The heat-inducible lysyl-tRNA synthetase from Escherichia coli (LysU; EC6/1/1/6.html) converts ATP to diadenosine tri- and tetraphosphates (Ap3A/Ap4A) in the presence of l-lysine/Mg2+/Zn2+. To understand LysU in more detail, 26 mutants were prepared: six of E264, four of R269 and sixteen mutants by alanine-scanning of the inner shell/motif 2 loop. In the presence of glycerol and absence of exogenously added Zn2+/l-lysine, we unexpectedly found that E264K catalysed the production of glycerol-3-phosphate, powered by ATP turnover to ADP. E264Q and E264N are also capable of this activity, but all three show little formation of Ap4A/Ap3A under normal conditions (additional Zn2+/l-lysine/Mg2+). By contrast, wild-type LysU has a weaker glycerol kinase-like capability in the absence of Zn2+ and is dominated by Ap4A/Ap3A synthesis in its presence. Kinetic and isothermal titration calorimetry results suggest that E264 is a crucial residue for Zn2+ promotion of Ap4A/Ap3A synthesis. This is consistent with the hypothesis that E264 provides an anchor point for a Zn2+ ion complexed to the active site, with simultaneous coordination to the enzyme bound lysyl-adenylate intermediate and secondary substrate ATP/ADP. The glycerol kinase-like activity is uncovered on disruption of this specific coordination.
Aminoacyl-tRNA synthetases are a heterogeneous family of 20 distinct housekeeping enzymes familiar for their role in protein synthesis. They catalyse the specific esterification of amino acids to the 2′/3′-hydroxyl group of the 3′-terminal adenosine of appropriate cognate tRNA(s) during the translation process. There are two distinct classes based on their conserved sequence motifs within the active site topologies . Class I synthetases have a catalytic domain based on a classical Rossman nucleotide-binding fold, to which ATP binds in an extended conformation. Class II synthetases are built around an antiparallel β-fold surrounded by α-helices and bind ATP in a bent conformation. Escherichia coli lysyl-tRNA synthetase is a class II synthetase and, unusually, exists in two distinct isoforms, LysS and LysU . These share a high degree of sequence identity (88%) and similar aminoacylation activities, but are regulated differently. LysS is constitutively expressed under normal growth conditions and appears to be responsible for the tRNA charging activity, whereas LysU is the product of a normally silent gene that is induced to high expression during heat shock [3, 4]. Although LysU is also capable of tRNA charging, it is an unusually effective catalyst of diadenosine 5′,5′′′-P1,P4-tetraphosphate (Ap4A) and diadenosine 5′,5′′′-P1,P3-triphosphate (Ap3A) sequential formation from ATP [5-7]. The combined mechanisms of LysU-mediated Ap4A and Ap3A synthesis are illustrated (Scheme 1). Both Ap4A and Ap3A are thought to be associated with a wide variety of inter- and intracellular signalling/modulatory mechanisms, although many roles require further investigation [8-13].
The involvement of Zn2+ is unclear given that LysU's sequence does not harbour any known consensus binding sites for this ion. Because free Zn2+ in solution readily leads to protein aggregation, the absence of an obvious and tightly binding site suggests transient coordination, enabling catalytic effect but not releasing significant amounts of free Zn2+ on product formation. A possible answer may be provided by the diadenosine polyphosphates . Ap4/5A are known to chelate divalent metal ions (Zn2+ in particular) via an N7–N7 binding mechanism (Scheme 2A) and Ap3A does the same, albeit less effectively. Such a mechanism may be predicted to enhance reaction between the lysyl-adenylate intermediate and a second nucleotide substrate (ATP/ADP) docked adjacent to it in the active site [6, 15].
The first stage of lysyl-adenylate formation is mechanistically well characterized. LysU is homodimeric with one active site associated with each monomeric polypeptide. Previous order-of-addition experiments [16, 17] show that LysU is initially loaded at each site with Mg2+ and l-lysine, giving a LysU:6Mg2+:l-lysine2 intermediate state with significant conformational distortion. Thereafter, ATP binds to a single active site (designated 2) developing a structural asymmetry culminating in distortions in the motif 2 loop at site 1. As a result, this site cannot bind ATP until the first lysyl-adenylate intermediate is generated in site 2. The turnovers of the active sites are thus out of step with respect to each other, a situation described as half-of-sites reactivity. The most obvious differences between the site geometries are the relative positions of R269. In site 2, R269 appears to act as a lid to close the binding pocket; in site 1, this lid has been displaced leaving the pocket open. Other active site/first shell amino acid residues of potential importance to Ap4A/Ap3A formation can be seen by inspection of the molecular dynamics simulation  and X-ray crystal structures  of LysU with bound substrates (Fig. 1A). With bound adenosine 5′-(β,γ-methylene)triphosphate (AMPPCP; an inert β,γ-methylene analogue of ATP) there is co-association with three Mg2+ ions: Mg2+(1) bridges the α- and β-phosphate residues of AMPPCP to amino acid residues E414 and E421; Mg2+(2) bridges the β- and γ-phosphates to E421 (and perhaps D376); and finally Mg2+(3) bridges the β- and γ-phosphates to E264 (Fig. 1B). Following lysyl-adenylate formation: Mg2+(1) is left bridging the α-phosphate residue of this intermediate to E414 and E421; whereas Mg2+(3) appears to shift away from E264, associating with the α- and β-phosphates of the pyrophosphate leaving group (Fig. 1C). Mg2+(2) is lost, leaving an ‘anionic pocket’ near E264 and the adenine-N7 of lysyl-adenylate [6, 15]. Accordingly, the motif 2 loop and these other named amino acid residues should play significant roles.
Here is described an attempt to explore LysU catalysis with a range of mutations involving E264, R269 and an alanine scan of the remaining residues in the motif 2 loop and other key inner shell positions. These investigations resulted in a more detailed understanding of the purpose and position of the Zn2+ ion within the active site. It is also noted that certain E264 mutants lose the capacity for Zn2+-promoted Ap4A/Ap3A formation, but demonstrate an unexpected glycerol kinase-like activity.
Twenty-six LysU single mutants were prepared: six of E264, four of R269 and sixteen in the inner shell/motif 2 loop. As with LysU WT (wild-type), the 26 mutants were overexpressed from plasmid pADH2 in an E. coli lysU deletion strain lysU2-17A, then purified as described previously [16, 18]. This procedure produced an average of 40–50 mg of > 95% pure, active protein per litre of growth culture. Structural characterization of mutants was necessary to prove integrity in advance of binding and catalysis studies. Using CD spectroscopy, all proteins were judged to have similar percentages of the main secondary structural elements: 37% α helix, 26% β sheet and 38% random-coil according to k2D analysis . CD and fluorescence titration experiments were then employed to follow the unfolding transitions of these LysU mutants in comparison with wild-type by monitoring ΔΔA222 (α helix minimum) as a function of temperature or ΔI340 (tryptophan fluorescence from W115 and W365) as a function of guanidine hydrochloride (GuHCl) concentration. Thermal denaturation temperature (Tm) and unfolding transition midpoints (Cm) were found to be equivalent to wild-type (Table S1). In the absence of significant changes, the LysU mutants appear to possess appropriate structural integrity relative to LysU WT. Consequently, any differences in catalytic behaviour were interpreted as emanating from the corresponding mutations.
Glycerol kinase-like activity
After purification, recombinant LysU was stored at −20 °C with glycerol (20% v/v) as a stabilizing agent. LysU stored in this fashion was known to be initially inactive with respect to Ap4A/Ap3A synthase activities and only regained catalytic action after 3–4 days of dialysis against a suitable buffer (normally 4 L of 50 mm Tris/HCl buffer, 2 mm β-mercaptoethanol, pH 8.0, at 4 °C). This has previously been accorded to the presence of trace nucleotides blocking the active site. Recently it was observed that insufficiently dialysed LysU WT was capable of slow but substantial ATPase activity, despite the absence of extraneously added l-lysine, Mg2+ or Zn2+ ions (X. Chen, M. Wright & A. D. Miller, unpublished results). This activity was not present with more extensively dialysed material. This observation suggested that LysU might harbour an additional catalytic capability similar to a phosphate transferase or kinase. After preparing enzyme without the use of glycerol as a stabilizing agent, a series of simple kinetic ATP-to-ADP turnover studies was carried out using LysU WT or mutants in the presence of glycerol, but in the absence of added Zn2+ or l-lysine (Fig. 2A). LysU WT and mutant mediated sequential catalysis of Ap4A/Ap3A in the presence of added l-lysine, Mg2+ and Zn2+ ions, but in the absence of added glycerol was also studied (Fig. 2B).
With respect to ATP-to-ADP turnover, a full range of mutant effects were observed that are categorized as follows: W-type mutations, caused an increase in kcat catalytic efficiency relative to wild-type; X-type mutations, caused a loss in kcat catalytic efficiency of > 50% compared with wild-type; Y-type mutations, caused a loss in kcat catalytic efficiency of > 90% compared with wild-type; and Z-type mutations, caused a complete loss in catalytic turnover (Table 1). Of these, the E264K mutation (W-type) is the most significant. This promotes enhanced ATP turnover by approximately threefold compared with wild-type and has almost completely lost Ap4A/Ap3A synthase capabilities. HPLC assay data demonstrate this clearly (Fig. 3). ATP turnover was subsequently shown to result from phosphorylation of glycerol by HPLC isolation of a low molecular mass product from the E264K reaction, using a shallow gradient of triethylammoninium hydrogen carbonate (0–10% w/v over 15 min). This product (yield > 95%) was identified as glycerol-3-phosphate by ESI-MS, 1H-NMR and 13C-NMR spectroscopy analyses (see 'Materials and methods').
Table 1. Classification of LysU mutants by glycerol kinase activity
Classification of mutation
Mutation – putative function of position
W-type: increased catalytic efficiency relative to wild-type
E264K/N/Q – H-bond/salt links to Mg2+, adenine-NH2
H270A – H-bond/salt links to γP of ATP, adenine ring of ATP
X-type: > 50% loss in catalytic efficiency compared with wild-type
F261A – structural residue pre N-terminal motif 2 loop
Similar kinetic analyses were then performed using hydrolysis-resistant analogues adenosine 5′-(β,γ-methylene) triphosphate (AMPPCP) or adenosine 5′-(α,β-methylene) triphosphate in the place of ATP. Consistently, no glycerol-3-phosphate or adenosine diphosphates were formed. Experiments were also performed with propane-1,2-diol, propane-1,3-diol, ethane-1,2-diol and d,l-isopropylidene glycerol in place of glycerol, but again no ADP or phosphorylated products were formed. Finally, turnover experiments were performed in the presence of pyrophosphate (PPi) at twice the concentration of the ATP (5 mm). In this case, ATP-to-ADP turnover was reduced 10-fold, indicating a PPi-competitive inhibition of ATP binding. These results suggest that LysU WT and (to a greater extent) the E264K/N/Q mutants possess the capability for a substantial glycerol kinase-like activity that is not Zn2+ dependent. In effect, phosphorylation of glycerol appears to represent a LysU function alternative to the catalysis of Ap4A/Ap3A formation, with the prevalence reaction depending on presence (or not) of co-enzymes and metal cations.
Zn2+ ions in Ap4A/Ap3A synthase activities
Zn2+ is indispensable to the secondary step of Ap4A/Ap3A synthesis, increasing production rate by at least an order of magnitude. Considering the likely effects of N7–N7 bridging (Scheme 2A), the role of Zn2+ mediated catalytic enhancement could be explained in terms of two combined effects (Scheme 2B): (a) co-association of lysyl-adenylate intermediate and a second nucleotide substrate (ATP or ADP) in close proximity due to N7–N7 bridging; and (b) electrophilic catalysis of the second-stage coupling reaction made possible by polarization of the α-phosphate of the intermediate. Such an arrangement could then be accommodated within a LysU active site during catalysis. As described above, lysyl-adenylate intermediate formation is associated with an increase in active site exposure and dissociation of Mg2+(2). This dissociation opens up an anionic pocket around E264 that may act as an anchor for Zn2+. This would then bind simultaneously to LysU, the lysyl-adenylate α-phosphate and adenine-N7, and also the adenine-N7 of the second nucleotide substrate (ATP or ADP) via N7–N7 bridging. In addition, the second substrate may also occupy the PPi-binding site position upon its vacation post intermediate formation. This would bring it into juxtaposition with the lysyl-adenylate intermediate.
The validity of this hypothesis was investigated by studying LysU-mediated sequential catalysis of Ap4A and Ap3A in the presence of added l-lysine, Mg2+ and Zn2+ (160 μm), but in the absence of glycerol (Fig. 2B). As previously noted, both Ap4A and Ap3A synthase activities involve consecutive, essentially identical mechanisms and are both Zn2+ ion dependent . Because the formation rate of Ap3A is slow and directly dependant on that of Ap4A, it is simpler to only describe the synthase rates for the latter. The effect of LysU mutations on the catalysis of Ap4A formation may be divided into three categories: A-type mutations, no discernable or substantial effects; B-type mutations, differential loss in catalytic efficiency of < 50% compared with wild-type; C-type mutations, differential loss in catalytic efficiency of > 50% compared with wild-type (Table 2). The majority of A-type mutations are motif 2 loop structural amino acid residue mutations. B-type mutations are also largely motif 2 loop structural mutants with the notable exception of mutation E264D. Finally, C-type mutations mostly involve clear functional amino acid residue mutations of which E264K, E264N, E264Q, E264R and E264A were all notable. These results strongly suggest a functional role for E264 in LysU-mediated catalysis of Ap4A/Ap3A formation promoted by Zn2+, a role partially retained in the conservative E264D mutant.
Table 2. Classification of LysU mutants by Ap4A synthase activity
Classification of mutation
Mutation – putative function of position
*Mutations are numerically borderline although their overall properties fit the chosen classification.
C-type: differential loss in catalytic efficiency of > 50% in Ap4A synthesis compared with wild-type
R262A – H-bond/salt links to αP of ATP
E264K/N/Q/R – H-bond/salt links to Mg2+(3), adenine-NH2, Zn2+
E264A – as above (*)
H270A – H-bond/salt links to γP of ATP, adenine ring of ATP (*)
E273A – structural residue post motif 2 loop
E414A – H-bond/salt links to Mg2+(1)
E421A – H-bond/salt links to α-, β-, γPs of ATP, α-, βPs of PPi, and Mg2+(2)
R480A – H-bond/salt links to γP of ATP and βP of PPi
First nucleotide substrate binding studies
Further evidence was obtained using isothermal titration calorimetry, a direct method for the determination of binding thermodynamics. This technique was used to investigate the effects of key mutations on the first stage of Ap4A/Ap3A formation, namely lysyl-adenylate intermediate formation. By studying the interaction between AMPPCP and LysU WT or mutants (E264K, E264N and E264Q) in the presence of Mg2+ and l-lysine, we observed that the mutants possessed only slightly reduced binding affinities relative to wild-type (Table 3). Otherwise binding and stoichiometry data were very similar to wild-type data reported here and previously [16, 17]. The use of AMPPCP avoids any interference from hydrolytic breakdown during the studies and allows for the inclusion of Mg2+ and l-lysine in the experimental buffer without risk of lysyl-adenylate formation. Under normal conditions, Mg2+ and l-lysine precede the binding of ATP and their inclusion is essential for ATP binding to take place at all . Because none of the E264 mutations investigated appear to have any substantial perturbation of initial substrate binding, it seems likely that they also have little impact upon the enzymes' capacity for intermediate formation. Unfortunately, this is difficult to study directly but it appears that the mutation effects will be largely confined to the second stages of Ap4A/Ap3A formation.
Table 3. Thermodynamic binding parameters for the interaction between AMPPCP and LysU wild-type or mutants involved in glycerol kinase activities
Stoichiometry to monomer
0.42 ± 0.03
1.49 ± 0.57
−3.0 ± 0.33
0.46 ± 0.06
4.3 ± 1.7
−1.7 ± 0.31
0.34 ± 0.10
4.7 ± 1.0
−2.53 ± 0.98
0.35 ± 0.08
8.31 ± 0.11
−0.45 ± 0.16
LysU is well established as an aminoacyl-tRNA synthetase with substantial secondary capabilities as an Ap4A/Ap3A synthase in the presence of Zn2+. Both functions are interlinked in mechanistic terms, although the biological reasons remain to be established. Here, we report on an additional glycerol kinase-like activity, for which the mechanistic relationship is also unclear. ATP-to-ADP turnover is observed by HPLC for LysU WT and 11 of the mutants, leading to the formation of glycerol-3-phosphate. Of these 11, E264N, E264Q and (especially) E264K demonstrated enhanced characteristics compared with wild-type, slightly weakened binding interactions with ATP and poor Ap4A/Ap3A production (Table 3). A further active mutant H270A also shows poor Ap4A/Ap3A synthesis activity, which is expected given the role of H270 in forming H-bond/salt links to the γ-phosphate of the first nucleotide ATP substrate, and in binding the adenine ring of the same. These results may be interpreted with the assistance of the X-ray crystal structure of the LysU active site following the binding of l-lysine and AMPPCP or lysyl-adenylate and PPi . These images were obtained from LysU prepared and crystallized in the presence of glycerol, so it is unsurprising that there are several molecules embedded within the protein structure. It should be noted that one of these is within the LysU active site and has a confirmation suggesting interactions with the bound ligands and their associated Mg2+ ions (Fig. 1B,C). This would likely also be present in LysU insufficiently dialysed from 20% glycerol stocks or in the presence of exogenous glycerol. This may allow the observed ‘glycerol kinase’ activity.
The bent conformation of AMPPCP seen in LysU is not appropriate for γ-phosphate transfer to the glycerol (compare the extended ATP in E. coli glycerol kinase; Fig. 1D) . Order of addition experiments have already shown that Mg2+ with l-lysine associate with both LysU active sites first and reorganize them prior to arrival of the first ATP substrate . Given that the resulting lysyl-adenylate intermediate appears to be well bound (it has not been reliably isolated) and that all X-ray structures of LysU  with a full set of ligands show the reaction stalled either just before intermediate formation (with AMPPCP or adenosine 5′-(α,β-methylene) triphosphate) or just after (with ATP), it is likely that LysU is purified in tight association with the lysyl-adenylate and its associated Mg2+ ions. On addition of ATP it would be free to associate with the active site at the location vacated by PPi dissociation. In the presence of Zn2+ the normal reaction to form Ap4A will occur (with concurrent release of l-lysine), but in the absence of Zn2+ an alternative reaction would be transfer of the ATP γ-phosphate to the local glycerol (Scheme 2C). This would be followed by release of glycerol-3-phosphate and ADP prior to continued ATP and glycerol turnover. In a reaction mixture containing ATP, l-lysine, Mg2+, Zn2+ and trace glycerol (e.g. due to insufficient dialysis) these two reactions would compete, leading to an apparently slow or stalled Ap4A/Ap3A synthase reaction. Storage of LysU either at −20 °C in the absence of glycerol or at 4 °C as a (NH4)2SO4-induced precipitate avoids this problem.
In all mutants, motif 2 loop/first shell amino acid residue mutations R262A, E414A, E421A and R480A substantially reduce Ap4A formation to the point of absence (Fig. 2B and Table 2). It is likely that these mutations substantially impair adenine, Mg2+ or phosphate-binding associations required for lysyl-adenylate formation (Scheme 1). Otherwise, the E264 mutants are notable since the wild-type residue was pinpointed hypothetically as a potentially important substrate/metal ion-binding site (Scheme 2B). E264 mutations (except E264D, a B-type mutation that is the most conservative mutation in this position) led to a collapse in Ap4A synthase activity to a basal level, usually observed in the absence of added Zn2+ ions (Table 2). These results are apparently consistent with E264 being a functionally important amino acid residue in Zn2+-induced LysU-catalysed Ap4A/Ap3A synthesis, for instance, by acting as an anionic Zn2+ binding site for catalysis as proposed (Scheme 2B). Enzyme-associated Zn2+ ions may bring both lysyl-adenylate and the second nucleotide substrate into close proximity by N7–N7 bridging, and to activate the α-phosphate of the lysyl-adenylate intermediate for substitution under Lewis acid catalysis. E264D is an active but weaker Ap4A synthesis catalyst than wild-type, an observation that can be explained if the homologous D264 residue functions in the same way as E264 but less effectively, owing to foreshortening of the side chain by a methylene group.
Although the core structures of aminoacyl-tRNA synthetases are highly conserved, there is accumulating evidence of multifunctionality. These include activities in amino acid biosynthesis , cellular signalling , RNA splicing , DNA translation regulation , inflammation , angiogenesis , apoptosis regulation  and virus assembly . Such a diverse collection of functions lead to the suggestion that aminoacyl-tRNA synthetases are mechanistically flexible, able to adapt non-canonical substrates by the alteration of only a few amino acid residues in the active site scaffold [29, 30]. Recently Ibba and coworkers reported the reengineering of Bacillus cereus LysRS by site-directed mutagenesis. One of the resulting mutants (A233S) showed both improved recognition of β-lysine over wild-type and successful formation of β-lysyl-tRNA . In a similar fashion, Guo et al. have used site-specific alteration of Methanococcus jannaschii TyrRS to incorporate an α-hydroxyacid analogue of l-tyrosine into proteins in E. coli . Yeast TyrRS mutant (Y43G) has also been developed by Nishikawa and coworkers, able to utilize several 3-substituted tyrosine analogues as substrates for aminoacylation . Although the biological relevance of these kinds of structural adaptations is matter of speculation, the ability of aminoacyl-tRNA synthetases to support multiple catalytic or noncatalytic activities is clear.
There are several questions that emerge for future consideration. First, given the evidence for LysU as a multifunctional enzyme, are there others that can be shown to be similarly diverse? It is known that a given fold (such as the β barrel) can catalyse a variety of different chemistries involving different enzymes (e.g. isomerase, mutase, kinase, racemase, reductase, deaminase, etc.). Perhaps LysU represents an example of an alternative scenario, where single enzymes constructed from single folds have the capacity to promote a variety of significantly different chemistries depending upon surrounding conditions. Second, the E264K mutation (and to a lesser extent E264Q and E264N) demonstrates how a point mutation in an enzyme structure can lead to a fundamental shift in activity (from Ap4A/Ap3A synthase to glycerol kinase). Although LysU WT has both capabilities, the E264K mutant loses an active site residue important for Zn2+ ion promotion of Ap4A/Ap3A synthesis allowing the apparent promotion of its kinase activity. This illustrates the potential for point mutations of enzymatic structure to lead to the stepwise evolution of new catalytic functions from a given fold.
In chemical terms, it would be interesting if further chemical synthesis could be achieved with the glycerol-3-phosphate product. Should direct acylation with fatty acids prove to be viable, then E264K LysU would provide a quick and relatively easy enzymatic route to the precursor for phosphatidic acids. This method of phosphatidic acid synthesis would not require the use of acyltransferases, which are limited to specific classes of substrates. Lapidot and Selinger have reported a method of direct acylation without the use of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide as a coupling agent which may be modified and applied to this work in the future . From our results, it appears that substrates bearing just one or two hydroxyls are not able to undergo phosphorylation. It would be of interest to determine whether substrates bearing three or more hydroxyls or other nucleophilic groups might do. These may include dl-glyceraldehyde, (±)-3-amino-1,2-propanediol, 1,2,3-heptanetriol and meso-1,2,3,4-tetrahydroxybutane. It would also be worth investigating whether the reaction can occur with γ-substituted ATP analogues such as Ap3A. Ultimately, if it were possible to synthesize a γ-bromo-ethyl ester form of ATP, which could undergo the phosphotransfer catalysed by E264K LysU, one might synthesis phosphatidylcholine directly from glycerol.
Materials and methods
All chemicals were obtained from Sigma Aldrich (St. Louis, MO, USA) unless otherwise stated. Tris/HCl buffer consists of 50 mm Tris/HCl, pH 8.0. Buffer A consists of 50 mm potassium phosphate pH 7.8, 2 mm EDTA, 2 mm β-mercaptoethanol, 0.6 mm benzamidine and an EDTA-free protease inhibitor cocktail tablet (Roche, Basal, Switzerland) added just before use. Buffer B is as buffer A with additional 1 m NaCl. Triethylammonium hydrogencarbonate buffer was made by bubbling CO2 (from dry ice) through 2 m triethylamine (aq.) at 4 °C until the solution was monophasic and the pH dropped to < 9. The triethylammoninium hydrogen carbonate was then sealed and stored at 4 °C. Mass spectra were recorded on an ESI-MS Esquire 3000 (Bruker, Billerica, MA, USA) set to negative ionization, the solvent was 1 : 1 acetonitrile/water with 0.1% acetic acid. 1H- and 13C-NMR spectra were recorded on an AMX-500 instrument (Bruker) at ambient temperature in D2O. Integrals are not reported due to excessive peak overlap. LysU concentrations refer to the dimer (unless otherwise stated) and were determined from A280 measurement using an extinction coefficient of 30 580 m−1·cm−1 .
LysU mutagenesis and purification
Site-directed mutagenesis was performed by PCR using QuikChange kits (Stratagene, La Jolla, CA, USA). The plasmid pADH2 was used as a template  and mutations were verified by sequencing. E. coli lysU deletion strain lysU2-17A was used for expression and protein purifications were performed, as described previously . In brief, cell extracts were obtained by sonication in buffer A and clarified by centrifugation. After treatment with DNAase/MgCl2/streptomycin, the protein lysate was then precipitated by fractionation with ammonium sulphate (20–60% saturation), redissolved and purified by size-exclusion chromatography (FPLC system and Sephacryl S300 column; GE Healthcare BioSciences, Piscataway, NJ, USA) followed by anion-exchange chromatography (HiTrap Q-Sepharose column; GE Healthcare), eluting with a gradient of buffer B. Finally, the Superdex 200 column was used for further purifying the LysU protein before concentrating to > 80 μm with a stirred cell. Purified proteins were estimated > 95% pure by SDS/PAGE (Coomassie Brilliant Blue stained) and either used directly or flash frozen in liquid nitrogen for storage at −20 °C.
RP-HPLCMS/MS shotgun analysis
The identities of each purified protein were verified using 1D LC MS/MS methods on an LTQ VELOS mass spectrometer (Thermo, San Jose, CA, USA). Appropriate volumes of protein sample were digested with trypsin at 37 °C overnight. The tryptic protein mixtures were separated with C18 RP column (0.15 × 150 mm; Column Technology Inc., Fremont, CA, USA). Mobile phases A (0.1% formic acid in water) and B (0.1% formic acid in acetonitrile) were selected and the mass spectrometer was set to positive-ion mode, with one full MS scan followed by three MS/MS scans on the most intense ions using dynamic exclusion setting. The acquired MS/MS spectra were automatically searched against an E. coli protein database using the sequest program (Thermo Fisher Scientific Inc., Waltham, MA, USA).
Fluorescence spectra were measured on a RF5301-PC spectrofluorophotometer (Shimadzu, Kyoto, Japan) using a quartz cuvette (1-cm path length) with excitation wavelength set to 295 nm and emission observed between 310 and 350 nm. LysU samples were incubated at 30 °C for 20 min in Tris/HCl buffer containing different concentrations of guanidinium chloride (GuHCl) as indicated. Post incubation, tryptophanic fluorescence intensity at 340 nm was measured as a function of GuHCl concentration in order to derive unfolding transition curves for wild-type and mutant LysUs.
CD studies were performed on a J-715 spectropolarimeter (Jasco, Tokyo, Japan) using a quartz cuvette (0.1 cm path length), 1 nm bandwidth and a response time 4 s for far-UV measurements. For the temperature-induced unfolding studies, aliquots of LysU (10 μm) in 50 mm Tris/Cl buffer, pH 8.0 were monitored during a temperature ramp from 20 to 80 °C with a scan rate of 60 °C·h−1. Far-UV CD spectra were recorded (190–250 nm) and ΔΔA222 plotted as a function of temperature to derive unfolding transition curves for wild-type and mutant LysU.
LysU activities monitored by HPLC
Ion-exchange HPLC studies were carried out on an Agilent 1100 system (Santa Clara, CA, USA) monitoring at 254 and 280 nm and equipped with a Tricorn column (0.5 × 10 cm; GE Heathcare, Pittsburgh, PA, USA) packed with 2 mL SOURCE 15Q anion-exchange medium (GE Healthcare). The column was loaded with Tris/HCl buffer and eluted with a 2 mL·min−1 gradient of 1 m NaCl .
In the case of Ap4A synthesis, LysU WT or mutants were dialysed thoroughly into Tris/HCl before use. Standard reaction mixtures (0.3 or 1 mL) were prepared with 10 μm LysU, 10 mm MgCl2, 2 mm l-lysine, 5 mm ATP and 6 units (7 μg) inorganic pyrophosphatase (E. coli) in Tris/HCl buffer. Each reaction mixture was treated with 160 μm ZnCl2 added slowly with vortex mixing, followed by incubation at 37 °C over 3 h. Aliquots (2 μL) were extracted and analysed using an HPLC program at 0, 5, 10, 15, 30, 60, 120 and 180 min. Relative concentrations of AMP, ADP, ATP, Ap4A and Ap3A were determined analytically (from eluted peak area) as a function of time, to determine catalytic efficacy [36, 37]. Components of the mixture were identified by retention time compared with standards.
The ATP to ADP turnover (glycerol kinase-like) studies were carried out analagously, using a reaction mixture of 10 μm LysU, 10 mm MgCl2, 20 mm glycerol and 5 mm ATP in Tris/HCl buffer. Subsequent larger scale preparations (7 and 60 mL mixtures) were also checked for completion by HPLC, using larger aliquots (50 μL). Control studies (without glycerol) indicated minimal formation of ADP and no glycerol phosphate.
Isolation of glycerol-3-phosphate
Larger scale reaction mixtures were frozen at −20 °C to denature LysU, thawed and loaded onto a 35 mL SOURCE 15Q ion-exchange column mounted on an FPLC system (GE Heathcare). The column was washed (5 ×) with deionized water and eluted at 5–7 mL·min−1 with 0–10% triethylammoninium hydrogen carbonate. Because glycerol-3-phosphate does not show any significant UV absorbance, the location of the relevant fraction was surmised to be before any nucleotide elution. These fractions were lyophilized and checked by ESI-MS. Appropriate fractions were combined and re-lyophilized to yield glycerol-3-phosphate as a white amorphous powder (0.140 g for 2 × 7 mL mixture, 0.515 g for 60 mL mixture); m/z 171.0 [M–H]−, 342.7 [M2–H]− (expected 171.1 and 343.1). A small portion (40 mg) was dissolved in a methanol (5 mL) and lyophilized overnight. This was redissolved in deionized water and loaded on to a prepared Dowex 50WX2 ion-exchange column. The Dowex resin (H+ form, 13.6 g) was hydrated in deionized water for 15 min, before being washed and suspended in 1 m NaOH (aq. 200 mL) for 1 h before being washed with water again. This resin was then introduced to a sintered glass dropping column (1 cm diameter), packed to a height of 10 cm and washed with water (50 mL) before use. After loading, the column was eluted with 20 mL of water. The eluent containing the resulting sodium salt was collected and lyophilized before NMR analysis: δH (500 MHz, D2O) 3.55 (dd, 2J 11.8 Hz, 3J 5.2 Hz, C-1), 3.63 (dd, 2J 11.8 Hz, 3J 5.2 Hz, C-1), 3,77 (m, C-2 and C-3); δC (126 MHz, D2O) 62.3 (s, C-1), 64.9 (d, 2JPC 4.2 Hz, C-3), 71.3 (d, 3JPC 6.9 Hz, C-2). These values compare well to the literature [38, 39].
Isothermal titration calorimetry
A VP- isothermal titration calorimeter (GE Healthcare Biosciences, Pittsburgh, PA, USA) was used in all binding studies to observe the heat energy gained or lost during binding events. Data was fitted using origin 5.0 (OriginLabs, Northampton, MA, USA). The reaction cell was maintained at 20 °C during all titrations and contained aliquots of LysU (20 μm, monomer) suspended in Tris/HCl buffer containing 10 mm MgCl2, and 1 mm l-lysine. A stock solution of appropriate nucleotide (250 μm) was prepared for injection in the same buffer. Titrations consisted of 58 injections (each 5 μL, 10 s), with a 4 min interval between injections. Titration results were corrected for heats of dilution and the LysU monomer concentration was used throughout to facilitate the determination of stoichiometry (N) of binding.
NB and ST wish to thank the Royal Thai Government for PhD studentships, while XC wishes to thank the China Scholarship Council for funding. This work was also supported by Mitsubishi Chemical Corporation, IC-Vec and ImuThes Ltd.