Kinetic and Catalytic Mechanism of HhaI Methyltransferase*

Kinetic and catalytic properties of the DNA (cyto-sine-5)-methyltransferase HhaI are described. With poly(dG-dC) as substrate, the reaction proceeds by an equilibrium (or processive) ordered Bi-Bi mechanism in which DNA binds to the enzyme first, followed by S-adenosylmethionine (AdoMet). After methyl trans- fer, S-adenosylhomocysteine (AdoHcy) dissociates followed by methylated DNA. AdoHcy is a potent com- petitive inhibitor with respect to AdoMet (Ki = 2.0 ELM) and its generation during reactions results in non- linear kinetics. AdoMet and AdoHcy significantly in-teract with only the substrate enzyme-DNA complex; they do not bind to free enzyme and bind poorly to the methylated enzyme-DNA complex. In the absence of AdoMet, HhaI methylase catalyzes exchange of the 5-H of substrate cytosines for protons of water at about 7-fold the rate of methylation. The 5-H exchange reaction is inhibited by AdoMet or AdoHcy. In the en- zyme-DNA-AdoHcy complex, AdoHcy also suppresses dissociation of DNA and reassociation of the enzyme with other substrate sequences. Our studies reveal that the catalytic mechanism of DNA (cytosine-5)-methyl- transferases involves attack of the C6 of substrate cytosines by an enzyme nucleophile and formation of a transient covalent adduct. Based on precedents of other enzymes which catalyze similar reactions and the sus- ceptibility of HhaI to inactivation by N-ethylmaleim-ide,

and AdoHcy. In mammalian cells, DNA (cytosine-5)-methyltransferases methylate certain CpG sequences which are believed to modulate gene expression and cell differentiation (for review see Ref. 1). Bacterial DNA (cytosine-5)-methyltransferases are a component of restriction-modification systems and serve as valuable tools for the manipulation of DNA structure and the analysis of protein-nucleic acid interactions.
We have undertaken studies to elucidate the mechanisms of catalysis, ligand interactions, and specificity of DNA (cytosine-5)-methyltransferases. Recently we proposed that the catalytic mechanism of DNA (cytosine-5)-methyltransferases involves formation of a transient covalent adduct between the enzyme and the 6-carbon of cytosines, in analogy to other enzymes which catalyze 1-carbon transfer to pyrimidines (2,3). This hypothesis is supported by observations that DNA containing azaC, a potent inhibitor of DNA (cytosine-5)methyltransferases, forms covalent complexes with mammalian and bacterial enzymes (4)(5)(6)(7). The azaC is presumed to form a stable adduct analogous to the proposed catalytic intermediate of the reaction.
In this paper, we describe the kinetic and catalytic mechanisms of the DNA (cytosine-5)-methyltransferase HhaI methylase, which recognizes the sequence GCGC. Using poly(dG-dC), a synthetic substrate of the enzyme (8), we show that the methylation reaction proceeds by an ordered kinetic scheme in which enzyme first binds to DNA. We also demonstrate a novel enzyme-induced exchange of the 5-H of cytosines in the absence of AdoMet and provide evidence for the formation of a transient covalent intermediate during methylation by HhaI methylase.

RESULTS
Properties of HhaI Methylase-HhaI methylase is a monomer of M, = 37,000 as determined by gel filtration chromatography and sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The enzyme is rapidly inactivated by 0.4 mM NEM (tLh < 10 s at 22 "C) and is protected by poly(dG-dC) (at 0.21 p~, 3 tzh -60 s; at 1.1 p~ tIh -120 sf. In contrast, *The "Materials and Methods" and Fig. 1-7 are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 86M-1588, cite the authors, and include a check or money order for $2.70 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press. To facilitate comparison between kinetic data for HhaI methylase and those for other DNA (cytosine-5)-methyltransferases, concentrations of poly(dG-dC) are expressed in units of double-stranded recognition sites. Hence, poly(dG-dC) at 1.0 hi in total nucleotides is 0.25 PM, or one-fourth of the nucleotide concentration, in recognition sites. This convention is followed throughout the text unless otherwise stated.
AdoMet up to 25 p~ (1500 X K,,,) does not protect the enzyme from inactivation by NEM. These data suggest that an active site sulfhydryl group may be involved in catalysis by HhaI methylase. One picomole of enzyme is equal to 1.3 units of activity based on estimates of protein concentration in a purified preparation and from inhibition of 5-3H exchange by AdoHcy using the method of Henderson (38).
Kinetics of HhaI Methylase-catalyzed Methylation of Poly(dG-de)-Early in this study we observed that the rate of HhaI-catalyzed methylation of poly(dG-dC) by AdoMet progressively decreased as the reaction proceeded (Fig. la). The decrease in rate was inversely related to the concentration of AdoMet and most noticeable at low concentrations of AdoMet (<50 nM). This behavior was not due to decay of enzyme activity under these conditions because <lo% loss of activity was lost in 60 min. One probable cause of rapid loss of linearity in this reaction is product inhibition by methylated DNA or AdoHcy. In the reactions of Fig. 1, the concentration of poly(dG-dC) substrate is much greater than the amount of methylated DNA formed (>lOOO-fold). Competitive inhibition by the methylated product would require its Ki to be on the order of 10-"-10"2 M (concentration in doublestranded recognition sites), which is unreasonably low. The inhibition appears to be competitive with respect to AdoMet, and the nonlinear kinetics are consistent with competition by the AdoHcy generated in the reaction. The progress curves shown in Fig. 1 fit Equation 1, which describes a reaction wherein a product, i.e. AdoHcy, shows competitive inhibition with respect to the substrate, i.e. AdoMet (9).
Here, P is the amount of AdoHcy formed at time t, So is the initial AdoMet concentration, and Kt is the dissociation constant of AdoHcy. The amount of AdoHcy formed corresponds to the amount of methylated DNA, the product which is measured in these experiments. Plots of P/t versus [In (So/ (So -P))]/t provide a series of lines with positive slopes (Fig.   lb), converging at a negative value on the ordinate. The positive slopes indicate that the K,,, of AdoMet is much larger than Ki. Replots of the slopes of these lines versus So yield kinetic constants according to Equation 2: A plot of the slope versus So gave a value of 1.0 k 0.5 nM for the K, of the generated AdoHcy, in good agreement with the value determined by other methods (see below). In principle, values for K,,, and V,,, could also be derived from these data, but the large K J K , ratio in this system made it difficult to assess these parameters from progress curves. Therefore, K , and V,,, were calculated from initial velocity experiments.
To determine initial velocities, product formation was measured under conditions such that the maximal final concentrations of AdoHcy generated in the reactions was less than 1 nM, and the highest AdoHcy/AdoMet ratio achieved was below 0.03. These constraints ensured that the overall inhibition of the reaction by AdoHcy was less than 5%.
The Km for AdoMet is 14.7 f 1.8 nM. The K,,, for poly(dG-dC) is 2.3 f 0.3 nM. Initial velocity experiments also confirmed that AdoHcy is a competitive inhibitor with respect to AdoMet (see Fig. 7a). The K , for AdoHcy is 2.1 0.1 nM, which is in acceptable agreement with the value obtained by progress curve analysis. The Vmax was calculated to be 33.8 nmol e min" mg-' enzyme. Using M, = 37,000 for HhaI, kcat is calculated to be 1.3 min". The kcat/Km for poly(dG-dC) in this reaction is -9.6 X lo6 M" s-'; for AdoMet, k,,,/K,,, -1.5 X lo6 M-' S -' .
Reverse Reaction-The ability of HhaI methylase to catalyze the reverse reaction, namely, the transfer of C5 methyl groups from poly(dG- [5-methyl]dC) to AdoHcy, was measured using p01y(dG-[5-~H-methyl]dC) (specific activity -10 pCi/nmol of 5-methylcytosine). Reactions (400 p1) contained 20 nM labeled copolymer (concentration in double-stranded recognition sites), 0, 2, or 200 p~ AdoHcy, and 3.3 units of HhaI methylase. Samples (50-pl duplicates) were removed at various time intervals, mixed with 50 pl of poly(dG-[14C]dC) (22,000 dpm total) calibration standard, and measured for 3H remaining on the DNA by adsorption on DE52 as described under "Materials and Methods." No tritium loss was observed in any reaction after 100 min at 37 "C. Using a 5% change in 3H content as the lower limit of detection (equivalent to -0.1 pmol of methyl group per 50 p1 assayed), we calculate that the reverse reaction is less than 0.08 pmol/min, or at least 400-fold slower than the forward reaction based on the enzyme concentration in these reactions. Hence, the methyl transfer reaction is, as far as we're concerned, irreversible.
Tritium Release from P01y(dG-[5-~H]dC) during Methylation-HhaI-catalyzed methylation of p0ly(dG-[5-~H]dC) is accompanied by release of tritium into solvent. In 20 pM [methyZ-'4C]AdoMet, tritium release is stoichiometric with and proceeds at a rate identical to the rate of incorporation of [14C]methyl groups into poly(dG-dC) (Fig. 2). In addition to the correspondence between 3H release and 14C incorporation, this experiment confirms that 295% of the poly(dG-dC) is methylated by HhaI, as reported by Mann and Smith (8).
Tritium Exchange from P0ly(dG-[5-~H]dC) in the absence of AdoMet-Incubation of p0ly(dG-[5-~H]dC) with HhaI methylase in the absence of AdoMet caused a rapid release of tritium from the polymer into water (Fig. 2). The exchange reaction exhibits first-order kinetics and is about 7-fold faster than the 3H release that accompanies methylation. More than 95% of the tritium in the copolymer can be washed out into solvent. Tritium release is not due to chemical modification of cytosine residues, because p0ly(dG-[5-~H]dC) from which S O % of the label has been exchanged can accept [3H]methyl groups from [~nethyl-~HIAdoMet. When determined by adsorption of products to DE52, >99% of the cytosines of p01y(dG-[5-~H]dC) was methylated by HhaI after the majority of the 5-3H was exchanged.
The exchange from p0ly(dG-[5-~H]dC) is specifically catalyzed by HhaI methylase. Other enzymes, for example HpuII methylase and the restriction endonucleases HhaI and BssHII, which also bind to poly(dG-dC), do not cause tritium release from the copolymer.
The K,,, for poly(dG-dC) in the exchange reaction is 2.1 nM in recognition sites, in good agreement with the value of 2.3 nM obtained in the methylation assay. The kcat was 8.85 min-', some 7-fold greater than that €or methylation. From these, kCat/K,,, was calculated to be -6.7 X lo7 M" s-'.
Ligand Interactions in the Exchange Reaction-As shown in Fig. 3, AdoHcy is an extremely potent inhibitor of the exchange reaction with Ki = 1.4 nM, in good agreement with its Ki (2.1 nM) in the methylation reaction. However, we do not know the extent of inhibition of exchange by AdoHcy in the ternary enzyme-DNA-AdoHcy complex (Scheme I). Equation 3 describes the HhaI-catalyzed exchange of poly(dG-[5-3H]dC) from the binary enzyme-DNA and ternary enzyme-DNA-AdoHcy complexes without prior assumptions about the contribution of each complex to the overall rate.

Mechanism of HhaI Methylase
The order of substrate binding and product release for the reaction catalyzed by HhaI methylase.
Here, k1 is the apparent first-order rate of exchange determined for the binary complex in the absence of inhibitor; kz is the rate of exchange from the ternary enzyme-DNA-AdoHcy complex; kbbS is the observed rate of exchange, and kd is the dissociation constant for AdoHcy from the ternary complex. A replot of the data from Fig. 3 illustrates the application of Equation 3 (Fig. 4). The plot yields an ordinate intercept (l/(klk p ) ) of 98 min which, based on K, = 1.02 X IO-' min", establishes that K2 C= 0, i.e. catalytic turnover of 5-H exchange in the ternary complex is negligible. In a separate experiment containing 0.20 p~ AdoHcy, saturating p0ly(dG-[5-~H]dC) (3.44 p~) , and HhaI methylase at -0.26 p~, the amount of tritium released after 30 min at 33 "C was <0.3% of the uninhibited reaction. This amount of exchange was indistinguishable from the control-omitting enzyme, and represented <3 mol % of enzyme concentration, showing that single turnover exchange from the ternary enzyme-DNA-AdoHcy complex is also suppressed by AdoHcy.
Tritium exchange was examined by varying p0ly(dG-[5-~H] dC) concentrations at different fixed concentrations of AdoHcy. Double-reciprocal plots of velocity uersus substrate yield an uncompetitive inhibition pattern (Fig. 5), consistent with a mechanism in which free enzyme associates only with DNA. Binary complexes of enzyme and AdoHcy do not form in these conditions. Inhibition of exchange results from the binding of AdoHcy to enzyme-DNA to form a "dead-end ternary complex.
Ligand Interactions in the Methylation Reaction-Initial velocities were determined for methylation reactions using various concentrations of [meth~l-~HJAdoMet and 32Ppoly(dG-dC). Double-reciprocal plots of l / u uersus 1/ [AdoMet] gave a series of lines intersecting at the ordinate (Fig. 6a). Plots of l / u uersus l/[poly(dG-dC)] gave lines that intersect to the left of the ordinate (Fig. 6b). The data have been fitted to Equation 4, which describes an equilibrium ordered mechanism (Scheme 11). DNA (designated A) initially forms a rapidly reversible EA complex, followed by binding of AdoMet (designated B); Kk is the dissociation constant for the EA complex and Kb is the Michaelis constant for AdoMet.
Other mechanisms were eliminated by demonstrating unac-ceptable fits of the data to the appropriate equations using the Fortran programs of Cleland (13). The corresponding equation for a nonequilibrium ordered mechanism has a KJB] term in the denominator (identical in form to that for a rapid equilibrium random mechanism) and is ruled out by the data. Hence, dissociation of enzyme-DNA is faster than the forward maximal velocity. A processive mechanism in which enzyme binds DNA by lateral diffusion can also be fitted to Equation 4 and cannot be distinguished from the conventional rapid equilibrium mechanism by the present work. AdoHcy inhibited the reaction competitively with respect to AdoMet and uncompetitively with respect to DNA (Fig. 7). Kinetics of inhibition by AdoHcy were best fitted to a modification of AdoMet did not show substrate inhibition at concentrations >1 mM (60,000-fold K,,,). Thus, it does not bind significantly to the enzyme-DNAM" product. Since AdoHcy is a competitive inhibitor with respect to AdoMet, AdoHcy and AdoMet must bind to the same enzyme form, which we have shown to be the substrate enzyme-DNA complex. This conclusion is in accord with results of AdoHcy binding in the 5-H exchange reaction. Since AdoHcy does not bind to free enzyme, the order of product release must be AdoHcy first, followed by methylated DNA. Furthermore, dissociation of AdoHcy from the methylated enzyme-DNA-AdoHcy product complex is thermodynamically favored as indicated by the poor binding of AdoHcy to the methylated DNA-enzyme product (in contrast to its affinity for substrate enzyme-DNA). Otherwise, AdoHcy would show noncompetitive inhibition kinetics with respect to AdoMet.

DISCUSSION
In this work we describe kinetic and catalytic properties of the DNA (cytosine-5)-methyltransferase HhaI, which methylates the internal cytosine residue of the tetranucleotide sequence GCGC. The enzyme is a monomer with native M, -37,000.
Enzyme Assays and Kinetic Parameters-Poly(dG-dC) was chosen as the substrate in these studies because it offers several practical advantages. ( a ) The alternating copolymer provides recognition sites that are not flanked by heterogeneous sequences. In other DNAs, such heterogeneity could complicate kinetic interpretations. (b) As shown here and elsewhere (S), over 95% of the cytosine residues of poly(dG-dC) can be methylated. Therefore, use of the copolymer allows high concentrations of substrate sites. The nearly complete methylation of poly(dG-dC) demonstrates that the enzyme also methylates internal cytosines in the partially methylated sequence GCGMeC. We recognize that partially methylated sequences could alter kinetic properties of the enzyme, but this is not apparent in our data. (e) For future studies of processivity, alternating sequences of poly(dG-dC) are ideal because the substrate sites are separated by a frame of only two nucleotides. (d) The polymer contains repeating units of the CG substrate sites of mammalian DNA (cytosine-5)methyltransferase; we anticipate that many results obtained with the bacterial enzyme can be directly correlated with those which will be obtained with the mammalian enzyme.
In addition to the standard assay which measures incorporation of labeled methyl groups into DNA from AdoMet, we use two other assays for HhaI methylase. In the first, we monitor the AdoMet-dependent release of tritium from [5-3H]cytosine of p01y(dG-[5-~H]dC) into water. In the second, we monitor the enzyme-catalyzed release of tritium from p0ly(dG-[5-~H]dC) in the absence of AdoMet. The latter exchange reaction has not been reported and has important implications for the catalytic mechanism of the methylase (see below). AdoMet inhibits the 5-3H exchange with protons of water. At saturating concentrations of AdoMet, the loss of tritium from p0ly(dG-[5-~H]dC) is accounted for by its release upon methylation.
The generation and buildup of the product AdoHcy causes a rapid, progressive decline in the rate of HhaI-catalyzed methylation of poly(dG-dC). Therefore, accurate determination of the kinetics of methylation required progress curve analysis (9) or stringent initial velocity conditions. Our results suggest that the practice of monitoring single time points during methylation could result in low and imprecise initial velocity determinations, with consequent errors in derived kinetic parameters.
The kcat for HhaI methylase (1.3 min-') is slow, but similar to those reported for other DNA methyltransferases (6,17). The K, for poly(dG-dC) in both the exchange and methylation reactions is about 2.3 nM in double-strand recognition sites. This number agrees well with K, values determined for other DNA methyltransferases (18,19). However, the K, for AdoMet (-15 nM) and the K, for AdoHcy (2.0 nM) were substantially lower (about 50-and 100-fold, respectively) than values reported for other DNA methyltransferases. We can estimate the rate of DNA association to HhaI methylase based on its K, and the turnover numbers. The kJK, for DNA (6.7 X lo7 M-' s-') in the exchange reaction represents a minimum estimate of the rate constant for the association of DNA to enzyme, and approaches the diffusioncontrolled limit (10s-109 M-' s-') for enzyme-ligand interactions (20). Based on this value and the dissociation constant K, for the enzyme-DNA complex derived for the methylation reaction, the lower limit for the dissociation rate constant is calculated to be 0.15 s-'. Thus, dissociation of the binary complex is significantly faster than kc,, of methylation (0.02 s-') and verifies other kinetic data (see below) which indicate that the rate-determining step for the methylation reaction occurs after formation of the enzyme-DNA complex. Our data do not allow assignment of the rate-determining step.
Kinetic Mechanism-The H h I methylase-catalyzed reaction proceeds by an ordered bi-bi mechanism as shown in Scheme 11. DNA first binds to enzyme, followed by AdoMet.
Kinetic data indicate the rapid equilibrium binding of DNA; interaction of enzyme with this substrate is more rapid than catalysis. Methyl transfer in the ternary enzyme-DNA-AdoMet complex yields the product complex of enzyme-DNAM"-AdoHcy. AdoHcy is subsequently released to give the enzyme-DNAMe complex. Whereas AdoHcy and AdoMet are tightly bound to the substrate enzyme-DNA complex, they are poorly bound to the product enzyme-DNAMe complex, which possesses a methylated cytosine residue in the substrate site. Consequently, release of AdoHcy from the product ternary complex is favorable, and high concentrations of AdoMet do not cause significant substrate inhibition. We cannot distinguish whether the enzyme fully dissociates from the product DNAMe after each methylation event, or processively moves to an adjacent substrate site on the same molecule of DNA. The two mechanisms are kinetically equivalent.
The arguments for the assigned kinetic mechanism are as follows: (a) Double-reciprocal plots of initial velocity data with AdoMet as the variable substrate and poly(dG-dC) as the fixed substrate give a pattern of lines intersecting at the reciprocal velocity axis. With poly(dG-dC) as the varied substrate, plots give lines that intersect to the left of the reciprocal velocity axis. The data are inconsistent with an equilib-rium ordered mechanism and indicate rapid association and dissociation of DNA to the enzyme. These results rule out a random mechanism as well as a ping-pong mechanism which might have indicated a methylated enzyme intermediate. They do not address whether the enzyme is processive or distributive, nor do they demonstrate the order of product release; the latter is established by data described below. (b) The fact that HhaI methylase catalyzes 5-3H exchange from p0ly(dG- [5-~H]dC) in the absence of AdoMet with a K, for DNA similar to that in the methylation reaction confirms the formation of a catalytically competent binary enzyme-DNA complex. Evidence for a binary enzyme-DNA complex is also provided by the protection which poly(dG-dC) affords HhaI methylase against inactivation by NEM. (c) Studies of inhibition demonstrate that AdoHcy binds to the enzyme-DNA binary complex, but not to free enzyme. AdoHcy is an uncompetitive inhibitor with respect to DNA in both 5-H exchange (Ki = 1.4 nM) and methylation (K; = 2.0 nM). Uncompetitive inhibition demonstrates the formation of an enzyme-DNA-AdoHcy complex and rules out the formation of a binary enzyme-AdoHcy complex. Since AdoHcy does not associate with free enzyme, dissociation of products in the methylation reaction must also be ordered. AdoHcy must first dissociate from the enzyme, followed by methylated DNA.
From the kinetics of methylation, AdoMet does not appreciably bind to the binary product complex enzyme-DNAMe. Formation of such a ternary dead-end complex would result in substrate inhibition at high concentrations of AdoMet. Based on the competitive kinetics of inhibition by AdoHcy with respect to AdoMet, we also conclude that AdoHcy does not bind tightly to the enzyme-DNAMe product complex. Formation of a tight enzyme-DNAMe-AdoHcy complex would have resulted in non-competitive inhibition by AdoHcy with respect to AdoMet in the methylation reaction, Thus, although AdoMet and AdoHcy bind tightly to substrate enzyme-DNA complexes, their affinity for the product enzyme-DNAMe complex is low.
Catalytic Mechanism-The methylation of DNA by DNA (cytosine-5)-methyltransferases is analogous to other enzyme-catalyzed transfers of 1-carbon units to the C5 of pyrimidine nucleotides. Examples of enzymes that catalyze this type of reaction include thymidylate synthase, dUMP and dCMP hydroxymethylases, certain RNA-modifying enzymes, and DNA (cytosine-5)-methyltransferases (see Ref. 3 for review). The mechanism of this class of enzymes has been established most thoroughly for thymidylate synthase, and several salient features of catalysis have emerged. The primary consideration is that the carbon at the 5-position of the pyrimidine is not sufficiently nucleophilic to react with biological donors of 1-carbon units. However, the heterocycle is susceptible to addition/elimination reactions which activate the 5-position for electrophilic substitution reactions. In thymidylate synthase, a nucleophile of the enzyme adds to C6 of dUMP to generate a 5,6-dihydropyrimidine intermediate with anionic character at the 5-position. This carbanion equivalent is sufficiently nucleophilic to condense with 1-carbon units; subsequent @-elimination of the 5-H and enzyme nucleophile generates the 5-substitutedpyrimidine and catalytically active protein. Most aspects of this mechanism have been verified with studies of model chemical counterparts (21,22). Thus, the hallmark of biological electrophilic substitution reactions at the 5-position of pyrimidines appears to be covalent catalysis involving addition of a nucleophile to the 6-carbon of the heterocycle.
Two approaches to detect covalent intermediates are relevant to DNA (cytosine-5)-methyltransferases. The first is SCHEME 111. The mechanism of cytosine methylation by HhaI methylase.
demonstration of an exchange of the 5-H of the substrate pyrimidine with solvent protons. Insofar as we know, this reaction can only occur via nucleophilic attack at the 6position and the formation and reversal of 5,6-dihydropyrimidine intermediates (21,(23)(24)(25). A second approach employs mechanism-based inhibitors to trap the covalent intermediate. These analogs form stable covalent adducts with the enzyme as a consequence of catalysis. The adducts represent analogs of a steady-state intermediate of the reaction and provide important tools for studying aspects of catalysis. 5-Fluoro-2"deoxyuridine 5'-monophosphate inhibition of thymidylate synthase represents a paradigm for this approach (3). Previous studies demonstrated covalent binding of several DNA (cytosine-5)-methyltransferases to DNA containing azaC (5,6,26). The mechanism of inhibition by azaC has been proposed to involve covalent addition at C6 of azaC residues in substrate sites by an enzyme nucleophile (2,5 ) .
We show here that HhaI methylase catalyzes 5-3H release from p0ly(dG-[5-~H]dC) in the absence of AdoMet. Other enzymes which catalyze 5-H exchange of their pyrimidine nucleotide substrates do so at rates much slower than those of the normal methylation reactions (23)(24)(25): In contrast, 5-H exchange catalyzed by HhaI methylase is about 7-fold faster than methylation and provides direct support for the formation of covalent intermediates during catalysis by HhaI methylase. The most reasonable mechanism is shown in Scheme 111. Upon binding of substrate(s), an enzyme nucleophile adds L. Hardy and D. V. Santi, unpublished results. Stereospecificity-The stereochemical configurations at C5 and C6 are determined by which face of the pyrimidine accepts the substituents. We expect addition of the enzyme nucleophile at C6 and subsequent methyl transfer from AdoMet to C5 to proceed in a stereospecific manner. 5-H exchange, however, does not require stereospecific addition/elimination of protons at C5 of intermediates 1 and 3, as discussed below.
DNA structure could influence the stereochemical course of nucleophilic addition to C6. We have used the molecular graphics computer program Insight (27) to examine the local environment of cytosine residues in DNA. In B-DNA, covalent addition to cytosine C6 is most favorable via the si face of the heterocycle (the facial assignment as defined at C6), because this face is exposed in the major groove of DNA. The C6 appears inaccessible from the re face, which is bounded closely by atoms of nejghboring base residues (average internuclear distance <4 A). Therefore, unless there is a large disruption of the native double helix upon association of the enzyme, covalent addition must occur from the si face. In Z-DNA, the C6 positions of cytosines are sterically blocked on both faces. This factor may account for the inability of HhaI to methylate Z-DNA (28, 29), although reduced binding interactions between the enzyme and Z-DNA could also explain the phenomenon. Access to C5 also is hindered on the re face, such that methyl transfer to give 2 may be required to proceed from the same face as nucleophilic addition. Thus the overall reaction may involve cis addition/trans elimination, which contrasts with the trans-addition/cis-elimination reaction established for thymidylate synthase (30).
Whatever the stereochemical course of addition of the enzyme nucleophile and methyl group across the 5,6-double bond of cytosine to give 2, subsequent elimination of the 5-H must occur from the face opposite that of methyl addition. Thus, trans addition would be followed by cis elimination or vice versa. However, the stereochemistry of the enzyme-catalyzed 5-H exchange is enigmatic. If the reaction is stereospecific and exchange simply involves a reversal of steps leading to formation of 3, then the same proton added to 1 to give 3 would be removed during reversal. The result is that formation and reversal of 3 would not be accompanied by 5-H exchange. Two alternative explanations may account for the observed 5-H exchange. The first is that proton addition to 1 and removal from 3 might occur in a non-stereospecific fashion from either face of the pyrimidine (cis or trans to the enzyme at C6). In this case, the rate of formation and reversal of 3 could actually be greater than the observed rate of 5-H exchange. The second explanation is that proton addition to 1 and removal from 3 follows the same stereochemical course as methylation and is completely stereospecific. Thus, like the methyl group of AdoMet, the proton from water would add to a single face to provide 3 having an asymmetric C5.
The tritium that was originally at C5 would be abstracted from the opposite face of the dihydropyrimidine 3, and subsequent @-elimination yields the 5-protio-cytosine and free enzyme. As with the methylation reaction, this mechanism for 5-H exchange requires trans addition/& elimination (or vice versa) across the 5,6-double bond of cytosine. We favor this hypothesis because it utilizes catalytic features of the normal enzymic reaction of methyl group transfer.
5-H Exchange from Ternary Complexes-We have shown that AdoHcy inhibits the HhaI-catalyzed 5-H exchange of p0ly(dG-[5-~H]dC). This inhibition is most simply explained by the ordered association of ligands with the enzyme: at a saturating concentration of the second substrate, dissociation of the first is suppressed. In this situation, 5-H exchange from multiple substrate sites requires the enzyme to release AdoHcy from the ternary complex before it can bind to a new DNA site. In the presence of saturating AdoHcy, the enzyme-DNA complex preferentially partitions to the enzyme-DNA-AdoHcy complex; dissociation of the binary complex is suppressed. Thus, catalysis of 5-H exchange is inhibited simply because the enzyme is locked onto the same tetranucleotide sequence which it initially bound (the consequences to processivity in this system will be addressed in a future report). In addition, binding of AdoHcy to the enzyme-DNA complex in itself causes complete suppression of 5-H exchange from the ternary complex (see below).
Unlike AdoHcy, AdoMet could not inhibit 5-H exchange by simply preventing dissociation and reassociation of the enzyme-DNA complex. Catalysis of methylation requires that dissociation of products and association of substrates occur at a rate commensurate with turnover. However, if the binary enzyme-DNA complex is the species responsible for 5-H exchange, then high concentrations of AdoMet will partition it toward the ternary complex. This effect would reduce the steady-state concentration of the binary complex and consequently the apparent rate of 5-H exchange. For example, using a lower limit of 1.5 X lo6 M" s" for the association rate constant of AdoMet and the kc,, for exchange, at 1 PM AdoMet the rate of 5-H exchange would be at most one-eleventh of exchange in the absence of AdoMet. This explanation does not address the issue of whether 5-H exchange occurs within the enzyme-DNA-AdoMet central complex. At saturating AdoMet concentrations, we could not detect tritium release beyond that accountable by methylation (see Fig. 2). Assuming a 10% difference in the rates of 3H release and methylation as the minimum detectable level, we estimate from the relative kcat values that 5-H exchange from the ternary complex is at least 70-fold slower than from the binary complex. Hence, the presence of AdoMet within the enzyme-DNA complex also appears to suppress 5-H exchange.
There are several possible mechanisms by which AdoHcy or AdoMet could inhibit single-turnover exchange from the respective ternary complex. First, AdoHcy may simply prevent covalent addition of the enzyme to C6 of cytosine, so that the essential intermediate 1 is not formed. However, this cannot be the mechanism for AdoMet-induced suppression of exchange because the covalent adduct 1 is formed during catalysis of methylation. Second, if exchange results from non-stereospecific formation and reversal of 3, then the presence of bound AdoMet or AdoHcy could simply enhance the stereospecificity of the reaction. In this case, 3 would be formed, but the proton released from C5 would be the same as the one received; hence no net exchange would be observed. A third possibility is that the presence of bound AdoMet or AdoHcy precludes accessibility of C5 to protons of solvent. In the presence of AdoHcy, there may be transient formation of 1, but it could only undergo reversal. In the presence of AdoMet, 1 could accept a methyl group from the cofactor but not a proton from solvent.
The Active-site Nucleophile and Its Role in Protein-Nucleic Acid Interactions-We have demonstrated that DNA (cytosine-5)-methyltransferase shares several catalytic features with other enzymes that catalyze 1-carbon transfer reactions at the 5-position of pyrimidines, and it is reasonable to expect that elements important in catalysis would be conserved in such enzymes. In thymidylate synthase, the paradigm for such enzymes, the nucleophile that initiates catalysis by addition to the 6-position of dUMP is a sulfhydryl of cysteine, which in all sources thus far examined is preceded by a proline (31)(32)(33). Our observations that HhaI methylase is inactivated by NEM and that the enzyme is protected from such inactivation by substrate DNA suggest that it may also utilize cysteine as the essential active-site nucleophile. In addition, we have noted that a Pro-Cys doublet found at residues 80-81 in HhaI methylase is also found in the deduced amino acid sequences of BsuRI, BspRI, and Bacillus subtilis phage SPR methylases (14,(34)(35)(36). This observation is significant for the following reasons: (i) the Pro-Cys doublets are embedded in a highly conserved region of the methylases (encompassing over 200 amino acids), (ii) the doublets are aligned among all the sequences, and (iii) this alignment contains the only cysteine which is invariant in these DNA (cytosine-5)-methyltransferases. Based on the above considerations, we propose that the common sulfhydryl found in the conserved Pro-Cys sequence of DNA (cytosine-5)-methyltransferases (residue 81 of H h I