Nudix hydrolases that degrade dinucleoside and diphosphoinositol polyphosphates also have 5-phosphoribosyl 1-pyrophosphate (PRPP) pyrophosphatase activity that generates the glycolytic activator ribose 1,5-bisphosphate.

A total of 17 Nudix hydrolases were tested for their ability to hydrolyze 5-phosphoribosyl 1-pyrophosphate (PRPP). All 11 enzymes that were active toward dinucleoside polyphosphates with 4 or more phosphate groups as substrates were also able to hydrolyze PRPP, whereas the 6 that could not and that have coenzyme A, NDP-sugars, or pyridine nucleotides as preferred substrates did not degrade PRPP. The products of hydrolysis were ribose 1,5-bisphosphate and P(i). Active PRPP pyrophosphatases included the diphosphoinositol polyphosphate phosphohydrolase (DIPP) subfamily of Nudix hydrolases, which also degrade the non-nucleotide diphosphoinositol polyphosphates. K(m) and k(cat) values for PRPP hydrolysis for the Deinococcus radiodurans DR2356 (di)nucleoside polyphosphate hydrolase, the human diadenosine tetraphosphate hydrolase, and human DIPP-1 (diadenosine hexaphosphate and diphosphoinositol polyphosphate hydrolase) were 1 mm and 1.5 s(-1), 0.13 mm and 0.057 s(-1), and 0.38 mm and 1.0 s(-1), respectively. Active site mutants of the Caenorhabditis elegans diadenosine tetraphosphate hydrolase had no activity, confirming that the same active site is responsible for nucleotide and PRPP hydrolysis. Comparison of the specificity constants for nucleotide, diphosphoinositol polyphosphate, and PRPP hydrolysis suggests that PRPP is a significant substrate for the D. radiodurans DR2356 enzyme and for the DIPP subfamily. In the latter case, generation of the glycolytic activator ribose 1,5-bisphosphate may be a new function for these enzymes.

The Nudix hydrolases are members of an enzyme family that was named after their ability to hydrolyze predominantly the pyrophosphate linkage in a variety of compounds having the general structure of a nucleoside diphosphate (Npp) linked to another moiety, X, with varying degrees of specificity (1,2). Thus, nucleoside triphosphates (Npp-p), dinucleoside polyphosphates (Npp-p n N), nucleotide sugars (Npp-sugar), NADH, and coenzyme A are examples of Nudix hydrolase substrates that fall into this category. In general terms, the members of this protein family are believed to rid the cell of potentially deleterious endogenous nucleotide metabolites and to modulate the accumulation of metabolic intermediates by diverting them into alternative pathways in response to biochemical need, although specific regulatory functions may also be associated with individual members (1).
The ability of this subset of Nudix hydrolases to utilize a sugar pyrophosphate as a substrate prompted us to test another such compound of known biological importance, 5-phosphoribosyl 1-pyrophosphate (PRPP). PRPP is both a substrate and regulator of purine, pyrimidine, and pyridine nucleotide synthesis (18 -21); in bacteria and lower eukaryotes it is also a precursor for histidine and tryptophan biosynthesis (22)(23)(24)(25). Furthermore, a potential product of pyrophosphatase activity acting upon PRPP is ribose 1,5-bisphosphate (Rib-1,5-P 2 ), which has recently been shown to be a physiological regulator of glycolysis and the fructose 6-phosphate/fructose 1,6-bisphosphate cycle (26 -28). Here, we show that Nudix hydrolases of the DIPP subfamily and the related Ap 4 A hydrolases all exhibit PRPP pyrophosphatase activity, whereas Nudix hydrolases previously shown to be specific for NDP-sugars, pyridine nucleotides, and coenzyme A are unable to hydrolyze PRPP.

MATERIALS AND METHODS
Assay of PRPP Pyrophosphatase Activity-Initial screening of enzyme preparations for their ability to release P i from PRPP was carried out using a phosphomolybdate colorimetric assay (33). Purified enzymes (5 g) were incubated for 15 min at 37°C in 50 mM Tris-HCl, pH 8.0, 5 mM Mg acetate, 1 mM dithiothreitol, and 0.2 mM PRPP in a total volume of 200 l and the reactions stopped by the addition of the molybdate detection reagent. Enzymes testing negative in an initial screen were retested with up to 12 g of protein per assay. Kinetic constants for PRPP hydrolysis by selected enzymes were calculated from initial rates determined using a sensitive continuous spectrophotometric assay based on the phosphate-dependent conversion of 2-amino-6-mercapto-7-methylpurine riboside to 2-amino-6-mercapto-7-methylpurine and ribose-1-phosphate catalyzed by purine nucleoside phosphorylase (34). For this, the EnzChek assay kit was used according to the manufacturer's instructions with the following modifications: reactions were preincubated for 5 min without substrate at 37°C in 50 mM BisTrisPropane buffer, pH 8.5, 5 mM Mg acetate, and 1 mM dithiothreitol and then incubated with substrate for up to 2 min. Final enzyme concentrations were 75 g/ml (human Ap 4 A hydrolase), 15.2 g/ml (DIPP-1), and 7 g/ml (D. radiodurans Ap n A hydrolase). In each case, controls lacking substrate or enzyme were performed.
HPLC Analysis of PRPP Hydrolysis Products-The products of PRPP hydrolysis by the D. radiodurans Ap n A hydrolase were generated by incubation of 0.2 mM PRPP with 7 g of enzyme in 50 mM Tris-HCl, pH 8.0, 5 mM Mg acetate, 1 mM dithiothreitol for 10 min at 37°C. Reactions (200 l) were stopped by freezing and then applied to a 1-ml Resource-Q column (Pharmacia) at 2 ml/min in 35 mM NH 4 HCO 3 , pH 9.6. The elution system consisted of a gradient of 5-100% buffer A (0.7 M NH 4 HCO 3 , pH 9.6) in water over 10 min. Fractions (0.5 ml) were collected and the presence of products determined by colorimetric determination of phosphate released after incubation with alkaline phosphatase or inorganic pyrophosphatase as required (33,35).
Other Methods-Ap 4 A hydrolase activity was determined luminometrically as previously described (3). Protein concentrations were estimated by the Coomassie Blue binding method (36). Positive ion electrospray mass spectrometry was performed as previously described (11).

Hydrolysis of PRPP by Nudix
Hydrolases-A total of 17 different Nudix hydrolases were tested for their ability to hydrolyze PRPP. The assay employed measures the release of inorganic phosphate. As can be seen from Table I, all 11 enzymes that can utilize dinucleoside polyphosphates with 4 or more phosphate groups as substrates are also able to hydrolyze PRPP, whereas the 6 that cannot and that have coenzyme A, NDP-sugars, or pyridine nucleotides as preferred substrates do not degrade PRPP. It should be stressed that the rates quoted of micromoles of P i produced min Ϫ1 ⅐micromoles protein Ϫ1 are not directly convertible to true k cat values because a standard set of conditions, including a fixed PRPP concentration of 200 M, was employed for all assays and so is not necessarily optimal for each enzyme. Therefore, the rank order of PRPP hydrolase activity should only be taken as a guide. Because PRPP is known to undergo spontaneous, Mg 2ϩ -dependent nonenzymic hydrolysis (37,38), it was important to establish that the observed activity was enzyme-catalyzed. This is clearly demonstrated by the lack of degradation by two active site mutants of the C. elegans Ap 4 A hydrolase (Table I). In common with other Nudix hydrolases, substitution of Glu residues in the Nudix motif by Gln dramatically reduced activity of this K m and k cat values for PRPP were determined for one prokaryotic Ap n A hydrolase, one eukaryotic Ap 4 A hydrolase, and one eukaryotic Ap 6 A hydrolase/DIPP, using a continuous spectrophotometric assay. D. radiodurans DR2356 Ap n A hydrolase is an enzyme that hydrolyzes a variety of nucleoside and dinucleoside polyphosphates, including p 4 A, p 5 A, Ap 4 A, Ap 5 A, and Ap 6 A. 5 It has no activity toward diphosphoinositol polyphosphates 4 ; however, it readily hydrolyzes PRPP with a K m of 1 mM and a k cat of 1.5 s Ϫ1 (Table II). For comparison, K m and k cat values for Ap 4 A were determined to be 30 M and 0.035 s Ϫ1 , respectively, resulting in similar specificity constants (k cat /K m ) for both substrates (Table II). Human Ap 4 A hydrolase also hydrolyzed PRPP with a k cat of 0.057 s Ϫ1 and a K m of 0.13 mM (Table II). In this case, however, the specificity constant with PRPP was some 20,000-fold lower than that found with Ap 4 A as substrate. Finally, human DIPP-1 was the most efficient of the three enzymes at PRPP hydrolysis in vitro with a k cat of 1.0 s Ϫ1 and a K m of 0.38 mM (Table II). The specificity constant with PRPP of 2,600 was 325-and 18,000-fold lower than those previously measured with Ap 6 A and PP-InsP 5 , respectively (5). The physiological significance of these data is discussed below.
Identification of Rib-1,5-P 2 as the Product of PRPP Hydrolysis-Because all Nudix hydrolases cleave pyrophosphate linkages, it was anticipated that they would remove the ␤-phosphate from the pyrophosphate moiety attached to the ribose C1. Using the D. radiodurans DR2356 Ap n A hydrolase as an example enzyme, the products of hydrolysis were first separated by anion-exchange HPLC, fractions collected and incubated with alkaline phosphatase, and the P i released determined colorimetrically. Two products, A and B, were observed that co-chromatographed with P i and PP i , respectively (Fig. 1). Peak area integration showed the ratio of phosphate released from PRPP, B, and A to be exactly 3:2:1. Because A did not co-chromatograph with ribose-1-phosphate (Rib-1-P) or ribose 5-phosphate (Rib-5-P) it must be P i . Product B could be Rib-1,5-P 2 , ribosyl 1-pyrophosphate, 5-phosphoribosyl 1,2-(cyclic) phosphate (37,38), or, less likely, PP i itself. Therefore, a sample of product B was subjected to TLC before and after acid treatment. Acid removes phosphate from the anomeric C1 but not from C5, and ribose and derivatives with an unesterified C1 hydroxyl can be detected after TLC by AgNO 3 treatment (38). Fig. 2 shows the TLC plate after AgNO 3 treatment. It can clearly be seen that product B was not detected by AgNO 3 before acid treatment because it has an esterified C1-OH (lane C); however, after acid treatment, product B generated a visible spot that co-chromatographed with Rib-5-P and acid-treated PRPP (lane F). This indicates that product B must be Rib-1,5-P 2 . This identification was confirmed by positive ion electrospray mass spectrometry. B had a mass of 333 Da, correspond-   ing to the monosodium salt of Rib-1,5-P 2 (not shown). A cyclic phosphate product would have had a mass 17 Da less. Therefore it can be concluded that the D. radiodurans DR2356 Ap n A hydrolase is a PRPP pyrophosphatase producing Rib-1,5-P 2 and P i from PRPP. The products of PRPP hydrolysis generated by the human Ap 4 A hydrolase had the same HPLC retention times as those produced by the D. radiodurans DR2356 Ap 4 A hydrolase. Therefore, given the conserved reaction mechanism employed by all Nudix hydrolases, it seems highly likely that Rib-1,5-P 2 and P i are the products in all cases.

DISCUSSION
The results reported here are important for two reasons. First, they have implications for the specificity and physiological function(s) of Nudix hydrolases, particularly the DIPP/ Ap n A hydrolases, and secondly, they impact on the metabolism of PRPP and the generation from it of the regulatory molecule Rib-1,5-P 2 . Regarding the enzymes themselves, the surprising ability of the active sites of the DIPPs to accommodate two seemingly unrelated sets of substrates, the diadenosine and diphosphoinositol polyphosphates, has already been highlighted (5). The use of PRPP as a substrate by these enzymes makes this easier to understand. Modeling of PRPP onto the crystal structure of the C. elegans Ap 4 A hydrolase (39) shows that it can fit readily into the substrate-binding cleft with its 5-phosphate located in the exact position occupied by the P 1 phosphate of Ap 4 A and the ␣-phosphate of the pyrophosphate moiety located where the attacked P 4 phosphate of Ap 4 A lies (Fig. 3). The same water (or hydroxyl) responsible for nucleophilic attack at P 4 could attack this ␣-phosphate. This model is consistent with the requirement for the catalytic residues Glu 52 and Glu 56 in the C. elegans Ap 4 A hydrolase. The ␤-phosphate of the pyrophosphate group can readily occupy the position of the ribose moiety of the 'AMP product' of Ap 4 A because this is already known to accommodate the P 5 phosphate of Ap 5 A and to lie outside the protein structure. The ribose ring of PRPP occupies the same position as the P 2 and P 3 phosphates of Ap 4 A. The substrate-binding cleft is wide at this point, and the residue side chains in this region are either small (Ala 5 , Ala 25 , Thr 33 , and Gly 37 ) or are highly mobile and lack electron density in the crystal structure of the C. elegans hydrolase (Tyr 27 ); therefore, a variety of structures may be accommodated in this region because it appears to provide few, if any, phosphatespecific contacts.
A major determinant of the stability of the C. elegans Ap 4 A hydrolase-PRPP complex is probably the binding of the 5-phosphate by salt bridges/H-bonds to His 31 , Lys 36 , Tyr 76 , and Lys 83 . These residues are important for binding the P 1 phosphate of Ap 4 A (39). Stacking of the adenosine moiety attached to P 1 between the aromatic rings of Tyr 76 and Tyr 121 does not seem to be as critical for Ap 4 A binding as was predicted from the crystal structure. We have found that mutation of both Tyr residues to Ala still yields active Ap 4 A hydrolase with a 20-fold lower k cat (1.1 s Ϫ1 ) and only a 4-fold higher K m (33 M) for Ap 4 A. 2 Thus, the binding of the PRPP phosphates on C5 and C1 as in Ap 4 A and the accommodation of the ribose in a relatively nonspecific region of the binding cleft appear to be sufficient to allow PRPP to behave as a substrate for the C. elegans Ap 4 A hydrolase. Similar arguments presumably apply to the other Ap 4 A hydrolases and DIPPs and to the ability of the DIPPs to bind diphosphoinositol polyphosphates. This model also explains why the Nudix hydrolases with specificities for NADH, NDP-sugars, and coenzyme A do not accept PRPP as a substrate. These enzymes do not have a second phosphate binding site (like P 1 ) located the required distance away from the catalytic binding site (like P 4 ). Only enzymes able to hydrolyze nucleotide substrates with four or more phosphates in the polyphosphate chain should bind PRPP. These results emphasize the point that certain Nudix hydrolases can also accept non-nucleotide substrates and should prompt the search for other such substrates that satisfy these minimal binding requirements.
Is PRPP a physiologically relevant substrate for any of the enzyme studies here? According to the measured specificity constants, at least for the eukaryotic enzymes, the diadenosine and diphosphoinositol polyphosphates appear to be highly favored over PRPP. However, this does not take into account the relative substrate concentrations in vivo. Literature values for the intracellular concentration of PRPP vary considerably and have been reported in different units. However, taking various measurements for prokaryotes (25,40,41) and eukaryotes (20,26,(42)(43)(44) and applying the unit conversion factors of Traut (45) suggests that prokaryotes typically have a PRPP steadystate concentration of around 1 mM, whereas in eukaryotes (excluding erythrocytes) it is 1-2 orders of magnitude lower, although it can be as high as 0.4 mM in mouse fibroblasts deficient in adenine and hypoxanthine phosphoribosyltransferases (43,45). The measured K m values for PRPP for all three enzymes are, therefore, within acceptable ranges if PRPP were to be considered a physiologically relevant substrate. The intracellular concentration of Ap 4 A in unstressed cells is typically 0.1-1.0 M (46), whereas PP-InsP 5 may be in the low micromolar range (47). There are no measurements of cytoplasmic Ap 6 A in mammalian cells; an estimate of 32 nM in platelets is an intracellular average because the Ap 6 A is concentrated in the dense granules (48). Indeed, it is likely to be even lower than the sole measured value for Ap 5 A of 4 nM in Schizosaccharomyces pombe (52). Taking the product of the specificity constant and substrate concentration as a better indication of potential substrate utilization in vivo shows that PRPP hydrolysis is likely to be a much more significant reaction in vivo for the D. radiodurans Ap n A hydrolase than is Ap 4 A hydrolysis (Table II). In contrast, the human Ap 4 A hydrolase is more likely to act upon Ap 4 A than on PRPP. For DIPP-1, PP-InsP 5 still appears to be the favored substrate by virtue of its ex- tremely low K m for this compound. Nevertheless, the combined activities of members of the DIPP subfamily could still have a significant impact on PRPP hydrolysis in vivo in tissues where more than one is expressed.
PRPP pyrophosphatase activity in cell extracts has been detected before. Divalent ion-dependent (49) and -independent (50) activities were ascribed to acid and alkaline phosphatase, respectively. Like the spontaneous degradation of PRPP, these reactions are believed to proceed via 1,2 and 1,5 cyclic derivatives to Rib-1-P and -5-P and ultimately to ribose with Rib-1,5-P 2 as a possible minor intermediate in one pathway (37,38). In view of the established mechanisms of Nudix hydrolases and the fact that Ap 4 A hydrolysis is known to proceed by direct in-line attack of water (51), a cyclic intermediate is unlikely; generation of Rib-1,5-P 2 most probably occurs directly by nucleophilic attack of water on the C1 ␣-phosphate rather than via a complex route involving the initial internal attack of a ribose hydroxyl. Recently, it has been clearly demonstrated (26) that the rapid rise in Rib-1,5-P 2 that occurs in macrophages under hypoxic conditions in parallel with the switch to anaerobic glycolysis is due to a rise in PRPP accompanied by the activation of an unidentified PRPP pyrophosphatase. This activity appears to be divalent ion-independent and may be activated by protein kinase C. Its possible relationship to any of the mammalian Nudix PRPP pyrophosphatases described here is unknown. Nevertheless, it is clear that, at least in mammalian cells, several Nudix hydrolases exist that have the ability to generate Rib-1,5-P 2 from PRPP. Like fructose 2,6-bisphosphate, this molecule is a potent activator of phosphofructokinase, is also an inhibitor of fructose 1,6-bisphosphatase, and is believed to be an important regulator of glycolysis (26 -28). The Nudix PRPP pyrophosphatases must be considered potential generators of Rib-1,5-P 2 in vivo and, therefore, regulators of glucose metabolism. Verification of this possibility will require measurements of PRPP and Rib-1,5-P 2 in cells in which the relevant Nudix hydrolase activities have been reduced by gene disruption or knockdown.