A Phosphate-binding Histidine of Binuclear Metallophosphodiesterase Enzymes Is a Determinant of 2′,3′-Cyclic Nucleotide Phosphodiesterase Activity*

Binuclear metallophosphoesterases are an enzyme superfamily defined by a shared fold and a conserved active site. Although many family members have been characterized biochemically or structurally, the physiological substrates are rarely known, and the features that determine monoesterase versus diesterase activity are obscure. In the case of the dual phosphomonoesterase/diesterase enzyme CthPnkp, a phosphate-binding histidine was implicated as a determinant of 2′,3′-cyclic nucleotide phosphodiesterase activity. Here we tested this model by comparing the catalytic repertoires of Mycobacterium tuberculosis Rv0805, which has this histidine in its active site (His98), and Escherichia coli YfcE, which has a cysteine at the equivalent position (Cys74). We find that Rv0805 has a previously unappreciated 2′,3′-cyclic nucleotide phosphodiesterase function. Indeed, Rv0805 was 150-fold more active in hydrolyzing 2′,3′-cAMP than 3′,5′-cAMP. Changing His98 to alanine or asparagine suppressed the 2′,3′-cAMP phosphodiesterase activity of Rv0805 without adversely affecting hydrolysis of bis-p-nitrophenyl phosphate. Further evidence for a defining role of the histidine derives from our ability to convert the inactive YfcE protein to a vigorous and specific 2′,3′-cNMP phosphodiesterase by introducing histidine in lieu of Cys74. YfcE-C74H cleaved the P-O2′ bond of 2′,3′-cAMP to yield 3′-AMP as the sole product. Rv0805, on the other hand, hydrolyzed either P-O2′ or P-O3′ to yield a mixture of 3′-AMP and 2′-AMP products, with a bias toward 3′-AMP. These reaction outcomes contrast with that of CthPnkp, which cleaves the P-O3′ bond of 2′,3′-cAMP to generate 2′-AMP exclusively. It appears that enzymic features other than the phosphate-binding histidine can influence the orientation of the cyclic nucleotide and thereby dictate the choice of the leaving group.

-Pase uses Mn 2ϩ to catalyze phosphoester hydrolysis with a variety of substrates, including phosphopeptides, phosphoproteins, nucleoside 2Ј,3Ј-cyclic phosphates, and "generic" organic phosphomonoesters and diesters such as p-nitrophenyl phosphate and bis-p-nitrophenyl phosphate. Although the physiological substrate(s) and biological function of -Pase remain obscure, other well studied members of the binuclear metallophosphoesterase superfamily play key physiological roles in cellular pathways of signal transduction (e.g. the phosphoprotein phosphatase calcineurin), DNA repair (e.g. the DNA nuclease Mre11), or RNA processing (e.g. the RNA debranching enzyme Dbr1) (8 -10).
The signature feature of the metallophosphoesterase superfamily is an active site composed of two metal ions (typically manganese, iron or zinc) coordinated with octahedral geometry by a cage of histidine, aspartate, and asparagine side chains (Fig. 1). The metals directly coordinate the scissile phosphate anion, as does the metal-binding asparagine. Plausible catalytic mechanisms have been proposed based on crystal structures of superfamily members with phosphate or sulfate in the active site (5,9,(11)(12)(13) and mutational studies of a few exemplary enzymes. We construe the active site configuration of the Mycobacterium tuberculosis Rv0805 protein bound to a phosphate anion (14) to mimic the substrate complex of the phosphoesterase reaction (Fig. 1, top panel). In this structure, a metal-bridging water is situated 3 Å from the phosphorus atom. The almost perfectly apical orientation of this water to the putative "leaving" oxygen atom implicates the metal-bridged water as the nucleophile in the hydrolysis reaction. A putative mimetic of the product complex is exemplified by the active site of the Escherichia coli YfcE protein (15) (Fig. 1,  middle panel). Here, the tetrahedral sulfate anion has undergone stereochemical inversion relative to the phosphate in Rv0805, and the former metal-bridged water is incorporated into the anion product.
The rapid pace of identification and structural/biochemical characterization of new members of the binuclear metallophosphoesterase superfamily via genome mining has not been matched by progress in understanding the biological functions and relevant substrates of most of these enzymes. The empirical approach is to produce recombinant protein and survey for activity with a broad a range of phosphoester substrates and then surmise a role based on the results. In this way, it was inferred that the M. tuberculosis Rv0805 enzyme depicted in Fig. 1 functions as a 3Ј,5Ј-cyclic nucleotide phosphodiesterase (16). By contrast, the E. coli YfcE protein catalyzed manganesedependent hydrolysis of bis-p-nitrophenyl phosphate (15) but had no activity with any natural phosphodiesters tested (nucleic acid, cyclic nucleotides, and phosphatidyl choline) or any of 57 natural phosphomonoester substrates (nucleotides, sugars, and amino acids).
Our studies have focused on Clostridium thermocellum polynucleotide kinase-phosphatase (CthPnkp) as a model to probe the binuclear metallophosphoesterase mechanism and the determinants of substrate specificity (7,(17)(18)(19)(20). CthPnkp catalyzes 5Ј and 3Ј RNA end-healing reactions that prepare broken RNA termini for sealing by RNA ligase (17,20). The central 3Ј end-healing domain of CthPnkp belongs to the binuclear metallophosphoesterase superfamily; extensive mutational analysis underscores the strong similarity of the active site of CthPnkp to that of -Pase with respect to the metal and phos-phate ligands (7,18,19). Biochemically, CthPnkp is a Ni 2ϩ / Mn 2ϩ -dependent phosphodiesterase/monoesterase, active on nucleotides (2Ј,3Ј-cAMP, 3Ј-AMP and 2Ј-AMP) and generic substrates (bis-p-nitrophenyl phosphate, p-nitrophenyl phosphate, and p-nitrophenyl phenylphosphonate). The phosphodiesterase and monoesterase reactions rely on overlapping but different ensembles of active site functional groups. The enzyme is remarkably plastic, insofar as CthPnkp can be transformed toward narrower metal and substrate specificities via mutations of the active site. For example, certain changes (e.g. replacing the metal-binding His 189 residue with aspartate) transform CthPnkp into a Mn 2ϩ -dependent phosphodiesterase devoid of monoesterase activity (19).
We have analyzed in depth the 2Ј,3Ј-cyclic phosphodiesterase activity of CthPnkp, in light of the fact that 2Ј,3Ј-cyclic phosphate termini are the predominant products of several known RNA damage pathways (7). We found that alanine, glutamine, or asparagine mutations at the phosphate-binding residue His 264 of CthPnkp (corresponding to His 76 in -Pase or His 98 in Rv0805; Fig. 1) crippled the 2Ј,3Ј-cyclic phosphodiesterase activity, whereas the same changes enhanced the generic phosphodiesterase activity of CthPnkp with bis-p-nitrophenyl phosphate.
Our results prompted speculation that binuclear metallophosphoesterases might evolve distinct biochemical specificities via subtle changes at the active site (7). In particular, we predicted a correlation between 2Ј,3Ј-cyclic nucleotide phosphodiesterase activity and the presence of a phosphate-binding histidine analogous to His 264 . In the present study, we tested this idea by comparing the repertoire of M. tuberculosis Rv0805, which has this histidine in its active site (His 98 ), with that of E. coli YfcE, which has a cysteine at the equivalent position (Cys 74 ). We find that Rv0805 has a vigorous 2Ј,3Ј-cyclic phosphodiesterase function (a property missed in earlier studies of this enzyme), and we characterize the activity here, especially its dependence on His 98 . By introducing a histidine at the YfcE active site in lieu of Cys 74 , we transform an inactive protein into a 2Ј,3Ј-cyclic phosphodiesterase.

Materials
p-Nitrophenyl phosphate, bis-p-nitrophenyl phosphate, p-nitrophenol, cAMP, cGMP, and cUMP were purchased from Sigma. Malachite green reagent was purchased from BIOMOL Research Laboratories.

Purification of Rv0805
The M. tuberculosis gene Rv0805 was amplified by two-stage overlap extension PCR (21) from genomic DNA with Pfu DNA polymerase using primers designed to eliminate an internal BamHI site while introducing an NdeI restriction site at the start codon and a BamHI site 3Ј of the stop codon. The PCR product was digested with NdeI and BamHI and inserted into pET16b to generate an expression plasmid encoding the 318amino acid Rv0805 polypeptide fused to an N-terminal His 10 tag. Missense mutations H98N and H98A were introduced into the Rv0805 open reading frame by PCR using the two-stage overlap extension method (21). The inserts were sequenced to Water is colored red. The phosphate-binding histidine in Rv0805 (His 98 ) is replaced by a cysteine in YfcE (Cys 74 ). The bottom panel shows models of two potential orientations of 2Ј,3Ј-cGMP in the active site of Rv0805. The 2Ј,3Ј-cGMP molecule was imported from Protein Data Bank code 1GSP. The cyclic phosphate was superimposed on the phosphate anion in the Rv0805 structure. When the ribose O2Ј is apical to the metal-bridged water nucleophile (left), the reaction yields a 3Ј-PO 4 nucleotide product. When the ribose O3Ј atom is apical to the water nucleophile (right), the product is a 2Ј-PO 4 nucleotide. The His 98 side chain is poised to donate a hydrogen bond to the leaving ribose oxygen atom in the modeled 2Ј,3Ј-cGMP substrate in either orientation.
verify that there were no unwanted coding changes. Wild type and mutant pET-Rv0805 plasmids were transformed into E. coli strain BL21(DE3). Cultures (200 ml) of E. coli BL21(DE3)/pET-Rv0805 were grown at 37°C in Luria-Bertani medium containing 0.1 mg/ml ampicillin until the A 600 reached ϳ0.6. The cultures were chilled on ice for 30 min, adjusted to 0.4 mM isopropyl-␤-D-thiogalactopyranoside and 2% (v/v) ethanol, and then incubated at 17°C for 16 h with continuous shaking. The cells were harvested by centrifugation, and the pellet was stored at Ϫ80°C. All of the subsequent procedures were performed at 4°C. Thawed bacteria were resuspended in 20 ml of buffer A (50 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 10% sucrose). Lysozyme, phenylmethylsulfonyl fluoride, and Triton X-100 were added to final concentrations of 1 mg/ml, 1 mM, and 0.1%, respectively. The lysates were sonicated to reduce viscosity, and insoluble material was removed by centrifugation. The soluble extracts were applied to 1-ml columns of nickel-nitrilotriacetic acid-agarose (Qiagen) that had been equilibrated with buffer A. The columns were washed with 8 ml of the same buffer and then eluted stepwise with 4-ml aliquots of 25, 50, 200, and 500 mM imidazole in buffer B (50 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 10% glycerol). The polypeptide compositions of the column fractions were monitored by SDS-PAGE. The His 10 -Rv0805 proteins adsorbed to the column and were recovered predominantly in the 200 mM imidazole eluates. Protein concentrations were determined by using the Bio-Rad dye reagent with bovine serum albumin as the standard. The Rv0805 preparations were stored at Ϫ80°C.

Purification of YfcE
The open reading frame encoding the 184-amino acid YfcE polypeptide was amplified from E. coli genomic DNA with primers that introduced an NdeI site at the start codon and a BamHI site 3Ј of the stop codon. The PCR product was digested with NdeI and BamHI and inserted into pET16b to generate an expression plasmid encoding His 10 -tagged YfcE. Missense mutations C74A, C74N, and C74H were introduced by PCR using the two-stage overlap extension method (21). The pET-YfcE plasmids were transformed into E. coli BL21(DE3). The His 10 -YfceE proteins were purified from soluble extracts of 200-ml cultures of isopropyl-␤-D-thiogalactopyranoside-induced bacteria as described above for His 10 -Rv0805.

Hydrolysis of p-Nitrophenyl Phosphate and Bis-p-nitrophenyl Phosphate
Reaction mixtures (25 l) containing 50 mM Tris-HCl (pH 8.5), 0.5 mM MnCl 2 , 10 mM p-nitrophenyl phosphate or bis-pnitrophenyl phosphate, and Rv0805 or YfcE as specified were incubated at 37°C. The reactions were quenched by adding either 20 mM EDTA (YfcE reactions) or 5% SDS (Rv0805 reactions), followed by 0.9 ml of 1 M Na 2 CO 3 . Release of p-nitrophenol was determined by measuring A 410 and interpolating the value to a p-nitrophenol standard curve.

Hydrolysis of Cyclic Nucleotides
Reaction mixtures (10 l) containing 50 mM Tris-HCl (pH 8.5), 0.5 mM MnCl 2 , 10 mM cyclic nucleotide as specified, and either Rv0805, YfcE, or calf intestine phosphatase (CIP) as spec-ified were incubated for 10 min at 37°C. (CIP was present in excess and did not limit the extent of phosphate release.) The reactions were quenched by adding 20 mM EDTA, followed by 1 ml of malachite green reagent. Release of phosphate was determined by measuring A 620 and interpolating the value to a phosphate standard curve.
Hydrolysis of 2Ј,3Ј-cAMP-Reaction mixtures (10 l) containing 50 mM Tris-HCl (pH 8.5), 0.5 mM MnCl 2 , 1 unit CIP, increasing concentrations (0.625, 1.25, 2.5, 5.0, or 10 mM) of 2Ј,3Ј-cAMP, and either 7.5 pmol Rv0805 (0.75 M enzyme) or 25 pmol YfcE-C74H (2.5 M enzyme) were incubated at 37°C for 15 min (YfcE-C74H) or 5 min (Rv0805). The enzyme concentrations and incubation times were chosen to ensure that Յ36% of the substrate was converted to product at the lowest substrate concentrations tested (the ranges were from 9 to 36% conversion). The extents of p-nitrophenol or P i production were first plotted as a function of substrate concentration. K m and k cat were then calculated from Eadie-Hofstee plots of the data. The K m and k cat values reported in Table 1 are averages from two independent substrate titration experiments Ϯ mean absolute error.

2Ј,3Ј-Cyclic Phosphodiesterase Activity of M. tuberculosis
Rv0805-Shenoy et al. (16) found that Rv0805 catalyzes Mn 2ϩdependent cleavage of bis-p-nitrophenyl phosphate and hydrolysis of 3Ј,5Ј-cAMP to 5Ј-AMP. The ability of Rv0805 to hydrolyze 2Ј,3Ј-cyclic nucleotides was not reported. Here, we produced Rv0805 in E. coli as a His 10 fusion and purified the enzyme from a soluble bacterial extract by nickel-agarose chromatography (Fig. 2). The recombinant protein hydrolyzed 10 mM bis-p-nitrophenyl phosphate in the presence of 0.5 mM MnCl 2 to yield p-nitrophenol; the extent of product formation was proportional to input enzyme (Fig. 3A). From the slope of the titration curve, we calculated a turnover number of 12.6 s Ϫ1 . Formation of p-nitrophenol by Rv0805 displayed a hyperbolic dependence on the concentration of bis-p-nitrophenyl phosphate (not shown). From an Eadie-Hofstee plot, we calculated a K m of 0.9 mM and k cat of 12.4 s Ϫ1 (Table 1). (The kinetic parameters reported previously by Shenoy et al. (16) were: K m ϭ 1.3 mM bis-p-nitrophenyl phosphate and k cat ϭ 4.2 s Ϫ1 .) Rv0805 displayed much weaker activity as a phosphomonoesterase (Fig. 3B). It hydrolyzed 10 mM p-nitrophenyl phosphate to p-nitrophenol with a specific activity of 0.5 s Ϫ1 . From the results of a substrate titration experiment, we calculated a K m of 1.7 mM p-nitrophenyl phosphate and k cat of 0.55 s Ϫ1 (Table 1). Thus, the catalytic efficiency (k cat /K m ) of Rv0805 was 43-fold greater for the phosphodiesterase substrate.
2Ј,3Ј-Cyclic nucleotide phosphodiesterase activity was tested by reacting Rv0805 with 10 mM 2Ј,3Ј-cAMP in the presence of 0.5 mM MnCl 2 . CIP was included in the reaction to liberate inorganic phosphate from any nucleoside phosphomonoesters formed by Rv0805. In the experiment shown in Fig. 4, Rv0805 converted 16% of the input 10 mM 2Ј,3Ј-cAMP to a CIP-sensitive phosphomonoester. By contrast, a control reaction with CIP alone released Ͻ1% of the P i from 2Ј,3Ј-cAMP. Also, no free phosphate was released from 2Ј,3Ј-cAMP by Rv0805 in the absence of CIP.
To query whether Rv0805 displays specificity toward a particular nucleotide and whether the enzyme discriminates between a 2Ј,3Ј-cyclic phosphate and a 3Ј,5Ј-cyclic phosphate, we reacted the enzyme with 10 mM 2Ј,3Ј-cGMP, 3Ј,5Ј-cAMP, 3Ј,5Ј-cGMP, or 3Ј,5Ј-cUMP substrates. Rv0805 converted 16% of the input 2Ј,3Ј-cGMP to a CIP-sensitive phosphomonoester; thus, the enzyme did not have a preference for 2Ј,3Ј-cAMP versus 2Ј,3Ј-cGMP. The salient finding was that Rv0805 displayed much weaker activity as a 3Ј,5Ј-cyclic phosphodiesterase, converting only 0.2, 1, and 1.6% of the input 3Ј,5Ј-cAMP, 3Ј,5Ј-cGMP, and 3Ј,5Ј-cUMP substrates to CIP-sensitive phosphomonoesters, respectively (Fig. 4A). An enzyme titration experiment (Fig. 3D) showed that the specific activity of Rv0805 as a 2Ј,3Ј-cyclic AMP phosphodiesterase was 150-fold higher than as a 3Ј,5Ј-cyclic AMP phosphodiesterase (estimated turnover numbers of 150 and 1 min Ϫ1 , respectively). The kinetic data reported by Shenoy et al. (16) for hydrolysis of 3Ј,5Ј-cAMP indicated a k cat value of ϳ1.7 min Ϫ1 , which agrees with the value we observe. Thus, the conclusion by Shenoy et al., that Rv0805 is a 3Ј,5Ј-cyclic nucleotide phosphodiesterase, is open to question in light of our determination that it has 2 orders of magnitude higher activity with 2Ј,3Ј-cyclic nucleotide substrates. Additional experiments revealed that Rv0805 had essentially no detectable ability to hydrolyze phosphomonoester substrates 5Ј-AMP, 3Ј-AMP, or 2Ј-AMP (10 mM) in Tris-HCl buffer (pH 8.5) the presence of 0.5 mM MnCl 2 . Specifically, a 10-l reaction mixture so constituted with 21 M Rv0805 released Յ0.1 nmol of inorganic phosphate (the lower limit of detection of the assay) as product from 100 nmol of input 5Ј-AMP, 3Ј-AMP, or 2Ј-AMP substrate during a 10-min incubation at 37°C, which corresponds to a turnover number of Յ0.05 min Ϫ1 , a value 3,000-fold lower than the activity of Rv0805 with 2Ј,3Ј-cAMP.
Formation of a CIP-sensitive adenylate by Rv0805 displayed a hyperbolic dependence on the concentration of 2Ј,3Ј-cAMP. From an Eadie-Hofstee plot, we calculated a K m of 1.6 mM and k cat of 2.8 s Ϫ1 (Table 1). To determine the chemical identity of the products of the reaction with 2Ј,3Ј-cAMP, we performed cellulose TLC analysis of the reaction mixture as a function of enzyme in the presence of Rv0805 only (no CIP). The TLC plate was developed with buffer containing saturated ammonium   NOVEMBER 7, 2008 • VOLUME 283 • NUMBER 45 JOURNAL OF BIOLOGICAL CHEMISTRY 30945 sulfate, 3 M sodium acetate, isopropanol (80/6/2), in which the order of migration away from the origin (R f ) is 2Ј,3Ј-cAMP Ͻ 3Ј-AMP Ͻ 2Ј-AMP (22). This experiment revealed a quantitative conversion of 2Ј,3Ј cAMP to a mixture of more quickly moving products. The major species corresponded to 3Ј-AMP, and the minor product migrated as 2Ј-AMP (Fig. 4B). Their proportions did not change over the range of Rv0805 concentrations tested, arguing against a precursor-product relationship between the major and minor species. Thus, we conclude that Rv0805 preferentially cleaves the P-O2Ј bond of 2Ј,3Ј-cAMP.

Metallophosphodiesterase Substrate Specificity Determinants
Effect of His 98 Mutations on the 2Ј,3Ј-Cyclic Phosphodiesterase Activity of Rv0805-We produced and purified Rv0805 mutants with alanine or asparagine in place of His 98 (Fig. 2). The specific activity of the H98A protein in hydrolysis of bis-pnitrophenyl phosphate was identical to that of wild type Rv0805, whereas the specific activity of H98N was 2-fold higher than the wild type value (Fig. 3A). By contrast, the H98A muta-tion virtually abolished phosphomonoesterase activity on the p-nitrophenyl phosphate substrate (Fig. 3B). The specific activity of the H98N protein with p-nitrophenyl phosphate (0.12 s Ϫ1 ) was one-fourth the wild type level (Fig. 3B). The salient findings were that the H98A and H98N changes both suppressed the 2Ј,3Ј-cAMP phosphodiesterase specific activity to one-fifth the level of wild type Rv0805 (Fig. 3C). These selective inhibitory effects of His 98 mutations on hydrolysis of 2Ј,3Ј-cAMP (and p-nitrophenyl phosphate) but not bis-p-nitrophenyl phosphate are in accord with the mutational effects seen for His 264 in CthPnkp (7,19).
Phosphodiesterase Activity of E. coli YfcE-Miller et al. (15) characterized E. coli YfcE protein as a Mn 2ϩ -dependent phosphodiesterase that cleaved bis-p-nitrophenyl phosphate, thymidine 5Ј-monophosphate-p-nitrophenyl ester, and p-nitrophenyl phosphorylcholine but was unable to hydrolyze 2Ј,3Ј-or 3Ј,5Ј-cyclic nucleotide phosphodiesters or any of 57 different phosphomonoesters, including p-nitrophenyl phosphate. The crystal structure of YfcE with metals and sulfate in the active site (15) reveals that YfcE has a cysteine (Cys 74 ) at the position equivalent to Rv0805 His 98 and CthPnkp His264 (Fig. 1). We seized on YfcE as a promising scaffold to test the key prediction of our substrate-specificity model, by engineering a gain-offunction in a metallophosphoesterase enzyme that ordinarily lacks the ability to hydrolyze 2Ј,3Ј-cyclic nucleotides.
We produced wild type YfcE as a His 10 fusion and purified it from a soluble bacterial extract by nickel-agarose chromatography (Fig. 2). YfcE hydrolyzed 10 mM bis-p-nitrophenyl phosphate in the presence of 0.5 mM MnCl 2 to yield p-nitrophenol; from the slope of the titration curve (Fig. 5A), we estimated a turnover number of 22 s Ϫ1 . From the results of a substrate titration experiment, we calculated a K m of 4.8 mM bis-p-nitrophenyl phosphate and a k cat of 18 s Ϫ1 (Table 1). (The kinetic parameters reported previously by Miller et al. (15) were: K m ϭ 9.7 mM bis-p-nitrophenyl phosphate and k cat of 20 s Ϫ1 .) Unlike Miller et al., we were able to detect a generic YfcE phosphomonoesterase activity. YfcE hydrolyzed 10 mM p-nitrophenyl phosphate to p-nitrophenol in an enzyme concentration-dependent manner (Fig. 5B). From the slope of the titration curve, we estimated a turnover number of 0.4 s Ϫ1 . Kinetic parameters for the YfcE phosphomonoesterase reaction determined from a substrate titration experiment were: K m ϭ 10.6 mM p-nitrophenyl phosphate and k cat ϭ 0.9 s Ϫ1 ( Table 1). The catalytic efficiency (k cat /K m ) of YfcE is thereby 44-fold greater for the generic phosphodiesterase substrate. YfcE had feeble activity in hydrolysis of 2Ј,3Ј-cAMP to a CIP-sensitive nucleoside monoester (Fig. 5C). From the slope of the titration curve, we estimated a turnover number of 0.03 s Ϫ1 .  Transformation of YfcE into a 2Ј,3Ј-Cyclic Nucleotide Phosphodiesterase-YfcE mutants with alanine, asparagine, or histidine in lieu of Cys 74 were purified (Fig. 2) and surveyed for phosphodiesterase and monoesterase activities (Fig. 5). The specific activities in hydrolysis of bis-p-nitrophenyl phosphate (Fig. 5A) varied within a 5-fold range according to the amino acid at position 74, as follows: His (56 s Ϫ1 ) Ͼ Asn (40 s Ϫ1 ) Ͼ Cys (22 s Ϫ1 ) Ͼ Ala (10 s Ϫ1 ). Greater salutary effects of the histidine and asparagine changes were observed for phosphomonoesterase activity with p-nitrophenyl phosphate (Fig. 5B), with the following hierarchy of specific activities: Asn (6 s Ϫ1 ) Ͼ His (2.5 s Ϫ1 ) Ͼ Cys (0.4 s Ϫ1 ) Ϸ Ala (0.3 s Ϫ1 ). The increase of generic Mn 2ϩdependent monoesterase activity in YfcE-C74H versus C74A reciprocated the loss of generic monoesterase function observed when the Rv0805 and CthPnkp histidines were mutated to alanine. The YfcE-C74N and C74H titration curves deviated downward from linearity at the higher levels of input enzyme at which Ն30% of the 10 mM p-nitrophenyl phosphate substrate was converted to p-nitrophenol and P i (Fig. 5B). This reflected product inhibition by P i . From a separate experiment entailing prior addition of increasing amounts of phosphate to a reaction mixture containing 10 mM p-nitrophenyl phosphate and 0.2 M YfcE-C74H, we determined an IC 50 value of 2.5 mM P i (data not shown).
To determine the chemical identities of the products of the reaction with 2Ј,3Ј-cAMP, we performed cellulose TLC analysis of the reaction mixture as a function of reaction time in the presence of YfcE-C74H only (no CIP). YfcE-C74H quantitatively converted the 2Ј,3Ј-cAMP substrate to a single product that comigrated with the 3Ј-AMP standard (Fig. 6B). We conclude that YfcE-C74H cleaves exclusively the P-O2Ј bond of 2Ј,3Ј-cAMP. Steady-state kinetic parameters for the YfcE-C74H cyclic phosphodiesterase derived from 2Ј,3Ј-cAMP titra-  tion experiments were as follows: K m ϭ 35 mM 2Ј,3Ј-cAMP and k cat ϭ 2.95 s Ϫ1 (Table 1). It is notable that the k cat of the "designed" YfcE-C74H 2Ј,3Ј-cAMP phosphodiesterase was similar to that of the "natural" Rv0805 enzyme, although the affinity of the YfcE-C74H protein for the substrate was 22-fold less than Rv0805.
These results attest to the transformative power of the active site histidine as a determinant of cyclic nucleotide phosphodiesterase activity. As discussed below, we surmise that enzymic groups other than the phosphate-binding histidine might contribute to cyclic nucleotide binding and orientation of the leaving group.

DISCUSSION
The present study supports a model, suggested by our studies of CthPnkp (7), that the presence of a phosphate-binding histidine in the active site of phosphodiesterase members of the binuclear metallophosphoesterase superfamily is a determinant of 2Ј,3Ј-cyclic nucleotide phosphodiesterase activity. Here we focused on two structurally characterized metallophosphoesterases that differ in having a histidine (Rv0805) or a cysteine (YfcE) at this active site position. Although Rv0805 had been dubbed a 3Ј,5Ј-cyclic nucleotide phosphodiesterase by other investigators (16), our characterization of this enzyme shows it to be 150-fold more active on 2Ј,3Ј-cAMP than 3Ј,5Ј-cAMP. Thus, the presence of an active site histidine in Rv0805 correctly predicted its heretofore unexamined capacity for 2Ј,3Ј-cNMP hydrolysis. Changing the histidine to alanine or asparagine suppressed 2Ј,3Ј-cAMP phosphodiesterase activity of Rv0805 without affecting the hydrolysis of a generic nonnucleotide phosphodiester substrate. Even more compelling evidence for the defining role of the histidine derives from our ability to convert the otherwise inactive YfcE protein into an active and highly specific 2Ј,3Ј-cNMP phosphodiesterase by introducing a histidine in lieu of Cys 74 .
The correlation of an active site histidine and 2Ј,3Ј-cNMP phosphodiesterase activity applies to other metallophosphoesterase superfamily members, including -Pase (7) and the recently characterized Deinococcus radiodurans enzyme DR1281 (23). DR1281 resembles Rv0805 in its ϳ35-fold higher k cat for Mn 2ϩ -dependent hydrolysis of bis-p-nitrophenyl phosphate versus p-nitrophenyl phosphate and its selective hydrolysis of 2Ј,3Ј-cNMPs versus 3Ј,5Ј-cNMPs (23). The correlation between a nonhistidine residue and the absence of cyclic phosphodiesterase activity seen here with YfcE is reminiscent of the properties of the structurally characterized Methanococcus jannaschii enzyme MJ0936 (24). MJ0936 has vigorous activity in hydrolyzing bis-p-nitrophenyl phosphate but is unable to cleave p-nitrophenyl phosphate. Although possessed of a generic phosphodiesterase activity, MJ0936 reportedly had no detectable cyclic phosphodiesterase activity with the 2Ј,3Ј-or 3Ј,5Ј-forms of cAMP or cGMP (24). The crystal structure of manganese-bound MJ0936 (Protein Data Bank code 1S3N) (24) reveals the similarity of its active site to that of Rv0805, except for the presence of an asparagine in lieu of the phosphate-coordinating histidine.
Although the histidine is clearly a major determinant of 2Ј,3Ј-cNMP phosphodiesterase activity, the outcomes of the hydrolysis step can differ significantly from one enzyme to another. CthPnkp-H189D (a diesterase-only mutant in which a metal-binding histidine is changed to aspartate) catalyzes hydrolysis of the P-O3Ј bond of 2Ј,3Ј-cAMP or -cGMP to yield exclusively 2Ј-AMP or 2Ј-GMP products (7,20). By contrast, YfcE-C74H (Fig. 6) and DR1281 (23) both catalyze hydrolysis of the P-O2Ј bond of 2Ј,3Ј-cAMP to yield 3Ј-AMP as the sole product. Rv0805, on the other hand, can hydrolyze either P-O2Ј or P-O3Ј to yield a mixture of 3Ј-AMP and 2Ј-AMP products, with a clear bias toward generation of 3Ј-AMP (Fig.  3). -Pase also hydrolyzes either P-O2Ј or P-O3Ј to yield both 3Ј-AMP and 2Ј-AMP products, albeit with a preference for 2Ј-AMP formation (7). We surmise that the histidine facilitates an aspect of phosphohydrolase chemistry common to both pathways (conceivably entailing proton donation to the ribose O2Ј or O3Ј leaving atom), but constituents other than the histidine dictate the reaction outcome by influencing substrate orientation and affinity.
The critical issue of substrate orientation in determining which products are formed is illustrated in Fig. 1 (bottom  panel), in which we have modeled a 2Ј,3Ј-cGMP molecule (imported from a crystal structure of RNase T1; Protein Data Bank code 1GSP) (33) into the active site of Rv0805 by superimposing the cyclic phosphate moiety of cGMP on the phosphate anion in the Rv0805 structure. This results in two potential configurations for opening the cyclic phosphate. When the ribose O2Ј is apical to the metal-bridged water nucleophile, the reaction yields a 3Ј-PO 4 nucleotide product. When the ribose O3Ј atom is apical to the water nucleophile, the product is a 2Ј-PO 4 nucleotide. The His 98 side chain is poised to donate a hydrogen bond to the leaving ribose oxygen atom in the modeled substrate complex with 2Ј,3Ј-cGMP in either orientation (Fig. 1, bottom panel). The asymmetry of the binding modes is apparent in the locations of the bulky purine base relative to the (pseudo-mirror-symmetrical) cyclic phosphate-ribose ring system. It is likely that steric constraints on the adoption of one orientation versus the other are responsible for stringent and opposite choice of leaving groups during the 2Ј,3Ј-cNMP phosphodiesterase reactions of CthPnkp-H189D versus YfcE-C74H and DR1281. We speculate that Rv0805 and -Pase are less constrained with respect to the binding orientations of the cyclic nucleotide, thereby allowing formation of either 2Ј-NMP or 3Ј-NMP products. Further evaluation of this hypothesis and elucidation of the structural elements that dictate substrate orientation, will hinge on determining crystal structures of metallophosphodiesterase enzymes with a 2Ј,3Ј-cyclic nucleotide bound in the active site.
Are there useful biological inferences to be drawn from the fact that a metallophosphodiesterase family member has a vigorous and relatively specific 2Ј,3Ј-cyclic nucleotide phosphodiesterase activity in vitro? In the case of CthPnkp, we have suggested that this activity is relevant to the repair of RNA 2Ј,3Ј-cyclic ends (7,20), which are natural intermediates in RNA processing (25) and RNA catabolism (26). RNA 2Ј,3Ј-cyclic ends are also the end products of RNA-cleaving toxins (27)(28)(29)(30). An initial clue to a nucleic acid repair role for CthPnkp was the covalent linkage of its metallophosphoesterase domain to a polynucleotide kinase module (17) that is known to catalyze 5Ј end healing reactions in RNA repair pathways (25,27). In the case of Rv0805 and -Pase, there are no physiologically instructive flanking domains, and there is, to our knowledge, no genetic evidence implicating either protein in a particular biological process in its native context. The prospect that M. tuberculosis Rv0805 acts on broken RNAs is worthy of consideration, given that M. tuberculosis encodes at least seven MazF-like endoribonuclease toxins that generate site-specific breaks with 2Ј,3Ј-cyclic ends (30 -32).
Finally, the correlation between a nonhistidine residue (Cys 74 ) in the YfcE active site and the absence of cyclic phosphodiesterase activity hints at a novel specificity within the superfamily. We think it is unlikely that the histidine substitution by cysteine in YfcE is a sporadic event resulting in a "crippled" metallophosphodiesterase diesterase enzyme. Indeed, Miller et al. (15) described a novel clade of bacterial YfcE-like binuclear metallophosphoesterases, embracing predicted polypeptides from diverse bacterial genera, each of which has a cysteine in lieu of the phosphate-binding histidine. A current data base search identifies members of this cysteine-containing bacterial subfamily in Shigella, Citrobacter, Enterobacter, Klebsiella, Yersinia, Serratia, Erwinia, Vibrio, Shewanella, Photobacterium, Aeromonas, Clostridium, Bacteroides, Ruminococcus, Thermotoga, Moorella, Treponema, and others. The fact that YfcE can hydrolyze 2Ј,3Ј-cNMPs when mutated to His 74 hints that this subfamily evolved toward a narrow substrate specificity. The "real" substrate for YfcE-like phosphodiesterases (and thus the specificity determining role of the active site cysteine) remains to be discovered.