Chemical modification and site-directed mutagenesis of cysteine residues in human placental S-adenosylhomocysteine hydrolase.

Human placental S-adenosylhomocysteine (AdoHcy) hydrolase (EC 3.3.1.1) was inactivated by 5′,5-dithiobis(2-nitrobenzoic acid) following pseudo-first-order kinetics. Modification of three of the 10 cysteine residues per enzyme subunit resulted in complete inactivation of the enzyme. The three modified cysteine residues were identified as Cys113, Cys195, and Cys421, respectively, by protein sequencing after modification with [1-14C]iodoacetamide. Of the three modifiable cysteines, Cys113 and Cys195 could be protected from modification in the presence of the substrate adenosine (Ado), which also protected the enzyme from inactivation. On the other hand, Cys421 was not protected by Ado, and modification of Cys421 alone did not affect the enzyme activity. To verify whether some of these cysteine residues are important for the enzyme catalysis, these three cysteine residues were replaced by either serine or aspartic acid using site-directed mutagenesis. Mutants of both Cys113 (C113S and C113D) and Cys421 (C421S and C421D) had enzyme activities similar to that of the wild-type enzyme, and only slight changes were observed in the steady-state kinetics measured in both the synthetic and hydrolytic directions. However, mutants of Cys195 (C195D and C195S) displayed drastically reduced enzyme activities, and kcat values were only 7 and 12% of that of the wild-type enzyme, respectively, resulting in a calculated loss in binding energy (ΔΔG) of approximate 1 Kcal/mol. The Cys195 mutants were capable of catalyzing both the 3′-oxidative and 5′-hydrolytic reactions, as evidenced by the reduction of E·NAD+ to NADH and formation of the 5′-hydrolytic product when incubated with (E)-5′,6′-didehydro-6′-deoxy-6′-chlorohomoadenosine at rates comparable with those catalyzed by the wild-type enzyme. However, mutations of the Cys195 severely altered the 3′-reduction potential as evidenced by the drastic reduction in the rate of [2,8-3H]Ado release from the E−NADH·[2,8-3H]3′-keto-Ado complex. Circular dichroism studies of the Cys195 mutants confirmed that the observed effects are not due to changes in secondary structure. These results suggested that the Cys195 is involved in the catalytic center and may play an important role in maintaining the 3′-reduction potential for effective release of the reaction products and regeneration of the active form (NAD+ form) of the enzyme; the Cys113 is located in or near the substrate binding site, but plays no role beneficial to the catalysis; and the Cys421 is a nonessential residue, which also explains why Cys421 does not occur in any other known AdoHcy hydrolases.

AdoHcy hydrolase has been cloned from a number of different sources, including Homo sapiens (9), Rattus species (10), Plasmodium falciparum (11), Rhodobacter capsulatus (12), Triticum aestivum, 2 Catharanthus roseus, 3 Petroselinum crispum (15), Leishmania donovani (16), Dictyostelium discoicrispum (17), and Caenorhabditis elegans (18). The amino acid sequences of the cloned AdoHcy hydrolases have been deduced from their cDNA sequences. Comparison of the amino acid sequences from these species shows a remarkable degree of conservation ranging from 64% identity between human and Rhodobacter capsulatus (12) to 97% identity between human and rat (9). All of the cloned AdoHcy hydrolases are tetramers with M r values between 180,000 and 200,000. AdoHcy hydrolase contains four tightly bound molecules of NAD ϩ and consists of structurally identical subunits that are catalytically equivalent and functionally independent (19).
The mechanism by which AdoHcy hydrolase catalyzes the conversion of AdoHcy to Ado and Hcy, as well as the reverse reaction, has been elucidated by Palmer and Abeles (20). In the hydrolytic direction, the first step involves oxidation of the 3Ј-hydroxyl group of AdoHcy (3Ј-oxidative activity) by enzymebound NAD ϩ (E⅐NAD ϩ ), followed by ␤-elimination of L-Hcy to give 3Ј-keto-4Ј,5Ј-didehydro-5Ј-deoxy-Ado. Michael addition of water to the 5Ј-position of this tightly bound intermediate (5Ј-hydrolytic activity) affords 3Ј-keto-Ado, which is then reduced by enzyme-bound NADH (E⅐NADH) to Ado (3Ј-reduction activity). In this case, the 5Ј-hydrolytic activity depends upon the 3Ј-oxidative activity.
In contrast to the extensive studies on mechanisms of the enzyme catalysis and the enzyme inactivation, little is known about the structural features of the active site of AdoHcy hydrolase. To date, attempts at crystallization of this enzyme for x-ray studies have been unsuccessful. Therefore, efforts to identify residues in the active site of AdoHcy hydrolase have been made by several laboratories using alternative approaches, including chemical modification (25)(26)(27)(28), affinity labeling (29,30), limited proteolytic digestion (31), and sitedirected mutagenesis (24,32). Accumulated information from these approaches will provide valuable insights into the structural features of the enzyme needed for catalysis and inactivation and may ultimately lead to the rational design of more potent inhibitors of AdoHcy hydrolase.
In this study, we have used a combination of chemical modification and site-directed mutagenesis to identify a cysteine residue that is critical to the catalytic function of human placental AdoHcy hydrolase and to elucidate its possible role(s) in the mechanism of the enzyme action. Overexpression and Purification of Wild-type AdoHcy Hydrolase-Overexpression and purification of wild-type AdoHcy hydrolase was carried out in the procedures described previously (24,33). Briefly, E. coli JM109 carrying the expression vector pPROKcd20 for recombinant human placental AdoHcy hydrolase were grown in 2 ϫ YT medium and induced with 1 mM IPTG. The cell-free extract was used for the enzyme purification through the following steps: DEAE-cellulose ion exchange (batch method), ammonium sulfate precipitation (60%), gel filtration on Sephacryl S-300 column, and ion exchange on DEAE-Sepharose column. The protein concentration was determined by the method of Bradford (34) using bovine serum albumin as a standard, and the subunit M r was used to calculate the molarity of enzyme solutions.

Materials
Assays for Enzyme Activities-The assay of AdoHcy hydrolase activity in the synthetic direction was performed as described previously (21). This assay measures the rate of formation of AdoHcy from Ado and Hcy using HPLC. One unit of enzyme activity is defined as the amount of enzyme that can synthesize 1 mol of AdoHcy/min.
The assay of AdoHcy hydrolase activity in the hydrolytic direction was performed spectroscopically by measuring the rate of the product (Hcy) formed by reaction with DTNB. To 800 l of the enzyme solution (800 l) containing 4.7 g of AdoHcy hydrolase and 4 units of Ado deaminase in 50 mM potassium phosphate buffer, pH 7.2, containing 1 mM EDTA (buffer A) was added 200 l of AdoHcy (500 M) containing 250 M DTNB in buffer A. The reaction mixture was maintained at 37°C for 2 min, and it was monitored at 412 nm continuously using an HP 8452 diode array UV spectrophotometer (Hewlett-Packard Co., Palo Alto, CA). The initial rate was obtained by fitting the data to zero-order kinetics, and an extinction coefficient of 13,600 M Ϫ1 cm Ϫ1 for 5-thio-2nitrobenzoate (TNB), the product of the DTNB reaction, was used to calculate the amount of Hcy formed (35). The blank was taken against a sample containing all other components except the enzyme. One unit of the enzyme activity is defined as the amount of enzyme that can hydrolyze 1 mol of AdoHcy/min.
The assay of the 5Ј-hydrolytic activity of AdoHcy hydrolase toward EDDClHA was conducted by mixing 150 l of enzyme solution (14.2 M) in buffer A with 50 l of 800 M EDDClHA. The reaction mixture was incubated at 37°C for various times, and the reaction was terminated by addition of 15 l of 5 N HClO 4 . After centrifugation, the supernatant was analyzed for the 5Ј-hydrolytic reaction product 6Ј-deoxy-6Ј-chloro-5Ј-hydroxyhomoAdo (DClHHA) by HPLC using a C18 reverse-phase column as described previously (23). The data were fitted to Equation 1, where P is the product DClHHA; k 5Ј is the rate constant of the 5Јhydrolytic reaction; k 3Ј is the rate constant of the enzyme inactivation, which is equal to the rate constant of the 3Ј-oxidation reaction; E 0 is the concentration of initial enzyme; and k app is the apparent rate constant of enzyme inactivation at this substrate concentration. Details of this equation have been reported previously by our laboratory (23). The assay of the 3Ј-oxidative activity toward EDDClHA was performed under the same conditions as that for the 5Ј-hydrolytic activity assay except that the reaction was stopped by addition of 3 volumes of 95% ethanol to release NAD ϩ /NADH from the enzyme. After centrifugation, the supernatant was analyzed for the remaining NAD ϩ contents by HPLC (30). Data were fitted to an exponential decay equation using the Ultrafit fitting program (Ultrafit, Cambridge, United Kingdom).
Enzyme Inactivation by DTNB-DTNB stock solution (10 mM) was prepared in 0.1 M sodium phosphate buffer, pH 7.0. AdoHcy hydrolase (1.5 M) was incubated with various concentrations of DTNB in a cuvette containing 3 ml of 0.1 M sodium phosphate buffer, pH 8.0, and 1 mM EDTA at 25°C. At intervals, an aliquot (30 l) was removed from the reaction mixture and added to 470 l of the enzyme assay mixture to measure the remaining enzyme activity in the synthetic direction. Controls containing no modification reagent were run concurrently at any given time, and the residual activity was calculated relative to the appropriate control.
The number of DTNB-modified cysteine residues was determined by measuring the absorbance at 412 nm on an HP 8452 diode array spectrophotometer using an extinction coefficient of 13,600 M Ϫ1 cm Ϫ1 for TNB.
Incorporation of [1-14 C]Iodoacetamide into AdoHcy Hydrolase-For specific modification of cysteine residues located in the substrate binding site, the enzyme (1.5 mg) was dissolved in 0.8 ml of 0.1 M sodium phosphate buffer, pH 8.0, containing 1 mM EDTA. To the enzyme solution was added 0.4 ml of 5 mM adenine (Ade), and the mixture was incubated at 25°C for 10 min. DTNB was then added to a final concentration of 50 M, and the reaction was continued in the dark. The extent of modification was monitored at 412 nm, and the reaction was stopped by gel filtration on a Sephadex G-50 spin column equilibrated with 0.1 M Tris-HCl buffer, pH 8.9, to remove excess DTNB when approximately one cysteine residue per subunit was modified. After being dialyzed against Tris-HCl buffer for 5 h to remove Ade, the DTNB-modified enzyme (1.2 ml) was mixed with 250 l of 10.2 mM [1-14 C]iodoacetamide (1 mCi/mmol). The reaction was carried out at 25°C in the dark for 1 h and was terminated by passing the reaction mixture through the spin column. The radioactivity incorporated into the enzyme was determined by liquid scintillation counting using a complete counting mixture 3a70B (Research Products International Corp., Mount Prospect, IL) after the modified protein was denatured with 8 M urea and passed through the spin column, which had been equilibrated with 1 M urea in the buffer A.
For specific modification of cysteine residues located outside of the substrate binding site, AdoHcy hydrolase (1.5 mg) was dissolved in 0.5 ml of buffer A containing 700 M Ado. After incubation at 37°C for 10 min, the mixture was transferred to 0.5 ml of 0.2 M Tris-HCl buffer, pH 8.9. To the Ado-protected enzyme was then added 200 l of 10.2 mM [1-14 C]iodoacetamide, and the mixture was incubated at 25°C in the dark for 30 min. The reaction was then terminated, and the radioactivity incorporated was determined as described above.
Isolation and Identification of [1][2][3][4][5][6][7][8][9][10][11][12][13][14] C]Carboxymethylated Peptides-The specific radiolabeled AdoHcy hydrolases obtained from the procedures described above were heated at 100°C for 3 min in the presence of 8 M urea. To the denatured protein, dithiothreitol was added to a final concentration of 1 mM and incubated at 50°C for 10 min. After cooling to room temperature, the protein was incubated with unlabeled iodoacetamide (10.2 mM) at 25°C for 15 min. The urea concentration was then reduced to 1 M by passing through the spin column equilibrated with 0.4 M NH 4 HCO 3 , pH 8.4, containing 1 M urea. A freshly prepared solution of TPCK-trypsin was added to give an enzyme-to-substrate ratio of 1:50 (by weight). After incubation for 5 h at 37°C, a second addition of the same amount of trypsin was made and the digestion was continued for another 5 h. After lyophilization, the trypsin digest was dissolved in 0.1% trifluoroacetic acid, and the peptides were analyzed on a Vydac C18 protein and peptide column (Vydac 218TP54, C18, 5 m, 0.46ϫ25 cm). The solvent system consisted of solvent I (0.1% trifluoroacetic acid) and solvent II (80% CH 3 CN, 20% H 2 O, 0.07% trifluoroacetic acid). The initial conditions were 2% solvent II with a linear gradient to 70% solvent II over 120 min at a flow rate of 0.5 ml/min. The UV absorbance of the eluted peptides was monitored at 220 nm. The radioactivity in the fractions collected (0.5 ml) was measured by liquid scintillation counting. Peptide peaks containing major radioactivity were collected and concentrated by Speed-Vac and rechromatographed on the same column using the initial conditions of 20% solvent II with a linear gradient to 60% solvent II over 60 min or 10% solvent II with a linear gradient to 40% solvent II over 60 min. Detection of peptides and measurement of radioactivity were the same as described above except that fractions were collected manually peak by peak.
The isolated peptides were sequenced by automated Edman degradation on an Applied Biosystems 473A protein sequencer in the Biotechnology Laboratory at Kansas State University, Manhattan, KS. At each sequencing cycle, the washings from the conversion flask and eluate from the HPLC column were collected and combined for determination of radioactivity.
Site-directed Mutagenesis-The EcoRI fragment from the pPROKcd20 plasmid was subcloned into M13mp19, and the mutants were generated in this vector using Amersham's Sculptor™ in vitro mutagenesis kit, which is based on the method of Taylor et al. (36), exploiting the inability of NciI to cleave a thio-containing DNA strand. The oligonucleotides used for generating the mutants were so designed that at each mutation site, specific nucleotide changes were made to create desired codons for specific amino acid. This in vitro mutated double-stranded DNA was then transformed into competent E. coli TG1 cells, and phage plaques were screened by DNA sequencing to identify phages containing the correct mutation. The EcoRI DNA fragment encoding the mutated AdoHcy hydrolase gene was subcloned into the EcoRI site of pPROK-1 (Clontech, Palo Alto, CA) expression vector and transformed into competent E. coli JM109 cells. Transformed cells were selected against 100 g/ml ampicillin. Ampicillin-resistant clones were grown in 10 ml of 2 ϫ YT medium containing 35 g/ml ampicillin for 2 h at 37°C and then induced with 1 mM IPTG overnight. Cell-free extracts were analyzed by SDS-polyacrylamide gel electrophoresis to identify clones that inducibly overproduced a protein band at M r of about 45,000.
Overexpression and purification of mutant forms of AdoHcy hydrolase were carried out in the same conditions as that for the wild-type enzyme as described above.
Steady-state Kinetics-Kinetic constants were determined in both the synthetic and hydrolytic directions. The initial velocities for the enzyme reactions were measured using the methods as described in the enzyme activity assay section, except shorter reaction time (30 s) was used in the synthetic direction. For mutant enzymes of low activities, 4 to 10 times higher protein concentrations were used than that for the wild-type enzyme. Data were fitted to the Michaelis-Menten equation to obtain k cat and K m values using a nonlinear least squares fitting program (Ultrafit) run on a Macintosh computer. Changes in free energy (⌬⌬G) for mutated enzymes were calculated using Equation 2 (37) (as follows).
Determination of E⅐NAD ϩ and E⅐NADH-E⅐NAD ϩ and E⅐NADH contents of wild-type and mutant AdoHcy hydrolases were determined by an HPLC method as described previously (30).
Circular Dichroism Spectra-Circular dichroism (CD) spectra were recorded at 13°C using a AVIV-60DS spectropolarimeter (AVIV Associates Inc. Lakewood, NJ) equipped with a data processing system. Measurements were made with a cylindrical quartz cell with a path length of 0.1 cm. The concentration of the protein samples was 10 M in buffer A. Five scans from 200 to 250 nm were recorded in 0.5-nm intervals for each sample. The scans for each sample were then averaged and corrected by subtracting a buffer base line. The relative percentages of ␣-helix, ␤-sheet, ␤-turn, and random coil structures were estimated using the AVIV version 3.1 computer program.  [2, H]3Ј-keto-Ado concomitantly with the reduction of E⅐NAD ϩ to E⅐NADH based on the mechanism of the enzyme action (20) and our recent experimental observations that after denaturation of the wild-type enzyme incubated with Ado, the major reaction product released was 3Ј-keto-Ado (data not shown). The protein concentration and radioactivity of the E -NADH ⅐[2,8-3 H]3Ј-keto-Ado complex were determined in order to calculate the stoichiometry of [2, H]Ado bound to AdoHcy hydrolase. The complex (ϳ1 ml) was then dialyzed against a large volume (500 ml) of buffer A at 4°C. At various time intervals, two samples (30 l each) were taken from the dialysis tubing to measure the protein concentration and radioactivity. The dialysis buffer was changed to new buffer A at each sampling time. The apparent rate constant for the 3Ј-reduction from E -NADH ⅐3Ј-keto-Ado to E -NAD ϩ and Ado was obtained by plotting the log of the percent [2, H]Ado bound versus the dialysis time, and the data were fitted to a linear least squares equation.

Kinetics of AdoHcy Hydrolase Inactivation by DTNB-Incu-
bation of AdoHcy hydrolase with DTNB in 0.1 M phosphate buffer, pH 8.0, at 25°C resulted in a time-dependent loss of enzyme activity, and the inactivation followed pseudo-first order kinetics (Fig. 1). After incubation for 50 min with 10 M DTNB, about a 98% loss of enzyme activity was observed. The double logarithmic plot of the apparent first order rate constants (k app ) versus DTNB concentrations yielded a straight line with a slope of 1.73, which represents the number of TNB bound per active site of the enzyme. When residual enzyme activity was plotted against the number of cysteine residues modified, it was shown that three cysteine residues per subunit were modified when the enzyme was completely inactivated (Fig. 2). The inactivation of AdoHcy hydrolase by DTNB was prevented in the presence of the substrate Ado or the competitive inhibitor Ade. Because of its tight binding with the enzyme, Ado was found to be an extremely strong protector against modification of the enzyme by DTNB. As shown in Fig. 2, in the presence of Ado (6.3-fold molar excess over the enzyme), only about one cysteine residue could be modified and the enzyme activity was almost completely protected when measured in the synthetic direction. This result indicated that the two cysteine residues protected by Ado may be located in or near the Ado binding site and could be essential for the enzyme activity. On the other hand, the cysteine residue that was not protected by Ado may be located outside of the Ado binding site and may not essential for the enzyme activity. Enzyme inactivation by DTNB could also be prevented by Ade, although higher concentrations of Ade (500 M) were required to achieve the same protective effect as that by Ado.
Isolation and Characterization of [1-14 C]Iodoacetamide-mod-ified Peptide Fragments of Tryptic Digestion-The enzymes that had been specifically modified with [1-14 C]iodoacetamide for the two cysteines located in Ado binding site and the one located outside of the Ado binding site were subjected to tryptic digestion. Radiolabeled peptides were separated by reversephase HPLC. Fig. 4a shows the HPLC chromatogram from the enzyme specifically modified for the two cysteine residues in the Ado binding site. About 60% of the radioactivity was recovered from two major fractions (a and b), which were in an approximately 1:1 ratio. These two radiolabeled fractions were collected and rechromatographed on the same column with different buffer conditions. As shown in Fig. 4a, insets I and II, fractions a and b contained one major and several minor components with the radioactivity associated only with the major component (fractions aЈ and bЈ). Similarly, one major fraction, which accounts for about 70% of the radioactivity, was observed from the tryptic digested enzyme modified for the cysteine residue located outside of the Ado binding site (Fig. 4b). Rechromatography of this major fraction generated several components, but only a single peak contained radioactivity (fraction cЈ) (Fig. 4b, inset). The HPLC-purified radiolabeled peptides (approximately 50 pmol of fraction aЈ, 60 pmol of fraction bЈ, and 55 pmol of fraction cЈ) were subjected to N-terminal amino acid sequencing; the results are shown in Table I. Neither fraction aЈ nor bЈ was found to be the complete tryptic digestion fragment since each contained one non-C-terminal lysine residue. In contrast, fraction cЈ was a complete tryptic fragment. Residues of Scarboxymethylcysteine (Cys(Cm)) in the sequences were identified as Cys 113 , Cys 195 , and Cys 421 , respectively, for fractions aЈ, bЈ, and cЈ based on the radioactivity counted at each sequencing cycle and the primary sequence deduced from the cDNA encoding the human placental AdoHcy hydrolase (9).
Site-directed Mutagenesis-Cysteine residues (Cys 113 , , at 25°C in the dark for different times. At the indicated times, aliquots (50 and 3 l) of the reaction mixture were removed for determination of stoichiometry of covalent binding and activity remaining. For stoichiometry of covalent binding, the modified enzyme was passed through a spin column (Sephadex G-50, 3 ml) to remove free [1-14 C]iodoacetamide. The protein was then denatured with 8 M urea and passed through another spin column equilibrated with 1 M urea to remove any noncovalently bound [1-14 C]iodoacetamide. The radioactivity and protein concentration were determined as described under the "Experimental Procedures." For the remaining activity assay, the modified enzyme was mixed with 500 l of buffer A containing 100 M Ado and 5 mM Hcy as described previously (21). Ç, remaining activity of the native enzyme; å, remaining activity of DTNB-modified Ade-protected enzyme.
Cys 195 , and Cys 421 ) identified by chemical modification were mutated to both serine (structurally conservative to cysteine) and aspartic acid (introducing a negative charge as DTNB does). All six mutants (C113S, C113D, C195S, C195D, C421S, and C421D) were overexpressed and purified by the same procedures used for the wild-type enzyme. No significant differences were observed in the levels of mutant enzyme expression or in the amount of mutant enzymes recovered in the purification compared with wild-type enzyme. One liter of the cell culture normally yielded 50 -60 mg of pure AdoHcy hydrolase (wild-type or mutant enzymes) in our laboratory. All the mutant enzymes had the same subunit size and tetramer structure as the wild-type enzyme as determined by SDS-polyacryl-amide gel electrophoresis and chromatography on Superose 12 gel filtration column in a fast protein liquid chromatography system (data not shown). E⅐NAD ϩ and E⅐NADH contents of mutant enzymes were similar to those of the wild-type enzyme, having approximately 0.8 mol of NAD ϩ and 0.2 mol of NADH/ mol of enzyme subunit except for C195D, which had less than a stoichiometric amount (0.6 mol of NAD ϩ and 0.1 mol of NADH/mol of subunit).
Steady-state Kinetic Properties of the Mutant Enzymes-Table II shows kinetic parameters of mutant and wild-type enzymes and changes in binding energy (⌬⌬G) of mutant enzymes. Michaelis-Menten constants were determined toward all three variable substrates: AdoHcy, Ado, and Hcy. Both Cys 113 mutants and Cys 421 mutants had only slightly increased K m values and slightly decreased k cat values toward all the three substrates compared with the K m and k cat values of the wild-type enzyme, indicating that mutating either Cys 113 or Cys 421 does not provoke any major changes in ground state binding and subsequent catalytic steps. In contrast, mutation of Cys 195 to either serine or aspartic acid led to large reductions in turnover numbers toward all three substrates. The k cat values of the C195D and C195S were 14-and 9-fold lower than those of the wild-type enzyme while there were relatively small decreases (1-3-fold) in K m values. Therefore, the catalytic efficiencies (k cat /K m ) for the three substrates were approximately 7-11-fold lower for the C195D, and 3-8-fold lower for the C195S than those of the wild-type enzyme, which results in a calculated loss in binding energy (⌬⌬G) of approximately 0.7-1.4 Kcal/mol (Table II). This indicates that the drastic losses of overall catalytic activities (93% loss for C195D and 88% loss for C195S) of the Cys 195 mutants are due mainly to effects on catalytic steps. The low k cat values of the Cys 195 mutants may have contributed to their relatively lower K m values. In addition, these losses of enzyme activity of the Cys 195 mutants were apparently not correlated to changes in the secondary structures of the enzymes, since there were no significant differences in the secondary structures between the mutants and the wild-type enzyme based on CD analysis (wild-type, ␣-helix ϭ 23.4%, ␤-sheet ϭ 16.7%, ␤-turn ϭ 32.6%, random coils ϭ 27.3%; C195S, ␣-helix ϭ 23.4%, ␤-sheet ϭ 16.6%, ␤-turn ϭ 33.3%, FIG. 4. HPLC tryptic peptide map of the [1-14 C]iodoacetamide-modified AdoHcy hydrolase. The tryptic-digested enzyme was applied to a Vydac C18 protein and peptide column (Vydac 218TP54, 18, 5 m, 250 ϫ 46 mm). Elution was carried out with a gradient of acetonitrile in 0.1% trifluoroacetic acid from 2 to 70% in 120 min with a flow rate of 0.5 ml/min. a, tryptic-digested enzyme specifically modified for cysteine residues located in the Ado binding site; b, tryptic-digested enzyme specifically modified for the cysteine residue located out side of the Ado binding site as described under "Experimental Procedures." Insets I and II in a, and the inset in b, were profiles of rechromatography of fractions a, b, and c, respectively, with a gradient of acetonitrile in 0.1% trifluoroacetic acid from 20 to 60% in 60 min for fractions a and b and with a gradient of acetonitrile in 0.1% trifluoroacetic acid from 10 to 40% in 60 min for fraction c. Shaded peaks represent radioactive peaks. random coils ϭ 26.7%; C195D, ␣-helix ϭ 25.0%, ␤-sheet ϭ 12.6%, ␤-turn ϭ 35.2%, random coils ϭ 27.2%).

Effects of the Cys 195
Mutations on the 3Ј-Oxidative and 5Ј-Hydrolytic Activities-In the hydrolytic direction, the overall catalytic activity of AdoHcy hydrolase requires two major sequential steps of catalysis: the 3Ј-oxidative activity and the 5Ј-hydrolytic activity. The 3Ј-oxidative activity can be measured by determining the rate of E⅐NAD ϩ to E⅐NADH conversion, and the 5Ј-hydrolytic activity can be measured independently from the 3Ј-oxidation by determining the rate of hydrolytic product formation using EDDClHA as a substrate (23). As seen in Fig. 5, both C195D and C195S showed slightly faster 3Ј-oxidation rates than that of the wild-type enzyme with apparent k 3 Ј values of 0.11, 0.13, and 0.15 min Ϫ1 for the wildtype, C195S, and C195D, respectively. However, due to the protection from the largely formed 5Ј-hydrolytic products (DClHHA and Ade) (23), the conversion of E⅐NAD ϩ to E⅐NADH by EDDClHA was not complete as shwon in Fig. 5.
The 5Ј-hydrolytic activities of the Cys 195 mutants and the wild-type enzyme were also shown to be similar, as seen in Fig.  6. From equation (1) and the apparent 3Ј-oxidation rate constants (k 3 Ј), the apparent 5Ј-hydrolytic rate constants (k 5 Ј) were calculated to be 0.72, 0.74 and 0.69 min Ϫ1 for the wildtype, C195S and C195D, respectively.
Effects of Cys 195 Mutations on the Rate of 3Ј-Reduction Reaction-Since the 3Ј-oxidative and 5Ј-hydrolytic reactions appeared not to be responsible for the severe loss of enzyme activity of the Cys 195 mutants, we focused our attention on the 3Ј-reduction step, which is the last chemical reaction step in both the synthetic and hydrolytic directions. The 3Ј-reduction reaction converts the tightly bound intermediates 3Ј-keto-Ado or 3Ј-keto-AdoHcy to the final reaction products Ado or AdoHcy concomitantly with the oxidation of E⅐NADH to E⅐NAD ϩ to regenerate the active enzyme (NAD ϩ form) for the next catalytic cycle. Comparison of the 3Ј-reduction rates between the Cys 195 mutants and the wild-type enzyme was accomplished by monitoring the rates of Ado release from the E -NADH ⅐3Ј-keto-

TABLE II Kinetic constants of wild-type and mutant AdoHcy hydrolases
The kinetic constants were determined in 50 mM potassium phosphate buffer, pH 7.2, containing 1 mM EDTA at 37°C with variable substrate concentrations as described under "Experimental procedures." Data were the average of two determinations, and the K m and k cat values were obtained by directly fitting the data into the Michaelis-Menten equation using a computer program. Free energy changes (⌬⌬G) were calculated from the equation: ⌬⌬G ϭ ϪRTln((k cat /k m )mutant/(k cat /k m )wild-type) (  Ado complex in a dialysis system. As the dissociation of Ado from E -NAD ϩ ⅐Ado is a fast step with a rate constant of 2 s Ϫ1 for wild-type enzyme (38), the measurement of the rate of Ado release from E -NADH ⅐3Ј-keto-Ado complex is essentially a measurement of the rate of the 3Ј-reduction (rate-determining step). As shown in Fig. 7, the apparent 3Ј-reduction rate constants for C195S and C195D were found to be 3.2 ϫ 10 Ϫ6 s Ϫ1 and 2.5 ϫ 10 Ϫ6 s Ϫ1 , respectively, which were 5.1-and 6.5-fold slower than that of the wild-type enzyme (16.3 ϫ 10 Ϫ6 s Ϫ1 ). It should be noted that the values of the apparent 3Ј-reduction rate constants obtained in this system were not the true values of the effective 3Ј-reduction rate constants for the enzymes during catalysis, since there was no substrate competition in the dialysis system. However, these apparent rate constants serve the purpose for comparison of the 3Ј-reduction rates between the mutants and the wild-type enzyme under the same conditions. DISCUSSION Due to lack of an x-ray crystal structure of AdoHcy hydrolase, little is known about the amino acid residues involved in the substrate binding and/or catalysis of the enzyme. In this study, we have utilized the thiol-specific reagents DTNB and iodoacetamide to probe the active site of AdoHcy hydrolase and have employed site-directed mutagenesis to characterize the cysteine mutants identified by chemical modification. From these studies, it is concluded that Cys 195 is important for the enzyme to exert its full catalytic activity.
There are 10 cysteine residues per subunit of human placental AdoHcy hydrolase. Incubation of the enzyme with DTNB or iodoacetamide resulted in complete loss of enzyme activity when three cysteine residues (Cys 113 , Cys 195 , and Cys 421 ) were modified. Kinetic studies on enzyme inactivation by DTNB showed that the loss of enzyme activity was due to modification of two (Cys 113 and Cys 195 ) of the three modifiable cysteine residues. Protection of these two cysteine residues with the substrate Ado or the competitive inhibitor Ade led to protection of the enzyme from inactivation, suggesting that Cys 113 and Cys 195 may be located in or near the enzyme active site. These results are consistent with earlier results, which indicated that Cys 112 in the rat liver AdoHcy hydrolase (29), comparable with the Cys 113 in human enzyme, and Glu 197 in human AdoHcy hydrolase (31), only two residues away from the Cys 195 , are located in or near the active site of AdoHcy hydrolase. Modification of Cys 421 alone did not effect the enzyme activity, indicating that Cys 421 is a nonessential residue; this is also supported by the site-directed mutagenesis results, since replacement of Cys 421 with other amino acids did not significantly effect the enzymatic activity. This also explains why Cys 421 does not occur in any other AdoHcy hydrolase including rat liver AdoHcy hydrolase, which is 97% homologous to the human enzyme (9,11).
Although the protection study showed that Cys 113 may be located in or near the Ado binding site, results from sitedirected mutagenesis demonstrated that Cys 113 plays no significant role in either substrate binding or catalysis, since mutations of this residue with serine and aspartic acid had little effects on the enzyme activity, which is consistent with mutational observations of Cys 112 in the rat AdoHcy hydrolase (32).
In contrast, mutations of the Cys 195 with serine or aspartic acid resulted in drastic loss of enzyme activity. This loss of activity is concluded to be mainly due to the perturbation of the 3Ј-reduction potential necessary for efficient release of the reaction product and regeneration of the active form (NAD ϩ form) of the enzyme. Direct evidence that supports this conclusion is that the rates of Ado release from the E -NADH ⅐3Ј-keto-Ado complex of the Cys 195 mutants are 5-6.5-fold slower than that of the wild-type enzyme. It seems that the decrease in the 3Ј-reduction rate alone does not account for the overall losses of the enzyme activities (9 -14-fold), especially for the C195D mutant. However, when the NAD ϩ content of the C195D mutant, which is 30% less than that of the wild-type enzyme, is taken into account, then the loss of enzyme activity of the C195D mutant is only 9.8-fold, which is quite close to its loss of the 3Ј-reduction rate. Other evidence that indirectly supports the conclusion includes: (i) the Cys 195 mutants have slightly higher 3Ј-oxidation activities and similar 5Ј-hydrolytic activities compared with the wild-type enzyme, suggesting that the loss of enzyme activity of the Cys 195 mutants is not caused by dysfunction of these two major catalytic steps, and (ii) the secondary structures of the Cys 195 mutant and the wild-type enzyme are basically the same, indicating that the loss of enzyme activity of the Cys 195 mutant is not correlated to alternations in secondary structures. Exclusion of these two inactivation possibilities led us to focusing on the 3Ј-reduction reaction.
The Cys 195 residue may be involved in maintaining the 3Јreduction potential by forming a hydrogen bond with the 3Ј-CAO or 3Ј-OH of the ribose ring of 3Ј-keto-Ado or Ado, respectively. It has been reported that, based on an x-ray crystal structure, the OSH group of Cys 35 in tyrosyl-tRNA synthetase forms a hydrogen bond with the 3Ј-OH of the ribose ring of ATP (37). Mutation of this Cys 35 to serine results in significant loss of enzyme activity and a calculated loss of free energy (⌬⌬G) of approximately 1 Kcal/mol, which is equal to the value of the binding energy of the hydrogen bond between the OSH group and the 3Ј-OH of ATP (37). Similar binding energy losses (ϳ1.2 Kcal/mol) were observed for the Cys 195 mutants, indicating that mutation of Cys 195 may lead to a loss of a hydrogen bond between the OSH group and the 3Ј-CAO of the ribose ring of 3Ј-keto-Ado. It is not surprising that loss of this hydrogen bond impairs the 3Ј-reduction reaction, since hydrogen bond formation between the OSH group and the 3Ј-CAO may help to facilitate the electron or hydride transfer from the pyridine ring of NADH to the 3Ј-carbonyl carbon. If this is the case, then the reverse reaction, i.e. the 3Ј-oxidation reaction, would be enhanced as the loss of the hydrogen bond between the OSH group and the 3Ј-OH would benefit the electron or hydride transfer from the 3Ј-OH and 3Ј-CH to the pyridine ring of NAD ϩ . This is also consistent with the observation that the 3Ј-oxidation rate constants of the Cys 195 mutants are slightly larger than that of the wild-type enzyme.
If the hydrogen bond between the OSH group of the Cys 195 and the 3Ј-CAO is critical to the 3Ј-reduction potential, it might be expected that a conservative mutation of Cys 195 to serine would have no significant effect on enzyme activity, since the OOH group of the serine can also form a hydrogen bond with the 3Ј-CAO of the ribose ring of 3Ј-keto-Ado. However, the optimum OH⅐⅐⅐⅐O hydrogen bond distance is at least 0.4 Å shorter than the corresponding SH⅐⅐⅐⅐O distance (13,14). Thus, since the OSH of the Cys 195 is in a position to make the optimal hydrogen bond with the 3Ј-CAO of the ribose ring of 3Ј-keto-Ado, a hydrogen bond between OOH of the serine and the 3Ј-CAO would be at least 0.5 Å longer than the optimum, which would contribute little to the binding energy and thus to the 3Ј-reduction potential.
In summary, this study has identified three DTNB modifiable cysteine residues per subunit of human AdoHcy hydrolase. The Cys 421 residue is nonessential and is not located in the Ado binding site. Identification of this nonessential cysteine residue has been found to be useful in molecular probing and derivatization of the enzyme. The Cys 113 residue is located in or near the Ado binding site, but plays no significant role beneficial to the enzyme binding or catalysis. The Cys 195 residue is involved in the catalytic center and is critical to the full catalytic function of the enzyme. Cys 195 is most likely involved in the 3Јreduction step in the overall catalytic pathway and may play a role in maintaining the 3Ј-reduction potential by hydrogen bond formation with the 3Ј-CAO of the ribose of 3Ј-keto-Ado. However, Cys 195 is not absolutely required for the enzyme activity, as demonstrated by the residual enzyme activities of the Cys 195 mutants and the evolutional mutation of Cys 195 that is replaced by isoleucine in nematode Caenorhabditis elegans (18), although it is conserved in all nine of the other known AdoHcy hydrolases from highly evolutionally divergent species (11).