Site-directed Mutagenesis of Diphosphoinositol Polyphosphate Phosphohydrolase, a Dual Specificity NUDT Enzyme That Attacks Diadenosine Polyphosphates and Diphosphoinositol Polyphosphates*

Diphosphoinositol polyphosphate phosphohydrolase (DIPP) hydrolyzes diadenosine 5′,5‴-P1,P6-hexaphosphate (Ap6A), a Nudix (nucleosidediphosphate attached-moiety “x”) substrate, and two non-Nudix compounds: diphosphoinositol pentakisphosphate (PP-InsP5) and bis-diphosphoinositol tetrakisphosphate ((PP)2-InsP4). Guided by multiple sequence alignments, we used site-directed mutagenesis to obtain new information concerning catalytically essential amino acid residues in DIPP. Mutagenesis of either of two conserved glutamate residues (Glu66 and Glu70) within the Nudt (Nudix-type) catalytic motif impaired hydrolysis of Ap6A, PP-InsP5, and (PP)2-InsP4 >95%; thus, all three substrates are hydrolyzed at the same active site. Two Gly-rich domains (glycine-rich regions 1 and 2 (GR1 and GR2)) flank the Nudt motif with potential sites for cation coordination and substrate binding. GR1 comprises a GGG tripeptide, while GR2 is identified as a new functional motif (GX 2GX 6G) that is conserved in yeast homologues of DIPP. Mutagenesis of any of these Gly residues in GR1 and GR2 reduced catalytic activity toward all three substrates by up to 95%. More distal to the Nudt motif, H91L and F84Y mutations substantially decreased the rate of Ap6A and (PP)2-InsP4 metabolism (by 71 and 96%), yet PP-InsP5 hydrolysis was only mildly reduced (by 30%); these results indicate substrate-specific roles for His91and Phe84. This new information helps define DIPP's structural, functional, and evolutionary relationships to Nudix hydrolases.

Diphosphoinositol polyphosphate phosphohydrolase (DIPP) hydrolyzes diadenosine 5,5ٟ-P 1 ,P 6 -hexaphosphate (Ap 6 A), a Nudix (nucleoside diphosphate attached-moiety "x") substrate, and two non-Nudix compounds: diphosphoinositol pentakisphosphate (PP-InsP 5 ) and bis-diphosphoinositol tetrakisphosphate ((PP) 2 -InsP 4 ). Guided by multiple sequence alignments, we used site-directed mutagenesis to obtain new information concerning catalytically essential amino acid residues in DIPP. Mutagenesis of either of two conserved glutamate residues (Glu 66 and Glu 70 ) within the Nudt (Nudix-type) catalytic motif impaired hydrolysis of Ap 6 A, PP-InsP 5 , and (PP) 2 -InsP 4 >95%; thus, all three substrates are hydrolyzed at the same active site. Two Gly-rich domains (glycine-rich regions 1 and 2 (GR1 and GR2)) flank the Nudt motif with potential sites for cation coordination and substrate binding. GR1 comprises a GGG tripeptide, while GR2 is identified as a new functional motif (GX 2 GX 6 G) that is conserved in yeast homologues of DIPP. Mutagenesis of any of these Gly residues in GR1 and GR2 reduced catalytic activity toward all three substrates by up to 95%. More distal to the Nudt motif, H91L and F84Y mutations substantially decreased the rate of Ap 6 A and (PP) 2 -InsP 4 metabolism (by 71 and 96%), yet PP-InsP 5 hydrolysis was only mildly reduced (by 30%); these results indicate substrate-specific roles for His 91 and Phe 84 . This new information helps define DIPP's structural, functional, and evolutionary relationships to Nudix hydrolases.
The mammalian diphosphoinositol polyphosphate phosphohydrolase (DIPP) 1 was originally characterized (1) as being responsible for the dephosphorylation of PP-InsP 5 and (PP) 2 -InsP 4 (Fig. 1), which are the most highly phosphorylated mem-bers of the inositide signaling family of molecules (2). PP-InsP 5 and (PP) 2 -InsP 4 are widely distributed across the phylogenetic spectrum; they have been identified in mammals (3,4), yeasts (5), slime molds (6 -8), plants (9), and free living amoebae (10,11). The dephosphorylation of PP-InsP 5 and (PP) 2 -InsP 4 results in a considerable free energy change, in no small part due to the relief of the severe electrostatic and steric constraints imposed by the high density of phosphate groups that are clustered around the inositol ring (7). Nevertheless, in vivo there is rapid turnover of both PP-InsP 5 and (PP) 2 -InsP 4 , so the levels of these compounds must be continually replenished at a robust rate. This ATP-consuming process therefore places a substantial burden upon the cell's energy reserves (12). These observations have led to the suggestion that PP-InsP 5 and (PP) 2 -InsP 4 turnover might act as a molecular switch for a very dynamic cellular process, for which vesicle trafficking has emerged as a candidate (13,14). Support for the idea that PP-InsP 5 in particular might regulate the movement of vesicles through the cell comes from demonstrations that some of the proteins that regulate this process bind PP-InsP 5 very tightly; examples include coatomer (5,15), AP-2 (see Ref. 3), and AP-3 (16). Another energy-consuming activity that may be nominated for being regulated by diphosphoinositol polyphosphates is mRNA export from the nucleus. Recent studies indicate that efficient mRNA export in yeasts requires the synthesis of inositol hexakisphosphate (17). However, rather than inositol hexakisphosphate participating in this process directly, as was originally proposed (17), the significance of inositol hexakisphosphate may actually be as the precursor of PP-InsP 5 and (PP) 2 -InsP 4 (12), the turnover of which can more readily account for some of the energy-dependent aspects of mRNA export.
The contention that it is the actual turnover of diphosphoinositol polyphosphates that is functionally significant is further supported by their cellular levels being regulated by specific cell-signaling events. These include the mobilization of certain categories of cellular Ca 2ϩ stores, which inhibits the synthesis of PP-InsP 5 (18). In addition, the metabolism of diphosphoinositol polyphosphates is regulated by receptor-mediated changes in levels of cAMP and cGMP (19). Since DIPP is the enzyme with primary responsibility for controlling the dephosphorylation of PP-InsP 5 and (PP) 2 -InsP 4 (1), this phosphohydrolase is a prime candidate for being regulated by these intracellular signaling processes.
Recently, the prominence of DIPP 2 was further enhanced by * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Given the importance of the two different catalytic activities of DIPP, it is of considerable enzymological significance to understand how this enzyme has the flexibility to hydrolyze both diadenosine polyphosphates and the structurally unrelated diphosphoinositol polyphosphates (Fig. 1). There would appear to be little scope for two separate catalytic centers in DIPP, considering its relatively small size of only 19.5 kDa (1). In fact, a single candidate catalytic core in DIPP has been proposed (1), on the basis of it matching a consensus for the catalytic motif of Nudix hydrolases (nucleoside diphosphate attached to another moiety, "x" (26)). The Nudix designation embraces Ap 6 A, but not PP-InsP 5 and (PP) 2 -InsP 4 . Thus, one of the goals of the current study was to ascertain whether amino acid residues within the Nudt (Nudix-type) catalytic core of DIPP were recruited for the hydrolysis of both classes of substrates. Yet we also queried whether there might be some substrate-dependent interactions with specific amino acid residues, in part because DIPP cleaves the internal polyphosphate chain of Ap 6 A, but the enzyme removes the terminal ␤-phosphates from PP-InsP 5 and (PP) 2 -InsP 4 (Fig. 1). Still further catalytic agility of DIPP must be accounted for because of its alternative positional specificity when either PP-InsP 5 or (PP) 2 -InsP 4 is a substrate (Fig. 1). It is the 5-␤-phosphate that is cleaved from PP-InsP 5 (3,4), yet the 5-␤-phosphate on (PP) 2 -InsP 4 is apparently protected, and it is the other ␤-phosphate that is predominantly hydrolyzed (3).
To identify candidate active-site residues in DIPP, we first compared the amino acid sequence of this enzyme with those of Aps1 and YOR163w, which are two yeast proteins that share DIPP's ability to actively hydrolyze diadenosine polyphosphates and diphosphoinositol polyphosphates (20). The sequences of these three proteins diverge considerably, except for a limited degree of conservation within the Nudt motif and short flanking regions (20). Thus, this homologous region of DIPP was the area of focus for this study; we compared this portion of DIPP with the corresponding sequences of the Nudix hydrolase family members. In this paper, we outline how this alignment procedure identified a subset of proteins with a number of conserved residues, and we go on to describe how we obtained new information concerning their functional significance, in the most detailed mutagenic study of a Nudt protein described to date.
Site-directed Mutagenesis of Recombinant DIPP-Wild-type, recombinant human DIPP was prepared as described previously (1). The recombinant DIPP mutants were prepared by using the Quick-Change TM site-directed mutagenesis kit from Stratagene, according to the manufacturer's instructions. Two complementary mutagenic primers, each containing one base mismatch, were synthesized with a Beckman Oligo 1000M DNA synthesizer (Table I); the preparation of the E70Q mutant was described previously (1). The cDNA construct for recombinant DIPP was used as template for PCR using the mutagenic primers. The wild-type strand in the resulting PCR product was removed by digestion with 10 units of DpnI at 37°C for 1 h before being used to transform Escherichia coli XL1-Blue supercompetent cells. The colonies from the transformation plates were screened by colony PCR with two primers flanking the DIPP gene, and plasmids were prepared from the colonies that generated the expected PCR product. The presence of the correct mutation in all of the constructs was verified by DNA sequencing using a PCR-based dRhodamine fluorescent dye method (Perkin-Elmer).
Expression and Purification of Recombinant DIPP and Its Mutants-The cDNAs for DIPP and each of the mutants were transformed into M15 competent cells, which were incubated overnight at 37°C in 10 ml of LB broth containing 100 g/ml of ampicillin and 25 g/ml of kanamycin. The overnight culture was inoculated into 100 ml of the same medium, and the incubation was continued until A 600 reached 0.9 -1.0, whereupon 1.5 mM isopropyl-␤-thiogalactopyranoside was added to initiate the expression. After an additional 5 h, the cells were harvested by The diphosphate group in mammalian PP-InsP 5 has been determined to be attached to the 5-carbon (4). Since 5-PP-InsP 5 is the precursor of (PP) 2 -InsP 4 , the latter also has a diphosphate group in the 5-position. However, the location of the second diphosphate group on (PP) 2 -InsP 4 has not been determined for any mammalian system; it is tentatively placed at the 6-position in the figure, since 5,6-(PP) 2 -InsP 4 has been identified in Dictyostelids (8,43). The solid arrows indicate the major sites of attack upon these substrates by DIPP. While Ap 6 A is predominantly converted to AMP and adenosine 5Ј-pentaphosphate, lesser quantities of ADP plus adenosine 5Ј-tetraphosphate are also formed (20); the latter is schematically depicted by the unfilled arrow.  centrifugation. The techniques used to lyse the cells and to purify recombinant DIPP and the various mutants using a nickel-nitrilotriacetic acid Superflow column are all as described previously (1). To increase the quantity of protein for analysis by circular dichroism, 10 ml of the overnight culture was inoculated into 250 ml of LB medium in the expression step. In addition, the peak fraction eluted from the nickelnitrilotriacetic acid column was placed in a 3-ml Slide-A-Lyzer cassette (Pierce) and dialyzed against 4 liters of 10 mM phosphate buffer (pH 7.5). The dialyzed protein was then filtered (0.45 m), and its protein concentration was determined according to Beer's law. The extinction coefficient at 280 nm was calculated (27) to be 27,670 M Ϫ1 cm Ϫ1 for recombinant DIPP and its mutants (except F84Y, which was 28,950 M Ϫ1 cm Ϫ1 ). The final concentration of the dialyzed mutants was between 9.1 and 32.2 mM. The purity of each protein preparation was determined on 4 -12% gradient NuPAGE gels ( Fig. 2) with MES SDS running buffer. Structural Analysis by Circular Dichroism Spectroscopy-Far-ultraviolet circular dichroism spectra were acquired on a Jasco J-600 spectropolarimeter equipped with a thermostatically regulated cuvette (set to 15°C) with a 1-mm path length, using a resolution of 0.1 nm, a bandwidth of 1.0 nm, and a response time of 2 s. Samples were diluted to 9.1-11.2 mM of protein in 10 mM sodium phosphate buffer, pH 7.5. Five CD spectra were acquired and averaged before the buffer background was subtracted. Estimates for ␣-helix and ␤-pleated sheet content were obtained using the Selcon fitting program.

Identification of Candidate Functional Elements of DIPP by
Multiple Sequence Alignments-Mammalian DIPP has dual specificity, in that it hydrolyzes diphosphoinositol polyphosphates (1) and diadenosine polyphosphates (20). Two yeast proteins (Aps1 and YOR163w; see Ref. 20 and Fig. 3) have been identified that also hydrolyze these two different groups of substrates. The amino acid sequences of these yeast proteins diverge considerably from that of DIPP, except for a limited degree of conservation within the central core that includes the Nudt catalytic motif and short flanking regions (20). We focused on this homologous internal region of DIPP, and aligned it with a large number of Nudix hydrolases, including 40 that were listed in a recent study (29). A small subgroup of these proteins were identified, in which there are varying degrees of conservation of several interesting features (Fig. 3): (i) highly conserved glutamate residues within the Nudt motif; (ii) Glyrich regions flanking the Nudt motif; (iii) a conserved phenylalanine residue; and (iv) a patch of amino acids with strongly positive electrostatic potential, in which a His residue is the most strongly conserved.
The identification of these conserved elements forms the basis for the mutagenic strategy that we have used in this study.
Glutamate Residues within the Nudt Motif Are Essential for the Hydrolysis of Diphosphoinositol Polyphosphates and Diadenosine Polyphosphates-In our sequence alignments (Fig. 3), five of the seven Glu residues in the Nudt region of DIPP are highly conserved; two of these aligned Glu residues have previously been shown to be catalytically essential in other Nudt contexts (Glu 52 in human MutT homologue type 1 (30) and Glu 57 in MutT (31)). We therefore targeted the equivalent Glu residues in DIPP (Glu 66 and Glu 70 ; see Fig. 3).
The E70Q mutant of recombinant DIPP had a greatly reduced rate of hydrolysis of PP-InsP 5 and (PP) 2 -InsP 4 ( Fig. 4 and Ref. 1). We have extended this observation by demonstrating that this same E70Q mutation also greatly impaired (by 98%) the ability of recombinant DIPP to hydrolyze Ap 6 A (Fig.  4). The E66Q recombinant DIPP mutant also had no significant activity toward either PP-[ 3 H]InsP 5 or (PP) 2 -[ 3 H]InsP 4 , and hydrolase activity against Ap 6 A was found to be only 2% as active as the wild-type enzyme (Fig. 4). The catalytic paradigm for the Nudt motif envisages these two Glu residues participating in both binding of the Mg 2ϩ -substrate complex and in general base catalysis (31,32). Our experiments are consistent with this catalytic process in DIPP being utilized for the hydrolysis of both diphosphoinositol polyphosphates and diadenosine polyphosphates.
We also used CD spectroscopy to examine whether these two mutations had any effect on the secondary structure of recombinant DIPP. The CD spectra of the E66Q and E70Q mutants were not significantly different from that of wild-type recombinant DIPP (data not shown). The CD spectrum for the wildtype enzyme (see below), when analyzed using the Selcon computer algorithm, indicated that the content of ␣-helix and ␤-sheet was 18 Ϯ 0.3 and 29 Ϯ 1.2%, respectively (n ϭ 4).
Substrate Recognition outside the Nudt Domain: Roles of Phe 84 and His 91 -Although the Nudt motif is well recognized to represent the catalytic core of the family of proteins in which it occurs (26), relatively little is known of the contributions to catalysis made by the amino acid residues that lie outside this conserved motif. Previously published sequence alignments of Nudix hydrolases have identified a conserved Phe residue that lies downstream of the eponymous catalytic motif (33); this residue is believed to ensure appropriate positioning of the nucleotide ring (31,32), although the impact of mutagenesis of this residue has not previously been studied. Our own multiple sequence alignments suggested that this conserved residue was equivalent to Phe 84 in DIPP (Fig. 3). Thus, an F84Y mu-

FIG. 2. Analysis by SDS-polyacrylamide gel electrophoresis of wild-type (WT) and mutant recombinant DIPPs.
Recombinant DIPP and the mutants described in A, B, and C were expressed and purified with the nickel-nitrilotriacetic acid column as described under "Experimental Procedures." Approximately 0.7 g of each purified protein was analyzed by SDS-polyacrylamide gel electrophoresis as described under "Experimental Procedures." The gel was stained with Coomassie Blue dye. Molecular weight standards (low range) were from Bio-Rad. The purity of the E70Q mutant is described elsewhere (1).
tant was constructed to further examine this idea. The CD spectra indicated that this mutation did not have a significant effect on the secondary structure of the protein (data not shown). However, the rates of hydrolysis of (PP) 2 -InsP 4 and Ap 6 A were substantially impaired (by 63 and 75%, respectively; Fig. 4). In contrast, the rate of hydrolysis of PP-InsP 5 was only slightly reduced in the F84Y recombinant DIPP mutant (by only 30%; Fig. 4), suggesting only a relatively minor role for Phe 84 in the dephosphorylation of that particular substrate.
A recurring characteristic to emerge from our multiple sequence alignment (Fig. 3) is the presence, in six of these eight proteins, of a patch of between two and five residues with strong positive electrostatic potential, beginning 16 -23 residues downstream of the carboxyl terminus of the Nudt motif. This kernel of positive charge is a good candidate for having electrostatic interactions with the negative charges carried by the phosphate groups on DIPP's substrates. The most conserved of this group of residues (His 91 ) was chosen as a target for further study of this particular issue. An H91L mutant of recombinant DIPP was constructed; CD spectra indicated that this amino acid substitution had no significant effect on the overall secondary structure (data not shown). With PP-InsP 5 as substrate, the H91L mutant expressed nearly normal phosphohydrolase activity (70% of the activity of the wild type recombinant DIPP; see Fig. 4). The hydrolysis of Ap 6 A was more substantially impaired in the H91L mutant (26% of the activity of the wild type enzyme; Fig. 4). However, the most profound effect of this mutation was observed with (PP) 2 -InsP 4 as substrate; the H91L mutant expressed only 4% of the activity of the wild-type enzyme (Fig. 4). The particular sensitivity of (PP) 2 -InsP 4 hydrolysis to the H91L mutation, especially when compared with its relatively small impact upon PP-InsP 5 metabolism, may be relevant to the differential positional specificity of DIPP toward these two substrates (Fig. 1); i.e. the 5-␤-phosphate is hydrolyzed from PP-InsP 5 (3,4), but this same phosphate group on (PP) 2 -InsP 4 is apparently protected from attack, and it is the other ␤-phosphate that is hydrolyzed (3); His 91 appears to have relatively specific interactions with (PP) 2 -InsP 4 .
The Identification and Potential Significance of Conserved Glycine-rich Domains Flanking the Nudt Motif of DIPP-DIPP contains two short Gly-enriched regions (GR1 and GR2; see Fig. 3) that extend out from each end of the Nudt motif in DIPP. Despite their short length, these two regions together contain over half of DIPP's total complement of 13 Gly residues (1). Our sequence alignments (Fig. 3) reveal that three of the four Gly residues in the GR2 region comprise a consensus sequence (GX 2 GX 6 G) that is conserved in both Aps1 and YOR163w, which, like DIPP, also express dual specificity toward diphosphoinositol polyphosphates and diadenosine polyphosphates FIG. 3. Candidate functional sequences in DIPP, identified from multiple sequence alignments with Nudt family members. The Nudt signature (26), in human DIPP (GenBank TM accession number AF062529), plus amino-and carboxyl-flanking sequences, were aligned with the following proteins: two DIPP homologues, namely Aps1 from Schizosaccharomyces pombe (Q09790) and YOR163w from Saccharomyces cerevisiae (Z75071); the predicted sll1058 protein from C. Synechocystis sp. strain PCC6803 (BAA16648); human MutT homologue type 1 (P36639); a human diadenosine tetraphosphate hydrolase, human APAH1 (U30313); the MutT protein from E. coli (P08337); and the E. coli dATPase (P24236). Boldface type is used to highlight residues both in DIPP and in the other proteins that are identical to those in DIPP. The alignment was created using the BioTools scoring matrix in Peptool version 1.1; hyphens denote gaps that were introduced. Residues that are underlined indicate those targeted for mutagenesis in this study (i.e. in DIPP) or those previously shown by mutagenesis to be important for catalytic activity, in human MutT homologue type 1 (30) or MutT (31,32,44). Arranged above this alignment is a schematic of candidate functionally significant domains in DIPP, namely two Gly-rich domains (GR1 and GR2), the Nudt motif, a conserved Phe residue, and a patch of amino acids with positive electrostatic potential.
FIG . 4. The catalytic activities of wild-type (W.T.) and E66Q, E70Q, F84Y, and H91L mutants of recombinant DIPP. Wild-type recombinant DIPP and E66Q, E70Q, F84Y, and H91L mutants of recombinant DIPP were assayed for phosphohydrolase activities against PP-InsP 5 (gray bars; n ϭ 3-5 for mutants, n ϭ 14 for wild type) or (PP) 2 -InsP 4 (black bars; n ϭ 3-6 for mutants, n ϭ 12 for wild type) or for hydrolase activity toward Ap 6 A (striped bars; n ϭ 3-7 for mutants, n ϭ 15 for wild type). S.E. values are also indicated. k Ϫ1 is the firstorder rate constant in the following rate equation: [S] ϭ [S] 0 e Ϫkt . The units for k Ϫ1 are min Ϫ1 g Ϫ1 . Data for PP-InsP 5 and (PP) 2 -InsP 4 metabolism by the E70Q mutant are similar to those published previously (1). (20). As for GR1, this tripeptide is unique to DIPP in our alignments, but Gly itself is represented between two and four times in the first nine residues of all the sequences in Fig. 3.
One clue as to the significance of these Gly clusters comes from predictive analyses of the general architecture of the Nudt motif, which suggests an ␣-helix flanked by loops (34,35). This arrangement was empirically confirmed by the determination of the solution structure of the prototypical Nudix hydrolase, the E. coli MutT protein that attacks 8-oxo-dGTP (32). The importance of Gly residues in GR1 and GR2 may be to construct the tight bends in the polypeptide backbone upon which the loop structure depends (36). However, the significance of Gly can extend beyond this structural role; there are circumstances in which these loops form compact ligand-binding pockets in which Gly itself directly participates. Two types of Glyligand interactions are suggested by the Nudix context in which GR1 and GR2 are found. First, one of the binding sites for the catalytically essential divalent cation in the E. coli MutT protein is the carbonyl carbon of Gly 38 (31,32). This residue is conserved in other Nudix hydrolases and corresponds to Gly 51 in the GR1 region of DIPP (Fig. 3). Second, since Ap 6 A is a nucleotide, it is significant that Gly-rich sequences are also a recurring feature of nucleotide-binding loops (36,37). 3 The absence of a side chain in the Gly residue permits a phosphate group to approach the backbone amides, to which the ligand is gripped by hydrogen bonds (38,39). This represents a potential role for GR2.
GR1: Mutagenesis of Gly 50 , Gly 51 , and Gly 52 -Our proposition that Gly 51 may be functionally significant in GR1 (see above) was pursued by conservatively mutating this residue to Ala. This mutant of recombinant DIPP had the same CD spectrum as the wild-type enzyme (data not shown), but its catalytic activity was severely compromised (Ͼ97% less than that of the wild-type enzyme; see Fig. 5). Assuming we are correct with our suggestion that the cation-binding function for the carbonyl carbon of Gly 38 in the E. coli MutT protein (31,32) is conserved for Gly 51 in DIPP (see above), then we can rationalize the effect of the G51A mutation; the steric bulk of the Ala side chain may prevent the cation from approaching the backbone carbonyl group of Gly 51 .
We also examined the importance to catalysis of the two Gly residues that flank Gly 51 . The rates of hydrolysis of diphosphoinositol polyphosphates by G50A and G52A mutants of recombinant DIPP was substantially (Ͼ85%) less than that of the wild type enzyme (Fig. 5). The hydrolysis of Ap 6 A was also decreased compared with wild type recombinant DIPP, although slightly less dramatically (by 60 -75%; see Fig. 5). The rate of Ap 6 A hydrolysis was further reduced (by 90 -99%; see Fig. 5) in G50V and G52V mutants of recombinant DIPP. The fact that Gly-to-Val mutants are impaired more dramatically than Gly-to-Ala substitutions is again consistent with the idea that the introduction of steric bulk in the GR1 region prevents proper ligand-protein interactions. Circular dichroism spectra indicated that the secondary structure of these additional mutants were not significantly different from that of the wild type recombinant DIPP (data not shown).
GR2: Mutagenesis of Gly 72 , Gly 75 , Gly 78 , and Gly 82 -Four separate Gly-to-Ala mutants of recombinant DIPP were constructed in the GR2 region: G72A, G75A, G78A, and G82A. The catalytic activity of the G78A mutant (Fig. 6) and its CD spectrum (data not shown) were not significantly different from that of the wild-type enzyme. This is an interesting observation, because Gly 78 is the only one of these four Gly residues that is not conserved in Aps1 and YOR163w. Note, however, that the additional steric bulk introduced in a G78V mutant was not tolerated, and then there was a substantial decrease in catalytic activity toward all three substrates (Fig. 6).
The CD spectrum obtained from the G82A mutant showed a substantial change from that of the wild-type enzyme (Fig. 7). The nadir in the spectrum at 203 nm broadened considerably and increased in value from Ϫ9000 to Ϫ3000 degrees⅐cm 2 ⅐dmol Ϫ1 (Fig. 7). The Selcon computer algorithm estimated that, compared with the wild-type recombinant DIPP, the content of ␣-helix in the G82A mutant was 36% lower, while the ␤-sheet content was 30% higher. Considering the very conservative nature of the Gly-to-Ala substitution, the consequences for the overall secondary structure of the protein are very dramatic. Further structural analysis will be required in order to account for this effect. However, in view of the fact that Gly plays an important role in the architecture of certain polypeptide loops (36), such as those that generally flank the Nudt motif (31,32,34,35), it is notable that there are examples of proteins where no other amino acid can replace Gly at particularly tight turns in a loop (40); this may be why the introduction of an G82A mutation promotes some unraveling of the normal secondary structure. Presumably as a consequence, the catalytic activity of this mutant recombinant DIPP is sub- stantially impaired (Fig. 6).
The CD spectra of the remaining two mutants, G72A and G75A, were both very similar to that of the wild-type enzyme (data not shown), indicating that the mutations had no signif-icant impact upon the secondary structure. However, the catalytic activities of the G72A and G75A mutants toward PP-InsP 5 , (PP) 2 -InsP 4 , and Ap 6 A were impaired by Ͼ95% (Fig. 6). Our mutagenesis of the GR2 region of DIPP has therefore identified a novel Gly-rich consensus, GX 2 GX 6 G, that is essential for this enzyme's dual specificity.
General Conclusions-The importance of this study primarily lies in its providing new information relevant to our understanding of the structural and functional relationship that DIPP has with Nudix hydrolases. This knowledge is particularly significant, because the metabolism of diphosphoinositol polyphosphates is an unorthodox activity for a protein that has a Nudt catalytic motif. Our study also breaks new ground by identifying some catalytically important residues that lie outside the Nudt consensus. It is these extramural residues that are believed to impart substrate specificity upon each of the individual members of this hydrolase family (26). We have found several residues in this category that make important contributions to the unusual dual catalytic specificity of DIPP.
A particularly striking illustration of these developments is our identification of a new functionally and structurally important array of Gly residues (the "GR2" motif, GX 2 GX 6 G) that projects out from the carboxyl terminus of the Nudt domain. The GR2 motif is conserved in Aps1 and YOR163w (Fig. 3), which are two yeast homologues of DIPP that also express dual specificity toward diadenosine polyphosphates and diphosphoinositol polyphosphates (20). Moreover, this GR2 motif is absent from two Nudix hydrolases that actively degrade diadenosine polyphosphates but do not attack diphosphoinositol polyphosphates, namely human APAH1 (20,35) and the IalA invasion protein of Bartonella bacilliformis (20,41,42). These comparisons suggest that the emergence of the GR2 motif has been an essential factor for the evolution of dual specificity toward diadenosine polyphosphates and diphosphoinositol phosphates. In this case, the GR2 motif may be a signature sequence that helps predict which other proteins with a Nudt motif may have a similar catalytic activity. We have already identified one candidate: the putative sll1058 protein from Cyanobacterium synechocystis (Fig. 3).
Among other catalytically essential features of DIPP that we identified in this study, His 91 was found to be of special significance; the mutation of this residue to Leu inhibited (PP) 2 -InsP 4 dephosphorylation by 96%, yet PP-InsP 5 metabolism was only reduced by 30% (Fig. 4). One other residue, Phe 84 , was also found to contribute more to the hydrolysis of (PP) 2 -InsP 4 compared with PP-InsP 5 (Fig. 4). These findings provide the first evidence for there being a substantial difference in active-site residues that participate in the hydrolysis of these two substrates. As such, this is an important step forward toward the goal of understanding how DIPP has the flexibility to express differential positional specificity toward these two inositolbased substrates (see Fig. 1).
FIG. 6. The effect of mutations within the GR2 motif on the catalytic activity of recombinant DIPP. Wild-type (W.T.) recombinant DIPP and the G72A, G75A, G78A, and G82A mutants of recombinant DIPP were assayed for phosphohydrolase activities against either PP-InsP 5 (gray bars; n ϭ 3-5 for mutants) or (PP) 2 -InsP 4 (black bars; n ϭ 3-6 for mutants) or for hydrolase activities toward Ap 6 A (striped bars; n ϭ 3-4 for mutants). S.E. values are also indicated. The units for k Ϫ1 are min Ϫ1 g Ϫ1 . Data obtained from wild-type recombinant DIPP are taken from Fig. 4 for comparison.
FIG. 7. Far-ultraviolet circular dichroism spectra of wild-type recombinant DIPP, and its G82A mutant. CD spectra were obtained as described under "Experimental Procedures" for the wild-type (closed circles) and the G82A mutant form of recombinant DIPP (open circles).