MutT from the fish pathogen Aliivibrio salmonicida is a cold-active nucleotide-pool sanitization enzyme with unexpectedly high thermostability.

Upon infection by pathogenic bacteria, production of reactive oxygen species (ROS) is part of the host organism's first line of defence. ROS damage a number of macromolecules, and in order to withstand such a harsh environment, the bacteria need to have well-functioning ROS scavenging and repair systems. Herein, MutT is an important nucleotide-pool sanitization enzyme, which degrades 8-oxo-dGTP and thus prevents it from being incorporated into DNA. In this context, we have performed a comparative biochemical and structural analysis of MutT from the fish pathogen Aliivibrio salmonicida (AsMutT) and the human pathogen Vibrio cholerae (VcMutT), in order to analyse their function as nucleotide sanitization enzymes and also determine possible cold-adapted properties of AsMutT. The biochemical characterisation revealed that both enzymes possess activity towards the 8-oxo-dGTP substrate, and that AsMutT has a higher catalytic efficiency than VcMutT at all temperatures studied. Calculations based on the biochemical data also revealed a lower activation energy (E a) for AsMutT compared to VcMutT, and differential scanning calorimetry experiments showed that AsMutT displayed an unexpected higher melting temperature (T m) value than VcMutT. A comparative analysis of the crystal structure of VcMutT, determined to 2.42 Å resolution, and homology models of AsMutT indicate that three unique Gly residues in loops of VcMutT, and additional long range ion-pairs in AsMutT could explain the difference in temperature stability of the two enzymes. We conclude that AsMutT is a stable, cold-active enzyme with high catalytic efficiency and reduced E a, compared to the mesophilic VcMutT.

Introduction

When microbes attack a host, phagocytes produce reactive oxygen species (ROS) as a first line of defence [1-3]. These, together with oxygen radicals generated through normal cellular metabolism or by ionising radiations, lead to oxidative stress in the bacterial cell. A common damage generated by ROS is the oxidation of guanine, resulting in the generation of 8-oxoG in DNA or 8-oxo-dGTP in the nucleotide pool, which later can be incorporated in DNA (Fig. 1) [4,5]. 8-oxoG base pairs with both adenine and cytosine in DNA, which consequently can induce A:T to C:G and G:C to T:A mutations that potentially are lethal to the organism [4,5]. Therefore, organisms possess efficient ROS scavenging and repair systems for preventing mutations [6].

Fig. 1

A common damage generated by ROS is the oxidation of guanine, and it is formed in DNA either by direct guanine oxidation, or by incorporation of 8-oxoG from the nucleotide pool. MutT catalyses the hydrolysis of 8-oxo-GTP by nucleophilic substitution at the β phosphorus site, yielding 8-oxo-dGMP and inorganic pyrophosphate (PPi). When 8-oxoG is incorporated in DNA against adenine or cytosine, the organism is equipped with a large mechanism system to counteract the mutagenic effects of 8-oxoG. Here, the glycosylases from base excision repair pathway MutM and MutY function to prevent T:A to G:C and G:C to T:A mutations respectively.

The mutator enzyme MutT hydrolyses the 8-oxoG containing deoxyribose or ribonucleoside triphosphates to their corresponding monophosphates and inorganic pyrophosphate (PPi) and thus plays an important role in the cell’s sanitization system to prevent incorporation of 8-oxo-dGTP into DNA (Fig. 1) [4,5]. In vitro experiments have shown that 8-oxo-dGTP is the best substrate for Escherichia coli MutT (EcMutT) with a Km value of 0.081 μM that is 13,500 fold lower than that of dGTP (1100 μM) [7]. It was also shown that the enzyme has high affinity for the reaction product 8-oxo-dGMP with a KD of 52 nM that is 34,600 fold lower than that of dGMP (1.8 mM) [8]. The tight binding of the reaction product was later confirmed by determination of both the solution and X-ray crystal structures of EcMutT in complex with 8-oxo-dGMP [9,10].

The MutT enzyme is widespread in nature and representatives are found in eukaryotes, prokaryotes, archaea and viruses [11]. It belongs to a superfamily of enzymes shown to require two divalent cations for activity [12,13]. Members of this family are recognised by a highly conserved 23-residue motif, or Nudix box, GX5EX7REUXEEXGU, where U is a bulky hydrophobic or aliphatic residue and X is any residue [14,15]. The Nudix box forms a loop-α-helix-loop structural motif [16] that constitutes the active site.

Structures of MutT enzymes have been determined from various sources. EcMutT was the first Nudix hydrolase to have its structure determined and mechanism studied in detail [9,10]. The structure of human MutT (MTH1) was solved by NMR [17], and although EcMutT and MTH1 share poor sequence similarity outside the conserved Nudix motif, they adopt a similar globular and compact structure with the parallel portion of a β-sheet sandwiched between two α-helices, forming an α + β fold.

The interest in studying MutT enzymes have drastically increased based on recent findings that inhibitors of MTH1 can be used as anti-cancer drugs [18,19]. By targeting MTH1 with inhibitors the cancer cells will not be able to clean up the oxidised nucleotide pool and thus increase the cancer cell’s sensitisation to cancer treatment. This finding also opens for investigation of use of MutT inhibitors for treatment of infections by pathogenic bacteria as an alternative to antibiotics.

Enzymes from cold-adapted organisms are characterised by an increased catalytic efficiency, lower temperature optimum for activity, lower temperature stability and reduced activation energy (Ea) compared to their mesophilic homologues (reviewed in [20]). It is suggested that these features originate from an increased structural flexibility [21], however the underlying molecular mechanisms for the cold-adapted properties are still not fully understood.

Here we have studied MutT from the psychrophilic fish pathogen Aliivibrio salmonicida (AsMutT) and the mesophilic human pathogen Vibrio cholerae (VcMutT) in order to obtain more information regarding their function as nucleotide sanitization enzymes and to study cold-adapted properties of AsMutT, through enzyme kinetic measurements, differential scanning calorimetry, X-ray structure determination and homology modelling. Our findings show that AsMutT possesses cold-adapted properties with a higher catalytic efficiency than VcMutT mainly caused by higher kcat values, lower Km, and lower Ea due to lower activation enthalpy (ΔH#). However AsMutT exhibits an unexpected higher overall stability than VcMutT. The comparative analysis of the crystal structure of VcMutT, determined to 2.42 Å resolution, and homology models of AsMutT shows that VcMutT possesses additional Gly residues in dynamic loops and AsMutT possesses additional long range ion-pairs which could explain the difference in stability. We conclude that both AsMutT and VcMutT are functional MutT enzymes within their respective organisms, and that AsMutT is able to function well in its natural cold environment due to optimised substrate interactions (lower Km) and high substrate turnover (higher kcat).

Materials and methods

Cloning

The genes encoding AsMutT (VSAL_I2636) and VcMutT (VC_2392) were inserted into the pDest14 expression vector using Gateway technology following the manufacturer’s instructions (Invitrogen), as described previously for Deinococcus radioduransuracil-DNA N-glycosylase [19]. The constructs contained nucleotides encoding a 6 × His-tag immediately upstream of the gene using the following primers (Sigma–genosys): FAsMutT: 5′-CAT CAC CAT CAC CAT CAC AAG CGT GTA CAT ATT GTG-3′, RAsMutT: 5′-GGG GAC CAC TTT GTA CAA GAA AGC TGG GTC TTA TCT TAA GTA ACA TGT CAT CAC-3′, FVcMutT: 5′-CAT CAC CAT CAC CAT CAC TTG AAA CGC ATC CAC ATT GTT GCA G-3′, RVcMutT: 5′-GGG GAC CAC TTT GTA CAA GAA AGC TGG GTC TTA ACC AAA TTG AGC AAT CAC TTG C-3′, FDRHISTAG: 5′-G GGG ACA AGT TTG TAC AAA AAA GCA GGC TTC GAA GAT AGA ACC ATG CAT CAC CAT CAC CAT CAC-3′. The genes were first amplified using the primers FAs/FVcMutT and RAs/RVcMutT in order to engineer the N-terminal 6 × His-tag (underlined). The resulting gene products were thereafter amplified using the FDRHISTAG (containing the attB1 site and nucleotides encoding the His-tag) and the RAs/RVcMutT primers and used in a BP reaction along with the pDONR201 vector, and in a LR reaction along with the pDEST14 vector. The sequence of the clones was confirmed by DNA sequencing (ABI 377 DNA sequencer, GE Healthcare).

Expression and purification

BL21 AI competent cells (Invitrogen) transformed with the pDest14 vector containing the genes encoding AsMutT and VcMutT were grown at 37 °C in 1 L Luria–Bertani broth containing 100 μg mL−1 ampicillin and 0.1% glucose until the optical density at 600 nm (OD600) reached 0.7. The temperature was set to 20 °C before expression was induced by 0.2% l-arabinose. The cells were harvested 16 h post-induction at 5000 rpm and resuspended in buffer A (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 10 mM MgCl2, and 5 mM 2-mercaptoethanol). Protease inhibitors (Roche) were added prior to cell disruption by sonication (Vibra-Cell VCX 750, IDE) and the lysate was cleared by centrifugation at 13,000 rpm for 20 min. The resulting protein extract was loaded onto a 1 mL HisTrap HP column (GE Healthcare) equilibrated with buffer A. The column was washed with 5% buffer B (Buffer A supplemented with 500 mM imidazole) to remove weakly bound proteins before the bound proteins were eluted using a 20 mL gradient to 100% buffer B. MutT containing fractions were further loaded onto a Superdex 75 10/300 GL column (GE Healthcare) using buffer A. Protein purity was determined by analysis on 4–20% SDS–PAGE (Pierce) stained with Simply Blue SafeStain (Invitrogen). The identity of the protein was confirmed using mass spectrometry (Waters® Micromass® MALDI micro MX).

Activity assay and enzyme kinetic measurements

The activity of MutT was monitored using an assay adapted from Svensson et al. [22]. By mixing 8-oxo-dGTP [10–50 μM] (Jena Bioscience), AsMutT [64 nM] or VcMutT [30 nM], enzyme buffer (250 mM Tris–HCl pH 7.5, 750 mM NaCl and 50 mM MgCl2) and pyrophosphatase [0.08 U] (Sigma) in a total volume of 100 μL, we were able to detect free phosphate in solution by adding malachite green working solution containing 10 parts MG-stock prepared according to Baykov et al. [23], 2.5 parts 8% ammonium molybdate (Sigma), and 0.2 parts 11% Tween-20 solution. Here a conjugate between malachite green, molybdate and phosphate makes it possible to measure absorbance at 630 nm (OD630) using Spectramax M2 (Molecular Devices). All measurements were performed in triplicate. OD630 values were converted to phosphate concentration for each sample using a standard curve established using various known phosphate concentrations. Vmax, kcat and Km values were determined by fitting the data points to the Michaelis–Menten equation using non-linear regression analysis in GraphPad Prism Software (version 6.0d).

Thermodynamic activation parameters were calculated using the following equations [21]:where ΔG# is the free energy of activation, ΔH# is the activation enthalpy, ΔS# is the activation entropy, R is the gas constant (8.314 J mol−1 K−1), kB is the Boltzmann constant (1.3805 × 10−23 J K−1), h the Planck constant (6.6256 × 10−34 J s) and T the temperature in kelvin.

To determine the Ea of AsMutT and VcMutT, the enzymes (AsMutT at 64 nM and VcMutT at 30 nM) were incubated with nine different substrate concentrations of 8-oxo-dGTP (Jena Bioscience) ranging from 10 to 50 μM at 4, 11, 18, 25, 32, and 37 °C in a total reaction volume of 100 μL (as described above). From these results we could calculate the Ea, by taking the natural logarithm of the Arrhenius equation (Eq. (4)) and plotting a linear regression curve (ln kcat as a function of 1/T). The Ea value can be derived from the slope of this curve (−Ea/R).

Stability measurements by DSC

DSC experiments were executed on a Nano-Differential Scanning Calorimeter III, model CSC6300 from Calorimetry Sciences Corporation (Lindon, USA). The enzymes were prepared by dialysis over night at 4 °C in a Pierce Slide-A-Lyzer dialysis cassette from Thermo Fisher Scientific Inc (Rockford, USA) with a 10 kDa cut-off, against 1 L of dialysis buffer containing 25 mM Hepes pH 7.5, 150 mM NaCl, 10 mM MgCl2, and 5 mM 2-mercaptoethanol. The enzymes were further filtered using a 0.2 μm syringe filter (Millipore, Billerica, USA), and diluted in dialysis buffer to a final enzyme concentration of 1 mg mL−1. The dialysis buffer was used as reference buffer, and like the enzyme, it was degassed for approximately 15 min. before being loaded into the reference and sample DSC cells, respectively. The scans were executed between 10 and 80 °C at heating and cooling rates of 1 °C min−1 at a constant pressure of 3 atm. The experiments were carried out in triplicates.

Crystallization and structure determination of VcMutT

VcMutT was crystallized with reservoir solution containing 0.2 M lithium sulphate, 0.1 M Tris–HCl pH 8.5, and 1.26 M ammonium sulphate by the hanging drop method, and the drops were made by mixing 1 μl 10 mg mL−1 protein and 1 μL reservoir solution. The quartz-shaped crystals were cryoprotected with reservoir solution supplemented with 12.5% ethylene glycol and flash-frozen in liquid nitrogen.

X-ray data was collected at the European Synchrotron Radiation Facility (ESRF), France, on the ID14-EH1 beamline with a crystal to detector distance of 254 mm, 1.5 s exposure per frame, 0.5° oscillation, 3 passes and the final dataset included 55° of data. The images were integrated and scaled with XDS [24] and the structure factors were obtained with Truncate in the CCP4 suite [25]. The crystal belonged to space-group P41212 or P43212 with cell axis of a = b = 81.15 Å and c = 90.57 Å. The structure was solved by molecular replacement, using the VcMutT sequence and the EcMutT-Mn2+ bound holo form (PDB: 3A6V) as search model. There was clearly one better solution in P41212 compared to the other possible space group. MolRep generated a homology model that was input for a second molecular replacement run in Phaser [26], and this solution was refined further with Translation/Libration/Screw (TLS) and restrained refinement REFMAC 5.5 [25]. There are two molecules in the asymmetric unit resulting in a solvent content of 51%. The model was inspected and finalised in Coot [27].

Molecular modelling of AsMutT and VcMutT

Two homology models of AsMutT were generated by target-template alignment in Swiss-Model [28]. The first model, referred to as AsMutT-open, was built using the crystal structure of VcMutT from this work as a reference that displays 58% sequence identity with AsMutT. The second model, referred to as AsMutT-closed, was generated using the EcMutT-8-oxo-dGMP-Mn (II) structure (PDB: 3A6U) as a template. A homology model of VcMutT in a closed conformation was also generated from the same EcMutT structure (PDB: 3A6U) and is denoted VcMutT-closed.

Structural analysis and interaction calculations

For the structural comparison, we included the two VcMutT monomers A and B, the AsMutT-open homology model and the four EcMutT crystal structures with 8-oxo-dGMP-Mn (PDB: 3A6U), 8-oxo-dGMP (PDB: 3A6T), Mn bound (PDB: 3A6V) and Apo (PDB: 3A6S).

The program HBPLUS v3.15 [29] was used to calculate hydrogen bonds satisfying the following criteria for the parameters donor (D), acceptor (A), acceptor Antecedents and (calculated) hydrogen (H): maximum distances for D–A, 3.5 Å and H–A, 2.5 Å; minimum angle for D–H–A, D–A–AA and H–A–AA of 90°.

Ion pair formation was investigated by the WHAT IF Web Interface (http://swift.cmbi.ru.nl/servers/html/index.html) [30] with interatomic distances <6 Å or <4 Å between the side chains of Asp or Glu to Arg, Lys or His residues.

Results and discussion

MutT from A. salmonicida and V. cholerae

The Nudix hydrolase MutT is part of the nucleotide sanitization system and is thus an important tool to protect pathogens from oxidative stress induced by heir host during bacterial infection. Here we have biochemically and structurally characterised the MutT enzymes from the fish pathogen A. salmonicida (AsMutT) and human pathogen V. cholerae (VcMutT) in order to determine their enzymatic, stability and structural properties. Primary sequence analysis of VcMutT, AsMutT and EcMutT enzymes shows that AsMutT and VcMutT are highly conserved displaying 58% sequence identity and more distantly related to EcMutT with sequence identities of 40.6% and 41.6% respectively (Fig. 2). When aligning the Nudix motif (residues 39–61) of these three enzymes, we see that it is highly conserved, with substitutions only in position 44, 47, and 56 where AsMutT has Asp44, Ala47 and Asn56, compared to Gly44, Arg47 and Glu56 in VcMutT.

Fig. 2

Sequence alignment of VcMutT, AsMutT and EcMutT with the secondary structure elements of VcMutT (top) and EcMutT (bottom, PDB: 3A6S). Nudix box residues 39–61 (GX5EX7REUXEEXGU, where U is a bulky hydrophobic residue and X is any residue).

Enzyme kinetics and activation energy calculations

In order to analyse the ability of AsMutT and VcMutT to perform hydrolysis of nucleoside- and deoxynucleoside triphosphate substrates, we performed activity assays of both enzymes with an 8-oxo-dGTP substrate in the presence of pyrophosphatase. Since we also wanted to identify potentially cold-adapted properties of AsMutT, the assay was performed at six different temperatures for both enzymes (4, 11, 18, 25, 32 and 37 °C). The results (Table 1) show that both AsMutT and VcMutT possess activity for 8-oxo-dGTP and can thus be considered as MutT enzymes. Cold-adapted enzymes usually show a higher catalytic efficiency than their mesophilic homologues [20,21,31], and the results here show that AsMutT indeed possesses significantly higher kcat/Km values compared to VcMutT at all temperatures studied. The higher catalytic efficiency for AsMutT is mainly caused by high kcat values, but also slightly lower Km values at 18, 25, 32 and 37 °C. We also observed that VcMutT is not active at 4 °C whereas AsMutT is highly active at this temperature, clearly indicating that AsMutT is adapted for function at its naturally low temperature environment, while VcMutT works better at its naturally higher temperature environments. Note that the kcat values are similar for both enzymes at their respective optimum growth temperatures, 12 °C for A. salmonicida and 37 °C for V. cholerae.

Table 1

Kinetic constants at 4–37 °C, and thermodynamic activation energy parameters calculated (kJ mol−1) for the psychrophilic AsMutT and the mesophilic VcMutT.

Enzyme	Ea	T	Km	kcat	kcat/Km	ΔG#	ΔH#	TΔS#
	(kJ mol−1)	(°C)	(μM)	(s−1)	(s−1 μM−1)	(kJ mol−1)
AsMutT	36.26 ± 0.14	4	1.78 ± 1.21	0.534 ± 0.029	0.30 ± 0.12	69.15	33.96	−35.20
		11	2.93 ± 1.42	0.713 ± 0.039	0.24 ± 0.03	70.28	33.90	−36.38
		18	3.21 ± 0.99	0.974 ± 0.039	0.303 ± 0.04	71.31	33.84	−37.47
		25	10.74 ± 1.80	1.54 ± 0.08	0.144 ± 0.045	74.09	33.78	−40.31
		32	14.97 ± 2.29	2.04 ± 0.11	0.114 ± 0.049	72.98	33.72	−39.26
		37	18.17 ± 0.90	2.83 ± 0.25	0.134 ± 0.057	73.38	33.68	−39.70
VcMutTa	45.67 ± 0.11	4	n/a	n/a	n/a	n/a	n/a	n/a
		11	1.99 ± 0.68	0.153 ± 0.004	0.077 ± 0.006	73.92	43.31	−30.61
		18	5.59 ± 0.71	0.252 ± 0.006	0.045 ± 0.008	74.59	43.25	−31.34
		25	12.84 ± 0.65	0.491 ± 0.008	0.038 ± 0.012	74.78	43.19	−31.59
		32	17.24 ± 2.54	0.573 ± 0.032	0.033 ± 0.013	76.21	43.13	−33.08
		37	25.65 ± 5.49	0.77 ± 0.074	0.030 ± 0.014	76.73	43.09	−33.64

n/a indicates no activity detected.

Cold-adapted enzymes are usually characterised by lower Ea than their mesophilic homologues in an enzymatic reaction [21,32]. Ea of AsMutT and VcMutT was derived from the slope (−Ea/R) of the Arrhenius plot (ln kcat as a function of 1/T) made from measurements of kcat at six different temperatures (Fig. 3A). The results show that AsMutT has an Ea of 36.26 kJ mol−1 compared to 45.67 kJ mol−1 for VcMutT, which allows AsMutT to perform catalysis more efficiently at low temperatures. The calculations of the ΔH# and ΔS# of activation revealed lower values for AsMutT (Table 1), as is typically observed for psychrophilic enzymes [33].

Fig. 3

(A) Arrhenius plot of the enzyme activity of VcMutT (open circles), and AsMutT (open triangles) using Eq. (4). The activation energies (Ea) were found to be 45.67  ± 0.11 kJ mol−1 (VcMutT) and 36.26 ± 0.14 kJ mol−1 (AsMutT). (B) Thermal stability measurements of mesophilic VcMutT (…) and psychrophilic AsMutT (---) revealed melting temperatures (Tm) of 54.70 ± 0.22 °C (VcMutT) and 57.80 ± 0.09 °C (AsMutT). The experiments were monitored by DSC at a scan rate of 1 °C min−1, and the thermograms were base line subtracted and normalised towards the protein concentrations.

Thermal stability of AsMutT and VcMutT

In order to compare the overall thermal stability of AsMutT and VcMutT, DSC experiments were performed. The results (Table 1 and Fig. 3B) show that VcMutT possesses a slightly lower overall stability, with a melting temperature (Tm) of 54.70 ± 0.22 °C compared to AsMutT, which has a Tm of 57.80 ± 0.09 °C. The thermal denaturation of both AsMutT and VcMutT enzymes was irreversible and aggregation was evident in the samples extracted from the calorimeter cells. Therefore, attempts to calculate more thermodynamic data were prevented.

The crystal structure of VcMutT

The crystal structure of VcMutT was determined to 2.42 Å resolution (Table 2; PDB: 4V14) by molecular replacement. Different EcMutT models were tested, and the open Mn bound form (PDB: 3A6V) gave higher rotational and translation scores and was thus used as the molecular replacement model. VcMutT adopts an α-β-α sandwich structure with seven β-strands (β1-7) and two α-helices (α1-2) as shown in Fig. 4A. This fold is similar to EcMutT and is conserved in the Nudix hydrolase family of proteins. In VcMutT, the Nudix motif (i.e. the MutT signature) starts with Gly39 and makes a strand-loop-helix-loop (SLHL) motif (β4-loop-α1-loop), terminating with Ile61. VcMutT was crystallized with two protein molecules (A and B) in the asymmetric unit, with average B-factors of 48.9 and 63.8 Å2 respectively (Table 2). The root mean square deviation (RMSD) between molecules A and B is 0.48 Å2 for the Cα atoms, and for simplicity only molecule A, with the lowest mean B-factor, will be referred to hereafter. There are some regions with poor electron density, hence the residues were left out of the final model (A1, A28-29, A76-A79, A132, B24-32, B77-79 and B132).

Table 2

Statistics from the X-ray data collection and refinement of VcMutT. The numbers in parentheses represent values in the highest of 10 resolution shells, and the resolution limits for these are indicated.

X-ray statistics	VcMutT
PDB entry	4V14
Beamline	ESRF, ID14-EH1
Space group	P41212
Unit cell (Å)	a = b = 81.15, c = 90.57
Resolution (Å)	50–2.42 Å
(highest bin)	(2.55–2.42)
Wavelength (Å)	0.934
No. of unique reflections	11 909
Multiplicity	4.4 (4.4)
Completeness (%)	99.1 (100)
Mean (<I>/<σI>)	21.1 (2.8)
Rmerge (%) a	5.2 (59.6)
Wilson B-factor (Å2)	55.69
Refinement
Resolution (Å)	10–2.42
R-factor (all reflections) (%)b	23.02
R-free (%)b	30.21
No. of protein atoms	1982
No. of water molecules	48
RMDS bond lengths (Å)	0.019
RMSD bond angles (°)	1.75
Average B-factor (Å2)
All atoms	56.2
Chain A/B	48.9/63.8
Water molecules	55.1

Rmerge = (∑∑|I(h) − |)/(∑∑(h)), where I(h) is the ith measurement of reflection h and is the weighted mean of all measurements of h.

∑||Fobs| − |Fcalc||/∑|Fobs|, where |Fobs| and |Fcalc| are observed and calculated structure factor amplitudes for all reflections (R-factor) and the reflections applied in the test R-free set (reflections not used in the structure refinement), respectively.

Fig. 4

(A) Ribbon diagram of the VcMutT X-ray structure with β-strands in firebrick red, α-helices in brown and loops in grey. (B) Homology model of VcMutT-closed (grey) made from EcMutT-8-oxo-dGTP-Mn superimposed on VcMutT (firebrick red). 8-oxo-dGMP and Mn2+ are modelled from the EcMutT-8-oxo-dGTP-Mn structure. (C) Hydrogen bonding interactions between 8-oxo-dGMP and VcMutT (crystal structure in firebrick red and closed homology model in grey). The super imposed AsMutT models (closed in dark-grey and open in sky-blue) are also shown. Residues involved in Mn2+ coordination and residues promoting conformational stabilization to the enzymes structure are indicated. (D) Hydrogen bonds between the VcMutT-closed model and 8-oxo-dGMP. Electrostatic surface of the homology model of (E) VcMutT-closed and (F) AsMutT-closed with important residue differences highlighted. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

For Chain A of VcMutT compared to EcMutT the RMSD and no of aligned Cα atoms are: 1.72 Å for 119 Cα atoms compared EcMutT-8-oxo-dGMP-Mn (PDB: 3A6U), 1.73 Å for 120 Cα atoms to EcMutT-8-oxo-dGMP (PDB: 3A6T), 1.33 Å for 115 Cα atoms compared to open EcMutT-Mn (Chain A, PDB: 3A6V), and 1.58 Å for 116 Cα atoms relative to EcMutT-Apo (Chain A, PDB: 3A6S).

Active site comparison of AsMutT, VcMutT and EcMutT

Several residues in the loops between β2–β3 and β5–β6 (28–29 and 76–79) of VcMutT were not visible in the observed electron density, indicating that these loops are flexible. For the open EcMutT structure bound to Mn, residues in the first loop are also missing, but upon substrate binding the enzyme adopts a more closed conformation, in which these loops fold over the active site interacting with each other and become well defined in the EcMutT complex structures (PDB: 3A6T, 3A6U). This ligand-induced conformational change is modelled in the VcMutT-closed and AsMutT-closed homology models based on EcMutT-8-oxo-dGMP-Mn (PDB: 3A6U) (Fig. 4B and C).

MutT belongs to a superfamily of enzymes requiring two divalent cations for activity [12,13], and it has been shown that MutT possesses higher activity when the magnesium concentration is increased up to 10 mM [34]. Although VcMutT was purified with 10 mM MgCl2, no electron density could be seen in the VcMutT structure corresponding to Mg2+. This is not a surprise knowing that the affinity of MutT to Mg2+ is very low (1/50 compared to Mn2+) [12]. However, when we model Mn2+ in VcMutT-closed and AsMutT-closed assuming it is bound in the same sites as Mn in EcMutT, we observe that residues Glu54, Glu58, and the main-chain oxygen of Gly38 are potential metal binding residues compared to EcMutT (PDB: 3A6U) [10,35] thus we believe Mg2+ can bind to our enzymes and thus contribute to the catalytic process (Fig. 4C). A second metal binding site could involve Glu54 and Glu58 in addition to water mediated interactions in both AsMutT and VcMutT, as shown for the open EcMutT with Mn (PDB: 3A6V) [10].

A central question in the field is how MutT enzymes favour the recognition of 8-oxo-dGTP over dGTP? Structural studies have demonstrated that MutT undergoes conformational changes upon substrate binding [8-10]. The analyses have suggested that Asn119 (Asn120 in AsMutT and VcMutT) is especially important for the affinity between EcMutT and 8-oxo-dGMP. Saraswat et al. (2004) [35] described this by performing an N119A-mutation, which led to a 1650-fold decrease in the affinity of the enzyme for 8-oxo-dGMP. The substrate-binding pocket is very well conserved when comparing AsMutT, VcMutT and also EcMutT, and the Asn120 is conserved, thus also discriminating the substrate binding in Vibrio MutTs (Fig. 4D). A close inspection of the substrate-binding pocket also shows that both AsMutT and VcMutT possess a tyrosine residue (Tyr76), which is in a position to hydrogen bond with the oxygen atom (O8) at C8 of 8-oxo-dGTP (Fig. 4D). This feature is not present in EcMutT, since it has a Phe in the same position, but Tyr76 may assist the AsMutT and VcMutT in their substrate discrimination. Additional interactions from AsMutT or VcMutT to 8-oxo-dGTP are identical and include: Arg24, His29, Phe36, Gly38, Lys40 and Lys79 (Fig. 4C and D). The phosphate group of 8-oxoG interacts with the guanidinium group of Arg24, through a metal mediated interaction with Gly38 O, and through waters to Lys79 NZ and Lys40 N atoms. The main chain atoms of Phe36 make three different hydrogen bonds to N2, N1 and O6. From the O3’ atom of the deoxyribose there is a direct bond to His29 (ND1) and a water-mediated interaction to Lys79 NZ.

Structural explanations for cold-active properties

Our enzymatic assays have shown that AsMutT displays cold-adapted properties with a high catalytic efficiency and a low Ea compared to its mesophilic homologue VcMutT. However, the thermal stability of AsMutT is unexpectedly higher compared to VcMutT. Here we will compare the crystal structures of VcMutT and EcMutT, and the homology models to find structural explanations for the observed biochemical differences.

Many psychrophilic enzymes possess higher Km values than their mesophilic counterparts, a phenomenon that has been explained by a more dynamic or flexible active site that binds its substrate weakly [33]. A low Km value, as found for AsMutT, indicates that it possesses a higher affinity for its substrate, which is unusual but has been observed for other cold-adapted DNA repair enzymes like uracil DNA N-glycosylase from Atlantic cod (cUNG) [36] and A. salmonicida [37] and for the anionic salmon trypsin (AST) [38]. Herein, the negatively charged DNA substrate binds very strongly to a positively charged active site of cUDG, and in the case of trypsin, the negative surface charge of AST was proposed to explain the tight binding of the positively charged substrate [38,39]. Upon substrate binding the surface is important and the calculated electrostatic surface potential of modelled VcMutT-closed and AsMutT-closed bound to 8-oxo-dGMP (Fig. 4E and F) illustrate this. The overall charges are not so different since AsMutT has a net charge of −7 and VcMutT −8 (Table 3), but the charge distribution is slightly altered. Here Lys27 is unique to AsMutT (His27 in VcMutT) and situated on the tip of the loop between β2 and β3, which changes conformation upon substrate binding [10], and moves closer to the substrate binding site in the closed models. Also, in the Nudix motif, Asn56 in AsMutT is substituted by a negatively charged Glu56 in VcMutT. Taken together, we believe that the presence of Lys27 in AsMutT and Glu56 in VcMutT can explain the slightly higher substrate affinity of AsMutT for the negatively charged triphosphate tail of the substrate.

Table 3

Sequence and structural characterisation of AsMutT (homology model), VcMutT and four EcMutT structures (PDB: 3A6U, 3A6S, 3A6T, 3A6V). Theoretical pI for AsMutT and VcMutT was calculated from the full-length sequences including the 6× His-tag. Residues included in the various class definitions are net charged, polar, hydrophobic, and aromatic residues.

	AsMutT	VcMutT		EcMutT
Chain, ligand		A	B	8-oxo-dGMP-Mn	8-oxo-dGMP	Mn	Apo
PDB ID	Model	4V14		3A6U	3A6T	3A6V	3A6S
Tm (°C)	57.80	54.70		–
No. amino acids per monomer in PDB (in gene construct)	129 (135)	124 (134)	119	126 (129)	127	122	125
Calculated pIa	5.77 (5.10)	5.71 (5.01)		5.01
Net chargeb	−7	−8		−7
Polar residuesc (%)	27.3	27.4		21.9
Hydrophobic residuesd (%)	38.7	39.7		47.3
Aromatic residuese (%)	11.5	11.5		10.1
No. of Met. residues	2	3		4
No. of Gly. residues	8	13		10
No. of Pro. residues	6	5		9
Arg/(Arg + Lys)	0.27	0.46		0.41
(Leu + Ile)/(Leu + Ile + Val)	0.66	0.63		0.74
Resolution	Model	2.42 Å	2.42 Å	2.56 Å	1.96 Å	2.00 Å	1.80 Å
No. of hydrogen bonds per residue in PDB file	0.77	0.80	0.73	0.83	0.83	0.77	0.80
No. SSf hydrogen bonds per residue	0.11	0.08	0.08	0.12	0.10	0.07	0.09
No. SMg hydrogen bonds per residue	0.12	0.12	0.08	0.17	0.16	0.14	0.14
No. MMh hydrogen bonds per residue	0.53	0.60	0.56	0.54	0.57	0.57	0.57
No. ion pairs <4/<6 Å	8/12	7/9	6/8	10/12	10/13	7/15	9/17
No. 2 member networks <4.0 Å	5	2	3	2	2	4	1
No. 3 member networks <4.0 Å	0	1	0	2	2	0	3
No. 4 member networks <4.0 Å	1	1	1	0	0	1	1
No. 5 member networks <4.0 Å	0	0	0	1	1	0	0

Calculated pI includes 6× His-tag for AsMutT and VcMutT. (Calculated pI without 6× His-tag).

Calculated with residues R, K, D and E.

Polar residues G, S, T, Y, N, Q and C.

Hydrophobic residues A, V, L, I, W, F, P and M.

Aromatic residues F, W and Y.

SS, side-chain to side-chain hydrogen bonds.

SM, side-chain to main-chain hydrogen bonds.

MM, main-chain to main-chain hydrogen bonds.

The Ea calculation of AsMutT and VcMutT enzymes was made based on the kcat values of the two enzymes, and shows that AsMutT possesses a lower Ea compared to VcMutT, and thus agrees with previous findings for other cold-adapted enzymes (reviewed in Feller et al. 2013) [40]). The main explanation for the reduced Ea lies in the ΔH# of the enzyme, which indicates that a reduced number of enthalpy related interactions have to be broken during transition-state formation [41] and that the kcat is thus less dependent on temperature. In order to find an explanation for our observed difference in ΔH#, we inspected the AsMutT-closed and VcMutT-closed models and analysed the protein-substrate interactions (Fig. 4C). The results show that all of the interactions between residues in the substrate binding pocket and the substrate are the same in AsMutT and VcMutT (Fig. 4C and D). This thus suggests that other parts of the enzyme most likely account for the observed enthalpy difference. Recently it was suggested that differences in enzyme–water interaction could explain reduced ΔH# for the cold-adapted AST compared to its mesophilic homologue Bovine trypsin [42]. An analysis of residues on the surface of AsMutT and VcMutT shows that there are many substitutions, which may change the protein–water interactions. We see in particular that AsMutT has more charged residues on its surface, where VcMutT has polar residues, e.g. for the surface residues 14, 17, 44, 62, 67, 70, 90, 126 (Fig. 2). This can result in reduced H-bonding network on the surface since the latter may prefer to interact with bulk solvent [42].

The higher negative value of ΔS# for AsMutT corresponds well to previous findings of other cold-adapted enzymes and suggests that the ground state enzyme-substrate complex has a greater degree of disorder than for VcMutT (reviewed in Feller et al. 2013) [40]. But a more disordered AsMutT does not relate to the observed higher thermal stability of the enzyme. The lower ΔS# of AsMutT is thus not rationalised in our current structure or models.

Structural elements promoting overall stability

Characterisation of the thermal stability by DSC showed that AsMutT possesses a Tm of 57.80 °C, which is 3.10 degrees higher than for VcMutT. It is unusual that cold-adapted enzymes have an increased overall stability compared to their mesophilic homologues, however it has previously been observed for some enzymes, e.g. isocitrate dehydrogenase from Desulfotalea psychrophila [43]. In order to identify specific structural determinants, which may explain the difference in temperature stability between AsMutT and VcMutT, and the unexpected high stability of AsMutT, we compared the amino acid compositions, the content of aromatic residues, ion pairs and hydrogen bonds of the two enzymes (Table 3), and compared these values to the mesophilic EcMutT. This analysis revealed that AsMutT possesses less glycine residues compared to VcMutT and EcMutT (8/13/10 Gly in As/Vc/EcMutT). The additional glycine residues in VcMutT are found within the β-strands β1 (Gly9) and β7 (Gly104), but the ones on the surface, which is Gly44 (in the loop between β4 and α1), Gly93 (between β6 and β7) and the disordered Gly132 in the C-terminal, are all likely to contribute to the reduced thermal stability of VcMutT (Fig. 5A). The (Leu + Ile)/(Leu + Ile + Val) ratio, where a high number reflect a higher thermal stability [44] is similar in all enzymes, but the Arg/(Arg + Lys) ratio is lowest in AsMutT (Table 3) indicating lower thermal stability. However, we also found fewer flexible methionine residues in AsMutT (2/3/4 Met in As/Vc/EcMutT) and one extra proline (Pro107) in AsMutT compared to VcMutT (6/5/9 Pro in As/Vc/EcMutT) both promoting higher stability of AsMutT.

Fig. 5

(A) Ribbon diagram of VcMutT with unique glycines (green) believed to contribute to reduced thermal stability. (B) Superimposition of X-ray structure of VcMutT (firebrick red) and the AsMutT-open homology model (sky-blue) with conserved aromatic residues in white. Unique aromatic residues are shown in firebrick red (VcMutT) and sky-blue (AsMutT). (C) Amino acids involved in ionic interactions in AsMutT-open (sky-blue) which include Lys90-Glu65 and (D) Lys18-Glu110, with the corresponding VcMutT residues (firebrick red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Aromatic interactions are important contributors to protein stability, and AsMutT and VcMutT have 12 identical aromatic residues located in the same positions in the sequences (Fig. 2) and both have an aromatic residue in position 131 (AsMutT Tyr131; VcMutT Phe131). In addition, AsMutT includes Tyr82, Phe86 and Tyr96, and VcMutT has Phe12, Phe72 and Phe74 (Fig. 5B). For the latter, Phe72 and Phe74 form stacking interactions with each other with a distance of 4.8 Å from the two aromatic rings, that are adjacent to both Phe83 and Phe85 which again are adjacent to Phe36 and Phe116. This makes a bigger aromatic cluster in VcMutT than in AsMutT and favours overall stability of the former. The unique AsMutT aromatic residues are all found at the surface, altering the protein water interactions. In total, the aromatic interactions favour thermal stability in VcMutT due to a bigger aromatic cluster than in AsMutT.

An analysis of ion pairs showed that there are some unique long ionic interactions in AsMutT-open, including Lys18-Glu110 and Lys90-Glu65 (Fig. 5C and D), which gives a total of 8/12 ion pairs (<4/<6 Å) in AsMutT-open compared to 7/9 pairs in VcMutT (Table 3). In EcMutT, there are more ion pairs (<4 Å) in the closed forms (PDB: 3A6U, 3A6T) than in the open apo form (PDB: 3A6S), but not the open Mn2+ bound structure (PDB: 3A6V). In all EcMutT structures, there is one large ion pair cluster with 4 or 5 members (Table 3). The same ion pair cluster is found in both AsMutT and VcMutT with Glu42, Arg53, Glu54 and Glu54 as shown in Fig. 4C, but since the residues are conserved in both AsMutT and VcMutT they do not explain the difference in Tm. Ionic interactions and increased formation of large ionic networks at the protein surface increases the thermal stability [43] and for AsMutT-open the two unique interactions could enhance the thermal stability. The number of H-bonds per residues in AsMutT-open, VcMutT and EcMutT show comparable numbers from 0.73 to 0.83 (Table 3) and thus seem similar in the compared structures.

MutT is a small and compact monomeric enzyme (15 kDa), which is well conserved throughout evolution and has been referred to as a “house-cleaning” enzyme [45]. The unexpected high thermostability of AsMutT does not seem to affect the catalytic efficiency of the enzyme at low temperatures, and there is no tradeoff for the enzyme function in its natural low temperature environment. Thus, we explain the high temperature stability of AsMutT with its high conservation in nature, small and compact structure and the need of having a functional MutT enzyme available at all times in order to maintain the A. salmonicida genome during ROS attacks from the hosts they are invading.

Concluding remarks

Here we have studied MutT enzymes from the psychrophilic fish pathogen A. salmonicida and the mesophilic human pathogen V. cholerae. Our results show that both enzymes are functional MutT enzymes, catalysing the degradation of 8-oxo-dGTP to 8-oxo-dGMP and PPi, thus both organisms have these important nucleotide sanitizing enzymes to fight ROS attacks from their host during infection. We have also shown that AsMutT is a cold-active enzyme with higher catalytic efficiency and lower energy of activation compared to VcMutT. The high efficiency is mainly caused by a high kcat, which is allowed by a low ΔH# caused by a reduced number of enthalpy related interactions, which have to be broken during transition-state formation. AsMutT showed unexpected high temperature stability, which through comparative analysis of the crystal structure of VcMutT and homology models of AsMutT, is explained by the absence of three Gly residues in loops (Gly44, Gly93, Gly132), one extra proline (Pro107) and two additional long-range ion-pairs on the enzyme surface.

Recent studies on human MutT (MTH1) have shown that there is potential for using MTH1 inhibitors as anticancer drugs [18,19]. Upon addition of MTH1 inhibitors to cancer cells, the cells were directed to apoptosis and died. Based on this finding and the current knowledge that exists regarding bacterial MutT and their importance for the bacterial resistance to ROS upon infection, we propose that there is a potential for using MutT inhibitors in treatment of bacterial infections both in humans and animals as an alternative to antibiotics.