Histidine Triad Nucleotide-binding Protein 1 (HINT-1) Phosphoramidase Transforms Nucleoside 5′-O-Phosphorothioates to Nucleoside 5′-O-Phosphates*

Nucleoside 5′-O-phosphorothioates are formed in vivo as primary products of hydrolysis of oligo(nucleoside phosphorothioate)s (PS-oligos) that are applied as antisense therapeutic molecules. The biodistribution of PS-oligos and their pharmacokinetics have been widely reported, but little is known about their subsequent decay inside the organism. We suggest that the enzyme responsible for nucleoside 5′-O-monophosphorothioate ((d)NMPS) metabolism could be histidine triad nucleotide-binding protein 1 (Hint-1), a phosphoramidase belonging to the histidine triad (HIT) superfamily that is present in all forms of life. An additional, but usually ignored, activity of Hint-1 is its ability to catalyze the conversion of adenosine 5′-O-monophosphorothioate (AMPS) to 5′-O-monophosphate (AMP). By mutagenetic and biochemical studies, we defined the active site of Hint-1 and the kinetic parameters of the desulfuration reaction (P-S bond cleavage). Additionally, crystallographic analysis (resolution from 1.08 to 1.37 Å) of three engineered cysteine mutants showed the high similarity of their structures, which were not very different from the structure of WT Hint-1. Moreover, we found that not only AMPS but also other ribonucleoside and 2′-deoxyribonucleoside phosphorothioates are desulfurated by Hint-1 at the following relative rates: GMPS > AMPS > dGMPS ≥ CMPS > UMPS > dAMPS ≫ dCMPS > TMPS, and during the reaction, hydrogen sulfide, which is thought to be the third gaseous mediator, was released.

NMPS and dNMPS (together denoted (d)NMPS)) are formed during the enzymatic hydrolysis of oligo(nucleoside phosphorothioate) (PS-oligos) that contain a sulfur atom attached in non-bridging positions to the phosphorus atom at each or selected internucleotide bond(s). Synthetic PS-oligos have been developed as antisense probes for genomic research and medicinal applications (1,2). These oligonucleotides are promising therapeutic molecules because they are much more stable against nucleolytic degradation in blood and various cellular systems than their natural, unmodified counterparts (3)(4)(5). Their hydrolysis in plasma, kidney, and liver proceeds mainly from the 3Ј end, resulting in the appearance of the mononucleoside 5Ј-phosphorothioates identified in urine from PS-oligo-injected animals (6,7). (d)NMPS may exert cytotoxic effects affecting cell proliferation, DNA or RNA synthesis, and other unknown processes (8,9). Recently, the phosphorothioate DNA segments have been identified in bacterial DNA (10), which makes investigations into PS-oligo metabolism even more important.
Although several reports have been published on the biodistribution of PS-oligos, little is known about the metabolism of the products of their degradation in vivo. It has been suggested that extracellular dNMPS and dNMP can be converted to the corresponding nucleoside by 5Ј-nucleotidase (ecto-5Ј-NT) (9). This membrane-bound enzyme preferably releases adenosine from extracellular AMP, but other purine and pyrimidine 5Ј-nucleotides are also dephosphorylated by this protein (11). Alkaline phosphatase has also been shown to be able to hydrolyze O-phosphorothioate monoesters, although much more slowly than their phosphate analogs (12,13). It is possible that nucleoside phosphorothioates are also substrates for this enzyme.
Other classes of compounds that can be transformed to (d)NMP and (d)NMPS in vivo are nucleoside phosphoramidates (2, Fig. 1, X ϭ O) and phosphoramidothioates (2, X ϭ S) carrying an N-alkyl residue, most often derived from amino acids (14 -16). In the prodrug approach, such molecules are converted intracellularly to the corresponding nucleoside 5Јmonophosphates (17). It is believed that the enzyme responsible for hydrolysis of the P-N bond in both classes of nucleotide phosphoramidates is histidine triad nucleotide-binding protein 1 (Hint-1) (14,18). The Hint-1 protein is a member of the histidine triad (HIT) protein superfamily. The family members act as hydrolases and transferases and contain an amino acid sequence motif, HXHXHXXX (H, histidine residue; X, hydrophobic amino acid residue), at their C termini (19). Interestingly, Hint-1 exerts its phosphoramidase activity toward several synthetic phosphoramidate substrates, such as AMP-N-⑀-lysine, AMP-N-alanine, AMP-N-morpholine, and adenosine 5Ј-phosphoramidate (AMP-NH 2 ) (20). Although the cellular function and biochemical relevance of that activity remains unknown, it has been suggested that mammalian Hint-1 has some tumor suppressor activity (21)(22)(23) and is involved in regulating apoptosis (24).
Crystallographic studies on human and rabbit Hint-1 revealed that the HIT motif is involved in nucleotide substrate binding and hydrolysis (25,26). The phosphoramidase activity of Hint-1 strongly depends on the presence of the conserved middle histidine of the histidine triad motif (His-112 for rHint-1) (20). In contrast to earlier suggestions, Ser-107 but not His-114 acts as the acid-base catalyst (27).
Another activity of Hint-1, the ability to hydrolyze adenosine 5Ј-O-phosphorothioate (AMPS), was initially described by Bieganowski et al. (20). During our studies on the stereochemistry of rabbit Hint-1-assisted hydrolysis of the P-diastereoisomers of adenosine 5Ј-O-[N-(tryptophanylamide)]phosphoramidothioate (AMPS-Trp, 2, Fig. 2, X ϭ S, RNH ϭ tryptophanyl residue), we observed that the stereoretentive process of the P-N bond cleavage in AMPS-Trp was followed by effective enzyme-assisted removal of the sulfur atom from AMPS, resulting in the formation of AMP (18).
In this report, we describe mutagenetic and biochemical studies executed to define the substrates, active site, and kinetic parameters of the P-S bond cleavage catalyzed by Hint-1, which is the putative enzyme responsible for (d)NMPS metabolism. In the catalysis reaction, we analyzed the participation of two cysteine residues (Cys-38 and Cys-34 for rHint-1), Ser-107 and the conserved third histidine of the HIT motif (His-114 for rHint-1). For that purpose, we engineered three cysteine mutants (C38A, C84A, and C38A/ C84A), a serine mutant (S107A), and two histidine mutants (H114D and H114N) and compared their activities toward AMP-N-⑀-lysine (a substrate for the Hint-1 phosphoramidase P-N bond cleavage activity) and AMPS (a substrate for the Hint-1 desulfurase P-S bond cleavage activity). Additionally, due to differences in their enzymatic activities, the crystal structures of the three cysteine mutants were compared.
Moreover, the analysis of Hint-1 activity toward new substrates indicated that not only AMPS but also GMPS, CMPS, UMPS, and corresponding phosphorothioate 2Ј-deoxyribonucleosides are desulfurated by Hint-1.
Mutagenesis and Purification of Proteins-Mutants of rabbit Hint-1 were generated by site-directed mutagenesis of the WT protein expression vector pSGA02-HINT (20)  To generate the C38A/C84A double mutant of rabbit Hint-1, site-directed mutagenesis was performed with primers, C84A on the C38A mutant, as described above. The sequences of all obtained plasmid constructs were confirmed by DNA sequencing (Institute of Biochemistry and Biophysics PAS, Warsaw, Poland).
Enzymatic Assays-For the enzymatic digestion, solutions of substrates at 200 M (AMP-N-⑀-lysine, NMPS, and dNMPS) or 1 mM (dCMPS, TMPS, AMPS(OMe), and dinucleotide d(ApsA)) were prepared in 20 mM Na-HEPES buffer (pH 7.2) or 66 mM sodium/potassium phosphate buffer (pH 7.2) containing 0.5 mM MgCl 2 and the enzyme (0.1-2.0 g) in a 10-l total volume. The samples were incubated for 0.1-24 h at 30°C. The reaction mixtures were quenched by cooling on ice and analyzed by RP-HPLC on a BDS Hypersil C18 column (5 m, 250 ϫ 4.6 mm; Thermo Electron Corporation) with a mobile phase, a gradient of acetonitrile in 0.1 M triethylammonium bicarbonate (pH 7.4) from 0 to 20% over 24 min at a flow rate of 1 ml/min. The specific activity of the Hint-1 enzyme was calculated from the reactions in which hydrolysis did not exceed 10%. Each experimental point represents the mean Ϯ S.E. from measurements performed in triplicate.
Kinetic Assay-To determine the kinetic parameters of the AMPS desulfuration by rHint-1, the initial rate assays were completed in a buffer containing 0.5 mM MgCl 2 and 20 mM Na-HEPES (pH 7.2) at 8 substrate concentrations, ranging from 50 M to 1 mM. The products were analyzed by RP-HPLC as described above. To calculate K m and k cat values, Lineweaver-Burk plots were used.
Circular Dichroism Measurements-The circular dichroism (CD) spectra were recorded on a CD6 dichrograph (Jobin-Yvon, Longjumeau, France) using cuvettes with a 0.1-mm path length, 2 nm bandwidth, and 1 to 2-s integration time. The protein samples (1 mg, molecular mass of 13,693 Da, 126 residues) were dissolved in 0.4 ml of buffer containing 20 mM Tris (pH 7.5) and 150 mM NaCl. The spectra were recorded in a range of 200 to 350 nm at 25°C and smoothed with a 25-point algorithm included in the manufacturer's software, version 2.2.1. The data in Table 1 were calculated using the algorithm K2D (a web server Computational Biology Group of European Molecular Biology Laboratory, Heidelberg, Germany), which uses the experimental CD spectrum obtained as described above.
Crystallization and X-ray Data Collection-C38A, C84A, and C38A/C84A rHint-1 mutants were crystallized as described for WT Hint-1 (25,26) by the vapor diffusion method in a hanging drop variant at 5°C using 10 mg/ml of protein solution and a precipitant solution containing 100 M sodium cacodylate buffer (pH 6.5), 20 to 30% (w/v) PEG 8000, both with or without the addition of 100 M sodium acetate. Crystallization drops were set up by mixing 2 l of the protein solution with 2 l of the precipitant solution and suspended over precipitant solution (1 ml). Crystals of typical dimensions (0.1 ϫ 0.2 ϫ 0.6 mm) and rhombic shape appeared after 48 to 72 h.
Crystals of rHint-1 cysteine mutants were flash-frozen by transferring them into 25% (v/v) PEG 400 (used as a cryoprotectant) followed by freezing in a nitrogen stream at Ϫ173°C. The diffraction data were collected using synchrotron radiation with a Rayonics MX-220 CCD detector on beamline MX-14-1 at BESSY, Berlin, Germany (C38A, C38A/C84A) and with an MAR CCD 165-mm detector on beamline I911-2 at the MAX-Lab Synchrotron, Lund, Sweden (C84A). The data were processed, integrated, and scaled with MOSFLM (29) and SCALA (30) from the CCP4 program package (31) or with DENZO and SCALEPACK from the HKL2000 program package (32). The data collection and processing statistics are given in Table 3.
Structure Determination, Refinement, and Analysis-All datasets were analyzed using programs from the CCP4 package. The structures were solved by the molecular replacement method using the program MOLREP (33). A search model for rHint-1 cysteine mutants was the structure of wild-type rHint-1 solved previously at 1.1-Å resolution (Protein Data Bank code 3LLJ). One monomer was identified and placed in the asymmetric unit of the crystal in each of the datasets. The structures were completed using alternate cycles of manual building, including main and side chain rebuilding, adding alternative residue conformations, loop fragments, and solvent molecules in COOT (34) and refinement in REFMAC5 (35). All refinement steps were monitored using R and R free values. The quality of each model was judged using the program PROCHECK (36). Values of mean temperature factors for the protein main and side chains as well as water molecules were calculated using the program BAVERAGE from the CCP4 program suite. Protein structures were compared using the program LSQKAB (37). Visualization of electrostatic potential surfaces was generated using PyMol (38).

H 2 S Evolution from the AMPS to AMP Transformation
Reaction-The Hint-1-assisted removal of the sulfur atom from AMPS resulted in the formation of AMP (18) and release of H 2 S into the reaction mixture; the characteristic rotten egg smell was sensed organoleptically (Fig. 3). The evolution of H 2 S to the reaction mixture containing rHint-1 and AMPS was confirmed by a highly sensitive (10 nmol) analytical reaction with AgNO 3 , leading to the formation of a black deposit of Ag 2 S. In an independent experiment, HEPES buffer, used in the AMPS to AMP transformation reaction, was excluded as a source of sulfur. Therefore, one could conclude that the produced H 2 S carried the sulfur atom originally present in the 5Ј-O-phosphorothioate moiety.
Recent studies suggest that hydrogen sulfide (H 2 S) is the third gaseous mediator, in addition to nitric oxide (NO) and carbon monoxide (CO), in mammalian cells (39). In most tissues, as well as in plasma, the physiological concentration of H 2 S is about 50 M, but it can be three times higher in the brain (40). Notably, the toxic concentration is only two times higher than the physiological concentration. H 2 S is involved in the regulation of many physiological processes, and its defi-   DECEMBER 24, 2010 • VOLUME 285 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 40811 ciency as well as its excess can cause physiological and even pathological effects (39,41,42).

Hint-1 Transforms (d)NMPS to (d)NMP
It is a widely accepted opinion that the P-S bond in phosphorothioate monoesters and diesters is quite stable under physiological conditions. In principle, the P-S bond in (d)NMPS can be converted into the P-O bond in (d)NMP by the following reactions: 1) oxidation with the participation of an R-SH group to form a disulfide P-S-S-R moiety, followed by hydrolysis; 2) Lewis acid-assisted hydrolysis; 3) alkylationpromoted hydrolysis; or 4) radical oxidation. For enzymatic conversion, the first two mechanisms should be considered more plausible. A sulfhydryl group of a cysteine residue may be involved in the initial oxidation, whereas the positively charged or proton-donating amino acid residues (e.g. Arg, Lys, and Ser) may act as the Lewis acid. The two other mechanisms seem to be much less likely, yet not impossible. Moreover, there are known enzymes (e.g. cysteine desulfurase) that catalyze the conversion of L-cysteine to L-alanine and sulfide via the formation of a protein-bound cysteine per sulfide in-termediate on a conserved cysteine residue (43,44). They are engaged in the biogenesis of iron-sulfur proteins. Due to their intrinsic redox properties, proteins with Fe/S clusters are ubiquitous in biological electron transport systems.
Cysteine Mutant Proteins: CD Spectra and Activity toward AMP-N-⑀-lysine and AMPS-To determine whether the desulfuration process occurs via the formation of a disulfide intermediate (path 1) with the participation of the sulfur atom of any of two cysteine residues present in the polypeptide chain of rHint-1, two single mutants, C38A and C84A, and one double mutant, C38A/C84A, were engineered. Although CD measurements showed that both single mutants had remarkably altered global secondary structures (the molar ellipticity decrease by ϳ20% in the 200 to 240 nm region of the spectrum) as compared with WT rHint-1 (Table 1, Fig. 4), their activities remained virtually unchanged. The recombinant proteins exerted phosphoramidase activity toward AMP-N-⑀-lysine and converted AMPS to AMP at rates comparable with those of the WT rHint-1 ( Table 2).
The data in Table 1 were calculated using the K2D algorithm (see "Experimental Procedures"), which deconvolutes the CD data. The results did not show any participation of the ␤-sheet in the rHint-1 structure. This finding is not compatible with the data derived from the crystal structure of the rHint-1 protein, where the presence of an antiparallel 10-stranded ␤-sheet has been demonstrated (26). Unfortunately, our samples could not be analyzed in a wavelength range of 185 to 200 nm, preventing the use of more accurate algorithms (e.g. Fasman-Johnson) for the secondary structure calculations of these proteins.
Designed by CD analysis, the secondary structure of the double mutant C38A/C84A was not very different from the   Fig. 4). Notably, it retained its activities for both substrates (AMPS and AMP-N-⑀-lysine), albeit at levels about 13 and 7 times lower, respectively, than those of WT rHint-1 and both single cysteine mutants (Table 2). However, this reduced activity is obscure because of small differences in the predicted structures of the three cysteine mutants designated on the base of the CD spectra.
Crystal Structures of rHint-1 Cysteine Mutants-To explain the discrepancy concerning the structures (similar for all three mutants) and low activity (toward both substrates) of the double cysteine C38A/C84A mutant, we also analyzed the crystal structures of the three cysteine recombinant proteins.
The coordinates and structure factor data have been submitted to the Protein Data Bank. The PDB identifiers are 3O1C for C38A, 3O1X for C84A, and 3O1Z for C38A/C84A. All datasets were collected to atomic resolution under cryoconditions. In two datasets (C38A and C84A), one molecule of adenosine per protein chain was localized in the binding cavity. This ligand was plausible because of its use in the purification of the protein by affinity chromatography. All residues have correct (, ) values in the most favored and additional favored regions of the Ramachandran plot. Many residues were observed in two or more side chain conformations, clearly defined at 2͉F o ͉ Ϫ ͉F c ͉ or ͉F o ͉ Ϫ ͉F c ͉ difference Fourier maps. As in the earliest published rHint-1 structures, the present data also show the 11 N-terminal residues are almost invisible on electron density maps because of high flexibility of this region of protein chain.
In search for explanations of the observed differences in activity toward AMP-N-⑀-lysine and AMPS among wild-type, single cysteine mutants, and the double cysteine mutant, a structural comparison was performed, based on the x-ray data. The root mean square differences were calculated for all C ␣ -atom positions in the WT rHint-1 monomer versus the three cysteine mutants. For C38A, C84A, and C38A/C84A components, root mean square difference values of 0.149, 0.111, and 0.173Å, respectively, were found, indicating close similarity between the analyzed structures, with negligible disorders of the polypeptide chain architecture. In addition, mutated residues did not play any important role in stabilizing the protein molecule structure, and they were located far from the ligand-binding cavity. Hint-1 protein gains enzymatic activity after dimerization (27). Coming from the second monomer, conserved Trp-123, which closes the binding site of the first monomer and interacts with the substrate across the dimer interface, does not change its position. The other amino acid locations and interactions between the two monomers are also identical in all three mutants. Moreover, the surface electrostatic potential remains unchanged, especially at the binding pocket. On the other hand, it is possible that water molecule interactions can be different for the double mutant because of the presence of two hydrophobic alanine residues instead of two cysteines. However, we were not able to observe any differences, probably because the water visibility on an electron density map depends on data resolution (worse for the double mutant, Table 3). Even so, in the

Hint-1 Transforms (d)NMPS to (d)NMP
DECEMBER 24, 2010 • VOLUME 285 • NUMBER 52 binding pocket of the C38A/C84A protein, water molecules are noticeable instead of the adenosine observed for both single mutants. The crystal structure comparison did not provide any hint as to the explanation for the lowered activity of the C38A/C84A mutant protein toward both substrates. However, this observation eliminated a mechanism of desulfuration, based on the formation of the disulfide intermediate and the contribution of the cysteines' methylene-mercapto residue in this reaction, because in all three cysteine mutants, the mutations changed the phosphoramidase and desulfurase activity to a similar extent. The above results allow us to argue for the participation of the same amino acids in the desulfuration reaction and in the phosphoramidase reaction, which indicates that at both activities, the histidine triad motif is engaged.
Histidine and Serine Mutant Proteins: Structure and Activity-Because the His-112-mutant protein was expected to be completely inactive in the hydrolysis process (20), the conserved third histidine was mutated to investigate participation of the HIT motif in the desulfuration process. Two histidine mutants, H114D and H114N, were constructed. The CD measurements showed that for both mutants, the molar ellipticity in the 200 -240 nm region was lowered by about 11%, compared with WT rHint-1 (Table 1, Fig. 4). The changes in the global secondary structures seem to be smaller than for cysteine mutants C38A, C84A, and C38A/C84A. However, the activities of mutants H114D and H114N toward both substrates were substantially reduced ( Table 2), up to 32 times toward AMP-N-⑀-lysine and 40 times to AMPS for Hint-H114D. For the Hint-H114N mutant, 8-and 5-fold reduction to AMP-N-⑀-lysine and AMPS, respectively, was observed. Such a low activity of the Hint-H114D mutant can be partially caused by a non-full homogeneous preparation (a small amount of impurities in the solution of the protein, Fig.  2). During the purification step, lower affinity of this mutant to the AMP resins compared to other studied proteins was noticed, and was revealed as additional proof of His-114 participation in the "holding" of the substrate during the hydrolysis. These results indicate that His-114 plays similar roles in phosphoramidase and desulfurase activities, which may suggest a common mechanism of action.
To test the hypothesis that Ser-107 plays a similar role in the desulfuration and phosphoramidase reactions, we used a rabbit S107A mutant protein and characterized its activity toward AMP-N-⑀-lysine and AMPS (Table 2). If the hydroxyl function of Ser-107 was similarly engaged in the binding and catalysis of both substrates in the same binding pocket, a reduction of both activities should be observed. Indeed, the S107A mutant exerted ϳ2 times reduced activity toward AMP-N-⑀-lysine and ϳ5 times toward AMPS in comparison to WT Hint-1. Thus, Ser-107 seems to serve as a hydrogen bond donor to both the sulfur atom in AMPS and the phosphoramidate nitrogen in the tested substrates.
Based on our results and crystal structure of rHint-1 (26,27), we propose a scheme for possible interactions between AMPS and amino acid residues engaged in hydrolysis (Fig. 5). The desulfuration of AMPS is expected to proceed through adenylated His-112 (Fig. 6). Presumably, the nitrogen atom of the imidazole group of His-114 forms a hydrogen bond with the bridging 5Ј-oxygen atom of the substrate (N-O distance ϳ0.312 nm, on the basis of the rHint-1 crystal structure from Ref. 27, PDB ID 1RZY or 0.307 nm, on the basis of the rHint-1 crystal structure from Ref. 26, PDB code 3RHN). Within this frame, only one orientation of the sulfur atom allows the reaction to occur because the position of the leaving group should be "in line" with the attacking His-112 residue (the first step). The distance requirement to form the hydrogen bond S⅐⅐⅐⅐H-O-Ser-107 (N-O distance ϳ0.32 nm (27) or 0.281 nm (26), where the nitrogen atom is replaced by a larger sulfur atom) can be fulfilled; thus, the sulfur atom, after protonation, becomes a good leaving group and hydrogen sulfide is released upon attack of His-112. The resulting intermediate product (with adenylated His-112) is identical to that of the phosphoramidate substrate, so the next hydrolytic step should be identical. Thus, we suggest that the mechanism of the "loss of sulfur" in AMPS is similar to cleavage of the P-N bond in AMP-N-⑀-lysine and involves two discrete steps, with participation of Ser-107 (rHint-1) acting as the acid-base catalyst (Fig. 6).
Kinetic Parameters of the AMPS Hydrolysis-Kinetic parameters (k cat and K m ) were determined for the rHint-1-catalyzed desulfuration of AMPS and compared with those for the hydrolysis of AMP-pNA and AMP-N-⑀-lysine (27). The k cat /K m ratio for the desulfuration (81.0 M Ϫ1 s Ϫ1 ) was more than 5 times higher than for the hydrolysis of AMP-pNA (14.0 M Ϫ1 s Ϫ1 ). On the other hand, Hint-1 converted AMPS to AMP with 500 times higher K m and 10 times lower k cat compared with AMP-N-⑀-lysine (Table 4). The higher K m values are reasonable because sulfur has a larger atomic radius and lower electronegativity than oxygen; thus, its H-bonding ability is weaker relative to the latter, and therefore, AMPS demonstrates weaker affinity to the enzyme than AMP-N-⑀-lysine and AMP-pNA.
NMPS and dNMPS as the Substrates for Hint-1-We have found that the desulfuration process is not specific toward AMPS. Also, treatment of the 5Ј-O-phosphorothioylated derivatives of guanosine, cytidine, and uridine as well as those of deoxyadenosine, deoxyguanosine, deoxycytidine, and thymidine with rHint-1 leads to the corresponding 5Ј-O-phosphates. However, the cleavage rate of the P-S bond in deoxyribonucleotide phosphorothioates is lower than for the corresponding ribonucleotide analogs ( Table 5). The data indicate that the rates of desulfuration decrease in the following order: It has been demonstrated for the phosphoramidase activity of Hint-1 that the enzyme prefers purine-based substrates rather than pyrimidines and that a hydroxyl group at the ribose 2Ј position of the substrate is essential for interactions with the ␥-oxygen of Asp-43 (rHint-1) of the enzyme in the binding pocket (45). We observed a similar tendency for rHint-1 desulfurating activity. However, our results show that the purine deoxyribose derivative, dAMPS, is a worse substrate than both pyrimidine ribose analogs, CMPS and UMPS. On the other hand, we noticed that dGMPS was an equally good substrate as CMPS. These data indicate that the preference of the enzyme substrate pocket to purine residues (such as dGMPS) is balanced by the presence of the 2Ј-OH group in CMPS, but the affinity to purines in dAMPS appears less im-portant than the influence of 2Ј-OH group of the ribose pyrimidine analogs. All desulfuration reactions catalyzed by rHint-1 occurred at greater rates than the spontaneous loss of sulfur from AMPS (Table 5).

Hint-1 Transforms (d)NMPS to (d)NMP
We also examined whether phosphate diester derivatives could be the substrates for Hint-1. In these reactions, neither AMPS(OMe) nor dinucleotide d(ApsA) were hydrolyzed by the enzyme (Table 5), and no products of desulfuration were observed.
We found that the Fhit protein, another enzyme of the HIT superfamily, which usually acts as an Ap 3 A and Ap 4 A hydrolase (19), can also desulfurate AMPS, although at a rate ϳ50  The atomic distances (Å) for the rHint-1 crystal structure are taken from PDB codes 3RHN (26) and 1RZY (27).