Use of oligodeoxyribonucleotides with conformationally constrained abasic sugar targets to probe the mechanism of base flipping by hhaI DNA (cytosine C5)-methyltransferase

Article abstract of DOI:10.1021/ja001989s

X-ray crystallographic studies of HhaI DNA (cytosine-C5)-methyltransferase (M.HhaI) covalently linked to methylated 5-fluorocytosine in DNA provided the first direct evidence that the cytosine residue targeted for methylation was "flipped" out of the heli

Full text of DOI:10.1021/ja001989s

12422
J. Am. Chem. Soc. 2000, 122, 12422-12434
Use of Oligodeoxyribonucleotides with Conformationally Constrained
Abasic Sugar Targets To Probe the Mechanism of Base Flipping by
HhaI DNA (Cytosine C5)-methyltransferase
Peiyuan Wang,|,† Adam S. Brank,|,‡ Nilesh K. Banavali,|,§ Marc C. Nicklaus,†
Victor E. Marquez,*,† Judith K. Christman,*,‡ and Alexander D. MacKerell, Jr.*,§
Contribution from the Laboratory of Medicinal Chemistry, DiVision of Basic Sciences,
National Cancer Institute, NIH, Bethesda, Maryland 20892, Department of Biochemistry and
Molecular Biology and UNMC/Eppley Cancer Center, 984525 UniVersity of Nebraska Medical Center,
Omaha, Nebraska 68198-4525, and Department of Pharmaceutical Sciences, School of Pharmacy,
UniVersity of Maryland, Baltimore, Maryland 21201
ReceiVed June 5, 2000
Abstract: X-ray crystallographic studies of HhaI DNA (cytosine-C5)-methyltransferase (M.HhaI) covalently
linked to methylated 5-fluorocytosine in DNA provided the first direct evidence that the cytosine residue
targeted for methylation was “flipped” out of the helix during the transfer reaction. Subsequent studies indicated
that removal of the target cytosine base, i.e., introduction of an abasic site, enhanced binding of M.HhaI to
DNA and that the conformation of the sugar-phosphate backbone at the abasic site in the resultant complexes
was the same as that of the sugar attached to a “flipped” cytosine. In the present study, pseudorotationally
constrained sugar analogues, based on bicyclo[3.1.0]hexane templates, were placed in DNA duplexes as abasic
target sites in the M.HhaI recognition sequence. Biochemical studies demonstrate that binding affinity of M.HhaI
for abasic sites increases when the abasic target sugar analogue is constrained to the south conformation and
decreases when it is constrained to the north conformation. In native gel-shift assays, M.HhaI exhibits a “closed”
conformation when bound to the abasic south or abasic furanose analogues, whereas an “open” conformation
predominates with the abasic north analogue. A structural understanding of these results was obtained via
molecular dynamics simulations of the DNA duplex alone and in ternary complex with M.HhaI and cofactor,
along with quantum mechanical calculations on model compounds representative of the abasic and modified
sugars. Binding affinities are shown to be related to the ability of the abasic sugar analogues to spontaneously
flip out of the DNA duplex. Enhanced binding of the abasic south analogue is suggested to be due to its
increased capacity for sampling the experimentally observed conformation of the DNA target site in the M.HhaI
ternary complex. Decreased binding of the north analogue is due to decreased flexibility of the phosphodiester
backbone associated with a north pseudorotation angle, thereby inhibiting spontaneous flipping of the sugar
moiety out of the DNA duplex. Spontaneous flipping of the sugar moiety out of the DNA duplex is also
suggested to facilitate formation of a “closed” complex between M.HhaI and DNA whereas partial or no
flipping favors the “open” conformation. These results show that introduction of structural constraints into
DNA that induce enhanced sampling of protein-bound conformations facilitate DNA-protein binding.
Implications of the present results with respect to the mechanism of base flipping in the M.HhaI catalytic
cycle are discussed.
Introduction
flipping process, the base is rotated completely out of the helix,
breaking the corresponding Watson-Crick base pair and
interrupting base stacking. Two mechanisms have been proposed
for this process: (1) an active process in which the enzyme
first pushes the base out of the helix and then pulls it into the
active site of the enzyme and (2) a passive process in which a
spontaneously formed extrahelical base is trapped by the protein
during the normal breathing of the DNA.8
One of the hallmarks of DNA (cytosine C5)-methyltrans-
ferases from HhaI and HaeIII is extrusion of a base from a
B-DNA helix into the active site of the enzyme by a base
flipping mechanism.1,2 Many DNA glycosylases that remove
abnormal bases from DNA have also been reported to use a
base flipping mechanism to exert their action.3-7 During the
* To whom correspondence should be addressed: (e-mail) marquez@
dc37a.nci.nih.gov; jchristm@unmc.edu; alex@outerbanks.umaryland.edu..
| The first three authors contributed equally to the present work and
should all be considered as first authors.
(3) Lau, A. Y.; Scharer, O. D.; Samson, L.; Verdine, G. L.; Lipscomb,
W. N. Cell 1998, 95, 249-258.
(4) Vassylyev, D. G.; Kashiwagi, T.; Mikani, Y.; Ariyoshi, M.; Iwai,
S.; Ohtsuka, E.; Morikawa, K. Cell 1998, 95, 773-782.
(5) Savva, R.; McAuley-Hecht, K.; Brown, T.; Pearl, L. Nature 1995,
373, 487-493.
(6) Slupphaug, G.; Mol, C. D.; Kavli, B.; Arvai, A. S.; Krokan, H. E.;
Tainer, J. A. Nature 1996, 384, 87-92.
(7) Barrett, T. E.; Savva, R.; Panayotou, G.; Barlow, T.; Brown, T.;
Jiricny, J.; Pearl, L. H. Cell 1998, 92, 117-129.
† National Cancer Institute.
‡ University of Nebraska Medical Center.
§ University of Maryland.
(1) Klimasauskas, S.; Kumar, S.; Roberts, R. J.; Cheng, X. Cell 1994,
76, 357-369.
(2) Reinisch, K. M.; Chen, L.; Verdine, G. L.; Lipscomb, W. N. Cell
1995, 82, 143-153.
10.1021/ja001989s CCC: $19.00 © 2000 American Chemical Society
Published on Web 12/01/2000

HhaI DNA (Cytosine C5)-Methyltransferase
J. Am. Chem. Soc., Vol. 122, No. 50, 2000 12423
From a comparison of NMR data and gel electrophoretic
behavior, it has been proposed that at least three protein-DNA
complexes are involved in the structural mechanism of action:
(a) an initial complex formed between the enzyme and a
normally stacked B-DNA helix, (b) an “open” complex of
enzyme and DNA which is comprised of an “ensemble of
flipped-out confomers”, and (c) a more compact or “closed”
complex in which the active site loop of the enzyme locks the
flipped-out target cytosine into the active site pocket.9 The first
two complexes appear to be the major contributors in a dynamic
equilibrium between the three states until cofactor binding shifts
the equilibrium in favor of the final complex.9 No experimentally
derived structural information is available for the first two binary
complexes although model structures have been constructed.9
Because the binding affinity of M.HhaI for its recognition
sequence in DNA is inversely related to the stability of the target
base pair, the first two complexes are proposed to be the major
determinants of binding affinity between enzyme and substrate
until cofactor binding shifts the equilibrium toward formation
of the closed complex. The tightest binding of M.HhaI occurs
when either the target base or the target nucleotide with both
its 5′- and 3′-phosphates is missing (abasic site or gap).10
Crystallographic analysis of ternary complexes containing
M.HhaI, the cofactor S-adenosylhomocysteine (AdoHcy), and
DNA containing an abasic target site revealed that the flexible
abasic furanose sugar is in the flipped-out conformation.11 The
sugar adopts a ring pucker pseudorotation angle of 33°,11
characteristic of a north conformation in the pseudorotational
cycle.12,13 In B-DNA, the sugar moieties favor pseudorotation
angles of ∼160°, characteristic of the south conformation.14
These observations suggested to us that studies of the effects
of target abasic sugars with different conformational constraints
on the interaction between M.HhaI and DNA would help to
elucidate the nature of the DNA-protein interactions and their
role in the mechanism of enzyme action.
Figure 1. Conformationally constrained location of the bicyclo[3.1.0]-
hexane templates in the sugar pseudorotation cycle.13
Scheme 1. (A) Sequence of the ds 13-Mer Oligonucleotide
Used in the Biochemical Experiments (M ) C5-Methylated
Cytosine, X ) Abasic Sugar). (B) Sequence of
Double-Stranded 12-Mer Used for MD Simulations
Marquez and co-workers have designed and synthesized
conformationally rigid nucleosides based on a bicyclo[3.1.0]-
hexane template to study the importance of ring pucker for
recognition and binding by various nucleoside and nucleotide
processing enzymes.15-31 In the present study, a logical exten-
sion of this approach was utilized to synthesize DNA fragments
(oligodeoxyribonucleotides, ODNs; see Scheme 1) containing
conformationally rigid abasic sugars (X) in place of the target
C residue in the M.HhaI recognition sequence (5′-GXGC-3′).
When the complementary strand contains a 5-methylcytosine
(M in the sequence 5′-GMGC-3′) the residue X, indicated by
an arrow in Scheme 1, becomes the only target for flipping.
Incorporation of the sugars locked in one of the two antipodal
hemispheres of the pseudorotational cycles (Figure 1) was
accomplished by using the phosphoramidites 5 and 15 (Schemes
1 and 2 of the Supporting Information). Structures of the ODNs
(8) Roberts, R. J.; Cheng, X. Annu. ReV. Biochem. 1998, 67, 181-198.
(9) Klimasauskas, S.; Szyperski, T.; Serva, S.; Wuthrich, K. EMBO J.
1998, 17, 317-324.
(10) Klimasauskas, S.; Roberts, R. J. Nucleic Acids Res. 1995, 23, 1388-
1395.
(11) O’Gara, M.; Horton, J. R.; Roberts, R. J.; Cheng, X. Nat. Struct.
Biol. 1998, 5, 872-877.
(12) Altona, C.; Sundaralingam, M. J. Am. Chem. Soc. 1972, 94, 8205-
8212.
(13) Saenger, W. Principles of Nucleic Acid Structure; Springer-
Verlag: New York, 1984.
(14) Plavec, J.; Thibadeau, C.; Chattopadhyaya, J. Pure Appl. Chem.
1996, 68, 2137-2144.
(15) Rodriguez, J. B.; Marquez, V. E.; Nicklaus, M. C.; Barchi, J. J., Jr.
Tetrahedron Lett. 1993, 34, 6233.
(24) Marquez, V. E.; Ezzitouni, A.; Russ, P.; Siddiqui, M. A.; Ford, H.,
Jr.; Feldman, R. J.; Mitsuya, H.; George, C.; Barchi, J. J., Jr. J. Am. Chem.
Soc. 1998, 120, 2780-2789.
(25) Marquez, V. E.; Ezzitouni, A.; Russ, P.; Siddiqui, M. A.; Ford, J.,
H.; Feldman, R. J.; Mitsuya, H.; George, C.; Barchi, J., J. J. Nucleosides
Nucleotides 1998, 17, 1881-1884.
(16) Rodriguez, J. B.; Marquez, V. E.; Nicklaus, M. C.; H., M.; Barchi,
J. J., Jr. J. Med. Chem. 1994, 37, 3389-3399.
(17) Ezzitouni, A.; Barchi, J. J., Jr.; Marquez, V. E. J. Chem. Soc., Chem.
Commun. 1995, 1345.
(18) Jeong, L. S.; Marquez, V. E.; Yuan, C.-S.; Borchardt, R. T.
Heterocycles 1995, 41, 2651.
(19) Siddiqui, M. A.; Ford, H., Jr.; George, C.; Marquez, V. E.
Nucleosides Nucleotides 1996, 15, 235.
(20) Jeong, L. S.; Marquez, V. E. Tetrahedron Lett. 1996, 37, 2353.
(21) Ezzitouni, A.; Marquez, V. E. J. Chem. Soc., Perkin Trans. 1 1997,
1073.
(26) Jeong, L. S.; Buenger, G.; McCormack, J. J.; Cooney, D. A.; Hao,
Z.; Marquez, V. E. J. Med. Chem. 1998, 41, 2572-2578.
(27) Bekiroglu, S.; Thibadeau, C.; Kumar, A.; Matsuda, A.; Marquez,
V. E.; Chattopadhyaya, J. J. Org. Chem. 1998, 63, 5447-5462.
(28) Moon, H. R.; Kim, H. O.; Chun, M. W.; Jeong, L. S.; Marquez, V.
E. J. Org. Chem. 1999, 64, 4733-4741.
(29) Marquez, V. E.; Russ, P.; Alonso, R.; Siddiqui, M. A.; Shin, K. J.;
George, C.; Nicklaus, M.; Dai, F.; Ford, J., H. Nucleosides Nucleotides
1999, 18, 521-530.
(30) Nandanan, E.; Jang, S.-Y.; Moro, S.; Kim, H.; Siddiqui, M. A.;
Russ, P.; Marquez, V. E.; Busson, R.; Herdewjin, P.; Harden, T. K.; Boyer,
J. L.; Jacobson, K. A. J. Med. Chem. 2000, 43, 829-842.
(31) Marquez, V. E.; Russ, P.; Alonso, R.; Siddiqui, M. A.; Hernandez,
S.; George, C.; Nicklaus, M. C.; Dai, F.; Ford, J., H. HelV. Chim. Acta
1999, 82, 2119-2129.
(22) Ezzitouni, A.; Russ, P.; Marquez, V. E. J. Org. Chem. 1997, 62,
4870.
(23) Marquez, V. E.; Ezzitouni, A.; Siddiqui, M. A.; Russ, P.; Ikeda,
H.; George, C. Nucleosides Nucleotides 1997, 16, 1431-1434.

12424 J. Am. Chem. Soc., Vol. 122, No. 50, 2000
Wang et al.
reaction mixture. After 5 min, additional Et2Zn in hexane (10.9 mL)
was added dropwise followed by the remaining half of the CH2I2
solution. The reaction mixture was stirred cold for ∼16 h and gradually
allowed to reach room temperature. It was cooled again over ice, poured
into 200 mL of aqueous saturated NH4Cl, and extracted with EtOAc
(3 × 250 mL). The combined organic extract was washed with aqueous
saturated NH4Cl, dried (MgSO4), and filtered. The filtrate was
concentrated to dryness, and the residue was purified by column
chromatography on silica gel with hexanes and EtOAc (20:1) to give
1
2 (1.51 g, 70.6%) as a syrup: [R]25 19.8° (c 0.57, MeOH); H NMR
D
(CDCl3) δ 7.36 (m, 5 H, PhH), 4.54 (s, 2 H, PhCH2O), 4.20 (d, J )
6.8 Hz, 1 H, H-3), 3.45 (dd, J ) 9.0, 6.1 Hz, 1 H, PhCH2OCHH), 3.27
(t, J ) 8.8 Hz, 1 H, PhCH2OCHH), 2.09-2.20 (m, 2 H, H-2 and H-4a),
1.77 (br s, 1 H, OH), 1.65 (dd, J ) 14.2, 1.5 Hz, 1 H, H-4b), 1.27 (m,
1 H, H-1 or H-5), 1.12 (uneven quartet, 1 H, H-5 or H-1), 0.59 (m, 2
H, H-6a,b); FAB MS m/z (relative intensity) 219 (MH+, 41), 91 (PhCH2+
,
100). Anal. Calcd for C14H18O2‚0.04H2O: C, 76.78; H, 8.32. Found:
C, 76.48; H, 8.34.
(1R,2R,3S,5R)-2-(Hydroxymethyl)bicyclo[3.1.0]hexan-3-ol (3). To
a stirred solution of Pd black (2 g) in MeOH (60 mL) was added a
solution of 2 (1.06 g, 4.9 mmol) in MeOH (30 mL). Formic acid (96%,
6 mL) was added, and the reaction mixture was stirred at 50-55 °C
for 1.5 h. The reaction mixture was filtered through Celite, and the
filtrate was concentrated to dryness. The residue was purified by column
chromatography on silica gel with 20 f 40% EtOAc in hexanes to
give 3 (0.434 g, 70%) as a solid: mp 68-69 °C; [R]25D 42.5 ° (c 0.16,
MeOH); 1H NMR (CDCl3) δ 4.24 (d, J ) 6.8 Hz, 1 H, H-3), 3.65 (dd,
J ) 10.6, 5.9, Hz, 1 H, PhCH2OCHH), 3.45 (dd, J ) 10.5, 8.3 Hz, 1
H, PhCH2OHH), 2.18 (m, 1 H, H-2), 2.05 (m, 1 H, H-4a), 1.94 (br s,
2 H, OH), 1.67 (d, 14.2 Hz, 1 H, H-4b), 1.39 and 1.22 (multiplets, 2 H,
H-1, and H-5), 0.60 (m, 2 H, H-6a,b); FAB MS of the diacetate derivative
m/z (relative intensity) 213 (MH+, 61), 153 (MH+ - AcOH, 100). Anal.
Calcd for C7H12O2: C, 65.50; H, 9.44. Found: C, 65.53; H, 9.51.
(1R,2R,3S,5R)-2-[[Bis(4-methoxyphenyl)phenylmethoxy]methyl]-
bicyclo[3.1.0]hexan-3-ol (4). A solution of 3 (0.082 g, 0.6 mmol) in
anhydrous CH2Cl2 (2.5 mL) containing DBU (0.25 mL, 1.64 mmol)
was treated with 4,4′-dimethoxytrityl chloride (0.30 g, 0.8 mmol) at 0
°C. The reaction mixture was then stirred at room temperature for 14
h. It was then concentrated, and the residue was purified by column
chromatography on silica gel with hexanes/EtOAc/Et3N (gradient from
94:4:2 to 90:8:2) to give 4 (0.25 g, 91%) as a syrup: [R]25D 18.89° (c
Figure 2. Model compounds used to study the γ-dihedral energy
surface in the abasic sugar moieties (A-C) and representations of the
abasic sugar moieties substituted in the DNA backbone (D-F).
Compound A, abasic furanose moiety; compound B, abasic north
carbocyclic moiety; compound C, abasic south carbocyclic moiety. The
dihedral degrees of freedom indicated by curved arrows in D are named
according to the standard convention for nucleic acids.13
containing sugar analogues in place of the target base X are
represented in Figure 2E and F, with the abasic furanose ODN
structure represented in Figure 2D. The stability of these ODN
duplexes was determined and compared to the control ODN by
melting temperature measurements. The interaction of the ODNs
with M.HhaI was characterized by determining their capacity
to inhibit methyl transfer from S-adenosylmethionine (AdoMet)
to target C residues in normal substrates and by nondenaturing
polyacrylamide gel-shift analysis of complexes formed between
M.HhaI and the ODNs in the presence and absence of the
cofactors, AdoHcy and AdoMet. Computational studies were
used to understand the conformational behavior of the abasic
ODNs in both aqueous solution and the protein-bound ternary
complex. The impact of the conformational properties of the
abasic sugar on interactions of M.HhaI with its substrates can
be explained on the basis of experimental and theoretical results
and is discussed within the general context of base flipping in
the M.HhaI catalytic cycle.
1
0.19, EtOAc); H NMR (CDCl3) δ 6.82-7.47 (m, 13 H, ArH), 4.19
(d, J ) 5.6 Hz, 1 H, H-3), 3.80 (s, 6 H, 2OCH3), 3.13 (dd, J ) 8.8, 6.1
Hz, 1 H, DMTOCHH), 2.93 (t, J ) 8.3 Hz, 1 H, DMTOCHH), 2.13
and 2.06 (multiplets, 2 H, H-2 and H-4a), 1.60 (dd, J ) 13.9, 1.5 Hz,
1 H, H-4b), 1.41, 1.27, and 1.14 (multiplets, 3 H, H-1, H-5, and H-6a),
0.60 (m, 1 H, H-6b); FAB MS m/z (relative intensity) 430.4 (MH•+, 6),
303 (trityl cation, 100). Anal. Calcd for C28H30O2‚0.6H2O: C, 76.20;
H, 7.13. Found: C, 76.08; H, 7.07.
Methods
(1R,2R,3S,5R)-3-[[2{[Bis(4-methoxyphenyl)phenylmethoxy]methyl]-
bicyclo[3.1.0]hex-3-yloxy][bis(methylethyl)amino]phosphinooxy}-
propanenitrile (5). To a solution of 4 (0.082 g, 0.2 mmol) in anhydrous
CH2Cl2 (5 mL) containing N,N-diisopropylethylamine (0.184 mL, 1.1
mmol) was added 2-cyanoethyl N,N-diisopropylchlorophosphoramidite
(0.118 mL, 0.5 mmol) at 0 °C. The reaction mixture was stirred at 0
°C for 1 h and then at room temperature for an additional hour. After
cooling again to 0 °C, the reaction was quenched with MeOH (1.5
mL) and concentrated to dryness under vacuum. The residue was
purified by column chromatography on silica gel with hexanes/EtOAc/
Et3N (gradient from 95:3:2 to 93:5:2) to give 5 (0.092 g, 77%) as a
syrup: FAB MS m/z (relative intensity) 303 (trityl cation, 100), 102
[(i-Pr)2NH2+, 67); 31P NMR (CDCl3) δ 144.23 and 144.34.
(1S,2S,3R,4S)-4-(2,2-Dimethyl-1,1-diphenyl-1-silapropoxy)-3-[(phen-
ylmethoxy)methyl]-2-(phenylselenamethyl)cyclopentan-1-ol (8). A
mixture of 622 (0.788 g, 1.8 mmol) and PhSeCl (0.404 g, 2.1 mmol) in
anhydrous DMSO (5 mL) was stirred under argon for 10 min at room
temperature and then cooled to 0 °C. Silver trifluoroacetate (0.465 g,
2.1 mmol) was added, and the reaction mixture was stirred at room
temperature for 12 h after which time TLC analysis (hexanes/benzene,
4:3) indicated no starting material present (Rf ) 0.51). The reaction
mixture containing crude 7 was treated with 5% NaOH in 95% EtOH
1. Synthetic Methods. All chemical reagents were commercially
available. Melting points were determined on a Mel-temp II apparatus
and are uncorrected. Column chromatography was performed on silica
gel 60, 230-400 mesh (E. Merck), analytical TLC was performed on
1
Analtech Uniplates silica gel GF. Routine IR, and H and 13C NMR
spectra were recorded using standard methods on a Bruker AC-250
instrument at 250 MHz. 31P NMR spectra were recorded on a Bruker
AMX500 using an inverse broad-band probe and were referenced to
external trimethyl phosphate. Specific rotations were measured in a
Perkin-Elmer model 241 polarimeter. Positive-ion fast-atom bombard-
ment mass spectra (FABMS) were obtained on a VG 7070E mass
spectrometer at an accelerating voltage of 6 kV and a resolution of
2000. Glycerol was used as the sample matrix, and ionization was
effected by a beam of xenon atoms. Elemental analyses were performed
by Atlantic Microlab, Inc., Norcross, GA.
(1R,2R,3S,5R)-2-[(Phenylmethoxy)methyl]bicyclo[3.1.0]hexan-3-
ol (2). A stirred solution of 123 (2 g, 9.8 mmol) in dry CH2Cl2 (81 mL)
was cooled at 0 °C and treated dropwise with Et2Zn (1 M solution in
hexane, 10.9 mL). After the addition, the reaction mixture was stirred
for 15 min. Separately, CH2I2 (1.8 mL, 22.6 mmol) was dissolved in
dry CH2Cl2 (10 mL) and half of this solution was added rapidly to the

HhaI DNA (Cytosine C5)-Methyltransferase
J. Am. Chem. Soc., Vol. 122, No. 50, 2000 12425
(9 mL) at 0 °C, stirred at room temperature for 30 min, poured into ice
water (25 mL), and extracted with Et2O (3 × 25 mL). The combined
organic extract was washed with brine, dried (MgSO4), and filtered.
The filtrate was concentrated to dryness, and the residue was purified
by column chromatography on silica gel with 5 f 8% EtOAc in
hexanes to give 8 (0.882 g, 81.5%) as a syrup: [R]25D -12.66° (c 0.79,
MeOH); 1H NMR (CDCl3) δ 7.10-7.62 (m, 20 H, Ph), 4.35 (m, 2 H,
H-2, H-1), 4.21 (s, 2 H, PhCH2O), 3.38 (dd, J ) 9.3, 3.4 Hz, 1 H,
H-4), 3.21 (t, J ) 7.8 Hz, 1 H, PhCH2OCHH), 3.09 (dd, J ) 9.3, 3.2
Hz, 1 H, PhCH2OCHH), 2.09-2.18 (m, 2 H, H-3, and H-5a), 1.76 (m,
1 H, H-5b), 1.02 (s, 9 H); FAB MS m/z (relative intensity) 617 (MH+,
2), 599 (MH+ - H2O, 7.2), 91 (PhCH2+, 100). Anal. Calcd for
C35H40O3SeSi: C, 68.27; H, 6.55. Found: C, 68.00; H, 6.52.
(1S,4S)-4-(2,2-Dimethyl-1,1-diphenyl-1-silapropoxy)-3-[(phenyl-
methoxy)methyl]cyclopent-2-en-1-ol (9). A solution of 8 (0.865 g,
1.4 mmol) and sodium metaperiodate (0.601 g, 2.8 mmol) in MeOH/
water (9:1 v/v, 30 mL) was stirred at room temperature for 22 h. The
reaction mixture was concentrated and coevaporated with EtOH to
dryness. The residue was triturated with EtOAc (60 mL), and the white
insolubles were removed by filtration. The filtrate was concentrated to
dryness, and the residue was purified by column chromatography on
silica gel with 10-15% EtOAc in hexanes to give 9 (0.470 g, 72.9%)
as a syrup [R]25D -47.50° (c 0.12, EtOAc); 1H NMR (CDCl3) δ 7.29-
7.67 (m, 15 H, PhH), 5.88 (s, 1 H, H-2), 5.04 (narrow multiplet, 1 H,
H-1), 4.86 (narrow multiplet, 1 H, 4-H), 4.47 (s, 2 H, PhCH2O), 4.10
(d, J ) 13.7 Hz, 1 H, PhCH2OCHH), 3.91 (d, J ) 13.9 Hz, 1 H, PhCH2-
OCHH), 2.11 (m, 1 H, H-5a), 1.81 (m, 1 H, H-5b), 1.50 (br s, 1 H,
OH), 1.06 (s, 9 H, tert-butyl); FAB MS m/z (relative intensity) 459
(MH+, 4), 441 (MH+ - H2O), 91 (PhCH2+, 100). Anal. Calcd for
C29H34O3Si: C, 75.94; H, 7.47. Found: C, 75.72; H, 7.54.
min. Separately, CH2I2 (0.186 mL, 2.31 mmol) was dissolved in dry
CH2Cl2 (1.7 mL), and half of this solution was added rapidly to the
reaction mixture. After 5 min, an additional amount of Et2Zn (1
M/hexane, 1.12 mL) was added dropwise followed by the remaining
half of the CH2I2 solution. The reaction mixture was stirred cold for
∼16 h and gradually allowed to reach room temperature. It was cooled
again over ice and poured into 20 mL of aqueous saturated NH4Cl and
extracted with EtOAc (3 × 25 mL). The combined organic extract was
washed with aqueous saturated NH4Cl, dried (MgSO4), and filtered.
The filtrate was concentrated to dryness, and the residue was purified
by column chromatography on silica gel with 10% EtOAc in hexanes
to give 12 (0.189 g, 88.3%) as a syrup: [R]25D 16.45 ° (c 0.93, MeOH);
1H NMR (CDCl3) δ 7.28-7.36 (m, 5 H, PhH), 4.52-4.58 (m, 3 H,
H-2, and PhCH2O), 3.76 (d, J ) 9.8 Hz, 1 H, PhCH2OCHH), 3.42 (d,
J ) 9.8 Hz, 1 H, PhCH2OCHH), 2.00-1.85 (m, 2 H, H-3a, and OH),
1.71-1.81 (m, 2 H, H-3b, and H-4a), 1.21 (m, 2 H, H-5, and H-4b),
0.94 (t, J ) 4.4 Hz, 1 H, H-6b), 0.45 (dd, J ) 7.8, 5.4 Hz, 1 H, H-6b);
FAB MS m/z (relative intensity) 219 (MH+, 21), 210 (MH+ - H2O,
17), 93 [MH+ - (PhCH2OH + H2O), 100]. Anal. Calcd for C14H18O2:
C, 77.03; H, 8.31. Found: C, 77.15; H, 8.32.
(1R, 2S, 5S)-1-(Hydroxymethyl)bicyclo[3.1.0]hexan-2-ol (13). To
a stirred suspension of Pd black (0.358 g) in MeOH (26 mL) was added
a solution of 12 (0.164 g, 0.8 mmol) in MeOH (14 mL). Formic acid
(96%, 0.9 mL) was added, and the reaction mixture was stirred at 50
°C for 1 h. After cooling to room temperature, the reaction mixture
was filtered through Celite and the filtrate was concentrated to dryness.
The residue was purified by column chromatography on silica gel with
30-50% EtOAc in hexanes to give 13 (0.077 g, 80%) as a syrup: [R]25
D
1
2.5 ° (c 0.90, MeOH); H NMR (CDCl3) δ 4.59 (t, J ) 8.2 Hz, 1 H,
H-2), 3.96 (d, J ) 10.7 Hz, 1 H, CHHOH), 3.56 (d, J ) 11.2 Hz, 1 H,
CHHOH), 1.72-2.02 (m, 3 H, H-4a, H-3a,b), 1.25 (m, 2 H, H-5, and
H-4b), 0.91 (t, J ) 4.6 Hz, 1 H, H-6a), 0.51 (dd, J ) 87.8, 5.6 Hz, 1
H, H-6b). Anal. Calcd for C7H12O2: C, 65.60; H, 9.44. Found: C, 65.77;
H, 9.48.
(5S)-[5-(2,2-Dimethyl-1,1-diphenyl-1-silapropoxy)cyclopent-1-
enyl](phenylmethoxy)methane (10). To a solution of 9 (1.426 g, 3.1
mmol) in anhydrous THF (15 mL) maintained under argon was added
sulfur trioxide-pyridine complex (0.743 g, 4.67 mmol). The suspension
was stirred at 0-3 °C for 3 h after which time analysis by TLC
(hexanes/EtOAc, 3:1) indicated that the starting material had disap-
peared (Rf ) 0.30). A solution of 1 M LiAlH4 in THF (18.66 mL, 18.6
mmol) was added at 0 °C while stirring for 1 h. The reaction mixture
was then allowed to reach 25 °C during the course of 3 h before
quenching. Quenching of the reaction was performed at 0 °C by the
successive addition of 0.7 mL of water, 0.7 mL of 15% aqueous sodium
hydroxide, and 2 mL of water. Et2O (131 mL) was added, and the
precipitate formed was filtered and washed with additional organic
solvent. The filtrate was concentrated to dryness, and the residue was
purified by column chromatography on silica gel with 20% benzene in
(1R,2S,5S)-1-{[Bis(4-methoxyphenyl)phenylmethoxy]methyl}-
bicyclo[3.1.0]hexan-2-ol (14). A solution of 13 (0.080 g, 0.6 mmol)
in anhydrous CH2Cl2 (2.5 mL) containing DBU (0.23 mL, 1.6 mmol)
was treated with 4,4′-dimethoxytrityl chloride (0.267 g, 0.8 mmol) at
0 °C. The reaction mixture was stirred for 6 h at room temperature
and then concentrated under reduced pressure. The residue was purified
by column chromatography on silica gel with hexanes/EtOAc/Et3N
(from 96:2:2 to 93:5:2) to give 14 (0.156 g, 58%) as a syrup: [R]25
D
5.60 ° (c 0.25, EtOAc); 1H NMR (CDCl3) δ 6.82-7.47 (m, 13 H, PhH),
4.49 (t, J ) 8.1 Hz, 1 H, H-2), 3.80 (s, 6 H, 2 OCH3), 3.38 (d, J ) 9.3
Hz, 1 H, DMTOCHHO), 3.07 (d, J ) 9.3 Hz, 1 H, DMTOCHHO),
2.5 (br s, 1 H, OH), 1.93 (m, 1 H, H-4a), 1.71 (m, 2 H, H-4b, and
H-3a,b), 1.11-1.30 (m, 2 H, H-4b, and H-5), 0.95 (t, J ) 4.6 Hz, 1 H,
H-6a), 0.41 (dd, J ) 7.8, 5.1 Hz, 1H, H-6b); FAB MS m/z (relative
intensity) 430 (M•+, 7), 303 (trityl cation, 100). Anal. Calcd for
C28H30O4: C, 78.11; H, 7.02. Found: C, 77.84; H, 7.18.
hexanes to give 10 (0.685 g, 50%) as a syrup: [R]25 -7.4° (c 0.50,
D
EtOAc); 1H NMR (CDCl3) δ 7.29-7.73 (m, 15 H, PhH), 5.83 (s, 1H,
H-2), 4.97 (uneven t, 1 H, H-5), 4.45 (s, 2 H, PhCH2O), 4.08 (AB m,
2 H, PhCH2OCH2), 2.33 (m, 1 H, H-3a), 2.1 (m, 1 H, H-3b), 1.94 (m,
1 H, H-4a), 1.77 (m, 1 H, H-4b), 1.09 (s, 9 H, tert-butyl); FAB MS m/z
(relative intensity) 441 (MH+ - H2, 5), 91 (PhCH2+, 100). Anal. Calcd
for C29H34O2Si: C, 78.68; H, 7.74. Found: C, 78.72; H, 7.80.
(1S)-2-[(Phenylmethoxy)methyl]cyclopent-2-en-1-ol (11). Under
an atmosphere of argon, a mixture of 10 (0.680 g, 1.5 mmol) in
anhydrous acetonitrile (23 mL) was treated with triethylamine trihy-
drofluoride (98%, 1.48 mL) and heated at reflux for 10 h. After reaching
room temperature, water (10 mL) was added and stirring was continued
for 0.5 h. The reaction mixture was reduced to dryness, dissolved in a
mixture of EtOH and benzene, and reconcentrated. The residue was
purified by column chromatography on silica gel with 9% EtOAc in
(1R,2S,5S)-3-{[1-[[Bis(4-methoxyphenyl)phenylmethoxy]methyl]-
bicyclo[3.1.0]hex-2-yloxy][bis(methylethyl)amino]phosphinooxy}-
propanenitrile (15). To a solution of 14 (0.156 g, 0.4 mmol) in
anhydrous CH2Cl2 (10 mL) containing N,N-diisopropylethylamine (0.38
mL, 2 mmol) was added 2-cyanoethyl N,N-diisopropylchlorophos-
phoramidite (0.242 mL, 1.0 mmol) at 0 °C. The reaction mixture was
stirred at 0 °C for 1 h and then at room temperature for an additional
hour. After the reaction was quenched with MeOH (1.5 mL) at 0 °C,
it was concentrated to dryness under vacuum. The residue was purified
by column chromatography on silica gel with hexanes/EtOAc/Et3N
hexanes to give 11 (0.304 g, 100%) as a syrup: [R]25 5.56° (c 0.14,
(from 96:2:2 to 93:5:2) to give 15 (0.096 g, 42%) as a syrup: FAB
D
+
MeOH); 1H NMR (CDCl3) δ 7.27-7.40 (m, 5 H, PhH), 5.86 (d, 1 H,
J ) 0.7 Hz, H-3), 4.87 (narrow multiplet, 1 H, H-1), 4.46 (s, 2 H,
PhCH2O), 4.21 (AB q, J ) 10.9 Hz, 2 H, PhCH2OCH2), 2.52 (m, 1 H,
H-4a), 2.30 (m, 2 H, H-4b, and H-5a), 2.15 (br s, 1 H, OH), 1.80 (m, 1
H, H-5b). Anal. Calcd for C13H16O2‚0.15H2O: C, 75.44; H, 7.94.
Found: C, 75.29; H, 7.85.
MS m/z (relative intensity) 303 (trityl cation, 100), 102 [(i-Pr)2NH2
,
15); 31P NMR (CDCl3) δ 144.75 and 145.59.
5′-TGT CAG XGC ATG G-3′ (ODN-south, X ) South Bicyclo-
[3.1.0]hexane Template, Scheme 1A). This oligo (15 OD units, 458
µg) was synthesized by Oligos Etc. Inc., Wilsonville, OR. The purity
of the material was estimated to be >90.0% by capillary electrophoresis.
5′-TGT CAG XGC ATG G-3′ (ODN-north, X ) North Bicyclo-
[3.1.0]hexane template, Scheme 1A). This oligo (2.8 OD units, 85.5
µg) was synthesized by Oligos Etc. Inc. The purity of the material was
estimated to be >90.0% by capillary electrophoresis.
(1R, 2S, 5S)-1-[(Phenylmethoxy)methyl]bicyclo[3.1.0]hexan-2-ol
(12). A stirred solution of 11 (0.200 g, 1.0 mmol) in dry CH2Cl2 (8
mL) was cooled at 0 °C and treated dropwise with Et2Zn (1 M/hexane,
1.12 mL). After the addition, the reaction mixture was stirred for 15

12426 J. Am. Chem. Soc., Vol. 122, No. 50, 2000
Wang et al.
2. Biochemical Methods. Binding Assays. For native gel-shift
assays and determination of dissociation of M.HhaI-ODN complexes,
reaction mixtures contained 30 nM 32P-end-labeled duplex ODN, 30
nM M.HhaI, 75 ng of poly(dAdT:dAdT) to inhibit nonspecific binding,
and 100 µM cofactor (as noted) in binding buffer (50 mM Tris (pH
7.5), 10 mM EDTA, 5 mM 2-mercaptoethanol, and 13% (v/v) glycerol).
Complexes formed during a 30-min incubation at 22 °C were separated
from free ODNs by electrophoresis at 150 V for 2.5-3 h on 10%
polyacrylamide gels that had been pre-electrophoresed at 100 V for 1
h in TBE. To test complex stability, 100-fold molar excess of the same
unlabeled ODN used in the initial binding reaction was added, and
incubation continued for 1 min-5 days to determine the time required
to exchange 50% of radiolabel in the complex (T1/2). Gels were held at
a constant temperature of 14 °C throughout electrophoresis. Dried gels
were exposed to Phosphor Imager (Molecular Dynamics) screens, and
the resulting images were used to quantitate radiolabeled ODNs in free
and complexed form.
Inhibition Assays. Reaction mixtures (50 µL in 50 mM Tris (pH
7.5), 10 mM EDTA, and 5 mM 2-mercaptoethanol) contained 8.6 nM
M.HhaI, 2.43 µM [3H]-AdoMet (specific activity 15 Ci/mmol), 640
nM ds-ODN substrate, and increasing concentrations (0, 10, 25, 65, or
165 nM) of ds-ODNs containing furan, north- or south-bicyclo[3.1.0]-
hexane pseudosugars at the target site. The substrate DNA, which
contains a single M.HhaI recognition site, was formed from fully
methylated ODN A (AMp) and unmethylated A′ by heating equimolar
amounts of the two strands to 90 °C for 10 min and then slowly cooling
to 50 °C for 60 min32 (AMp, 5′-ATTGMGCATTCMGGATCMGM-
GATC-3′, and A′, 3′-TAACGCGTAAGGCCTAGGCGCTAG-5′- M.H-
haI; recognition site in boldface type, target underlined, M )
5-methylcytosine). The double-stranded (ds) ODNs with modified
sugars were formed from ODN-south, ODN-north, or ODN-furan
(5′TGTCAGXGCATGG3′, M.HhaI site in boldface type, X indicates
substituted target site) and its complement, 5′CCATGMGCTGACA3′.
Reactions were incubated at 37 °C for 5 min and terminated by the
addition of NaOH to a concentration of 0.4 M. DNA was processed
for quantitation of [3H]-methyl incorporation as previously described.32
Highly purified M.HhaI, used for all experiments reported here, was
the generous gift of Dr. Xiaodong Cheng.33
Melting Temperature Determinations. ODN-north, ODN-south,
ODN-furan, or ODN-C (cytosine in target site; see below) and the
complementary ODN containing 5-mC at position 6 were taken up in
0.3 × SSC. The final concentration of each ODN was 1.5 ( 0.05 µM.
The solutions were heated at 37 °C for 60 min and slowly cooled.
Initial OD measurements indicated that these conditions allowed full
annealing of the duplex. A Perkin-Elmer UV-visible spectrophotometer
Lambda 10 was used to monitor changes in absorbance with each 0.1
°C change in temperature from 25 to 80 °C. Data from four cycles of
melting and annealing were obtained for each of the four duplexes using
PE Temp Lab software. Tm determinations were made from each cycle
by graphic analysis of plots and confirmed by analysis of first-derivative
data.
3. Computational Methods. Ab initio calculations were carried out
using the Gaussian 9434 or Gaussian 9835 programs with the HF/6-
31+G* basis set on model compounds A-C shown in Figure 2A-C.
The definitions of the various dihedrals in the DNA backbone of the
abasic systems are illustrated in Figure 2D. Model compounds A-C
were chosen to obtain the γ dihedral energy surface in the three abasic
sugar moieties included in this study. This surface was obtained by
constraining the C3′-C4′-C5′-O5′ dihedral in 30° increments from
0 to 360°. In all cases, the following additional constraints were
included: C-O-P-O5′ ) 262°, O-P-O5′-C5′ ) 298°, P-O5′-
C5′-C4′ ) 168°, and C4′-C3′-O3′-H ) 187° while the remainder
of the molecule was allowed to relax. These constraints correspond to
modal values of the ú, R, â, and ꢀ crystallographic dihedral distributions
in B-DNA36 and are introduced to prevent variations in the other
dihedral degrees of freedom from interfering with the γ energy surface,
as previously performed.37 For the abasic furanose, additional constraints
were necessary to sample specific regions of the sugar psuedorotation
surface: the C3′-C4′-O4′-C1′ dihedral was constrained to 0° to
sample the C2′ endo conformation favored in B-DNA and the C4′-
O4′-C1′-C2′ was constrained to 0° to sample the C3′ endo conforma-
tion favored in A-DNA.37,38
Molecular mechanics force field calculations were carried out using
the CHARMM program,39,40 and the analysis of DNA parameters was
done using the FREEHELIX program41 modified to read CHARMM-
generated MD trajectories. The CHARMM27 parameters for DNA36,42
and the CHARMM22 parameters for proteins43 were used for all
calculations along with the CHARMM-modified TIP3P water model44,45
and the sodium parameters from Beglov and Roux.46 New parameters
had to be developed for the north and south carbocyclic sugar moieties.
This was performed based on ab initio data on various model
compounds representative of carbocyclic and abasic furanose sugar
moieties in the context of their substitution into DNA.47 The new
parameters are included with the Supporting Information. Unique
parameters for AdoHcy were generated by direct transfer of related
parameters from the nucleic acid and protein sets.
MD simulations were carried out on two types of systems: (1)
systems containing DNA alone in aqueous solution and (2) the partially
solvated protein-DNA-cofactor ternary complex. The DNA and
ternary complex MD simulations were each performed on four systems
consisting of the DNA dodecamer sequence 5′-CCATGCGCTGAC-3′
(Scheme 1B, note the omission of the 3′-terminal position base pair in
these calculations), hemimethylated at the position 6 cytosine of strand
1 (M in Scheme 1). The four systems consisted of (1) an unmodified
cytosine base, (2) an abasic furanose sugar, (3) an abasic north
(35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M.
W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.;
Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, Revision A.6
ed.: Pittsburgh, PA, 1998.
(36) Foloppe, N.; MacKerell, A. D., Jr. J. Comput. Chem. 2000, 21, 86-
104.
(37) Foloppe, N.; MacKerell, A. D., Jr. J. Phys. Chem. B 1999, 103,
10955-10964.
(38) Foloppe, N.; MacKerell, A. D., Jr. J. Phys. Chem. B 1998, 102,
6669-6678.
(39) Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J.,
Swaminathan, S., Karplus, M. J. J. Comput. Chem. 1983, 4, 187-217.
(40) MacKerell, A. D., Jr.; Brooks, B.; Brooks, C. L., III.; Nilsson, L.;
Roux, B.; Won, Y.; Karplus, M. CHARMM: The Energy Function and Its
Paramerization with an OVerView of the Program; Schleyer, P. v. R.;
Allinger, N. L.; Clark, T.; Gasteiger, J.; Kollman, P. A.; Schaefer, H. F.,
III; Schreiner, P. R., Eds.; John Wiley & Sons: Chichester, 1998; Vol. 1,
pp 271-277.
(41) Dickerson, R. E. Nucleic Acids Res. 1998, 26, 1906-1926.
(42) MacKerell, A. D., Jr.; Banavali, N. J. Comput. Chem. 2000, 21,
105-120.
(32) Christman, J. K.; Sheikhnejad, G.; Marasco, C. J.; Sufrin, J. R. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 7347-7351.
(33) Kumar, S.; Cheng, X.; Pflugrath, J. W.; Roberts, R. J. Biochemistry
1992, 31, 8648-8653.
(34) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Raghavachari, K.; Al-
Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski,
J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala,
P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts,
R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart,
J. J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, C.3
ed.; Gaussian, Inc.: Pittsburgh, PA, 1996.
(43) MacKerell, A. D., Jr.; Bashford, D.; Bellott, M.; Dunbrack, R. L.,
Jr.; Evanseck, J.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; Joseph,
D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo,
T.; Nguyen, D. T.; Prodhom, B.; Reiher, I., W. E.; Roux, B.; Schlenkrich,
M.; Smith, J.; Stote, R.; Straub, J.; Watanabe, M.; Wiorkiewicz-Kuczera,
J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998, 102, 3586-3616.
(44) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.;
Klein, M. L. J. Chem. Phys. 1983, 79, 926-935.
(45) Reiher, W. E., III. Theoretical Studies of Hydrogen Bonding; Harvard
University: Cambridge, MA, 1985.
(46) Beglov, D.; Roux, B. J. Chem. Phys. 1994, 100, 9050-9063.
(47) Banavali, N. K.; MacKerell, A. D., Jr., manuscript in preparation.

HhaI DNA (Cytosine C5)-Methyltransferase
J. Am. Chem. Soc., Vol. 122, No. 50, 2000 12427
carbocyclic sugar analogue, and (4) an abasic south carbocyclic sugar
analogue, all at position 6 of strand 2 (X in Scheme 1B) of the DNA.
Systems 2, 3, and 4 correspond to Figure 2D, E, and F, respectively.
The four DNA MD simulation systems will be referred to henceforth
as ODN-C, ODN-furan, ODN-north, and ODN-south, respectively. The
corresponding four protein-DNA-cofactor systems will be referred
to as the ternary-C, ternary-furan, ternary-north, and ternary-south
systems, respectively.
the water oxygen atoms using the miscellaneous mean field potential
(MMFP) module46 with an offset distance of 23.5 Å, a well depth of
the potential of -0.25 kcal/mol, and the p1 parameter of the potential
set to 2.25. The system was first minimized for 500 steepest descent
(SD) steps with mass-weighted harmonic constraints of 2.0 kcal mol-1
Å-1 on the remaining unconstrained non-hydrogen DNA, protein, and
AdoHcy atoms. The minimized system was then subjected to a 20-ps
NVT MD simulation keeping the above-mentioned harmonic constraints
in order to equilibrate the solvent around the complex. The resulting
system was minimized for 500 SD steps without the additional
constraints, and the final structure was used to initiate the production
trajectories. Production MD simulations were performed for 2 ns in
the NVT ensemble at 300 K with the Leapfrog integrator using the
Nose-Hoover method for temperature scaling with the magnitude of
coupling, qref, set to 100.0.53,54 All calculations were performed using
SHAKE51 to constrain covalent bonds containing hydrogen, an integra-
tion time step of 0.002 ps, and treating long-range electrostatic
interactions using atom truncation. Atom truncation was performed by
using the force shift and force switch methods to truncate the
electrostatic and LJ terms, respectively.55 Nonbond interactions were
heuristically updated, truncated at 12 Å with the switching function
initiated at 10 Å, and the nonbond interaction lists maintained to 14 Å.
Analysis of the MD simulations was carried out on the last 1500 ps
of the simulations, allowing 500 ps of equilibration time, unless
otherwise stated. Averages and standard errors were calculated by
dividing the data into five blocks of 300 ps each and carrying out
statistical analysis on the block average values. The division of the
data into individual 300-ps blocks is done with the assumption that
each block is an independent sample of the MD trajectory and allows
for calculation of statistical parameters that are not just representative
of the fluctuations in the trajectory.56
C1′-C1′ and C4′-C4′ interstrand distances have previously been
used to monitor the flipping-out transition in order to distinguish
between sugar movements and backbone phosphate movements.57 These
two distances were observed to be highly correlated in the flipping out
transition in that study. Since the O4′ atom connects the C1′ and C4′
atoms in the deoxyribose sugar, we chose the sugar O4′-O4′ interstrand
distance between strand 1, residue 7 and strand 2, residue 6 as the
single parameter to measure the extent of flipping. In the north and
south sugars, the carbon atom used instead of the O4′ atom for the
calculation of this distance corresponds to the C* atom shown in Figure
2. This distance will henceforth be referred to as the sugar O4′-O4′
distance. Residue 6 of strand 2 of the DNA is referred to as the flipped
position with its sugar moiety being referred to as the flipped sugar.
The residues adjacent to the flipped position on the 5′ side (strand 2,
residue 5) and the 3′ side (strand 2, residue 7) are referred to as the 5′
position and the 3′ position, respectively.
The abasic furanose and north carbocyclic and south carbocyclic
sugars were generated from the canonical B-form of DNA48 or from
the DNA in the experimental ternary structure11 by deletion of the
cytosine base at position 6 of strand 2 and substitution with a hydrogen
atom. For the conformationally constrained north and south carbocyclic
sugars, the O4′ atom of the furan ring of the sugar was substituted
with a carbon atom, and a cyclopropane ring was generated at the
appropriate position as illustrated in Figure 2E and F.
Simulations of the DNA systems were initiated from canonical
B-form DNA overlaid with a preequilibrated solvent box consisting of
water and sodium ions, which extended at least 8.0 Å beyond the DNA
solute. All solvent molecules having a non-hydrogen atom within 1.8
Å of the DNA were deleted. The number of sodium ions was adjusted
to ensure electrostatic neutrality of the system by adding sodium ions
at random positions in the box or deleting the sodium ions furthest
from the DNA, as required. Periodic boundary conditions were used
in all subsequent calculations with the images generated using the
CRYSTAL module.49 The system was minimized for 500 Adopted
Basis Newton-Raphson (ABNR) steps with mass-weighted harmonic
constraints of 2.0 kcal mol-1 Å-1 on the non-hydrogen DNA atoms.
The minimized system was then subjected to a 20-ps constant volume,
isothermal (NVT) ensemble MD simulation keeping the same harmonic
constraints in order to equilibrate the solvent around the DNA. The
resulting system was minimized for 500 ABNR steps without any
constraints, and the final structure was used to initiate the production
trajectories. Production MD simulations were performed for 2 ns in
the isobaric, isothermal (NPT) ensemble50 at 300 K with the Leapfrog
integrator. All calculations were performed using SHAKE51 to constrain
covalent bonds containing hydrogen, using an integration time step of
0.002 ps, and treating long-range electrostatic interactions using the
particle mesh Ewald (PME) approach.52 PME calculations were
performed using real space and Lennard-Jones (LJ) interaction cutoffs
of 10 Å, with nonbond interaction lists maintained and heuristically
updated out to 12 Å. The fast Fourier transform grid densities were set
to ∼1 Å-1 using a fourth-order smoothing spline. The screening
parameter (κ) was set to 0.35.
The experimental ternary abasic furanose M.HhaI-DNA-AdoHcy
complex structure11 was used as the starting point for the three abasic
ternary complex simulations. The ternary-C simulation was initiated
from the experimental ternary structure that has the cytosine base in
the flipped orientation.1 For all the ternary simulations, the starting
structures were overlaid with a preequilibrated sodium-water sphere
of radius 25 Å which was centered with respect to the two terminal
C1′ atoms on strand 1 of the DNA. All solvent molecules having a
non-hydrogen atom within 1.8 Å of DNA and 2.5 Å of protein non-
hydrogen atoms were deleted. The number of sodiums was adjusted to
yield a neutral system as mentioned above. Constraints applied
throughout the subsequent calculations were as follows. Protein residues
with one or more atoms outside the solvent sphere were fixed, and
protein residues with one or more atoms in the concentric region of
the solvent sphere extending from 21 to 25 Å radius were harmonically
constrained using a 2.0 kcal mol-1 Å-1 mass-weighted force constant.
A quartic spherical solvent boundary potential used to minimize artificial
boundary conditions at the surface of the water sphere was applied to
Results
1. Synthetic Studies. (1S,2R)-2-[(Benzyloxy)methyl]cyclo-
pent-3-enol (1) was selected as the chiral starting material for
the synthesis of the conformationally locked south- and north-
type phosphoramidites 5 and 15 (See text and Schemes 1 and
2 of the Supporting Information). Phosphoramidites 5 and 15
were used in a conventional DNA synthesizer to synthesize the
two 13-mers containing the south (ODN-south) and north (ODN-
north) abasic sites. The sequence of the target oligos synthesized
corresponds to 5′-TGT CAG XGC ATG G-3′, where X is either
the south or the north bicyclo[3.1.0]hexane template, respec-
tively (Scheme 1A, strand 2). The coupling yields for the
modified abasic pseudosugars were rather low; however, suf-
ficient material was obtained to perform the necessary biological
(48) Arnott, S.; Hukins, D. W. L. J. Mol. Biol. 1973, 81, 93-105.
(49) Field, M. J.; Karplus, M. CRYSTAL: Program for Crystal Calcula-
tions in CHARMM; Harvard University: Cambridge, MA, 1992.
(50) Feller, S. E.; Zhang, Y.; Pastor, R. W.; Brooks, R. W. J. Chem.
Phys. 1995, 103, 4613-4621.
(51) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys.
1977, 23, 327-341.
(52) Darden, T. A.; York, D.; Pedersen, L. G. J. Chem. Phys. 1993, 98,
10089-10092.
(53) Nose´, S. J. Chem. Phys. 1984, 81, 511-519.
(54) Hoover, W. G. Phy. ReV. 1985, A31, 1695.
(55) Steinbach, P. J.; Brooks, B. R. J. Comput. Chem. 1994, 15, 667-
683.
(56) Loncharich, R. J.; Brooks, B. R.; Pastor, R. W. Biopolymers 1992,
32, 523.
(57) Barsky, D.; Foloppe, N.; Ahmadia, S.; Wilson, D. M., III; MacKerell,
A. D., Jr. Nucleic Acids Res. 2000, 28, 2613-2626.

12428 J. Am. Chem. Soc., Vol. 122, No. 50, 2000
Wang et al.
Table 1. Melting Temperatures and IC50 Values for the Inhibition
of Methyltransferase Activity of M.HhaI by Oligodeoxynucleotides
Containing a Cytosine (control) or Abasic Target
melting tempa (°C)
IC50 value (nM)
1. control
2. furan
3. north
4. south
56.1 (0.0)
37.1 (0.2)
39.6 (0.3)
38.2 (0.5)
ndb
48
∼180
14
b
a Standard deviations in parentheses. nd, not determined as this
ODN is a substrate.
experiments. Both double-stranded ODNs gave similar melting
curves with Tm values comparable to a standard abasic double-
stranded ODN containing tetrahydrofuran at the abasic site
(Table 1 and Methods).
2. Biochemical Studies. Two approaches were used to
determine the effect of conformational constraint of an abasic
target site on the function of M.HhaI. The first was to compare
the inhibitory capacity of abasic ds-ODNs-north, -south, and
-furan. The second was to directly compare the extent to which
M.HhaI could bind to each of the ds-ODNs in the presence
and absence of cofactor and to determine whether the M.HhaI
in binary or ternary complexes took on an “open” or “closed”
conformation. Inhibition curves for M.HhaI methyltransferase
activity (Figure 1 of Supporting Information) showed that ODN-
south has the highest inhibitory activity with an IC50 value of
14 nM, followed by ODN-furan with an IC50 value of 48 nM
(Table 1). ODN-north shows the least inhibitory activity. It has
no detectable inhibitory activity at concentrations below 75 nM
and an estimated IC50 value of ∼180 nM.
Figure 3. Comparison of the formation of binary (no cofactor) and
ternary (with AdoMet or AdoHcy) complexes between M.HhaI and
ODNs-furan (lanes 2-4), -north (lanes 6-8), and -south (lanes 10-
12). All ODNs were radiolabeled on the methylated complement strand
so that the amount of the different ODNs bound into complexes could
be directly compared without need to adjust for variations in efficiency
of radiolabeling. Lanes 1, 5, and 9 show the migration of the ds-ODNs
in the absence of enzyme. Lanes 2, 6, and 10 are minus cofactor. Lanes
3, 7, and 11 are +AdoMet and lanes 4, 8, and 12 are +AdoHcy. Position
of “open” and “closed” complexes is indicated by arrowheads. Details
for complex formation and electrophoretic separation are given in
Methods.
ODN binding to M.HhaI was assessed by incubating the
enzyme with each ODN (-north, -south, and -furan) in the
absence (binary complex) or presence of cofactor, AdoMet or
AdoHcy (ternary complex), for 30 min. The complexes with
bound ODN were separated from free ODNs by polyacrylamide
gel electrophoresis under nondenaturing conditions (Figure 3).
Since the same 32P-radiolabeled complementary strand was used
to form each of the ds-ODNs used in this study, the amount of
complex formation with each type of ds-ODN can be compared
directly by autoradiographic quantitation of the amount of
radiolabel associated with the complex. The inference that
M.HhaI bound to DNA is in an “open” or “closed” conformation
is made on the basis of the complex’s electrophoretic mobility
in the gel.9 Ternary complexes between M.HhaI, ds-ODN-C
(cytosine target) and AdoHcy migrate faster than binary
complexes formed without cofactor.10,58 The more rapidly
migrating complex is presumed to be more compact because it
has a structure corresponding to the crystallographically deter-
mined structure of ternary complexes between M.HhaI, ds-
ODN-C, and AdoHcy. In this complex, two regions of the
enzyme, the flexible active site loop (amino acid residues 80-
99) and the small domain recognition loop (amino acid residues
231-238), have undergone conformational changes relative to
their positions in the X-ray crystal structure of M.HhaI with
cofactor alone. The loops “close” around the DNA strand with
the completely flipped target base (or abasic sugar moiety),
occupying its position in the helix and “locking” it into the active
site pocket. The more slowly migrating binary complex is
presumed to be in a less compact “open” or nonlocked
conformation.9
suggesting that the active site loop can assume a conformation
similar to that observed in the crystal structure of the ternary
complex.1 Complexes containing ODN-north bound to M.HhaI
display the unique characteristic of binding primarily as an
“open” more slowly migrating conformation under all three
conditions (lanes 6-8). Although a small proportion (∼10%)
of the more rapidly migrating complex is detectable, its
formation is not enhanced in the presence of cofactor. This
suggests that cofactor cannot facilitate the conformational
change of the flexible, active site loop of the enzyme around
the helical structure formed by ODN-north. However, the
enzyme appears capable of adopting a closed conformation upon
binding to a small population of north abasic targets that have
achieved a fully flipped conformation despite the restraints
imposed by the constrained north pseudosugar (see below).
ODN-north also differs from ODN-furan and ODN-south in
its binding capacity. In either the presence or absence of
cofactors, binding of M.HhaI to ODN-south or ODN-furan in
a closed conformation was quantitative (∼98% input ) 100-
(ODN bound/ODN bound + ODN free) with equimolar
concentration of enzyme and ODN. Under the same conditions,
54% of ODN-north was bound in the “open” complex in the
absence of cofactor, 55% in the presence of AdoMet, and 47%
in the presence of AdoHcy. Binding to ODN-south and ODN-
furan is essentially irreversible. In the absence of cofactor, it
takes more than 110 h for 50% dissociation of ODN-south and
55 h for ODN-furan to dissociate from M.HhaI when incubated
in the presence of 100-fold excess of cold competitor ODNs.
In contrast, the dissociation of the 10% of closed binary complex
formed by ODN-north has a T1/2 of ∼14 h. under the same
conditions. The T1/2 for the “open” complex of ODN-north is
too small to be measurable (data not shown).
Interestingly, both ODN-south and ODN-furan bind to
M.HhaI and form a “closed” complex regardless of whether
cofactor is present or not (Figure 3, lanes 2-4 and 10-12),
(58) Mi, S.; Alsono, D.; Roberts, R. J. Nucleic Acids Res. 1995, 23, 620-
627.

HhaI DNA (Cytosine C5)-Methyltransferase
J. Am. Chem. Soc., Vol. 122, No. 50, 2000 12429
Table 2. Averages of Bend Angles between Base Pair 2 and Base
Pair 11 Normal Vectors and RMSD from B-DNA and A-DNA in
the Final 1500 ps of DNA System Simulationsa
mean bend angle
(deg)
RMSDb (Å)
from B-DNA
RMSDb (Å)
from A-DNA
1. control
2. furan
3. north
4. south
18.1 (0.4)
16.6 (0.8)
20.1 (0.8)
21.4 (1.4)
3.3 (0.1)
2.9 (0.1)
3.2 (0.1)
3.1 (0.1)
4.8 (0.1)
4.4 (0.1)
4.2 (0.1)
4.2 (0.0)
a Standard errors in parentheses. b RMSD, root-mean-square devia-
tions.
It should be noted that the binding affinity of these ODNs
for M.HhaI (south > furan . north) is directly related to the
inhibitory activity of the three abasic ODNs (south > furan .
north). However, it is not directly related to their melting
temperatures, which are ranked north > south > furan (Table
1). Furthermore, the melting temperatures of all three abasic
ODNs are reduced by ∼18 ( 1.4 °C relative to ODNs with the
same structure containing a cytosine base as target. It is unlikely
that a 1.4 °C difference in melting temperature between ODN-
south and ODN-furan would decrease the T1/2 for dissociation
only 2-fold while the same difference in melting temperature
between ODN-north and ODN-furan would decrease the T1/2
from more than 50 h for dissociation of ODN-furan complexes
to less than 1 min for dissociation of the bulk of ODN-north
complexes. The results are consistent with formation of a highly
stable “closed” complex with ODN-south and ODN-furan that
leads to rapid enzyme inactivation. They also suggest that some
intrinsic property of the north-constrained abasic sugar, other
than its influence on the stability of the helix, is responsible
for the observed preponderance of the relatively unstable “open”
complex of enzyme with ODN-north.
Computational studies of the ODNs in aqueous solution were
undertaken to determine how constraining the abasic sugar
conformation influences the structural and dynamic properties
of the DNA, thereby leading to the biochemical properties
discussed in the preceding section. Calculations on the ternary
complexes were also performed to investigate the impact of the
constrained pseudosugars when the ODN is bound to M.HhaI.
DNA-protein interactions are also expected to impact the
biochemical results.
3. Computational Studies. 3.1. DNA in Aqueous Solution.
Substitution of an abasic sugar into the backbone has been
shown to cause minor overall structural deformation of the
double-helical form of the DNA.57 To identify possible large-
scale structural deformations in the present systems, two
parameters were assessed: (a) the bend angles between the
penultimate base pairs of the ODNs and (b) the root-mean-
square difference (RMSD) values with respect to canonical
B-DNA and A-DNA for non-hydrogen ODN atoms (Table 2).
There were no significant differences between these parameters
for the ODN-abasic systems and the ODN-C system. This
indicates that the absence of a base in the abasic systems does
not cause significant overall distortion of the DNA duplex. Even
the presence of a conformationally constrained A-DNA-like
north carbocyclic sugar in the otherwise B-DNA-like duplex
does not cause any significant bending in the DNA. The RMSD
values also indicate that the DNA adopts a form intermediate
to the canonical B-DNA and A-DNA forms, consistent with
previous studies.42
Figure 4. Probability distribution of sugar O4′-O4′ distances in the
four DNA system MD simulations. The lines and symbols used for
the different systems are as follows: control DNA system, line; furan
DNA system, dashed line; north DNA system, (b); south DNA system,
(0). Sugar O4′-O4′ distance in angstroms; bold vertical line, furan
ternary crystal structure sugar O4′-O4′ distance; dashed vertical line,
canonical B-DNA structure sugar O4′-O4′ distance.
MD simulations; in the north and south sugars the carbon-
labeled C* in Figure 2 is used instead of the O4′ atom (see
Methods). Figure 4 shows the probability distribution of the
sugar O4′-O4′ distances observed for the four ODNs along
with vertical lines indicating the O4′-O4′ distance in canonical
B-DNA and the abasic DNA-M.HhaI complex. It is clear that
the sugar which shows the maximum propensity for flipping
out of the helix is that in ODN-furan followed by ODN-south.
At certain time points, however, the ODN-south flips out to a
similar extent as the completely flipped ODN-furan (i.e., an
O4′-O4′ distance of ∼18 Å). ODN-north, on the other hand,
shows a lesser tendency to flip out and dwells predominately
inside the duplex structure. The ODN-north does, however, show
short excursions to a partially flipped-out state (O4′-O4′
distance ∼15 Å) without ever flipping out completely. ODN-
C, with a sharp peak centered around ∼12.8 Å, shows no
tendency to flip out spontaneously on the observed time scale
of the simulation, consistent with experimental studies showing
a millisecond time scale of spontaneous base flipping.59,60
To identify the local phosphodiester dihedrals that may impact
flipping, phosphodiester dihedrals were obtained from the
flipped region of the DNA for nine DNA-methyltransferase
crystal structures.1,2,10,61-64 The average values over those nine
structures with the target site in the flipped-out form along with
B-DNA modal values are presented in Figure 5 along with a
diagram showing the corresponding dihedrals in DNA. Com-
parisons of flipped and B-form dihedrals show the differences
to be localized to a few torsions. These are presented in boldface
type in Figure 5. For the flipped sugar, there are significant
differences in γ and ú. Beyond the flipped sugar position,
significant changes also occur in the adjacent 5′-ꢀ and 3′-â
dihedrals. On the basis of the similarity of these differences in
(59) Gue´ron, M.; Kochoyan, M.; Leroy, J.-L. Nature 1987, 328, 89-
92.
(60) Moe, J. G.; Russu, I. M. Biochemistry 1992, 31, 8421-8428.
(61) O’Gara, M.; Klimasauskas, S.; Roberts, R. J.; Cheng, X. J. Mol.
Biol. 1996, 261, 634-645.
(62) O’Gara, M.; Roberts, R. J.; Cheng, X. J. Mol. Biol. 1996, 263, 597-
606.
(63) Kumar, S.; Horton, J. R.; Jones, G. D.; Walker, R. T.; Roberts, R.
J.; Cheng, X. Nucleic Acids Res. 1997, 25, 2773-2783.
(64) Sheikhnejad, G.; Brank, A.; Christman, J. K.; Goddard, A.; Alvarez,
E.; Ford Jr., H.; Marquez, V. E.; Marasco, C. J.; Sufrin, J. R.; O’Gara, M.;
Cheng, X. J. Mol. Biol. 1999, 285, 2021-2034.
Flipping out of the target sugar from the interior of the
canonical DNA duplex structure is critical to the methylating
activity of M.HhaI.1 The extent of flipping was estimated by
measuring the sugar O4′-O4′ interstrand distance during the

12430 J. Am. Chem. Soc., Vol. 122, No. 50, 2000
Wang et al.
Figure 6. Potential energy as a function of the γ-dihedral in model
compounds. Symbols are as follows: abasic furanose compound A with
the constrained south sugar conformation, dashed line; abasic furanose
compound A with the constrained north sugar conformation, bold line;
north carbocyclic compound B (b); and south carbocyclic compound
C (0). Dihedral angles in degrees.
Figure 5. Averages of flipped region backbone dihedrals and modal
B-DNA values. The diagram on the right indicates the position of these
dihedrals along the backbone with the dihedral degrees of freedom
denoted by curved arrows. Significantly distorted dihedrals indicated
in boldface type with a following asterisk. Note: Distortion in the
δ-dihedral of the flipped sugar is associated with the pseudorotation
angle of the sugar. The pseudorotation angle properties are discussed
later and therefore the δ-dihedral is not indicated in boldface type here.
The individual dihedral values for the experimental structures, from
which the averages shown here were obtained, are listed in Table 1 of
the Supporting Information. Standard errors are in parentheses.
ab initio quantum mechanical calculations on model compounds
A, B, and C (Figure 2). Potential energy profiles for γ for the
abasic furanose compound A, north carbocyclic compound B,
and south carbocyclic compound C are shown in Figure 6. Due
to the flexibility of the furanose sugar moiety and to allow for
systematic analysis of the effect of sugar conformation on the
behavior of the γ-dihedral, an additional ring constraint was
applied to explicitly sample the C2′ endo and C3′ endo ring
conformations in compound A. Comparison of the four surfaces
shows the chemical structure of the sugar moiety to significantly
influence the energetic properties. Common to all surfaces are
the three minima that occur around 60, 180, and 300°,
corresponding to the gauche+, (g+), trans (t), and gauche- (g-)
states, respectively. The g+ minimum corresponds to the
conformation observed in experimental DNA duplex struc-
tures.13 Of note is the energy of the g- minimum in compound
B (north), which is ∼6 kcal/mol greater than that of both
compound A (furan) with the south constraint and compound
C (south). Interestingly, the high energy of the g- minimum
(∼4 kcal/mol greater than the g+ minimum) is also observed in
compound A when the sugar moiety is constrained to be north,
indicating that the observed high energy of the g- state is closely
associated with the north sugar conformation. The elevated
energy g- minimum in compound B is suggested to disallow
the transition to the g- γ-conformation in ODN-north. In the
MD simulations, ODN-furan and ODN-south, which show
significant flipping, exhibit sampling of the g- state of the
γ-dihedral (γ-dihedral value ∼300° in Figure 2 of the Supporting
Information). ODN-C and ODN-north, which do not show
significant flipping of the sugar, do not show sampling of the
g- state. On the basis of these observations, the transition to
the g- state of the γ-dihedral may contribute to the flipping
process. The inability of the north carbocyclic sugar to attain
the g- state due to its intrinsic energetics is suggested to hinder
flipping of the north carbocyclic sugar out of the DNA duplex.
The role of the north conformation of the sugar on the g- state
is supported by the high energy of that state when the sugar in
compound A is constrained to the north conformation. This is
consistent with results from the MD simulation where the ODN-
furan sugar conformation remains south during the flipping
process (not shown), thereby avoiding the high energy g- state
associated with the north conformation.
the nine structures, as evidenced by the small standard errors
(see Table 1 of the Supporting Information for primary data),
these specific dihedrals would be expected to have the largest
impact on the flipping event.
The pathway of flipping in all three abasic DNA systems
was through the minor groove, as confirmed by visual inspection
of intermediate conformations during the flipping process (not
shown). The three ODNs, however, show significant differences
in the sequence of changes in the phosphodiester backbone
dihedral angles asssociated with flipping of the sugar. Analysis
of the O4′-O4′ distance time series for the flipped sugars along
with the time series for the phosphodiester backbone dihedrals
that show significant distortions in the crystal structures (Figure
2 of Supporting Information) indicates differences that are
consistent with the survey results. For ODN-furan and ODN-
south, changes are observed in γ; however, significant differ-
ences in the time series are also present in the other dihedrals,
indicating that flipping occurs via different pathways. No
corresponding dihedral transitions occur in ODN-C or ODN-
north, consistent with the lack of significant flipping of the sugar
in these systems. Thus, the dihedral angles showing alterations
associated with sugar flipping in the simulations are consistent
with those observed to be distorted in experimental crystal
structures. Furthermore, on the basis of the pseudorotation angle
time series (not shown), no significant change in sugar pucker
from a south conformation is required for flipping.
Of the differences in phosphodiester backbone dihedrals
observed both experimentally and in the calculations, it may
be assumed that the structure of the sugar will impact the
inherent conformational properties of the dihedrals closest to
the sugar moiety. On the basis of this assumption, changes in
the inherent conformational properties of the γ-dihedral of the
flipped sugar were suspected to influence the different flipping
properties of the sugar in the three abasic simulations.
Investigation of the influence of sugar structure on the
conformational properties of the γ-dihedral was performed via
Probability distributions of the four dihedrals that differ from
canonical values in the crystal structures (see Figure 5) for the

HhaI DNA (Cytosine C5)-Methyltransferase
J. Am. Chem. Soc., Vol. 122, No. 50, 2000 12431
Table 3. Average Interaction Distances (Å) between Gln237 and
Ser87 or Guanine 7 from the Ternary Simulationsb
interaction
control
furan
north
south
Gln237 Nꢀ2 to 4.05 (0.15) 4.81 (0.07) 5.20 (0.13) 3.25 (0.02)
Ser87 Oγ
Gln237 N to
Gua7 O6
2.89 (0.01) 2.83 (0.00) 3.03 (0.02) 2.85 (0.01)
Gln237 Oꢀ1 to 2.85 (0.01) 2.88 (0.01) 2.87 (0.00) 2.83 (0.00)
Gua7 N1
Gln237 Oꢀ1 to 3.09 (0.02) 3.31 (0.03) 3.05 (0.03) 3.31 (0.02)
Gua7 N2
b Standard errors in parentheses.
mental average. For the flipped position ú-dihedral, ternary-C
and ternary-south show a preference for the g+ state but with a
lower value than experimental average values. It is to be noted
that ternary-south does show some sampling of a state inter-
mediate to the g+ and t states of the ú-dihedral. In ternary-
furan, the ú-dihedral populates both the g- and g+ states. In
ternary-north, the ú-dihedral occupies the t state exclusively.
For all four dihedrals selected, ternary-south shows the best
agreement with the experimental flipped-out average values.
Previous studies have implicated a role of Gln237 in
stabilizing the ternary complex.58 This stabilization by Gln237
is facilitated by interactions with the “orphan” guanine base
and with the flexible active site loop via residue Ser87. Analysis
of these interactions from the ternary MD simulations is
presented in Table 3 as average interaction distances. Consistent
with the biochemical binding data, the shortest interaction
distance involving Ser87 occurs with ternary-south and the
longest with ternary-north, with the ternary-furan intermediate.
Of note is the ternary-C distance being shorter that the ternary-
furan distance, suggesting that the increased flexibility of this
abasic furan may destabilize this particular interaction. Concern-
ing interactions between the guanine base and Gln237, the
interaction distances are similar in all four systems, indicating
the lack of the target base and the different sugar structures to
not significantly impact these interactions.
3.3. Pseudorotation Angle. The flipped-out abasic sugar
adopts a north conformation in the crystal structure,11 in contrast
to the south conformation characteristic of B-DNA. Therefore,
this pseudorotation behavior may be related to the conforma-
tional preference of the flipped-out sugar after conclusion of
the flipping event. Probability distributions of the pseudorotation
angle of the flipped position for all MD simulations were
determined (Figure 5 of the Supporting Information). For the
DNA simulations, it is clear that the south conformation is
generally preferred in ODN-C, ODN-furan, and ODN-south.
In ODN-north, the pseudorotation angle of the abasic sugar
moiety exclusively populates the north conformation, consistent
with the constrained nature of the carbocyclic sugar. In ODN-
south, the sugar pseudorotation angle shows minor sampling
of the north form due to a boat-chair transformation of the
bicyclo[3.1.0]hexane ring system. The north form of the south
carbocyclic sugar is very short-lived as expected from ab initio
studies and experimental observations.15-31,47 The wide distribu-
tion of the pseudorotation angle in ODN-furan is indicative of
the abasic furanose ring being more flexible than the abasic
north and south carbocyclic pseudosugar rings. Overall, the
sugar moiety primarily samples the south form, consistent with
B-form structures, except with the north DNA system where
the chemical structure of the carbocyclic moiety enforces the
north conformation.
Figure 7. Probability distribution of sugar O4′-O4′ distances in the
four ternary system MD simulations. The symbols for the different
systems are as follows: control ternary system, line; furan ternary
system, dashed line; north ternary system (b); south ternary system
(0). Sugar O4′-O4′ distance in angstroms; bold line, furan ternary
crystal structure sugar O4′-O4′ distance.11
last 1500 ps of the ODN simulations along with values observed
in B-form and flipped-out DNA experimental structures were
analyzed (see Figure 3 of the Supporting Information). For the
5′-ꢀ and 3′-â dihedrals, only ODN-furan and ODN-south show
significant sampling of states around the DNA-M.HhaI com-
plex average values. This is consistent with the observation that
only these two ODNs show complete flipping of the abasic
sugar. For the γ-dihedral, only ODN-furan and ODN-south
populate the t state, though the amount of sampling is small.
ODN-south alone shows some sampling of the g+ state of the
ú-dihedral at a value lower than the experimental average. These
probability distributions indicate that only ODN-furan and ODN-
south spontaneously sample the backbone conformations seen
in the DNA-M.HhaI complexes with ODN-south showing a
greater tendency to sample those conformations.
3.2. DNA in Ternary Complex. Understanding of the
structural and dynamic behavior of the DNA in the ternary
complexes, in addition to its behavior alone in aqueous solution,
is necessary to interpret the biochemical observations. All
ternary complex MD simulations were started from the flipped
out form of the sugar seen in the crystal structures. The sugar
O4′-O4′ distance, used to monitor the flipping-out transition
in the DNA MD simulations, was assessed to detect any
characteristic behavior for the ternary complexes. Figure 7 shows
the probability distributions of the sugar O4′-O4′ distances for
the four ternary complex simulations along with the experi-
mental structure value. The ternary-north system shows a small
shift toward a lower value for the sugar O4′-O4′ distance as
compared to the other three systems. The presence of the
deviation indicates that the flipped region in the north ODN is
attempting to revert back to the non-flipped-out conformation.
Probability distributions of the four dihedrals that differ from
canonical values in the crystal structures (see Figure 5) were
determined for the ternary simulations (Figure 4 of the Sup-
porting Information). Of the four dihedrals, significant deviations
from the experimental flipped-out structure average values are
observed in the flipped position γ- and ú-dihedrals. In both
ternary-C and ternary-furan, the flipped position γ-dihedral
shows a tendency to populate a region close to the g+ state
seen in experimental B-DNA structures. In ternary-north, the
γ-dihedral populates the t state but at a lower value ( ∼150 °)
than the experimental t-state γ-value. The ternary-south γ-di-
hedral populates both the t and g- states with the t-state
population showing excellent correspondence with the experi-
In the ternary complexes, ternary-C samples pseudorotation
angles intermediate to the north and south forms. A tendency

12432 J. Am. Chem. Soc., Vol. 122, No. 50, 2000
Wang et al.
to sample the north conformation is indicated with brief
excursions to the north form that occurs in the crystal structure.1
In ternary-furan, the sugar makes a transition from the north
form to the south form after ∼300 ps (not shown) and remains
in the south form for the remainder of the simulation. This is
contrary to what is observed in the crystal structure and is
probably a reflection of the more flexible nature of the abasic
furanose sugar moiety. This presence of the south conformation,
however, does not affect the sugar O4′-O4′ distance (Figure
7). For ternary-north, the expected north conformation is
sampled throughout the simulation. Finally, the south ternary
system shows the south conformation to dominate, consistent
with the constrained carbocyclic structure. It is observed,
however, that significant sampling of the north conformation
occurs. These results indicate that, in the ternary complex, the
north conformer of the sugar may be favored. However, the
sampling of the south conformation in the furan ternary complex
is not consistent with the observed crystal structure with the
M.HhaI-abasic DNA-AdoHcy ternary complex (see below).
carbocyclic sugar. Thus, enhanced binding of ODN-south and
ODN-furan is due to the ability of the sugar to spontaneously
flip out of the DNA duplex. Such flipping cannot occur with
ODN-north due to hindered rotation about the γ-dihedral due
to the ring conformation, leading to its significantly decreased
binding.
Analysis of the conformations of the phosphodiester backbone
of the target region in both DNA in solution (Figure 3 of the
Supporting Information) and in the DNA-M.HhaI-AdoHcy
complex (Figure 4 of the Supporting Information) from the MD
simulations yielded information on the enhanced binding of
ODN-south over ODN-furan. From the probability distributions
of these dihedrals it was observed that ODN-south assumed
conformations more similar to those that occur in DNA-
M.HhaI-AdoHcy crystal structures than ODN-furan. Improved
sampling of the bound conformation of the DNA phosphodiester
backbone by ODN-south, therefore, is suggested to lead to the
improved binding over ODN-furan. Thus, the overall binding
profile from the biochemical studies is predicted to result from
the capacity of the target site in the ODN to assume the protein-
bound (flipped) conformation in solution and maintain that
conformation when bound to the enzyme.
Discussion
Influence of Abasic Site on Duplex Stability. Tm measure-
ments (Table 1) indicate that the presence of any abasic site is
sufficient to strongly destabilize the duplex DNA, probably due
to the loss of base stacking and Watson-Crick hydrogen
bonding. Among the abasic ODNs, the order of stability was
north > south > furan. These results are inversely correlated
with the propensity for flipping calculated for the three systems
(Figure 4), where the order is furan > south > north, but not
the inhibitory capacity, where the order is south > furan > north.
Although it is clear that the difference in Tm is not sufficient to
account for the difference in binary and ternary complex stability
observed in our biochemical studies, this result does suggest
that spontaneous flipping can have a small but measurable
impact on the overall stability of the DNA duplex.
Inhibition of M.HhaI by DNA Containing Abasic and
Carbocyclic Pseudosugar Analogues at the Target Site. From
the synthetic and biochemical experiments, the order of inhibi-
tory potency was found to be ODN-south > ODN-furan .
ODN-north (Table 1 and Figure 1 of the Supporting Informa-
tion). This result was surprising due to crystallographic data
indicating that M.HhaI bound to DNA with either a flipped
cytosine or an abasic site at the target position has the sugar in
the north conformation.11 To understand the rank order of
binding, MD simulations were performed on the ODNs alone
in solution and in a ternary complex with M.HhaI. The MD
simulations reported here suggest that the sugar moiety in ODN-
south and ODN-furan flips out of the DNA duplex spontane-
ously (Figure 4), a process that does not occur in ODN-north.
To understand the cause of this difference, dihedrals in the
phosphodiester backbone that change significantly upon binding
of DNA to M.HhaI based on crystal structures were identified
(Figure 5) and shown to undergo transitions in the MD
simulations in which flipping occurs (Figure 2 of the Supporting
Information). Of these dihedrals, γ is directly adjacent to the
sugar moiety and, therefore, most likely to be influenced by
the sugar structure. QM calculations on model compounds
(Figure 2, compounds A-C) showed that the presence of the
north carbocyclic sugar disfavors the g- conformer of the
γ-dihedral (Figure 6), thereby hindering flipping of the north
carbocyclic sugar out of the DNA duplex. Constraining the
abasic sugar to the north conformation yielded a similar result
(Figure 6), indicating the hindered rotation to be due to sugar
conformation rather than the chemical structure of the north
Propensity for the “Open” versus “Closed” Conforma-
tions of M.HhaI. An essential step in the mechanism of methyl
transfer by M.HhaI is converting from an “open” complex with
DNA to a catalytically competent “closed” complex when
cofactor is available.9 The gel-shift assay experiments reported
here (Figure 3) show that M.HhaI assumes a closed conforma-
tion when bound to ODN-south and ODN-furan either in the
presence or absence of cofactor. In contrast, M.HhaI bound to
ODN-north is found predominantly in the open state in all cases.
The formation of closed binary complexes between ODN-south
and ODN-furan and M.HhaI is especially interesting since the
normal substrate, ODN-C, requires the presence of AdoHcy to
form a complex with closed conformation. From MD simula-
tions it was observed that both ODN-south and ODN-furan
assume a flipped out conformation in solution (Figure 4) and
maintain the conformation in the ternary complex (Figure 6).
This suggests that formation of the closed complex requires a
totally flipped-out state. In ODN-north, since the carbocyclic
sugar does not flip out of the duplex during the time of
observation (Figure 4), this observation would predict that a
closed conformation cannot form readily. In the ternary complex,
the north sugar moiety shifts to shorter O4′-O4′ distances
(Figure 6), suggesting that it does not want to assume that
flipped-out conformation, which is consistent with the native
gel-shift assay results on the ternary complexes.
Despite the predictions of the model, a small amount of closed
conformation of ODN-north is observed in the gel assay. While
the possibility that the north sugar moiety can assume a
completely flipped-out state, compatible with the closed con-
formation, cannot be excluded due to limitations in the present
calculations, this divergence between theory and experiment
suggests that additional interactions may be contributing. One
possibility is that interactions between Gln237 and the unpaired
G residue at the abasic site not only favors the partial flipping
that leads to formation of the “open” complex but allows the
north constrained abasic to sample the completely flipped state.
This is supported by the presence of this interaction in the
ternary-north simulation (Table 3). Alternatively, in the absence
of base, the enzyme may possibly take an active role in
“flipping”. What is clear is that, once achieved, the ODN-north
in the “closed” complex becomes resistant to rapid exchange
as evidenced by the T1/2 for dissociation of ∼14 h.

HhaI DNA (Cytosine C5)-Methyltransferase
J. Am. Chem. Soc., Vol. 122, No. 50, 2000 12433
Preferred Sugar Pseudorotation Angles. As shown in
Figures 2-4 of the Supporting Information, the present MD
simulations reproduce a number of structural properties observed
in DNA-M.HhaI crystal structures. Inconsistencies between the
crystal structures and ternary MD simulations are, however,
observed in the flipped furanose sugar pseudorotation angles
(Figure 5 of the Supporting Information). On the basis of the
ab initio calculations, the flipping process appears to be favored
by the south sugar pucker (vide infra); however, the crystal
structure11 of the abasic ternary complex indicates the flipped
sugar to assume a north conformation. In the ternary-furan MD
simulation, however, the sugar primarily samples the south
conformation. This tendency may be attributed to any of the
following reasons: (a) overstabilization of the C2′ endo form
of the sugar in the CHARMM force field parameters; (b) the
limited time scale of the simulation that is not sufficient to
represent the proper balance between the south and north
conformations; (c) inherent flexibility in the conformation of
the abasic sugar in the ternary complex not captured in the
crystal structure;11 and (d) the limited resolution of the crystal
structure (2.39 Å), 11 which may make unambiguous assignment
of the sugar pucker difficult. In this regard, it is of interest to
note that recently published APE1-DNA complex structures
show abasic flipped-out furanose sugars in the south conforma-
tion.65 Although the test ODN had a different sequence from
the one used in our studies, this finding suggests that for the
flexible furan moiety either the north or south conformation is
compatible with the flipped-out orientation of the sugar and that,
during crystallization, either form could be trapped. Interestingly,
brief excursions to the north conformation in the south car-
bocyclic sugar occur in the ternary-south simulation, which are
not expected given the significantly greater stability of the rigid
south conformation (Figure 5 of the Supporting Information).
Moreover, the control ternary flipped sugar populates pseu-
dorotation angles intermediate between the south and north
forms, while the north carbocyclic sugar pseudorotation angle
remains in the expected north conformation. These observations,
along with the north conformation observed in X-ray crystal
structure, suggest that, in the ternary complex, the north form
of the flipped sugar is favored over the south form, though the
present results support the possibility that the south conformation
is compatible with the completely flipped-out state of the abasic
sugar.
Model for the Structural Mechanism of M.HhaI. A
proposed structural pathway for formation of a catalytically
competent complex between M.HhaI, substrate DNA, and
cofactor involves formation of at least two structurally distinct
complexes prior to the attainment of the structural relationship
between substrate, enzyme, and cofactor that is required for
methyl transfer.9 They are a recognition complex in which the
target base remains stacked in an intact B-helix and an “open”
conformation with substrate conformers with varying degrees
of target base flipping. By combining information obtained in
our biochemical and theoretical studies, we can expand on the
following aspects of this model.
Enzymatic Influence on Direction of Flipping. On the basis
of considerations of steric crowding, flipping of a target base
through the major groove has been predicted to be more
energetically favorable than flipping out through the minor
groove.66 Some enzymes, such as uracil glycosylase use a major
groove-based mechanism to flip out their target bases6 and may
actively participate in base flipping.67 However, the structural
models for M.HhaI and M.HaeIII make it likely that DNA
cytosine (C5)-methyltransferases flip target cytosines through
the minor groove.1 Although NMR studies suggest that M.HhaI
does not accelerate flipping of the target cytosine,9 they cannot
rule out the possibility that the enzyme is necessary to direct
flipping of the target base toward the minor groove side.
The MD simulations in this study indicate that, at least in
the absence of a base, sugar flipping occurs through the minor
groove of DNA without requiring interaction with the enzyme.
On the basis of visual inspection, when flipping occurs via the
minor groove, the partially flipped-out form of the sugar appears
to occupy the spatial region that the flexible active site loop
(residues 80-99) closes around. Our calculations are not
sufficient to give information about protein-DNA interactions
in the recognition or the “open” complex. However, if we make
the assumption that during the flipping process the flexible active
site loop occupies the conformation seen in the binary M.HhaI-
AdoMet complex, we can predict that this loop can only lock
the enzyme in the “closed” conformation when the sugar (and
base) is completely flipped out of the helix. This observation is
consistent with the present biochemical studies that indicate that
the enzyme in the ternary-north complex is found primarily in
an “open” conformation, which is due to incomplete flipping
of the north constrained analogue.
Recognition and Formation of “Open” Complexes. The
initial binding of the M.HhaI recognition sequence GCGC in
DNA is thought to be mediated primarily via major groove
interactions with the small domain (residues 194-275) of the
enzyme.1 This region contains most of the sequence-specific
contacts and, on the basis of ternary complex crystal structures,
is proposed to remain in contact with the nontarget strand
throughout the process of methyl transfer. This is consistent
with the observation that, in the ternary complexes, where all
sugars and the control cytosine are in the completely flipped
configuration, the interaction distances between Gln237 and the
orphan guanine base are similar for all four systems (Table 3).
All of our simulations of target flipping in solution indicate
that the north constrained abasic target remains primarily inside
the duplex DNA with short excursions to a partially flipped-
out state (Figure 4, O4′-O4′ distances out to 15 Å). It is never
observed in the completely flipped-out state during the time of
observation. This is consistent with the gel-shift experiments
where at least 50% of the ODN is found migrating as an “open”
conformation while less than 10% is in the “closed” conforma-
tion in either the binary or ternary complexes. Previous studies
with a base-paired C target have shown that the small amount
of binary complex formed is in the open conformation while,
upon formation of the ternary complex, the closed conformation
dominates.9 These results suggest that the open conformation
is due to the sugar (and base in the normal substrate) occupying
unflipped or partially flipped states that disallow closure of the
flexible, active site loop around the target sugar and base. In
the normal C substrate, the presence of cofactor stabilizes the
interaction of the base with the protein, yielding a totally flipped
conformation of the base and sugar, thereby allowing full closing
of the flexible loop.
Formation of “Closed” Complexes. If, as postulated above
on the basis of simulation data, complete flipping of the target
sugar is necessary for formation of the “closed” conformation
between ODN and M.HhaI, it would be predicted that the furan
and south constrained abasic sites would preferentially form
(65) Mol, C. D.; Izumi, T.; Mitra, S.; Tainer, J. Nature 2000, 403, 451-
456.
(66) Ramstein, J.; Lavery, R. Proc. Natl. Acad. Sci. U.S.A. 1988, 85,
7231-7235.
(67) Stivers, J. T.; Pankiewicz, K. W.; Watanabe, K. A. Biochemistry
1999, 38, 952-963.

12434 J. Am. Chem. Soc., Vol. 122, No. 50, 2000
Wang et al.
closed conformations. This is exactly what was observed in our
native gel-shift analysis of complex formation. The fact that
there is no evidence for an effect of cofactor on the equilibrium
level of “closed” conformation indicates that cofactor is not
necessary for either establishing or stabilizing the “closed”
conformation of the enzyme when it is bound to the furan or
south sugar analogues. In essence, the furan and abasic south
targets may invite closure of the flexible and recognition loops
around themselves by adopting a completely flipped-out orienta-
tion. The enhanced stability of “closed” complexes of M.HhaI
with ODN-south relative to ODN-furan is suggested to be due
to its ability to better sample the bound DNA conformation and
the stronger interaction between Ser87 in the flexible loop and
Gln287 in the recognition loop (Table 3).
A similar lack of need for cofactor in formation of the highly
stable “closed” conformation was noted with ODNs containing
the transition-state analogue 5,6-dihydo-5-azacytosine (DZCyt)
as target base.64 This suggests that formation of the closed state
is based on the ability of the base to be stabilized in the active
site pocket. DZCyt being a transition-state mimic allows this
stabilization to occur in the absence of cofactor. In contrast,
the normal C substrate requires cofactor to stabilize the
interaction of the completely flipped-out base with the large
domain of M.HhaI.
conformation in the sugar moiety but also results in changes in
conformational flexibility of the γ-dihedral adjacent to the sugar
substituent. Such alterations in intrinsic energetic properties
associated with localized changes in DNA structure result in
substantially different structural and dynamic behavior of the
DNA. These structural and dynamic changes result in biochemi-
cal consequences including altered duplex DNA stability and
changes in interactions of the DNA with M.HhaI. The combined
synthetic, biochemical, and computational approach yields a
unique view of how structural and dynamic properties of
biomolecules impact biochemical properties. The present results
are consistent with and extend a current model describing the
structural mechanism(s) used by M.HhaI to ensure specificity
and optimal alignment of target base and methyl donor during
the methylation reaction.
Abbreviations: M.HhaI, HhaI DNA (cytosine-5)-methyl-
transferase; ODN, oligodeoxyribonucleotide; ss, single-stranded;
ds, double-stranded; QM, quantum mechanics; MD, molecular
dynamics; AdoHcy, S-adenosylhomocysteine; AdoMet, S-
adenosylmethionine; 5mC or M, 5-methylcytosine; TBE, 89 mM
Tris borate (pH 8.0), 2 mM EDTA; SSC, 45 mM sodium
chloride, 4.5 mM sodium citrate; OD, optical density.
In the case of mismatched base pairs,10 the proposed model
also explains why the open conformation is observed in the
absence of cofactor while the closed conformation is observed
in its presence. There is a high probability that a mismatched
base pair populates a variety of flipped states in the binary
complex after initial binding of the protein. The partially flipped-
out states of the target base prevent the closing of the flexible
loop around the DNA, leading to the observation of the open
state of the protein. The presence of the cofactor leads to greater
stabilization of the completely flipped-out target base in the
active site pocket, allowing the flexible loop to close around
the DNA and form the closed complex. However, as described
previously,64 this complex is less stable than that formed with
cytosine, presumably due to less than ideal interactions with
the active site pocket.
Acknowledgment. This work has been supported by NIH
Grant GM-51501 and computational support from the Pittsburgh
Supercomputing Center and the National Partnership for Ad-
vanced Computational Infrastructure to A.D.M.,Jr.; grant support
from DAMD Breast Cancer Program-17-98-1-8215, the Susan
G. Komen Breast Cancer Fund and the Nebraska Department
of Health to J.K.C.; fellowship support from the Graduate
College of UNMC to A.S.B.
Supporting Information Available: Text and two schemes
describing results of the synthetic procedures, five figures and
a table listing the values of the phosphodiester backbone in the
vicinity of the flipped based for currently available crystal
structures of the DNA-M.HhaI complexes, and the empirical
force field parameters for the carbocyclic sugar analogues (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
Conclusion
The presence of conformationally constrained sugar substit-
uents in DNA not only enforces the designated pseudorotation
JA001989S