A novel thymidylate synthase from the Vibrionales, Alteromonadales, Aeromonadales, and Pasteurellales (VAAP) clade with altered nucleotide and folate binding sites

Genome used and thymidylate synthase orthologous genes

We used the thymidylate synthase Vibrio parahaemolyticus (GenBank WP_100088861.1) found in the FIM-S1708+ strain genome deposited as GenBank JPLV00000000.1 (Gomez-Jimenez et al., 2014) as a seed to identify the homologous TS, using blast searches with an E-value cutoff of 1.0^e−15 against the proteomes encoded in 38 genomes belonging to the gamma-proteobacteria class (File S1). All proteins that shown an alignment ≤70% of their length were kept for further analysis. We collected these genomes for two reasons: First, to get a better perspective of the variation of the TS across the Gammaproteobacteria (primarily focused on the active site) and second, these genomes have been used in phylogenomic analyses of that bacterial class (Williams et al., 2010).

Phylogeny among the thymidylate synthase

To conduct a Maximum Likelihood (ML) phylogeny, we created a protein Multiple Sequence Alignment (MSA) using MUSCLE with 50 iterations (Edgar, 2004). We ran the ML phylogeny specifying the LG+G+I model, as determined by ProtTest3 (Darriba et al., 2011) and using non-parametric bootstrap analysis (100 replicates) to establish the support for the clades. The phylogenetic tree was colored and edited using FigTree v1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/).

VpTS expression and purification

VpTS was expressed in Escherichia coli using a codon-optimized synthetic gene cloned into the T7-promoter expression vector pJexpress414 which contains the ampicillin resistance gene (DNA2.0). We used the amino acid sequence of Vibrio parahaemolyticus thymidylate synthase (GenBank WP_100088861.1) including in the N-terminus a 10 His-tag and the cutting site for the PreScission Protease (GE Healthcare, Little Chalfont, UK). The total theoretical mass would be 35,944 Da and the tag could be removed upon treatment with the protease.

The E. coli BL21(DE3)-SI strain was used to express VpTS. To facilitate yield of soluble protein we used a plasmid containing the chaperones groES-groEL-Tig (TAKARA plasmid PG-Tf2). For this, the bacteria were transformed first with plasmid pPG-Tf2 with chloramphenicol selection at 20 µg/mL, and later with the pJexpress414-VpTS plasmid using chloramphenicol and ampicillin at 100 µg/mL. Both antibiotics were used in all further procedures.

A 5 mL of a starter culture of transformed bacteria was incubated overnight and used to inoculate 1 L of LB broth without NaCl. The bacteria were incubated in an orbital shaker at 250 rpm at 37 °C. When an optical density of 0.4 at 600 nm was obtained, chaperone expression was induced by addition of tetracycline to a final concentration of 9 ng/mL. When optical density reached 0.6 units (λ = 600 nm), the T7-promoter was induced with IPTG and NaCl to a final concentration of 0.1 mM and 0.3 M respectively. Cell growth was continued for 24 h at 25 °C. The bacterial pellet was obtained by centrifugation at 7,500× g at 4 °C, and the biomass was lysed by sonication in 20 mM potassium phosphate buffer pH 7.5, 5 mM dithiothreitol (DTT), 0.5 mM phenyl methyl sulfonyl fluoride (PMSF), and 5 mM benzamidine. The bacterial lysate was clarified at 25,000× g for 20 min at 4 °C.

For metal affinity chromatography, a 5 mL Ni-His Trap column (GE Healthcare) was equilibrated with 20 mM phosphate buffer pH 7.5, 0.5 M NaCl (buffer A) in an Äkta Prime chromatographer (GE Healthcare) at 1 mL/min. The clarified lysate was loaded into the column and washed with at least five volumes of buffer A until the absorbance at 280 nm returned to baseline. Elution was done with a linear gradient of buffer A and buffer A plus 500 mM imidazole. Fractions of 3 mL were collected. Purification was followed by 12% SDS-PAGE using pre-casted TGX stain-free gels (BioRad) and images were recorded on a GelDoc Easy Imaging system (BioRad). The TGX stain-free detection system has a sensitivity comparable to silver staining (Gilda & Gomes, 2013).

To further purify VpTS and confirm the oligomeric state, size exclusion chromatography was used using a Superdex 75 10/300 column in an Äkta Pure chromatographer (GE Healthcare) at 1 mL/min with Tris HCl 20 mM pH 7.5 and NaCl 100 mM as running buffer. Molecular weight (MW) standards were aprotinin (6.5 kDa), ribonuclease A (13.7 kDa), carbonic anhydrase (29 kDa), ovalbumin (44 kDa), conalbumin (75 kDa) and Blue Dextran (2,000 kDa) to determine the column void volume. VpTS native molecular weight was calculated from linear regression of a K_av vs. log MW plot, using the formula ${\displaystyle K_{av}=\left(V_t-V_o\right)∕\left(V_f-V_o\right)}$ where V_t is the total column volume, V_o is the void column and V_f is the volume where the sample or a molecular weight standard is eluted. Fractions were analyzed by 12% SDS-PAGE as mentioned above. Purified protein was quantitated using the bicinchoninic acid method (BCA kit; Pierce, Waltham, MA, USA).

VpTS kinetics and IC₅₀ inhibition

TS kinetics were determined using a spectrophotometric assay following the formation of dihydrofolate at 340 nm (ε_340nm = 6,400 M⁻¹ cm⁻¹) in a CARY-50 (Varian) UV-vis spectrophotometer as previously described (Arvizu-Flores et al., 2009). The reaction buffer contained 50 mM Tris-HCl pH 7.5, 5 mM DTT, 1 mM EDTA, 25 mM MgCl₂. The substrates were mTHF (Schircks Laboratories, Jona, Switzerland) and dUMP (Sigma-Aldrich, St. Louis, MO, USA). The assay was done in a final volume of 1 mL at 30 °C. The reaction was started by addition of 0.3 µM of recombinant VpTS, and the initial rates were determined by recording the absorbance increase at 340 nm, changing the substrate concentration for mTHF (0–300 µM) and dUMP (0–150 µM). Experimental data were obtained in triplicate and fitted to the Michaelis–Menten equation by a non-linear regression analysis using the GraFit software (Erithacus Software).

Steady-state IC₅₀ inhibition was determined with trimethoprim (TRIM, Sigma-Aldrich, St. Louis, MO, USA) as an antifolate in standard activity conditions (Arvizu-Flores et al., 2009) with saturating substrate concentrations (mTHF 300 µM and dUMP 150 µM) by triplicate. Data were adjusted to a dose–response model using non-linear fit with the GraFit software.

Structure modeling

The VpTS structural model was built using Protein Homology/analogY Recognition Engine V Phyre2.0 server (Kelley et al., 2015). The best model was selected by the highest sequence identity, alignment coverage and model confidence factor within the closed conformation structures. The most similar bacterial structure in the dUMP-bound closed conformation, was E. coli TS (32% of sequence identity; PDB 1AXW) as the reference structure. The structural analysis was performed in CCP4MG V7.0 program (McNicholas et al., 2011)

Expression and purification of VpTS

VpTS was co-expressed with the set of chaperones groES-groEL-Tig in E. coli strain BL21-SI, observing that the protein was best expressed at 24 h post induction. This was observed in the SDS-PAGE follow-up of expression, where VpTS migrated to an approximate mass of 35 kDa (Fig. 1A). This value is within the range of the theoretical molecular weight of 35,944 Da taking into account the 10-His tag. Besides VpTS, there was a highly co-expressed protein at 60 kDa corresponding to the GroEL chaperone molecular weight (Fig. 1A).

VpTS was purified by IMAC chromatography purification as seen in the SDS-PAGE of the chromatography fractions (Fig. 1B). VpTS eluted at an imidazole concentration of 375 mM, as observed VpTS in the asterisk lane of Fig. 1B. The yield of VpTS after IMAC chromatography was 40 mg of protein per liter.

We also included an extra purification step using size-exclusion chromatography, to confirm the native dimeric TS quaternary structure. A molecular weight of 70 kDa was interpolated from the analysis of the after mentioned calculation (Fig. 1C). The gel filtration purified protein had the same specific TS activity, therefore for enzymatic assays we used VpTS purified by IMAC chromatography only. This has been found also for varicella zoster TS, confirming that IMAC was sufficient for VpTS kinetic studies (Hew et al., 2015). Instead of Coomassie Blue or silver staining, TGX stain-free 12% polyacrylamide gels (BioRad) were used during all the purification steps, since they provide a good method to validate protein purity (Gilda & Gomes, 2013; Rivero-Gutiérrez et al., 2014).

Enzymatic and inhibition assay

Kinetic parameters for dUMP were 27.3 ± 4.3 µM for K_m, 0.3 s⁻¹ for k_cat, and 0.01 µM⁻¹ s⁻¹for kinetic efficiency (Fig. 2). For mTHF, the values were K_m 96.3 ± 18 µM, k_cat 0.3 s⁻¹, and a kinetic efficiency of 0.0031 µM⁻¹ s⁻¹ (Fig. 3). These values are higher than those reported for other TS in previous works (Table 1), clearly indicating that a loss of affinity occurs for both substrate and cofactor compared to prokaryotic and eukaryotic TSs (Greene et al., 1994; Spencer, Villafranca & Appleman, 1997; Fox et al., 1999; Sergeeva et al., 2003; Pozzi et al., 2012). Inhibition of VpTS was studied using TRIM, which is a broad range inhibitor of bacterial TS and dihydrofolate reductase (DHFR). Inhibition data were adjusted to a non-linear dose–response model obtaining an IC₅₀ value of 106 µM, with a standard error of 11.36 µM (Fig. 4).

Phylogenetic analysis of thymidylate synthases among the Gammaproteobacteria

To get a more compressive view of the molecular evolution of the TS among the gammaproteobacteria, we constructed an ML-phylogeny using 38 genomes (File S1), which represent the main orders of the Gammaproteobacteria. We have aligned selected TS amino acid sequences and highlighted in yellow the active site residues (File S2). The first thing we noticed is that two main clades were formed and the small clade (orange) corresponds to the VAAP: Vibrionales, Alteromonadales, Pasteurellales and Oceanospirales (Fig. 5). Interestingly, the TS of Shewanella amazonensis, Shewanella denitrificans, Shewanella baltica, Shewanella sp. MR-4, Shewanella putrefaciens and Pseudoalteromonas tunicate, do not reflect a history of the species according to the species tree of these genomes previously reported (Williams et al., 2010). These results suggest that these TS and other genes could be acquired by horizontal gene transfer in these phylogenetic clades.

Moreover, from the amino acid sequence alignment (File S2), we confirmed that all residues that form the canonical TS active site are invariant, except for the TSs in the VAAP clade (Fig. 5, orange rectangle). One of them was a conservative change of lysine to arginine at position 50 in VpTS (Table 2). This residue is important for folate binding as it makes an ionic contact between the mTHF γ-glutamate and its mutation reduces folate affinity (Arvizu-Flores et al., 2008). The other change is more dramatic, a glycine for an arginine at position 141′ in VpTS (Fig. 5, orange rectangle). This change leaves three arginines to coordinate the dUMP phosphate group and possibly reduces nucleotide affinity (Table 2). Other authors have also found that position Arg179′ (equivalent to VpTS Gly141′) was relatively permissive to mutations (Kawase et al., 2000).

Molecular modeling of the dUMP binding site in VpTS

The sequence of VpTS was modeled with Phyre2 to predict the three-dimensional structure of the enzyme. TS has a well-described conformational change from an open active site when unbound or dUMP is present. A closed conformation occurs when the nucleotide and a folate analog are bound (Finer-Moore, Santi & Stroud, 2003). We modeled a dUMP-bound complex by superposing the Phyre2-VpTS model with an E. coli TS structure (PDB 1AXW) with a 100% confidence level and 99% alignment coverage. We confirmed that the nucleotide phosphate was coordinated by three invariant arginine residues in VpTS: Arg22, Arg 140′ and Arg180 (Fig. 6). For clarity, Arg22 was not included the figure. As mentioned above, VpTS lacks the fourth arginine residue (Ec Arg127′; Lc Arg179′; hArg176′; see Table 2) that is important to stabilize the phosphate group by two hydrogen bonds. The presence of glycine in this position (Gly141′) is recurrent in the VAAP clade (File S2).

TRIM inhibition of VpTS

Some treatments against proliferative, microbial, inflammatory and parasitic diseases have been focused on folate metabolism inhibition (Gonen & Assaraf, 2012). TRIM is an antifolate with structural differences respect to methotrexate, which increase their specificity for the bacterial DHFR (Navarrete et al., 2013). In this work, the inhibitory capacity by TRIM for VpTS was evaluated. The IC₅₀ value was 106 µM, whereas values for bacterial DHRF are usually very low, such as 0.001 µM for Haemophilus influenzae and 0.01 µM for E. coli (Wax, 2008). Although TRIM is not a potent inhibitor of TS, due to its commercial availability, it is possible to improve its affinity to TS by rational drug design that would allow inhibition values comparable to the specific compounds for TS.