Two Alternative Conformations of S-Adenosyl-L-homocysteine Bound to Escherichia coli DNA Adenine Methyltransferase and the Implication of Conformational Changes in Regulating the Catalytic Cycle*

The crystal structure of the Escherichia coli DNA adenine methyltransferase (EcoDam) in a binary complex with the cofactor product S-adenosyl-l-homocysteine (AdoHcy) unexpectedly showed the bound AdoHcy in two alternative conformations, extended or folded. The extended conformation represents the catalytically competent conformation, identical to that of EcoDam-DNA-AdoHcy ternary complex. The folded conformation prevents catalysis, because the homocysteine moiety occupies the target Ade binding pocket. The largest difference between the binary and ternary structures is in the conformation of the N-terminal hexapeptide (9KWAGGK14). Cofactor binding leads to a strong change in the fluorescence of Trp10, whose indole ring approaches the cofactor by 3.3Å. Stopped-flow kinetics and AdoMet cross-linking studies indicate that the cofactor prefers binding to the enzyme after preincubation with DNA. In the presence of DNA, AdoMet binding is ∼2-fold stronger than AdoHcy binding. In the binary complex the side chain of Lys14 is disordered, whereas Lys14 stabilizes the active site in the ternary complex. Fluorescence stopped-flow experiments indicate that Lys14 is important for EcoDam binding of the extrahelical target base into the active site pocket. This suggests that the hexapeptide couples specific DNA binding (Lys9), AdoMet binding (Trp10), and insertion of the flipped target base into the active site pocket (Lys14).


EXPERIMENTAL PROCEDURES
Protein Purification and Mutagenesis-N-terminally Histagged EcoDam used for crystallization was expressed in HMS174(DE3) cells, by the method of autoinduction (17), and purified utilizing Ni 2ϩ affinity, UnoS, and S75-Sepharose sizing columns. Site-directed mutagenesis was performed as described (18). EcoDam wild type and its variants used for the biochemical experiments were purified as described (4,9). Because we observe that residues close to the N terminus of EcoDam are involved in AdoMet binding (Trp 10 in this work) and recognition of Gua1 (Lys 9 in Ref. 4), we cloned EcoDam with C-terminal His tag and purified the protein. Comparison of the K m values for AdoMet and the relative methylation of substrates, which carry a nucleotide exchange at position Gua1, showed that the N-terminal His tag did not interfere with AdoMet binding and Gua1 recognition (data not shown).
Crystallography-Concentrated binary complexes were mixed with a 12-bp oligonucleotide duplex (synthesized by the New England Biolabs, Inc) at a protein to DNA ratio of ϳ2:1 and allowed to stand on ice for at least two h before crystallization; final protein concentration for crystallization trials was ϳ15 mg/ml. Small hexagonal crystals (Ͻ50 m) were observed under high salt conditions of ϳ1.2-1.6 M NH 4 /Li(SO 4 ) 2 , 50 mM MgSO 4 , 100 mM buffer (MES or HEPES) (pH 6.8 -7.2). Attempts to obtain crystals of the EcoDam-cofactor binary complex in the absence of DNA resulted in small particles unsuitable for x-ray crystallographic study.
The structure was determined by molecular replacement with the program REPLACE (19), using an EcoDam protomer from the refined ternary complex structure (PDB 2G1P) as a search model. Although the oligonucleotides were present, the high salt crystal form only contains three binary complexes of EcoDam-AdoHcy. During the refinement, using the program CNS (20), the 3-fold noncrystallographic restraints were imposed at the earlier stage of refinement and were released at the later cycles to account for different interaction environments of crystal packing with each molecule ( Table 1). The two AdoHcy conformations in molecules ␤ and ␥ were refined with equal occupancy (50%), resulting in similar crystallographic temperature factors (in the range of ϳ30 Å 2 ). The folded AdoHcy conformation in molecule ␣ was refined in 90% occupancy.
DNA Methylation Studies-Oligonucleotide substrates were purchased from Thermo Electron (Dreieich, Germany) in the purified form. Methylation experiments were performed in 50 mM HEPES (pH 7.5), 50 mM NaCl, 1 mM EDTA, 0.5 mM dithiothreitol, 0.2 g/l bovine serum albumin containing 3.6 M [methyl-3 H]AdoMet (PerkinElmer Life Sciences) at 37°C as described (4,9,21) using 1 M 20-bp oligonucleotide and 0.25 M enzyme. The 20-bp oligonucleotide containing one GATC target site was a duplex of 5Ј-GC GAC AGT GATC GGC CTG TC-3Ј and 5Ј-GA CAG GCC GMTC ACT GTC GC-3Ј, where M is N 6 -methyl-Ade. Methylation experiments were repeated at least three times. The data shown are an average of all experiments. For analysis, the data were fitted to a curve describing a single exponential followed by a linear phase as shown in Equation 1, where k 1 is the rate constant of the exponential phase; F is amplitude of the exponential phase; k 2 is the rate constant of the linear phase; and B is experimental background.
For error analysis the experiments were analyzed separately. Results are given as averages of the individual analyses together with their respective standard deviations. Steady-state kinetics were determined by using various concentrations of AdoMet (0.6 -6 M), 0.5 M, 20-bp DNA, and 50 nM EcoDam. For each individual condition, the time course of DNA methylation was measured between 1 and 30 min. The slope of the initial part of the reaction progress curve was determined by linear regression, and the slopes were fitted to the Michaelis-Menten model to derive the K m and k cat values. For error analysis, by changing all other parameters K m and k cat were minimized and maximized separately, until the deviation of the data points from the fit was significantly worse than that of the best fit as determined by Student's t test using p ϭ 0.05.
UV Cross-linking AdoMet Binding Studies-AdoMet binding to EcoDam was studied by UV cross-linking as described (22). Briefly, enzyme (4 M) and AdoMet (7.2 M) were incubated in the absence and presence of the specific oligonucleotide (2 M) for 10 min on ice. The mixture was then subjected to UV crosslinking and AdoMet binding to the enzyme analyzed by SDS-PAGE. To avoid DNA methylation in the experiments, a catalytically inactive EcoDam variant (D181A) was used, which was purified as described (12).
2-Aminopurine Fluorescence Studies-The kinetics of base flipping were investigated by stopped-flow experiments performed in an SF-3 stopped flow device (BioLogic, Claix, France) as described (12) using enzyme and DNA at equal concentra- a Crystals, acquired with a nylon loop (Hampton), were quickly transferred to mother liquor containing either 25% glycerol or ethylene glycol before being flashfrozen directly in liquid nitrogen or in a cold nitrogen gas stream at 100 K. AUGUST 3, 2007 • VOLUME 282 • NUMBER 31 Tryptophan Fluorescence Stopped-flow Studies-Changes of tryptophan fluorescence upon AdoMet binding were determined by stopped-flow experiments performed essentially as described above. In these experiments 20 bp of DNA and enzyme (0.35 M each) were premixed in buffer containing 50 mM HEPES (pH 7.5) and 50 mM NaCl and rapidly mixed with AdoMet in the same buffer. In alternative setups, the experiments were carried out in the absence of DNA. The tryptophan fluorescence was excited at 295 nm, and emission was observed using a 320 nm cut-off filter. The dead time of the experiments was 3.1 ms. Typically, 15-20 shots were accumulated, averaged, and analyzed. All stopped-flow experiments were carried out at least in triplicate.

RESULTS
We used an autoinduction procedure (17) that yielded ϳ7 mg of purified EcoDam from a 0.5-liter bacterial culture. In the last purification step and during concentration (see "Experimental Procedures"), AdoHcy was added to the protein. During the trials for obtaining crystals of ternary complexes, concentrated binary complexes were mixed with oligonucleotide duplex. Hexagonal crystals were observed under high salt conditions. Although the oligonucleotides were present in solution, the high salt crystal form contains only the binary complexes of EcoDam-AdoHcy.
Overall Structure of EcoDam-The crystallographic asymmetric unit contains three molecules of EcoDam (Fig. 1A), named ␣, ␤, and ␥, that are highly similar to each other with a pairwise root-mean-square deviation of less than 0.3 Å comparing 242 pairs of C-␣ atoms. The binary structure is also highly similar to the ternary structure, with a root-mean-square deviation of 0.5 Å between molecule ␣ (binary) and the protein component from the ternary structure of EcoDam-DNA-AdoHcy (4) (see Fig. 1D for an extract of the superposition). The largest difference in protein conformation between the binary and the ternary structures lies in the hexapeptide ( 9 KWAGGK 14 ) ending with Lys 14 whose side chain is disordered in the binary structure (Fig. 1D), with the C-␣ atoms of Gly 12 and Gly 13 moved more than 2.0 Å. This hexapeptide sequence motif is highly conserved with two invariant residues of GXK among the GATC-related DNA-(adenine-N 6 )-MTase orthologs (see Fig. 1 of Ref. 8), and one residue of that stretch (Lys 9 ) forms the only sequence-specific contact of EcoDam to the Gua of the GATC sequence (4).
Conformations of AdoHcy-We can model two different conformations of AdoHcy in molecules ␤ and ␥, with an estimated 50% occupancy for each conformation (Fig. 1C). The extended conformation (green sticks in Fig. 1C) is the same one described for the ternary structure, as well as for many other DNA MTases (13,14). In molecule ␣, the folded conformation (yellow sticks in Fig. 1B) is the predominant conformation. However, residual and broken electron density exists for the extended conformation in molecule ␣, with an estimated occupancy less than 10%.
The adenine and ribose rings of AdoHcy in both conformations occupy the same positions (Fig. 1C). One side of the adenine ring forms face-to-edge van der Waals contacts with the phenyl ring of Phe 35 (Fig. 1, B and C); this interaction is highly conserved among DNA MTases via the FXGXG motif I (23). On the other side of the adenine ring lies the side chain of Ile 55 of motif II. A strongly conserved acidic residue (Asp 54 of motif II) forms two hydrogen bonds with both the ribose 2Ј and 3Ј hydroxyls ( Fig. 1, B and C); this is nearly universal to class I MTases (13). These interactions have been experimentally shown to be essential for AdoMet binding for many DNA MTases, including the bacterial M.EcoRV (24) and M.HhaI enzymes (25) and the murine Dnmt3a enzymes (22). A unique interaction of EcoDam involves the indole group of Trp 10 , whose ring nitrogen forms a hydrogen bond with one of the ribose hydroxyls (Fig. 1, B and C). This residue had been shown before to be in close proximity to the cofactor by UV crosslinking reaction (26).
The dihedral angle, C-4Ј-C-5Ј-S␦-C␥, begins to define whether the AdoHcy is extended (ϳ160°) or folded (ϳ80°). In the extended conformation (green sticks in Fig. 1, C-E), the peptide backbone atoms of GXG of motif I lie underneath the homocysteine moiety, whereas Asp 181 interacts with the amino group (NH 3 ϩ ). In the folded conformation (yellow sticks in Fig. 1, B-D), the amino group of AdoHcy makes a cationinteraction with the aromatic ring of Tyr 184 and forms a salt bridge with the carboxyl group of Asp 181 . In addition, the AdoHcy carboxyl oxygen atoms (COO Ϫ ) interact with the backbone amide of Tyr 184 (Fig. 1, B and C). The S␦-C␥ bond occupies a position corresponding to the S␦-CH 3 bond in AdoMet as observed in M.RsrI (14) (Fig. 2A). Thus, in the folded conformation the homocysteine moiety of the AdoHcy occupies the active site pocket, where the target Ade would bind to after being flipped out from the double helix. Previously, two distinct binding conformations (one for AdoMet and the other for AdoHcy) were observed in M.RsrI (14) and M.TaqI in the absence of DNA (27). Superimposition indicates that the extended conformation of AdoHcy in EcoDam represents the AdoMet conformation in M.RsrI, whereas the folded AdoHcy conformation in EcoDam corresponds to the AdoHcy conformation in M.RsrI ( Fig. 2A) and M.TaqI (data not shown).
Solution Studies on AdoMet and AdoHcy Binding-We suspected that the close proximity of cofactor and Trp 10 might induce changes in the fluorescence of this tryptophan upon binding of cofactor or its analog. As shown in Fig. 2B, the fluorescence of tryptophan in EcoDam was reduced by about 10% after rapid mixing of AdoMet and EcoDam-DNA complex. To test if this change indeed is because of the interaction of AdoMet with Trp 10 , EcoDam variants were prepared in which Trp 10 or Trp 236 (the only other tryptophan in this protein) was mutated to tyrosine (W10Y) or leucine (W236L). Both variants exhibited wild type-like catalytic activity (data not shown). However, the W10Y variant did not show a change in fluorescence upon AdoMet addition, whereas the W236L variant behaved like the wild type enzyme. This indicates that it is the close approximation of the cofactor and Trp 10 that causes the change of fluorescence in EcoDam.
The fluorescence stopped-flow signal was employed to study the kinetics of AdoMet/AdoHcy binding to EcoDam and EcoDam-DNA complexes. As shown in Fig. 2C, neither AdoMet nor AdoHcy binding to the free enzyme was detectable. Although it cannot be ruled out completely that AdoMet or AdoHcy bind to the free enzyme as well but do not cause a change in fluorescence at Trp 10 , this is very unlikely because the close contact between cofactor and Trp 10 is seen in the binary (this work) and ternary complexes (4). In this context it is noteworthy that DNA was present in the initial mixture used for crystallization; the high salt conditions probably disrupted the protein-DNA interaction, but the cofactor stayed bound to the enzyme after the DNA disassociation. AdoMet binding was detectable when the enzyme was preincubated with the 20-bp oligonucleotide substrate. Preferential binding of AdoMet to EcoDam-DNA complexes was confirmed by UV cross-linking cofactor binding studies using the catalytically inactive D181A variant (Fig. 2E).
Given these results our data suggest that in the catalytic cycle of EcoDam, AdoMet usually associates with enzyme bound to DNA (either at a specific site or at nonspecific sites). This agrees with the observation that the catalytic efficiency of the enzyme preincubated with DNA and then mixed with AdoMet is higher than if the enzyme is preincubated with AdoMet (28).
On the basis of the total fluorescence changes observed in the stopped-flow cofactor binding experiments, in the presence of DNA, binding to AdoMet was 1.8 Ϯ 0.2-fold stronger than binding to AdoHcy, if the enzyme was bound to DNA. This  AUGUST 3, 2007 • VOLUME 282 • NUMBER 31 result was confirmed by stopped-flow AdoMet/AdoHcy binding experiments conducted in the presence of specific DNA at different concentrations of AdoMet or AdoHcy (supplemental Fig. 2). The fluorescence traces were fitted to mono-exponential curves, and the apparent rates were re-plotted against the cofactor concentrations to determine the rate constants of cofactor binding and release. This analysis indicates that k on was ϳ2-fold higher for AdoMet (1.9 ϫ 10 4 s Ϫ1 M Ϫ1 ) than for AdoHcy (8.4 ϫ 10 3 s Ϫ1 M Ϫ1 ), whereas k off was the same for both (AdoMet, 0.29 s Ϫ1 ; AdoHcy, 0.28 s Ϫ1 ). One caveat of these experiments is that DNA methylation takes place in the presence of specific DNA and AdoMet. Then the AdoMet is turned over to AdoHcy during the experiment, indicating that the preference for AdoMet might be even higher than determined here.

Two AdoHcy Conformations in EcoDam
Side Chain of Lys 14 Communicates between DNA Binding, Cofactor Binding, and Base Flipping-In the ternary EcoDam-DNA-AdoHcy complex (4), the side chain of Lys 14 interacts with side chains of Asp 181 and Tyr 184 of the DPPY motif (Fig.  1E). The same DKY interactions were observed in the structures of T4 Dam (8), where the corresponding lysine residue Lys 11 is close to the ring N-1 atom of the target Ade (9). Substitution of arginine for the corresponding lysine in M.EcoRV (K16R) altered the specificity toward the target base suggesting that the lysine residue is close to the active site in this enzyme as well (29). Taken together, the conserved DKY interaction is likely to be critical for normal function of EcoDam by providing additional stability to the active site. In the binary structure presented here, the DKY interaction is disrupted, and the side chain of Lys 14 in all three molecules (␣, ␤, and ␥) is disordered. Comparison of the ternary and binary structures of EcoDam suggests that in the presence of specific DNA, AdoHcy prefers the extended conformation, and Lys 14 helps to stabilize the binding pocket for the flipped base.
We studied the kinetic properties of a K14A variant to define the role of Lys 14 in base flipping. At a ratio 1:4 of enzyme (0.25 M) to substrate (1 M), DNA methylation by EcoDam showed a burst phase (first turnover) followed by a linear phase (following turnovers in steady state) (Fig. 3A). During the burst, the K14A variant methylated DNA at about a 2-fold reduced rate compared with wild type EcoDam; in addition, the magnitude of the burst was reduced 3-fold indicating some partitioning of the complex into nonproductive conformations. Under steadystate conditions (Fig. 3B), the K m value for AdoMet of the K14A variant was similar to wild type (wild type, 2.8 Ϯ 0.3 M; K14A, 2.3 Ϯ 0.1 M), which is in agreement with the structural data indicating no role of Lys 14 in AdoMet binding. However, the k cat value of the mutant was reduced by 8-fold, from 3.3 Ϯ 0.4 min Ϫ1 for the wild type to 0.4 Ϯ 0.02 min Ϫ1 . This ratio is very similar to the combined effect observed under single turnover conditions, the 2-fold reduced single turnover rate and the 3-fold reduced size of the burst (Fig. 3A). We conclude that the K14A variant has difficulty reaching the transition state of the methylation reaction that is manifested in a reduced single turnover rate and a reduced size of the burst in single turnover kinetics.
Based on the observation that Lys 14 stabilizes the base binding active site pocket in the ternary structure, we investigated if this residue has any role in the insertion of the flipped base into the pocket. Base flipping of DNA MTases can be studied using the modified base analog 2AP, which fluoresces after rotating (flipping) out of the DNA helix (30,31). Previously, we have performed fluorescence stopped-flow studies using a 20-bp DNA substrate that has 2AP in place of the target Ade residue (4,12). A biphasic fluorescence change was observed in the  (14) (PDB 1NW5 and 1NW7) and EcoDam (this study). In M.RsrI (14), the sugar configuration adopted C-1Ј exo in AdoMet, C-1Ј exo/ C-2Ј endo in AdoHcy, and C-2Ј endo in sinefungin (not shown). The superimposition shown here indicates the sugar configuration difference (if any) is very minor among the two AdoMet and two AdoHcy conformations, whereas the difference in sinefungin is obvious (14). B, tryptophan fluorescence of EcoDam and its variants observed after rapid mixing of the preincubated enzyme-DNA complexes with the cofactor. Protein and DNA were used at 350 nM, and AdoMet was at 10 M. Fluorescence was excited at 295 nm and detected Ͼ320 nm. C, cofactor binding to EcoDam and EcoDam-DNA complexes. In these experiments the enzyme was rapidly mixed with cofactor (E ϩ AdoMet; E ϩ AdoHcy), or the enzyme was preincubated with 20 bp of DNA and rapidly mixed with cofactor (E,DNA ϩ AdoMet; E,DNA ϩ AdoHcy). D, the experiment shown in C was carried out at least in triplicate at different days, and the sizes of the fluorescence changes were calculated. The figures shows the average effect Ϯ S.E. E, AdoMet binding to the EcoDam D181A variant in the absence and presence of 20 bp of oligonucleotide. The enzyme was incubated with radioactively labeled AdoMet, cross-linked by UV irradiation, and the sample split in 2 aliquots. One was run on an SDS-polyacrylamide gel and analyzed for radioactivity to determine the amount of AdoMet bound to the enzyme, and the second part was run on another SDS gel and stained with Coomassie to check for equal loading. Densitometric analysis of the radioactive bands indicated that binding to the free EcoDam D181A was at least five times weaker than binding to the EcoDam D181A-DNA complex.
presence of AdoMet, in which a fast phase of fluorescence increase (corresponding to flipping of the target base) was followed by a slower phase of fluorescence decrease. The slow phase represents the insertion of the flipped base into the active site pocket, where it regains some stacking interaction with Tyr 184 . Interestingly, although the initial rate of flipping by the K14A variant was not altered, the second slow phase of fluorescence decrease was not observed (Fig. 3C), even in experiments lasting for up to 70 s (data not shown). We conclude that loss of Lys 14 side chain in K14A prevents or at least strongly slows down the insertion of the flipped 2-aminopurine into the active site pocket.
At first glance, this result appears to be inconsistent with the moderately reduced enzymatic activity of the K14A variant. However, we note that 2AP is a base analog that lacks the exocyclic 6-amino group, the target for methylation, that interacts with the active site DPPY tetrapeptide of adenine-N 6 MTases (9, 32). The absence of this interaction between the target amino group and the DPPY as well as the presence of the additional amino group at position 2 is expected to make binding of 2AP into the active site pocket less favorable. In this regard, it should be noticed that the time scale of insertion of the flipped 2-aminopurine into the binding pocket is not in agreement with the single turnover rate constant, which is also the case for the wild type enzyme, where the rate constant of binding 2AP to the active site is smaller than 0.1 s Ϫ1 (12), whereas the single turnover rate constant is in the order of 0.15 s Ϫ1 (4). These observations point toward a limitation in using 2AP for base flipping studies as far as analyzing binding of the modified base into the active site pocket is concerned. 2AP is well suited to probe the initial flipping of the base into the extrahelical state, but the rates of insertion of the flipped base into the active site pocket might not be the same as the rates for a flipped adenine. Moreover, depending on the specificity of the active site pocket for the target base, binding of 2-aminopurine could be delayed; and the magnitude of this difference may vary among MTases.

DISCUSSION
In enzymatic catalysis binding of substrates, release of products, and conformational changes of the enzyme and substrate must be coordinated to avoid population of inactive dead-end complexes. The catalytic cycle of DNA MTases includes several steps as follows. DNA has to be bound and contacted sequence specifically, and AdoMet has to enter the active site. The target base has to be flipped out of the DNA helix into an extrahelical position and then bound into the active site pocket of the enzyme as well. The latter step should only occur if the correct DNA sequence is bound and if the active site contains AdoMet, because the base would block access of AdoMet.
We have previously shown that in EcoDam cofactor binding, DNA binding and binding of the flipped target base into the active site pocket are coupled (4,12). Here, on the basis of a new binary structure of EcoDam in complex with AdoHcy and bio- chemical studies, we describe some of the structural and mechanistic details of this process. Evidently, the N-terminal loop of EcoDam comprising Lys 9 -Lys 14 is involved in all three of these binding events; it contacts the DNA by a Lys 9 -Gua1 base-specific H-bond and Gly 12 -phosphate interaction as shown previously (4), and it contacts the cofactor by Trp 10 (as shown here) and supports binding of the flipped target base into the active site pocket by Lys 14 (as shown here). Superposition of the protein conformation in the binary structure reported here and the ternary complex with specific DNA (4) shows a small movement in the positions (ϳ2 Å) of the hexapeptide (Lys 9 to Lys 14 ) in response to specific DNA binding. In addition, the side chain of Lys 14 is disordered in the binary structure and AdoHcy occupies two conformations, one extended (similarly as seen in the ternary structure) and one folded. Our data indicate that these slight conformational changes couple the three binding events of cofactor binding, DNA binding, and binding of the flipped target base into the active site pocket of EcoDam, which are essential for enzymatic activity of EcoDam.
The conformation of the N-terminal loop seen in the binary structure is not compatible with specific DNA binding, because the phosphate group between Gua and Ade of GATC would clash with Gly 12 in the binary conformation. In the ternary complex, the peptide chain at Gly 12 and Gly 13 is moved by 2 Å and the side chain of Lys 14 is oriented toward the active site and interacts with Asp 181 and Tyr 184 (4). Therefore, the conformational change is a direct consequence of the close approach of the DNA to the enzyme in the specific complex. Unfortunately, we do not know about the structure of the hexapeptide in complex to nonspecific DNA and in free enzyme. The observation that DNA binding stimulates coenzyme binding, although the interaction between Trp 10 and the cofactor is the same in the binary structure obtained in the presence of DNA and the ternary structure, suggests that in the free enzyme the hexapeptide has a conformation different from those seen in the structures. Given the partially disordered state of the hexapeptide in the binary complex, a disordered conformation is possible in the absence of coenzyme and DNA.
Modeling an AdoMet (by adding a methyl group onto the sulfur atom of the AdoHcy in correct stereochemistry) in the extended conformation indicates that in the ternary complex the methyl group would be positioned such that its hydrogen atoms are in contact with the backbone carbonyl oxygen atoms of Ala 11 and Gly 12 (C-H⅐⅐⅐O bonds). Because the AdoMet S ϩ -CH 3 hydrogen atoms are positively polarized, the dipolar interaction between the carbonyl oxygen and the CH 3 hydrogens will stabilize the extended conformation. In contrast, in the folded conformation there are no interaction partners available for the AdoMet methyl group in the ternary conformation. Such preferential interaction with AdoMet is in agreement with the preferred binding of AdoMet as compared with AdoHcy shown here. In agreement with these considerations, in all DNA MTases AdoMet adopts the extended conformation, and in the only examples where AdoMet and AdoHcy binding can be directly compared (M.TaqI and M.RsrI) AdoMet is extended whereas AdoHcy is folded (14,33). Therefore, the extended conformation most likely is the preferred conformation for AdoMet, whereas AdoHcy can adopt both the extended and the folded conformations in the binary complex.
Modeling an AdoHcy (or AdoMet) in the folded conformation into the ternary complex conformation of EcoDam would bring the amino group of the homocysteine (or methionine) moiety into steric conflict with the ⑀-amino group of Lys 14 . This makes the extended conformation favorable if the protein is in specific complex with DNA. We have shown that the side chain of Lys 14 is important for the insertion of the flipped target base. On the basis of the structure this effect might be caused by the stabilization of the base binding pocket by Lys 14 via the bridging interaction to Tyr 184 and Asp 181 or by forcing the cofactor into the extended conformation or both. Irrespective of the exact mechanism, the conformation of Lys 14 is an important trigger that couples DNA recognition and AdoMet binding to target base insertion.
Our observation of alternative AdoHcy conformations in the binary complex might explain the earlier observation that AdoMet could bind to EcoDam in two different "environments" (11) and having different roles (10). In addition to serving as a methyl group donor, AdoMet was found to play the role of an allosteric effector, which increases the affinity of the enzyme for the DNA. This result can be interpreted on the basis of our data, because AdoMet binding will pre-structure the N-terminal loop and thereby improve DNA binding, which is thermodynamically linked to our finding that DNA binding improves AdoMet binding.