Introduction

Tuberculosis remains a leading cause of death by infectious disease worldwide1. The growing incidence of Mycobacterium tuberculosis drug-resistant strains urges for a deeper understanding of the pathogen's biology and discovery of novel essential pathways that can grant new drug targets. Although the pathogen's genome has been decoded more than a decade ago, a function remains to be assigned to many of the genes2.

Mycobacteria synthesize unusual polysaccharides containing α–(1,4)-linked methylated hexoses that are slightly hydrophobic and with an helical conformation typical of amylose3. One of these polysaccharides is the methylglucose lipopolysaccharide (MGLP) and the other is a methylmannose polysaccharide (MMP)4,5. MGLP contains 10–20 hexose units (two branching glucose units), some of which are acylated with acetate, propionate, isobutyrate, succinate or octanoate4; the mycobacterial MMP is linear, also has variable chain length and methylation, but does not seem to contain acyl groups5. These polysaccharides form stable 11 complexes with fatty acids and modulate the activity of fatty acid synthase I in vitro6. While MGLP has been found in both slow- and rapid-growing mycobacteria and also in Nocardia otitidiscaviarum, MMP has been detected in rapid-growing mycobacteria and in Streptomyces griseus4,7,8,9,10.

MGLP contains glyceric acid linked to the first glucose forming glucosylglycerate (GG)4,11. The GG moiety is linked to a second glucose to form glucosyl–(1,6)–glucosylglycerate (diglucosylglycerate, DGG), which is attached to the glucose that initiates the main α–(1,4)–linked MGLP chain7. While the genes for the first step (Rv1208) and for the elongation (Rv3032) and possibly the gene for 6-O-methylation (Rv3030) of MGLP have been identified12,13, those for the additional steps remain to be identified (Fig. 1). These genes were considered essential for M. tuberculosis growth rendering this biosynthetic pathway a promising target for drug development14. Free GG and DGG were detected in M. smegmatis and proposed to be the precursors for MGLP synthesis15. More recently, GG has also been identified in different organisms ranging from methanogenic archaea and thermophilic bacteria to cyanobacteria and proteobacteria, where it serves as compatible solute during salt stress16,17,18.

Figure 1
figure 1

Proposed pathway for the synthesis of the MGLP in M. tuberculosis.

Confirmed activities are shaded in blue. White boxes indicate putative/deduced enzyme activities. Genes linked to the MGLP pathway by mutagenesis studies are indicated in red. GpgS, glucosyl-3-phosphoglycerate synthase; GpgP, glucosyl-3-phosphoglycerate phosphatase; DggS, di-glucosylglycerate synthase; GT, glucosyltransferases; MeTr, methyltransferases; AcTr, acyltransferases.

Glucosylglycerate can be synthesized by condensation of NDP-glucose and D-glycerate19. Alternatively, a phosphorylated intermediate (glucosyl-3-phosphoglycerate, GPG) is formed by a glucosyl-3-phosphoglycerate synthase (GpgS) and dephosphorylated by a glucosyl-3-phosphoglycerate phosphatase (GpgP)20. Mycobacterial GpgSs (Rv1208 in M. tuberculosis H37Rv) have very low sequence identity with GpgSs from the above mentioned organisms12. Since the M. tuberculosis gene had been considered essential for growth14, we have determined the protein's three-dimensional structure, laying the molecular basis for structure-based drug design21. Given that GpgS synthesizes GPG, a phosphatase was deemed necessary to yield GG, the putative primer for MGLP synthesis. However, a gene for an archetypal GpgP was absent from mycobacterial genomes20.

Herein, we report the purification of the native GpgP from M. vanbaalenii cell extracts and the identification of the corresponding gene. The M. tuberculosis GpgP gene (Rv2419c) was expressed in E. coli and the recombinant enzyme was purified and characterized. The assignment of a new function to a mycobacterial gene represents an important contribution into the knowledge of MGLP biosynthesis.

Results

Identification and sequence analyses of the mycobacterial GpgP (mGpgP)

BLAST analyses with glucosyl-3-phosphoglycerate phosphatase (GpgP) or related mannosyl-3-phosphoglycerate phosphatase (MpgP) sequences showed no homologues in mycobacterial genomes20,22. To detect glucosyl-3-phosphoglycerate-dephosphorylating activity we tested cell-free extracts from two species of Mycobacterium. Unlike the Mycobacterium smegmatis extract that only partially dephosphorylated glucosyl-3-phosphoglycerate (GPG), that from Mycobacterium vanbaalenii completely dephosphorylated GPG to GG (results not shown) leading us to select this species for purification of the native GpgP. We performed chromatography to isolate fractions with GpgP activity and one of the purest fractions contained 8 protein bands that were analyzed by mass spectrometry (peptide mass fingerprinting) (Fig. 2A). One of the proteins below the 25 kDa standard was identified as a putative phosphoglycerate mutase (PGM, Mvan_3924) (Fig. 2A). Since PGMs are part of the large histidine phosphatase superfamily, which includes several phosphatases with different specificities23, this protein was considered a likely GpgP candidate. The homolog from M. tuberculosis (Rv2419c) was expressed and functionally confirmed to be a glucosyl-3-phosphoglycerate phosphatase (GpgP, EC 3.1.3.70) since GPG was, by far, the preferred substrate (Table 1).

Table 1 Substrate specificity of the recombinant GpgP (Rv2419c) from M. tuberculosis
Figure 2
figure 2

Flow diagrams of the purification processes for the native and recombinant GpgPs.

(A) Purification of the native GpgP from M. vanbaalenii and SDS-PAGE gel with the purest fractions obtained. (B) Purification from E. coli extracts of the M. tuberculosis recombinant GpgP and SDS-PAGE gel with the pure recombinant enzyme (three consecutive fractions eluting from the final step).

The gpgP gene from M. vanbaalenii contained 678 bp coding for a polypeptide with 225 amino acids with a calculated molecular mass of 24.2 kDa and a isoelectric point of 5.6, while the M. tuberculosis gpgP gene contained 672 bp encoding a protein with 223 amino acids with a calculated molecular mass of 24.2 kDa and a isoelectric point of 6.1. Gel filtration indicated that the recombinant His-tagged mGpgP behaved as a dimeric protein in solution, with a molecular mass of about 46.0 ± 2.8 kDa (results not shown).

mGpgP had close homologues in the available mycobacterial genomes and in other actinobacteria. BLAST analyses with the M. tuberculosis GpgP sequence revealed homologues in M. bovis (100% amino acid identity), M. leprae (86%), M. marinum (84%), M. intracellulare (84%), M. ulcerans (83%), M. avium (83%), M. parascrofulaceum (82%), M. kansasii (80%), M. smegmatis (77%), M. vanbaalenii (77%), M. abscessus (75%) and M. gilvum (73%) and also in Rhodococcus equi (62%), Nocardia farcinica (61%), Corynebacterium diphtheriae (49%) and Streptomyces griseus (42%). The amino acid identity of mGpgP (Rv2419c) with known GpgPs and MpgPs (EC 3.1.3.70), typically associated to the two-step pathways for synthesis of GG or MG20,22, was negligible. Although the flanking genes have not been assigned to known functions, the genetic environment in the region containing the gpgP gene was highly conserved in mycobacterial genomes and in those of members of the genera Rhodococcus, Nocardia and Corynebacterium (Fig. 3).

Figure 3
figure 3

Genetic context of the gpgP genes in mycobacteria and closely related actinobacteria.

gpgP – glucosyl-3-phosphoglycerate phosphatase gene. nadD – probable nicotinate-nucleotide adenylyltransferase gene. orf, open reading frame of unknown function. Amino acid identity between GpgP homologues is indicated.

Five paralogs with relevant homology to mGpgP (Rv2419c) were identified (Table 2); the C-terminal domain of Rv2228c (43% amino acid identity), Rv3214 (33%), Rv3837c (32%), Rv0525 (32%) and Rv0489 (30%), all contained the distinctive “RHG” motif of the histidine phosphatase superfamily24,25,26 and some have assigned putative PGM functions. However, only the Rv0489 was structurally characterized as a PGM26. A sixth hypothetical PGM was annotated in the M. tuberculosis genome (Rv2135c) but no relevant homology with mGpgP was detected with the BLAST tool.

Table 2 GpgP (Rv2419c) paralogs in Mycobacterium tuberculosis H37Rv genome

A phylogenetic tree constructed from the alignment between mGpgP and other known GpgPs/MpgPs shows that the mycobacterial protein groups within the histidine phosphatase superfamily cluster (Fig. 4A). The additional M. tuberculosis enzymes annotated as putative PGMs also cluster within this group while other known GpgPs cluster with the Haloacid dehalogenase-like hydrolases (HAD-like hydrolase) (Fig. 4A). However, mycobacterial GpgPs exhibit the “RHG” motif of the histidine phosphatase superfamily (Fig. 4B).

Figure 4
figure 4

(A) Unrooted phylogenetic tree based on the amino acid sequences of identified and putative GpgPs/MpgPs (EC 3.1.3.70), PGMs (EC 5.4.2.1) and of other enzymes of the histidine phosphatase superfamily.

Organisms where GG (or MG) has been detected are shaded in yellow. Organisms with enzymes shown to dephosphorylate GPG (or MPG) are boxed. E. coli enzymes with confirmed function are highlighted in blue. The mycobacterial GpgPs studied in this work are highlighted in red. Peptide accession numbers (NCBI) are indicated. Scale bar, 0.2 changes per site. (B) Alignment of the N-terminal amino acid sequences of the PGM from E. coli, the GpgPs from M. vanbaalenii (Mvan_3924) and from M. tuberculosis (Rv2419c) and its paralogs. The typical “RHG” motif of the histidine phosphatase superfamily is boxed.

Properties of the recombinant mGpgP from M. tuberculosis

Expression of the M. tuberculosis gpgP gene in E. coli resulted in high level production of recombinant His-tagged protein, which was purified to homogeneity (Fig. 2B). Among the compounds tested as possible substrates for the recombinant mGpgP, GPG was by far the preferred substrate. Lower activity could also be measured with MPG (<10%) and MGPG (<5%), while the activity with p-nitrophenyl phosphate was extremely low (Table 1, Fig. 5A). Mycobacterium tuberculosis mGpgP exhibited Michaelis-Menten kinetics at 37°C with a Km for GPG of 0.35±0.03 mM and Vmax of 67.2±1.4 µmol/min.mg protein (Table 3). A comparison between mGpgP and other isofunctional GpgPs (HAD-like) is indicated in Table 320,27,28. mGpgP showed only trace PGM activity (0.96±0.08 µmol/min−1mg−1 protein) in the 3-PGA to 2-PGA direction representing <2% of the activity measured with GPG (Table 1, Fig. 5A), while the human PGM converted 3-PGA to 2-PGA with a specific activity of 23.0±1.5 µmol/min.mg protein. No PGM activity was detected in the 2-PGA to 3-PGA direction (Table 1, Fig. 5A).

Table 3 Properties of known GpgPs and one MpgP (with GpgP activity). The M. tuberculosis GpgP belongs to the histidine phosphatase superfamily (His-Phos) while the other proteins belong to the HAD-like hydrolase superfamily
Figure 5
figure 5

(A) Substrate preference of the recombinant GpgP from M. tuberculosis, (B) Temperature profile and (C) pH dependence.

GPG, MPG, MGPG and pNPP were tested as substrates for dephosphorylation; 3-PGA and 2-PGA were tested as substrates for PGM activity.

The recombinant mGpgP was active between 20 and 50°C, with maximal activity at about 45°C (Fig. 5B). At 37°C, the enzyme was active between pH 4.0 and 10.0, optimally at pH 7.0 (Fig. 5C). mGpgP activity was not dependent on divalent cations , although the addition of 5 mM EDTA reduced the activity to about 70% of maximal rate. There were no relevant differences in mGpgP activity with Mg2+ (1–10 mM), Mn2+ (1–10 mM), Co2+ (1–5 mM) or water. However, higher Co2+ concentrations (10–50 mM) progressively inhibited enzyme activity and Cu2+ was inhibitory at any concentration tested (1–50 mM). The enzyme retained full activity after 1 week storage on ice and at −20°C in 50 mM BTP with 50 mM NaCl (pH 7.0).

Discussion

The structural complexity and different linkages between hexoses in the mycobacterial methylglucose lipopolysaccharide (MGLP) require different enzymes for its biosynthesis7,8,15,29. Although most of the genes involved in this pathway remain unknown, the M. tuberculosis H37Rv gene (Rv3032) for the α-(1,4)-glycosyltransferase required for the elongation of MGLP and the methyltransferase gene (Rv3030) for 6-O-methylations have been identified (Fig. 1)13. It was also suggested that other genes in the Rv3030-Rv3037c cluster including putative glycoside hydrolase and acyltransferase genes might catalyze additional steps in the pathway13,30. We have recently identified the gene (Rv1208) for the first step in MGLP biogenesis and characterized the corresponding glucosyl-3-phosphoglycerate synthase (GpgS)12,21. Biochemical evidence was supported by genetic confirmation of the involvement of Rv1208 in the MGLP pathway31. This gene was located within a cluster (Rv1208-Rv1213) containing a α-(1,4)-glycosyltransferase gene (Rv1212c) likely involved in the elongation of MGLP and a NDP-glucose pyrophosphorylase gene (Rv1213) that may supply NDP-glucose for polymerization13,32.

Since GpgS synthesizes glucosyl-3-phosphoglycerate (GPG), the existence of a phosphatase to convert this phosphorylated precursor into GG was anticipated12,30. In most organisms that synthesize GPG, the dephosphorylation is carried out by a specific glucosyl-3-phosphoglycerate phosphatase (GpgP, EC 3.1.3.70) encoded by a gene normally contiguous or located at close distance from the GpgS gene20,27. Despite the fact that no typical GpgP could be detected in mycobacterial genomes, M. vanbaalenii extracts completely dephosphorylated GPG allowing the purification of a highly specific GpgP, which was designated mGpgP (mycobacterial GpgP). The corresponding gene (Rv2419c in M. tuberculosis) was annotated as a putative phosphoglycerate mutase (PGM) for the interconversion of 3-phosphoglycerate and 2-phosphoglycerate, but our results indicate that the enzyme retained only residual PGM activity. Phosphoglycerate mutases are part of a large group of functionally diverse enzymes included in the histidine phosphatase superfamily, as they share a conserved catalytic motif “RHG” where the histidine plays a crucial role in catalysis, being phosphorylated during the reaction23. Although PGMs are the best studied members of this superfamily, phosphatases largely dominate the group, which now also includes mGpgP with a confirmed function. Interestingly, all GpgPs characterized to date share conserved motifs of the haloacid dehalogenase-like hydrolase superfamily (HAD-like hydrolase)20,27,33 and lack the “RHG” motif found in mGpgP23. HAD-like phosphatases, which also include mannosyl-3-phosphoglycerate phosphatases (MpgP, EC 3.1.3.70) that are GpgP homologues, are also sometimes referred to as “DDDD” phosphohydrolases due to the presence of four invariant aspartate residues33,34. Structural studies have attributed PGM function to Rv048926, which shares 30% amino acid identity with Rv2419c (mGpgP). Four other proteins in M. tuberculosis have relevant amino acid identity to Rv0489 and to Rv2419c and all share the “RHG” motif. One of these paralogs, Rv3214, was annotated as a putative PGM but has recently been shown to be an acid phosphatase without PGM activity, although the specific substrate has not been identified25. While function based on sequence similarity alone is not a reliable approach to the unambiguous identification of enzyme activity, histidine phosphatases present additional challenges to automated genome annotation as they are remarkably diverse23,35.

The mycobacterial GpgP had homologues in other actinobacterial genomes, namely in Rubrobacter xylanophilus, so far the only actinobacterium known to accumulate the GG-related solute mannosylglycerate (MG)36. In this organism, an actinobacterial-type MpgS/GpgS with dual specificity could synthesize the phosphorylated precursors for both MG and GG in vitro even though neither GG nor MGLP have, so far, been detected in vivo37. A classical MpgP (HAD-like) involved in MG synthesis in hyper/thermophiles, was also absent from this organism's genome22,28. The present study allowed the detection of two mGpgP homologues in this organism, which may dephosphorylate mannosyl-3-phosphoglycerate (MPG) to MG.

Several of the above MpgPs were functionally interchangeable as they efficiently dephosphorylated MPG and GPG alike20, while mGpgP preferentially hydrolyzed GPG. Moreover, the mycobacterial GpgS was also specific for glucose donors12. Since both enzymes participate in the same pathway, it was not surprising that the temperature and pH profiles for the M. tuberculosis GpgS and GpgP activities in vitro were comparable. However, GpgS activity was absolutely dependent on magnesium ions, while mGpgP activity was independent of cations, unlike all known HAD-like GpgPs/MpgPs that require divalent cations for activity12,20,22,27,28. The GpgS for the first step in the MGLP pathway was predicted to be essential for M. tuberculosis growth, while the mGpgP was not14. This suggests that one of the above Rv2419c paralogs, or a phosphatase with broader specificity, could replace mGpgP activity. However, we did not detect an alternative dephosphorylating activity in M. vanbaalenii extracts during purification.

The Rv2419c sequence motifs place this novel mGpgP in the histidine phosphatase superfamily. The high specificity for GPG points toward convergent evolution with the non-homologous isofunctional GpgPs of the HAD-like hydrolase superfamily, indicating that these types of enzymes have emerged in evolution more than once38. Ongoing structural studies will add further insights on the catalytic mechanism and evolution of both types of GpgPs.

The apparent ubiquity of MGLP in pathogenic mycobacteria and the predicted essentiality of some of the putative biosynthetic genes14 suggest a fundamental role for this structure in mycobacterial physiology and that its absence or deficient assemblage could lead to cell death. Here, we provide biochemical evidence for the second step in MGLP biosynthesis12,30. The identification of this function in the genome of M. tuberculosis with the functional assignment of Rv2419c to GpgP, not only expands the already broad activity range of the histidine phosphatase superfamily, but also furthers our understanding of the MGLP biosynthetic pathway, which might help us devise new strategies to fight tuberculosis.

Methods

Bacterial strains and growth conditions

Mycobacterium smegmatis mc2155 (ATCC 700084) was obtained from LGC Standards S.L.U. (Barcelona, Spain). Mycobacterium vanbaalenii (DSM 7251) was obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Germany). M. smegmatis and M. vanbaalenii biomass for enzymatic assays was obtained from 1-L cultures grown with a glycerol-based medium in an orbital shaker, at 30°C for 24 h. The medium contained per litre: 20 g glycerol (Merck), 5 g casaminoacids (Difco), 1 g fumaric acid (Sigma), 1 g K2HPO4, 0.3 g MgSO4, 0.02 g FeSO4 and 2 g Tween 80 (Sigma), at pH 7.039. Mycobacterium vanbaalenii biomass for purification of the native glucosyl-3-phosphoglycerate phosphatase (GpgP) was obtained from an 18-L culture under the conditions described above. Escherichia coli BL-21 (Novagen) was grown at 37°C in LB medium at pH 7.0 in a 5-L fermentor with continuous aeration and stirred at 180 rpm. Kanamycin was added at a final concentration of 30 μg/mL.

Preparation of cell-free extracts

Cells were harvested by centrifugation (8000 × g, 10 min, 4°C) during the late exponential phase of growth and the pellet suspended in 20 mM Bis-tris propane buffer (BTP) (Molekula) at pH 7.5, containing 5 mM MgCl2, 10 µg/ml DNAse I and protease inhibitors (Roche). Cells were disrupted twice in a French-press followed by centrifugation to remove debris (15000 × g, 30 min, at 4°C). For the analysis of GpgP activity, cell extracts were dialysed overnight against 50 mM BTP at pH 7.5.

GPG synthesis

Glucosyl-3-phosphoglycerate (GPG) was synthesized after overnight reaction of the pure recombinant M. smegmatis GpgS with a mixture containing 20 mM (each) of UDP-glucose, 3-PGA and MgCl2 in 50 mM BTP, pH 7.5 at 37°C12. The concentration of GPG was determined as previously described20. The synthesis of GPG was monitored by thin-layer chromatography (TLC) with the solvent system ethyl acetate/acetic acid/water/ammonia 25% (6621, v/v). Glucosylglycerate (GG), UDP–glucose and D-glucose standards were used for comparative purposes.

Enzyme assays

The activities of the GpgPs from M. smegmatis and from M. vanbaalenii in crude extracts and during purification and the activity of the recombinant GpgP from M. tuberculosis, were detected with reaction mixtures (25 μl) containing 2.0 mM of GPG and 2 mM MgCl2 in 20 mM BTP, pH 7.0, after 10 min incubations at 30°C and 37°C, respectively. The hydrolysis of GPG was monitored by TLC with a solvent system composed by chloroform/methanol/acetic acid/water (305084, v/v). Standard GG and GPG were used for comparative purposes. Protein concentration was determined by the Bradford method40.

Purification of the native GpgP from Mycobacterium vanbaalenii

Mycobacterium vanbaalenii cell extracts were loaded onto a DEAE-Sepharose column equilibrated with 20 mM Bis-Tris propane buffer (BTP) at pH 7.5 (Buffer A) (Fig. 2A). Elution was carried out with a linear gradient of 0 to 1 M NaCl and GpgP activity was determined by TLC as described above. The active fractions were pooled, concentrated by centrifugation in 30 kDa cutoff centricons (Amicon), loaded onto a DEAE-Sepharose equilibrated with 20 mM BTP, pH 6.5 (Buffer B) and eluted by a linear gradient of 0 to 0.7 M NaCl. Active fractions were pooled and diluted 10-fold with buffer A, loaded onto a Q-Sepharose column equilibrated with the same buffer followed by a linear gradient of 0 to 0.7 M NaCl and by a second Q-Sepharose column equilibrated with buffer B and a linear gradient of 0 to 0.5 M NaCl. Desalted samples were loaded onto two sequential Resource-Q columns, equilibrated with buffer A and buffer B, respectively and eluted with linear gradients of 0 to 0.4 M NaCl. Active fractions were equilibrated with buffer A containing 200 mM NaCl and loaded onto a Superdex 200 column equilibrated with the same buffer. Two additional Resource-Q columns and a final Superdex 200 column were used as final purification steps as described above (Fig. 2A). GpgP activity was assessed by TLC and fraction purity was determined by SDS-PAGE (Fig. 2A). Each fraction (50 µl) was frozen and stored at −20°C for protein identification by mass spectrometry as described below.

Identification and sequence analyses of the mycobacterial gpgP gene

Each of the M. vanbaalenii purest fractions (50 µl) obtained after the last purification step were loaded onto a large SDS-PAGE gel (20 cm×20 cm) and separated under constant voltage (100 V). A total of 8 bands were excised for peptide mass fingerprinting analyses (Proteomics Unit, Biocant, Cantanhede). Each of the peptide sequences obtained from the 8 bands was analyzed with the BLAST tool at NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The ClustalX2 program (http://www.clustal.org) was used for sequence alignments and the MEGA4 program (www.megasoftware.net) was used to generate the phylogenetic trees.

Cloning and functional overexpression of gpgP from M. tuberculosis in E. coli

The M. tuberculosis Rv2419c (gpgP) gene sequence was optimized for expression in E. coli and synthesized (GenScript Corp.). The artificial gene was cloned between the NdeI and HindIII restriction sites of pET30a (Novagen) and transformed into E. coli BL21 using standard molecular biology procedures. Recombinant E. coli were grown to mid-exponential phase of growth (OD610 = 0.8); IPTG (Zymo Research) was added at a final concentration of 0.5 mM to induce gene expression and the temperature was lowered to 20°C. Cells were harvested 18 h later by centrifugation (8000×g, 10min, 4°C).

Purification of the recombinant mGpgP from E. coli extracts

The His-tagged recombinant GpgP from M. tuberculosis was purified with a pre-packed Ni-Sepharose high-performance column (His-Prep FF 16/10) equilibrated with 20 mM sodium phosphate, pH 7.4, 0.5 M NaCl and 20 mM imidazole (Fig. 2B). Elution was carried out with 500 mM imidazole and the purity of the fractions was determined by SDS-PAGE. The purest active fractions were pooled, concentrated, equilibrated with Buffer A and loaded onto a Q-Sepharose column equilibrated with the same buffer and eluted by a linear gradient of NaCl (0 to 1 M). Purity was determined by SDS-PAGE and the purest active fractions were pooled, concentrated, equilibrated with Buffer A with 200 mM NaCl and loaded onto a Superdex 200 column equilibrated with the same buffer (Fig. 2B). After SDS-PAGE analysis (Fig. 2B), the pure active fractions were concentrated and equilibrated with 50 mM BTP, pH 7.0 with 50 mM NaCl. Protein content of the samples was determined by the Bradford assay40.

Characterization of the recombinant mGpgP from M. tuberculosis

The substrate specificity of the recombinant GpgP was determined using 2-phosphoglycerate, 3-phosphoglycerate, 2,3-bisphosphoglycerate, 3-glycerol phosphate, phosphoenolpyruvate, α-glucose-1-phosphate, β-glucose-1-phosphate, glucose-6-phosphate, glucose-1,6-diphosphate, mannose-1-phosphate, mannose-6-phosphate, fructose-6-phosphate, trehalose-6-phosphate, ATP, UTP and ADP (all from Sigma-Aldrich), glucosyl-3-phosphoglycerate (GPG), mannosyl-3-phosphoglycerate (MPG) and mannosyl-(1,2)-glucosyl-3-phosphoglycerate (MGPG). GPG, MPG and MGPG were synthesized as previously described41. The general phosphatase substrate p-nitrophenyl phosphate (Sigma-Aldrich) was also tested as a possible substrate for GpgP.

Reaction mixtures (25 µl) containing pure recombinant GpgP (0.025 µg), 3 mM of substrate and 2.5 mM MgCl2 in 20 mM BTP at pH 7.0, were incubated at 50°C for 10 min. The reaction products were visualized by TLC as described above; the dephosphorylation of 2-phosphoglycerate (2-PGA), 3-phosphoglycerate (3-PGA), 2,3-bisphosphoglycerate, 3-glycerol-phosphate and phosphoenolpyruvate was quantified by the release of free phosphate measured by the Ames method42. The p-nitrophenol released was determined at 405 nm as previously described and alkaline phosphatase was used as positive control43.

To assess whether the recombinant GpgP retained PGM activity, both in the forward (3-PGA to 2-PGA) or reverse (2-PGA to 3-PGA) directions, reactions were carried out as previously described44. A recombinant PGM from human muscle (Affymetrix) was used as positive control. Phosphoglycerate mutase activities were monitored using a standard method coupled to NADH oxidation45.

Temperature, pH and cation dependence profiles were determined by the addition of known amounts of pure GpgP to 25 μl reaction mixtures containing the appropriate buffer and 2 mM GPG and stopped at different times by cooling on ethanol-ice. Free phosphate was quantified using the Ames reaction42. The effect of cations was examined by incubation of the reaction mixtures with the chloride salts of Mg2+, Mn2+, Co2+ (1 to 10 mM), with 5 mM EDTA or with no additions, at 37°C. The temperature profile was determined between 20 and 60°C in 20 mM BTP, pH 7.0. The effect of pH was determined at 37°C in 20 mM acetate buffer (pH 4.0 and 5.0), 20 mM BTP (pH 6.0 to 9.0) and 20 mM CAPS buffer (pH 10.0).

The Km value for GPG was determined at 37°C as follows: the reactions were initiated by the addition of the M. tuberculosis GpgP to reaction mixtures with 50 mM BTP at pH 7.5, containing GPG (0.25–20.0 mM) and stopped at different times. The phosphate released was measured as described above and the kinetic parameters Km and Vmax were calculated from Lineweaver–Burk plots. All experiments were performed in triplicate.

The molecular mass of the recombinant GpgP was estimated by gel filtration on a Superdex 200 column and molecular mass standards used were albumin (67 kDa), aldolase (158 kDa), catalase (232 kDa), ferritin (440 kDa). Blue dextran 2000 was used to determine the void volume (GE Healthcare).