The duality of LysU, a catalyst for both Ap4A and Ap3A formation


A. D. Miller, Imperial College Genetic Therapies Centre, Department of Chemistry, Flowers Building, Armstrong Road, Imperial College London, London, SW7 2AZ, UK
Fax: +44 20 75945803
Tel: +44 20 75945773


Heat shock inducible lysyl-tRNA synthetase of Escherichia coli (LysU) is known to be a highly efficient diadenosine 5′,5′′′-P1,P4-tetraphosphate (Ap4A) synthase. However, we use an ion-exchange HPLC technique to demonstrate that active LysU mixtures actually have a dual catalytic activity, initially producing Ap4A from ATP, before converting that tetraphosphate to a triphosphate. LysU appears to be an effective diadenosine 5′,5′′′-P1,P3-triphosphate (Ap3A) synthase. Mechanistic investigations reveal that Ap3A formation requires: (a) that the second step of Ap4A formation is slightly reversible, thereby leading to a modest reappearance of adenylate intermediate; and (b) that phosphate is present to trap the intermediate (either as inorganic phosphate, as added ADP, or as ADP generated in situ from inorganic phosphate). Ap3A forms readily from Ap4A in the presence of such phosphate-based adenylate traps (via a ‘reverse-trap’ mechanism). LysU is also clearly demonstrated to exist in a phosphorylated state that is more physically robust as a catalyst of Ap4A formation than the nonphosphorylated state. However, phosphorylated LysU shows only marginally improved catalytic efficiency. We note that Ap3A effects have barely been studied in prokaryotic organisms. By contrast, there is a body of literature that describes Ap3A and Ap4A having substantially different functions in eukaryotic cells. Our data suggest that Ap3A and Ap4A biosynthesis could be linked together through a single prokaryotic dual ‘synthase’ enzyme. Therefore, in our view there is a need for new research into the effects and impact of Ap3A alone and the intracellular [Ap3A]/[Ap4A] ratio on prokaryotic organisms.


aminoacyl-tRNA synthetase


2-amino-6-chloropurine ribonucleoside


2-amino-6-mercapto-7-methylpurine ribonucleoside


adenosine 5′-(α,β-methylene) diphosphate


adenosine 5′-(β,γ-methylene) triphosphate


diadenosine 5′,5′′′-P1,P3-triphosphate


diadenosine 5′,5′′′-P1,P4-tetraphosphate


methylene-substituted Ap4A analogue





Aminoacyl-tRNA synthetases (aaRSs) are a heterogeneous family of around 20 distinct enzymes involved in maintaining the fidelity of protein synthesis through the specific esterification of an amino acid to the 2′- or 3′-hydroxyl group of the 3′-terminal adenosine of the cognate tRNA(s) during the translation process [1]. Most prokaryotic aaRSs are usually coded for by single, unique genes (unlike eukaryotes), but Escherichia coli lysyl-tRNA synthetase (LysRS) is an exception in that it exists as two distinct synthetase isoforms, LysS and LysU. These two isoforms share a high degree of sequence identity (88%) but appear to have evolved for different purposes. LysS is constitutively expressed under normal growth conditions and is responsible for the normal tRNA charging activity, whereas LysU is the product of a normally silent gene, but can be induced to a high-level expression under selected physiological conditions, including heat shock, oxidative stress, and anaerobiosis [2,3]. LysU is a highly efficient diadenosine 5′,5′′′-P1,P4-tetraphosphate (Ap4A) synthase, and the production of Ap4A under conditions of cellular stress appears to be a primary function [4–7].

The synthesis of Ap4A catalysed by LysU occurs in two steps (Scheme 1). The first (step 1) involves the formation of a lysyl-adenylate intermediate in which the amino acid is activated through combination with the α-phosphate of the first nucleotide substrate ATP, in a process involving the simultaneous displacement of pyrophosphate. In step 2, the γ-phosphate of a second nucleotide substrate ATP combines with enzyme-bound lysyl-adenylate, thereby generating Ap4A and liberating free l-lysine [8]. Step 1 is highly specific and conservative; ATP can only be replaced by deoxy-ATP nucleotide substrates. Step 2 is much more catholic, and the second nucleotide substrate ATP can be replaced by a variety of diphosphate, triphosphate or tetraphosphate nucleotide substrates. This flexibility in step 2 means that LysU can be an efficient platform catalyst for the synthesis of a wide variety of natural and artificial polyphosphates [9–12].

Figure Scheme 1..

 LysU-catalysed Ap4A and Ap3A synthesis. Both synthetic mechanisms are catalysed in the presence of Mg2+, Zn2+ and inorganic pyrophosphatase. The pathways share a common step 1 in the formation of a lysyl-adenylate intermediate (a) from ATP (b) and l-lysine with release of pyrophosphate (cleaved to phosphate by inorganic pyrophosphatase). Steps 2 and 3 are the combination of this intermediate and either ATP to form Ap4A (c), or ADP (if present) to form Ap3A (d), with the concurrent release of l-lysine. Once the extraneous ATP (b) has been exhausted, partial reversal of step 2 results in the cleavage of Ap4A (c) back to lysyl-adenylate (a) and ATP (which in turn forms further intermediate). In this situation, step 4 allows the produced lysyl-adenylate (a) to slowly react with inorganic phosphate to give ADP (e), which then allows the formation of Ap3A (d). The overall reaction is thus: 2ATP→Ap4A + 2Pi→Ap3A + 3Pi.

Most previous studies of LysU catalysis have focused on Ap4A synthase activity, but there have also been several reports concerning the unexpected coisolation of diadenosine 5′,5′′′-P1,P3-triphosphate (Ap3A) from LysU catalysis mixtures [8,9,13,14]. Since ADP is not a first nucleotide substrate for LysU (that is, LysU cannot catalyse the formation of lysyl-adenylate directly from ADP and l-lysine), this Ap3A has been assumed to be a minor side product emanating from the combination of the lysyl-adenylate intermediate with residual ADP present in ATP. Therefore, we decided to determine whether Ap3A was indeed a minor side product of LysU catalysis or in fact a genuine second main product alongside Ap4A. At the same time, we intended to investigate the behaviour of LysU with respect to phosphorylation, looking for potential linkages between LysU-mediated catalysis and phosphorylation state. The results of our investigations are reported here.

Results and Discussion

Ap4A/Ap3A synthase duality

Typically, when a standard LysU catalysis mixture is prepared (comprising LysU, excess l-lysine and inorganic pyrophosphatase in a buffer containing Mg2+ and Zn2+), added ATP (> 99%, purified by ion exchange chromatography) will be converted into Ap4A after 20–30 min at 37 °C. Thereafter, significant quantities of Ap3A will be formed at the expense of Ap4A, such that triphosphate may easily become the major product after 1 h (depending upon the concentrations of LysU and substrate involved). This phenomenon was clearly observed using an assay based on a SOURCE 15Q ion exchange column attached to an HPLC system set up to take repeated aliquots from enzyme incubation mixtures every 9 min, thereby allowing for the quantitative measurement of individual nucleotide substrate and diadenosine polyphosphate concentrations as a function of time. The results obtained from the incubation of ATP (5 mm) with LysU in a typical catalysis mixture are shown in Fig. 1. From multiple data sets of this kind, we observed that the fast initial Ap4A synthesis (170 ± 40 min−1) converted the majority (perhaps all) of the available ATP to tetraphosphate. This was followed by a slower process of Ap4A-to-Ap3A conversion (12 ± 3 min−1) with concurrent rise of ADP levels to about 0.1 mm (initially negligible) over the next 1.5 h. This dual catalytic phenomenon runs counter to the general expectation of LysU as a primary Ap4A synthase and deserves an explanation. In the absence of initially available ADP, it is apparent that ADP is somehow generated in situ and then acts as an alternative second nucleotide substrate in step 2, combining with lysyl-adenylate to form Ap3A (Scheme 1, step 3). However, what is the origin of this in situ ADP?

Figure 1.

 (A) Ion exchange HPLC measurement of mononucleotide and dinucleotide levels observed in a LysU catalysis mixture containing 5 mm ATP over 2 h. Traces show A260 during each separation. (B) Integration and normalization of the traces to give relative concentrations for each component. Compounds are identified as: bsl00084, ATP; ○, Ap4A; •, ADP; bsl00072, Ap3A; □, AMP. Errors are ± 0.05 mm.

Initially, we considered two possibilities: option 1, that LysU may act as an ATP-to-ADP hydrolase; or option 2 that LysU may be a symmetrical Ap4A hydrolase. The rapidity of ATP consumption in the first stage of polyphosphate formation with no apparent formation of ADP appeared to rule out option 1. In fact, significant ADP was not seen in reaction mixtures until > 30 min from the start, by which time no ATP remained (Fig. 1). Instead, the kinetics of ADP appearance and Ap4A disappearance suggested that Ap4A may be the source of ADP according to option 2. Consistent with option 2, when a mixture of ATP and adenosine 5′-(β,γ-methylene) triphosphate (AMPPCP) 1 : 1 (m/m) is combined with a LysU catalysis mixture, AppCH2ppA (a methylene-substituted Ap4A analogue) forms without apparent formation of Ap4A, and subsequently Ap3A is not isolated even after 3 h at 37 °C (beyond which time, LysU begins to denature) [10]. Even a single bisphosphonate methylene linkage is known to confer resistance to hydrolysis [15,16], and AMPPCP is unable to function as a first nucleotide substrate for LysU (Scheme 1, step 1). It is, however, an apparently more potent second nucleotide substrate than ATP, able to trap the lysyl-adenylate intermediate in a nonreversible equivalent to step 2 (Scheme 1), and leading to the exclusive formation of AppCH2ppA [9]. The subsequent failure to generate Ap3A could then be accounted for by the inability of AppCH2ppA to undergo symmetrical hydrolysis to form ADP, also owing to the presence of the bisphosphonate methylene linkage.

In order to obtain further evidence for a putative ‘symmetrical Ap4A hydrolase’ activity, Ap4A was incubated with LysU in a standard catalysis mixture (comprising Tris/HCl buffer) minus inorganic phosphate or any other nucleotide. Against expectations, no ADP was generated over 3 h (Fig. 2A). However, when a mixture of ADP and Ap4A 1 : 1 (m/m) was incubated with LysU in an identical catalysis mixture (minus inorganic phosphate or any other nucleotide), then Ap3A was generated rapidly with an estimated turnover number for the conversion of Ap4A to Ap3A of 90 ± 5 min−1 (10-fold higher than observed previously) (Fig. 2B). The turnover rate was observed to increase further as the ADP-to-Ap4A ratio was increased. This latter result is consistent with a partial reversibility of Ap4A formation (Scheme 1, step 2), giving rise to an adenylate intermediate in situ that can then be trapped by ADP to give Ap3A at the expense of Ap4A (Scheme 1, step 3). The absence of any observed accumulation of ATP formed from reverse step 2 suggests that this is rapidly recombined with lysine to form an adenylate intermediate (Scheme 1, step 1) that would then be rapidly trapped by ADP once again to give Ap3A. Evidence in support of this reverse-trap process was obtained by substituting for AMPPCP or adenosine 5′-(α,β-methylene) diphosphate (AMPCP) for ADP. Trap products AppCH2ppA and ApCH2ppA, respectively, were generated exclusively, completely consistent with our proposed reverse-trap mechanism. Hence, faced with such evidence that LysU is clearly not an Ap4A hydrolase (Fig. 2A), we needed to come up with an alternative explanation for the source of ADP generated in situ.

Figure 2.

 Evidence for LysU catalytic duality. (A) A LysU catalysis mixture containing 5 mm Ap4A shows no significant conversion to Ap3A nor hydrolysis to ADP over 2 h. A constant AMP background is seen due to trace contaminants in the Ap4A stock, but no ADP or Ap3A is produced. (B) A catalysis mixture containing 5 mm Ap4A and 5 mm ADP shows rapid loss of ADP and Ap4A (fitting exponential decay curves), and concurrent synthesis of Ap3A. The turnover of ADP is about twice that of Ap4A, which matches the Ap4A + 2.ADP→2.Ap3A + 2.Pi stoichiometry. (C) A LysU catalysis mixture containing 5 mm Ap4A and made up in 50 mm potassium phosphate buffer, pH 7.8. Apparent phosphate attack on lysyl-adenylate results in greatly increased turnover of Ap4A to ADP and Ap3A. Longer incubation times (> 2 h) show that the conversion of Ap4A to Ap3A continues, while the concentration of ADP stabilizes at approximately 1 mm. (D) A LysU catalysis mixture containing 5 mm ATP and made up in 50 mm potassium phosphate buffer. The presence of inorganic phosphate disrupts the formation of Ap4A, and the major product under these conditions is ADP. Compounds are identified as:bsl00084, ATP; ○, Ap4A; •, ADP; bsl00072, Ap3A; □, AMP. Errors are ± 0.05 mm.

Returning to the original Ap4A synthesis scenario (Fig. 1), we came to the realization that there could be an alternative option (option 3). During the initial synthesis of Ap4A, the main byproduct of step 1 is inorganic phosphate formed through cleavage of pyrophosphate by the action of an inorganic pyrophosphatase enzyme (contained in the LysU catalysis mixture). Therefore, we considered that this inorganic phosphate may also be capable of trapping residual adenylate intermediate arising from the partial reversibility of Ap4A formation (Scheme 1, step 4), thereby generating ADP in situ. An equivalent process to this (albeit much slower) has been remarked upon in studies on E. coli glycyl-tRNA synthetase [17]. Thereafter, ADP thus formed in situ would be in a position either to trap further adenylate deriving from the partial reversibility of Ap4A formation or to combine with adenylate generated in the normal way from ATP, thereby forming Ap3A in either case. All the proposed mechanistic steps are summarized in Scheme 1. Given this, when ATP is incubated in a standard LysU catalysis mixture (Fig. 1), then the observed variations in concentrations of mononucleotide and dinucleotide species with time can be accounted for in the following way. Early ATP consumption and Ap4A formation are rapid, consistent with an initial process in which step 1 is committed and step 2 is kinetically very favourable. Thereafter, the partial reversibility of step 2 allows the reverse-trap mechanism to give ADP (from inorganic phosphate) and then Ap3A (from ADP; Scheme 1). The appearance of the lysyl-adenylate intermediate would appear to be rate-limiting (at least in part) for the formation of both ADP and Ap3A, although the former clearly must convert readily into the latter (Scheme 1, step 3) at the expense of Ap4A, keeping the overall solution concentration of ADP at a minimum while the concentration of end-product Ap3A begins to rise steadily.

Evidence in support of our proposed mechanism and the adenylate-trapping function of inorganic phosphate was obtained in the following way. Ap4A was incubated with LysU in a catalysis mixture containing 50 mm potassium phosphate but minus any nucleotides. In this instance, > 20% of Ap4A was observed to convert to Ap3A over 2 h. This contrasts with the previous situation where Ap4A alone incubated in a LysU catalysis mixture (Tris/HCl buffer, no inorganic phosphate) failed to convert into Ap3A over a 3-h period (compare Fig. 2C with Fig. 2A). Furthermore, when ATP alone was incubated with LysU in a catalysis mixture comprising 50 mm phosphate buffer, then Ap3A/Ap4A synthase activities as a whole were adversely affected, with fully 30% of the initial ATP being converted to ADP via attack of inorganic phosphate on the adenylate intermediate (compare Fig. 2D with Fig. 1B). Curiously, unlike ADP and inorganic phosphate, AMP did not appear to be able to trap adenylate. The reasons for this are unclear but are likely to be due to LysU active site topographies and unfavourable steric contacts [18]. Finally, the summarized mechanistic steps (Scheme 1) can be used to provide an alternative explanation for our original observation that AppCH2ppA is the exclusive product when ATP and AMPPCP 1 : 1 (m/m) are combined with a LysU catalysis mixture. As stated above, ATP is the only possible first nucleotide substrate of LysU, whereas both ATP and AMPPCP could be second nucleotide substrates [9,10]. Given the situation in which the formation of AppCH2ppA is essentially irreversible but the formation of Ap4A is partially reversible (as described above), the final outcome is consistent with the formation of only transient Ap4A, since adenylate intermediate regeneration at its expense would then be trapped by excess remaining AMPPCP. Therefore, our data describing the exclusive formation of AppCH2ppA can be seen as a simple variation of our newly proposed reverse-trap mechanism for the Ap3A synthase activity of LysU.

Phosphorylation state duality

Previous purification protocols for LysS and LysU have described the use of hydroxyapatite (HA) medium columns to subfractionate samples of LysU and LysS into different charged ‘isoforms’[4,5,19]. HA makes this possible, as it is a crystalline form of calcium phosphate that interacts differently with globular proteins depending upon their charge [20]. In our hands, LysU (previously purified by S300 and Q-Sepharose chromatography [14]) could be easily resolved into two main subfractions by elution through an HA column, the first subfraction eluting at 20% and the second at 30% from a linear gradient of potassium phosphate (10–300 mm) (Fig. 3). Both subfractions were confirmed to be greater than 95% pure LysU by SDS/PAGE. However, since HA resolves proteins by charge, we concluded that these two different subfractions should contain LysU isoforms of different overall charge at neutral pH.

Figure 3.

 Hydroxyapatite (HA) chromatography. The HA column was loaded with LysU and then eluted with a stepwise gradient of potassium phosphate buffer, pH 6.5 (10–300 mm); fractions were analysed for absorbance at 280 nm. Illustration of the elution profile monitored by absorbance showing two main subfractions of LysU eluting at 20% and 30% gradient.

Differences in protein phosphorylation state were considered the most likely explanation for the emergence of these two different LysU isoforms. Therefore, separate aliquots of each subfraction were analysed for phosphate by means of the malachite green assay [21] calibrated against a standard curve generated with known concentrations of inorganic phosphate. The results clearly suggest that the 20% subfraction was unphosphorylated whereas the 30% subfraction contained LysU phosphorylated at the level of a single phosphate per monomer (Fig. 4), in line with our explanation. Furthermore, western blot analyses of the two subfractions using a phosphothreonine antibody confirmed that not only was the 30% subfraction phosphorylated, but also was the position of phosphorylation on the hydroxyl group of a threonine residue (Fig. 4). LysU has a number of surface-accessible threonine residues according to the X-ray crystal structure [2]. Therefore, enzyme digestion and fragment MS is likely to be the most effective way to determine precisely which threonine residue is involved.

Figure 4.

 LysU phosphorylation. (A) SDS/PAGE of 20% and 30% subfractions (see Fig. 3), both showing clean bands matching the expected mass for LysU monomer. (B) Calibration of Malachite green assay. (C) Anti-phosphothreonine western blot of 20% and 30% subfractions, with only the 30% subfraction showing positive results.

Phosphorylated forms of aminoacyl-tRNA synthetase enzymes are known. For example, phosphorylated forms of eukaryotic lysyl-tRNA from rat liver have been characterized [22,23], as have phosphorylated forms of other aminoacyl-tRNA synthases (multiply phosphorylated on the serine amino acid residue) from rabbit reticulocytes [24]. The role of these phosphorylation events has not been explored to any great extent, although nonphosphorylated yeast LysRS is known to be more specific for lysine than its threonine-phosphorylated equivalent [22]. Furthermore, with regards to Ap4A synthesis, phosphorylation of rabbit reticulocyte SerRS and ThrRS has been shown to enhance the catalytic rates of Ap4A synthesis rates by two-fold and six-fold, respectively [25]. Consequently, we elected to characterize the 20% and 30% subfractions of LysU in order to determine if there were any major differences. In particular, in view of the dual catalytic behaviour of LysU described here, we were curious to characterize the effects or otherwise of phosphorylation upon the catalytic rates for Ap4A and Ap3A synthesis. A variety of catalysis characterization techniques were employed to identify differences. A radioactive assay using [14C]ATP was used to accurately measure total polyphosphate synthesis over a fixed period, allowing for the calculation of average turnover under kcat conditions (moles Ap3/4A per min per mole LysU at 37 °C). Typical results were 134 ± 10 min−1 (20% subfraction) and 150 ± 12 min−1 (30% subfraction), suggesting little difference between phosphorylated LysU and the nonphosphorylated isoform within experimental error. On the other hand, phosphorylated LysU was able to retain activity (70 ± 5% activity after 7 days) for significantly longer after storage at 4 °C than unphosphorylated LysU (30 ± 5% activity after 7 days), suggesting that phosphorylated LysU is significantly more stable than nonphosphorylated LysU.

A 1H-NMR spectroscopy assay was then used to characterize the conversion of ATP to Ap4A by the 20% and 30% subfractions of LysU, respectively. For each individual NMR assay, conversion of ATP into Ap4A can be observed with time in an NMR tube by monitoring the disappearance as a function of time of the H2 and H8 proton signals of adenine (of ATP) matched by the appearance of the H2 and H8 proton signals of adenine (of Ap4A). Ap4A adenine signals are shifted slightly upfield due to the π-stacking of adenine rings in Ap4A (not observed with ATP) [26]. Assuming steady-state conditions, initial rates of catalysis of Ap4A formation may be calculated from H2 and H8 proton peak integrations, and used to determine the main kinetic steady-state parameters for ATP and l-lysine [14,27]. The effects of [ATP] and [l-lysine] on Ap4A formation rates were determined by taking one substrate in excess and varying the concentration of the other (excess concentrations estimated from previous results as 10 mm ATP and 2 mm lysine), keeping the concentrations of 20% or 30% subfractions of LysU constant (400 nm, dimer concentration) (Table 1). Unsurprisingly the kcat constants of the two LysU isoforms were found to be essentially identical within experimental error, although the specificity constant (kcat/KM) of phosphorylated LysU for l-lysine appears to be 2–3-fold higher than that of nonphosphorylated LysU. The reason appears to be that phosphorylated LysU has both a marginally higher kcat for l-lysine and a lower value of KM, indicative of a situation wherein LysU phosphorylation has enabled l-lysine to be an improved substrate under these given assay conditions.

Table 1.   Kinetic constants derived from 1H-NMR Ap4A synthesis assays of 20% and 30% subfractions.
  KMkcat (s−1)kcat/KM (s−1·mm−1)
LysU 20% subfraction
 For ATP7 ± 4 mm2.7 ± 0.40.4 ± 0.3
 For lysine23 ± 12 µm1.8 ± 0.178 ± 40
LysU 30% subfraction
 For ATP6 ± 3 mm2.0 ± 0.20.3 ± 0.2
 For lysine10 ± 5 µm2.9 ± 0.2290 ± 120

Next, a colorimetric-coupled assay was used to characterize the formation of the lysyl-adenylate intermediate by the 20% and 30% subfractions of LysU, respectively. In each individual assay, the formation of inorganic phosphate (produced by the action of inorganic pyrophosphatase following lysyl-adenylate formation) was coupled with the release of a UV-active dye 2-methyl-6-mercapto-7-methylpurine (MMMP) [28,29] from a chromogenic substrate 2-amino-6-mercapto-7-methylpurine ribonucleoside (AMMPR). The coupling enzyme purine nucleoside phosphorylase catalyzes the combination of inorganic phosphate with AMMPR, leading to MMMP release. This coupled reaction must occur faster than the formation of lysyl-adenylate. Accordingly, only a low fixed LysU concentration (20 nm dimer concentration) was used per assay, together with a correspondingly high excess of ATP or l-lysine and a low incubation temperature of 20 °C (which significantly slows the catalysis of Ap4A synthesis). The concentrations of AMMPR, purine nucleoside phosphorylase and inorganic pyrophosphatase required were determined by experiment, and the assay was calibrated against known concentrations of potassium phosphate. The kinetic parameters determined are shown (Table 2). Since this coupled assay monitors specifically the formation of lysyl-adenylate alone and not Ap4A synthesis, these parameters can be attributed solely to the binding and chemical equilibrium of step 1 (Scheme 1) alone. Once again, values of kcat were found to be essentially identical within experimental error, but the value of the specificity constant (kcat/KM) for ATP of phosphorylated LysU was between one- and two-fold higher than that of nonphosphorylated LysU. The reason for this is that phosphorylated LysU has a lower value of KM for ATP, indicative of a situation wherein phosphorylation of LysU has enabled ATP to be a mildly improved substrate for LysU-mediated catalysis under these assay conditions. Finally, a complete repetition of the SOURCE 15Q ion exchange chromatography assays (as described earlier) suggested that the rates of formation of Ap3A from Ap4A were unaffected by the state of LysU phosphorylation (results not shown).

Table 2.   Kinetic constants derived from lysyl-adenylate, enzyme-coupled assays of 20% and 30% subfractions.
 KMkcat (s−1)kcat/KM
LysU 20% subfraction
 For ATP9.6 ± 3 mm16.0 ± 2.51.7 ± 0.8 s−1·mm−1
 For lysine0.27 ± 0.14 µm8.6 ± 1.132 ± 20 s−1·µm−1
LysU 30% subfraction
 For ATP4.0 ± 1.2 mm12.2 ± 1.53.0 ± 1.3 s−1·mm−1
 For lysine0.34 ± 0.11 µm10.1 ± 0.730 ± 11 s−1·µm−1

Roles of dualities

We have demonstrated that dimeric LysU has dual Ap4A and consecutive Ap3A synthase activities, and that it exists in either a phosphorylated or nonphosphorylated state. The phosphorylated LysU appears to be a more robust enzyme than nonphosphorylated LysU, although in catalytic terms they do not appear to have substantially different activities. This suggests that modifying LysU-catalysed Ap4A or Ap3A formation is probably not the main function of phosphorylation. Nevertheless, in our view, the Ap4A/Ap3A synthase and phosphorylation state dualities of LysU do appear to be linked, albeit not strongly. The Ap3A synthase activities of LysU appear to be the product of a number of possible contributory mechanisms. Both ADP and inorganic phosphate are present in cells, so Ap3A could be formed directly from ATP and ADP (where ADP replaces ATP as the second nucleotide substrate in Scheme 1, step 2), or from Ap4A by the illustrated reverse-trap mechanism (Scheme 1, reverse step 2, then steps 3 and 4). However, there is insufficient information in the current literature to make sense of this catalytic behaviour. In eukaryotic cells, there seems to be a general tendency for Ap4A and Ap3A to have antagonistic effects in vivo. In particular, the relative concentrations of these polyphosphates appear to be indicative of cellular status, with human cultured cells showing high Ap4A/Ap3A ratios when undergoing induced apoptosis and the reverse during differentiation [30,31]. Ap4A 10 µm was also shown to be sufficient to trigger apoptosis in a variety of human and mouse cell lines, a concentration not significantly higher than that seen in human adrenal vein blood serum [32].

Similarly, Ap3A has been reported to act as a coinducer of cell differentiation when combined with protein kinase C activators. This antagonistic behaviour has also been seen in a number of other studies. For example, submicromolar concentrations of Ap3A have been reported to induce platelet aggregation, in contrast to Ap4A which causes disaggregation [33], and these compounds have been shown to have opposing effects on rabbit interocular pressure as well [34]. Regrettably, there has been no evidence produced to date to suggest that Ap4A and Ap3A should have antagonistic effects in vivo in prokaryotes as well as in eukaryotes. Therefore, although Ap4A may appear to act as an immediate modulator of stress responses in prokaryotes [30,35,36], analogous links between Ap3A and longer-term prokaryotic stress accommodation, or high concentrations of Ap4A and failure of the stress response, must remains hypothetical. Our data presented here suggest that Ap3A and Ap4A biosynthesis are linked together through a single prokaryotic dual ‘synthase’ enzyme. Therefore, in our view, there is now an urgent need for new research into the effects and impact of Ap3A alone and the intracellular [Ap3A]/[Ap4A] ratio on prokaryotes in order to make sense of the dual enzymology of LysU in a fuller, more complete biological context.


LysU has dual Ap3A/Ap4A synthase activities and dual phosphorylation state behaviour. These dualities are linked, but only weakly. There is insufficient knowledge about the role of Ap3A in biological prokaryotic systems to understand the implications of these dualities. More research into the function of Ap3A in prokaryotes is necessary.

Experimental procedures


LysU enzyme was overexpressed and purified according to previously published protocols [18]. All LysU concentrations are of the dimer. LysU concentration was determined from A280 measurement using an extinction coefficient of 30 580 m−1·cm−1[37]. All compounds used were obtained from Sigma Aldrich (St Louis, MO) unless otherwise stated.

Ion exchange HPLC assays

A custom 2 mL SOURCE 15Q packed ion exchange column (Amersham Biosciences HR 5/10; Piscataway, NJ) was attached to an Agilent 1100 series HPLC column (Agilent, Palo Alto, CA), equipped with an autosampler, a thermostatted sample chamber (37 °C) and a UV absorbance detector set at 260 nm. The column was loaded (injection aliquots 10 µL) in 5 mm Tris/HCl buffer, pH 8.0, and eluted with a gradient of NaCl (0–45%) over 5 min. The standard LysU catalysis mixture (500 µL) comprised 50 mm Tris/HCl, pH 7.8, 10 mm MgCl2, 160 µm ZnCl2, 5 mm nucleotides, 2 mm l-lysine, 5 units of inorganic pyrophosphatase (Roche, Laval, Canada) and 1 µm LysU. The contents of fractions postelution were identified by their retention time compared with standards (AMP 3.85 min, AMPPCP 4.39 min, ADP 4.43 min, ATP 4.72 min, Ap4A 5.13 min). Mononucleotide and dinucleotide concentrations were calculated from peak area, assuming 260 nm extinction coefficients to be roughly equivalent for AMP through ATP and double for Ap3A and Ap4A. Ap3A was synthesized by incubating 5 mm ATP in a standard LysU mixture overnight, and then extracted from the mixture using a SOURCE 15Q ion exchange column (50 mL) (Amersham Biosciences), loaded in water and eluted with a gradient of 2 m triethylammonium hydrogencarbonate buffer (0–70%) [17]. Appropriate fractions were combined and freeze dried. Contents were characterized by ESI-MS (Bruker Esquire 3000, negative ionization; Billerica, MA), giving a PMI of 754.9 m/z, and ion exchange HPLC, giving a single peak at 4.71 min retention time.

HA column

An HA biogel column (3 × 20 cm) was mounted on an FPLC (Amersham Biosciences, Piscataway, NJ) system and equilibrated with 10 mm potassium phosphate, pH 6.5, containing 2 mmβ-mercaptoethanol, at a flow rate of 1.5 mL·min−1. LysU (3 mg) was dialysed to phosphate buffer and loaded onto the column. The column was eluted at 1.5 mL·min−1 with a gradient (10–300 mm) of potassium dihydrogen phosphate, holding at 10%, 20% and 30% to collect fractions (detected by A280 absorbance). The first 10% peak was discarded and the remainder were concentrated separately using a stirred cell with a 30 kDa molecular weight cutoff filter, before being stored at 4 °C in 20% glycerol.

Malachite green assay

The reagent was made up from one part 0.2% Malachite green in 1 m HCl to five parts 1 m HCl, and two parts 10% ammonium molybdate in 3 m HCl. The reagent was centrifuged to clear precipitate, and an aliquot (100 µL) was added to an aliquot of LysU (300 µL in water). A810 was measured after 5 min. Equal amounts of protein solution and 10% Mg(NO3)2 in 95% ethanol were combined and then vigorously dehydrated in a glass flask over a strong flame. The ‘ash’ was redissolved in 0.5 m HCl prior to assay [38].

Anti-threonine western blot

SDS/PAGE gels of the two LysU fractions (20% and 30%) were electroblotted onto nitrocellulose membranes at 25 V, 0.2 mA for 2 h [15], in 20 mm Tris/HCl, pH 8.0, containing 152 mm glycine and 20% methanol. Nonspecific binding was blocked with 10% free BSA in TBS (20 mm Tris/HCl, 137 mm NaCl), pH 8.2, 0.05% Tween-20 for 1 h at room temperature. Primary antibodies (mouse monoclonal IgG2b anti-phosphothreonine, Sigma-Aldrich) were added at 1 : 1000 dilution, and the whole mixture was incubated for 12 h at 4 °C with gentle rocking. The membrane was then washed with TBS (20 mL), pH 8.2, Tween-20 0.05%; 5 × 10 min washes were performed at room temperature. Thereafter, the secondary antibody (goat anti-mouse IgG conjugated to horse radish peroxidase; Santa Cruz Biotechnology, Santa Cruz, CA) was added at 1 : 5000 dilution and the whole mixture was incubated for 1 h. Finally, the membrane was washed again and then incubated with 2 × 5 mL chemiluminescence reagents (Santa Cruz) for 1 min before wrapping in clear film and exposing to photographic film.

Radioactive [14C]ATP turnover assay

Initially, a LysU reaction mixture of 20 mm Hepes, pH 7.8, was prepared, comprising 150 mm KCl, 25 mm ATP, 25 mm MgCl2, 0.5 mm l-lysine, and 150 µm ZnCl2. Reaction buffer (50 µL) was added to an Eppendorf tube (1.5 mL), followed by [14C]ATP (2.5 µL; 0.125 µCi), 0.25 µg of inorganic pyrophosphatase and a concentrated aliquot of LysU (5 µL). This LysU catalysis mixture was then incubated at 37 °C for 40 min and then briefly boiled to denature the enzymes. After this, 10 units of alkaline phosphatase was added to degrade remaining nucleotides and the mixture was incubated at 37 °C for a further 2 h. An aliquot (40 µL) of the mixture was then washed through three-ply DE-81 filters (ion exchange paper) with 5 × 1 mL of fresh 25 mm ammonium bicarbonate. The filters were then transferred to 5 mL of scintillation fluid [Brady's fluid: 10% (v/v) methanol, 6% (w/v) naphthalene, 2% (v/v) ethylene glycol, 0.4% (w/v) diphenyloxazole in 1,4-dioxane] and shaken vigorously for 20 min. The average LysU turnover was calculated from the scintillation count (minus background) as moles Ap4A per min per mole of LysU at 37 °C.

1H-NMR assays

Individual NMR assays were preformed at 37 °C in a 5 mm NMR tube using a 400 MHz Bruker NMR spectrometer set up with spectra taken every 2 min over a period of 20 min. Each spectrum was a compilation of 64 scans acquired over a period of 1 min. Initially, LysU reaction mixtures of 20 mm Hepes, pH 7.8, 150 mm KCl, 5 mm MgCl2, 150 µm ZnCl2, 1–50 mm ATP, 0.05–5 mm l-lysine, 0.03 mg inorganic pyrophosphatase and 20% D2O were prepared. For each NMR assay, an aliquot (600 µL) of an appropriate LysU reaction mixture was equilibrated at 37 °C in the NMR tube, after which a LysU aliquot (400 nm) was introduced by needle and spectral acquisition was begun. ATP 1H-NMR signals were seen initially at: 8.55 p.p.m. (H2) and 8.29 p.p.m. (H8), and this was followed by the emergence of Ap4A signals with time at: 8.40 p.p.m. (H2) and 8.18 p.p.m. (H8). The H8 signals were seen to be the most reliable indicator of relative concentrations; therefore, the initial rate of conversion of ATP into Ap4A observed in each given NMR assay was determined from H8 signal integrations. Initial rate data were then processed by standard Michaelis–Menten data-fitting to give values of KM and kcat for ATP conversion into Ap4A.

AMMPR-coupled assay

This assay was performed on a thermostatted Ultraspec III (Amersham Biosciences) set to detect A360 at 20 °C from a 0.7 mL quartz cuvette. This assay uses AMMPR, which was prepared in a three-step, one-pot reaction from the reagent 2-amino-6-chloropurine ribonucleoside (ACPR, 1 g). ACPR was dissolved in distilled dimethyl formamide (4 mL) and the solution was then placed in a stirred flask (50 mL) under nitrogen. Methyl iodide (2 mL) was added, and the mixture was stirred for 20 h at room temperature before thiourea (1 g) was added. The mixture was stirred for a further 30 min, after which 2 m ammonia in methanol was added dropwise until the mixture reached neutral pH. Finally, the mixture was poured into stirred acetone (200 mL), causing AMMPR to precipitate out of solution. The pale yellow solid was filtered and dried under nitrogen before storage at − 20 °C. The yield was 64%, and spectral characterization matched the literature [16].

Purine nucleotide phosphorylase enzyme (Sigma-Aldrich) was repurified before use by means of a Mono-Q column (1 × 10 cm) packed in 50 mm Tris/HCl, pH 7.6, and eluted with a gradient of NaCl (0–1.5 m) at a flow rate of 3 mL·min−1. Individual reactions were tested for activity by incubation with 5 mm potassium phosphate and 150 µm AMMPR. Active fractions were pooled, concentrated using 3 kDa cutoff centricon concentrators and dialysed into 20 mm Hepes, pH 7.8, before storage as a precipitate in 3.2 m ammonium sulphate at 4 °C.

Initially for the AMMPR assay, aliquots (500 µL each) of a LysU reaction mixture comprising 20 mm Hepes, pH 7.8, 150 mm KCl, 25 mm MgCl2, 150 µm ZnCl2, 0.25 µg of inorganic pyrophosphatase and 0.03 mg of purine nucleotide phosphatase, were prepared. Each aliquot was then used in two sets of experiments, one set with l-lysine (1–200 µm) and fixed ATP (25 mm), and the other with ATP (0.1–20 mm) and fixed l-lysine (0.5 mm). For each assay, the LysU reaction mixture with added l-lysine and ATP was transferred to the cuvette and equilibrated at 20 °C for 2 min, after which AMMPR (400 µm final concentration) and LysU (20 nm final concentration) were added, the mixture was vigorously agitated, and A360 was followed as a function of time. After background correction, these absorbance data were converted into initial rate data. These initial rate data were then processed by standard Michaelis–Menten data-fitting to give values of KM and kcat for ATP and lysine conversion into lysyl-adenylate.


MW would like to thank IC-Vec, and NB would like to thank the Royal Thai Government for personal support. We would also like to thank IC-Vec and the Mitsubishi Chemical Corporation for their support of the Imperial College Genetic Therapies Centre.