The Diadenosine Hexaphosphate Hydrolases fromSchizosaccharomyces pombe andSaccharomyces cerevisiae Are Homologues of the Human Diphosphoinositol Polyphosphate Phosphohydrolase

Aps1 from Schizosaccharomyces pombe(Ingram, S. W., Stratemann, S. A., and Barnes, L. D. (1999) Biochemistry 38, 3649–3655) and YOR163w fromSaccharomyces cerevisiae (Cartwright, J. L., and McLennan, A. G. (1999) J. Biol. Chem. 274, 8604–8610) have both previously been characterized as MutT family hydrolases with high specificity for diadenosine hexa- and pentaphosphates (Ap6A and Ap5A). Using purified recombinant preparations of these enzymes, we have now discovered that they have an important additional function, namely, the efficient hydrolysis of diphosphorylated inositol polyphosphates. This overlapping specificity of an enzyme for two completely different classes of substrate is not only of enzymological significance, but in addition, this finding provides important new information pertinent to the structure, function, and evolution of the MutT motif. Moreover, we report that the human protein previously characterized as a diphosphorylated inositol phosphate phosphohydrolase represents the first example, in any animal, of an enzyme that degrades Ap6A and Ap5A, in preference to other diadenosine polyphosphates. The emergence of Ap6A and Ap5A as extracellular effectors and intracellular ion-channel ligands points not only to diphosphorylated inositol phosphate phosphohydrolase as a candidate for regulating signaling by diadenosine polyphosphates, but also suggests that diphosphorylated inositol phosphates may competitively inhibit this process.

Following the discovery of dinucleoside polyphosphates in biological systems over 30 years ago (1), these compounds have been studied extensively in prokaryotic and eukaryotic organisms. Several important intracellular and extracellular signaling functions have now been ascribed to the diadenosine com-pounds, Ap 3 A, 1 Ap 4 A, Ap 5 A, and Ap 6 A (2-4). Indeed, the ultimate fate of cell lineages and the very survival of an organism may depend upon the tight control of cellular diadenosine polyphosphate metabolism. For example, the intracellular level of Ap 4 A has long been known to be associated with cell proliferation (5). Moreover, it was recently proposed (6) that Ap 3 A has an antiproliferative role when complexed with the putative tumor suppressor Fhit protein, an Ap 3 A hydrolase (7,8). Thus, the Ap 3 A/Ap 4 A ratio may be an important factor in determining the alternative cellular fates of proliferation, differentiation, and apoptosis (3,9). In higher eukaryotes, Ap n A appear also to be intracellular mediators of certain extracellular stimuli; they respond to glucose in pancreatic ␤-cells (10). Ap n A may also regulate ATP-sensitive K ϩ channels in ␤-cells and cardiac muscle (11)(12)(13) and intracellular ryanodine-binding Ca 2ϩrelease channels in cardiac and skeletal muscle, and in the brain (14,15). Finally, extracellular Ap 5 A and Ap 6 A have also been identified as neurotransmitters (2) and vasomodulators (16,17).
In addition to these physiological functions, Ap n A respond to heat shock and oxidative stress with an increase in concentration (18). If allowed to accumulate, they could prove toxic through their ability to inhibit nucleotide kinases (19,20), protein kinases (21,22), and other enzymes (23). Thus, there is considerable interest in the enzymes that control the synthesis and catabolism of diadenosine polyphosphates.
The MutT/Nudix motif represents one general solution to the challenge of regulating the levels of metabolic intermediates that either act as cellular signals, or can be deleterious to cell function (24). This motif, which appears in a number of proteins from across the phylogenetic spectrum, is characterized by the following (or closely related) sequence: GX 5 EX 7 REUXEEXGU, where U is usually either I, L, or V (24). There is a family of so-called "Ap 4 A" hydrolases, which contain the MutT motif; these enzymes hydrolyze Ap 4 A in preference to 1 The abbreviations used are: Ap n A, diadenosine 5Ј,5ЈЈЈ-P 1 ,P n -oligophosphate (n ϭ 3-6); Aps1, Ap six A hydrolase; p 4 A, adenosine 5Ј-tetraphosphate; p 5 A, adenosine 5Ј-pentaphosphate; PP-InsP 5  The search for mammalian homologues of the yeast Ap 5 A and Ap 6 A hydrolases has now drawn us to the observation that there is a human MutT-type protein with some limited sequence similarity to both Aps1 and YOR163w (Fig. 1). However, this particular human enzyme has not previously been shown to have any significant activity toward nucleoside phosphates: for example, dATP (31) and dGTP 2 are not physiologically significant substrates (6). Instead, this enzyme has been shown to have a quite different catalytic activity; it was identified (31) as a diphosphoinositol polyphosphate phosphohydrolase (DIPP). DIPP's substrates, PP-InsP 5 and [PP] 2 -InsP 4 (31), which are the most highly phosphorylated members of the inositol-based cell-signaling family, are metabolically unrelated to the diadenosine polyphosphates. However, it is of interest that PP-InsP 5 and [PP] 2 -InsP 4 have themselves been strongly implicated as playing important roles in signal transduction. For example, cellular levels of PP-InsP 5 act as a sensor to a specific mode of Ca 2ϩ pool depletion (32). Second, [PP] 2 -InsP 4 turnover is regulated by a cAMP-and cGMP-dependent process that operates independently of A and G kinases (33).
We now describe our discovery that Aps1, YOR163w, and DIPP have overlapping substrate specificities; these enzymes catalyze the hydrolysis of diadenosine polyphosphates and diphosphorylated inositol polyphosphates. This overlapping specificity of an enzyme for two completely different classes of substrate is not only of enzymological significance, but in addition this finding provides important new information pertinent to the structure, function and evolution of the MutT motif.

EXPERIMENTAL PROCEDURES
Materials-Ap 6 A was synthesized as described previously (29). Ap 5 A, Ap 4 A, and Ap 3 A were purchased from Sigma. Non-radiolabeled PP-InsP 5 was synthesized as described previously (34 (28), and DIPP (31) were all obtained as described previously.
Enzyme Assays-For the analysis of Ap n A (n ϭ 3-6) hydrolysis, the substrate was incubated with enzyme in buffer containing 50 mM HEPES (pH 7.6), 1 mM MnCl 2 , and 100 g/ml bovine serum albumin at 37°C . The mass of enzyme, incubation time, and substrate concentrations were varied as described previously to determine substrate specificity, time courses, and substrate saturation curves (29). Assay solutions were analyzed by HPLC to resolve individual nucleotides as described previously (29). For the analysis of PP-InsP 5 and [PP] 2 -InsP 4 hydrolysis, the substrate was incubated with enzyme in buffer containing 50 mM KCl, 50 mM HEPES (pH 7.2), 4 mM CHAPS, 50 g/ml bovine serum albumin, 1 mM Na 2 EDTA, 2 mM MgSO 4 . At the appropriate time, aliquots were quenched, neutralized and subsequently analyzed by HPLC as described previously (31), except that the gradient was generated by mixing buffer A (1 mM Na 2 EDTA) and buffer B (buffer A plus 1.3 M (NH 4 ) 2 HPO 4 , pH 3.85 with H 3 PO 4 ) as follows: 0 -5 min, 0% B; 5-10 min, 0 -50% B; 10 -60 min, 50 -100% B; 60 -70 min, 100% B. Fractions were collected at 1-min intervals, beginning 15 min into the gradient. Fig. 1 compares the sequences of three enzymes that each contain the MutT motif. Two of these proteins, Aps1 from S. pombe (29) and YOR163w from S. cerevisiae (30), have been characterized as Ap 6 A/Ap 5 A hydrolases. The third protein is from Homo sapiens and has been characterized as a PP-InsP 5 / [PP] 2 -InsP 4 phosphohydrolase (31). Much of the DIPP sequence (around 70%) showed no significant similarity to either of the two yeast proteins (Fig. 1). However, the region from Val 34 to Glu 85 in DIPP, which includes the MutT motif and short flanking regions, is 46% identical to the corresponding regions of YOR163w (Val 47 to Glu 99 ) and Aps1 (Val 57 to Lys 108 ). This comparison was of particular interest because DIPP has not been shown to metabolize nucleoside phosphates, and neither Aps1 nor YOR163w has been shown to metabolize diphosphoinositol polyphosphates. We therefore determined if DIPP and the Ap 5 A/Ap 6 A hydrolases could metabolize the other enzyme's 2 S. T. Safrany, unpublished data.

Structural Similarities among Aps1, YOR163w, and DIPP-
FIG. 1. Sequence alignment of Aps1, YOR163w, and DIPP. Two Ap 6 A hydrolases, namely Aps1 from S. pombe (accession no. Q09790) and YOR163w from S. cerevisiae (Z75071), were aligned with human DIPP (AF062529), using the PILEUP algorithm (with the caveat that the only gap permitted within the MutT motif was to accommodate an additional asparagine residue in YOR163w). The MutT motif is shaded gray. Stars denote amino acids in either of the two yeast proteins that were identical to corresponding residues in DIPP; physicochemically similar residues are marked with a cross.
substrates. For these experiments, we used pure, recombinant preparations of each enzyme (see "Experimental Procedures").
Reactivity of DIPP toward Diadenosine Polyphosphates-Ap 6 A was found to be actively metabolized by DIPP (Fig. 2), with multiple products being formed (Fig. 3A). In fact, when incubated with 100 M Ap 6 A, DIPP was virtually as efficient as an Ap 6 A hydrolase (1.1 mol/min/mg) as is either Aps1 or YOR163w, both of which hydrolyze 100 M Ap 6 A at a rate of 1-2 mol/min/mg (29,30). The rank order of relative specific activities of DIPP toward the various diadenosine polyphosphates (Ap 6 A Ͼ Ap 5 A Ͼ Ap 4 A Ͼ Ap 3 A; Fig. 2) mirrored that for Aps1 (29). In terms of relative substrate affinities, YOR163w is slightly different from both Aps1 and DIPP, since YOR163w does not hydrolyze Ap 4 A, possibly due to the insertion of an extra asparagine residue within the putative substrate-binding MutT motif ( Fig. 1) (30). It is significant that DIPP provides the first example of a hydrolase in the animal kingdom that expresses a preference for Ap 5 A and Ap 6 A over other diadenosine polyphosphates.
We also determined k cat and K m values for the hydrolysis of Ap 6 A and Ap 5 A by DIPP (Table I). The affinities of DIPP for these particular substrates (K m ϭ 6 -8 M) were slightly higher than the affinity of Aps1 (approximately 20 M; Ref. 29 The reaction mechanisms of Aps1, YOR163w, and DIPP promise to be interesting to unravel and compare. For the yeast enzymes, it has been noted that Ap 6 A can be hydrolyzed to three different sets of products. Aps1 primarily hydrolyzes Ap 6 A asymmetrically to yield p 4 A and ADP, but there is also some symmetrical hydrolysis to yield 2 ATP as a minor product (29). YOR163w also generates p 4 A and ADP as major products from Ap 6 A, but p 5 A and AMP are also formed as minor products (30). These different modes of Ap 6 A hydrolysis have been attributed to Ap 6 A having some mobility within the active site (30).
The time course of Ap 6 A hydrolysis by DIPP (Fig. 4A) provides insight into this enzyme's reaction mechanisms. In these assays, the rate of consumption of Ap 6 A was matched by the rate of accumulation of AMP plus ADP, at a ratio of about 4:1 (Fig. 4A). These results suggest that the predominant route of Ap 6 A hydrolysis is to AMP plus p 5 A, with the formation of ADP plus p 4 A being a more minor reaction. Although we estimate that approximately 80% of the Ap 6 A was degraded to p 5 A plus AMP, the p 5 A did not accumulate, other than for a small amount after 2 min (Fig. 4A). Presumably, p 5 A is dephosphorylated almost as rapidly as it is formed; p 5 A that was generated from Ap 6 A was also rapidly hydrolyzed (to p 4 A plus P i ) in previous experiments with YOR163w (30). DIPP also hydrolyzed the p 4 A that was formed during Ap 6 A hydrolysis (Fig.  4A). Between the 15-and 35-min time points, the decrease in p 4 A levels was matched by a corresponding increase in ATP levels. Thus, we conclude DIPP hydrolyzed p 4 A to ATP and P i . Furthermore, once all the original Ap 6 A had been consumed (after approximately 15 min), there were only relatively minor increases in the levels of ADP and AMP. Thus, we also conclude that there is relatively little hydrolysis of either p 4 A or ATP to either ADP or AMP. Finally, by the end of the time course, the total amount of [AMP ϩ ADP] that was formed was approximately equivalent to the amount of Ap 6 A added (Fig. 4A). Thus, all the ATP that accumulated can be accounted for by further   (29,30). However, one difference between these two enzymes is that ADP plus ATP are the major reaction products for Aps1 (29), whereas for YOR163w, p 4 A and AMP are the major products (30). When DIPP was incubated with Ap 5 A, this substrate was primarily hydrolyzed to AMP plus p 4 A (Figs. 3B and 4B). ADP accounted for no more than 4% of total nucleoside products (Fig. 4B), indicating that there was relatively little cleavage of Ap 5 A to ADP plus ATP. Thus, the ATP that did accumulate in these reactions (Fig. 4B) must have arisen from the further hydrolysis of p 4 A.
Reactivity of Aps1 and YOR163w toward Diphosphorylated Inositol Polyphosphates-We next discovered that Aps1 and YOR163w expressed phosphohydrolase activity toward PP-InsP 5 (Fig. 5A). Reaction products were identified by HPLC. InsP 6 was formed (Fig. 5A), which indicates that both yeast enzymes cleaved the ␤-phosphate from the diphosphate group of PP-InsP 5 , just as is the case for DIPP (31). No InsP 5 was produced (Fig. 5A), indicating that the pyrophosphate group was not removed from PP-InsP 5 ; the absence of InsP 5 in these reactions also demonstrates that InsP 6 was not a substrate. The yeast enzymes also did not hydrolyze any of the monoester phosphates of PP-InsP 5 , since no PP-InsP 4 was formed (Fig.  5A). Both Aps1 and YOR163w had only an 8-fold lower sub-strate affinity for PP-InsP 5 , compared with DIPP (Table II). k cat values for the three enzymes were also quite similar (Table II).

Reactivity of Other Ap n A Hydrolases toward PP-InsP 5 and [PP] 2 -InsP 4 -The results described above indicate that two
MutT motif Ap 6 A/Ap 5 A hydrolases from yeasts can also hydrolyze PP-InsP 5 or [PP] 2 -InsP 4 . We therefore examined if this overlapping substrate specificity was also a feature of other MutT-type Ap n A hydrolases. In Fig. 1, we noted a region of DIPP (from Val 34 to Glu 85 ) that was 46% identical in sequence to corresponding domains in the yeast Ap 6 A/Ap 5 A hydrolases. This same region of DIPP was only 33% identical to a human Ap 4 A hydrolase (Fig. 6). Although the latter enzyme did hydrolyze PP-InsP 5 and [PP] 2 -InsP 4 , the reactions proceeded relatively slowly (for PP-InsP 5 hydrolysis, k Ϫ1 ϭ 0.005 g Ϫ1 min Ϫ1 (0.0002% of DIPP activity) 2 ; for [PP] 2 -InsP 4 hydrolysis, k Ϫ1 ϭ 0.006 g Ϫ1 min Ϫ1 (0.007% of DIPP activity, see above)).
The IalA invasion protein of B. bacilliformis is another MutT motif protein, and this enzyme expresses broadly similar catalytic activities toward Ap 4 A, Ap 5 A and Ap 6 A (27, 28). The IalA protein, when incubated at up to a 120-fold higher protein concentration than was used for Aps1, showed no detectable hydrolysis of diphosphorylated inositol polyphosphates (data not shown). Indeed, outside the MutT motif, IalA has no significant similarity to DIPP (Fig. 6). Further evidence that the MutT motif does not generally impart the capacity to metabolize PP-InsP 5 and [PP] 2 -InsP 4 comes from experiments showing that these compounds are not significant substrates of hMTH1, 3 a human homologue of the prototypical MutT protein (which instead hydrolyzes 8-oxo-dGTP; Ref. 39).
Nature has developed more than one solution to the problem of how to regulate cellular levels of diadenosine polyphosphates. As well as Aps1, YOR163w, and DIPP, which belong to the MutT motif family (see above), some Ap n A hydrolases belong to the HIT protein family and are characterized by the presence of a histidine triad motif, HxHxH (7,8,40). We incubated one such enzyme, the human Fhit Ap 3 A hydrolase (7) with either PP-InsP 5 or [PP] 2 -InsP 4 . Neither substrate was metabolized by a concentration of Fhit that was 1000-fold greater than was used for Aps1 (data not shown).
In view of these results, we conclude that the remarkable overlapping substrate specificity seen in Aps1, YOR163w, and DIPP is neither a general property of MutT-type proteins, nor is it a general feature of Ap n A metabolizing enzymes.

DISCUSSION
The MutT paradigm is based on this motif recurring in a series of nucleoside phosphate hydrolases that act as "guardians" of cellular integrity (24). Members of this enzyme family have been shown to hydrolyze nucleoside phosphates that are directly hazardous, such as the mutagenic 8-oxo-dGTP (39), and they may also protect the cell against the potentially dangerous consequences of inappropriate increases in the levels of intracellular signals, such as dATP (41). To achieve efficiency in this role, it has previously seemed that individual MutT motif proteins have each evolved exclusive catalytic specificities toward restricted groups of nucleoside phosphates (24). Our results with Aps1, YOR163w, and DIPP are therefore exceptional, because the diadenosine polyphosphates, Ap 5 A and Ap 6 A, and the diphosphorylated inositol phosphates, PP-InsP 5 and [PP] 2 -InsP 4 , are two unrelated classes of substrates. Furthermore, these two groups of metabolites have independently emerged as participants in various aspects of signal transduction (see the Introduction). Since specificity of action is paramount for an intracellular signal, it is remarkable that one enzyme can hydrolyze these different molecules, thereby potentially regulating their levels and affecting their signaling strength. Thus, our observations not only draw an unexpected link between two disparate areas of signal transduction, but they are also pertinent to the evolution and functions of the MutT motif.
Evidence has accumulated that Ap 6 A and Ap 5 A are cellular signals that may act both outside (2,16,17,42) and inside (12)(13)(14)(15) the cell. There is therefore considerable interest in characterizing the enzymes that are responsible for metabolizing these two diadenosine polyphosphates. DIPP represents the first hydrolase in the animal kingdom shown to prefer Ap 6 A and Ap 5 A over Ap 4 A. In fact, the kinetic parameters associated with DIPP-dependent Ap 6 A/Ap 5 A hydrolase activity are not only similar to those of Aps1 and YOR163w (Tables I and II), they are also in line with the characteristics of other MutT motif Ap n A hydrolases (25)(26)(27)(28)30). In that sense, DIPP is a typical Ap 6 A/Ap 5 A hydrolase. This conclusion raises the important question as to whether this is a function of DIPP in vivo. Our in vitro data suggest that PP-InsP 5 is preferred as a substrate over either Ap 6 A or Ap 5 A (Tables I and II). However, we do not know the subcellular compartments in which DIPP is present, nor do we have reliable estimates of the concentrations, in these same compartments, of any of DIPP's substrates. Nevertheless, if DIPP were closely associated with the secretory vesicles in which Ap 6 A and Ap 5 A appear to be concentrated, at least in platelets, chromaffin cells and synapses (2), this enzyme could regulate the signaling strength inside the vesicles. If some DIPP were present on the cell surface, it could contribute to terminating the interactions of Ap 6 A and Ap 5 A with purinergic receptors (2). Finally, DIPP that is free in the cytosol may be a site for metabolic competition between Ap 6 A, Ap 5 A, PP-InsP 5 , and [PP] 2 -InsP 4 .
The fact that, to date, yeasts have not been reported to contain either Ap 5 A or Ap 6 A has raised questions concerning the possible biological functions of Aps1 and YOR163w (29,30). While the search for Ap 5 A and Ap 6 A in yeasts continues, it is worth noting that both S. pombe (43) and S. cerevisiae (44) are known to synthesize PP-InsP 5 and [PP] 2 -InsP 4 . Thus, another important feature of our studies is that PP-InsP 5 and [PP] 2 -InsP 4 represent a useful, new focus for further studies into the functions of YOR163w and Aps1. The identification of a PP-InsP 5 /[PP] 2 -InsP 4 phosphohydrolase in a genetically tractable organism such as yeast also provides us with new opportunities to rapidly increase our insight into the roles of these particular compounds. This is a particularly timely opportunity in the light of a recent observation from studies with S. pombe; the size of this yeast's InsP 6 metabolic reservoir, which in turn dictates the amount of PP-InsP 5 and [PP] 2 -InsP 4 inside cells (37), was recently shown to be acutely sensitive to hyperosmotic stress (43).
Another important feature of our study relates to the value of comparing related enzymes from different species, which can provide fresh insight into relationships between protein structure and function. The sequence alignment of DIPP with yeast Ap 6 A/Ap 5 A hydrolases (Fig. 1) identifies a number of conserved amino acid residues both inside the MutT motif, and in the immediately adjacent flanking regions. A systematic study into the contribution made by these residues, by site-directed mutagenesis for example, will be of great value in future studies into the catalytic process. Thus, our studies represent an important step toward the goal of understanding how the tertiary structures of this class of MutT-type proteins can accommodate two different substrate structures so successfully. Further in- formation on structure/function relationships will also come from the comparison of the sequence of DIPP with the sequences of other MutT-type enzymes that we have now shown not to attack PP-InsP 5 and [PP] 2 -InsP 4 (see Fig. 6 and "Results"). A popular viewpoint concerning the evolution of catalytic motifs envisages relatively nonspecific progenitors becoming duplicated and then adapted to perform specific tasks (45). The MutT motif has been proposed to exemplify this evolutionary process (24). It has been proposed that, from a primordial ancestor, a family of proteins with MutT motifs each have evolved subtly different phosphohydrolase specificities toward particular nucleoside phosphates (24). A contrary view states that a protein motif that provides a particular advantage can arise independently, in otherwise unrelated proteins, by convergent evolution (46). It is arguable that the latter idea is supported by the considerable divergence in the protein sequences that lie outside the eponymous motif of the MutTrelated proteins (see Figs. 1 and 6 and Ref. 24). However, sequence is less tightly conserved during divergent evolution than is structure (47). In view of these conflicting ideas, Aps1, YOR163w, and DIPP represent models of particular interest for further studies into the evolution of functions and tertiary structures of the MutT family.
It has been suggested that "Nudix" is a more accurate name for the "MutT" motif, in part because of the earlier data indicating that this motif imparted catalytic specificity toward compounds containing a nucleoside diphosphate attached to another moiety, "x" (24). Thus, the original discovery of a MutT motif in DIPP created an important precedent; DIPP became the first member of this class of enzymes shown to actively metabolize substrates that do not conform to the above structure. Our new data with Aps1 and YOR163w extend this exception. The unexpected catalytic promiscuity of this group of enzymes is an important new factor that will have to be taken into account in our ongoing efforts to understand how cells regulate the cellular levels and the biological functions of both diadenosine polyphosphates and diphosphorylated inositol phosphates.