Caught in the act: visualization of an intermediate in the DNA base-fl
http://www.100md.com
《核酸研究医学期刊》
1 Department of Biochemistry and 2 Graduate Program in Biochemistry, Cell, and Development Biology, Emory University School of Medicine, 1510 Clifton Road, Atlanta, GA 30322, USA, 3 Department of Pharmaceutical Sciences, School of Pharmacy, University of Maryland, 20 Penn Street, Baltimore, MD 21201, USA, 4 Laboratory of Medicinal Chemistry, Center for Cancer Research, NCI-Frederick, 376 Boyles Street, Bldg 376, Rm 104, Frederick, MD 21702-1201, USA and 5 Isis Pharmaceuticals, Inc., 2292 Faraday Avenue, Carlsbad, CA 92008, USA
* To whom correspondence should be addressed. Tel: +1 404 727 8491; Fax: +1 404 727 3746; Email: xcheng@emory.edu
Present addresses: Nilesh K. Banavali, Department of Biochemistry and Structural Biology, Weill Medical College of Cornell University, 1300 York Avenue, New York, NY 10021, USA
Niu Huang, Department of Biopharmaceutical Sciences, UCSF, 600, 16th Street, Suite N474E, San Francisco, CA 94143, USA
PDB ID 1SKM
ABSTRACT
Rotation of a DNA or RNA nucleotide out of the double helix and into a protein pocket (‘base flipping’) is a mechanistic feature common to some DNA/RNA-binding proteins. Here, we report the structure of HhaI methyltransferase in complex with DNA containing a south-constrained abasic carbocyclic sugar at the target site in the presence of the methyl donor byproduct AdoHcy. Unexpectedly, the locked south pseudosugar appears to be trapped in the middle of the flipping pathway via the DNA major groove, held in place primarily through Van der Waals contacts with a set of invariant amino acids. Molecular dynamics simulations indicate that the structural stabilization observed with the south-constrained pseudosugar will not occur with a north-constrained pseudosugar, which explains its lowered binding affinity. Moreover, comparison of structural transitions of the sugar and phosphodiester backbone observed during computational studies of base flipping in the M.HhaI–DNA–AdoHcy ternary complex indicate that the south-constrained pseudosugar induces a conformation on the phosphodiester backbone that corresponds to that of a discrete intermediate of the base-flipping pathway. As previous crystal structures of M.HhaI ternary complex with DNA displayed the flipped sugar moiety in the antipodal north conformation, we suggest that conversion of the sugar pucker from south to north beyond the middle of the pathway is an essential part of the mechanism through which flipping must proceed to reach its final destination. We also discuss the possibility of the south-constrained pseudosugar mimicking a transition state in the phosphodiester and sugar moieties that occurs during DNA base flipping in the presence of M.HhaI.
INTRODUCTION
Many DNA-binding proteins are non-catalytic and exert their effects by binding at appropriate locations of the double helix. The result of this interaction is, in some cases, alterations in local DNA conformation. Other proteins such as DNA methyltransferases (MTases) recognize specific nucleotide sequences while simultaneously bringing the catalytic side chains into proximity of the target nucleotide(s). All catalytically active proteins balance the requirements for recognition against those for catalysis, but for enzymes that act on specific DNA sequences, this balancing act is particularly demanding. DNA MTases, which share the core structure of the great majority of S-adenosyl-L-methionine (AdoMet)-dependent MTases (1), have had to adapt to act on their giant DNA substrates. Initially, it was difficult to understand how DNA MTases acted on a target atom (C5 or N4 of cytosine or N6 of adenine) that is held at the solvent-accessible major groove surface by base pairing and stacking, and seemingly inaccessible to the concave active site pocket. The answer to this puzzle came when the crystal structure of the HhaI MTase (M.HhaI) complexed with a synthetic DNA duplex was solved (2). In a process termed ‘base flipping’, the enzyme simply rotates the target DNA base 180° (along an axis parallel to the DNA major axis) on its flanking phosphodiester bonds such that the base projects into the catalytic pocket (Figure 1A). This strategy is used by other enzymes, such as those involved in DNA base excision repair, and helps to explain the widespread use of base flipping when an enzyme needs access to an individual base in double-stranded DNA or RNA substrates (3,4).
Figure 1. Close up views of the 5'-GCGC-3'/5'-GXGC-3' sequence. The top panels show the view from the minor groove side, while the middle panels view along the helical axis. The O4*–O4* distances between Watson–Crick base-pair partners (A, B and D) and the corresponding O4*–C4* interstrand distance (C) are labelled. (A) The target X = Cyt (PDB 1FJX ), (B) X = abasic furanose sugar (PDB 9MHT ), (C) X = abasic south carbocyclic sugar (PDB 1SKM) and (D) B-DNA (PDB 1CGC ). (E–G) Chemical representations of the abasic sugar moieties substituted in the DNA backbone. (H) The omit electron density map (contoured at 3.5 above the mean, where the south sugar and both the 5' and 3' phosphates were omitted in the structure factor calculation.
Considering its biological importance, it is surprising how much remains unknown about the mechanism of base flipping, including how the movement is initiated and advanced. Previous studies with M.HhaI revealed two basic points. First, the enhanced binding affinity of M.HhaI to DNA resulting from the presence of an abasic furanose sugar (Figure 1E) at the target recognition sequence increases >3-fold when this sugar is replaced by an analogue constrained to the south conformation (Figure 1G), but decreases by a similar amount when the sugar is replaced by an antipodal analogue constrained to the north conformation (Figure 1F) (5,6). The rigid bicylohexane scaffold in these structures effectively locks the conformation of the pseudosugar moiety into either the north (3'-endo) and south (2'-endo) conformations that normally characterize the sugar moieties of standard nucleosides (6). The difference in binding affinity between the constrained south analogue (Figure 1G) and the constrained north analogue (Figure 1F) is >10-fold (7). Second, the crystal structures of M.HhaI/DNA complexed with either a flipped-out base (cytosine, adenine or uracil) or an abasic furanose sugar (PDB 9MHT ; Figure 1B) display a common sugar-phosphate backbone conformation with the target sugar moiety in the north conformation (2,8). From these studies, it is clear that the sugar–protein and phosphodiester backbone–protein interactions make significant contributions to rotation of the target base out of the DNA double helix by base-flipping enzymes. However, it seems contradictory that the north abasic pseudosugar decreases the binding affinity whereas the fully flipped-out sugar appears always in the north conformation.
To resolve this apparent discrepancy and to examine the effect of having a conformationally constrained sugar on the interaction between M.HhaI and DNA, we describe here the structure of M.HhaI in a ternary complex with AdoHcy and a 13mer non-palindromic DNA duplex containing a 5'-GSGC-3'/5'-GCGC-3', where ‘S’ is a south bicyclohexane, a pseudorotationally constrained sugar analogue (6), at the target position on one strand, whereas a normal Cyt is at the target position on the complementary strand. The ‘target position’ corresponds to that of the nucleotide that would be flipped out and methylated by M.HhaI. In the present structure, which was determined at a resolution of 2.2 ? (PDB code 1SKM), the abasic bicyclohexane sugar analogue provides an all carbon scaffold (Figure 1G). Supplementing the crystallographic data are the results from molecular dynamics (MD) simulations suggesting that the constrained south pseudosugar induces a conformation on the phosphodiester backbone that mimics the mid-point in the trajectory of the base-flipping pathway, possibly representing a transition state in the DNA phosphodiester backbone that occurs during flipping.
MATERIALS AND METHODS
Methods of expression and purification were similar to those published earlier (9). Double-stranded DNA oligonucleotides were prepared by annealing two single-stranded 13mers d(TCCATGCGCTGAC) and d(TGTCAGSGCATGG), where S is a south bicyclohexane, with one 5'-thymine overhang. The synthesis of the constrained sugar-containing strand was as described previously (6,10). New England Biolabs (Beverly, MA) synthesized the normal strand. Concentrated M.HhaI (15 μl at 20 mg/ml) was incubated with a 50% molar excess of AdoHcy at 16°C for 20 min. Oligonucleotide, also in 50% molar excess to the protein, was then added into the mixture and incubated for 2 h at 4°C. The final protein concentration was 9 mg/ml for crystallization, carried out using the hanging drop method.
A single large crystal in a droplet, under the conditions of 25% polyethylene glycol (PEG-6000), 50 mM sodium cacodylate (pH 6.5) and 75 mM magnesium acetate, was observed after weeks and kept in the mother liquor for months before being used for X-ray data collection, when the crystal was flash frozen in a cold nitrogen stream in well solution that included 25% glycerol. Data were collected at beamline X26C at National Synchrotron Light Source (crystal-to-detector distance was 210 mm, wavelength was 1.10 ?, 0.5° rotation and 240 s exposure per image, and a total 129 images were used). The crystal was in space group R32 with a unit cell a = b = 95.68 ?, c = 315.68 ?. Data were collected to 2.2 ? with an average redundancy of 2.9, = 18.4 and overall Rsymm = 6.7%. The dataset is better than 95% complete down to 2.7 ?; there the data falls off with the last shell (2.25–2.20 ?) being only 36.9% complete with a Rsymm of 0.179. The structure was solved by molecular replacement using the REPLACE program (11); refinement proceeded using CNS (12). Final Rfactor and Rfree are 18.3 and 22.1%, respectively, with the final model including 2591 protein atoms, 499 DNA atoms, 26 AdoHcy atoms and 257 waters. The estimated coordinate error from the Luzzati plot is 0.23 ?.
MD simulations used a protocol identical to the published DNA–M.HhaI ternary complex simulations (6) using the program CHARMM (13) with the CHARMM27 all-atom nucleic acid force field (14,15). Starting coordinates for the south and north carbocyclic sugar simulations were those of the new reported crystal structure. In the case of the north simulation, the sugar moiety was prepared by applying the crystallographically identified south sugar atoms directly to the north sugar. With the remaining atoms in the system fixed, the north sugar atoms were minimized in the presence of mass weighted harmonic restraints of 10 kcal/mol/? for 50 steepest descent steps, followed by removal of the harmonic restraints and minimization of the sugar for 50 conjugate gradient steps in the presence of a dihedral harmonic restraint of 1000 kcal/mol/rad on the C4'-C3'-C2'-C1' set at 30°, thereby forcing the sugar to assume the north pucker. It should be noted that the timescale of the present simulations, 2 ns, is not of adequate duration for larger scale structural changes, including dissociation of the north DNA from the protein to occur, although, as is evident, it is adequate to yield interesting observations allowing for better interpretation of the experimental data on this system.
Coordinates
Coordinates have been deposited in the Protein Data Bank (accession code 1SKM).
RESULTS
The present study confirms that M.HhaI indeed forms a stable ternary complex (DNA–protein–AdoHcy) when the target cytosine is replaced by an abasic south-constrained pseudosugar. The resulting crystal structure shows the south-constrained target pseudosugar to be trapped on the DNA major groove side by a 90° rotation about its flanking phosphodiester bonds (Figure 1C). This corresponds to the mid-point along the flipping pathway, between the non-flipped native B-DNA (0° rotation; Figure 1D) and the completely flipped state . This result helps to confirm the suggested major groove flipping pathway based on the MD free-energy calculations (16) (see Discussion). The identity of the south-constrained pseudosugar was confirmed by an omit electron density map (Figure 1H). The introduction of the constrained sugar did not grossly distort the structure of the complex. The protein component of the complex is nearly identical to that of previously solved DNA–M.HhaI–cofactor ternary structures: least squares superposition of 327 pairs of C atoms gave a root-mean-square deviation from 0.327 ? (PDB 1FJX ) to 0.721 ? (PDB 9MHT ). Significant changes in protein side chains were only observed near the constrained pseudosugar (Thr250 and Ser252) and in the active site due to omission of a flipped base.
The south pseudosugar
When base flipping occurs, the previously paired nucleotide on the complementary strand (the ‘orphan’ base) must be accommodated by M.HhaI. Residues Gln237 and Ser87, which restore the hydrogen-bonding network to the orphan guanine by penetrating into the DNA helix (Figure 2A), are in positions indistinguishable from those in each of the previously solved DNA–M.HhaI–cofactor ternary structures, indicating that flipping had been initiated. The south-constrained pseudosugar is rotated about its flanking phosphodiester bonds, 90° from its initial position in B-form DNA, but short of the completely flipped position (180° rotation). Superimposition of the DNA strand containing the constrained pseudosugar with that of 9MHT containing the unconstrained abasic furanose reveals the largest difference to involve the sugar and its associated phosphodiester bonds, particularly on the 3' side (Figure 2B). Table 1 includes selected dihedrals and the O4*–C4* interstrand distance from the new crystal structure, along with values of the corresponding O4*–O4* distances between Watson–Crick base pair partners and angles from a previous survey of flipped DNA–M.HhaI crystal structures (6) and a survey of B-DNA crystal structures (17). Four dihedrals were previously shown to differ significantly between B-DNA and the M.HhaI_survey data (‘a’ in Table 1), and the crystal structure reported here approximates three of the four changes. These include the 5' epsilon, gamma and zeta dihedrals of the target sugar. The change in the 3' beta dihedral is not seen in the new crystal structure, but resembles the value in native B-DNA. Notably, the beta dihedral for the target sugar, the dihedral angle along O5'–C5' bond (see Figure 3), has a value of 89°, which is nearly 90° away from both the M.HhaI_survey data (193 ± 4°) and the B-DNA (168°) values. This reinforces the interpretation that the south-constrained pseudosugar analogue is in a position between those characterizing B-DNA and the flipped-out state. An important internal standard is provided by the value of delta, which corresponds to the sugar pseudorotation (ring puckering) angle. In the new structure, delta is 149°, which corresponds to a south pseudorotation angle. This is the expected value for a south-constrained carbocyclic sugar and is similar to the value for B-DNA. In contrast, the delta value from the M.HhaI_survey for the flipped conformation of the unconstrained sugar is 77°, which corresponds to a north sugar pucker.
Figure 2. Stereo views of M.HhaI–DNA containing south constrained sugar analogue. Atomic bonds are depicted as green stick for DNA and grey sticks for M.HhaI. Nitrogen, oxygen, sulphur and phosphorus atoms are shown as blue, red, yellow and magenta balls, respectively. Carbon atoms are shown as green (for DNA) and grey (for protein) balls, respectively. Specific interactions are displayed as dashed single line (hydrogen bonds) and strapped lines (Van der Waals contacts). The oxygen atoms of water molecules are labelled as w. (A) Detailed structure in the vicinity of the active site and the orphaned guanine. (B) Superimposition of the DNA strands containing the south sugar from the new structure (coloured) and the M.HhaI–DNA containing an abasic furanose sugar (grey). (C) Detailed interactions between the DNA strand containing the south sugar and M.HhaI. (D) Detailed interactions between the target sugar and 5' Gua. (E) Without changing the phosphodiester bonds, superimposing the chemically modelled north pseudosugar onto the observed south conformation suggests that the specific interactions between the triangle and the side chain methyl group of Thr250 would not exist for the north sugar.
Table 1. Comparison of selected dihedrals (in degrees) and the O4*–C4* (or O4*–O4*) distance
Figure 3. Schematical representation of interactions between M.HhaI and the south sugar. The distance for specific contact is listed in Table 2. Curved arrows denote the dihedrals along the phosphodiester backbone. The letters in parentheses indicate the individual dihedral value is close to the flipped conformation (F), the B-DNA (B) or unique to the south conformation (S). Double dashed arrows indicate Van der Waals contacts and dashed lines indicate hydrogen bonds. Main chain atoms are in grey and side chain atoms in black.
Table 2. List of contacts between M.HhaI and the target sugar
M.HhaI–south sugar interactions
In the present crystal structure, we identified protein–DNA contacts closer than 4 ? as listed in Table 2 and shown schematically in Figure 3. Although the 3' and 5' phosphates flanking the south pseudosugar interact with M.HhaI similar to those flanking the abasic furanose, interactions with the pseudosugar itself are strikingly different. The south pseudosugar is buried and stabilized by a network of interactions involving Ser85 of motif IV in the active-site loop, Val121 of ENV in motif VI and the tripeptide Thr250-Leu251-Ser252 in the DNA recognition domain (Figure 2C). This network of interactions involves five distinct components. First, the side chain methyl group of Thr250 pointing perpendicularly to the triangular plane formed by C1', C4* and C6' of the cyclopropane ring. The side chain of Thr250 is rotated along the C–C? bond, in comparison with the 9MHT structure, maximizing interaction between its methyl group and the triangle of sugar carbons, and also allowing its hydroxyl to form a hydrogen bond with Arg163 (Figure 2A). In addition to the side chain conformational change of Thr250, the rotameric change of Ser252 allows its side chain hydroxyl to hydrogen bond to the 3' phosphate oxygen atom and main chain amide nitrogen of Gly255. Second, the main chain carbonyl of Leu251 is 3.1 ? away from the sugar C1', forming a C–HO=C type of hydrogen bond (18–20). Third, in addition to its interaction with the 5' phosphate oxygen, the Ser85 hydroxyl is in Van der Waals contact with C4' and C5', on the opposite side of the tripeptide Thr250-Leu251-Ser252. The interaction between C4' and Ser85 hydroxyl is unique to the south bicyclohexane, while the distance between C5' and Ser85 hydroxyl is longer with the furanose of 9MHT structure (Table 2). Furthermore, the C? atom of Ser85 interacts with the 3' Gua deoxyribose C5' atom. The two DNA atoms interacting with Ser85, C5' of the south pseudosugar and C5' of the 3' Gua, are involved in two dihedrals being either B-DNA (3' beta) or 90° from B-DNA (the target beta). Fourth, the C5' atom of the south pseudosugar is also in contact with Val121, a unique interaction to the south bicyclohexane (Table 2). Finally, besides the protein–DNA interactions, the south pseudosugar is also in contact with the deoxyribose ring of the 5' Gua. The Van der Waals interactions between C4' (south pseudosugar) and O3' (Gua), and between C4* and C3', make the C4'–C4* bond of the south pseudosugar nearly parallel to the O3'–C3' bond of the 5' Gua (Figure 2D).
Therefore, in the middle of the flipping pathway lies an extensive network of stabilizing interactions (primarily Van der Waals contacts) that trap the south-constrained pseudosugar halfway to the final flipped-out state. Emphasizing the importance of these interactions is the observation that residues Ser85 (motif IV: PCQXFS), Val121 (motif VI: ENV) and Thr250-Leu251-Ser252 of the DNA recognition domain are invariant in a large number of 5mC MTase sequences examined (21–23). This is in sharp contrast to the non-conserved state of Gln237 and Ser87, which penetrate the DNA helix. Consistent with the present observations is a previous mutagenesis study suggesting that Thr250 constrains the conformation of the DNA backbone (24).
The M.HhaI active site
In the absence of a flipped target, the active site of M.HhaI contains a few water molecules (Figure 2A). For example, water w1 interacts with the side chain of Arg165, and w2 interacts with the side chain of Glu119 and the main chain carbonyl oxygen between Phe79 and Pro80. The water site w1 corresponds to the position of O4* of the abasic furanose sugar in the 9MHT structure, while the w2 site corresponds to the exocyclic amino nitrogen N4 of the flipped cytosine (8). Interestingly, Arg165 and Glu119 form an ion pair that was present in both the M.HhaI structure without DNA (25) and in the structure with DNA containing only the abasic furanose at the flipped position (8), but was absent in all other M.HhaI–DNA complexes containing an entire nucleotide (whether cytosine, uracil or adenine) at the active site. Clearly, the presence of a flipped-out base breaks the salt bridge, resulting in both Glu119 and Arg165 interacting with the base (Glu119...N3, Arg165...O2 and Arg165...O4*). Such structural changes in the active site due to the presence of the base may contribute to stabilization of the base in the fully flipped-out orientation.
Molecular dynamics simulations
MD simulations were performed on the DNA–M.HhaI–AdoHcy ternary complex starting with the new crystal structure to facilitate an understanding of the interactions between the protein and the constrained pseudosugars and how they contribute to the enhanced binding of the south-constrained pseudosugar versus the decreased binding affinity of the north constrained pseudosugar. In the case of the constrained south pseudosugar (Figure 4A), the probability distribution for the O4*–C4* distance (peaked at 16.5 ?) stays close to the present crystallographic value (15.6 ?, thick dashed line), being 1 ? greater, while in the constrained north simulation the distances (18 ?) shifts to significantly longer values, 2.5 ? longer than the crystallographic value. Interestingly, the increase in the O4*–C4* distance with the north conformation yields better agreement with the average value (18.5 ?) from the DNA–M.HhaI crystal structures with a completely flipped conformation (thin line).
Figure 4. (A) O4*–C4* distance histograms from the MD simulations with a constrained carbocyclic south (closed circles) or north sugar (open circles) for the ternary complex initiated from the (south carbocycle)–DNA–M.HhaI–AdoHcy crystal structure. Included as vertical lines are the O4*–C4* distances from the survey of flipped DNA–M.HhaI crystal structures (6) (thin line), canonical B-DNA (6) (thick line) and the (south carbocycle)–DNA–M.HhaI crystal structure (thick dashed line). Note that C4* is the atom in the south carbocycle that corresponds to the O4* atom in the normal furanose sugar (see Figure 1E–G). (B) Sampling of the target Cyt sugar pucker pseudorotation angle or (C) beta dihedral angle versus the pseudodihedral angle defining the extent of flipping of the target Cyt (16) in the M.HhaI–DNA–AdoHcy ternary complex (2). Shown are the values of the sugar pseudorotation or beta dihedral angle sampled every 1 ps from each window in the flipping freeenergy calculation reported (16).
Consistent with the extensive contacts between the carbocyclic sugar moiety and the protein observed in the crystal structure is the stability of the O4*–C4* distance and dihedral probability distributions (see Supplementary Figure 1) in the south MD simulation. In contrast, with the north pseudosugar there are significant deviations from the crystallographic values for the O4*–C4* distance and dihedral distributions are significantly broader than those with the south-constrained pseudosugar (see Supplementary Figure 1). These results indicate that the protein cannot accommodate the north-constrained pseudosugar (e.g. see Figure 2E), contributing to its lower binding affinity. Also, the increase in the O4*–C4* distances in the north simulation is consistent with the crystallographic observations that north conformations are required for binding in the fully flipped state.
DISCUSSION
The new structure of M.HhaI with the constrained south pseudosugar at the target site of the recognition sequence provides important clues for DNA–protein interactions and the mechanism of base flipping. To see the south-constrained pseudosugar frozen in the major groove side was surprising (Figure 1C) because it was initially thought that flipping occurred through the minor groove (2). This supposition was based on the location of the M.HhaI DNA recognition domain, which approaches the DNA from the major groove side, and thus, sequence-specific interactions in the major groove were thought likely to block the major groove-flipping pathway. In addition, DNA-only simulations of abasic systems supported the minor groove pathway (6). However, the new crystal structure indicates that base flipping occurs most likely through the major groove of the DNA. Such a pathway has been suggested based on free-energy calculations of flipping in DNA complexed to M.HhaI (16).
The presence of the partially flipped pseudosugar in the major groove of the new crystal structure is consistent with the large conformational change of the M.HhaI active-site loop upon DNA binding (2). This loop moves towards the DNA from the minor groove side, where it would probably interfere with flipping through the minor groove. If flipping occurs via the minor groove before closing at the active-site loop, solvation of the base should occur in a manner similar to DNA in aqueous solution. The energetics of base flipping from DNA in aqueous solution indicate that movement of the target Cyt out of the helix and into an aqueous environment leads to a drastic increase in the free energy, thus disfavouring flipping (26,27). Therefore, it was hypothesized that the protein must supply an environment eliminating the unfavourable energetics associated with flipping into an aqueous solution (16). Such an environment may be supplied by the DNA-binding domain via a major groove-flipping pathway. It was noted earlier that M.HhaI–DNA interactions involve many main chain atoms of the DNA-binding domain, creating an intimate contact surface in the major groove (2). Of course, we cannot completely exclude the minor groove pathway, though we consider it unlikely. The south pseudosugar conformation we observe could be viewed as having flipped through the minor groove, passed through the active site (because in the absence of the Arg165–O4* bonding and other interactions the enzyme would not hold the sugar at the active site) and trapped in the observed location.
The present observations combined with structural results from previously published free-energy calculations (16) can shed additional light on the mechanism of flipping. Presented in Figure 4B is the sugar pseudorotational angle P of the target Cyt sugar as a function of the extent of flipping. As may be seen, the sugar is predominantly sampling south conformations (P = 160°) in the Watson–Crick base-paired state (Center Of Mass COM pseudodihedral angle = 10 or 370°, depending on whether flipping begins from the minor groove or the major groove, respectively). The south conformation continues to be sampled as flipping occurs via the major groove (i.e. COM Pseudodihedral angle decreasing from 370°). At 285°, which is 85° from the Watson–Crick base-paired state and hence midway to a fully flipped-out state, the sugar pucker starts to decrease and continues to decrease as flipping continues to the fully flipped state at a COM angle of 190° at which the sugar has assumed a north pucker (P 10°), consistent with the crystal structure of the flipped Cyt in the ternary complex (2). Interestingly, COM angle of 285° corresponds to the small free-energy barrier that remains to flipping in the presence of the protein . Since that barrier occurs approximately halfway between the Watson–Crick and fully flipped states and the sugar is still assuming a south conformation, consistent with the new crystal structure, it is suggested that the new crystal structure corresponds to an intermediate stage of the sugar and phosphodiester backbone moieties halfway along the flipping corridor. Consistent with this is the sampling of the target beta dihedral during flipping observed in the free-energy calculations (Figure 4C). In the vicinity of the mid-point of the flipping trajectory (COM Pseudodihedral angle = 285°), beta is sampling a wide range of values, including the value observed in the present crystal structure (90°, Table 1), while in both the Watson–Crick and fully flipped states beta is sampling values in the vicinity of 180°, consistent with that observed in B-DNA and in the M.HhaI–DNA complexes (Table 1). Thus, the stronger binding of the south abasic pseudosugar suggests that this moiety is possibly mimicking a transition state (see below) of the phosphodiester backbone that occurs during the base-flipping pathway.
The critical question is: How are the interactions between M.HhaI and the DNA phosphodiester backbone facilitating flipping? Based on surveys of crystal structures of canonical DNA there is minimal sampling of beta in the region of 90°, suggesting that this conformation is energetically unfavourable, which is supported by quantum mechanical calculations on model compounds representative of the phosphodiester backbone (17). Thus, if flipping requires a transition through values of beta near 90°, as indicated in the free-energy calculations (Figure 4C), and this region is of high energy, then the enzyme can facilitate flipping by favourably binding this orientation. As shown in Figure 3, this would involve interactions between residues Ser85, Val121, Thr250 and Ser252 and the phosphodiester backbone. Additional interactions between the carbocyclic south pseudosugar and the protein (i.e. the cyclopropane moiety with Thr250 and Leu251) stabilize the complex so that it does not move beyond the mid-point stage to the fully flipped state. Such a model is consistent with respect to interactions of an abasic furanose or north constrained carbocyclic sugar with M.HhaI. With the abasic furanose the conformational changes required for flipping can occur, leading to the fully flipped structure observed in crystal structures; however, the flexibility of the furanose moiety disallows stabilization of conformations at the mid-point of the flipping trajectory, leading to the decreased binding affinity as compared to the carbocyclic south pseudosugar. In the case of the north pseudosugar, it appears that the conformational changes required for flipping are disallowed such that favorable interactions between the enzyme and DNA cannot occur leading to the unfavourable experimental binding affinity (6). Note that in the present MD simulations the movement of the north constrained pseudosugar into a conformation beyond that of the flipped orientation is probably an artefact associated with the starting conformation being the partially flipped state from the present crystal structure (see Figure 2E).
In summary, based on the present structure it is hypothesized that M.HhaI facilitates flipping, in part, by stabilizing an energetically unfavourable conformation of the phosphodiester backbone that may represent a possible transition state in the flipping process. The idea of tight binding of transition states by enzymes being responsible for their catalytic capabilities was first proposed by Pauling (28). This concept has subsequently been applied to understand chemical catalysis by enzymes as well as being exploited in the rational design of small molecular weight enzyme inhibitors as structural mimics of transition states (30). However, this concept has been mostly applied to transition states involved in bond making or bond-breaking events, in contrast to the conformational change occurring in the present case. Similar to catalysis of chemical reactions, facilitation of base flipping involves lowering the free-energy barrier to flipping. This represents a change in the conformational dynamics of a macromolecule (i.e. DNA) responding to an external mechanical force imposed by another macromolecule (i.e. the M.HhaI enzyme). Although only a >10-fold increase in binding affinity between oligodeoxynucleotides containing conformationally locked north and south abasic pseudosugars occurs (7), versus the 3–4 order of magnitude increase in affinities generally observed for transition-state mimics of chemical reactions, it is interesting to speculate that similar transition state principles may be applied to macromolecular conformational changes that go through discrete conformational states.
SUPPLEMENTARY MATERIAL
ACKNOWLEDGEMENTS
X.C. is grateful to Dr Richard J. Roberts (New England Biolabs) and Dr Robert M. Blumenthal (Medical College of Ohio) for their continued support and encouragement. We thank Susan Sunay (Emory University) for help with purification and crystallization, Annie Heroux (Brookhaven National Laboratory) for help with X-ray data collection at beamline X26C in the facilities of the National Synchrotron Light Source. Work was supported in part by National Institutes of Health Grants GM49245 to X.C. and GM51501 to A.D.M.
REFERENCES
Schubert,H.L., Blumenthal,R.M. and Cheng,X. ( (2003) ) Many paths to methyltransfer: a chronicle of convergence. Trends Biochem. Sci., , 28, , 329–335.
Klimasauskas,S., Kumar,S., Roberts,R.J. and Cheng,X. ( (1994) ) HhaI methyltransferase flips its target base out of the DNA helix. Cell, , 76, , 357–369.
Roberts,R.J. and Cheng,X. ( (1998) ) Base flipping. Annu. Rev. Biochem., , 67, , 181–198.
Cheng,X. and Roberts,R.J. ( (2001) ) AdoMet-dependent methylation, DNA methyltransferases and base flipping. Nucleic Acids Res., , 29, , 3784–3795.
Klimasauskas,S. and Roberts,R.J. ( (1995) ) M.HhaI binds tightly to substrates containing mismatches at the target base. Nucleic Acids Res., , 23, , 1388–1395.
Wang,P., Nicklaus,M.C., Marquez,V.E., Brank,A.S., Christman,J., Banavali,N.K. and MacKerell,A.D.,Jr ( (2000) ) Use of oligodeoxyribonucleotides with conformationally constrained abasic sugar targets to probe the mechanism of base flipping by HhaI DNA (cytosine C5)-methyltransferase. J. Am. Chem. Soc., , 122, , 12422–12434.
Marquez,V.E., Wang,P., Nicklaus,M.C., Maier,M., Manoharan,M., Christman,J.K., Banavali,N.K. and Mackerell,A.D.,Jr ( (2001) ) Inhibition of (cytosine C5)-methyltransferase by oligonucleotides containing flexible (cyclopentane) and conformationally constrained (bicyclohexane) abasic sites. Nucleos. Nucleot. Nucleic Acids, , 20, , 451–459.
O'Gara,M., Horton,J.R., Roberts,R.J. and Cheng,X. ( (1998) ) Structures of HhaI methyltransferase complexed with substrates containing mismatches at the target base. Nature Struct. Biol., , 5, , 872–877.
Kumar,S., Cheng,X., Pflugrath,J.W. and Roberts,R.J. ( (1992) ) Purification, crystallization, and preliminary X-ray diffraction analysis of an M.HhaI–AdoMet complex. Biochemistry, , 31, , 8648–8653.
Maier,M., Choi,Y., Gaus,H., Marquez,V.E. and Manoharan,M. ( (2004) ) Synthesis and characterization of oligonucleotides containing conformationally constrained bicyclohexane pseudosugar analogue. Nucleic Acids Res., , 32, , 3642–3650.
Tong,L. ( (1993) ) REPLACE, a suite of computer programs for molecular-replacement calculations. J. Appl. Cryst., , 26, , 748–751.
Brunger,A.T., Adams,P.D., Clore,G.M., DeLano,W.L., Gros,P., Grosse-Kunstleve,R.W., Jiang,J.S., Kuszewski,J., Nilges,M., Pannu,N.S. et al. ( (1998) ) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr., , 54, , 905–921.
Brooks,B.R., Bruccoleri,R.E., Olafson,B.D., States,D.J., Swaminathan,S. and Karplus,M. ( (1983) ) CHARMM: a program for macromolecular energy, minimization, and dynamics calculations. J. Comput. Chem., , 4, , 187–217.
Foloppe,N. and MacKerell,A.D.,Jr ( (2000) ) All-atom empirical force field for nucleic acids: 1) Parameter optimization based on small molecule and condensed phase macromolecular target data. J. Comput. Chem., , 21, , 86–104.
MacKerell,A.D.,Jr and Banavali,N.K. ( (2000) ) All-atom empirical force field for nucleic acids: 2) Application to solution MD simulations of DNA. J. Comput. Chem., , 21, , 105–120.
Huang,N., Banavali,N.K. and MacKerell,A.D.,Jr. ( (2003) ) Protein-facilitated base flipping in DNA by cytosine-5-methyltransferase. Proc. Natl Acad. Sci. USA, , 100, , 68–73.
Foloppe,N. and MacKerell,A.D.,Jr ( (1999) ) Contribution of the phosphodiester backbone and glycosyl linkage intrinsic torsional energetics to DNA structure and dynamics. J. Phys. Chem., , B103, , 10955–10964.
Sutor,D.J. ( (1962) ) The C–HO hydrogen bond in crystals. Nature, , 195, , 68–69.
Huggins,M.T. and Lightner,D.A. ( (2001) ) A C–HO=C hydrogen bond? Intramolecular hydrogen bonding in a novel semirubin. J. Org. Chem., , 66, , 8402–8410.
Matsuura,H., Yoshida,H., Hieda,M., Yamanaka,S.-Y., Harada,T., Shin-Ya,K. and Ohno,K. ( (2003) ) Experimental evidence for intramolecular blue-shifting C–HO hydrogen bonding by matrix-isolation infrared spectroscopy. J. Am. Chem. Soc., , 125, , 13910–13911.
Posfai,J., Bhagwat,A.S., Posfai,G. and Roberts,R.J. ( (1989) ) Predictive motifs derived from cytosine methyltransferases. Nucleic Acids Res., , 17, , 2421–2435.
Lauster,R., Trautner,T.A. and Noyer-Weidner,M. ( (1989) ) Cytosine-specific type II DNA methyltransferases. A conserved enzyme core with variable target-recognizing domains. J. Mol. Biol., , 206, , 305–312.
Kumar,S., Cheng,X., Klimasauskas,S., Mi,S., Posfai,J., Roberts,R.J. and Wilson,G.G. ( (1994) ) The DNA (cytosine-5) methyltransferases. Nucleic Acids Res., , 22, , 1–10.
Vilkaitis,G., Dong,A., Weinhold,E., Cheng,X. and Klimasauskas,S. ( (2000) ) Functional roles of the conserved threonine 250 in the target recognition domain of HhaI DNA methyltransferase. J. Biol. Chem., , 275, , 38722–38730.
Cheng,X., Kumar,S., Posfai,J., Pflugrath,J.W. and Roberts,R.J. ( (1993) ) Crystal structure of the HhaI DNA methyltransferase complexed with S-adenosyl-L-methionine. Cell, , 74, , 299–307.
Giudice,E., Várnai,P. and Lavery,R. ( (2001) ) Energetic and conformational aspects of A:T base pair opening within the DNA double helix. Chem. Phys. Chem., , 11, , 673–677.
Banavali,N.K. and MacKerell,A.D.,Jr. ( (2002) ) Free energy and structural pathways of base flipping in a DNA GCGC containing sequence. J. Mol. Biol., , 319, , 141–160.
Pauling,L. ( (1946) ) Molecular architecture and biological reactions. Chem. Eng. News, , 24, , 1375–1377.
Garcia-Viloca,M., Gao,J., Karplus,M. and Truhlar,D.G. ( (2004) ) How enzymes work: analysis by modern rate theory and computer simulations. Science, , 303, , 186–195.
Wolfenden,R. ( (1976) ) Transition state analog inhibitors and enzyme catalysis. Annu. Rev. Biophys. Bioeng., , 5, , 271–306.(John R. Horton1, Gary Ratner1,2, Nilesh )
* To whom correspondence should be addressed. Tel: +1 404 727 8491; Fax: +1 404 727 3746; Email: xcheng@emory.edu
Present addresses: Nilesh K. Banavali, Department of Biochemistry and Structural Biology, Weill Medical College of Cornell University, 1300 York Avenue, New York, NY 10021, USA
Niu Huang, Department of Biopharmaceutical Sciences, UCSF, 600, 16th Street, Suite N474E, San Francisco, CA 94143, USA
PDB ID 1SKM
ABSTRACT
Rotation of a DNA or RNA nucleotide out of the double helix and into a protein pocket (‘base flipping’) is a mechanistic feature common to some DNA/RNA-binding proteins. Here, we report the structure of HhaI methyltransferase in complex with DNA containing a south-constrained abasic carbocyclic sugar at the target site in the presence of the methyl donor byproduct AdoHcy. Unexpectedly, the locked south pseudosugar appears to be trapped in the middle of the flipping pathway via the DNA major groove, held in place primarily through Van der Waals contacts with a set of invariant amino acids. Molecular dynamics simulations indicate that the structural stabilization observed with the south-constrained pseudosugar will not occur with a north-constrained pseudosugar, which explains its lowered binding affinity. Moreover, comparison of structural transitions of the sugar and phosphodiester backbone observed during computational studies of base flipping in the M.HhaI–DNA–AdoHcy ternary complex indicate that the south-constrained pseudosugar induces a conformation on the phosphodiester backbone that corresponds to that of a discrete intermediate of the base-flipping pathway. As previous crystal structures of M.HhaI ternary complex with DNA displayed the flipped sugar moiety in the antipodal north conformation, we suggest that conversion of the sugar pucker from south to north beyond the middle of the pathway is an essential part of the mechanism through which flipping must proceed to reach its final destination. We also discuss the possibility of the south-constrained pseudosugar mimicking a transition state in the phosphodiester and sugar moieties that occurs during DNA base flipping in the presence of M.HhaI.
INTRODUCTION
Many DNA-binding proteins are non-catalytic and exert their effects by binding at appropriate locations of the double helix. The result of this interaction is, in some cases, alterations in local DNA conformation. Other proteins such as DNA methyltransferases (MTases) recognize specific nucleotide sequences while simultaneously bringing the catalytic side chains into proximity of the target nucleotide(s). All catalytically active proteins balance the requirements for recognition against those for catalysis, but for enzymes that act on specific DNA sequences, this balancing act is particularly demanding. DNA MTases, which share the core structure of the great majority of S-adenosyl-L-methionine (AdoMet)-dependent MTases (1), have had to adapt to act on their giant DNA substrates. Initially, it was difficult to understand how DNA MTases acted on a target atom (C5 or N4 of cytosine or N6 of adenine) that is held at the solvent-accessible major groove surface by base pairing and stacking, and seemingly inaccessible to the concave active site pocket. The answer to this puzzle came when the crystal structure of the HhaI MTase (M.HhaI) complexed with a synthetic DNA duplex was solved (2). In a process termed ‘base flipping’, the enzyme simply rotates the target DNA base 180° (along an axis parallel to the DNA major axis) on its flanking phosphodiester bonds such that the base projects into the catalytic pocket (Figure 1A). This strategy is used by other enzymes, such as those involved in DNA base excision repair, and helps to explain the widespread use of base flipping when an enzyme needs access to an individual base in double-stranded DNA or RNA substrates (3,4).
Figure 1. Close up views of the 5'-GCGC-3'/5'-GXGC-3' sequence. The top panels show the view from the minor groove side, while the middle panels view along the helical axis. The O4*–O4* distances between Watson–Crick base-pair partners (A, B and D) and the corresponding O4*–C4* interstrand distance (C) are labelled. (A) The target X = Cyt (PDB 1FJX ), (B) X = abasic furanose sugar (PDB 9MHT ), (C) X = abasic south carbocyclic sugar (PDB 1SKM) and (D) B-DNA (PDB 1CGC ). (E–G) Chemical representations of the abasic sugar moieties substituted in the DNA backbone. (H) The omit electron density map (contoured at 3.5 above the mean, where the south sugar and both the 5' and 3' phosphates were omitted in the structure factor calculation.
Considering its biological importance, it is surprising how much remains unknown about the mechanism of base flipping, including how the movement is initiated and advanced. Previous studies with M.HhaI revealed two basic points. First, the enhanced binding affinity of M.HhaI to DNA resulting from the presence of an abasic furanose sugar (Figure 1E) at the target recognition sequence increases >3-fold when this sugar is replaced by an analogue constrained to the south conformation (Figure 1G), but decreases by a similar amount when the sugar is replaced by an antipodal analogue constrained to the north conformation (Figure 1F) (5,6). The rigid bicylohexane scaffold in these structures effectively locks the conformation of the pseudosugar moiety into either the north (3'-endo) and south (2'-endo) conformations that normally characterize the sugar moieties of standard nucleosides (6). The difference in binding affinity between the constrained south analogue (Figure 1G) and the constrained north analogue (Figure 1F) is >10-fold (7). Second, the crystal structures of M.HhaI/DNA complexed with either a flipped-out base (cytosine, adenine or uracil) or an abasic furanose sugar (PDB 9MHT ; Figure 1B) display a common sugar-phosphate backbone conformation with the target sugar moiety in the north conformation (2,8). From these studies, it is clear that the sugar–protein and phosphodiester backbone–protein interactions make significant contributions to rotation of the target base out of the DNA double helix by base-flipping enzymes. However, it seems contradictory that the north abasic pseudosugar decreases the binding affinity whereas the fully flipped-out sugar appears always in the north conformation.
To resolve this apparent discrepancy and to examine the effect of having a conformationally constrained sugar on the interaction between M.HhaI and DNA, we describe here the structure of M.HhaI in a ternary complex with AdoHcy and a 13mer non-palindromic DNA duplex containing a 5'-GSGC-3'/5'-GCGC-3', where ‘S’ is a south bicyclohexane, a pseudorotationally constrained sugar analogue (6), at the target position on one strand, whereas a normal Cyt is at the target position on the complementary strand. The ‘target position’ corresponds to that of the nucleotide that would be flipped out and methylated by M.HhaI. In the present structure, which was determined at a resolution of 2.2 ? (PDB code 1SKM), the abasic bicyclohexane sugar analogue provides an all carbon scaffold (Figure 1G). Supplementing the crystallographic data are the results from molecular dynamics (MD) simulations suggesting that the constrained south pseudosugar induces a conformation on the phosphodiester backbone that mimics the mid-point in the trajectory of the base-flipping pathway, possibly representing a transition state in the DNA phosphodiester backbone that occurs during flipping.
MATERIALS AND METHODS
Methods of expression and purification were similar to those published earlier (9). Double-stranded DNA oligonucleotides were prepared by annealing two single-stranded 13mers d(TCCATGCGCTGAC) and d(TGTCAGSGCATGG), where S is a south bicyclohexane, with one 5'-thymine overhang. The synthesis of the constrained sugar-containing strand was as described previously (6,10). New England Biolabs (Beverly, MA) synthesized the normal strand. Concentrated M.HhaI (15 μl at 20 mg/ml) was incubated with a 50% molar excess of AdoHcy at 16°C for 20 min. Oligonucleotide, also in 50% molar excess to the protein, was then added into the mixture and incubated for 2 h at 4°C. The final protein concentration was 9 mg/ml for crystallization, carried out using the hanging drop method.
A single large crystal in a droplet, under the conditions of 25% polyethylene glycol (PEG-6000), 50 mM sodium cacodylate (pH 6.5) and 75 mM magnesium acetate, was observed after weeks and kept in the mother liquor for months before being used for X-ray data collection, when the crystal was flash frozen in a cold nitrogen stream in well solution that included 25% glycerol. Data were collected at beamline X26C at National Synchrotron Light Source (crystal-to-detector distance was 210 mm, wavelength was 1.10 ?, 0.5° rotation and 240 s exposure per image, and a total 129 images were used). The crystal was in space group R32 with a unit cell a = b = 95.68 ?, c = 315.68 ?. Data were collected to 2.2 ? with an average redundancy of 2.9, = 18.4 and overall Rsymm = 6.7%. The dataset is better than 95% complete down to 2.7 ?; there the data falls off with the last shell (2.25–2.20 ?) being only 36.9% complete with a Rsymm of 0.179. The structure was solved by molecular replacement using the REPLACE program (11); refinement proceeded using CNS (12). Final Rfactor and Rfree are 18.3 and 22.1%, respectively, with the final model including 2591 protein atoms, 499 DNA atoms, 26 AdoHcy atoms and 257 waters. The estimated coordinate error from the Luzzati plot is 0.23 ?.
MD simulations used a protocol identical to the published DNA–M.HhaI ternary complex simulations (6) using the program CHARMM (13) with the CHARMM27 all-atom nucleic acid force field (14,15). Starting coordinates for the south and north carbocyclic sugar simulations were those of the new reported crystal structure. In the case of the north simulation, the sugar moiety was prepared by applying the crystallographically identified south sugar atoms directly to the north sugar. With the remaining atoms in the system fixed, the north sugar atoms were minimized in the presence of mass weighted harmonic restraints of 10 kcal/mol/? for 50 steepest descent steps, followed by removal of the harmonic restraints and minimization of the sugar for 50 conjugate gradient steps in the presence of a dihedral harmonic restraint of 1000 kcal/mol/rad on the C4'-C3'-C2'-C1' set at 30°, thereby forcing the sugar to assume the north pucker. It should be noted that the timescale of the present simulations, 2 ns, is not of adequate duration for larger scale structural changes, including dissociation of the north DNA from the protein to occur, although, as is evident, it is adequate to yield interesting observations allowing for better interpretation of the experimental data on this system.
Coordinates
Coordinates have been deposited in the Protein Data Bank (accession code 1SKM).
RESULTS
The present study confirms that M.HhaI indeed forms a stable ternary complex (DNA–protein–AdoHcy) when the target cytosine is replaced by an abasic south-constrained pseudosugar. The resulting crystal structure shows the south-constrained target pseudosugar to be trapped on the DNA major groove side by a 90° rotation about its flanking phosphodiester bonds (Figure 1C). This corresponds to the mid-point along the flipping pathway, between the non-flipped native B-DNA (0° rotation; Figure 1D) and the completely flipped state . This result helps to confirm the suggested major groove flipping pathway based on the MD free-energy calculations (16) (see Discussion). The identity of the south-constrained pseudosugar was confirmed by an omit electron density map (Figure 1H). The introduction of the constrained sugar did not grossly distort the structure of the complex. The protein component of the complex is nearly identical to that of previously solved DNA–M.HhaI–cofactor ternary structures: least squares superposition of 327 pairs of C atoms gave a root-mean-square deviation from 0.327 ? (PDB 1FJX ) to 0.721 ? (PDB 9MHT ). Significant changes in protein side chains were only observed near the constrained pseudosugar (Thr250 and Ser252) and in the active site due to omission of a flipped base.
The south pseudosugar
When base flipping occurs, the previously paired nucleotide on the complementary strand (the ‘orphan’ base) must be accommodated by M.HhaI. Residues Gln237 and Ser87, which restore the hydrogen-bonding network to the orphan guanine by penetrating into the DNA helix (Figure 2A), are in positions indistinguishable from those in each of the previously solved DNA–M.HhaI–cofactor ternary structures, indicating that flipping had been initiated. The south-constrained pseudosugar is rotated about its flanking phosphodiester bonds, 90° from its initial position in B-form DNA, but short of the completely flipped position (180° rotation). Superimposition of the DNA strand containing the constrained pseudosugar with that of 9MHT containing the unconstrained abasic furanose reveals the largest difference to involve the sugar and its associated phosphodiester bonds, particularly on the 3' side (Figure 2B). Table 1 includes selected dihedrals and the O4*–C4* interstrand distance from the new crystal structure, along with values of the corresponding O4*–O4* distances between Watson–Crick base pair partners and angles from a previous survey of flipped DNA–M.HhaI crystal structures (6) and a survey of B-DNA crystal structures (17). Four dihedrals were previously shown to differ significantly between B-DNA and the M.HhaI_survey data (‘a’ in Table 1), and the crystal structure reported here approximates three of the four changes. These include the 5' epsilon, gamma and zeta dihedrals of the target sugar. The change in the 3' beta dihedral is not seen in the new crystal structure, but resembles the value in native B-DNA. Notably, the beta dihedral for the target sugar, the dihedral angle along O5'–C5' bond (see Figure 3), has a value of 89°, which is nearly 90° away from both the M.HhaI_survey data (193 ± 4°) and the B-DNA (168°) values. This reinforces the interpretation that the south-constrained pseudosugar analogue is in a position between those characterizing B-DNA and the flipped-out state. An important internal standard is provided by the value of delta, which corresponds to the sugar pseudorotation (ring puckering) angle. In the new structure, delta is 149°, which corresponds to a south pseudorotation angle. This is the expected value for a south-constrained carbocyclic sugar and is similar to the value for B-DNA. In contrast, the delta value from the M.HhaI_survey for the flipped conformation of the unconstrained sugar is 77°, which corresponds to a north sugar pucker.
Figure 2. Stereo views of M.HhaI–DNA containing south constrained sugar analogue. Atomic bonds are depicted as green stick for DNA and grey sticks for M.HhaI. Nitrogen, oxygen, sulphur and phosphorus atoms are shown as blue, red, yellow and magenta balls, respectively. Carbon atoms are shown as green (for DNA) and grey (for protein) balls, respectively. Specific interactions are displayed as dashed single line (hydrogen bonds) and strapped lines (Van der Waals contacts). The oxygen atoms of water molecules are labelled as w. (A) Detailed structure in the vicinity of the active site and the orphaned guanine. (B) Superimposition of the DNA strands containing the south sugar from the new structure (coloured) and the M.HhaI–DNA containing an abasic furanose sugar (grey). (C) Detailed interactions between the DNA strand containing the south sugar and M.HhaI. (D) Detailed interactions between the target sugar and 5' Gua. (E) Without changing the phosphodiester bonds, superimposing the chemically modelled north pseudosugar onto the observed south conformation suggests that the specific interactions between the triangle and the side chain methyl group of Thr250 would not exist for the north sugar.
Table 1. Comparison of selected dihedrals (in degrees) and the O4*–C4* (or O4*–O4*) distance
Figure 3. Schematical representation of interactions between M.HhaI and the south sugar. The distance for specific contact is listed in Table 2. Curved arrows denote the dihedrals along the phosphodiester backbone. The letters in parentheses indicate the individual dihedral value is close to the flipped conformation (F), the B-DNA (B) or unique to the south conformation (S). Double dashed arrows indicate Van der Waals contacts and dashed lines indicate hydrogen bonds. Main chain atoms are in grey and side chain atoms in black.
Table 2. List of contacts between M.HhaI and the target sugar
M.HhaI–south sugar interactions
In the present crystal structure, we identified protein–DNA contacts closer than 4 ? as listed in Table 2 and shown schematically in Figure 3. Although the 3' and 5' phosphates flanking the south pseudosugar interact with M.HhaI similar to those flanking the abasic furanose, interactions with the pseudosugar itself are strikingly different. The south pseudosugar is buried and stabilized by a network of interactions involving Ser85 of motif IV in the active-site loop, Val121 of ENV in motif VI and the tripeptide Thr250-Leu251-Ser252 in the DNA recognition domain (Figure 2C). This network of interactions involves five distinct components. First, the side chain methyl group of Thr250 pointing perpendicularly to the triangular plane formed by C1', C4* and C6' of the cyclopropane ring. The side chain of Thr250 is rotated along the C–C? bond, in comparison with the 9MHT structure, maximizing interaction between its methyl group and the triangle of sugar carbons, and also allowing its hydroxyl to form a hydrogen bond with Arg163 (Figure 2A). In addition to the side chain conformational change of Thr250, the rotameric change of Ser252 allows its side chain hydroxyl to hydrogen bond to the 3' phosphate oxygen atom and main chain amide nitrogen of Gly255. Second, the main chain carbonyl of Leu251 is 3.1 ? away from the sugar C1', forming a C–HO=C type of hydrogen bond (18–20). Third, in addition to its interaction with the 5' phosphate oxygen, the Ser85 hydroxyl is in Van der Waals contact with C4' and C5', on the opposite side of the tripeptide Thr250-Leu251-Ser252. The interaction between C4' and Ser85 hydroxyl is unique to the south bicyclohexane, while the distance between C5' and Ser85 hydroxyl is longer with the furanose of 9MHT structure (Table 2). Furthermore, the C? atom of Ser85 interacts with the 3' Gua deoxyribose C5' atom. The two DNA atoms interacting with Ser85, C5' of the south pseudosugar and C5' of the 3' Gua, are involved in two dihedrals being either B-DNA (3' beta) or 90° from B-DNA (the target beta). Fourth, the C5' atom of the south pseudosugar is also in contact with Val121, a unique interaction to the south bicyclohexane (Table 2). Finally, besides the protein–DNA interactions, the south pseudosugar is also in contact with the deoxyribose ring of the 5' Gua. The Van der Waals interactions between C4' (south pseudosugar) and O3' (Gua), and between C4* and C3', make the C4'–C4* bond of the south pseudosugar nearly parallel to the O3'–C3' bond of the 5' Gua (Figure 2D).
Therefore, in the middle of the flipping pathway lies an extensive network of stabilizing interactions (primarily Van der Waals contacts) that trap the south-constrained pseudosugar halfway to the final flipped-out state. Emphasizing the importance of these interactions is the observation that residues Ser85 (motif IV: PCQXFS), Val121 (motif VI: ENV) and Thr250-Leu251-Ser252 of the DNA recognition domain are invariant in a large number of 5mC MTase sequences examined (21–23). This is in sharp contrast to the non-conserved state of Gln237 and Ser87, which penetrate the DNA helix. Consistent with the present observations is a previous mutagenesis study suggesting that Thr250 constrains the conformation of the DNA backbone (24).
The M.HhaI active site
In the absence of a flipped target, the active site of M.HhaI contains a few water molecules (Figure 2A). For example, water w1 interacts with the side chain of Arg165, and w2 interacts with the side chain of Glu119 and the main chain carbonyl oxygen between Phe79 and Pro80. The water site w1 corresponds to the position of O4* of the abasic furanose sugar in the 9MHT structure, while the w2 site corresponds to the exocyclic amino nitrogen N4 of the flipped cytosine (8). Interestingly, Arg165 and Glu119 form an ion pair that was present in both the M.HhaI structure without DNA (25) and in the structure with DNA containing only the abasic furanose at the flipped position (8), but was absent in all other M.HhaI–DNA complexes containing an entire nucleotide (whether cytosine, uracil or adenine) at the active site. Clearly, the presence of a flipped-out base breaks the salt bridge, resulting in both Glu119 and Arg165 interacting with the base (Glu119...N3, Arg165...O2 and Arg165...O4*). Such structural changes in the active site due to the presence of the base may contribute to stabilization of the base in the fully flipped-out orientation.
Molecular dynamics simulations
MD simulations were performed on the DNA–M.HhaI–AdoHcy ternary complex starting with the new crystal structure to facilitate an understanding of the interactions between the protein and the constrained pseudosugars and how they contribute to the enhanced binding of the south-constrained pseudosugar versus the decreased binding affinity of the north constrained pseudosugar. In the case of the constrained south pseudosugar (Figure 4A), the probability distribution for the O4*–C4* distance (peaked at 16.5 ?) stays close to the present crystallographic value (15.6 ?, thick dashed line), being 1 ? greater, while in the constrained north simulation the distances (18 ?) shifts to significantly longer values, 2.5 ? longer than the crystallographic value. Interestingly, the increase in the O4*–C4* distance with the north conformation yields better agreement with the average value (18.5 ?) from the DNA–M.HhaI crystal structures with a completely flipped conformation (thin line).
Figure 4. (A) O4*–C4* distance histograms from the MD simulations with a constrained carbocyclic south (closed circles) or north sugar (open circles) for the ternary complex initiated from the (south carbocycle)–DNA–M.HhaI–AdoHcy crystal structure. Included as vertical lines are the O4*–C4* distances from the survey of flipped DNA–M.HhaI crystal structures (6) (thin line), canonical B-DNA (6) (thick line) and the (south carbocycle)–DNA–M.HhaI crystal structure (thick dashed line). Note that C4* is the atom in the south carbocycle that corresponds to the O4* atom in the normal furanose sugar (see Figure 1E–G). (B) Sampling of the target Cyt sugar pucker pseudorotation angle or (C) beta dihedral angle versus the pseudodihedral angle defining the extent of flipping of the target Cyt (16) in the M.HhaI–DNA–AdoHcy ternary complex (2). Shown are the values of the sugar pseudorotation or beta dihedral angle sampled every 1 ps from each window in the flipping freeenergy calculation reported (16).
Consistent with the extensive contacts between the carbocyclic sugar moiety and the protein observed in the crystal structure is the stability of the O4*–C4* distance and dihedral probability distributions (see Supplementary Figure 1) in the south MD simulation. In contrast, with the north pseudosugar there are significant deviations from the crystallographic values for the O4*–C4* distance and dihedral distributions are significantly broader than those with the south-constrained pseudosugar (see Supplementary Figure 1). These results indicate that the protein cannot accommodate the north-constrained pseudosugar (e.g. see Figure 2E), contributing to its lower binding affinity. Also, the increase in the O4*–C4* distances in the north simulation is consistent with the crystallographic observations that north conformations are required for binding in the fully flipped state.
DISCUSSION
The new structure of M.HhaI with the constrained south pseudosugar at the target site of the recognition sequence provides important clues for DNA–protein interactions and the mechanism of base flipping. To see the south-constrained pseudosugar frozen in the major groove side was surprising (Figure 1C) because it was initially thought that flipping occurred through the minor groove (2). This supposition was based on the location of the M.HhaI DNA recognition domain, which approaches the DNA from the major groove side, and thus, sequence-specific interactions in the major groove were thought likely to block the major groove-flipping pathway. In addition, DNA-only simulations of abasic systems supported the minor groove pathway (6). However, the new crystal structure indicates that base flipping occurs most likely through the major groove of the DNA. Such a pathway has been suggested based on free-energy calculations of flipping in DNA complexed to M.HhaI (16).
The presence of the partially flipped pseudosugar in the major groove of the new crystal structure is consistent with the large conformational change of the M.HhaI active-site loop upon DNA binding (2). This loop moves towards the DNA from the minor groove side, where it would probably interfere with flipping through the minor groove. If flipping occurs via the minor groove before closing at the active-site loop, solvation of the base should occur in a manner similar to DNA in aqueous solution. The energetics of base flipping from DNA in aqueous solution indicate that movement of the target Cyt out of the helix and into an aqueous environment leads to a drastic increase in the free energy, thus disfavouring flipping (26,27). Therefore, it was hypothesized that the protein must supply an environment eliminating the unfavourable energetics associated with flipping into an aqueous solution (16). Such an environment may be supplied by the DNA-binding domain via a major groove-flipping pathway. It was noted earlier that M.HhaI–DNA interactions involve many main chain atoms of the DNA-binding domain, creating an intimate contact surface in the major groove (2). Of course, we cannot completely exclude the minor groove pathway, though we consider it unlikely. The south pseudosugar conformation we observe could be viewed as having flipped through the minor groove, passed through the active site (because in the absence of the Arg165–O4* bonding and other interactions the enzyme would not hold the sugar at the active site) and trapped in the observed location.
The present observations combined with structural results from previously published free-energy calculations (16) can shed additional light on the mechanism of flipping. Presented in Figure 4B is the sugar pseudorotational angle P of the target Cyt sugar as a function of the extent of flipping. As may be seen, the sugar is predominantly sampling south conformations (P = 160°) in the Watson–Crick base-paired state (Center Of Mass COM pseudodihedral angle = 10 or 370°, depending on whether flipping begins from the minor groove or the major groove, respectively). The south conformation continues to be sampled as flipping occurs via the major groove (i.e. COM Pseudodihedral angle decreasing from 370°). At 285°, which is 85° from the Watson–Crick base-paired state and hence midway to a fully flipped-out state, the sugar pucker starts to decrease and continues to decrease as flipping continues to the fully flipped state at a COM angle of 190° at which the sugar has assumed a north pucker (P 10°), consistent with the crystal structure of the flipped Cyt in the ternary complex (2). Interestingly, COM angle of 285° corresponds to the small free-energy barrier that remains to flipping in the presence of the protein . Since that barrier occurs approximately halfway between the Watson–Crick and fully flipped states and the sugar is still assuming a south conformation, consistent with the new crystal structure, it is suggested that the new crystal structure corresponds to an intermediate stage of the sugar and phosphodiester backbone moieties halfway along the flipping corridor. Consistent with this is the sampling of the target beta dihedral during flipping observed in the free-energy calculations (Figure 4C). In the vicinity of the mid-point of the flipping trajectory (COM Pseudodihedral angle = 285°), beta is sampling a wide range of values, including the value observed in the present crystal structure (90°, Table 1), while in both the Watson–Crick and fully flipped states beta is sampling values in the vicinity of 180°, consistent with that observed in B-DNA and in the M.HhaI–DNA complexes (Table 1). Thus, the stronger binding of the south abasic pseudosugar suggests that this moiety is possibly mimicking a transition state (see below) of the phosphodiester backbone that occurs during the base-flipping pathway.
The critical question is: How are the interactions between M.HhaI and the DNA phosphodiester backbone facilitating flipping? Based on surveys of crystal structures of canonical DNA there is minimal sampling of beta in the region of 90°, suggesting that this conformation is energetically unfavourable, which is supported by quantum mechanical calculations on model compounds representative of the phosphodiester backbone (17). Thus, if flipping requires a transition through values of beta near 90°, as indicated in the free-energy calculations (Figure 4C), and this region is of high energy, then the enzyme can facilitate flipping by favourably binding this orientation. As shown in Figure 3, this would involve interactions between residues Ser85, Val121, Thr250 and Ser252 and the phosphodiester backbone. Additional interactions between the carbocyclic south pseudosugar and the protein (i.e. the cyclopropane moiety with Thr250 and Leu251) stabilize the complex so that it does not move beyond the mid-point stage to the fully flipped state. Such a model is consistent with respect to interactions of an abasic furanose or north constrained carbocyclic sugar with M.HhaI. With the abasic furanose the conformational changes required for flipping can occur, leading to the fully flipped structure observed in crystal structures; however, the flexibility of the furanose moiety disallows stabilization of conformations at the mid-point of the flipping trajectory, leading to the decreased binding affinity as compared to the carbocyclic south pseudosugar. In the case of the north pseudosugar, it appears that the conformational changes required for flipping are disallowed such that favorable interactions between the enzyme and DNA cannot occur leading to the unfavourable experimental binding affinity (6). Note that in the present MD simulations the movement of the north constrained pseudosugar into a conformation beyond that of the flipped orientation is probably an artefact associated with the starting conformation being the partially flipped state from the present crystal structure (see Figure 2E).
In summary, based on the present structure it is hypothesized that M.HhaI facilitates flipping, in part, by stabilizing an energetically unfavourable conformation of the phosphodiester backbone that may represent a possible transition state in the flipping process. The idea of tight binding of transition states by enzymes being responsible for their catalytic capabilities was first proposed by Pauling (28). This concept has subsequently been applied to understand chemical catalysis by enzymes as well as being exploited in the rational design of small molecular weight enzyme inhibitors as structural mimics of transition states (30). However, this concept has been mostly applied to transition states involved in bond making or bond-breaking events, in contrast to the conformational change occurring in the present case. Similar to catalysis of chemical reactions, facilitation of base flipping involves lowering the free-energy barrier to flipping. This represents a change in the conformational dynamics of a macromolecule (i.e. DNA) responding to an external mechanical force imposed by another macromolecule (i.e. the M.HhaI enzyme). Although only a >10-fold increase in binding affinity between oligodeoxynucleotides containing conformationally locked north and south abasic pseudosugars occurs (7), versus the 3–4 order of magnitude increase in affinities generally observed for transition-state mimics of chemical reactions, it is interesting to speculate that similar transition state principles may be applied to macromolecular conformational changes that go through discrete conformational states.
SUPPLEMENTARY MATERIAL
ACKNOWLEDGEMENTS
X.C. is grateful to Dr Richard J. Roberts (New England Biolabs) and Dr Robert M. Blumenthal (Medical College of Ohio) for their continued support and encouragement. We thank Susan Sunay (Emory University) for help with purification and crystallization, Annie Heroux (Brookhaven National Laboratory) for help with X-ray data collection at beamline X26C in the facilities of the National Synchrotron Light Source. Work was supported in part by National Institutes of Health Grants GM49245 to X.C. and GM51501 to A.D.M.
REFERENCES
Schubert,H.L., Blumenthal,R.M. and Cheng,X. ( (2003) ) Many paths to methyltransfer: a chronicle of convergence. Trends Biochem. Sci., , 28, , 329–335.
Klimasauskas,S., Kumar,S., Roberts,R.J. and Cheng,X. ( (1994) ) HhaI methyltransferase flips its target base out of the DNA helix. Cell, , 76, , 357–369.
Roberts,R.J. and Cheng,X. ( (1998) ) Base flipping. Annu. Rev. Biochem., , 67, , 181–198.
Cheng,X. and Roberts,R.J. ( (2001) ) AdoMet-dependent methylation, DNA methyltransferases and base flipping. Nucleic Acids Res., , 29, , 3784–3795.
Klimasauskas,S. and Roberts,R.J. ( (1995) ) M.HhaI binds tightly to substrates containing mismatches at the target base. Nucleic Acids Res., , 23, , 1388–1395.
Wang,P., Nicklaus,M.C., Marquez,V.E., Brank,A.S., Christman,J., Banavali,N.K. and MacKerell,A.D.,Jr ( (2000) ) Use of oligodeoxyribonucleotides with conformationally constrained abasic sugar targets to probe the mechanism of base flipping by HhaI DNA (cytosine C5)-methyltransferase. J. Am. Chem. Soc., , 122, , 12422–12434.
Marquez,V.E., Wang,P., Nicklaus,M.C., Maier,M., Manoharan,M., Christman,J.K., Banavali,N.K. and Mackerell,A.D.,Jr ( (2001) ) Inhibition of (cytosine C5)-methyltransferase by oligonucleotides containing flexible (cyclopentane) and conformationally constrained (bicyclohexane) abasic sites. Nucleos. Nucleot. Nucleic Acids, , 20, , 451–459.
O'Gara,M., Horton,J.R., Roberts,R.J. and Cheng,X. ( (1998) ) Structures of HhaI methyltransferase complexed with substrates containing mismatches at the target base. Nature Struct. Biol., , 5, , 872–877.
Kumar,S., Cheng,X., Pflugrath,J.W. and Roberts,R.J. ( (1992) ) Purification, crystallization, and preliminary X-ray diffraction analysis of an M.HhaI–AdoMet complex. Biochemistry, , 31, , 8648–8653.
Maier,M., Choi,Y., Gaus,H., Marquez,V.E. and Manoharan,M. ( (2004) ) Synthesis and characterization of oligonucleotides containing conformationally constrained bicyclohexane pseudosugar analogue. Nucleic Acids Res., , 32, , 3642–3650.
Tong,L. ( (1993) ) REPLACE, a suite of computer programs for molecular-replacement calculations. J. Appl. Cryst., , 26, , 748–751.
Brunger,A.T., Adams,P.D., Clore,G.M., DeLano,W.L., Gros,P., Grosse-Kunstleve,R.W., Jiang,J.S., Kuszewski,J., Nilges,M., Pannu,N.S. et al. ( (1998) ) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr., , 54, , 905–921.
Brooks,B.R., Bruccoleri,R.E., Olafson,B.D., States,D.J., Swaminathan,S. and Karplus,M. ( (1983) ) CHARMM: a program for macromolecular energy, minimization, and dynamics calculations. J. Comput. Chem., , 4, , 187–217.
Foloppe,N. and MacKerell,A.D.,Jr ( (2000) ) All-atom empirical force field for nucleic acids: 1) Parameter optimization based on small molecule and condensed phase macromolecular target data. J. Comput. Chem., , 21, , 86–104.
MacKerell,A.D.,Jr and Banavali,N.K. ( (2000) ) All-atom empirical force field for nucleic acids: 2) Application to solution MD simulations of DNA. J. Comput. Chem., , 21, , 105–120.
Huang,N., Banavali,N.K. and MacKerell,A.D.,Jr. ( (2003) ) Protein-facilitated base flipping in DNA by cytosine-5-methyltransferase. Proc. Natl Acad. Sci. USA, , 100, , 68–73.
Foloppe,N. and MacKerell,A.D.,Jr ( (1999) ) Contribution of the phosphodiester backbone and glycosyl linkage intrinsic torsional energetics to DNA structure and dynamics. J. Phys. Chem., , B103, , 10955–10964.
Sutor,D.J. ( (1962) ) The C–HO hydrogen bond in crystals. Nature, , 195, , 68–69.
Huggins,M.T. and Lightner,D.A. ( (2001) ) A C–HO=C hydrogen bond? Intramolecular hydrogen bonding in a novel semirubin. J. Org. Chem., , 66, , 8402–8410.
Matsuura,H., Yoshida,H., Hieda,M., Yamanaka,S.-Y., Harada,T., Shin-Ya,K. and Ohno,K. ( (2003) ) Experimental evidence for intramolecular blue-shifting C–HO hydrogen bonding by matrix-isolation infrared spectroscopy. J. Am. Chem. Soc., , 125, , 13910–13911.
Posfai,J., Bhagwat,A.S., Posfai,G. and Roberts,R.J. ( (1989) ) Predictive motifs derived from cytosine methyltransferases. Nucleic Acids Res., , 17, , 2421–2435.
Lauster,R., Trautner,T.A. and Noyer-Weidner,M. ( (1989) ) Cytosine-specific type II DNA methyltransferases. A conserved enzyme core with variable target-recognizing domains. J. Mol. Biol., , 206, , 305–312.
Kumar,S., Cheng,X., Klimasauskas,S., Mi,S., Posfai,J., Roberts,R.J. and Wilson,G.G. ( (1994) ) The DNA (cytosine-5) methyltransferases. Nucleic Acids Res., , 22, , 1–10.
Vilkaitis,G., Dong,A., Weinhold,E., Cheng,X. and Klimasauskas,S. ( (2000) ) Functional roles of the conserved threonine 250 in the target recognition domain of HhaI DNA methyltransferase. J. Biol. Chem., , 275, , 38722–38730.
Cheng,X., Kumar,S., Posfai,J., Pflugrath,J.W. and Roberts,R.J. ( (1993) ) Crystal structure of the HhaI DNA methyltransferase complexed with S-adenosyl-L-methionine. Cell, , 74, , 299–307.
Giudice,E., Várnai,P. and Lavery,R. ( (2001) ) Energetic and conformational aspects of A:T base pair opening within the DNA double helix. Chem. Phys. Chem., , 11, , 673–677.
Banavali,N.K. and MacKerell,A.D.,Jr. ( (2002) ) Free energy and structural pathways of base flipping in a DNA GCGC containing sequence. J. Mol. Biol., , 319, , 141–160.
Pauling,L. ( (1946) ) Molecular architecture and biological reactions. Chem. Eng. News, , 24, , 1375–1377.
Garcia-Viloca,M., Gao,J., Karplus,M. and Truhlar,D.G. ( (2004) ) How enzymes work: analysis by modern rate theory and computer simulations. Science, , 303, , 186–195.
Wolfenden,R. ( (1976) ) Transition state analog inhibitors and enzyme catalysis. Annu. Rev. Biophys. Bioeng., , 5, , 271–306.(John R. Horton1, Gary Ratner1,2, Nilesh )