Mycobacterium tuberculosis Phosphoribosylpyrophosphate Synthetase: Biochemical Features of a Crucial Enzyme for Mycobacterial Cell Wall Biosynthesis

The selection and soaring spread of Mycobacterium tuberculosis multidrug-resistant (MDR-TB) and extensively drug-resistant strains (XDR-TB) is a severe public health problem. Currently, there is an urgent need for new drugs for tuberculosis treatment, with novel mechanisms of action and, moreover, the necessity to identify new drug targets. Mycobacterial phosphoribosylpyrophosphate synthetase (MtbPRPPase) is a crucial enzyme involved in the biosynthesis of decaprenylphosphoryl-arabinose, an essential precursor for the mycobacterial cell wall biosynthesis. Moreover, phosphoribosylpyrophosphate, which is the product of the PRPPase catalyzed reaction, is the precursor for the biosynthesis of nucleotides and of some amino acids such as histidine and tryptophan. In this context, the elucidation of the molecular and functional features of MtbPRPPase is mandatory. MtbPRPPase was obtained as a recombinant form, purified to homogeneity and characterized. According to its hexameric form, substrate specificity and requirement of phosphate for activity, the enzyme proved to belong to the class I of PRPPases. Although the sulfate mimicked the phosphate, it was less effective and required higher concentrations for the enzyme activation. MtbPRPPase showed hyperbolic response to ribose 5-phosphate, but sigmoidal behaviour towards Mg-ATP. The enzyme resulted to be allosterically activated by Mg2+ or Mn2+ and inhibited by Ca2+ and Cu2+ but, differently from other characterized PRPPases, it showed a better affinity for the Mn2+ and Cu2+ ions, indicating a different cation binding site geometry. Moreover, the enzyme from M. tuberculosis was allosterically inhibited by ADP, but less sensitive to inhibition by GDP. The characterization of M. tuberculosis PRPPase provides the starting point for the development of inhibitors for antitubercular drug design.


Introduction
Mycobacterium tuberculosis, which is the etiologic agent of tuberculosis (TB), was discovered in 1882 by the German physician Robert Koch. TB was already then considered one of the most dangerous infectious diseases but, continues to still be, unfortunately, a major cause of death in underdeveloped nations, and a re-emerging disease in developed countries. Moreover, TB is currently endemic in the regions of sub-Saharan Africa, where susceptibility of HIV-infected people in developing the disease continuously increases [1].
According to the World Health Organization (WHO), in 2006 there were 9.2 million new cases of TB, and 1.7 million deaths from the disease, of which 95% occurred in low-income countries [2]. TB treatment is made more difficult by the emergence of multidrug resistant strains (MDR-TB), i.e. strains resistant to two of the first-line drugs, either isoniazid or rifampicin. MDR-TB demands treatment with second-line drugs [3][4]. Lately, a still more dangerous form of tuberculosis, i.e. extensively drug-resistant tuberculosis (XDR-TB), has been identified in all regions of the world and is becoming an alarming growing global health problem [5].
For these reasons, an emergence of a global plan to stop TB is necessary and needs the designing of new drugs and the identification of new molecular targets [6][7].
Recent studies have shown that, because of the mycobacterial cell wall's importance as a virulence factor in pathogenicity, it is thus rich in promising drug targets [8]. The mycobacterial cell wall structure is very complex and highly hydrophobic. It is characterized on the outer side by a mycolic acid layer and on the inner side by a peptidoglycan layer. These two layers are linked together by an arabinogalactan complex. It has been demonstrated that enzymes involved in arabinogalactan biosynthesis are essential for the livelihood of M. tuberculosis [9]. This makes these enzymes ideal targets for designing new antitubercular drugs.
Recently, Makarov et al. [10] demonstrated that benzothiazinones, which are a new generation class of antitubercular drugs, act inhibiting M. tuberculosis DprE1 activity, an essential membrane associated enzyme [11][12] that works in concert with the DprE2 enzyme in catalyzing the epimerization of decaprenylphosphorylribose (DPR) to decaprenylphosphoryl-arabinose (DPA), which is a precursor for arabinan synthesis [12]. It is noteworthy that without DPA, a complete mycobacterial cell wall cannot be produced [12].
Within the DPA biosynthesis pathway, other enzymes could be considered potential antitubercular targets such as the phosphoribosylpyrophosphate synthetase (PRPPase).
Three different classes of PRPPase have been described so far with distinctive enzymatic properties, such as the requirement of phosphate ions for activity and allosteric regulation and specificity for the diphosphoryl donor. Most PRPPases belong to class I, and are also named ''classical'' PRPPases. These enzymes, which require phosphate and Mg 2+ ions, are allosterically inhibited by ADP and, possibly, by other nucleotides, and exclusively use ATP or, in some instances, also dATP as diphosphoryl donors [15][16][17]. Class II PRPPases are specific for plants and are characterized by the independence of phosphate ions and the lack of allosteric inhibition by purine ribonucleoside diphosphates. Moreover, class II PRPPases have a broad specificity for diphosphoryl donors using GTP, CTP or UTP in addition to ATP and dATP [18][19][20]. Finally, a new class III PRPPase has been recently described, from the archaeon Methanocaldococcus jannaschii. This enzyme is activated by phosphate and uses ATP as a diphosphoryl donor. Conversely, it is devoid of the allosteric site for ADP [21].
The crystal structures of Bacillus subtilis and human isoform 1 (class I) [22][23], as well as M. jannaschii (class III) PRPPase have been solved [21]. Class I enzymes are hexamers of identical subunits, which consist of two domains that are organized as a propeller with the N-terminal domains at the centre and the C- terminal domains on the outside. The substrates binding sites are located at the interface between the domains of each subunit, whereas the allosteric sites are at the interface between the three subunits of the hexamer. On the contrary, the class III PRPPase is tetrameric. The active sites are at the interface between the domains of the subunits, although no allosteric sites have been found [21].
Our laboratory is aimed at producing enzymes involved in the DPA synthesis, such as DprE1 [10], for structural studies and drug design, as we believe that the enzymes belonging to this pathway could represent a ''weak ring of the chain'' [24].
In this context, the PRPPase enzyme seems very promising being essential as shown by Himar1-based transposon mutagenesis in the M. tuberculosis H37Rv strain [25] and is furthermore involved in two important pathways: the DPA, and purine/ pyrimidine nucleotides biosyntheses.
In this work, the biochemical characterization of the M. tuberculosis PRPPase obtained in recombinant form is reported, as a basis for the identification of a potential antitubercular drug target.

Strains and Growth Conditions
All cloning steps were performed in Escherichia coli DH5a grown in Luria-Bertani (LB) broth or on LB agar. The expression strain was E. coli BL21(DE3)pLysS. When necessary, antibiotics (Sigma) were added at the following concentrations: ampicillin, 100 mg/ ml; chloramphenicol, 34 mg/ml; kanamycin, 50 mg/ml. All strains were grown aerobically at 37uC with shaking at 200 rpm.
Cloning of rv1017c Gene in pET28-a Expression Vector The rv1017c gene (prsA) encoding MtbPRPPase, was amplified by PCR from the genomic DNA of M. tuberculosis H37Rv using Taq DNA Polymerase (Qiagen) with primers Rv101728aF (59-TTGGATCCTTGAGCCACGACTGG-39; BamHI restriction site is underlined) and Rv1017R (59-TTAAGCTTCTATGCG-TCCCCGTCG-39; HindIII restriction site is underlined). The PCR reaction was performed by using the MJ Mini Personal Thermal Cycler (BioRad). The resulting amplified fragment (981 bp) was purified with a Wizard PCR Prep mini-column (Promega), digested with BamHI and HindIII restriction endonucleases, and cloned into pET28-a expression vector (Novagen) by means of T4 DNA ligase in order to form the pET28-a/rv1017 construct which carries a fusion of six histidine residues at its Nterminus [26]. Restriction enzymes and T4 DNA ligase were purchased from GE-Healthcare and used following the manufacturer's instructions.

MtbPRPP Synthetase Heterologous Production and Purification
E. coli BL21(DE3)pLysS cells were electroporated with the pET28-a/rv1017 construct and grown on LB agar plates containing kanamycin (50 mg/ml) and chloramphenicol (34 mg/ ml). Roughly 100 colonies were inoculated in 2 litres of ZYP-5052 autoinducing medium [27] containing kanamycin (50 mg/ml) and chloramphenicol (34 mg/ml), and incubated at 37uC for 3 hrs and at 17uC o. n. with orbital shaking at 200 rpm. Cells were collected by centrifugation (at 60006g for 10 min at 4uC), washed with cold PBS and stored at 220uC. In order to purify the enzyme, frozen cells were suspended in 250 ml buffer A (sodium phosphate pH 8.0, 300 mM NaCl, 10 mM imidazole), supplemented with a protease inhibitor cocktail (Sigma-Aldrich), sonicated at 800 W for 6 minutes, cleared by ultracentrifugation, and the supernatant was applied to a HisTrap HP column (GE-Healthcare) equilibrated in buffer A. Proteins were eluted with scalar concentration (20 to 500 mM) of imidazole in buffer A and fractions containing MtbPRPPase activity were collected, concentrated and applied to a HiLoad 16/60 Super-dex-200 column (GE-Healthcare) equilibrated in buffer B (50 mM potassium phosphate pH 8.0, 100 mM KCl). The enzyme was eluted by buffer B and fractions containing MtbPRPPase activity were checked by 12% SDS-PAGE and pooled. Protein concentration was determined according to Lowry et al. [28].

Enzyme Activity Assay
MtbPRPPase activity was assayed with a HPLC-based method developed in our laroratory (unpublished data), and following the AMP rate formation. The standard reaction mixture contained 50 mM potassium phosphate pH 8.0, 100 mM KCl, 2 mM Mg-ATP, 2 mM R5P, in a final volume of 100 ml. After incubation at 37uC, the reaction was stopped by adding 10% (w/v) ice-cold trichloroacetic acid, and neutralized with 200 mM K 2 CO 3 . After centrifugation, samples (10 ml) were loaded onto a Supelcosil LC-18 column (25064.6 mm, 5 mm particle size, Supelco Analytical). Isocratic separation was performed in 20 mM potassium phosphate pH 8.0 at a flow rate of 0.8 ml/min. Analytes were monitored at 254 nm.
The nmoles of AMP produced were determined using a calibration curve obtained by injecting scalar amounts (0.06 to 20 nmol) of AMP, treated in the same way as that adopted for the enzyme assay. One unit is defined as the amount of enzyme catalyzing the production of 1 mmol of AMP per minute under conditions here described.

Kinetic Analyses
Unless otherwise indicated, enzymatic activity was assayed at 37uC by using various concentrations of R5P and Mg-ATP under conditions identical to those described above except for substrates and effectors.  The kinetic parameters were determined for R5P at 10 mM Mg-ATP and for Mg-ATP at 2 mM R5P. In all cases the reaction was initiated by adding R5P, and the enzyme activity was assayed at least with 12 different concentrations of substrate. All measurements were performed at least in triplicate. The plot of Lineweaver-Burk was used to determine V max and apparent K m values. The Hill plot obtained by the Enzyme Kinetic Module 1.1 (SPSS Science Software) was used to determine the apparent S 0.5 and n H values.
For the assessment of the activation by phosphate or sulfate ions, the enzyme stored in buffer B was diluted in 50 mM Tris HCl pH 8.0 buffer, containing 2mM Mg-ATP, lowering the phosphate concentration to 0.25 mM. The enzyme activity was then immediately assayed at saturating concentrations of substrates, and using as assay buffer 50 mM Tris-HCl pH 8.0, 100 mM KCl, in the presence of different concentrations of potassium phosphate or ammonium sulfate.

Thermal Stability Assays
Thermal stability was measured by incubating the enzyme (100 mg/ml) at given temperatures in buffer B, in the absence and in the presence of ligands. Samples were removed at intervals and immediately assayed as described above.
Relative activity was expressed as percentage of the enzyme activity before the incubation. t 1/2 is the time required by the enzyme to lose 50% of its initial activity at a given temperature.
The thermal denaturation was also measured by circular dichroism spectropolarimetry. Thermal unfolding was followed by continuous measurements of ellipticity at 220 nm at the temperature range 50-90uC under a constant heating rate of 1uC/ min, and with a Jasco J-710 spectropolarimeter (Jasco Europe, Cremella, Italy) equipped with a Neslab RT-11 programmable water bath (Thermo Fisher Scientific, Waltham, MA, USA) and a 1 mm path-length cuvette. Protein concentration was 0.1 mg/ml in buffer B. The midpoint temperatures (T m ) were calculated from curves fitting.

Homology Modelling of MtbPRPPase
The three dimensional structure of MtbPRPPase was modelled using, as the template, the atomic coordinates of the X-ray crystal structure of the human ortholog in complex with AMP, cadmium and sulfate ion (PDB code 2HCR) [23]. The program SWISS-PDBviewer in conjunction with the SWISS-MODEL server (http://www.expasy.org/spdbv/) was employed for building and optimizing the model. The stereochemistry of the predicted structure has been assessed with the program PROCHECK [29]. 92.0% of residues felt in the most favoured region of the Ramachandran plot, 8.0% in the additional allowed region with

Heterologous Expression and Purification of M. tuberculosis PRPPase
The recombinant MtbPRPPase was expressed in E. coli BL21(DE3)pLysS cells, and purified to homogeneity as described in the ''Material and Methods'' section. The typical yield was about 20 mg of purified MtbPRPPase from 1 litre of culture. The specific activity, under standard conditions, was 59.7 U/mg. No detectable activity was found with Mg-GTP used as substrate. As phosphate (P i ) has been reported to be indispensable in preserving protein stability of PRPPases, the MtbPRPPase was maintained in 50 mM phosphate, pH 8.0 [16][17]23]. In actual fact, dialysis against buffers such as 50 mM Tris-HCl, pH 8.0 or 50 mM Hepes-NaOH, pH 8.0 resulted in a protein precipitation and complete loss of activity. The addition of 50 mM ammonium sulfate or 5 mM Mg-ATP to Tris-HCl, pH 8.0 allowed the enzyme to preserve 20% of initial activity after a period of 16 hours, whereas full activity was maintained with the addition of 50 mM P i .

Main Characteristics of MtbPRPPase
Oligomeric state-The enzyme migrated in 12% SDS-PAGE as a protein of apparent molecular mass of approximately 35 kDa ( Fig. 2A) and eluted from a Superose 6 column as a single simmetric peak, corresponding to a 220 kDa protein (Fig. 2B). These results indicated that the recombinant MtbPRPPase was a hexamer of identical subunits.
Dependence on pH-The pH-activity profile for MtbPRPP is shown in Figure 3. The enzyme exhibited preference for high pH values, showing an optimum at a pH value close to 8, and possessing nearly 70% of its maximal activity at pH 9.5. The activity at pH 7 was only 57% of the maximal one. The pH profile exhibited by MtbPRPPase approached that of B. subtilis enzyme [31] Requirements for inorganic phosphate-PRPPases are known to require phosphate for their activity [16][17]23]. MtbPRPPase resulted to be actually dependent on P i for its activity: the optimal P i concentration ranged from 10 mM to 40 mM; higher concentrations of P i were inhibitory (50% inhibition at 100 mM P i ) (Fig. 4A). SO 4 22 ions were also able to stimulate the enzyme activity, but with respect to P i , were less effective and required higher concentrations (40-60 mM) in order to exhibit maximal activation (Fig. 4A). On the contrary, SO 4 22 , at concentrations up to 100 mM, were only faintly inhibitory.
Activation by divalent cations-It has been reported that PRPPases are activated by free divalent cations. At subsaturating Mg-ATP concentrations, MtbPRPPase reached half-maximum activation at approximately 1 mM free ions (Mg 2+ and Mn 2+ , 1.2 mM and 1.1 mM, respectively), although the maximal activity reached in the presence of 5 mM Mg 2+ resulted to be roughly 80% of that in the presence of 5 mM Mn 2+ (Fig. 4B)

Steady State Kinetics as a Function of Substrates Concentration
Steady state kinetics of the recombinant MtbPRPPase as a function of R5P and Mg-ATP, are shown in Figure 5. Main kinetic parameters are summarized in Tables 1 and 2.
At saturating concentration of Mg-ATP, the enzyme exhibited hyperbolic response to R5P (Fig. 5A), with an apparent K m of 0.071 mM. On the contrary, at saturating R5P concentration, it showed sigmoidal behaviour towards Mg-ATP (Fig. 5B), with an apparent S 0.5 of 1.71 mM and a Hill coefficient (n H ) of 2.6.
The presence of 5 mM free Mg 2+ in kinetics towards R5P did not alter the curve profile, whereas 5 mM Mn 2+ raised the maximal activity to 120% (Fig. 5A). As for the response of the enzyme towards Mg-ATP, the presence of 5 mM free Mg 2+ converted the sigmoid curve into a hyperbole, lowering the apparent S 0.5 value and leaving the V max value unchanged ( Fig. 5B and Table 2). A similar effect was obtained by the presence of 5 mM Mn 2+ to the kinetics versus Mn-ATP ( Fig. 5B and Table 2). Notably, the presence of 5 mM Mn 2+ in the kinetics versus Mg-ATP (curve profile not shown) led to kinetic parameters which were nearly identical to those obtained for the kinetics towards Mn-ATP (Table 2).

Inhibition by Divalent Cations
Divalent cations, such as Ca 2+ or Cd 2+ , are reported to inhibit PRPPases [31]. Figure 6A reports the inhibition curves of CuCl 2 , CaCl 2 and FeCl 2 at 5mM Mg-ATP. All ions resulted to be inhibitory, Cu 2+ being the most effective, with an IC 50 (inhibitor concentration lowering enzyme activity to 50%) value of 0.02 versus 0.4 and 0.8 mM of Fe 2+ and Ca 2+ , respectively. The presence of Cu 2+ , Ca 2+ or Fe 2+ at a concentration equal to their IC 50 left the affinity for Mg-ATP unchanged or even slightly increased, as shown by the kinetics towards this substrate (Fig. 6B, Tables 3 and  4). In addition, these ions reduced, but did not completely abolish, the cooperativity towards Mg-ATP (n H value reduced up to 1.4 in the case of Cu 2+ , Table 3). The inhibition was not even removed by using fully activating concentrations of free MgCl 2 , although in the presence of Mg 2+ the curves vs Mg-ATP became hyperbolic. V max values remained similar to those obtained in the presence of inhibitory ions alone (Fig. 6B, Table 3). Comparable inhibitory effects were also observed when Mn-ATP was used as the variable substrate, although the V max values were slightly reduced. The addition of free Mn 2+ abolished the enzyme cooperativity towards the nucleoside triphosphate, leaving the V max values almost unchanged (Fig 6C, Table 4).

Inhibition by ADP
Class I PRPPases are reported to be allosterically inhibited by ADP or by GDP [17]. The inhibition curves of Mg-ADP and Mg-GDP at subsaturating concentrations of Mg-ATP and in the presence of 50mM P i (Fig. 7A) showed that MtbPRPPase was weakly sensitive to GDP (IC 50 .5 mM), whereas it was highly inhibited by ADP (IC 50 0.4 mM). The degree of inhibition by ADP was higher at lower concentration of P i (IC 50 , 0.26 mM at 5 mM P i , Fig. S1), suggesting that ADP inhibition hindered P i in its activatory ability. Thus, inhibition by ADP and activation by P i resulted to occur by competition for binding to the same site.
To prove that ADP was actually an allosteric inhibitor of MtbPRPPase, we assayed the enzyme activity at varying Mg-ATP concentration, in the presence of either 0.5 mM or 1 mM Mg-ADP, with and without 5 mM MgCl 2 (Fig 7B). The presence of the nucleoside diphosphate lowered the V max of the enzyme, without affecting both the apparent S 0.5 and the n H values. The inhibition by Mg-ADP was not removed by the presence of the activating cation (V max values unchanged), although the response towards Mg-ATP became hyperbolic with an affinity for the substrate similar to that displayed in the presence of Mg 2+ without Mg-ADP (Fig. 7B, Table 5). As for the kinetics towards R5P, the presence of Mg-ADP gave effects similar to those observed when the Mg-ATP was used as the variable substrate (Fig 7C), the V max being the only kinetic parameter affected ( Table 6). As far as other potential inhibitors are concerned [31], it is worth mentioning that no inhibitory effects were shown by the presence of pyrimidine nucleoside mono-or diphosphates or of histidine, up to 2 mM (data not shown).

Thermal Stability
The enzyme thermal stability was assessed either by measuring the activity at intervals after incubation at 62uC, or by monitoring the thermal unfolding at increasing temperature with circular dichroism spectropolarimetry.
MtbPRPPase resulted to be a highly stable enzyme, losing 50% of its activity in 10 minutes of incubation at 62uC, and showing a T m of 69.3uC (Table 7). Mg-ATP greatly increased the protein stability, allowing the enzyme to preserve full activity for more than one hour when incubated in the presence of this substrate. A protective effect was also exerted by R5P, although to a lesser extent (t 1/2 22 minutes), whereas no protection was observed in the presence of Mg 2+ ion (Fig. 8A). Similarly, the midpoint temperatures were shifted by the presence of substrate (70.8 and 74.5uC for ATP and R5P, respectively), but not by MgCl 2 (Fig. 8B).

MtbPRPPase Three Dimensional Structure Prediction
We are acutely aware of the issue of selectivity of drug action for inhibitors targeting the MtbPRPPase, as the mycobacterial enzyme shares a significant degree of sequence identity with human counterpart (sequence identity of 44%). Although the identification of possible peculiar structural features to be exploited for the design of specific inhibitors must wait for the determination of the X ray crystal structure of the MtbPRPPase, we carried out a prediction of its structure based on homology modelling. As expected, the overall structural organization of the mycobacterial and human enzymes appeared to be strongly conserved ( Fig. 9A and 9B) as demonstrated by the observation that the two structures can be optimally superimposed with a r.m.s.d. of only 0.5 Å based on 303 Ca pairs. However, the analysis of the ATP binding pocket revealed interesting differences between the two enzymes ( Fig. 9C and 9D). In particular, two major substitutions in the residues that define the nucleoside triphosphate binding site can be identified. In the MtbPRPPase a glutamic acid (Glu113) occupies the structurally equivalent position of Ala105 in the human enzyme; moreover a histidine residue (His109) replaces Asp101 in the human PRPP synthetase. Since MtbPRPPase shows a strong cooperativity for ATP binding, we cannot quantify the impact of these substitutions based on our predicted structure.

Discussion
The biosynthesis pathway of decaprenylphosphoryl-arabinose has been proved to be an optimal target for antitubercular drugs [10,12]. In this context, the characterization of M. tuberculosis phosphoribosylpyrophosphate synthetase, which is the enzyme catalysing the second step of this metabolic pathway, is reported. Noticeably, PRPP, which is the product of the PRPPase catalysed reaction, is also a key metabolite for the nucleotides and for the amino acids histidine and tryptophan synthesis. The rv1017c gene, which encodes PRPPase, is thus essential for M. tuberculosis growth [25].
MtbPRPPase was expressed as recombinant form, purified to homogeneity and biochemically characterized. Although the biochemical characterization of the MtbPRPPase was performed using the enzyme with a hexahistidine tag attached to its Nterminus, as shown in Figure S2, the tag did not affect the main kinetic properties (see Materials and Methods S1).
The enzyme exhibited a hexameric quaternary structure, specificity for Mg-ATP as substrate and requirement of phosphate for its activity. These features allowed us to label MtbPRPP as class I enzyme. SO 4 22 mimicked the activation by P i , although to a lower extent (56%). On the other hand, the inhibitory effect exhibited by P i at high concentrations was negligible in the case of SO 4 22 . In this respect, MtbPRPP turned out to be quite similar to the enzyme from B. subtilis and mammals [22,[32][33].
PRPPAses require both free Mg 2+ ion as an essential activator and Mg-ATP as a substrate. Free ion may induce and properly stabilize the open conformation of the so-called flexible loop which binds Mg-ATP at the active site [34][35]. In the absence of free Mg 2+ , MtbPRPPase showed homotropic cooperativity towards Mg-ATP, which was the cause of a relatively low affinity for this substrate (apparent S 0.5 , 1.71 mM). The presence of free Mg 2+ abolished the cooperativity versus Mg-ATP (n H , 1) and lowered the apparent S 0.5 , suggesting that it activated the enzyme, behaving as an allosteric effector. Moreover, the kinetic properties displayed by MtbPRPPase in the absence and in the presence of the activator Mg 2+ could fulfil the requirements of the K-type allosteric enzyme of the model described by Monod [36]. Comparable heterotropic activation was also exerted by Mn 2+ , which resulted even more effective than Mg 2+ (Table 2) whether the enzyme used Mg-ATP or Mn-ATP as a variable substrate. In this respect, MtbPRPP showed to be different from other class I enzymes, which display maximal activation in the presence of free Mg 2+ ions [17,31,37].
Thermal stability assays allowed us to evidence conformational changes caused by the presence of ligands (Fig. 8). Whereas MtbPRPPase exhibited a more stable conformation in the presence of Mg-ATP (t 1/2 , .2hrs versus 109200 of the enzyme in the absence of ligands), the presence of free Mg 2+ ions did not lead to any increased protein stability (t 1/2 , 119400), suggesting that the binding of the free activating ion did not induce large rearrangements of the protein. Thus, keeping in consideration previous data obtained from crystallographic studies on B. subtilis enzyme [22,35], we hypothesize that the binding of the free Mg 2+ to its site would induce a local conformational change at the active site of the single subunits, stabilizing the open conformation of the flexible loop and abolishing the cooperativity of the Mg-ATP binding sites, but leaving the overall conformation of the enzyme unchanged. On the other hand, the binding of Mg-ATP to one subunit would lead to overall enzyme conformational changes, thus inducing the stabilization of the open active site conformation in the next subunits, and increasing their affinity for Mg-ATP.
Divalent cations, such as Ca 2+ and Cd 2+ , have been reported to inhibit PRPPase activity [32,34]. MtbPRPPase was inhibited by Ca 2+ (IC 50 , 0.8 mM), but the effect of this ion resulted to be less effective than that observed in B. subtilis and human enzymes [32,34]. In actual fact, a higher inhibition was found when the enzyme activity was assayed in the presence of Cu 2+ ions (IC 50 , 0.02 mM). However, in all cases, the reduction of the activity was accompanied by a decrease in the cooperativity towards Mg-ATP and a slight increase in the affinity for this substrate ( Table 3). The inhibition was only partially removed by the addition of either free  [31,32,37], thus suggesting a different geometry of the free cation binding site. Figure 10 shows the sequence alignment of the human, B. subtilis and M. tuberculosis cation binding site, as deduced from the B. subtilis structure [35], and obtained using Multalin 5.4.1 [38]. Arg 180 (B. subtilis numbering), in the absence of cation, establishes a hydrogen bonding network with two aspartic acid residues (Asp 174 and Asp 223 ) devoted to the free Mg 2+ binding, and moves away to a new aspartic acid residue (Asp 133 ) in the presence of the ion. In the MtbPRPPase, Arg 180 , which is also conserved in the human enzyme, is substituted by an isoleucine, whereas two arginines are located one and three residues behind, respectively. These structural differences could very likely be the reason for a different free cation site topology, thus accounting for the different ion specificity.
It is known that class I enzymes are allosterically inhibited by purine diphosphate nucleosides [31][32]. MtbPRPPase acted as the enzymes of this class (Fig. 7A), with non-competitive inhibition by Mg-ADP, either in the absence or in the presence of free Mg 2+ . Similarly to the B. subtilis and Salmonella typhimurium enzymes [31,39], MtbPRPPase was only weakly inhibited by Mg-GDP, distinguishing itself from the mammal enzymes which were more affected by this nucleotide (IC 50 , 10-fold higher) [32][33]. On the other hand, MtbPRPPase was more sensitive to inhibition by ADP than B. subtilis enzyme (IC 50 , 4-fold lower) [31], to this respect behaving like mammal enzymes [32][33]. Interestingly, the concentration of the ADP needed by MtbPRPPase for halfmaximal inhibition increased with increasing P i concentration, thus supporting the conclusions of previous studies that indicate the presence of a regulatory site to which both inhibitory ADP and activatory P i could bind [22]. That MtbPRPPase was regulated by ADP in an allosteric manner resulted by the kinetic responses to substrates concentrations at two different concentrations of ADP. In fact ( Figure 7A and 7B, Table 5 and 6) V max was the only parameter affected. Therefore, MtbPRPPase underwent the inhibition by ADP fully meeting the uncommon requirements of the V-type allosteric enzyme described by Monod et al. [36].
In conclusion, the biochemical investigation on PRPPase from M. tuberculosis allows us to add a well-characterized member to class I enzymes, and to contribute to the elucidation of the regulatory properties of this complex enzyme involved in nucleotides and in the mycobacterial cell wall biosynthesis. The picture emerging from these studies is that of a ''chameleon'' enzyme which adopts different conformations in response to a variety of allosteric effectors, either positive or negative, thus finely adapting the synthesis of PRPP to the variable cell demands. The enzyme characterization may represent the starting point for the development of inhibitors for antitubercular drug design, also in the light of the structural differences with respect to the human counterpart, as suggested by the MtbPRPPase three dimensional structure prediction. Our model supports the notion that the different kinetics shown by the mycobacterial and human PRPPase are likely due to peculiar structural traits of the nucleoside triphosphate binding pocket and suggests that the  identification of selective ligands can be challenged. In this respect, it is worth mentioning that M. tuberculosis ATP phosphoribosyl transferase (HisG) (the enzyme catalysing a reaction one step downstream PRPPase along the same pathway and also showing a significant degree of sequence identity with the human ortholog), has been successfully approached for the discovery of inhibitors selective toward the M. tuberculosis enzyme by exploiting the PRPP binding site in structure based virtual screening [40]. Therefore, although we recognise that the issue of the selectivity of inhibitor action is a major concern in the case of MtbPRPPase, both our extensive biochemical investigation as well as a foreseen more robust structural characterization, may prove to be useful for the design of potent and highly specific inhibitors. Author Contributions