|Home | About | Journals | Submit | Contact Us | Français|
We report the first enzymatic synthesis of D-tagatose-1-phosphate (Tag-1P) by the multi-component PEP-dependent:tag-PTS present in tagatose-grown cells of Klebsiella pneumoniae. Physicochemical characterization by 31P and 1H NMR spectroscopy reveals that, in solution, this derivative is primarily in the pyranose form. Tag-1P was used to characterize the putative tagatose-1-phosphate kinase (TagK) of the Bacillus licheniformis PTS-mediated D-Tagatose catabolic Pathway (Bli-TagP). For this purpose, a soluble protein fusion was obtained with the 6 His-tagged trigger factor (TFHis6) of Escherichia coli. The active fusion enzyme was named TagK-TFHis6. Tag-1P and D-fructose-1-phosphate (Fru-1P) are substrates for the TagK-TFHis6 enzyme, whereas the isomeric derivatives D-tagatose-6-phosphate (Tag-6P) and D-fructose-6-phosphate (Fru-6P) are inhibitors. Studies of catalytic efficiency (kcat/Km) reveal that the enzyme specificity is markedly in favor of Tag-1P as substrate. Importantly, we show in vivo that the transfer of the phosphate moiety from PEP to the B. licheniformis tagatose-specific enzyme II (EIITag) in E.coli is inefficient. The capability of the PTS general cytoplasmic components of B. subtilis, HPr and EI, to restore the phosphate transfer is demonstrated.
In many bacterial species, from both Gram-positive [Reizer et al., 1988; Thompson, 1987] and Gram-negative genera [Meadow et al., 1990; Postma et al., 1993], glycans are phosphorylated simultaneously with vectorial translocation via sugar-specific phosphoenolpyruvate: sugar phosphotransferase systems (PEP-PTS) [Deutscher et al., 2006]. In this multi-component complex (see, Fig.1B) the high-energy phosphoryl moiety of PEP is transferred sequentially to two general cytoplasmic proteins Enzyme I (EI) and HPr (a histidine-containing phosphocarrier protein). Subsequently, the phosphate moiety is transferred by group translocation to the incoming sugar, via multi-domain sugar: specific membrane proteins designated Enzymes II (EII). For discussion and nomenclature of PTS components, see, [Saier and Reizer, 1992]. Recent NMR studies by Clore and colleagues [Cai et al., 2003; Clore and Venditti, 2013], have provided molecular level insight to the structure, dynamics, and biophysical interactions of the protein-protein complexes comprising the bacterial PEP-PTS.
The lactose: PEP-PTS is present, and has been studied in many microorganisms including: industrially important Group N streptococci [Bissett and Anderson, 1974; Thompson, 1979], Lactobacillus casei [Chassy and Thompson, 1983], oral Streptococci [Hamilton and Lebtag, 1979; Hamilton and Lo, 1978] and significantly, Staphylococcus aureus [Bissett and Anderson, 1980a, b; Bissett et al., 1980]. The multi-cistronic genes encoding the proteins of the lactose (lac)-operon may have a plasmid or chromosomal locus. The product of translocation via the lac-PTS is lactose-6’-phosphate (Lac-6’P). Intracellularly, the phosphorylated disaccharide is cleaved by β-D-phospho-galactoside galactohydrolase EC 220.127.116.11 (P-β-gal), to yield galactose-6-phosphate (Gal-6P) and glucose. After ATP-dependent phosphorylation, the latter hexose (glucose-6P) may directly enter the glycolytic pathway. Conversely, Gal-6P must first be converted to D-tagatose-6-phosphate (Tag-6P) by the D-tagatose pathway prior to glycolytic fermentation. First reported by Bissett and Anderson in 1974, the three-stage D-tagatose pathway comprises: galactose-6P isomerase, EC 18.104.22.168 [Bissett et al., 1980], D-tagatose 6-phosphate kinase, EC 22.214.171.124 [Bissett and Anderson, 1980a] and class I D-ketohexose 1,6-bisphosphate (1,6-BP) aldolase, EC 126.96.36.199 [Bissett and Anderson, 1980b]. The structural genes comprising the lac-operon and Tag-6P pathway have been characterized, cloned and sequenced from several bacterial species [Rosey et al., 1991; Rosey and Stewart, 1992; van Rooijen et al., 1991].
For many years, the D-tagatose-6P pathway was considered to be the primary (if not the sole) route, for the dissimilation of tagatose by microorganisms. However in 2004, studies by Shakeri-Garakani and colleagues [Shakeri-Garakani et al., 2004] revealed an unsuspected PTS-mediated D-tagatose catabolic pathway in certain species of Enterobacteriaceae, including: Klebsiella oxytoca, Klebsiella pneumoniae and Salmonella enterica. For these Gram-negative bacteria, it seemed that sugar-specific EIIA/EIIBCTag proteins catalyzed the PEP-dependent translocation and phosphorylation of tagatose into the cell. Together with the PTS genes, the putative tag-operon encoded additional genes, designated tagK and gatYZ. Comparative sequence alignment of amino acids encoded by tagK, revealed significant homology with enzymes assigned to the large ribokinase family, particularly with D-fructose-1-phosphate kinases and D-tagatose-6P kinases [Bork et al., 1993]. The protein encoded by gatYZ possessed the characteristics of a hetero-dimeric class II tagatose 1,6-BP aldolase. Based largely on functional and sequence relatedness of PTS proteins and metabolic enzymes, Shakeri-Garakani et al. [Shakeri-Garakani et al., 2004] suggested that: (i) Tag-1P (not Tag-6P) was the likely product of PTS activity, (ii) Tag-1P was phosphorylated by the tagK-encoded enzyme (tagatose-1P kinase) to form tagatose 1,6-bisphosphate (Tag 1,6-BP), and (iii) that the bisphosphate derivative was hydrolyzed by Tag 1,6-BP aldolase to yield the glycolytic metabolites, glyceraldehyde 3-phosphate (G3P) and dihydroxyacetone phosphate (DHAP).
In 2013, we reported the existence of a closely related PTS-mediated D-tagatose catabolic pathway in the Gram-positive bacterium Bacillus licheniformis (Bli-TagP) ([Van der Heiden et al., 2013], and fig.1). Although consistent with experimental data, in neither of these reports [Shakeri-Garakani et al., 2004; Van der Heiden et al., 2013] was unequivocal evidence presented to support two crucial hypotheses. First, that the tag-PTS product was in fact Tag-1P and not the isomeric (and more common) Tag-6P derivative. Second, no data were presented to establish the specificity of the putative Tag-1P kinase, nor was it verified that Tag-1P was a specific substrate for this enzyme. The purpose of our investigation was to obtain experimental evidence to validate (or refute) these hypotheses. To this end, we report the first PEP-PTS catalyzed synthesis, and physicochemical characterization of Tag-1P. Importantly, we describe the purification, and substrate specificity of the putative Tag-1P kinase from B. licheniformis.
D-tagatose and D-tagatofuranose-6-phosphate, lithium salt (Tag-6P) were purchased from Carbosynth US LLC, San Diego, CA. Phospho(enol)pyruvate (PEP) monosodium salt, barium acetate (99%), NADH, Triose phosphate isomerase from baker’s yeast (Saccharomyces cerevisiae), Glycerophosphate dehydrogenase Type I (rabbit muscle), and D-fructose 6-phosphate, disodium salt hydrate (Fru-6P) were obtained from Sigma-Aldrich. ATP disodium trihydrate was from Amresco and D-fructose 1-phosphate, sodium salt (Fru-1P) was from Santa Cruz biotechnology.
Staphylococcus aureus subsp. aureus ATCC 25923 and Bacillus licheniformis ATCC 14580 were obtained from the American Type Culture Collection, Manassas, VA. Bacillus subtilis 168 was from the Bacillus Genetic Stock Center (BGSC accession number 1A1). E. coli BL21(DE3) strain (Stratagen, La Jolla, CA) was used to overexpress proteins.
Klebsiella pneumoniae subsp. pneumoniae ATCC 23357 was obtained from the American Type Culture Collection. This strain was used for the enzymatic synthesis of Tag-1P. The organism was grown in a defined medium containing (per liter): Na2HPO4, 7.1g; KH2PO4, 1.5g; (NH4)2SO4, 3g; MgSO4.7H2O, 0.1g and FeSO4.7H2O, 5mg. Filter-sterilized tagatose was added to autoclaved medium to a final concentration of 0.4 % (w/v).
The organism was grown (without aeration) at 37 °C in 3 × 1-liter bottles, each containing 800 ml of medium. After growth to stationary phase (18 h), the cells were harvested by centrifugation (13,000 × g for 10 min at 5 °C) and washed twice in 25 mM Tris-HCl buffer (pH 7.5) containing 1 mM MgCl2.6H2O. The yield was ~2 g wet weight of cells / liter.
The enzymatic synthesis of this novel hexose phosphate was catalyzed via the multi-component PEP-dependent: tag-PTS present in tagatose-grown cells of K. pneumoniae (see fig. 1B). The procedure, with slight modification, is essentially that described previously for the biosynthesis of a variety of 6’-O-phosphorylated disaccharides [Thompson et al., 2001a; Thompson et al., 2001b]. Tagatose-grown cells were added to 5 ml of 25 mM Tris-HCl buffer (pH 8) containing 1 mM MgCl2 to a density of 10 mg dry weight/ml. After chilling on ice, the cells were permeabilized by the addition of 50 µl of a mixture of acetone/toluene (9:1 v/v), and the suspension was agitated vigorously for 30 s on a Vortex mixer. This procedure was performed three times, and the permeabilized cell suspension was then placed in ice. For preparative purposes, 15 such suspensions were prepared. Thereafter, PEP (330 mg) and tagatose (1 g) were dissolved in 12 ml of 25 mM Tris-HCl buffer (pH 8) and, after adjustment to pH 8 with ~ 0.5 ml of 5 N NaOH, water was added to a final volume of 15 ml. Subsequently, 1 ml of PEP/tagatose solution was added to each of the permeabilized cell suspensions to provide approx. 100 µmol PEP and 350 µmol tagatose per reaction mixture. The 15 suspensions were transferred to a 37 °C water bath and, after 1.5 h of incubation (with occasional inversion), the preparations were chilled to 0 °C. During incubation there was usually a decrease of 0.5 units in the pH of the reaction mixture. The suspensions were pooled and cells were removed by centrifugation. The supernatant fluid was adjusted to pH 8.2 with NaOH, and 8 ml of aqueous barium acetate (25 % w/v) was added slowly with continuous stirring. The mixture was chilled on ice for 30 min, and the heavy white precipitate of water-insoluble Ba2+ compounds was removed by centrifugation. The clarified supernatant was filtered through a 0.45 µm pore-size membrane, and 4 volumes of ice-cold absolute ethanol were added. The solution, containing a white flocculent precipitate of the ethanol-insoluble Ba2+ salt of Tag-1P was placed overnight in a cold room.
The barium salt of Tag-1P was collected by centrifugation, and upon drying (at 37 °C) the ethanolic white pellet yielded a yellow/white crystalline residue. Ba2+ ions were exchanged for H+ by addition of 2–3 ml of an aqueous suspension of Bio-Rad AG 50W ×2 (H+-form) resin. Exchange resin was removed by filtration, and after adjustment to pH 7.2 (by addition of NaOH or NH4OH) the solution was frozen and lyophilized. Further purification of Tag-1P and removal of trace contaminants (e.g; nucleotides, Pi and residual PEP), was effected by descending paper chromatography (Whatman 3 MM, 20 h) in a solvent containing n-butanol/glacial acetic acid/H2O (5:2:3 v/v). The location of Tag-1P was visualized as a vivid white zone against a pink background by the Wade-Morgan stain [Wade and Morgan, 1953]. The appropriate strips containing Tag-1P were excised from the chromatogram, and the hexose monophosphate was collected by centrifugal elution with distilled water. The pH of the eluate was adjusted to 7.2 with NaOH or NH4OH, and the solution was frozen and lyophilized. As outlined, the procedure usually yielded 20–30 mg of white crystalline Tag-1P salt.
Thin-layer chromatographic analysis provided evidence for the purity of the compound. When 20–100 nmol of Tag-1P were applied to 0.1 mm layers of Polygram Cellulose MN 300 and eluted using the previously described solvent, only a single white spot was revealed by Wade-Morgan stain. Under these conditions, the migration positions of Tag-1P and Tag-6P were virtually identical. High-resolution negative mode electrospray mass spectra data for Tag-1P were acquired on a Thermo Q-Exactive orbitrap. Calculated for C6H12O9P [M-H]−1, 259.0224; measured 259.0224. 1H, 13C and 31P-NMR spectral data for Tag-1P are presented in the Results section.
All NMR data were collected in a Bruker Avance II, 700 MHz NMR spectrometer, equipped with a QXI xyz-gradient, room temperature probe.
For tagatose chemical shift assignment, two samples of tagatose were prepared: 1) a 200 mM tagatose solution in 90:10 H2O:D2O and 2) a 200 mM tagatose in 50:50 H2O:D2O solution. The pH was measured at 25 °C and adjusted to 6.5 with NaOH. The data were collected at −10 °C. Low temperatures were utilized to ensure sample preservation and the observation of hydroxyl proton signals [Battistel et al., 2014]. Chemical shift assignment was carried out utilizing a series of 2D HSQC-TOCSY experiments (10, 30 and 60 ms of DIPSI-2 isotropic mixing [Rucker and Shaka, 1989]. These experiments also afforded hydroxyl proton assignments, as previously reported [Battistel et al., 2013].
For the enzymatically synthesized tagatose phosphate, a 50 mM aqueous solution, with 10% D2O, was prepared. The pH was adjusted to 7.5 with NaOH, at 25 °C. Chemical shift assignments were carried out as for tagatose at −10 °C and 25 °C (table 1). A pH of 7.5 was required to decrease the OH exchange rate with the solvent and allow observation of the isotope shift in 13C signals. 1H-1H coupling constants were estimated from HSQC traces or from phase sensitive COSY experiments.
For control experiments with Tag-6P, a 3.8 mM aqueous solution 10% D2O pH 7.5 was prepared. Chemical shift assignments were obtained as for tagatose and Tag-1P.
Position of phosphate substitution was determined using three methods: 1) We measured and assigned the OH groups in tagatose and compared them to phosphorylated tagatose; 2) We measured the isotope effect on the 13 C chemical shifts of the carbohydrate ring in a 50:50 H2O:D2O mixture and in 90:10 H2O:D2O and 3) We measured two- and three bond 31P-13C coupling constants (2J13C-31P, 3J13C-31P, respectively, table 1). It has been shown [Eyrisch et al., 1993] that 3J13C-31P are often larger than 2 J13C-31P, thus these NMR observables can help in establishing which carbon atoms are two or three covalent bonds removed from the phosphate group.
When OH signals cannot be directly observed, deuterium isotope effects can be useful in establishing which OH groups are substituted by other functional groups. Deuterium two-bond isotope effects on 13 C can be observed if the hydroxyl proton exchange rate is slow enough (exchange rate constant, kex < |2*Δδ13C|, where Δδ13C is the frequency difference between 13C measured in H2O with respect to that measured in D2O). If the exchange is fast (kex >> |2*Δδ13C|, then a single peak will be observed, however its chemical shift will change depending on the D2O: H2O ratio used.
Because OH groups could not be directly observed at the sample concentration available, the H/D isotope effect was monitored as follows: the 90:10 H2O:D2O, tagatose phosphate sample was used to conduct 13 C 1D NMR experiments. Thereafter, the sample was lyophilized and then dissolved in 50:50 H2O:D2O, pH 7.5 at 25 °C, and 13C 1D data were again collected. Carbon atoms with OH groups in slow exchange with the solvent will display two 13C signals because of isotope shifts originating from two different kinds of 13C atoms: one signal for 13C attached to OH and other signal for 13C attached to OD. Several factors may account for the absence of signal duplication for a given 13C, including: 1) OH is in fast exchange with the solvent. 2) the OH is phosphorylated, alkylated or substituted by some other functional group without an exchangeable hydrogen atom, or 3) the exchangeable hydrogen atom was lost during formation of a sugar ring or a glycosidic linkage. In the first case, the existence of an OH group can be verified by varying the D2O percentage. The presence of an OH group will lead to a change in the apparent 13C chemical shift, whereas its absence will not. In the second case, other NMR observables can be measured to determine which type of substituent is present. In this report, we have measured 13C-31P coupling constants (table 1) and collected 1D 31P and 2D 1H, 31P NMR experiments (Fig. S1B). In the third case, one can rely on available data e.g. chemical shifts, coupling constant of reference compounds and molecular formulae.
The structure of fructose 1-phosphate kinase from Bacillus halodurans (PDBID: 2ABQ) was used to generate fig. 2. Unfortunately, the structure of 2ABQ has not yet been published, nor has the Fru-1P binding site been established. Tag- 1P kinase has not yet been crystallized, and axiomatically the Tag-1P binding site is also unknown. However, because fructose differs from tagatose only in the orientation of the OH moiety at C4, we proceeded to use 2ABQ as a potential model, since it may bear structural resemblance with Tag-1P kinase. The chemical structures for Tag-1P, Tag-6P and Tag 1,6-BP were created in HyperChem software package (Gainesville, FL). The model depicted in fig. 2 was created in PyMOL (the PyMOL Molecular Graphics System, Version 188.8.131.52 Schrödinger, LLC), the magnesium ion and the glycans chemical structures coordinates were modified, with the translate command, such to appear approaching and leaving the protein surface.
The expression plasmids pET28a(+) and pColdTF were purchased respectively from Novagen and Takara. The plasmid pLysS was purchased from Novagen. Restriction endonucleases, Phusion High-Fidelity DNA Polymerase kit and In-Fusion HD Cloning kit were purchased respectively from Promega, Fisher Scientific and Clontech. Oligonucleotides primers were from Eurogentec, Seraing, Belgium.
The constitutive B. subtilis ptsHI operon was amplified by PCR with oligonucleotides primers ptsHI forward 5’-CCCGTCCTGTGGATCCGGGTGTTAGTACGCCGTGC-3’ and ptsHI reverse 5’-CCGGCGTAGAGGATCCCTTACTTATGAAAAAAACCAGACAG-3’ using the genomic DNA of B. subtilis 168 as template DNA. The PCR product was subcloned between the two BamHI sites of pLysS plasmid replacing the T7 lysozyme coding sequence yielding pPtsHI. Construct was verified by DNA sequencing (GIGA, Seraing, Belgium). The pLysS plasmid confers resistance to chloramphenicol (CamR) and contains the p15A origin compatible with that of the pDGTAG plasmid (pBR322 derived-plasmid) previously introduced in the E. coli DH5α strain [Van der Heiden et al., 2013]. This strain, harboring the pDGTAG plasmid expressing the Bli-TagP (see fig.1A), was transformed with pPtsHI or the pLysS plasmid (negative control).
E. coli DH5α strain harboring pDGTAG and either pPtsHI or pLysS plasmid was grown in a defined medium (DM) previously described [Van der Heiden et al., 2013] supplemented with either tagatose 20 mM or glucose 20 mM or desionised H20 (sugar negative control). Spectinomycin and chloramphenicol antibiotics were added in the DM at respectively 50 µg/ml and 12.5 µg/ml. Cultures were performed in 30 ml of the supplemented DM with 600 µl of preculture under agitation (250 rpm), at 28°C. Growth of the cultures was monitored by the increase in absorbance at 600 nm during 40 h.
The fruK2 gene was amplified by PCR with oligonucleotide primers FruK2-UP 5’-TCGAAGGTAGGCATATGATCTATACATGCACGATGAACACCGCAG-3’ and FruK2-RP 5’-ATTACCTATCTAGACTGCAGTTATAAATGAGTGATTTGAATTTGTGAAGCTGTTTG ATC-3’ using the genomic DNA from B. licheniformis ATCC 14580 as the template DNA. The PCR product was cloned into the NdeI and PstI sites of pColdTF yielding pCold-TagK-TFHis6. The gatY gene was amplified by PCR with oligonucleotide primers ALD-UP 5’-AGGAGATATACCATGGCCCTGACAAATACGAAAAAATGCTGCTGGACC-3’ and ALD-RP 5’-GGTGGTGGTGCTCGAGGTATCTTGTCATTGCTCATGCACATC-3’ using the genomic DNA from B. licheniformis ATCC 14580 as the template DNA. The PCR product was cloned into the NcoI and XhoI sites of pET28a yielding pET28a-GatYHis6. The lacD gene was amplified by PCR with oligonucleotide primers LacD-UP 5’-AGGAGATATACCATGGCGAAATCGAATCAAAAAATCGCATC-3’ and LacD-RP 5’-GGTGGTGGTGCTCGAGTTTTCTTTGTTTCCAAGATGTCGC-3’ using the genomic DNA from Staphylococcus aureus subsp. aureus ATCC 25923 as the template DNA. The PCR product was cloned into the NcoI and XhoI sites of pET28a yielding pET28a-LacDHis6. The three resulting constructs were verified by DNA sequencing (GIGA).
E. coli BL21(DE3) cells were transformed with each construct for protein overexpression. Cultures were grown at 37°C under agitation (250 rpm) in liquid 2XYT medium (10 g tryptone, 10 g yeast extract, 5 g NaCl for 1 L of medium) supplemented with ampicillin (100 µg/ml) or kanamycine (50 µg/ml) for respectively the pColdTF and pET28a constructs. Medium for the overexpression of GatYHis6 was supplemented with 0.5 mM ZnSO4. When the absorbance (600 nm) of the cultures reached 0.6–0.8, the overexpression of TagK-TFHis6 was induced by the addition of IPTG 0.5 mM and the culture was incubated overnight at 15°C under agitation (250 rpm). Similarly, the overexpression of GatYHis6 and LacDHis6 were induced by the addition of 1 mM IPTG and cultures were incubated during 3h at 28°C.
Cell pellets of the culture were resuspended in Tris-HCl 50 mM buffer (pH 8.0) and lysed with a high-pressure homogenizer (Emulsiflex C3, Avestin Europe GmbH). The supernatants were recovered after 30 min of centrifugation at 25,000 g. For protein purification, we used an automated two-step Profinia system composed of a Bio-scale Mini Profinity IMAC cartridge followed by desalting cartridge (Biorad). Internal programmed method “Native IMAC and Desalting” was applied.
Elution from the desalting cartridge was performed in buffer Tris-HCl 50 mM (pH 8.0), KCl 2.7 mM and NaCl 137 mM for TagK-TFHis6, in Tris-HCl 20 mM (pH 8.0) for GatYHis6 and in Tris-HCl 20 mM (pH 8.0), NaCl 200 mM for LacDHis6. Concentrations of the purified proteins were determined spectrophotometrically by measuring the absorbance at 280 nm and were 0.7 mg/ml for GatYHis6, 2.5 mg/ml for LacDHis6 and 2.1 mg/ml for TagK-TFHis6. Purity of His-tagged proteins was estimated to be no less than 90%. GatYHis6 and LacDHis6 can be conserved at −20°C at least 6 months without the needed of additive while TagK-TFHis6 requires the addition of 5% glycerol before freezing.
Phospho-tagatokinase and phospho-fructokinase activities of TagK-TFHis6 were measured by coupling the formation of Tag 1,6-BP or fructose 1,6-bisphosphate (Fru 1,6-BP) to NADH oxidation via class I ketohexose 1,6-BP aldolase (LacD), triose phosphate isomerase (TPI) and glycerophosphate dehydrogenase Type I (G3PDH) (fig. 3). The assay contained: 50 mM Tris-HCl (pH 8.0), 10 mM MgCl2, 0.2 mM NADH, 2 mM ATP and variable concentrations of substrate. None-rate limiting amounts of LacD (13 µM), of TPI (49.5 U/ml), and of G3PDH (9 U/ml) were added to the assay mixture. The concentration of TagK-TFHis6 was rate limiting (2.4 µM) in the assay. Enzyme activities were monitored spectrophotometrically at 340 nm during 12 minutes with a microplate reader (Tecan Infinite M200Pro) in a total volume of 200 µl at 25°C.
Fru-1P, Fru-6P, Tag-1P or Tag-6P was added at a final concentration of 2 mM or 10 mM to test the specificity of TagK-TFHis6. For the inhibition assays, the Tag-1P concentration was fixed at 2 mM and Fru-1P, Fru-6P or Tag-6P were then each added to the reaction, at 2 mM or 10 mM. Rate of the reaction with the Tag-1P at 2mM was designated 100% maximum activity. Subsequent assays were expressed as a relative activity according to the formula: Relative activity (%) = [slope (assay)/slope (Tag-1P 2mM)].100. All assays were performed in triplicate.
For the determination of the kinetic parameters, Tag-1P was added at 8 mM, 4 mM, 2 mM, 1 mM, 0.5 mM or 0.25 mM in the mixture assay. Fru-1P was added at 80 mM, 40 mM, 20 mM, 10 mM, 5 mM, 2.5 mM and 1.25 mM. The respective kcat and Km values were calculated by nonlinear fit of the Michaelis-Menten curve with the help of the GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA). Kinetic parameters were determined from three independent experiments.
Tag-6P and Tag 1,6-BP have previously been synthesized and characterized spectroscopically by Eyrisch et al. [Eyrisch et al., 1993]. However, to our knowledge, this is the first report of the synthesis and physicochemical characterization of Tag-1P. Furthermore, the tagatose hydroxyl proton signals have not been assigned, nor have their observation and chemical shifts been utilized for structural characterization. We assigned the hydroxyl protons of tagatose because they enabled us to easily distinguish between α and β anomers due to their characteristic chemical shifts [Battistel et al., 2014] and because their absence can be used to infer functionalization (fig. 4). Additionally, hydroxyl signals can be used to uncover substitution pattern in glycans. As explained in Materials and Methods, the absence of a hydroxyl hydrogen signal may indicate the formation of a glycosidic linkage or a ring, when its absence is considered in conjunction with other data. In the case of tagatose, the molecular formula is known and the hydroxyl group positions were determined. Therefore, we can infer that the absence of hydrogen signal for the group at C6 and the presence of a hydroxyl hydrogen signal at C5, indicates that tagatose exists primarily in the pyranose rather than the furanose form in solution at −10 °C. Conversely, the furanose form would show a hydroxyl signal at C6 but not at C5. Both α and β anomers were detected, in agreement with Freimund et al. [Freimund et al., 1996], however, no furanose or open forms signals were observed under the experimental conditions tested. Hydroxyl assignments for tagatose are presented in table 1. Additionally, tagatose serves as chemical shift reference and helps establish the phosphorylation site. Phosphorylation produces chemical shift changes at the site of phosphorylation and upon neighboring atoms, with respect to its un-phosphorylated form. Thus, this reference enables us to easily establish the phosphorylation locus (vide infra).
Previously, phosphorylation of tagatose has been shown [Eyrisch et al., 1993] to occur at the C1 (when C6 is already phosphorylated) and C6 positions (fig. 5A). Determination of the phosphorylation position is only the first step, because one has also to establish the glycan’s major conformation in solution (e.g; open form, or 5 or 6 member ring formation in either α or β configuration, fig. 5B). Three lines of evidence lead to the conclusion that biosynthesized tagatose phosphate is phosphorylated at C1. First, comparison between HSQCs for tagatose and tagatose phosphate (fig. 4B), shows a pronounced change for H1s,C1 cross peaks but not for H6s,C6. One would expect a pronounced chemical shift change at the site of phosphorylation. This result alone strongly indicates that the phosphate group is at position 1. Second, a 31P-13 C coupling between C1-P (4 Hz); and C2-P (8 Hz) indicate that the phosphate is either attached to C1 or C2 (fig. S1A). The 1H, 31P long-range HSQC experiment (fig. S1B) shows that both 1H’s at position 1 are coupled to phosphorus as well, which further supports the incorporation of a phosphate at either C1 or C2 and not C6. However, the chemical shift change observed upon phosphorylation for 13C signals at C1 and not at C2 or C6 indicates that the phosphate is at C1. Larger values for vicinal vs. geminal 13C-31P coupling constants have been reported and agree with the values presented in this manuscript [Eyrisch et al., 1993; Thompson et al., 1999]. Finally, comparison of the deuterium isotope effect observed on 13C signals for tagatose and tagatose phosphate showed that there is no hydroxyl group at C6 and at C1 and C6, respectively. Fig. 6 shows the deuterium two-bond isotope effect observed in the overlaid 13C NMR spectra of tagatose phosphate in 50:50 H2O:D2O and 10% D2O. The isotope effect is most significant at C4 and C5 (fig. 6A and 6C, respectively), indicating that OH exchange is slow, and that long-range isotope effects from neighboring OH groups may also be present. It is evident from the spectra that no isotope shift is observed at C6 (fig. 6D). As explained earlier in the text, the absence of isotope shift is expected for substituted carbon atoms bearing no exchangeable H's, such as phosphate. Because we demonstrated that there is no phosphate group at C6 of phosphorylated tagatose, therefore, the exchangeable proton in the hydroxyl group at C6 must have been lost while forming a pyranose ring. Thus, this result supports the conclusion that the compound in question is in the pyranose form. It may be argued that another possibility is that a phosphate group has been incorporated at this position, however there is no 31P -13C coupling involving C6. Line broadening and small chemical shift changes are observed at C2 and C3, indicating that OH at C2 and C3 are in fast exchange with solvent (fig. 6B and 6C, respectively).
Moreover, 1H 13C chemical shifts more closely match those of the α anomer of tagatose (table 1 and fig. 4) than the β anomer. Homonuclear 1H,1H coupling constants are also consistent with a pyranose form (table 1), where H3,H4 coupling is small ~3 Hz (2.6, 3.5, 4.8 and 4.7 Hz for the α pyranose, β pyranose, α furanose and β furanose, respectively ([Freimund et al., 1996], fig. 4A). A small coupling constant (0–3 Hz) for H3,H4 is expected if the coupled hydrogen atoms are oriented ca. 90 degrees from each other (pyranose forms, fig. S2A) and larger if the angle between them is ~ 0 (furanose forms fig. S2B). Thus, the above results indicate that the PEP:PTS-generated product is Tag-1P, and that the compound exists primarily in a pyranose form with 1H and 13C chemical shifts indicative of the α anomer.
Additionally, as a control, we collected NMR experiments on Tag-6P under the same conditions as for tagatose and Tag-1P. As expected, 31P -13C coupling constants are observed for C6 and C5 and not for C1 and C2 (table 1).
Enzyme activity was tested with Tag-1P, Fru-1P, Tag-6P and Fru-6P as potential substrates. The specificity of TagK is clearly in favor of the Tag-1P substrate compared to the Fru-1P substrate. When the activity of TagK is fixed to 100 % with 2 mM of Tag-1P as substrate, the relative activity calculated for the enzyme with 10 mM of Fru-1P as substrate is 31%. Because no significant activities on Tag-6P and Fru-6P were detected (fig. 7), it was of interest to test the inhibitory effects of these compounds on the activity of TagK with Tag-1P 2 mM as substrate. As expected, Tag-6P and Fru-6P are inhibitors of TagK but not Fru-1P. The effect is particularly significant with 10 mM of Tag-6P or Fru-6P with a residual activity of 43% and 69% in the assay condition, respectively. Relative activities and standard deviation are depicted in figure 7.
From the nonlinear fit of the Michaelis-Menten curve, the apparent Km and kcat values for Tag-1P, at constant level of ATP (2 mM), were determined to be 3.2 ± 0.4 mM and 0.44 ± 0.01 s−1, respectively. For Fru-1P, the apparent Km was 6.2 ± 0.6 mM and the apparent kcat was 0.021 ± 0.001 s−1. The catalytic efficiency (apparent kcat/apparent Km) of the enzyme for Tag-1P (137.5 M−1s−1) was 40-fold higher than that for Fru-1P (3.4 M−1s−1) confirming the enzyme as a Tag-1P-specific kinase (TagK).
In a previous report, we showed that E. coli DH5α (initially Tag− phenotype) metabolized tagatose after transformation with the pDGTAG vector harboring the Bli-TagP (Tag+ phenotype) (fig. 1A) [Van der Heiden et al., 2013]. However, growth of the transformant was poor, and we suspected that HPr from E. coli might be unable to transfer the phosphate group to the EIIAtag of B. licheniformis (fig. 1B). To test the hypothesis, DH5α was then transformed with a second plasmid expressing the ptsHI operon of B. subtilis (pPtsHI plasmid). The E. coli DH5α strain harboring both pDGTAG and pLysS grows considerably more slowly in the DM supplemented with tagatose, than with glucose. This observation is consistent with our previous results (fig. 8A) [Van der Heiden et al., 2013]. As we hypothesized, the growth rates of E. coli DH5α harboring both pDGTAG and pPtsHI plasmids were essentially the same in DM supplemented with either tagatose, or glucose as carbohydrates sources (fig. 8B).
Results from the present investigation augment our knowledge of the mechanisms of glycans transport and metabolism in microorganisms. Importantly, by enzymatic synthesis and physicochemical analyses, we validate the first of the hypotheses presented in the Introduction, by confirming that Tag-1P is the product of the PEP-dependent PTS transport of tagatose by cells of B. licheniformis and K. pneumoniae. Physicochemical analysis involved primarily conventional 1H, 13C, and 31P NMR experiments with the unprecedented incorporation of hydroxyl signal and 2H/1H isotope effects for the determination of ring size and phosphorylation position.
Validation of the second hypothesis required the synthesis of Tag-1P in substrate amount, and expression in active form of the protein YP_006714842 encoded by the fruK2 gene (renamed tagK hereafter) of the tagatose gene cluster from B. licheniformis (fig. 1A).
Initial attempts to obtain the wild type Tag-1P kinase were unsuccessful due to the production of insoluble protein inclusion bodies. However, fusion of YP_006714842 with the ‘trigger factor’ (TF) from E. coli plus a His6-tag enabled the expression, and purification of a soluble and enzymatically active fusion protein, TagK-TFHis6. The second hypothesis (see, Introduction) was confirmed by the finding that Tag-1P is a substrate phosphorylated by the ATP and Mg2+-dependent Tag-1P kinase (TagK-TFHis6). In addition, specificity and inhibition assays reveal that the 1P-derivatives of tagatose and fructose are substrates of the TagK-TFHis6 enzyme, whereas the corresponding 6P-isomers (Tag-6P and Fru-6P) are inhibitors. Since the catalytic efficiency with Tag-1P as substrate is 40 fold higher than that with Fru-1P, it is clear that the specificity of the TagK-TFHis6 enzyme is in favor of Tag-1P. For that reason, and because of the significant amino acid sequence identity (39%) between the B. licheniformis fruK2 product (YP_006714842) and the K. oxytoca M5a1 tagK product (AAL60166), we propose to rename the fruK2 gene of B. licheniformis as tagK, and its product FruK2 (TagK).
Our NMR analyses show that in solution, Tag-1P exists primarily in the pyranose conformation. However, we cannot discount the presence of minor concentrations of the furanose, open-chain or other tautomeric forms of the hexose phosphate, that were undetected by NMR. Because there are no reported structures for Tag-1P-kinase complex, the molecular structure of the compound that actually binds to Tag-1P kinase is thus uncertain. If (as seems reasonable), the pyranose form of Tag-1P is recognized (see, Fig. 2), then this ring must open to assume a furanose conformation in order to permit ATP-dependent phosphorylation to occur at C6, thus yielding Tag 1,6-BP. A binding preference for a pyranose as opposed to the furanose ring is consistent with the higher catalytic efficiency observed for Tag-1P vs. Fru-1P. Presently, there is no evidence to suggest the presence of a single catalytic site, or whether separate domains facilitate the sequential binding, ring-opening, and phosphorylation of Tag-1P. Solution of the X-ray structure of Tag-1P kinase, and co-crystallization of the enzyme with its ligands (Mg2+, ATP and Tag-1P) may clarify these issues and provide insight to the catalytic mechanism.
As noted earlier (see, Introduction), the PTS-mediated D-tagatose catabolic pathway of B. licheniformis is closely related to that of some enteric bacteria [Shakeri-Garakani et al., 2004; Van der Heiden et al., 2013]. The tagatose-specific PTS protein constituents of B licheniformis are EIIBCTag and EIIATag, which belong to the PTS fructose-mannitol (Fru) family [Saier, 2000]. As EIIBCTag of K. oxytoca encoded by tagT (EIIBCTag,Kox or TagT, AAL60167), EIIBCTag of B. licheniformis encoded by the fruA2 gene (EIIBCTag,Bli, YP_006714843) consists of two domains: IIB and IIC. Both EIIBC proteins exhibit 54% of identity. In contrast, EIIATag,Bli encoded by the orf 51 (Genbank accession number for the B. licheniformis ATCC 14580 genome: NC-006322) has no homology with the EIIATag,Kox (TagH), encoded by the tagH gene, which is a EIIA-pseudoHPr-like protein. This means that the EIIATag,Kox is constituted of a EIIA domain (from the Fru family) followed by a HPr-like domain also called TPr [Shakeri-Garakani et al., 2004]. This domain is absent from the EIIATag,Bli component. Genes encoding for the anionic cell wall polymers synthesis are already named tagT and tagH in B. subtilis [Kawai et al., 2011] and are also found in the B. licheniformis genome. For these reasons, we suggest renaming of orf 51 encoding EIIATag,Bli as tagM, and the gene product itself TagM. In addition, the fruA2 gene should be renamed tagL and its product, the EIIBCTag,Bli, TagL. A gene (orf 48) encoding for a regulator from the RpiR family is also present in the tagatose gene cluster (Fig.1A). The RpiR family regulators contain a N-terminal helix-turn-helix DNA binding motif and a C-terminal SIS (Sugar ISomerase) domain, which is found in many phosphosugar isomerases and phosphosugar binding proteins [Yamamoto et al., 2001]. We propose also to rename this orf 48 in tagR.
The class I ketohexose 1,6-BP aldolase (LacD) from S. aureus was chosen as the second enzyme in the coupled enzyme assay (fig. 3), because this class of enzyme is nonspecific with respect to the configuration of hydroxyl groups at carbon atoms 3 and 4. LacD is thus able to cleave both Tag 1,6-BP and Fru 1,6-BP (C4-isomers) into DHAP and G3P [Bissett and Anderson, 1980b]. This enzyme allowed us to test fructose and tagatose-derivatives in the assays. However, if LacD was replaced by the B. licheniformis Zn2+-requiring class II Tag 1,6-BP aldolase GatYHis6 (5.4 µM) in the coupled enzyme assay, then a decrease of NADH absorbance was also measured with Tag-1P (2 mM) used as substrate for TagK-TFHis6 (slope similar to that measured with LacD). No NADH absorbance decrease was detected if Fru-1P (10 mM) is used as substrate for the TagK-TFHis6 enzyme (data not shown). Our results show that whilst GatYH6 is highly active on Tag 1,6-BP, the enzyme exhibits poor to negligible activity upon Fru 1,6-BP. Similarly, the high specificity of class II Tag 1,6-BP aldolases from enteric bacteria for Tag 1,6-BP and a low one for Fru 1,6-BP have been highlighted in previous papers [Brinkkotter et al., 2002; Eyrisch et al., 1993]. In this context, our results establish that GatY is the functional enzyme that cleaves Tag 1,6-BP (to DHAP and G3P) in the PTS-mediated D-tagatose catabolic pathway of B. licheniformis (fig. 1B).
In our previous paper, we obtained evidence for a function role of Bli-TagP during supplementation, and transformation experiments, with E. coli DH5α and B. subtilis 168 that were initially unable to use tagatose as a carbohydrate source (Tag− phenotype) [Van der Heiden et al., 2013]. Interestingly, the growth rates of a stable Bli-TagP transformed B. subtilis 168 (Tag+ phenotype) in the presence of D-tagatose or the PTS sugar D-glucose were similar. Identical growths were also observed with B. licheniformis owing the Bli-TagP in its genome (fig. 1A). In contrast, the growth rate of E. coli DH5α, harboring the Bli-TagP on the pDGTAG plasmid (Tag+ phenotype) in the presence of D-tagatose was rather similar to that observed in presence of the non-PTS sugar D-galactose [Van der Heiden et al., 2013]. Our previous results suggested the possibility that the phosphate transfer between the endogenous protein HPr from E. coli to the exogenous EIIATag,Bli was not efficient. Support for this hypothesis is provided in this study, by the results of transformation of E. coli DH5α (pDGTAG) with a second plasmid (pPtsHI) encoding the PTS general cytoplasmic components HPr and EI of B. subtilis. The constitutive ptsHI operon of B. subtilis [Gonzy-Treboul et al., 1989] was chosen for the construction of this second plasmid because HPr proteins from B. subtilis and from B. licheniformis present 98% of identity. E. coli harboring Bli-TagP and exogenous HPr from B. subtilis is now able to catabolize tagatose at the same rate than glucose (fig. 8B). This result demonstrates that HPr is sufficient for the phosphate transfer to EIIATag,Bli in the Bli-TagP, and is consistent with the fact that a TPr domain is lacking in the EIIATag,Bli component. Interestingly, our results contrast with functional heterologous interactions reported between HPr of E. coli and glucose-specific enzyme EIIAGlc of B. subtilis [Reizer et al., 1992]. Nevertheless, it is well known that precise electrostatic interactions assure correct amino acid orientation for in-line phosphotransfer reaction. Further investigations should be undertaken to study the residues implicated in the interactions between both proteins EIIATag,Bli and HPr in B. licheniformis, in order to compare data to previously reported interfaces required for interactions between HPr and EIIAGlc of E. coli, and between HPr and EIIAGlc of B. subtilis [Herzberg, 1992; Wang et al., 2000].
Studies by Koh and colleagues [Koh et al., 2013] demonstrated that probiotic strains such as Lactobacillus Rhamnosus GG and Lactobacillus casei catabolize tagatose via tagatose-specific PTS. This study highlights the potential of tagatose as supplement (synbiotic partner) for L. casei and/or L. Rhamnosus strain GG in dairy foods and therapeutic dietary regimens. The study reports differential expression of genes of Lactobacillus Rhamnosus GG (ATCC 53103) in the presence of tagatose compared those present in glucose-grown (control) cells. For example, tagT and fruK gene expressions were dramatically increased 31 and 56-fold, respectively. The tagT gene (LGG_02647) codes for the PTS system fructose-specific transporter subunit IIABC, and the fruK gene (LGG_02648) encodes a 1-phosphofructokinase [Koh et al., 2013]. The IIABC protein comprises domains B, C and A. Whereas we found 54% identity between IIBCTag,Bli and IIBCTag,Kox, we found that the IIABC exhibits 53% identity with IIBCTag,Bli and 47% identity with IIBCTag,Kox. In addition, FruK protein from L. Rhamnosus exhibits 39% amino-acid sequence identity with both TagK of K. oxytoca and B. licheniformis. Presumably, a PTS-mediated D-tagatose pathway including a Tag-1P intermediate is also extant in other Gram-positive strains, including nutritionally important probiotic organisms, such as L. casei and L. Rhamnosus.
This work was supported in part by the Intramural Research Program of the NIDCR, National Institutes of Health, Bethesda, MD. Edwige Van der Heiden was the recipient of a FRIA (Fonds de la Recherche pour l’Industrie et l’Agriculture) fellowship. We thank Eric Anderson, Noel Whittaker and Andreas Pikis for mass spectrometry data and technical assistance. We thank Régine Freichels for the gift of LacD enzyme.