Escherichia coli orf17 codes for a nucleoside triphosphate pyrophosphohydrolase member of the MutT family of proteins. Cloning, purification, and characterization of the enzyme.

The product of the Escherichia coli orf17 gene is a novel nucleoside triphosphate pyrophosphohydrolase with a preference for dATP over the other canonical (deoxy)nucleoside triphosphates, and it catalyzes the hydrolysis of dATP through a nucleophilic attack at the β-phosphorus to produce dAMP and inorganic pyrophosphate. It has a pH optimum between 8.5 and 9.0, a divalent metal ion requirement with optimal activity at 5 mM Mg2+, a Km of 0.8 mM and a kcat of 5.2 s−1 at 37°C for dATP. dAMP is a weak competitive inhibitor with a Ki of approximately 4 mM, while PPi is a much stronger inhibitor with an apparent Ki of approximately 20 μM. The enzyme contains the highly conserved signature sequence GXVEX2ETX6REVXEEX2I designating the MutT family of proteins. However, unlike the other nucleoside triphosphate pyrophosphohydrolases with this conserved sequence, the Orf17 protein does not complement the mutT− mutator phenotype, and thus must serve a different biological role in the cell.

Members of the MutT family of proteins are categorized by a conserved amino acid motif originally identified as an important functional region in the Escherichia coli MutT and the Streptococcus pneumoniae MutX antimutator proteins (1). This conserved amino acid signature sequence was found, by computer search, to be present in a number of open reading frames broadly distributed throughout nature, from viruses to humans (1,2), and one of these, orf17 (GenBank D10165, 1992), is the subject of this paper. The orf17 gene is located at 41 min on the E. coli chromosome (3) just upstream of the ruvC gene (4,5), which codes for a Holliday junction-specific endonuclease (4).
Both the MutT and MutX proteins are nucleoside triphosphatases (6,7), as are the corresponding proteins from Proteus vulgaris (8) and from humans (9), and all four are implicated in preventing mutations. On the other hand, two other proteins containing the MutT signature sequence are not nucleoside triphosphatases. One of them, a nucleoside pyrophosphatase, prefers NADH as its substrate (10), while the other, GDPmannose mannosyl hydrolase, prefers GDP-mannose (11). Nei-ther of these latter two enzymes has been linked to a mutagenic pathway. A feature common to all six enzymes containing the signature sequence is the hydrolysis of a substrate containing a nucleoside diphosphate linkage.
In this paper, we describe the cloning and expression of the orf17 gene, and the purification and characterization of the Orf17 protein. Like the first four enzymes described above, it is a nucleoside triphosphatase. However, unlike those four enzymes, which prefer dGTP as a substrate over the other canonical (deoxy)nucleoside triphosphates, Orf17 has a unique preference for dATP, and it does not appear to be involved in antimutagenesis.

Materials
Enzymes-Thermus aquaticus DNA polymerase was from Perkin-Elmer, and other enzymes used in standard cloning procedures were from Life Technologies, Inc., Stratagene, or U. S. Biochemical Corp. Yeast inorganic pyrophosphatase was from Sigma.
Nucleic Acids-Oligonucleotide primers were from Integrated DNA Technologies, plasmid vectors pET11b and pTRC99A were from Novagen and Pharmacia, respectively, and plasmid pHS234 containing the orf17 gene was a gift from H. Shinagawa.
Bacterial Strains-E. coli HB101 and SB3 were from our laboratory stock, E. coli HMS174(DE3) was from Novagen, and E. coli strains AB1155 and HRS1024 were gifts from H. Shinagawa.
Chemicals-Nucleotides, ampicillin, and streptomycin sulfate were from Sigma. Nalidixic acid was from Winthrop, and IPTG 1

Methods
Cloning-The orf17 gene was subcloned from plasmid pHS234 (4) provided by H. Shinagawa. It was amplified in the polymerase chain reaction using oligonucleotide primers, d(CCGCATATGAAGGATA-AAGTGTATAAGCGT) and d(CGCGGATCCAGATAGCCCTGCCTGT-TCAGG), designed to incorporate an NdeI site at the start of the gene, a BamHI site at its end, and an ATG start codon. The amplified gene was digested with NdeI and BamHI, purified by gel electrophoresis, and ligated into the NdeI and BamHI sites of pET11b to place the orf17 gene under control of a T7 lac promoter for overexpression. The resultant plasmid, pETorf17, was used to transform E. coli strains HB101 for storage and HMS174(DE3) for expression.
Purification of the Orf17 Protein: Growth and Expression-Four liters of LB medium containing 100 g/ml ampicillin were inoculated with HMS174(DE3):pETorf17, grown at 37°C to an A 600 of 1.0, induced with 1 mM IPTG, and grown for an additional 2.5 h.
Crude Extract-The cells were harvested, washed by suspension in 2 volumes of a saline solution (0.5% KCl, 0.5% NaCl), recentrifuged, and frozen at Ϫ80°C. Freezing the cells was essential for preparing the extract. The cells were extracted in 2 volumes of buffer A (50 mM Tris, pH 7.5, 1 mM EDTA, 0.1 mM DTT), centrifuged, and re-extracted, and the supernatants were combined. A typical extract (Fraction I) yielded approximately 100 mg of protein from 10 g of cells.
Ammonium Sulfate Fractionation-Fraction I was brought to 41% saturation by the addition of 26 ml of saturated ammonium sulfate to 37.5 ml of crude extract. The precipitate was removed by centrifugation and discarded, and an additional 22.5 ml of ammonium sulfate were added to the supernatant to give a 57% saturated solution. This precipitate, containing the Orf17 protein, was collected and dissolved in 1 ml of buffer A (Fraction II).
Sephadex G-50 -Fraction II was loaded onto a 1.5 ϫ 55-cm Sephadex G-50 gel filtration column and eluted with buffer A. The fractions containing the Orf17 protein were combined (19.5 ml), concentrated by precipitation in 60% saturated ammonium sulfate, and dissolved in 1 ml of buffer A, yielding 18 mg of Orf17 protein, which were stored at Ϫ80°C (Fraction III).
Assay for Mutator Phenotype-The plasmid, pETorf17, was digested with XbaI and HindIII, and the fragment containing the orf17 gene was ligated into pTRC99A to produce pTRCorf17. The plasmid, pTRCmutT, was constructed in a similar way from pETmutT (12) substituting BamHI for XbaI. E. coli strain SB3, lacking a functional MutT protein, was transformed with pTRC99A, pTRCmutT, or pTRCorf17. These strains were grown in LB medium containing 100 g/ml ampicillin and 1 mM IPTG, and mutation frequencies were determined by plating aliquots on LB plates containing either 150 g/ml streptomycin or 20 g/ml nalidixic acid. Similarly, E. coli strains, AB1157 (orf17 ϩ ) and HRS1024 (orf17 Ϫ ), both gifts from H. Shinagawa (4), were grown in LB medium and mutation frequencies were determined on nalidixic acid plates as above. Mutation frequencies are reported as the average of five separate determinations.
Enzyme Assays-Two standard enzyme assays, both based on the production of PP i from dATP, were used.
Colorimetric Assay-The standard reaction mix contained in 50 l: 4 mM dATP, 50 mM Tris, pH 9, 10 mM MgCl 2 , 1 mM DTT, 0.1 mg/ml bovine serum albumin, 1% glycerol, 0.5 unit of inorganic pyrophosphatase, and 0.3-3 milliunits of Orf17 enzyme. The reaction was run at 37°C for 15 min, terminated by the addition of 50 l of a mixture of four parts of Norit® (20% packed volume) and one part of 7% HClO 4 , centrifuged, and an aliquot of the supernatant was analyzed for inorganic orthophosphate by the method of Ames and Dubin (14). A unit of activity is 1 mol of PP i /min under these conditions. Radioactive Assay-This is similar to the standard colorimetric assay with the following modifications. [␥-32 P]ATP (about 500 cpm/nmol) was used as the substrate. The reaction was terminated as in the colorimetric assay (except that 5 N HCl was substituted for the HClO 4 ) and an aliquot of the supernatant was analyzed for radioactivity. Since ATP is hydrolyzed at half the rate of dATP (see Table I), a unit of enzyme in this assay is comparable to 2 units in the colorimetric assay.
Acquisition and Processing of NMR Spectra-Experiments were done at 32°C on a Varian Unityplus 600 NMR spectrometer equipped with a 5-mm broad band probe with field/frequency locking on the D 2 O resonance. The 31 P NMR spectra were recorded with a transmitter frequency of 243 MHz, a spectral width of 12,000 Hz, and proton decoupling. Each spectrum was recorded with an acquisition time of 2.75 s and a relaxation delay of 6 s. The enzymatic reaction was monitored continuously in blocks of 64 -256 transients for 4 h. The spectrum in this paper was collected with 256 scans between 2.5 and 3 h after the addition of enzyme. Resonances were assigned by comparing reaction spectra to those of standards of dAMP, dADP, dATP, sodium pyrophosphate, and sodium phosphate, and chemical shifts were measured relative to an external standard of 85% phosphoric acid. The data were subjected to Fourier transformation after processing with line broadening of Ϫ1 Hz and a shifted sine bell function for resolution enhancement.

Subcloning, Expression, and Purification
The orf17 gene was subcloned as described under "Methods." Its sequence agrees with Takahagi et al. (4) except for the original GTG start site, which we modified to ATG as a matter of convenience. Expression of the gene from pETorf17 results in the appearance of a major band on SDS gels corresponding to a 17-kDa polypeptide (Fig. 1), which, when treated according to the outlined purification procedure, results in an essentially pure protein (Ͼ95% as visualized on the gel). Attention is called to a faint band in the high molecular weight region of the gel. This most likely represents an aggregated form of the protein, not a contaminant. It is consistently seen in varying amounts in different preparations of the enzyme. It is not present in extracts of cells transformed with the vector lacking the insert (lane 2) and a sequence analysis of the N-terminal 10 amino acids of the final preparation indicates that the enzyme is at least 98% homogeneous with respect to these residues (data not shown). It is noteworthy that the bulk of the Orf17 protein is released from the cells merely by freezing them and extracting with a dilute buffer. This results in a crude extract much more highly enriched in Orf17 than procedures involving more complete disruption of the cells, such as sonication. One other member of the MutT family of proteins, the GDP-mannose hydrolase (11), is also readily extractable from previously frozen cells. Perhaps these two expressed enzymes share a common cellular compartment such as the periplasmic space, which may be breeched in the freeze-thaw cycle, or they may share some other feature not readily apparent. At any rate, this mild extraction procedure facilitates the subsequent purification, because the bulk of cellular proteins is left behind in the cell pellet.

Properties of the Enzyme
Specificity-As shown in Table I, the enzyme hydrolyzes all eight of the canonical (deoxy)nucleoside triphosphates with a marked preference for dATP. The next most favored substrate is approximately 6-fold lower in catalytic efficiency (k cat /K m ). The enzyme recognizes not only the nucleotide base, but the sugar, as well, since in each case, the ribo nucleotide is less active than its corresponding deoxy equivalent. Especially noteworthy is the low activity of dGTP relative to dATP, amounting to an order of magnitude difference in catalytic efficiency. This is in marked contrast to the MutT nucleoside triphosphatase, which prefers dGTP, hydrolyzing it at a 30-fold higher rate (7) and with a catalytic efficiency 2 orders of magnitude higher than with dATP (13).
pH Optimum-The enzyme is maximally active at pH 8.6, diminishing to 50% activity at pH 7.6 or 9.9 (data not shown). This distinctly alkaline optimum is similar to the pH versus rate profiles of three other members of the E. coli MutT family i.e. MutT (7), NADH pyrophosphatase (10), and GDP-mannose hydrolase (11).
Divalent Cation Requirement-As with all other members of the MutT family, a divalent cation is required for activity, and Mg 2ϩ is most effective. The optimal concentration is 5 mM for free Mg 2ϩ , and substitution of Mn 2ϩ , Ca 2ϩ , or Zn 2ϩ at 1 mM or 5 mM reduces the rate 10-fold or more. Mg 2ϩ did not activate the enzyme until its concentration exceeded that of the substrate, suggesting that two metal ions are involved in the catalysis, as has been demonstrated for the MutT enzyme (15). The salts, NaCl or KCl, reduced the enzymatic activity by 50% at concentrations of 200 mM.
Products and Mechanism of the Reaction-The nature of the products formed during the enzymatic hydrolysis of dATP were determined in standard colorimetric assay mixtures scaled up 20-fold. Two reactions were run in parallel, except for the omission of yeast inorganic pyrophosphatase in one of them. Aliquots were removed over a 30-min period and analyzed for nucleotides by HPLC and inorganic orthophosphate as described under "Methods." In the absence of inorganic pyrophosphatase, no inorganic orthophosphate was detected throughout the incubation. Fig. 2 reports the results of the analysis of the incubation mixture containing yeast inorganic pyrophosphatase. The disappearance of dATP is reflected in the appearance of dAMP and PP i (calculated as 2 P i ) in approximately equivalent amounts. No dADP was detected during the course of the incubation and no P i was detected in the absence of inorganic pyrophosphatase. Thus, the reaction describing the hydrolysis may be written as follows.
The nature of the products are the same as those reported for two other related enzymes, the MutT and MutX nucleoside triphosphatases.
It had been demonstrated previously that the mechanism of hydrolysis of dGTP by MutT involves a nucleophilic attack on the ␤-phosphorus of the triphosphate, rather than on the more generally favored ␣-phosphorus, which would yield the same products (16). It was therefore of interest to determine whether the Orf17 enzyme, having a different specificity but producing similar products, also utilizes the same mechanism of attack. Accordingly, the hydrolysis of dATP by Orf17 was conducted in 25% H 2 18 O and the products were analyzed by 31 P NMR spectroscopy. The proton decoupled 243-MHz 31 P NMR spectrum of the reaction mixture is shown in Fig. 3. The course of the reaction, as measured by NMR spectroscopy, independently confirms the direct conversion of dATP to dAMP and PP i . Peaks corresponding to the substrate dATP and the products dAMP and PP i are observable, whereas no peaks corresponding to P ␣ (Ϫ12.4 ppm) or P ␤ (Ϫ8.3 ppm) of dADP are observable and the peak corresponding to P i (0 ppm) is negligible. The spectrum for the reaction consists of a doublet centered at Ϫ13.0 ppm corresponding to the ␣-phosphorus of dATP, a triplet centered at Ϫ21.4 ppm corresponding to the ␤-phosphorus of dATP, and a doublet centered at Ϫ8.0 ppm corresponding to the ␥-phosphorus of dATP, as well as resonances corresponding to dAMP (1.3 ppm) and inorganic pyrophosphate (Ϫ7.8 ppm). The resonance at 1.3 ppm corresponding to the phosphorus of dAMP shows a single peak, indicating no 18 2. Products of the reaction. A standard colorimetric assay (see "Methods") was scaled up 20-fold and was monitored by HPLC to quantify dATP (ࡗ) and dAMP (q). No dADP was detected during the course of the reaction. PP i (E) was converted to P i by the inorganic pyrophosphatase included in the reaction mixture and was calculated as one-half of the P i measured in the Fiske-Subbarow assay (31). No P i was formed in a parallel reaction in the absence of inorganic pyrophosphatase. It was passed through a Chelex 100 column prior to the addition of MgCl 2 , EDTA, and enzyme. Orf17 dATPase was dialyzed against 50 mM Tris, pH 7.5, 1 mM EDTA, 0.1 mM DTT, and 5 g/liter Chelex 100. The spectrum was acquired and processed as described under "Methods." The peaks for dAMP and PP i are expanded and shown to the right of each respective peak in the spectrum. The resonance corresponding to pyrophosphate consists of two peaks, one downfield peak for pyrophosphate containing only 16 O and one upfield for pyrophosphate labeled by 18 O. The resonance for dAMP appears as a single peak, indicating no 18 O labeling. unlabeled phosphorus (17,18). In pyrophosphate, the two phosphorus atoms are strongly coupled such that the resonances for unlabeled and 18 O-labeled phosphorus collapse into an apparent singlet 0.012 ppm upfield of the resonance for unlabeled pyrophosphate (19). The intensity ratio of the resonances for unlabeled PP i and [ 18 O]PP i is 3:1 consistent with 25% H 2 18 O in the reaction mixture (16,19). These results show that hydrolysis of the P ␣ -P ␤ linkage of dATP occurs by attack on the ␤-phosphorus rather than by the more frequently observed attack on the ␣-phosphorus (20). The observations were substantiated by repeating the reaction in 65% H 2 18 O. Again, one peak was detected for dAMP and two peaks were detected for inorganic pyrophosphate with an intensity ratio of approximately 1:2 in respect to unlabeled PP i and [ 18 O]PP i (data not shown). Thus, Orf17 dATPase is similar to the MutT dGTPase in catalyzing a nucleophilic attack by H 2 O (or initially by the enzyme, followed by H 2 O) on the ␤-phosphorus of dATP or dGTP, respectively.
Kinetics-Standard procedures were used to determine the general kinetic parameters of the dATPase. Initial estimates of the K m and V max were obtained from double-reciprocal plots of substrate versus initial velocity (21) and then refined using a nonlinear least squares analysis weighted to substrate concentration (22) (data not shown). The values are reported in Table II.
An interesting aspect of the reaction emerged early in the project, when we noticed that the rate of dATP hydrolysis decreased markedly during the course of the incubation. Further investigation implicated product inhibition as the cause, since inclusion of inorganic pyrophosphatase in the reaction mixture restored initial rates. This is shown in Fig. 4A, where the time course of dAMP formation is compared in the presence and absence of inorganic pyrophosphatase. That PP i and not dAMP is the product responsible for the inhibition is pointed up by the relative K i values determined from the plots in Fig. 4 (B  and C). Note that the concentrations of dAMP and PP i differ by 100-fold in the two graphs, and the K i values calculated from these data are approximately 4 mM for dAMP and 20 M for PP i . MutT nucleoside triphosphatase, which shares many of the properties of the Orf17 enzyme, has a 50-fold higher K i for PP i and relatively little product inhibition. 2 We have searched for possible covalent enzyme-substrate intermediates involving dAM 32 P or 32 PP i without success, and neither of these labeled products exchanges with dATP within the limits of our measurements during the course of the reaction. Failure to detect a covalent deoxyadenylate-enzyme intermediate by exchange of 32 PP i into dATP is consistent with a nucleophilic attack at the ␤-phosphorus rather than the ␣-phosphorus, as found by 18  In this regard, it is interesting to speculate about the Orf17 dATPase activity within the E. coli cell. Josse (23,24), in his elegant studies on E. coli inorganic pyrophosphatase, has shown that the enzyme is constitutive and accounts for approximately 0.2% of the cellular protein. It hydrolyzes PP i at the rate of about 1.5 mmol min Ϫ1 mg Ϫ1 , and it has a K m of 5 M. Thus one might expect that the intracellular PP i concentration would be kept well below the 20 M K i for the Orf17 dATPase, and that this enzyme would be active in the cytoplasm. However, Kukko and Heinonen (25) and Kukko-Kalske et al. (26) have shown that the intracellular concentration of PP i in E. coli

TABLE II
Kinetic parameters for Orf17 dATPase Standard assay conditions were used (see "Methods"). K m and V max were determined from a Lineweaver-Burk plot (21) using a nonlinear least squares fit weighted to substrate concentration (22). k cat was calculated from V max assuming one active site per monomer. K i values were calculated from the data in Fig. 4. 4. Product inhibition of the Orf17 dATPase. A, time course of a reaction under standard conditions for the radioactive assay (see "Methods") in the presence and absence of inorganic pyrophosphatase. B, inhibition by dAMP was measured in the colorimetric assay using 0.75 milliunit of the dATPase and 0.8 mM or 1.6 mM dATP. C, inhibition by PP i was measured in the radioactive assay modified to contain 10 or 20 M [␥-32 P]ATP (1.5 ϫ 10 4 cpm/nmol), 2.5 milliunits of dATPase, and no inorganic pyrophosphatase. The K i values calculated from these data were approximately 4 mM and 20 M for dAMP and PP i , respectively (see Table II). is 0.5 mM and is independent of very large fluctuations in the amount of inorganic pyrophosphatase in the cells. At 0.5 mM PP i , the dATPase would be expected to be almost completely inhibited. Perhaps the location of the dATPase is an important consideration. As pointed out in the purification scheme (see "Methods"), the enzyme is readily released from cells merely by a freeze-thaw cycle, suggesting that it may reside in the periplasmic space. If this is its normal residence, the consideration of cytoplasmic PP i concentrations may not be germane to its in situ activity. Further investigations will be necessary to elucidate the mechanism of inhibition by PP i , and perhaps its significance will emerge when the physiological role of the dATPase is understood.
What Is the Role of Orf17 dATPase?-Although the orf17 gene is located close to ruvC, it probably is not involved in ruvC function (4). Our original interest in the Orf17 protein derived from a small region of amino acid homology common to it and MutT, as shown in Fig. 5A. This same region is highly homologous to MutX of S. pneumoniae (1), to a MutT homologue of P. vulgaris (8), and to a human 8-oxo-dGTPase (9). All of these latter four proteins are involved in preventing mutations of the AT 3 CG variety, and all of the genes coding for these proteins complement mutT Ϫ strains. In addition, they all hydrolyze nucleoside triphosphates to PP i and nucleoside monophosphates. For these reasons, it was of interest to determine whether or not Orf17, which catalyzes similar reactions, is also involved in preventing mutations. Accordingly, a plasmid containing orf17 was used to transform a mutT Ϫ strain of E. coli. Although the transformed strain produced large quantities of the enzyme (data not shown), the results in Table III clearly demonstrate that orf17 does not complement mutT. Thus, it is most likely not the putative suppressor of mutT referred to by Ray (27) or Desiraju (28). Furthermore, a strain of E. coli, lacking the orf17 gene, has no significant enhancement in mutation frequency when compared to its wild type parent (Table III). Thus Orf17 does not share the antimutator properties of MutT. A broader spectrum of tests will have to be performed in order to see if orf17 is involved in other mutational pathways.
On the other hand, it is quite possible that Orf17 dATPase does not play a role in antimutagenesis. We have included in Fig. 5B three additional proteins with the highly homologous MutT signature sequence, yet having no nucleoside triphosphatase activity and no apparent role in preventing mutations. Orf257 and Orf1.9 are newly discovered enzymes, which hydrolyze NADH (10) and GDP-mannose (11), respectively, and the FIG. 5 A, comparison of the MutT and Orf17 amino acid sequences. The amino acid sequences of the MutT and Orf17 proteins were aligned to maximize homology using the program ALIGN (32) from the FASTA program package (33,34). Gaps in the amino acid sequences inserted to optimize alignment are indicated by dashes. Vertical lines indicate identical amino acids, and dots indicate conservative amino acid changes. Conservative changes are defined as substitutions within the following groups: AGPST, FYW, ILVM, HKR, and EQDN (1,35). The conserved signature sequence of the MutT family is highlighted in boldface. B, enzymes of the MutT family of proteins. The first five enzymes are nucleoside triphosphatases. Orf257 is an NADH pyrophosphatase, Orf1.9 is a GDP-mannose hydrolase, and A(p) 4 Aase is a P 1 -P 4 -di(adenosine-5Ј)-tetraphosphatase. Numerical superscripts refer to the position of each polypeptide within the complete protein.
is a strain lacking a functional MutT protein, AB1157 (orf17 ϩ ) is a wild type parent strain of HRS1024 (orf17 Ϫ ), and HRS1024 (orf17 Ϫ ) is a strain lacking a portion of the orf17 gene.
b Numbers represent the average (Ϯ standard deviation) mutation frequency of five separate cultures. nal r and strep r are colonies resistant to nalidixic acid or streptomycin, respectively. c -, not determined because both AB1157 and HRS1024 already contain a streptomycin resistance factor. gene for another related enzyme hydrolyzing diadenosine tetraphosphate has recently been cloned from human tissue (29). We are presently characterizing two new enzymes bearing the MutT signature sequence, which also are not nucleoside triphosphatases. 3 The feature common to all these enzymes is their recognition and hydrolysis of metabolites containing a nucleoside diphosphate linkage. It seems likely that the MutT signature sequence represents a nucleoside diphosphate binding and catalytic site conserved during evolution for the implementation of diverse metabolic reactions involving nucleoside diphosphates. The solution structure of the MutT protein reveals this signature sequence to be located in the loop 1-helix 1 motif of an ␣ ϩ ␤ sandwich (30). Recently, crystals of Orf17 dATPase have been obtained suitable for x-ray analysis, 4 and it will be of interest to see how the structural features of the signature sequence common to both enzymes compare in these otherwise very diverse proteins. At present, no specific physiological function can be ascribed to the Orf 17 dATPase. However, the availability of isogenic strains of E. coli with and without the enzyme, along with a clone overexpressing it, provides an excellent system for comparative experiments designed to elucidate the metabolic role of the enzyme.
Finally, the results presented here introduce a note of caution in ascribing MutT antimutator function to newly discovered proteins containing the MutT signature sequence. The Orf17 dATPase described here, as well as the NADH pyrophosphatase (10), the GDP-mannose hydrolase (11), the diadenosine tetraphosphatase (29), and two recently discovered enzymes 3 all contain the MutT sequence, but differ markedly in their substrate preferences and detailed reactions. It will be of interest to discover how broadly this amino acid motif is employed in nature to participate in diverse metabolic pathways.