer of the FBP. Formation of this diastereoisomer requires that the G3P C1 carbonyl be in a trans-configuration with respect to its C2 hydroxyl which can be stabilized through the C1 carbonyl interacting with the His83 Nδ (Figure 2A). This interaction is consistent with the position of the Zn ion at buried site I to maintain its coordination with the His83 via its Nε atom and presenting the His83 Nδ for hydrogen bonding with ligands. Additional density around the G3P C1 carbonyl loci afforded interpretation of a second conformation consistent with the D-G3P aldehyde aligned re face with respect to DHAP, or cis configuration. In this configuration, the G3P carbonyl interacts weakly with the Zn ion at site II (3.1Å). By FTIR, a significant frequency shift at 1706 cm -1 was observed for the G3P carbonyl in the yeast and E. coli holoenzymes and implying significant polarization of the G3P carbonyl by the intrinsic metal cofactor (20, 21). This frequency shift, however, was observed with G3P bound to the free holoenzyme which could allow unrestricted approach by the Zn ion to a new site to tightly coordinate the G3P carbonyl. Displacement by the metal cofactor to such a site was observed in a catalytically incompetent enzymatic complex of G. lamblia aldolase (23). Hexitol bisphosphate. To test whether the Zn ion is required for ligand binding, crystals of E. coli apoaldolase were soaked with HBP, namely a mixture of non-cleavable analogues, D-mannitol 1,6- bisphosphate (MBP) and D-glucitol 1,6-bisphosphate (GBP). Structure determination corroborated the absence of Zn ion in the active site and showed preferential binding by GBP with the displacement of the Zn 2+ chelating residue His226 equivalent to His180 in H. pylori aldolase, and is shown in Figure 2B. E. coli aldolase is a class IIa aldolase as is yeast aldolase (37) which has a 6.5-fold preference for GBP over MBP (Ki 60 µm and 400 µm, respectively) (38). Superposition of the HBP-soaked structure from E. coli apoenzyme either with structures of the E. coli holoenzyme-dhap complex therein (RMSD 0.27 Å), the G. lamblia holoenzyme-tbp complex (RMSD 1.3 Å) or complexes of the M. tuberculosis holoenzyme with triose-ps (DHAP/G3P) and FBP (RMSD 0.82 Å and 0.81 Å, respectively), shows identical binding loci for phosphate oxyanions and homologous hydroxyl groups in the active site, even though in the holoenzyme structures the active site ligand and the metal cofactor interact (22, 23). Based on the apoenzyme structure, competent active site binding therefore does not appear to depend on the presence of the metal cofactor. His180/His226 side chain. In H. pylori and E. coli aldolase, His180 and His226, respectively, are both located at the juncture of the mobile β6-α8 loop and undergo the largest displacement of the Zn metal chelating residues upon active site ligand binding (Figure 3). Simulated-annealing omit map encompassing all three chelating histidine residues and metal cofactor showed weak density for the His180 side chain in DHAP-bound structures of H. pylori aldolase, even though the electron density map allowed unambiguous tracing of its peptide backbone. In fact, the side chain of His180 could not be modelled due to the positional disorder of side chains associated with residues of the loop β6-α8. Clear electron density delineating the His180 side chain was only observed in the crystal structures for tight binding competitive inhibitors, PGHPP and PGH, in H. pylori aldolase (26), where the metal cofactor at site II interacted solely with its coordinating residues. Our structural data indicates that during relocation by the Zn ion cofactor from site I to site II, the metal cofactor must weaken or disrupt its interaction with His180 to bind its cognate triose-p substrate. Kinetics of native and mutant aldolases. The Michaelis-Menten steady state parameters, k cat and K m, were determined at various ph for native H. pylori aldolase. The resultant ph activity profile is best described by a bell-shaped curve, shown in Figure 4A, with calculated pka 1 and pka 2 of 7.34 and The pka for the acidic limb is consistent with a histidine residue implicated in general acid-base catalysis. To assess the contribution by His180 to the catalytic activity, the residue was mutated to a glutamine residue to retain steric proportions while being unable to participate in a proton transfer at physiological ph. In addition, Glu142 and Glu149 in H. pylori aldolase were independently mutated to alanine, as was previously done for residues Glu181 and Glu182 in E. coli aldolase, to better portray a catalytic role attributed to the conserved residue (Glu142/Glu182) as opposed to the non-conserved residue (Glu149/Glu181) of the same loop (28). Michaelis-Menten steady state kinetics were used to analyze the initial rate velocities of the variants and their k cat and K m values are shown in Table 2. We observed a significant loss in activity for both H180Q and E142A variants. 5

6 To quantify the production of enediolate intermediate in both native H. pylori aldolase and variants, enediolate oxidation was analyzed in terms of hexacyanoferrate (III) reduction (39) using Michaelis- Menten kinetics and the k cat values are shown in Table 2. The turnovers for substrate in the oxidation reaction were lower than for the cleavage reaction and comparable to the enediolate oxidation in presence of DHAP (Table 2) and is not inconsistent with inefficient enediolate oxidation in presence of the nascent G3P in the native enzyme. The enediolate oxidation from FBP cleavage was lowest for the H180Q mutant, indicating the H180Q mutation exerts an inhibitory effect at the cleavage step with the limiting rate step likely being the proton exchange at the FBP C4 hydroxyl group. The reduction in enediolate formation from a DHAP substrate compared to the native enzyme suggests that the variant also perturbs enediolate formation. In the E142A variant, the rate step for enediolate formation from DHAP is greatly affected compared to native enzyme. The hundred-fold reduction in enediolate formation from DHAP is consistent with this variant impacting enediolate turnover in the active site, as noted previously (28). The two-fold reduction in enediolate oxidation at the cleavage step compared to native enzyme is consistent with efficient G3P dissociation preceding the enediolate oxidation; nevertheless, the variant impacts the cleavage step albeit to a lesser extent. The role of Asp82 was investigated in the D82N variant of H. pylori aldolase using a combination of FBP cleavage activity and the structure determination of D82N variant crystals soaked with substrates. This conserved residue had been previously implicated for its role in the proton abstraction of the C4 hydroxyl (21). No activity was detected for the isosteric D82N variant even though a hundred-fold greater quantity of enzyme was used compared to the native enzyme. Thus, the D82N variant could not be characterized kinetically; given our limit of detection, we can estimate at least a 30,000-fold reduction in activity. The structure determination of this variant revealed it be isostructural with the native enzyme (RMSD 0.51 Å) and with no loss of either the Zn 2+ metal cofactor or the Na + ion. We were unable to trap any ligand in the active site under soaking conditions comparable to the native enzyme, with all structures determined showing that same conformation as that of the free enzyme. The only significant observed difference between the native and the variant structure was that the variant Asn82 side chain was significantly less surface exposed than the native Asp82 due to its inability to form a hydrogen bond with Asn23. The Asn82 side chain conformation would induce steric clash through its carboxamide moiety with the DHAP and the FBP C3 hydroxyl groups, thereby precluding competent active site binding (Figure S3). Hexose binding. Aldolase crystals of the H180Q variant were soaked in FBP or TBP solutions under identical conditions which in native aldolase resulted in FBP cleavage. A simulated annealing difference omit map shows unambiguous density corresponding to the ligands fully bound in both subunits, as shown in Figure 5. In case of the FBP bound variant in Figure 5A, active site interactions by the substrate with the enzyme were identical, as was the polypeptide fold, when compared to the HBP-soaked native E. coli aldolase structure (RMSD 1.00Å). Importantly, the high-resolution data delineated unambiguously the active site interactions made by FBP that are essential for substrate recognition. Binding of mono- or bisphosphate ligands is dictated through active site interaction with the phosphate moieties, while interactions with hydroxyl groups and carbonyl groups serve to discriminate ligands. Conserved interactions independent of triose-p or FBP binding are as follows: Gly211 backbone amide hydrogen bonding with DHAP/FBP C2 carbonyl, Asp82 hydrogen bonding with DHAP/FBP C3 hydroxyl group and Asp255 hydrogen bonding with G3P/FBP C2/C5. The remaining interaction involving the G3P C1 carbonyl differs from that of the FBP C4 hydroxyl in the active site. The C4 hydroxyl interacts with His83 oriented such that it also chelates Zn 2+ sequestered at site I. While the G3P C1 carbonyl in adopting cis or trans configurations can interact either with His83 (trans) in the same manner as the FBP C4 hydroxyl or directly with the Zn 2+ cation at site II (cis) (Figure 2A). No Zn ion was observed in the FBP H180Q variant structure at site II, even though site II was occupied in case of the DHAP bound variant structure. In addition to Zn present at the buried site I in the FBP H180Q variant, the Zn ion also occupied an intermediate third site (site III), whilst globally retaining full occupancy. At site III, the Zn ion no longer has His210 in its first coordination sphere and is coordinated by His83, Glu134 and water molecules, adopting an octahedral geometry. The ensemble of structures clearly illustrates the plasticity of the active site in readily accommodating different spheres of coordination by the same Zn ion at overlapping loci. In the H180Q structure, the FBP C4 hydroxyl is hydrogen bonded through a relay of two water molecules, W3 and W4, with W3 coordinated to the Zn ion at site III (Figure 5A). To gain insight as to whether this relay of water molecules is responsible for the residual activity of this variant, its ph-activity profile was determined. An acidic limb in the ph profile having a calculated pka for this variant, shown in Figure 4B, is consistent with that of a water molecule ionized through its coordination with the Zn ion (40). The ionized water molecule W3 would act as general base 6

7 and through a proton transfer relay mediated by the intervening water molecule W4, similar to that found in carbonic anhydrase (41), initiates proton abstraction at the FBP C4 hydroxyl resulting in C-C bond cleavage. Binding by TBP ligand in the H180Q structure is shown in Figure 5B. The intrinsic Zn ion occupied sites I and II, with respective refined occupancies of 0.7 and 0.3. Ligand interactions and enzyme fold were identical with FBP-soaked H180Q variant structure (RMSD 0.13 Å). The TBP-bound structure was also identical to the structure of G. lamblia aldolase complexed with TBP (RMSD 0.66Å) (23). Both FBP and TBP diastereoisomers show the identical active site binding modes, with a RMSD of 0.45Å (all diastereoisomer heavy atoms) and RMSD 0.069Å without the O4 atom. In case of the bound TBP, the (S)-C4 hydroxyl is not implicated directly in any enzyme-ligand interactions, different from the FBP (R)-C4 hydroxyl that hydrogen bonds with His83. Open and closed conformations. Upon active site ligand binding, loop β6-α8 ( ) in H. pylori aldolase undergoes a significant conformational rearrangement (RMSD of 9.9±0.4 Å), shown in Figure 3. The loop closure facilitates active site access by the loop β5-α7 (residues ), containing residues Glu142 and Glu149, as in the open conformation, loop β6-α8 may interfere with active site access. Loop β6-α8 closure correlates with P 1-oxyanion binding in the active site. Lys184 in loop β6-α8 together with residues Ser213 and Thr256 and backbone amides of Gly181, Ser213 and Thr256 grasp the P 1 phosphate and thereby anchors the loop in a closed conformation. In other aldolases such as T. aquaticus aldolase (34), M. tuberculosis aldolase (22) and E. coli aldolase (27), a monocation replaces Lys184 for its role in loop stabilization upon P1-oxyanion binding through binding to backbone carbonyls of the equivalent loop. The loop closures facilitate the relocation of the catalytic Zn 2+ ion from its buried site I to the surface exposed site II. DISCUSSION The ability to cryotrap reaction intermediates in the crystalline state provided a unique opportunity to delineate the essential mechanistic features associated with proton transfers during substrate turnover and concomitant triose-phosphate discrimination. Crucial for the mechanistic description was the reduced turnover in the variants that enabled cryotrapping of enzymatic complexes along the aldol/retroaldol reaction coordinate. Our structural analysis examined catalytic intermediates in two bacterial aldolases each belonging to a distinct subfamily of class II FBP aldolase in eubacteria (37). Members of each subfamily exhibit 40% sequence similarity, while there is 25% 30% sequence similarity among all class II FBP aldolases. Active site residues are conserved and make identical interactions with active site ligands indicating that the same reaction mechanism catalyzing reversible aldol addition is likely exploited by all class II FBP aldolases. Structural-function analysis of H. pylori aldolase identified His180 as the residue responsible for initiating proton abstraction at the FBP C4 hydroxyl, the requisite step prior to the substrate cleavage (4 3). The kinetic data for the H180Q mutation showed a greatly reduced turnover in the retro-aldol reaction and a substantial rate decrease in enediolate formation from FBP. The variant E142A had a lesser effect on retroaldol reaction but showed a much greater decrease in the rate of enediolate formation from DHAP, consistent with its role in proton exchange (2 1) in enediolate formation (28). Although the residues His180 and Glu142 in H. pylori aldolase facilitate substrate cleavage and enediolate formation, respectively, the possibility of Glu142 activating His180 thereby increasing its pka and rendering proton transfer more efficient at the C4 hydroxyl group and in the enediolate cannot be excluded. These proton transfers steps are tightly regulated through conformational changes and the Zn 2+ cofactor displacement. The role of Asp82 in the active site appears to be structural in nature. The structural analysis of the D82N variant is consistent with strong inhibition of substrate and DHAP binding due to a putative close contact by the Asn82 carboxamide group with either the C3 hydroxyl of FBP or DHAP. The inability of this variant to bind the substrate or DHAP makes kinetics analysis not possible and is consistent with activity in this variant being too low to be detected by our assays. Substrate binding results in extensive hydrogen bonding interactions at both the P1 and the P6 phosphate binding loci. Side chains of conserved Ser213/267 and Thr256/289 residues interact with the P1 oxyanion, while conserved Arg280/331 and Ser49/61 residues interact with the P6 oxyanion in H. pylori/e. coli aldolases, respectively. Additional oxyanion interactions at these loci involve backbone atoms and intervening water molecules. To fine position active site ligands, hydrogen bonding interactions are made by hexose oxygens with other conserved residues. For FBP and its analogue GBP, this involves the C2 carbonyl hydrogen bonding with backbone of Gly211/265, the C3 hydroxyl group with Asp82/109, and the C5 hydroxyl group with Asp255/288. When the Zn ion is positioned at site I, the C4 hydroxyl hydrogen bonds with His83/110 Nδ1 (Figure 5A). This interaction is precluded with TBP diastereoisomer as the C4- hydroxyl has (S) instead of (R) chirality. Substrate/product recognition. Recognition and discrimination of the triose-phosphate substrates is critical in the H. pylori aldolase whose primary function 7

8 is FBP synthesis by the aldolization of D-G3P and DHAP during gluconeogenesis. Each triose-phosphate, although chemically different, are structurally similar. Different from the triose-phosphate isomerase reaction, the aldol reaction in aldolase requires each triosephosphate to interact with a different binding locus. The indiscriminate attachment by DHAP and G3P at either triose-phosphate binding locus is precluded in the active site of the H. pylori aldolase. The enzymatic mechanism explicitly recognizes and discriminates G3P and DHAP at the level of donor/acceptor hydrogen bonds implicating the carbonyl and the hydroxyl oxygens of the triose-p molecules. The carboxylate moiety of the conserved active site residue Asp82 makes a charged hydrogen bond with the DHAP C3 hydroxyl group, shown in Figure 1, thereby discriminating against aldehyde triose-phosphate binding as the equivalent G3P C1 carbonyl interaction would be repulsive at physiological ph. Furthermore, keto moieties are considered stronger hydrogen bond acceptors than hydroxyl groups, thus binding by the DHAP C2 carbonyl with the backbone amide Gly211 would provide an additional level of discrimination. The binding locus corresponding to the cognate aldehyde, G3P, specifically discriminates against DHAP through the charged hydrogen bond made by the conserved Asp255 with the G3P C2 hydroxyl group, shown in Figure 2A and Figure S2B, as the equivalent DHAP C2 carbonyl interaction would be repulsive. Substrate recognition does not appear to be mediated by interaction with the Zn metal cofactor. The DHAP-bound E. coli aldolase structure showed that even though the ligand is present at full occupancy in the active site, the Zn ion has only half occupancy at site I (Figure 1B). An FBP analogue, GBP, was found to bind in the active site of the apoenzyme (depleted of Zn ion), implicating the same interactions with active site residues as was made with a natural substrate in the H. pylori aldolase (Figure 2B). Nuclear relaxation studies using paramagnetic metal cofactors with acetolphosphate as ligand in class II yeast aldolase yielded a binding geometry where the metal cofactor did not interact with the ligand (42). The [2-13 C] enriched acetol-phosphate exhibited only a 13 C resonance corresponding to carbonyl interaction that would exclude direct ligand binding with the Zn ion at site II. The location of the metal cofactor at site I in presence of the ligand would be consistent with such an interpretation. The soaking of FBP into H. pylori aldolase crystals, which resulted in the trapping of the triose-ps, DHAP and G3P, indicates crystallization of a catalytically active conformer that provided a structural explanation for stereospecific mechanistic discrimination in class II aldolases. The aldol/retroaldol reaction catalyzed by the FBP class II aldolase is highly stereospecific, preferentially forming FBP than the TBP diastereoisomer (43, 44). Although by stabilizing the enediolate form, the metal cofactor interacts weakly at best with the G3P aldehyde in the cis configuration, as is seen in Figure 2A and Figure S2B. Consequently, the metal cofactor would not polarise the G3P carbonyl sufficiently for efficient protonation of the carbonyl O1 oxygen, and thereby disfavouring TBP formation that would result from the aldol reaction of the enediolate with the cis configuration of the aldehyde. G3P release from the active site does not entail additional conformational changes, as structures corresponding to DHAP binding and to FBP/triose-Ps binding are essentially identical (RMSD 0.3Å). Unencumbered G3P release preceding DHAP dissociation would be in agreement with the obligate binding sequence in catalytically competent class II aldolases (45). The presence of the metal cofactor at site II would facilitate G3P release as His83 Nδ 1 cannot interact with the G3P carbonyl in this rotamer configuration. Only when the Zn ion is relocated to site I with chelating His83 imidazole rotated 180 does His83 Nδ 1 face the carbonyl enabling interaction with G3P and trapping it in the trans-configuration. Nucleophilic attack by the DHAP carbanion would require G3P in the transconfiguration to form FBP. Binding by the obligate triose-p involves recognition of the carbonyl form of DHAP as binding by the gem-diol form was not observed and is consistent with NMR analysis of DHAP binding in the E. coli aldolase showing primarily or exclusive binding by the carbonyl form (24). Upon attachment, the P1 oxyanion triggers a cascade of conformational changes, including the closure of the loop β6-α8, enabling it to grasp the phosphate moiety. Asp82 then guides the DHAP molecule via its interaction with the C3 hydroxyl group into the active site. The C3 hydroxyl group of the enediolate intermediate is coordinated by the Zn metal ion at site II while its C2 carbonyl engages in a close contact with the Zn ion consistent with a bidentate coordination, as observed with the tight binding phosphoglycolohydroxamate inhibitor (27). The closest approach made by the Zn ion was found in a highresolution structure of M. tuberculosis aldolase, crystallized under acidic conditions, where the distances from the bound DHAP to the Zn ion were found to be 2.12Å (C3) and 2.20Å (C2) (22). The Zn ion, however, was not strictly coplanar with the putative enediolate deviating slightly from the enediolate plane by ~ 0.55Å. Although 13 C NMR data clearly showed a significant shift by the [2-13 C] DHAP carbonyl upon Zn ion coordination in the E. coli aldolase, it could not distinguish whether interaction by the metal cofactor with the DHAP carbonyl was direct or indirect (24). The interaction of the DHAP C2 carbonyl with the backbone amide of Gly211 would, however, polarize the keto 8

9 moiety facilitating electronic rearrangements leading to enediolate formation. Moreover, the low pka ~ 8.0 determined for the enediolate intermediate formation in E. coli aldolase by NMR (24) and in model compounds (25) would be consistent with at least partial trapping of DHAP in enediolate form under our basic crystallization conditions. As planarity of O2-C2-C3-O3 is not an exclusive feature of the enediolate (46), distinction between DHAP and its enediolate form depends on ligand interaction with the Zn ion. In both E. coli and H. pylori aldolase, the bound DHAP retains the same coplanar geometry even when the Zn ion occupies either site I in the E. coli aldolase or the apogee of the major axis of the elongated electron density at site II, distant from both the DHAP C2 carbonyl and the C3 hydroxyl. Active site binding by DHAP thus stabilizes a precursor to incipient cis-enediolate formation as the dominant enzymatic intermediate in both class II aldolases that does not entail metal cofactor interaction. Enediolate formation from the precursor would then be stabilized through metal cofactor relocation. Conformational transitions prior to cleavage. In both H. pylori and E. coli aldolase, DHAP attachment at the P1 phosphate binding site remodels the active site (Figure 3). The conformational process orders the loop β6-α8 (residues in H. pylori aldolase and residues in E. coli aldolase) thereby stabilizing binding by the P1 phosphate oxyanion. His180 in the H. pylori aldolase and equivalent His226 in the E. coli aldolase which are located at the inception of this loop undergo conformational displacement, corresponding to a translation by the respective Cα atoms of 5 Å and 4 Å respectively, which disrupts their ability to chelate the metal cofactor at site I. The crystallographic data suggests that a water molecule found coordinated with the Zn ion at site II would be consistent with an exchange process that ultimately displaces His 180/226. However, FBP binding does not stabilize a comparable conformational change by the Zn ion and involves a lesser loop β6-α8 movement (which includes His180). In the variant H180Q, the Zn ion remains buried, seen in Figure 5A, and does not entail relocation of the Zn ion to site II. Here, FBP binding is stabilized by the His83 Nδ 1 via a suboptimal hydrogen bond with the FBP C4 hydroxyl which hinders rotation of the Zn ion chelating residue, thereby inhibiting the metal cofactor relocation from the buried site I. However, in the TBP bound structure of the H180Q variant, the Zn ion relocation does occur as the TBP C4 hydroxyl cannot interact with the His83 Nδ 1 due to the (S) chirality of the C4 hydroxyl, thereby allowing imidazole rotation. Zn ion relocation is thus consistent with an exchange between sites I and II that is regulated by the competing interactions between His83 and FBP C4 hydroxyl versus that of the Zn ion with the enediolate or its precursor at site II. The formation of the enediolate species upon C3-C4 bond cleavage would advantageously stabilize the Zn ion at site II due to the enediolate negative charge as compared to its interaction with the uncharged nascent enediolate precursor in TBP and thus would favor Zn relocation to site II during FBP cleavage. The key to active site remodeling is the intrinsic ability of the Zn 2+ metal ion to readily accommodate different coordination geometries with ligands in the active site. The Zn ion although frequently tetrahedrally coordinated in catalytic sites may readily adopt pentaor hexa-coordinate geometry (47). Energy differences between corresponding four- to six- oxygen coordinated Zn ion complexes appear to differ by less than 0.4 kcal/mol (48). Such relatively low energy barriers to interconversion among Zn ion coordinated complexes would allow Zn ion relocation among sites I, II, and III to be driven by conformational processes that readily exchange Zn ion coordination spheres. The same conformational plasticity can give rise to artifactual Zn coordination due to inhibitory crystallization conditions used to trap the enzymatic complexes. Reversible substrate cleavage. A reaction mechanism which promotes cleavage of the C-C bond in class II FBP aldolases that integrates enzymological and structural data is shown in Figure 6. In this mechanism, we propose an unconventional role for a Zn ion chelating histidine residue namely as the residue responsible for the FBP C3-C4 bond cleavage. A catalytic role has not been attributed to His180 to date (Figure 6A). A schematic of the same mechanism implicating His226 of the E. coli aldolase is shown in Figure 6B. Critical to the proposed mechanism is the substrate binding which induces a conformational change that displaces His180 from the Zn ion at site I through its exchange with a water molecule, thereby freeing His180 to abstract the FBP C4-hydroxyl proton. In silico modelling of His180 onto the H180Q variant, yielded a reactant geometry capable of facilitating substrate cleavage as illustrated in Figure 6A and required adjustment of χ 1 and χ 2 torsion angles of His180, together with a slight rotation of the His83 imidazole. Proton abstraction of the FBP C4 hydroxyl group by His180 then initiates a rearrangement resulting in cleavage of the substrate C3-C4 bond. This mechanism is consistent with a trajectory of least atomic motion requiring only slight torsion librations involving primarily rotational motions of the histidine side chain. Histidine residues coordinated to a Zn 2+ ion manifest high pka ~13 (49). In the context of the proposed mechanism, this would allow the partial positive charge at the Nδ 1 atom of the chelating histidine His83/His110 to stabilize the negative charge at the C4-9

10 hydroxyl upon proton abstraction at physiological ph. This mechanism does not rule out a concerted rearrangement whereby the Zn ion relocates to site II upon or during proton abstraction and is stabilized by the resulting enediolate species. The Zn ion relocation would induce a repulsive interaction between the nascent G3P C1 carbonyl and the chelating His83 Cδ 2 thereby favouring the triose-p dissociation. His180 does not appear to play a structural role as His180 is not required for Zn ion relocation nor for ligand binding. The Zn ion relocation at sites I and II takes place in absence of His180, with a water molecule replacing His180 in the first coordination sphere, as is seen in the DHAP-bound H180Q aldolase structure (Figure 1C) and in the TBP-bound H180Q aldolase structure (Figure 5B). Refinement suggests a high B- factor for the His180 side chain consistent with considerable mobility by the imidazole side chain in the DHAP-bound native structure, the DHAP-bound E142A variant structure and the FBP-soaked E142A variant structure (68.8 Å 2, 74.0 Å 2 and 88.6 Å 2, respectively). The adjacent conserved glycine containing motif H 180G 181 in class II aldolases may help to promote this mobility. In this context, the histidine residue behaves as a mobile catalyst. TBP inhibition. The TBP ligand in the H180Q variant binds identically to FBP in the active site of the same variant, shown in Figure 5B, and is shown superposed on FBP in Figure 6A. The mechanism proposed for C3-C4 bond cleavage predicts that TBP is not a substrate for the enzyme but acts as an inhibitor (7). The C4 hydroxyl group in the TBP diastereoisomer precludes stabilization of the negative charge developing during proton abstraction by His83/His110. Proton abstraction is further disfavoured by the unfavourable geometry made between the C4 hydroxyl group with His180/His226. A catalytic role by a Zn ion chelating histidine residue in class II aldolases represents an entirely novel enzymatic mechanism of the C-C bond cleavage. The reaction mechanism is structurally dynamic using conformational changes by active site loops that drive the catalytic cycle to remodel the active site and generate a conformer competent for the C-C bond cleavage. The resultant conformer facilitates relocation of the Zn cofactor that advances the reaction by its stabilization of the nascent enediolate species. Our finding of a catalytic His residue on the mobile loop β6- α8 reinforces recent efforts inhibiting class II aldolases by interference with this loop displacement (50, 51). Furthermore our proposed mechanism offers new insights into understanding substrate specificity and conformational states requisite for catalysis - critical knowledge in drug design, protein engineering, and more recently in the synthesis of rare sugars (52). EXPERIMENTAL PROCEDURES Hexitol 1,6-bisphosphate (HBP) preparation. HBP was prepared by the NaBH 4 reduction of FBP and assayed by organic phosphate analysis as previously described, with FBP depletion confirmed by enzymatic assays (53). This reaction produces a mixture of two diastereoisomers: (2R,3R,4R,5R)-D-mannitol 1,6- bisphosphate (MBP) and (2S,3R,4R,5R)-D-glucitol 1,6- bisphosphate (GBP). Purification and crystallization of aldolases. Native H. pylori class II aldolase as well as aldolase variants Glu142Ala, Glu149Ala, Asp82Asn or His180Gln were purified and crystallized as previously reported (26), as was also the case for purification and crystallization of E. coli class II aldolase (54). Mutations were obtained via PCR cassette and validated from their molecular weight as determined by mass spectrometry. Data collection and processing. Aldolase crystals were soaked in mother liquor buffer (12% PEG 1000, 12% PEG 8000, 0.1 M calcium acetate, and 50 mm Tris/AcOH ph 8) containing 10 mm of ligand. At this concentration, FBP and other hexose bisphosphates become insoluble and a fraction of which precipitated in mother liquor buffer. Hence, the actual concentration of soluble FBP in the mother liquor will be lower in the crystal soaks. Native aldolases were soaked for 10 min to 4 h with DHAP, while mutants were soaked for 4 h as shorter times did not yield high ligand occupancy. Wild-type and H180Q variant of H. pylori aldolase were soaked for 24 h in FBP, under saturation conditions with no evidence for crystal cracking. The E142A and E149A variants of H. pylori aldolase were soaked for 1 h with FBP. In case of E. coli aldolase, crystals were soaked for 10 min with HBP and 10mM EDTA was added to the mother liquor to chelate the Zn 2+ metal ion. Prior to data collection, crystals were cryoprotected by transfer through a cryobuffer solution (soaking solution plus 10% glycerol) and immediately flash cooled in liquid nitrogen, or gaseous stream on site, at 100K. Diffraction datasets were collected at beamlines X8C, X12B, X25 and X29A of the National Synchrotron Light Source (Brookhaven National Laboratory, Upton, USA), and at beamlines 08B1-1 and 08ID-1 of the Canadian Light Source (Saskatoon, Canada). A fluorescence energy scan about the Zn Kα edge ( Å), collected at beamline X8C (NSLS), corroborated that the H. pylori aldolase crystals contained Zn, although no exogenous zinc was added during purification or crystallization of the protein. All data sets were processed with HKL2000 (55) and the results are summarized in Table I. Structure solution and refinement. Crystal structures of complexes formed by soaking H. pylori aldolase crystals with DHAP, FBP or TBP were solved using the PHENIX molecular replacement program (AutoMR) (56) with reference structures of either native 10

11 aldolase, PGH bound to native aldolase or N-(3- hydroxypropyl)-glycolohydroxamic acid bis-phosphate (PGHPP) structure bound to native aldolase, PDB entry codes 3C4U, 3C52, and 3C56, respectively (26). The best solution was used as starting point for refinement. Each structure was refined by iterative rounds of refinement using the phenix.refine module (57, 58) and model building in Coot (59). All structures reported belong to either the triclinic space group P1 or the monoclinic space group P2 1 and contain one aldolase homodimer in the asymmetric unit, consistent with the biologically active form of the enzyme. Molprobity server (60) and the Coot validating tools were used to optimize the structures during the refinement. Loops β5-α7, β6-α8 and α9-α10 were often associated with regions of weak electron density and their corresponding atomic positions were not modelled. At least one dataset per aldolase variant was analyzed using anomalous difference maps calculated from the anomalous scattering component for Zn to further corroborate the Zn ion positions. D82N variant of H. pylori aldolase crystallized in space group C2. Structures of this variant were solved using a single subunit of the native aldolase structure complexed with DHAP, as packing consideration in these crystals indicate only one subunit per asymmetric unit. The orientation of Asn82 carboxamide was chosen in the electron density map to allow the carboxamide oxygen to hydrogen bond with its backbone amide as was observed in all native structures. The alternate orientation would result in an unfavourable close contact with the both its backbone amide and a Na ion bound in the β-barrel interior. Crystal structures of E. coli aldolase were determined and refined by the same strategies using the E. coli aldolase structure in complex with PGH, PDB 1B57 (27), as molecular replacement template. Ligand modelling was based on interpretation of electron density shapes from 2Fo-Fc and Fo-Fc simulated annealing omit maps. Binding by phosphorylated ligands were readily discernable initially by density at phosphate binding loci. Additional controls included B-factor concordance of equivalent atoms in different subunits, real space fit (RSR) or phenix ligand Fit (61, 62) and were used to discriminate between different ligands and their conformations. Occupancy of active site ligands was assessed on the basis of B-factor agreement with adjacent active site residues. Final model statistics, calculated with PHENIX, Molprobity and SFCHECK (63), are shown in Table I. The coordinates and structure factors of relevant H. pylori aldolases and E. coli aldolases have been deposited with the Protein Data Bank (64). Ramachandran plot analysis with Molprobity placed at least 98 % of non-glycine and non-proline residues in all 14 structures in the most favourable region and with the remainder found in allowed regions, attesting to good model geometry. Comparisons. Structure alignment was performed with the program PyMOL (65). Root mean square deviations (RMSD) are reported based on superposition of equivalent Cα atoms between appropriate subunits. The enediolate plane was calculated by singular value decomposition of DHAP C2, O2, C3 and O3 atoms position (minimization of atom-plane distances) and the Zn ion distance from said plane was measured (66). Enzyme kinetics. Enzymatic activity of FBP cleavage was monitored by spectrophotometry using a coupled assay which follows NADH oxidation at 340 nm (67, 68). Enediolate production was assessed by measuring hexacyanoferrate (III) reduction monitored by the decrease in absorbance at 420 nm (39, 69). Kinetic assays were done at 25 C in buffer (ph 7.4) of TrisH (50mM) and potassium acetate (100mM). Initial rate measurements of enzymatic activity were concentration dependent and followed saturation kinetics for both the cleavage assay and the enediolate oxidation. The kinetic constants were determined from the initial rate velocity measurements using GraFit 5.0 Data Analyzing Software (70). ph profile assay. The following buffers (50 mm) were used to assay maximal velocities at the indicated ph intervals: PIPES (ph ), TrisH (ph ) and CHES (ph ). The ph data obtained at 25 C was fitted to equation (1) by non-linear regression in GraFit 5.0 to extract the pka of titratable groups. Activity = Limit 1 + (1) (Limit 2 Limit 1 ) 10 ph pka 1 10 ph pka (Limit 3 Limit 2 ) 10 ph pka 2 10 ph pka ACKNOWLEDGEMENTS This work was supported by Discovery Grant from the Natural Science and Engineering Research Council of Canada to JS (RGPIN ). Diffraction data for this study were collected in part on beamlines X12B, X25 and X29 of the National Synchrotron Light Source. Financial support comes principally from the Offices of Biological and Environmental Research and of Basic Energy Sciences of the US Department of Energy (Contract No. DE-AC02-98CH10886), and from the National Center for Research Resources (P41RR012408) and the National Institutes of Health (P41GM103473). Kind assistance by beamline personnel, Drs L. Flaks, D.K. Schneider, A. Soares, A. Héroux and H. Robinson, is gratefully acknowledged. 11

12 Diffraction data was also collected using beamline 08ID-1 and 08B1-1 at the Canadian Light Source, which is supported by the Natural Sciences and Engineering Research Council of Canada, the National Research Council Canada, the Canadian Institutes of Health Research, the Province of Saskatchewan, Western Economic Diversification Canada, and the University of Saskatchewan. Data was collected with the assistance of Shaun Labiuk. The atomic coordinates and structure factors (codes 5UCK, 5UCN, 5UCP, 5UCS, 5UCZ, 5UD0, 5UD1, 5UD2, 5UD3, 5UD4, 5VJD, 5VJE and 5VJF) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ ( We thank Dr Alexandra Furtos of the Université de Montréal Mass Spectrometry Facility for mass spectral analysis. We thank Oktavian Toka for his assistance in the kinetic characterization of native and mutant H. pylori aldolases. CONFLICT OF INTEREST The authors declare that they have no conflicts of interest with the contents of this article. AUTHOR CONTRIBUTIONS B. J. and M. C. helped design, collected and analyzed the crystallographic data, performed the biochemical experiments, and helped write the manuscript. J.S. conceived, helped design, analyzed the crystallographic data and biochemical experiments, and wrote and edited the manuscript. All authors approved the final version. REFERENCES: 1. Kobes, R. D., Simpson, R. T., Vallee, B. L., and Rutter, W. J. (1969) Functional role of metal ions in a class II aldolase. Biochemistry (Mosc.). 8, Rutter, W. J. (1964) Evolution of aldolase. Fed. Proc. 23, Gerdes, S. Y., Scholle, M. D., Campbell, J. W., Balazsi, G., Ravasz, E., Daugherty, M. D., Somera, A. L., Kyrpides, N. C., Anderson, I., Gelfand, M. S., Bhattacharya, A., Kapatral, V., D Souza, M., Baev, M. V., Grechkin, Y., Mseeh, F., Fonstein, M. Y., Overbeek, R., Barabasi, A.- L., Oltvai, Z. N., and Osterman, A. L. (2003) Experimental Determination and System Level Analysis of Essential Genes in Escherichia coli MG1655. J. Bacteriol. 185, Wehmeier, U. F. (2001) Molecular cloning, nucleotide sequence and structural analysis of the Streptomyces galbus DSM40480 fda gene: the S. galbus fructose-1,6-bisphosphate aldolase is a member of the class II aldolases. FEMS Microbiol. Lett. 197, Singer, M., Rossmiessl, P., Cali, B. M., Liebke, H., and Gross, C. A. (1991) The Escherichia coli ts8 mutation is an allele of fda, the gene encoding fructose-1,6-diphosphate aldolase. J. Bacteriol. 173, Singer, M., Walter, W. A., Cali, B. M., Rouviere, P., Liebke, H. H., Gourse, R. L., and Gross, C. A. (1991) Physiological effects of the fructose-1,6- diphosphate aldolase ts8 mutation on stable RNA synthesis in Escherichia coli. J. Bacteriol. 173, Galkin, A., Kulakova, L., Melamud, E., Li, L., Wu, C., Mariano, P., Dunaway-Mariano, D., Nash, T. E., and Herzberg, O. (2007) Characterization, kinetics, and crystal structures of fructose-1,6-bisphosphate aldolase from the human parasite, Giardia lamblia. J Biol Chem. 282, Marshall, B. J. (2001) One Hundred Years of Discovery and Rediscovery of Helicobacter pylori and Its Association with Peptic Ulcer Disease. in Helicobacter pylori: Physiology and Genetics (Mobley, H. L., Mendz, G. L., and Hazell, S. L. eds), ASM Press, Washington (DC), [online] (Accessed October 7, 2016) 9. Mitchell, H. M. (2001) Epidemiology of Infection. in Helicobacter pylori: Physiology and Genetics (Mobley, H. L., Mendz, G. L., and Hazell, S. L. eds), ASM Press, Washington (DC), [online] (Accessed October 7, 2016) 10. Motta, C. R. A., Cunha, M. P. S. S., Queiroz, D. M. M., Cruz, F. W. S., Guerra, E. J. C., Mota, R. M. S., and Braga, L. L. B. C. (2008) Gastric Precancerous Lesions and Helicobacter pylori Infection in Relatives of Gastric Cancer Patients from Northeastern Brazil. Digestion. 78, Lee, Y.-C., Lin, J.-T., Chen, T. H.-H., and Wu, M.-S. (2008) Is Eradication of Helicobacter pylori the Feasible Way to Prevent Gastric Cancer? New Evidence and Progress, but Still a Long Way to Go. J. Formos. Med. Assoc. 107, Fukase, K., Kato, M., Kikuchi, S., Inoue, K., Uemura, N., Okamoto, S., Terao, S., Amagai, K., Hayashi, S., and Asaka, M. (2008) Effect of eradication of Helicobacter pylori on incidence 12

13 of metachronous gastric carcinoma after endoscopic resection of early gastric cancer: an open-label, randomised controlled trial. The Lancet. 372, Talley, N. J. (2008) Is it time to screen and treat H pylori to prevent gastric cancer? Lancet Lond. Engl. 372, Doig, P., Jonge, B. L. de, Alm, R. A., Brown, E. D., Uria-Nickelsen, M., Noonan, B., Mills, S. D., Tummino, P., Carmel, G., Guild, B. C., Moir, D. T., Vovis, G. F., and Trust, T. J. (1999) Helicobacter pylori Physiology Predicted from Genomic Comparison of Two Strains. Microbiol. Mol. Biol. Rev. 63, Marais, A., Mendz, G. L., Hazell, S. L., and Mégraud, F. (1999) Metabolism and Genetics of Helicobacter pylori: the Genome Era. Microbiol. Mol. Biol. Rev. 63, Chalk, P. A., Roberts, A. D., and Blows, W. M. (1994) Metabolism of pyruvate and glucose by intact cells of Helicobacter pylori studied by 13C NMR spectroscopy. Microbiology. 140, Stingl, K., Uhlemann, E.-M., Deckers-Hebestreit, G., Schmid, R., Bakker, E. P., and Altendorf, K. (2001) Prolonged Survival and Cytoplasmic ph Homeostasis of Helicobacter pylori at ph 1. Infect. Immun. 69, Hoffman, P. S., Goodwin, A., Johnsen, J., Magee, K., and Zanten, S. J. V. van (1996) Metabolic activities of metronidazole-sensitive and -resistant strains of Helicobacter pylori: repression of pyruvate oxidoreductase and expression of isocitrate lyase activity correlate with resistance. J. Bacteriol. 178, Chalker, A. F., Minehart, H. W., Hughes, N. J., Koretke, K. K., Lonetto, M. A., Brinkman, K. K., Warren, P. V., Lupas, A., Stanhope, M. J., Brown, J. R., and Hoffman, P. S. (2001) Systematic identification of selective essential genes in Helicobacter pylori by genome prioritization and allelic replacement mutagenesis. J. Bacteriol. 183, Belasco, J. G., and Knowles, J. R. (1983) Polarization of substrate carbonyl groups by yeast aldolase: investigation by Fourier transform infrared spectroscopy. Biochemistry (Mosc.). 22, Plater, A. R., Zgiby, S. M., Thomson, G. J., Qamar, S., Wharton, C. W., and Berry, A. (1999) Conserved residues in the mechanism of the E. coli class II FBP-aldolase1. J. Mol. Biol. 285, Pegan, S. D., Rukseree, K., Franzblau, S. G., and Mesecar, A. D. (2009) Structural basis for catalysis of a tetrameric class IIa fructose 1,6- bisphosphate aldolase from Mycobacterium tuberculosis. J Mol Biol. 386, Galkin, A., Li, Z., Li, L., Kulakova, L., Pal, L. R., Dunaway-Mariano, D., and Herzberg, O. (2009) Structural insights into the substrate binding and stereoselectivity of giardia fructose- 1,6-bisphosphate aldolase. Biochemistry (Mosc.). 48, Szwergold, B. S., Ugurbil, K., and Brown, T. R. (1995) Properties of fructose-1,6-bisphosphate aldolase from Escherichia coli: an NMR analysis. Arch. Biochem. Biophys. 317, Kimura, E., Gotoh, T., Koike, T., and Shiro, M. (1999) Dynamic Enolate Recognition in Aqueous Solution by Zinc(II) in a Phenacyl-Pendant Cyclen Complex: Implications for the Role of Zinc(II) in Class II Aldolases. J. Am. Chem. Soc. 121, Fonvielle, M., Coinçon, M., Daher, R., Nicolas Desbenoit, Kosieradzka, K., Barilone, N., Brigitte Gicquel, Sygusch, J., Jackson, M., and Therisod, M. (2008) Synthesis and biochemical evaluation of selective inhibitors of class II fructose bisphosphate aldolases: towards new synthetic antibiotics. Chemistry. 14, Hall, D. R., Leonard, G. A., Reed, C. D., Watt, C. I., A. Berry, and Hunter, W. N. (1999) The crystal structure of Escherichia coli class II fructose-1, 6-bisphosphate aldolase in complex with phosphoglycolohydroxamate reveals details of mechanism and specificity. J Mol Biol. 287, Zgiby, S., Plater, A.., Bates, M.., Thomson, G.., and Berry, A. (2002) A functional role for a flexible loop containing Glu182 in the class II fructose-1,6-bisphosphate aldolase from Escherichia coli. J. Mol. Biol. 315, Collins, K. D. (1974) An activated intermediate analogue. The use of phosphoglycolohydroxamate as a stable analogue of a transiently occurring dihydroxyacetone phosphate-derived enolate in enzymatic catalysis. J Biol Chem. 249, Lewis, D. J., and Lowe, G. (1973) Phosphoglycollohydroxamic acid: an inhibitor of class I and II aldolases and triosephosphate isomerase. A potential antibacterial and antifungal agent. J Chem Soc Chem Commun /C Pegan, S. D., Rukseree, K., Capodagli, G. C., Baker, E. A., Krasnykh, O., Franzblau, S. G., and 13

14 Mesecar, A. D. (2013) Active site loop dynamics of a class IIa fructose 1,6-bisphosphate aldolase from Mycobacterium tuberculosis. Biochemistry (Mosc.). 52, Kabsch, W., and Sander, C. (1983) Dictionary of protein secondary structure: Pattern recognition of hydrogen-bonded and geometrical features. Biopolymers. 22, Touw, W. G., Baakman, C., Black, J., te Beek, T. A. H., Krieger, E., Joosten, R. P., and Vriend, G. (2015) A series of PDB-related databanks for everyday needs. Nucleic Acids Res. 43, D364 D Izard, T., and Sygusch, J. (2004) Induced fit movements and metal cofactor selectivity of class II aldolases: structure of Thermus aquaticus fructose-1,6-bisphosphate aldolase. J Biol Chem. 279, Islam, Z., Kumar, A., Singh, S., Salmon, L., and Karthikeyan, S. (2015) Structural Basis for Competitive Inhibition of 3,4-Dihydroxy-2- butanone-4-phosphate Synthase from Vibrio cholerae. J. Biol. Chem. 290, Trentham, D. R., McMurray, C. H., and Pogson, C. I. (1969) The active chemical state of d- glyceraldehyde 3-phosphate in its reactions with d-glyceraldehyde 3-phosphate dehydrogenase, aldolase and triose phosphate isomerase. Biochem. J. 114, Plaumann, M., Pelzer-Reith, B., Martin, W. F., and Schnarrenberger, C. (1997) Multiple recruitment of class-i aldolase to chloroplasts and eubacterial origin of eukaryotic class-ii aldolases revealed by cdnas from Euglena gracilis. Curr. Genet. 31, Mabiala-Bassiloua, C.-G., Zwolinska, M., Therisod, H., Sygusch, J., and Therisod, M. (2008) Separate synthesis and evaluation of glucitol bis-phosphate and mannitol bisphosphate, as competitive inhibitors of fructose bis-phosphate aldolases. Bioorg. Med. Chem. Lett. 18, Healy, M. J., and Christen, P. (1973) Mechanistic probes for enzymic reactions. Oxidationreduction indicators as oxidants of intermediary carbanions (studies with aldolase, aspartate aminotransferase, pyruvate decarboxylase, and 6- phosphogluconate dehydrogenase). Biochemistry (Mosc.). 12, Steiner, H., Jonsson, B. H., and Lindskog, S. (1975) The catalytic mechanism of carbonic anhydrase. Hydrogen-isotope effects on the kinetic parameters of the human C isoenzyme. Eur. J. Biochem. 59, Lindskog, S. (1997) Structure and mechanism of carbonic anhydrase. Pharmacol. Ther. 74, Smith, G. M., Mildvan, A. S., and Harper, E. T. (1980) Nuclear relaxation studies of the interaction of substrates with a metalloaldolase from yeast. Biochemistry (Mosc.). 19, Tung, T. C., Ling, K. H., Byrne, W. L., and Lardy, H. A. (1954) Substrate specificity of muscle aldolase. Biochim. Biophys. Acta. 14, Zgiby, S. M., Thomson, G. J., Qamar, S., and Berry, A. (2000) Exploring substrate binding and discrimination in fructose1, 6-bisphosphate and tagatose 1,6-bisphosphate aldolases. Eur. J. Biochem. FEBS. 267, Rose, I. A., O Connell, E. L., and Mehler, A. H. (1965) Mechanism of the Aldolase Reaction. J. Biol. Chem. 240, Sharma, A., Reva, I., Fausto, R., Hesse, S., Xue, Z., Suhm, M. A., Nayak, S. K., Sathishkumar, R., Pal, R., and Guru Row, T. N. (2011) Conformation-Changing Aggregation in Hydroxyacetone: A Combined Low-Temperature FTIR, Jet, and Crystallographic Study. J. Am. Chem. Soc. 133, Dudev, T., Lin, Dudev, M., and Lim, C. (2003) First Second Shell Interactions in Metal Binding Sites in Proteins: A PDB Survey and DFT/CDM Calculations. J. Am. Chem. Soc. 125, Bock, C. W., Katz, A. K., and Glusker, J. P. (1995) Hydration of Zinc Ions: A Comparison with Magnesium and Beryllium Ions. J. Am. Chem. Soc. 117, Martin, R. B. (1974) Pyrrole hydrogen ionization of imidazole derivatives in metal ion complexes and carbonic anhydrase. Proc. Natl. Acad. Sci. 71, Capodagli, G. C., Sedhom, W. G., Jackson, M., Ahrendt, K. A., and Pegan, S. D. (2014) A noncompetitive inhibitor for Mycobacterium tuberculosis s class IIa fructose 1,6-bisphosphate aldolase. Biochemistry (Mosc.). 53, Capodagli, G. C., Lee, S. A., Boehm, K. J., Brady, K. M., and Pegan, S. D. (2014) Structural and Functional Characterization of Methicillin- Resistant Staphylococcus aureus s Class IIb Fructose 1,6-Bisphosphate Aldolase. Biochemistry (Mosc.). 53, Lee, S.-H., Hong, S.-H., An, J.-U., Kim, K.-R., Kim, D.-E., Kang, L.-W., and Oh, D.-K. (2017) Structure-based prediction and identification of 4-epimerization activity of phosphate sugars in class II aldolases. Sci. Rep. 7,

15 53. Ginsburg, A., and Mehler, A. H. (1966) Specific Anion Binding to Fructose Diphosphate Aldolase from Rabbit Muscle*. Biochemistry (Mosc.). 5, Blom, N. S., Tétreault, S., Coulombe, R., and Sygusch, J. (1996) Novel active site in Escherichia coli fructose 1,6-bisphosphate aldolase. Nat Struct Biol. 3, Otwinowski, Z., and Minor, W. (1997) [20] Processing of X-ray diffraction data collected in oscillation mode (Enzymology, B.-M. in ed), pp , Macromolecular Crystallography Part A, Academic Press, 276, Zwart, P., Afonine, P., Grosse-Kunstleve, R., Hung, L.-W., Ioerger, T., McCoy, A., McKee, E., Moriarty, N., Read, R., Sacchettini, J., Sauter, N., Storoni, L., Terwilliger, T., and Adams, P. (2008) Automated Structure Solution with the PHENIX Suite. in Structural Proteomics (Kobe, B., Guss, M., and Huber, T. eds), pp , Methods in Molecular Biology TM, Humana Press, / _ Afonine, P., Grosse-Kunstleve, R., and Adams, P. (2005) The Phenix refinement framework. CCP4 Newsl. 58. Adams, P. D., Grosse-Kunstleve, R. W., Hung, L.-W., Ioerger, T. R., McCoy, A. J., Moriarty, N. W., Read, R. J., Sacchettini, J. C., Sauter, N. K., and Terwilliger, T. C. (2002) PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr. D Biol. Crystallogr. 58, Emsley, P., and Cowtan, K. (2004) Coot: modelbuilding tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, Lovell, S. C., Davis, I. W., Arendall, W. B., de Bakker, P. I. W., Word, J. M., Prisant, M. G., Richardson, J. S., and Richardson, D. C. (2003) Structure validation by Cα geometry: ϕ,ψ and Cβ deviation. Proteins Struct. Funct. Bioinforma. 50, Terwilliger, T. C., Adams, P. D., Moriarty, N. W., and Cohn, J. D. (2007) Ligand identification using electron-density map correlations. Acta Crystallogr. D Biol. Crystallogr. 63, Terwilliger, T. C., Klei, H., Adams, P. D., Moriarty, N. W., and Cohn, J. D. (2006) Automated ligand fitting by core-fragment fitting and extension into density. Acta Crystallogr. D Biol. Crystallogr. 62, Vaguine, A. A., Richelle, J., and Wodak, S. J. (1999) SFCHECK: a unified set of procedures for evaluating the quality of macromolecular structure-factor data and their agreement with the atomic model. Acta Crystallogr. D Biol. Crystallogr. 55, Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N., and Bourne, P. E. (2000) The Protein Data Bank. Nucleic Acids Res. 28, Schrödinger, L. The PyMOL Molecular Graphics System, Version Mura, C., McCrimmon, C. M., Vertrees, J., and Sawaya, M. R. (2010) An Introduction to Biomolecular Graphics. PLOS Comput Biol. 6, e Warburg, O., and Christian, W. (1943) [Isolation and crystallisation of the enzyme zymohexose]. Biochem Z. 314, Richards, O. C., and Rutter, W. J. (1961) Preparation and Properties of Yeast Aldolase. J. Biol. Chem. 236, Qamar, S., Marsh, K., and Berry, A. (1996) Identification of arginine 331 as an important active site residue in the Class II fructose-1,6- bisphosphate aldolase of Escherichia coli. Protein Sci. 5, Leatherbarrow, R. J. (2001) GraFit Version 5, Erithacus Software Ltd., Horley, U.K. 15

16 Table 1. Data collection and refinement statistics (1 of 3) Enzyme EcA EcA HpA HpA HpA Ligand DHAP GBP - DHAP DHAP+G3P PDB ID 5VJD 5VJE n/a 5VJF 5UCK Data collection Wavelength (Å) Resolution range (Å) ( )* ( ) ( ) ( ) ( ) Space group P21 P21 P1 P21 P1 Unit cell parameters a (Å), b (Å), c (Å), α ( ), β ( ), γ ( ) Unique reflections / Multiplicity / / / 1.8 (3891 / 1.6) / / 1.9 (4429 / 1.77) Completeness (%) 92.9 (82.0) 95.1 (84.0) 91.0 (70.9) 91.9 (82.0) (71.26) Mean I/ (I) 19 (1.7) 28.2 (2.3) 7.4 (1.8) 16.5 (3.3) 9.9 (1.9) Wilson B-factor R-merge 0.08 (0.60) 0.09 (0.40) 0.06 (0.30) 0.06 (0.21) 0.06 (0.41) Refinement statistics Reflections used in refinement / free / 2762 (2977 / 119) / 2104 (6493 / 193) / 1728 (3420 / 129) / 5111 (1276 / 135) / 2077 (3441 / 160) R-cryst (%) 14.8 (32.6) 13.6 (24.3) 19.2 (33.4) 13.7 (18.9) 18.7 (31.5) R-free (%) 17.3 (33.4) 16.4 (29.0) 23.4 (39.5) 17.4 (27.0) 22.7 (35.4) #Atoms Protein Ligand Solvent Bonds RMS Length (Å) Angle ( ) Ramachandran Favored (%) Allowed (%) Average B-factor (Å 2 ) * All values in parentheses are given for the highest resolution shell R merge = hkl i I i(hkl) - Ī(hkl) / hkl iī(hkl), with i running over the number of independent observations of reflection hkl. R cryst = Σ hkl F o(hkl) - F c(hkl) /Σ hkl F o(hkl). R free = Σ hkl T F o(hkl) - F c(hkl) /Σ hkl T F o(hkl), where T is a test data set randomly selected from the observed reflections prior to refinement. Test data set was not used throughout refinement and contained 5-10% of the total unique reflections. 16

17 Table 1. Data collection and refinement statistics (2 of 3) Enzyme HpA-E142A HpA-E142A HpA-E149A HpA-E149A HpA-E149A Ligand DHAP FBP/DHAP+G3P - DHAP DHAP+G3P PDB ID 5UCN 5UCP 5UCS 5UCZ 5UD0 Data collection Wavelength (Å) Resolution range (Å) ( )* ( ) ( ) ( ) ( ) Space group P1 P1 P21 P21 P21 Unit cell parameters a (Å), b (Å), c (Å), α ( ), β ( ), γ ( ) Unique reflections / Multiplicity / 2.6 (4344 / 2.0) / 2.1 (7703 / 1.9) / 4.0 (9006 / 3.7) / 3.5 (4426 / 2.6) / 3.7 (5384 / 3.3) Completeness (%) 91.2 (75.2) 88.6 (71.6) 93.4 (81.4) 93.3 (76.4) 94.5 (84.1) Mean I/ (I) 10.3 (1.9) 8.6 (1.6) 10.5 (1.8) 12.5 (1.9) 13.2 (2.0) Wilson B-factor R-merge 0.09 (0.53) 0.06 (0.54) 0.07 (0.59) 0.08 (0.71) 0.06 (0.55) Refinement statistics Reflections used in refinement / free / 2213 (3535 / 179) / 3530 (6026 / 278) / 4326 (7366 / 379) / 2010 (3596 / 163) / 2734 (4719 / 243) R-cryst (%) 16.5 (26.6) 15.5 (25.5) 15.8 (23.3) 17.8 (27.9) 17.2 (22.1) R-free (%) 20.2 (31.2) 19.0 (28.6) 18.2 (26.9) 22.1 (31.1) 21.6 (27.7) #Atoms Protein Ligand Solvent Bonds RMS Length (Å) Angle ( ) Ramachandran Favored (%) Allowed (%) Average B-factor (Å 2 ) * All values in parentheses are given for the highest resolution shell R merge = hkl i I i(hkl) - Ī(hkl) / hkl iī(hkl), with i running over the number of independent observations of reflection hkl. R cryst = Σ hkl F o(hkl) - F c(hkl) /Σ hkl F o(hkl). R free = Σ hkl T F o(hkl) - F c(hkl) /Σ hkl T F o(hkl), where T is a test data set randomly selected from the observed reflections prior to refinement. Test data set was not used throughout refinement and contained 5-10% of the total unique reflections. 17

18 Table 1. Data collection and refinement statistics (3 of 3) Enzyme HpA-H180Q HpA-H180Q HpA-H180Q HpA-H180Q Ligand - DHAP FBP TBP PDB ID 5UD1 5UD2 5UD3 5UD4 Data collection Wavelength (Å) Resolution range (Å) ( )* ( ) ( ) ( ) Space group P1 P1 P1 P1 Unit cell parameters a (Å), b (Å), c (Å), α ( ), β ( ), γ ( ) Unique reflections / Multiplicity / 1.8 (3940 / 1.7) / 2.1 (3750 / 1.9) / 1.8 (8306 / 1.8) / 1.8 (6978 / 1.6) Completeness (%) 89.0 (78.5) 89.4 (66.9) 91.5 (80.1) 90.2 (75.2) Mean I/ (I) 9.4 (1.9) 8.6 (2.0) 9.7 (2.7) 7.5 (2.0) Wilson B-factor R-merge 0.06 (0.37) 0.08 (0.32) 0.05 (0.25) 0.08 (0.26) Refinement statistics Reflections used in refinement / free / 1860 (3254 / 171) / 2086 (3140 / 151) / 3480 (7363 / 293) / 3538 (6126 / 296) R-cryst (%) 20.5 (27.3) 18.9 (25.8) 13.3 (16.6) 13.1 (17.5) R-free (%) 25.1 (32.9) 21.4 (29.9) 16.3 (21.4) 17.4 (24.9) #Atoms Protein Ligand Solvent Bonds RMS Length (Å) Angle ( ) Ramachandran Favored (%) Allowed (%) Average B-factor (Å 2 ) * All values in parentheses are given for the highest resolution shell R merge = hkl i I i(hkl) - Ī(hkl) / hkl iī(hkl), with i running over the number of independent observations of reflection hkl. R cryst = Σ hkl F o(hkl) - F c(hkl) /Σ hkl F o(hkl). R free = Σ hkl T F o(hkl) - F c(hkl) /Σ hkl T F o(hkl), where T is a test data set randomly selected from the observed reflections prior to refinement. Test data set was not used throughout refinement and contained 5-10% of the total unique reflections. 18

19 Table 2: Steady state kinetic parameters for native and mutant H. pylori aldolases dreived from initial rate velocities of substrate cleavage and enediolate oxidation. Reaction Cleavage Enediolate oxidation FBP DHAP+G3P FBP ENE DHAP ENE Enzyme k cat (s -1 ) K m (µm) k cat (s -1 ) K m (µm) k cat (s -1 ) K m (µm) Native 0.71± ± ± ±1 0.75± ±50 H180Q ± ± ± ± ± ±5 E142A ± ± ± ± ± ±70 E149A 0.66± ±8 ND ND ND ND D82N < ND ND ND ND ND ENE refers to enediolate species ND: Not determined 19

20 Figure 1: Electron density of trapped dihydroxyacetone-p intermediates in the active site of Class II aldolases. Panels show A- dihydroxyacetone-p bound in the active site of the native H. pylori aldolase, B- dihydroxyacetone-p bound in the active site of the native E. coli aldolase and C- dihydroxyacetone-p bound in the active site of the H180Q variant H. pylori aldolase. Electron density encompassing dihydroxyacetone-p was calculated from a simulated annealing Fo-Fc omit map and contoured at 3.0σ for A and B, and 1.5σ for C. In panel A, the divalent zinc ion is depicted as a transparent ellipsoid and represents the probability displacement calculated from the anisotropic B-factor of the zinc metal ion in the H. pylori aldolase. In panels B and C, the zinc ion is shown as two brown spheres for two mutually exclusive sites I and II with respective occupancies of 0.35 and 0.65 in the E. coli aldolase and with respective occupancies of 0.30 and 0.70 in the H180Q variant of H. pylori aldolase. Water molecules are shown as red spheres and the sodium ion in purple. Water molecule denoted by Wp interacts with phosphate group at site P1, while W2 denotes a water molecule liganded by the zinc ion at site II. View is looking into the β-barrel from the carboxyl side of the β-strands. 20

21 Figure 2: Electron density of hexose bisphosphate and trioses phosphate trapped in the active sites of Class II aldolases. Panels show stereo-images of A- dihydroxyacetone-p and glyceraldehyde-3-p trapped in the active site of the H. pylori aldolase, showing mutually exclusive cis and trans geometries for G3P in light and dark pink, respectively, and B- glucitol-1,6-bisphosphate trapped in the active site of the E. coli aldolase. Electron density enclosing ligands was calculated from a simulated annealing Fo-Fc omit map and contoured at 2.5σ for glucitol-1,6-bisphosphate and at 1.5σ for triose-ps. Water molecules are shown as red spheres and the sodium ion in purple. The zinc ion is labelled for sites I and II. Electron densities were drawn taking into account the partial occupancies. View is looking into the β-barrel from the carboxyl side of the β-strands. 21

22 Figure 3: Conformational changes induced upon DHAP binding in class II aldolases. H. pylori aldolase and E. coli aldolase share the loops β5-α7 (residues and , respectively), β6-α8 (residues and ) and β7-α9 (residues and ). Class IIb H. pylori aldolase has another loop displaced concomitantly by the loop β6-α8, namely the loop α9-α10 (residues ). The putative catalytic residue, His180 for the H. pylori aldolase and His226 for the E. coli aldolase, are part of the loop β6-α8, while the catalytic residue responsible for enediolate protonation, Glu142 for the H. pylori aldolase and Glu182 for the E. coli aldolase (28), are found in the unobserved region of loop β5-α7. The H. pylori aldolase structures are coloured in orange shades with pink DHAP and the E. coli aldolase structures are in blue shades with magenta DHAP. Respective lighter colours correspond to the unbound structures. Zinc ions are shown as spheres. 22

23 Figure 4: ph activity profiles for the H. pylori aldolase. Panels show the ph profile for A- the native enzyme and B- the variant H180Q. Bars represent the standard errors of the turnover determined from a nonlinear regression analysis of the fit of the Michaelis-Menten steady state rate equation to the initial rate velocities measured at each ph. 23

24 Figure 5: Diastereoisomer binding in the active site of the H. pylori aldolase variant H180Q. Panels show stereo images of A- fructose-1,6-bisphosphate and B- tagatose-1,6-bisphosphate trapped in active site of the H180Q variant of H. pylori aldolase. Electron density enclosing ligands was calculated from a simulated annealing Fo-Fc omit map and contoured at 3.0σ. Water molecules are shown as red spheres and the sodium ion in purple. The zinc ion is labelled both at site I or III. Site III shows the zinc ion octahedrally coordinated by H83, D104, E134 and 3 water molecules. Explicitly identified in panel A are two water molecules (W3, W4) that bridge through hydrogen bonds the FBP C4 hydroxyl group with the Zn ion at site III. View is looking into the β-barrel from the carboxyl side of the β-strands. 24