AThTP has recently been discovered in E. coli, where it accumulates as a result of carbon starvation . Here, we describe for the first time the existence, the partial purification and the kinetic properties of an AThTP-synthesizing enzyme. The activity was low in E. coli extracts (supernatant obtained after sonication and centrifugation): 4.1 pmol·min-1·mg-1 of protein under standard incubation conditions. We tried to concentrate the enzyme, by using different precipitation procedures, including ammonium sulfate, acetone, polyethylene glycol and isoelectric point precipitation, but all of them were unsuccessful; each time a 5–10-fold reduction in enzyme activity was observed. We considered that this loss of activity could be due either to inactivation of the enzyme or to the removal of an essential activator. It seems that both factors are important. We found that addition of a boiled bacterial extract to the incubation medium containing salted-out enzyme led to 40–50% increase in the reaction rate (data not shown), suggesting the requirement for an activator, whose chemical nature remains unknown, though it seems to be a low molecular weight compound resistant to heating. Recent studies show an important accumulation of cAMP and phosphoenolpyruvate during carbon starvation , but none had an activating effect on AThTP synthesis (not shown).
Nevertheless, even in the presence of the activator, the specific activity of the enzyme preparation remained low. We faced the same problem testing conventional adsorption chromatography methods and resins – Phenyl-Sepharose, DEAE-Sephacel, Blue-Sepharose, 2',5'ADP-Sepharose and hydroxyapatite. Each time, the activity was lost after chromatography.
It should be noted that this loss of activity was not the consequence of proteolysis or unfavorable buffer composition, as the extract could be kept at +4°C for several days without any significant loss of activity. Moreover, the use of common protective reagents such as glycerol or dithiothreitol did not lead to increased enzyme recovery. It seems that the AThTP-synthesizing enzyme is extremely sensitive to any separation by procedures based on precipitation or adsorption. On the other hand, this enzyme was rather resistant to heating and liquid-liquid chromatography such as gel filtration. Therefore, we set up a procedure for the partial purification in two steps, heat treatment and gel filtration on Sephadex-G-200.
During the gel-filtration step, two peaks of activity were eluted from the column, with molecular masses of 355 and 190 kDa respectively. It should be noted that the respective peak areas depended on the experimental conditions, especially on the time of sample processing and protein concentration, the high molecular mass peak being predominant in most cases. One may suppose that these peaks correspond to aggregated and dissociated forms of the enzyme, based on the ratio of their molecular masses. The effect of ADP concentration (Fig. 5) gave a sigmoid curve with a Hill coefficient around 2, which is in agreement with the assumption that the 355-kDa enzyme complex contains two cooperative subunits.
The AThTP-synthesizing enzyme also appears to undergo time-dependent reorganization. As shown in Fig. 2, it took several hours before the activity reached a steady-state, pointing to slow conformational changes in the molecular structure of the enzyme. Such a behavior is characteristic of a special class of allosteric enzymes capable of slow association-dissociation processes induced by substrate, ligands or protein concentration . It is possible that further dissociation of the 190 kDa species into smaller subunits is responsible for the low stability of the AThTP-synthesizing enzyme during precipitation and adsorption procedures, indicating a weak binding between its subunits.
The partially purified enzyme exhibited a maximal activity at pH 6.5–7.0 (Fig. 3), the enzyme being active in a rather broad range of pH. The enzyme is sensitive to the nature of the buffer as well as the buffer concentration. At pH 6.5, for example, the activity in 50 mM maleate was 1.2 times higher than in 50 mM Bis-Tris-propane, and the latter, in turn, was 1.5 times higher than in the 100 mM Bis-Tris-propane.
The enzyme showed hyperbolic saturation with respect to ThDP concentration at a fixed ADP concentration, with an apparent Km of approximately 5 mM (Fig. 4). Though this is a very high value compared to the free intracellular ThDP concentration in bacteria , it has a physiological meaning. Indeed, under such conditions ([S] Km) the reaction follows first order kinetics with respect to ThDP: the more ThDP is available, the more AThTP can be synthesized. This could explain why in intact bacteria, AThTP accumulates during carbon starvation . In this situation, where catabolic processes become prevalent, cellular proteins are degraded, probably leading to the dissociation of enzyme-bound ThDP and a substantial increase in the cytosolic concentration of free ThDP.
On the other hand, a sigmoid saturation curve, a typical feature of allosteric enzymes, was obtained for ADP at a fixed ThDP concentration (Fig. 5). This might indicate a regulatory mechanism. However, ADP concentration in bacterial cells is around 1 mM  indicating that the enzyme is saturated under normal physiological conditions.
The enzyme has an absolute dependence on divalent metal ions such as Mg2+or Mn2+. A sigmoid saturation curve was observed when the rate of AThTP synthesis was plotted against the total Mg2+ concentration (Fig. 6). As Mg2+ is known to form complexes with polyphosphates, the concentration of free Mg2+ is not equal to its total concentration and this could explain the reason for the sigmoid behavior. Indeed, if we replace total Mg2+ by free Mg2+ concentration estimated from the dissociation constant of 457 μM  for the Mg- ADP- complex, the shape of the plot becomes less sigmoid (Fig. 6). In addition, complexes with ThDP (Kd = 420 μM ) and buffer ions (maleate) are also generated, leading to a further decrease in free Mg2+ content, especially in the range of its low concentrations. It is thus likely that there are no cooperative effects of Mg2+ ions.
This enzyme is highly specific for ThDP among thiamine phosphates, but it is able to use both ATP and ADP as the second substrate. As AThTP is synthesized only under conditions of carbon starvation , i.e. when ATP content is low, ADP is probably the physiologically relevant substrate. The replacement of ADP (ATP) by GDP, UDP or CDP gave no product formation. It could be argued that, as the apparent affinity for ThDP is low, ThDP might not be the physiological substrate. An obvious possibility is that the real substrate is a second ADP molecule instead of ThDP. An analogous phenomenon has been reported in the case of adenylate kinase 1: ThDP can replace ADP at one site but is a much poorer substrate . If this were the case for our enzyme an important synthesis of diadenosine triphosphate should be observed with ADP as the sole substrate, but this was not the case. Moreover, the synthesis of AThTP should be impaired when ADP is in excess over ThDP. However, data in Fig. 5 show that there is no tendency to inhibition of AThTP synthesis by excess ADP. Those data strongly suggest that ADP does not bind to the ThDP-binding site with high affinity. This does not exclude, however, that some unknown substrate might replace ThDP.
Concerning the correct systematic name of AThTP-synthesizing enzyme, it should belong to EC subgroup of 2.7.7 of nucleotidyl transferases as a nucleotidyl moiety is transferred to ThDP. Subgroup 2.7.7 comprises a set of enzymes carrying out a nucleotidyl transfer and release of inorganic phosphate (with an NDP as substrate) or pyrophosphate (with an NTP as substrate): X- P- P(- P) + P- Y ⇔ X- P- P- Y + P(- P). Correspondingly, the AThTP-synthesizing enzyme could be named ADP (ATP): thiamine diphosphate adenylyl transferase (EC 220.127.116.11). As both ADP and ATP can act as substrates, we recommend the name of ThDP adenylyl transferase (THAT).