Active Site and Remote Contributions to Catalysis in Methylthioadenosine Nucleosidases

Abstract

5′-Methylthioadenosine/S-adenosyl-L-homocysteine nucleosidases (MTANs) catalyze the hydrolysis of 5′-methylthioadenosine to adenine and 5-methylthioribose. The amino acid sequences of the MTANs from Vibrio cholerae (VcMTAN) and Escherichia coli (EcMTAN) are 60% identical and 75% similar. Protein structure folds and kinetic properties are similar. However, binding of transition-state analogues is dominated by favorable entropy in VcMTAN and by enthalpy in EcMTAN. Catalytic sites of VcMTAN and EcMTAN in contact with reactants differ by two residues; Ala113 and Val153 in VcMTAN are Pro113 and Ile152, respectively, in EcMTAN. We mutated the VcMTAN catalytic site residues to match those of EcMTAN in anticipation of altering its properties toward EcMTAN. Inhibition of VcMTAN by transition-state analogues required filling both active sites of the homodimer. However, in the Val153Ile mutant or double mutants, transition-state analogue binding at one site caused complete inhibition. Therefore, a single amino acid, Val153, alters the catalytic site cooperativity in VcMTAN. The transition-state analogue affinity and thermodynamics in mutant VcMTAN became even more unlike those of EcMTAN, the opposite of expectations from catalytic site similarity; thus, catalytic site contacts in VcMTAN are unable to recapitulate the properties of EcMTAN. X-ray crystal structures of EcMTAN, VcMTAN, and a multiple-site mutant of VcMTAN most closely resembling EcMTAN in catalytic site contacts show no major protein conformational differences. The overall protein architectures of these closely related proteins are implicated in contributing to the catalytic site differences.

5′-Methylthioadenosine/S-adenosyl-L-homocysteine nucleosidase (MTAN, EC 3.2.2.16) from Vibrio cholerae catalyzes the hydrolysis of 5′-methylthioadenosine (MTA) to adenine and 5-methylthioribose (MTR).¹ MTANs are involved in the pathways of autoinducer-1 and autoinducer-2 quorum sensing in Gram-negative bacteria. Thus, inhibition of MTANs interferes with pathogenicity functions associated with quorum sensing.² In a few microorganisms, including Helicobacter pylori, MTAN is also required for menaquinone synthesis, an essential pathway for electron transport. In these organisms, MTAN inhibition provides a unique antibiotic approach by targeting menaquinone synthesis.³⁻⁵

The amino acid sequence of VcMTAN is 60% identical and 75% similar to that of EcMTAN (Figure S1 of the Supporting Information). Both MTANs are homodimers with similar protein folds and a single active site per monomer. The energetics of binding of picomolar affinity transition-state analogues to V. cholerae MTAN (VcMTAN) indicate that favorable entropic contributions dominate binding, with smaller favorable enthalpic contributions.⁶ In contrast, binding of the inhibitor to Escherichia coli MTAN (EcMTAN) is dominated by a large favorable enthalpy accompanied by smaller but also favorable entropic factors.⁶

Analysis of the amino acid sequence and X-ray crystal structures of VcMTAN and EcMTAN in complex with 4′-deaza-1′-aza-2′-deoxy-1′-(9-methylene)-Immucillin-A (DADMe-ImmA) and similar compounds revealed only two differences in the amino acid residues directly in contact with the bound inhibitor (Figure 1 and Figure S1 of the Supporting Information). At position 113, there is an alanine in VcMTAN and a proline in EcMTAN. The residue at position 113 is located on a loop structure that interacts with the 5′-region of the bound substrate or inhibitor at the active site of the adjacent monomer. Ala113 contributes a hydrophobic interaction to the 5′-group of the inhibitor near the solvent interface of the active site.⁷

Figure 1.

Structures of select residues of EcMTAN (green) and VcMTAN (blue) in the active sites bound to transition-state analogues. The top label corresponds to the residue numbering and identity in VcMTAN, while the bottom label (in parentheses) corresponds to analogous residues in EcMTAN. (A) Structural orientation of selected active site residues in complex with BuTDIA in the active sites of VcMTAN and EcMTAN. (B) Structural orientation of selected active site residues in complex with MTDIA in the active sites of VcMTAN and EcMTAN.

In EcMTAN, residue 152 is an isoleucine, while its analogous residue in VcMTAN is Val153. The main chain nitrogen and oxygen of Val153 in VcMTAN form hydrogen bonds to N1 (H-bond length of 3.1 Å) and N6 (H-bond length of 2.9 Å), respectively, of the deazaadenine base of BuTDIA (Figure S3 of the Supporting Information). If either of these hydrogen bonds is weakened in VcMTAN_V153I, one would expect a decrease in binding energy for BuTDIA.

In silico mutations of Ile153 and Ala113 in VcMTAN to the analogous residues found in EcMTAN (Val152 and Pro113, respectively) are predicted to be compatible with the overall structure, with the preservation of similar hydrogen bonding distances to the bound inhibitor (Figure S4 of the Supporting Information). However, the in silico single mutant, Val153Ile, of VcMTAN predicted that one of the isoleucine rotamers had the potential for steric hindrance with the neighboring Arg158 located at the dimer interface (Figure 1 and Figure S4 of the Supporting Information). Significantly, the residue corresponding to Arg158 in EcMTAN is glycine. On the basis of these differences, we also investigated the effect of the triple mutant, VcMTAN_{A113P/V153I/R158G} (VcMTAN_TRIPLE), all of which mutate the VcMTAN structure toward EcMTAN. Analyses of 17 individual active site mutations in EcMTAN have been performed in prior studies, yet the effects of mutations at positions 113, 153, and 158 (VcMTAN numbering) have not been reported for any MTAN.⁸

Structural comparison of VcMTAN-BuTDIA [Protein Data Bank (PDB) entry 3DP9]⁹ with EcMTAN-MTDIA (PDB entry 1Y6Q)⁷ demonstrated similar overall structures with a root-mean-square deviation (rmsd) of 0.56 Å, in an all-atom comparison, with 2117 atoms accepted for alignment.¹⁰ Previous molecular dynamic simulations of these enzymes in complex with BuTDIA suggested that an increased level of dynamic motion in the EcMTAN complex was responsible for the binding of transition-state analogues being tighter than to VcMTAN.¹¹ Here, in addition to the mutagenesis analysis, we determined the crystal structures of EcMTAN-BuTDIA (PDB entry 4WKC), VcMTAN-MTDIA (PDB entry 4WKB), and VcMTANTRIPLE-MTDIA (PDB entry 4×24) to permit direct comparisons between MTANs in complexes with the same inhibitors and to probe the effects of multiple mutations on the structures.

We measured the thermodynamic and kinetic properties of VcMTANs with active site residues mutated to be identical to those residues found in EcMTAN. The original hypothesis was that the catalytic site residue differences between VcMTAN and EcMTAN are responsible for the differences in ligand binding affinities and thermodynamic properties. Mutation of the active site residues of VcMTAN to match those of EcMTAN was anticipated to shift the catalytic properties of the modified VcMTANs to more closely resemble those of EcMTAN. Instead, the inhibitor affinity and the thermodynamic binding contributions in the altered VcMTANs became more dissimilar to the properties of EcMTAN. The mutations also altered catalytic site cooperativity, revealing a functional path for dimer−subunit communication in VcMTAN.

MATERIALS AND METHODS

Chemicals

Transition-state analogue inhibitors were synthesized, purified, and characterized as previously reported.^12,13 [¹⁴C]MT-DADMe-ImmA [MTDIA*, 4′-deaza-1′-aza-2′-deoxy-1′-(9-[¹⁴C]methylene)-Immucillin-A] was synthesized with the incorporation of [¹⁴C]formaldehyde to introduce the radiolabel at the methylene bridge.⁶

Enzyme Expression

The VcMTAN (WT) gene, contained in a pDEST14 (Life Technologies) vector, was expressed and the protein purified as previously reported.⁶ Mutant VcMTANs were generated using mutagenic polymerase chain reaction (PCR) primers (Life Technologies) and the GENEART Site-Directed Mutagenesis System (Life Technologies). The PCR product was transformed into One Shot TOP10 Electrocomp Cells (Life Technologies) and plated onto LB agar plates containing the appropriate antibiotic. Plasmids from isolated colonies were sequenced across the MTAN region to confirm the presence of the mutation and then transformed into BL21(DE3) competent cells (Life Technologies). Proteins were expressed in cells grown in LB, induced with 1 mM IPTG (37°C, 10 h), and purified on a Ni-NTA benchtop column according to a previously published protocol.⁶ The proteins were >90% pure as assessed by sodium dodecyl sulfate−polyacrylamide gel electrophoresis.

Protein Preparation for ITC

MTANs are copurified with tightly bound adenine, and it was necessary to remove this ligand prior to ITC measurements. Adenine was removed by dialysis against phosphate buffer and activated charcoal as previously described.⁶ The dimer concentrations used in ITC experiments ranged from 60 to 120 μM.

Kinetic Analysis

The K_M and k_cat values were determined using a modified protocol for a luminescence assay that converts the product adenine to ATP in a coupled assay in which the ATP is detected by luciferase.¹⁴ Each reaction was performed in 50 μL in a 96-well plate format, and luminescence was detected at 570 nm using a SpectraMax L Luminescence Microplate reader (Molecular Devices). Reaction wells contained 50 mM Tris acetate, 1 mM NH₄Cl, 1 mM PEP, 1 mM PRPP (pH 7.7), 8 μL of ATPlite (PerkinElmer), and 10 mM MgCl₂. The coupling enzymes pyruvate phosphate dikinase (Clostridium symbiosum PPDK,¹⁵ EC 2.7.9.1) and adenine phosphoribosyltransferase (Saccharomyces cerevisiae APRTase,¹⁶ EC 2.4.2.7) were added in excess at levels of 320 and 20 microunits/well, respectively. The MTAN concentrations used ranged from 5 to 25 pM. The reactions were initiated by the addition of varying concentrations of MTA (0.02−1 μM).

Catalytic Site Titration

MTDIA has a high affinity for native MTANs and therefore exhibits stoichiometric binding in ITC or static preincubation experiments. MTDIA was added to MTAN at known molar ratios and the residual activity recorded to explore catalytic site stoichiometry. A direct UV−vis spectrophotometric assay was used to measure the disappearance of MTA as a function of time at 25°C and 274 nm.

K_i Determination, Isothermal Titration Calorimetry Studies, and Processing of ITC Data

The experimental procedures for quantitation of kinetic parameters, catalytic site titration, and isothermal titration calorimetry (ITC) for MTAN have been described in detail.⁶ The K_i values were determined using varying concentrations of the inhibitor in the presence of 0.8 mM MTA and using 1.0 unit of xanthine oxidase (Sigma-Aldrich), monitored at 305 nm. MTAN concentrations were between 5 and 20 pM. For VcMTANV153I, the k_cat is reduced and higher concentrations of enzyme and inhibitor were used to compensate for lower catalytic activity. VcMTAN_V153I enzyme activity was unstable in the long assays required for slow-onset K_i measurements, and these values were not obtained.

In brief, ITC studies were conducted using a VP-ITC (MicroCal) microcalorimeter. Inhibitors were dissolved in the dialysate from protein preparation, and all solutions were filtered (Millipore, 0.2 μm) and degassed (Microcal Thermo-vac) for 15 min. The reference cell was filled with dialysate; 1.5 mL of enzyme was loaded into the sample cell, and 250 μL of the inhibitor solution was loaded into the injection syringe. Titration injection volumes were 5−8 μL. All titrations were conducted at 25°C. The ITC data were fit to the equations provided in Origin7, which describe two distinct independent binding sites. The K_i* values were used in eq 1 to generate ΔG. ΔH is the integral of the power needed to maintain the temperature of the sample cell at the same temperature as the reference cell after injection. The entropy term was determined using the Gibbs free energy equation (eq 2).

$$\mathrm{\Delta }G=RT\ln (K_{\mathrm{i}}\ast )$$ (1)

$$\mathrm{\Delta }G=\mathrm{\Delta }H-T\mathrm{\Delta }S$$ (2)

Titration of MTDIA to MTANs

[¹⁴C]MTDIA (MTDIA*) was incubated with the enzyme at varying MTDIA*:MTAN dimer ratios. The quantity of bound MTDIA* was measured after separation with a spin filter (Amicon) that retained enzyme and bound ligand and allowed unbound MTDIA* to pass through the filter. Initial and final MTDIA* concentrations were measured via liquid scintillation counting on a Tricarb 2910 TR instrument (PerkinElmer).

Kinetic Constants for ITC Analysis

ITC titrations for native EcMTAN and VcMTAN with MTDIA were reported previously and used K_i values based on K_M values obtained at high substrate concentrations, corresponding to second-site filling.⁶ Here we use the more sensitive luciferase assay at low substrate concentrations that reports on the K_M values at the first catalytic site. Thus, the kinetic and thermodynamic parameters for native EcMTAN and VcMTAN differ from those reported previously.

Crystallization of E. coli MTAN-BuTDIA, V. cholerae MTAN-MTDIA, and V. cholerae MTAN_TRIPLE-MTDIA

Enzymes were concentrated to 10 mg/mL and incubated with 2 equiv of the respective inhibitors. Both proteins were crystallized with the C-terminal TEV cleavage site and His₆ tag intact. All crystallization experiments were performed at 22°C using the sitting drop vapor diffusion technique. Rod-shaped crystals (0.02 mm × 0.03 mm × 0.27 mm) of EcMTAN-BuTDIA were obtained overnight using 0.2 M ammonium acetate, 0.1 M BIS-TRIS (pH 5.5), and 25% (w/v) polyethylene glycol 3350. Block-shaped crystals (0.04 mm × 0.16 mm × 0.28 mm) of VcMTAN-MTDIA were obtained over 2 weeks using 0.1 M diammonium hydrogen citrate and 15% (w/v) polyethylene glycol 3350. Rod-shaped crystals (0.03 mm × 0.03 mm × 0.35 mm) of VcMTAN_TRIPLE-MTDIA were obtained over a week using 0.2 M ammonium chloride (pH 6.3) and 20% (w/v) polyethylene glycol 3350.

Data Collection and Processing

The inhibitor-bound EcMTAN and VcMTAN crystals were transferred to a cryoprotectant solution containing 20% (v/v) glycerol and the respective mother liquors [80% (v/v)] prior to being flash-cooled in liquid nitrogen. Diffraction data for EcMTAN-BuTDIA and VcMTAN-MTDIA crystals were collected with 1.075 Å wavelength radiation at the X29A beamline (Brookhaven National Laboratory, Upton, NY) on an ADSC Q315 CCD X-ray area detector to 1.64 and 1.37 Å resolution, respectively. Diffraction data for VcMTAN_TRIPLE-MTDIA crystals were collected with 0.9793 Å wavelength radiation at the LRL-CAT beamline (Argonne National Laboratory, Argonne, IL) on a Rayonix 225 HE CCD detector to 1.50 Å resolution. Diffraction intensities were integrated and scaled with Denzo and Scalepack.¹⁷ The diffraction data statistics are summarized in Table 1.

Table 1.

Data Collection and Refinement Statistics for EcMTAN·BTDIA and VcMTAN·MTDIA

	EcMTAN-BTDIA	VcMTAN-MTDIA	VcMTANTRIPLE-MTDIA
Data Collection
space group	C2221	P21	P21
no. of molecules in the asymmetric unit cell dimensions	1	2	2
a, b, c (Å)	71.40, 91.39, 70.64	53.19, 72.00, 61.65	52.90, 72.71, 61.62
β (deg)	90	110.3	110.0
resolution (Å)a	50.0–1.64 (1.67–1.64)	50.0–1.37 (1.39–1.37)	50.0–1.50 (1.53–1.50)
no. of unique reflectionsa	28677 (1394)	85031 (4510)	69602 (3351)
Rmergea	0.083 (0.537)	0.048 (0.220)	0.082 (0.685)
I/σIa	32.9 (4.2)	27.3 (6.8)	18.7 (1.8)
completeness (%)a	99.9 (98.3)	93.0 (99.1)	99.1 (96.8)
Refinement
resolution (Å)	30.0–1.64	30.0–1.37	30.0–1.50
Rcryst	0.159	0.164	0.169
Rfree	0.184	0.183	0.187
rmsd
bond lengths (Å)	0.008	0.007	0.008
bond angles (deg)	1.37	1.37	1.40
no. of atoms (average B factor)
protein	1735 (17.6)	3579 (13.4)	3544 (17.1)
inhibitor	23 (13.6)	40 (9.1)	40 (12.4)
waters	169 (27.4)	431 (24.3)	360 (26.6)
other	13 (34.4)		10 (43.3)
PDB entry	4WKC	4WKB	4×24

^aNumbers in parentheses indicate values for the highest-resolution shell.

Structure Determination

The structures were determined by molecular replacement with Molrep,¹⁸ using a single monomer from the previously published structures of EcMTAN-MTDIA (PDB entry 1Y6Q) and VcMTAN-BuTDIA (PDB entry 3DP9) as search models. For all structures, the refinement of the initial solution and subsequent refinements were conducted using a restrained refinement performed with Refmac,¹⁹ using all data between 30.0 Å and the respective high-resolution cutoffs. Manual model rebuilding was conducted using Coot.²⁰ Difference Fourier maps calculated with F_obs − F_calc coefficients revealed ordered water molecules and strong unambiguous density difference, corresponding to the MTDIA and BuTDIA inhibitors. The coordinates for both inhibitors were present in the Coot monomer library, having monomer codes TDI for MTDIA and BIG for BuTDIA. Water molecules with proper hydrogen bonding coordination and electron densities greater than one rmsd and three rsmds in maps calculated with 2F_obs − F_calc and F_obs − F_calc coefficients, respectively, were included in the model. For the VcMTAN and VcMTAN_TRIPLE structures, all of the VcMTAN residues (residues 1−231) and six of the 13 residues in the cleavage/His₆ tag (residues 232−237) were sufficiently ordered to be included in the model. Likewise for the EcMTAN structure, all residues (residues 1−232) were included in the model; however, none of the residues in the cleavage/His₆ tag could be modeled. Several residues in each structure were modeled with alternate conformations. The EcMTAN-BuTDIA structure was refined to an R_cryst of 15.9% and an R_free of 18.4%, the VcMTAN-MTDIA structure to an R_cryst of 16.5% and an R_free of 18.3%, and the VcMTAN_TRIPLE-MTDIA structure to an R_cryst of 16.9% and an R_free of 18.7%. Analysis of the structures in Coot revealed good stereochemistry with no residues falling into the disallowed regions of the Ramachandran plot in either of the VcMTAN structures and one outlier in the EcMTAN structure (Ser155). The refinement statistics are listed in Table 1.

RESULTS

Structural Similarity between VcMTAN and EcMTAN

The crystal structures reported here permit direct comparison of the Ec- and VcMTANs bound to the transition-state inhibitors MTDIA and BuTDIA. All four inhibitor-bound proteins are structurally similar. The overall rmsd between an all-atom comparison of VcMTAN and EcMTAN in complex with MTDIA is 0.57 Å (2541 atoms accepted for comparison), and the same comparison for the complexes with BuTDIA is 0.61 Å (2495 atoms accepted for comparison). The overall rmsd between an all-atom comparison of VcMTANTRIPLE and EcMTAN in complex with MTDIA is 0.54 Å (2539 atoms). The conformations of residues at positions 113, 153, and 158 in VcMTAN and analogous positions in EcMTAN are highlighted in Figure 1, where VcMTAN residues (blue) are overlaid with EcMTAN residues (green). The transition-state analogues are colored according to the active site in which they are bound.

The hydrogen bonding distances and hydrophobic interactions to bound inhibitors are similar between VcMTAN and EcMTAN (Figure S3 of the Supporting Information). Five key hydrogen bonds to the inhibitors are found in all five structures: Ile152/Val153 to N1 and N6, Asp197/198 to N6 and N7, and inhibitor N1′ to the nucleophilic water molecule. Additionally, Phe151/152 participates in π-stacking interactions with the deazapurine moiety, having centroid−centroid distances of ∼4.4 ± 0.2 Å (Figure S4 of the Supporting Information). A related interaction is observed in purine nucleoside phosphorylase (PNP), where Phe200 shares a related interaction with 9-deazahypoxanthine from transition-state analogues bound in the catalytic sites.^21,22

Hydrophobic interactions involving the 5′-alkyl chains of MTDIA and BuTDIA are similar in all five structures, despite the increased steric bulk of Pro113 in VcMTAN compared to that of Ala113 in EcMTAN. The distances between hydrophobic interactions vary between 4.1 and 4.9 Å in the BuTDIA-bound structures and between 4.5 and 4.7 Å in the MTDIA-bound structures (Figure S3 of the Supporting Information). Overall, the position and conformations of MTDIA and BuTDIA bound to Ec- and VcMTAN are nearly superimposable with respect to their binding sites, with only the 5′-butyl chains varying in orientation (Figure S6 of the Supporting Information).

As mentioned above, Arg158 is near Val153 in VcMTAN, whereas Gly157 and Ile152 form this comparable pair in EcMTAN. Static modeling of the respective structures suggested that VcMTANV153I might have side chain clashes with Arg158 (Figure S4 of the Supporting Information). Given this observation and the large change in side chain volume, a Arg158Gly mutant was also investigated in VcMTAN.

Steady-State Kinetic Properties

The sensitivity of the luminescence assay for MTAN activity permitted determination of the nanomolar K_M value for first-catalytic site filling in MTANs (Table 2 and Figure S7 of the Supporting Information). The affinity of VcMTAN for MTA (61 nM) is greater than the affinity of EcMTAN for MTA (139 nM), while the catalytic efficiencies are similar at 2.3 or 3.0 × 10⁶ M⁻¹ s⁻¹ (Table 2). The K_M values for VcMTAN_A113P and VcMTAN_A113P/V153I (VcMTAN_DOUBLE) mutants are similar (262 and 202 nM, respectively) but are increased to 430 nM for VcMTAN_V153I. The k_cat values were unchanged in VcMTAN_A113P and decreased 22-fold in VcMTAN_V153I (Table 2). In VcMTAN_DOUBLE, a fraction of the substrate affinity was restored (improved K_M) and the catalytic efficiency improved 20-fold compared to that of VcMTAN_V153I. VcMTAN_{A113P/V153I/R158G} (VcMTAN_TRIPLE) most closely resembles the catalytic site of EcMTAN, yet its K_M value is significantly increased. None of the VcMTANs mutated toward EcMTAN displayed catalytic efficiencies approaching that of EcMTAN.

Table 2.

Kinetic Constants for VcMTAN and Mutants Determined Using the Luminescence-Based Assay

MTAN	KM (nM)	kcat (s−1)	kcat/KM (×106 M−1 s−1)	Ki,MTDIA (pM)	Ki,BuTDIA (pM)	Ki (MTDIA/BuTDIA)
VcMTAN
WT	61 ± 6	0.18 ± 0.01	3.0 ± 0.3	2.7 ± 0.2	0.21 ± 0.04	13
A113P	262 ± 29	0.17 ± 0.01	0.7 ± 0.1	62 ± 7	2.4 ± 0.5	26
V153I	430 ± 47	(7.9 ± 4) × 10−3	(1.8 ± 2) × 10−2	NAa	NAa	NAa
DOUBLE	202 ± 23	(7.4 ± 3) × 10−2	0.4 ± 0.1	3.0 ± 0.5	10.3 ± 1.2	0.3
TRIPLE	536 ± 122	0.11 ± 0.01	0.2 ± 0.1	166 ± 22	3.3 ± 0.5	50
EcMTAN	139 ± 18	0.32 ± 0.01	2.3 ± 0.3	0.16 ± 0.02	0.20 ± 0.04	0.8

^aNot available.

Dissociation Constants for Transition-State Analogues with EcMTAN and VcMTANs

Single and combined mutations in VcMTAN at positions 113, 153, and 158 altered the dissociation constants for MTDIA and BuTDIA (Table 2). The K_i value for MTDIA is 6 times larger for VcMTAN than for EcMTAN, and the K_i values for BuTDIA are similar for both enzymes. Increasing the 5′-alkyl chain length from methyl (MTDIA) to butyl (BuTDIA) has little effect on EcMTAN but results in a 13-fold change for VcMTAN. VcMTAN_DOUBLE has a K_i similar to that of the native enzyme for MTDIA, despite its reduced K_M value. The K_i values for MTDIA are increased in the A113P mutant and further increased in the triple mutant. VcMTAN mutants exhibited increased K_i values for BuTDIA in all mutants tested compared to those of its parent.

A Conservative Active Site Mutation Affects Cooperativity

Catalytic site titrations of VcMTANs with MTDIA reveal that both active sites of VcMTAN and VcMTAN_A113P must be filled with inhibitor to cause complete catalytic inhibition (Figure 2). In contrast, inhibitor bound to one active site of either VcMTAN_V153I or VcMTAN_DOUBLE causes complete inhibition (Figure 2). Native and mutant VcMTANs migrated the same on nondenaturing gels (Figure S8 of the Supporting Information). Thus, altered quaternary structure in either VcMTAN_DOUBLE or VcMTAN_V153I is unlikely to account for the half-the-sites inhibition. Complete inhibition of dimeric mutant MTANs by inhibitor binding to a single monomer raises the question of inhibitor binding to the second catalytic site. MTDIA* titrations sufficient to saturate one or two catalytic sites of VcMTAN and VcMTAN_A113P showed complete binding at the first site and near-stoichiometric binding at the second site. Thus, both sites bind inhibitor in stoichiometric titrations, but binding to the second active site is weaker than binding to the first (Figure 3). In VcMTAN_DOUBLE and VcMTAN_V153I, titration to two MTDIA* molecules per dimer left more MTDIA* unbound as both the first and second sites were titrated (Figure 3). VcMTAN_DOUBLE and VcMTAN_V153I bind inhibitor less well at both the first active and second catalytic sites than VcMTAN or VcMTAN_A113P.

Figure 2.

Fractional activity of the MTAN homodimer following titration with the tight binding inhibitor MTDIA. Wild-type VcMTAN and VcMTANA113P require two inhibitor molecules to be bound per homodimer to abolish MTAN activity. VcMTAN_V153I and VcMTAN_A113P/V153I require one inhibitor molecule bound per homodimer to abolish activity.

Figure 3.

Fraction of unbound radiolabeled MTDIA as a function of dimer catalytic site titration in native (WT) or mutated VcMTANs (A113P, V153I, and DOUBLE). Fraction of MTDIA unbound (y-axis) following titration of the MTAN dimer with varying concentrations of MTDIA (x-axis). An MTDIA:MTAN dimer value of 1 indicates sufficient inhibitor to fill one of the two catalytic sites. An MTDIA:MTAN dimer value of 2 indicates an inhibitor stoichiometry that can fill both catalytic sites, if the affinity is high. MTAN dimer concentrations were 27−33 μM for all experiments except for WT, for which the concentrations were 27−85 μM.

Isothermal Calorimetry of the Transition-State Analogue Binding to Mutant VcMTANs

Isothermal calorimetry (ITC) experiments measure distinct energetic contributions to inhibitor binding. Enthalpic changes occurred as both catalytic sites were filled in VcMTAN and VcMTAN_A113P, but only for first-site filling with V153I, DOUBLE, and TRIPLE mutants (Table 3). Plotting the isotherms together emphasizes the differences in the energetics of binding to the VcMTANs (Figure 4B).

Table 3.

Thermodynamic Constants for Binding of MTDIA to Wild-Type and Mutant VcMTANs

MTAN + MTDIA titrant pair	sitea	ΔGb (kcal mol−1)	ΔHc (kcal mol−1)	TΔSd (kcal mol−1)
Vc-WT	first	−15.8 ± 0.1	−6.0 ± 0.2	−9.8 ± 0.2
	second	−14.3 ± 0.2	−2.6 ± 0.3	−11.7 ± 0.4
Vc-A113P	first	−13.9 ± 0.1	−2.9 ± 0.1	−11.0 ± 0.1
	second	−13.5 ± 0.8	−1.8 ± 0.2	−11.7 ± 0.8
Vc-V153I	single	−10.6 ± 0.2	−4.9 ± 0.1	−5.7 ± 0.2
Vc-DOUBLE	single	−15.7 ± 0.1	−3.9 ± 0.1	−11.8 ± 0.1
Vc-TRIPLE	single	−13.3 ± 0.1	−4.8 ± 0.2	−8.5 ± 0.2
Ec-WT	first	−17.4 ± 0.9	−18.9 ± 0.2	1.5 ± 0.9
	second	−10.2 ± 0.1	−16.8 ± 0.2	6.6 ± 0.2

^aSite represents which dimer site contributed most to the indicated enthalpic and associated thermodynamic properties using the best fit described in Materials and Methods (each monomer of the dimer has a single active site). Single indicates only the first site has an enthalpic signature as detected by ITC.

^bΔG was calculated from the K_i* (Table 1) using the equation ΔG = RT ln K_i*.

^cΔH was determined from the ITC titrations of enzyme with inhibitor for each of the two catalytic site of the enzyme dimer.

^d−TΔS was calculated from the Gibbs free energy equation.

Figure 4.

Comparison of heat isotherms for titration of MTDIA against VcMTAN (WT and mutants) (top six panels). In the bottom panel, the energetics are shown on a common scale to emphasize the difference between the systems.

The enthalpic isotherm generated for binding of MTDIA to VcMTAN indicated negative cooperativity. As inhibitor filled the second catalytic site the heat released by the system decreased (Figure 4A). Isotherms for VcMTAN-MTDIA and VcMTAN_A113P-MTDIA are similar and end with an abrupt loss of heat production when catalytic site stoichiometry is reached, as expected for tight-binding inhibitors. The enthalpy generated in VcMTAN_A113P is 3 kcal mol⁻¹ less favorable than the enthalpy generated from its parent. The enthalpies generated by VcMTAN_DOUBLE, VcMTAN_TRIPLE, and VcMTAN_V153I are more favorable than that of VcMTAN_A113P by −1.0, −1.9, and −2.0 kcal mol⁻¹, respectively. Thus, all VcMTAN catalytic site changes toward EcMTAN resulted in unfavorable enthalpy contributions to MTDIA binding compared to that of native VcMTAN (Figure 5).

Figure 5.

Representation of the ΔH and −TΔS values for binding of MTDIA to EcMTAN, VcMTAN, and VcMTAN mutants at 25 °C. The gray dotted arrow indicates the direction either ΔH or −TΔS must shift for the energetic binding profile of the VcMTAN mutants to more closely match that of EcMTAN.

In contrast to the unfavorable enthalpic changes, VcMTAN_A113P and VcMTAN_DOUBLE displayed favorable entropic gains while VcMTAN_V153I and VcMTAN_TRIPLE displayed entropic penalties compared to native VcMTAN (Figure 5). The entropic gains with VcMTAN_A113P and VcMTAN_DOUBLE were overcome by an even larger enthalpic penalty compared to that of VcMTAN.

Potential Origins for Altered Thermodynamic Parameters

We explored if the changes in binding thermodynamics might be attributed to alterations in the active site water reorganization or hydrophobicity due to altered active site volumes. Comparison of the cavities of VcMTAN, EcMTAN, and VcMTANTRIPLE indicated that the cavity volumes of VcMTAN_TRIPLE at the active sites (both A and B subunits) remained unchanged, while the cavity volume located at the dimer interface increased in VcMTAN_TRIPLE (Figure S9 of the Supporting Information). The cavity volumes of active site A and active site B are similar between VcMTAN and VcMTAN_TRIPLE with a <3% variation in volume. The active site volumes of VcMTAN_TRIPLE are 18% (monomer A) and 6% (monomer B) smaller than that of EcMTAN (Table 4). The most striking difference occurs at the dimer interface of VcMTAN_TRIPLE, where the cavity volumes are increased 1.9-and 2.2-fold relative to those of VcMTAN and EcMTAN, respectively (Table 4). This doubling of cavity volume at the dimer interface in VcMTAN_TRIPLE suggests a possible mechanism for the disruption of catalytic site cooperativity. The similar catalytic site volumes make it less likely that changes in solvent access to the catalytic sites are responsible for altered thermodynamics.

Table 4.

Cavity Volumes of VcMTAN_TRIPLE, VcMTAN, and EcMTAN at Monomer A, Monomer B, and the Dimer Interface^a

	volume (Å3)
	monomer A	monomer B	dimer interface
VcMTANTRIPLE	185.0	185.3	1028.4
VcMTAN	180.4	181.0	543.2
EcMTAN	224.3	197.4	478.2

^aVolumes calculated using a 2.5 Å query diameter on CASTp.

DISCUSSION

Mutations toward Catalytic Site Similarity

Attempts to convert the catalytic site of VcMTAN in terms of sequence and structural identity to those of EcMTAN led to catalytic dysfunction. The Val153Ile mutation caused VcMTAN to acquire half-the-sites inhibitor susceptibility. Thus, the Val153 region is involved in facilitating catalytic communication between VcMTAN monomers. Although large effects due to conservative mutations are surprising, they are documented in other systems, including dihydrofolate reductase (DHFR). E. coli DHFR (EcDHFR) catalyzes the reduction of 7,8-dihydrofolate (DHF) to 5,6,7,8-tetrahydrofolate (THF). In the active site, a conservative residue change of Asp27Glu, to mimic the Glu at this position in the human enzyme (Glu31), affects catalysis, producing a similar k_cat; however, the KM(DHF) is increased 20-fold, and methotrexate affinity is decreased 176-fold.^23,24 Asp27 exhibits this effect through an intervening water molecule required for proton transfer.^25,26 The presence of Glu31 in hDHFR is proposed to alter enzyme dynamics, which is linked to differences in catalytic properties and metabolic flux between EcDHFR and hDHFR.²⁷

Possible Path for Cooperativity

The Val153Ile mutation in the VcMTANs gives rise to half-the-sites inhibition, with inhibitor binding at one site inhibiting catalysis at both sites. The protein connectivity among the catalytic sites of EcMTAN, VcMTAN, and VcMTAN_TRIPLE by the most direct route includes His98, Asp99, Gly149, Asp150, Ala151, and Phe152 (yellow) and to a lesser extent Ala100, Asp101, and Val102 (Figure 6, blue). These residues are conserved between the Ec- and VcMTANs and are highly similar in all of the structures, with a rmsd for the Cα peptide atoms of 0.162 Å between the Ec- and VcMTANs and 0.143 Å between Ec- and VcMTAN_TRIPLE. From this analysis, there is no obvious structural reason for the change from independent active sites to half-the-sites reactivity. However, the kinetic analysis makes it clear that in the Val153Ile mutant, the second site is rendered catalytically nonfunctional by inhibitor binding at only one site.

Figure 6.

Stick figure of EcMTAN, VcMTAN, and VcMTAN_TRIPLE MTANs with MTDIA bound, illustrating connectivity between active sites. The transition-state analogue MTDIA is colored in magenta. The residues at position 153 are colored green. Amino acids in a direct route between active sites are colored yellow, whereas those more distal are colored blue.

Mutations in Groups

VcMTAN_TRIPLE is the mutant most closely related to EcMTAN, but the changes resulting in a catalytic site that is more similar to that of EcMTAN cause K_M, k_cat, and inhibitor binding to be less efficient than those of the parent VcMTAN and make it even less like EcMTAN. These changes occur without any significant changes in the overall protein structure of the triple mutant. Although catalytic site function is impaired in VcMTAN_V153I, the combination of Val153Ile with Ala113Pro partly restores inhibitor and substrate binding. However, this second mutation is insufficient to restore catalytic independence at the active sites or improve the catalytic properties toward those of EcMTAN.

With VcMTAN_V153I as the reference point, changes toward an active site that is more similar to the active site of EcMTAN result in enthalpic gains in inhibitor binding but are offset by larger entropic penalties. The evolution of the Ala113-Val153-Arg158 sites in VcMTAN appears to be group-dependent. Thus, perturbation of any member of the group results in a loss of catalytic fitness. Execution of a Basic Local Alignment Search Tool (BLAST) with wild-type VcMTAN as the query sequence indicates that the Ala113-Val153-Arg158 sites and Ala-Val-Xxx (Xxx represents a variable amino acid) sites are exclusively present in the family Vibrionaceae. In contrast, the corresponding sites are Pro-Ile-Gly in EcMTAN, and this sequence is present exclusively in the Enterobacteriaceae family, with its members belonging to the genera Escherichia, Salmonella, and Citrobacter (Figure S10 of the Supporting Information).²⁸ The corresponding Pro-Ile-Xxx residues are found across multiple species of the Gram-positive firmicutes and the Gram-negative α-proteobacteria, δ-proteobacteria, ɛ-proteobacteria, and γ-proteobacteria (Figure S10 of the Supporting Information). The Ala-Val-Xxx and Pro-Ile-Xxx groups apparently co-evolved with a complementary protein scaffold. Simply altering the catalytic site groups in VcMTAN without altering the protein architectural scaffold causes a decrease in catalytic and binding functions.

Similar Overall Structures between Enzymes

Comparing complexes of MTDIA or BuTDIA bound to either MTAN shows that different inhibitors cause small rmsds in the complexes, 0.35 and 0.21 Å for EcMTAN and VcMTAN, respectively. Additionally, the rmsds between the EcMTAN and VcMTAN proteins bound to MTDIA or BuTDIA are also small, 0.57 and 0.60 Å, respectively. Taken together, significantly different conformational geometries are unlikely to account for differences in inhibitor affinity for these enzymes. The active site of the VcMTAN_TRIPLE mutant is nearly identical to that of EcMTAN (EcMTAN and VcMTAN_TRIPLE bound to MTDIA shown in Figure 7). Bound MTDIA overlaps from structures in both enzymes, and both show geometry consistent with the sp³-hybridized N1′ that acts as the cationic mimic of the C1′ ribocation of the transition state. A single exception to the close structural similarity occurs at Phe105 with a 30° rotation of the phenyl ring. Phe105 is found on the loop of the adjacent monomer, which occludes the active site entrance and might play a role in stabilizing the inhibitor deazapurine ring in the active site (Figure 7). The absence of active site differences between VcMTAN_TRIPLE and EcMTAN therefore implicates whole protein dynamic contributions to account for the differences in catalytic properties.

Figure 7.

Structural comparison (stereoview) of the active sites of EcMTAN (green) vs VcMTAN_TRIPLE (gray) with transition-state analogue MTDIA and a nucleophilic water molecule bound.

Enzyme Flexibility: Perturbations in Short Time Scale Motions

Molecular dynamics (MD) simulations of apo and BuTDIA-bound VcMTAN and EcMTAN have suggested that the EcMTAN complex retains more Cα dynamic motion than the VcMTAN complex.¹¹ The increased flexibility was invoked to explain differential inhibitor affinities of BuTDIA for EcMTAN and VcMTAN, as retained system dynamics upon inhibitor binding decreases entropic loss in the bound complex. The average total numbers of hydrogen bonds between VcMTAN and EcMTAN and their bound BuTDIA ligand were similar at 12.85 and 12.45, respectively, suggesting system dynamic contributions beyond catalytic site contacts.¹¹

Other Examples of Catalytic Site Connectivity

Examples in which modifications to regions distinct from the active site affect catalytic properties are commonplace in enzymology. In one such example, exhaustive mutagenesis performed in a 110-residue ribonuclease barnase indicated that substitutions at only 15 of 109 positions were not well tolerated. Furthermore, 12 of those 15 critical substitution-sensitive positions did not directly interact with the substrate.²⁷ In another study, a hybrid β-lactamase (TEM-1), which differed from the parent only at the C-terminus, resulted in an inactive protein despite the C-terminus being distinct from the active site and poorly conserved.²⁸ These examples provide support that residues apart from the active site can be deterministic in key aspects of the catalytic cycle.

Ec- and VcMTANs possess similar overall protein architecture but small catalytic site differences. By mutationally matching the catalytic site sequences of VcMTAN to that of EcMTAN, we attempted to equalize catalytic site function, but the catalytic properties became even more disparate. The results establish the significance of remote interactions to create catalytic function at sites with the same crystallographic contacts with reactants. The evolutionary distance between Ec- and VcMTANs has resulted in drift in both the catalytic sites and the remote protein architecture, working together to optimize the protein architecture−catalytic site interplay. Altering the catalytic site without the overall dynamic architecture solves only part of the evolutionary difference. Clearly, catalytic site sequence design alone will be insufficient to design and engineer efficient enzymatic catalysts. Incorpo ration of appropriate catalytic site functionality requires the appropriate dynamic architecture.²⁹⁻³¹

CONCLUSIONS

VcMTAN_DOUBLE and EcMTAN share 100% active site residue identity, yet considerable differences in kinetic properties were observed. There is a 6-fold decrease in catalytic efficiency and a 20−50-fold decrease in transition-state analogue affinities in VcMTAN_DOUBLE, despite having matched all inhibitor-interacting active site residues to those identified in EcMTAN. The VcMTAN_DOUBLE mutant also became more dissimilar to EcMTAN in thermodynamic terms. Although VcMTAN_DOUBLE did not become more similar to EcMTAN, it displayed improved affinity for MTA compared to that of VcMTAN_V153I and displayed improved binding to the transition-state analogues relative to the binding of either of the single mutants (VcMTAN_V153I or VcMTAN_A113P).

The presence of the Val153Ile mutation disrupted catalysis at the second active site of the homodimer when the first site was inhibited. Val153 is not located at the dimer interface, nor is it sufficiently close to another residue from the second monomer to exert direct influence on its neighbor. Val153 is likely to be involved in a network of residues involved in facilitating conformational interactions between monomers.

Alterations remote from the catalytic site impose strong catalytic function on similar catalytic sites and thus make design of de novo enzymes a difficult goal. Dynamic motions remote from the catalytic site are likely responsible for differences in binding and catalytic properties of the MTANs investigated here. Full protein scaffold dynamical motions are critical for optimized functioning of the MTANs and likely other enzymes. The results reported here, and related work, add growing support to the position that dynamical motions are integral to catalysis.^29,32−34

ABBREVIATIONS

MTAN

5′-methylthioadenosine/S-adenosyl-L-homocysteine nucleosidase

MTA

5′-methylthioadenosine

MTR

5′-methythioribose

VcMTAN

V. cholerae MTAN

EcMTAN

E. coli MTAN

DADMe-ImmA

4′-deaza-1′-aza-2′-deoxy-1′-(9-methylene)-Immucillin-A

rmsd

root-mean-square deviation