© 1993 Oxford University Press
Nucleic Acids Research, 1993, Vol. 21, No. 24
5623-5629
The crystal structure of N4-methylcytosine-guanosine
base4
pairs in the synthetic hexanucleotide d(CGCGm CG)
Aldo R.Cervi, Andre Guy1, Gordon A.Leonard, Robert Teoule1 and William N.Hunter*
Department of Chemistry, University of Manchester, Oxford Road, Manchester M13 9PL, UK and
1
Service d'Etudes des Systemes et Architectures Moleculaires, Departement de Recherche
Fondamentale de la Matiere Condensee, Centre d'Etudes Nucleaire de Grenoble, BP 85X,
F38041 Grenoble Cedex, France
Received September 9, 1993; Revised and Accepted October 11, 1993
PDB no. 133D
ABSTRACT
The structure of d(CGCGm4CG) where m4C = N4methylcytosine has been determined by crystallographic methods. The crystals are multifaced prisms,
with orthorhombic space group P212121 and unit cell
dimensions of a = 17.98, b = 30.77 and c = 44.75A. The
asymmetric unit consists of one duplex of hexanucleotide and 49 waters. The R-factor is 0.189 for 1495
reflections with F > a(F) to a resolution limit of 1.8A.
The double helix has a Z-DNA type structure which
appears to be intermediate in structure to the two
previously characterised structure types for Z-DNA
hexamers. The two m4C G base-pairs adopt structures
that are very similar to those of the equivalent basepairs in the structure of the native sequence
d(CGCGCG) except for the presence of the methyl
groups which are trans to the N3 atoms of their parent
nucleotides and protrude into the solvent region. The
introduction of the modified base-pairs into the
d(CGCGCG) duplex appears to have a minimal effect
on the overall base-pair morphology of the Z-DNA
duplex.
INTRODUCTION
The DNA base N4-methylcytosine (m4C) is found almost
exclusively in thermophilic bacteria which have high G-C content
DNA (1). It has been proposed that the base might replace
5-methylcytosine (m5C) in these bacteria as the temperature for
their optimal growth (50—70°C) would accelerate the
deamination of m5C to yield thymine. This could result in the
formation of potentially mutagenic G.T mismatch base-pairs in
the bacterial genome.
Alternating d(CG)n sequences are prone to adopt the lefthanded Z-DNA conformation, especially at high salt
concentrations (2,3) and it has been shown that the presence of
m5C in these sequences results in a large enhancement of the
propensity of these sequences to assume this conformation (4).
;
To whom correspondence should be addressed
An explanation put forward to account for this observation is
that the creation of a hydrophobic region on the DNA surface
stabilises the Z-form relative to the B-form (5). In contrast, while
the midpoint for the B —Z transition for d(CGCGCG) occurs at
approximately 3M NaCl concentration (5), that for
d(CGm4CGCG) is not evident even at 5M NaCl (6). This
suggests that, unlike m5C, m4C does not stabilise Z-DNA with
respect to the B-form and thus, in thermophilic bacteria, fewer
Z-DNA stretches are to be expected. However, we have
crystallised the synthetic DNA hexamer d(CGCGm4CG) and
found that, like d(CGCGCG), it assumes the Z-DNA
conformation and commensurate with our interest in the structures
of DNA duplexes containing non-standard base-pairs we report
here the results of a single crystal X-ray analysis of
d(CGCGm4CG).
MATERIALS AND METHODS
The hexanucleotide d(CGCGm4CG) was synthesized using
techniques described previously (7). Crystals were grown at 4°C
by vapour diffusion from droplets sitting in Corning glass plates
(8). Well formed colourless prisms grew after three weeks from
23/iL droplets containing the hexanucleotide (4.4mM), sodium
cacodylate (22mM, pH 6.5) magnesium chloride (9mM),
spermine hydrochloride (0.4mM) and 2-methyl—2,4-pentanediol
(4% v/v). The droplets were equilibrated against an external
reservoir of 50/iL aqueous 2-methyl-2,4-pentanediol (75% v/v).
A single crystal of dimensions 0.10x0.15 X0.40 mm was sealed
in a glass capillary and used for X-ray data collection. The data
were measured at 22 °C on a Rigaku AFC5 diffractometer using
a Rigaku RU200 rotating anode operating at 50 KV, 100mA and
producing CuK a radiation (graphite monochromator,
X=1.5418A, 0.5 mm focal spot). A 0.5mm incident collimator
was used and the crystal to detector distance was set at 400mm
with a continuously evacuated beam tunnel being employed to
reduce absorption by air. The lattice parameters were determined
from a least-squares fit of 25 reflections in the range 4° < 20
5624 Nucleic Acids Research, 1993, Vol. 21, No. 24
< 12° to be a= 17.98, b=30.77, c=44.75A and, from the Laue
symmetry and systematic absences, the space group was
determined to be P212121. Intensities were measured with w
scans of width (0.9 + O.3tan0)° at a speed of 4c/minute. Three
standard reflections were monitored every 150 measurements and
indicated no significant crystal decay. Intensities were corrected
for Lorentz and polarisation factors. An empirical absorption
correction (9) was also applied (minimum=0.56, average=0.72).
All unique reflections (2994) to a resolution of 1.5A were
measured and those with F/er(F) < la were scanned in triplicate
to improve counting statistics. However, the diffraction data
above 1.8A resolution was extremely weak and it was therefore
excluded from the refinement procedure. The software for data
collection and processing was provided by the Molecular
Structure Corporation, Texas, USA (10).
The unit cell parameters and space group symmetry suggested
that the structure of d(CGCGm4CG) is isomorphous with an
orthorhombic form of d(CGCGCG). As we employed only a very
small amount of spermine in our crystallisation conditions the
refined coordinates for d(CGCGCG) crystallised in the presence
of magnesium only (11) were retrieved from the Brookhaven
Protein Databank (12) and used as the starting model for the
refinement. In order to correctly position the model in the unit
cell the starting model was refined as rigid body using a modified
version of SHELX (13). Starting with data in the region
10.0—7.0A the resolution limit of the data was increased in steps
of 1A until all data with F > 0 in the range 10.0-2.5A was
included. This converged with a crystallographic residual R=0.37
for 844 reflections (R is defined as E|F O -F C |/2|F O | where F o
and F c represent the observed and calculated structure factor
amplitudes respectively). The refinement was continued using
the Konnert-Hendrickson restrained least-squares procedure (14)
as modified for nucleic acids in NUCLSQ (15) using data in the
range 8.0— 1.8A with F > 2CT(F). Isotropic thermal parameters
at die onset of mis part of the refinement were set at 8A2 for
the^ atoms in the bases, 10A2 for the atoms in the sugars and
12A2 for die atoms in die phosphate groups. 10 cycles of
positional refinement reduced R from 0.33 to 0.31 and a further
5 cycles in which isotropic temperature factors were included
resulted in a residual of 0.27 for 1151 reflections. Fourier electron
density (2F 0 —F^ and difference density (Fo—Fc) maps were
men calculated and displayed on an Evans and Sutherland ESV30
graphics workstation using the program FRODO (16).
Manipulation of the model at various points of the sugarphosphate backbone to improve the fit to the density maps was
carried out and solvent molecules were included into the model.
The major change to the DNA was the addition of a methyl group
to the N4 atoms of the cytosine residues 5 and 11 with die
change being consistent with die observed density. Solvent
molecules were included on the criteria that they displayed
approximately spherical density in both (FQ-FC) and Q.¥o-F^
maps formed reasonable hydrogen bonds widi neighbouring
functional groups and/or existing solvent molecules. The
refinement was continued widi NUCLSQ (15) and periodic
examination of bom the model and density maps on the graphics
system. A total of 26 solvent molecules were included as die
refinement converged at R=0.19.
The number of 2a(F) data in die resolution range 8.0—1.8A
represents only 46% of die total meoretical data and in order
to improve the parameters/observations ratio in our refinement
it was decided to introduce data with F > o(F). As expected,
this resulted in an increase in the R-factor to 0.213. The model
was then refined with a simulated annealing protocol using the
X-PLOR software package (17,18) after which R was reduced
to 0.195. Several further rounds of NUCLSQ refinement were
carried out and a number of additional solvent molecules were
added to the model. This resulted in a final R-factor of 0.189
for 1495 reflections with F > s(F) in the resolution range
8-1.8A. This represents 60% of the theoretical data within die
limits stated, 42% between 1.9A and 1.8A. The final model,
which consists of the DNA duplex (242 atoms) and 49 solvent
molecules, has good geometry with a root-mean-square deviation
(r.m.s) from ideal bond lengths for sugar/base groups of 0.019A.
For phosphate groups the r.m.s. deviation is 0.028A. The angle
associated distances for sugar/base groups have an r.m.s deviation
from ideal values of 0.025A while mat for me phosphate groups
is 0.029A. An example of the excellent fit between the final model
and the electron density is shown in Figure 1. Atomic coordinates
and structure factors have been deposited with the Brookhaven
Protein Databank (12).
RESULTS AND DISCUSSION
Structure of the double helix
In the Z-DNA duplex the nucleotides are labelled C(l) to G(6)
on strand 1 and C(7) to G(12) on strand 2 in me 5' to 3' direction.
The solvent molecules are labelled W13 to W61.
d(CGCGCG) has been crystallised under diree distinct sets of
conditions, using magnesium only (11), magnesium and spermine
together (3) and spermine only (19). This has been discussed in
some detail by Schneider et al., (20) and in this present work
each crystal type will be referred to as forms I, n and HI
respectively. In crystals of form in the conformation of the
sugar/phosphate backbone of the duplex is Zj for all residues.
In crystals grown in the presence of magnesium (forms I and
II) the backbone conformation of residue 5 is Z n Although the
unit cell dimensions and symmetry are essentially die same the
duplexes in form in pack differently from the other two forms.
In view of the crystallisation conditions from which our sample
had been grown we had expected d(CGCGm4CG) to be
isostructural widi forms I and II. The values of the torsion angle
f for d(CGCGm4CG) (Table 1) indicate that the backbone
conformation for all me residues in the double helix is Zj and
there is no Z n conformation at residue 5. The duplex structure
of d(CGCGm4G) would therefore seem to be more similar to
that of d(CGCGCG) crystallised in the presence of spermine only
(form HI, 19) which also has the backbone conformation of all
its residues as Zx. However, an examination of die packing of
die different duplexes shows that die duplexes in die structure
of d(CGCGm4CG) pack in the same manner as duplexes of
d(CGCGCG) crystallised in die presence of magnesium (forms
I and H). Figure 2 clearly shows that bom the position of die
d(CGCGm4CG) duplex in the unit cell and its orientation is
different from mat of form in. It appears dierefore mat the
structure of d(CGCGm4CG) is slighdy different to me two types
of structure previously reported for d(CGCGCG) (3,11,19).
Further evidence for this is found when die structure of
d(CGCGm4CG) is superimposed on the structure of the
magnesium form, the magnesium/spermine form and die
spermine form of d(CGCGCG). The r.m.s. deviations in atomic
positions are 0.44A, 0.57A and 0.47 A respectively and suggest
mat the structure of d(CGCGm4CG) is equally different to bom
the previously reported structure types. Schneider et al., (20)
in their very diorough and detailed analysis of Z-form
Nucleic Acids Research, 1993, Vol. 21, No. 24 5625
Figure 1. The G(2)m 4 C(ll) base-pair superimposed on the relevant section of the 2F O -F C electron density (purple chicken wire) calculated using the final model
from the refinement procedure. The map is contoured at approximately 1.5 r.m.s. deviations from the average density in the unit cell. The three inter-base hydrogen
bonds are shown as dashed lines. The colour scheme for the atoms is as follows: Carbon, yellow; nitrogen, blue; oxygen, red.
conformation in relation to hydration and crystal packing have
concluded that solvation effects, involving water molecules
bridging symmetry related duplexes, play a major role in
determining the precise nature of the structure observed. The
resolution of our structure determination is good enough to
distinguish between the Zj and Z n conformations and it is worth
noting that the model used to initiate the refinement had a Z n
conformation at residue 5. This was only altered during the course
of the refinement when the electron density maps indicated an
improved fit could be obtained. We conclude that the intermediate
conformation presented here is determined by our crystallisation
conditions and not due to the incorporation of a modified base.
Regardless of structure type the base-pair morphology of ZDNA hexamers remains similar (20). Table 2 presents the
parameters defining the base-pair morphology for
d(CGCGm4CG). Usually in Z-DNA the buckle of the base-pairs
alternates between highly positive and highly negative. Here this
is not the case and at one end of the duplex the base-pairs are
reasonably flat. It is tempting to attribute this difference to the
presence of the m4C G base-pairs. However, as the flattening
of the base-pairs does not occur at both ends of the duplex, this
is unlikely to be the case and the difference that we observe in
base-pair buckle is more likely to be an artefact of the refinement
procedure or a natural variation in Z-DNA structure. In general
the other geometrical parameters follow he same pattern as
previously observed for Z-DNA hexamers (20) and the
introduction of the two m4C G base-pairs does not appear to
affect the structure to any significant extent.
Conformation of the m4C G base-pairs
The two m 4 CG base-pairs adopt a Watson-Crick type
conformation typical of that adopted by C G base-pairs in ZDNA (Figure 1) and are identical within the accuracy of our
crystal structure analysis. Solution studies of N4-methylcytosine
have shown that the methyl group can be either cis (proximal)
or trans (distal) to N3 (21,22). When the methyl group is in the
proximal conformation it will not allow the formation of
Watson-Crick base-pairs and will result in the destabilisation
of the duplex in which it is contained. This is observed in solution
studies of d(CGm"CGCG) (6). We, however, do not see any
evidence of non Watson—Crick base-pairs involving m4C and
this may result from a tendency of the most stable form of a DNA
5626 Nucleic Acids Research, 1993, Vol. 21, No. 24
Table 1. Sugar/phosphate backbone torsion angles (°), glycosyl torsion angles (°) and sugar conformations
as observed in the structure of (CGCGm4CG)
a
Residue
X
Cl
G2
C3
G4
-148
60
-159
55
m4C5
-162
63
G6
C7
G8
C9
G10
-173
62
-156
73
m 4 Cll
-145
G12
90
_
62
176
79
213
130
—
57
202
92
192
95
0
_
183
220
163
198
124
189
197
174
250
161
y
6
80
180
88
176
51
163
259
188
59
161
68
169
131
92
128
99
140
152
139
98
146
91
132
174
-93
-104
-69
-95
-64
—
-103
-103
-84
-110
-92
—
f
Sugar Conformation
90
-68
68
-49
61
89
-55
63
-83
66
—
Cl'-exo
C2'-exo
C2'-endo
C2'-exo
C2'-endo
C3'-exo
C2'-endo
C2'-exo
C2'-endo
C2'-exo
C2'-endo
C3'-exo
Main chain and glycosyl torsion angles are as defined in reference 25. The deoxyribose conformations have
been determined from the analysis of a \/d torsion angle diagram from the same reference.
i
*
Figure 2. Orthogonal views of the superpositions of the structures of the pure magnesium form of d(CGCGCG) (form I, red), the pure spermine form of d(CGCGCG)
(HI, black) and the d(CGCGm4CG) double helix (grey). The superposition is based on the origins of their respective unit cells. The position of the duplex of the
pure spermine form in the unit cell is clearly different from that of the other two duplexes. For clarity only the pure magnesium and pure spermine forms of d(CGCGCG)
are shown in the left hand figure.
Table 2. Geometrical properties of individual base-pairs and base steps for d(CGCGm4CG)
Base-pair
4
Base-step Roll(°) Slide(A) Twist(°) Rise(A) Tilt(°) BuckleO Propeller Twist(°)
1.2
5.2
13.1
3.8
0.0
0.5
-0.9
48.0
3.6
-1.1
2.0
5.3
9.8
3.7
0.4
0.8
-0.5
52.5
3.8
2.2
0.3
5.5
11.2
3.6
-0.4
G(2)m C(ll)
C(3)G(10)
G(4)C(9)
m4C(5)-G(8)
G(6)C(7)
-3.2
-3.4
-0.9
1.9
-1.9
-2.8
7.6
2.9
-1.1
-1.1
12.0
5.5
All parameters were calculated using the NEWHEL92 computer program distributed by R.E.Dickerson
via the Brookhaven Protein Databank. Nomenclature is according to reference 27.
Nucleic Acids Research, 1993, Vol. 21, No. 24 5627
Figure 3. A stereoview of the base stacking for each base step involving the G(2)m 4 C(ll) base-pair.
Figure 4. A stereoview depicting the van der Waals representation of the d(CGCGm4CG) double helix viewed down the helix axis. Note how the N 4 methyl group
on the N4-methylcytidine bases (labelled C4M5 and shaded) protrudes into the minor groove.
duplex to crystallise out in preference to other possible
conformations that exist in solution.
Figure 3 shows the base stacking interactions involving the
G(2)m 4 C(ll) base-pair as found in the structure of
d(CGCGm4CG). There is little difference between this and what
is usually observed in Z-DNA structures. At the CpG step there
is considerable interstrand overlap of the pyrimidine residues
while the guanine bases are stacked onto the furanose moiety of
the preceeding cytosine nucleotide. At the GpC step there is
considerable intrastrand overlap of the purine and pyrimidine
residues. The methyl groups themselves are not involved in any
base-stacking and protrude out into the solvent regions (Figure
4) in a similar fashion to that found for the methyl groups in
N6-methyladenosinethymidine base-pairs (23). If m4C G basepairs adopt a similar configuration in B-DNA as they do in ZDNA (i.e. the methyl group is well removed from the rest of
the base-pair) then this would allow easy recognition of these
base-pairs in the DNA of diermophilic bacteria by restriction and
modification enzymes specific to this type of DNA. Equally this
type of conformation for the base-pair in the DNA of nonthermophilic bacteria may help to confirm the hypothesis of
Frederick et al. (23) as to the reasons why the methylation of
nucleic acid bases can result in the non-recognition of DNA by
similar restriction enzyme systems.
5628 Nucleic Acids Research, 1993, Vol. 21, No. 24
Figure 5. Top: The minor groove hydration of the the d(CGCGm4CG) double helix. The double ribbon of first shell water molecules (black spheres) are bridged
by second shell solvents (green spheres) to form pentagonal arrangements of hydrogen bonds in the minor groove of each base-pair. Bottom: A close up of a typical
arrangement of hydrogen bonds (dashed lines) in the minor groove. The base pair shown is G(2) m 4 C(l 1). DNA atoms are coloured as follows, oxygens red, nitrogens
blue, carbons stipled black, phosphorous pink. Note that the pentagonal arrangements of hydrogen bonds is completed by the minor groove side inter-base hydrogen
bond. Of the seven hydrogen bonds shown here the shortest distance is 2.5A, the longest 3.4A.
Hydration of the double helix
We have been able to elucidate the positions of a reasonable
number of well defined solvent molecules during the refinement
procedure all of which were assigned as the oxygen atoms of
water molecules. Surprisingly for Z-DNA crystallised in the
presence of Mg 2+ ions we could find no evidence of these
binding to the DNA even though the resolution of the structure
is good enough to allow the unequivocal assignment of such ions.
Less surprising perhaps, given the extremely small quantities,
is the failure to observe spermine binding to the DNA.
The most interesting facets of the hydration of this DNA duplex
concern the anionic phosphate oxygens and the deep minor groove
of the helix. The majority of phosphate oxygens are involved
in hydrogen bonding to solvent water molecules. Four, out of
eight, adjacent phosphate groups on the same strand are bridged
by networks of water molecules. In the minor groove the first
hydration shell consists of a double ribbon of water molecules
running the length of the duplex. These ribbons of water
molecules have been observed in previous studies of the structure
of Z-DNA (24). In the present work the water molecules are
generally within hydrogen bonding distance of either the cytosine
O 2 atoms or the guanine N 2 amino groups on the minor groove
side of the base-pairs. A second shell of solvent molecules then
bridges each pair of first shell water molecules to form a series
of pentagonal arrangements of hydrogen bonds. The corners of
the pentagons are the three solvent molecules and the two minor
groove functional groups mentioned above (Figure 5). In some
cases, the intersolvent distances in these pentagonal arrangements
of hydrogen bonds are of the order that one might expect if the
second shell solvent were in fact a Mg 2+ ion. In none of these
instances is there any further evidence that the solvent should
be assigned as an ion and in all cases they have been assigned
as water molecules. Pentagonal arrangements of hydrogen bonds
are commonplace in the structures of biologically important
molecules (25,26) and are thought to confer extra stability. It
is surprising that their presence in the well ordered Z-form
structures has so far escaped comment.
CONCLUSIONS
The single-crystal X-ray structure analysis of the hexamer duplex
d(CGCGm4CG) establishes that, like the parent d(CGCGCG),
it adopts a Z-type structure under the crystallisation conditions
used even though the presence of m4C in d(CG)n sequences is
not thought to favour the Z-conformation. The structure of
d(CGCGm4CG) would appear to be intermediate between the
two previously observed structural types for Z-DNA hexamers
which are crystallised either in the presence or absence of
magnesium ions. The effect on the base-pair morphology of
introducing m4C into the d(CGCGCG) double helix is minimal
and there are no perturbations of the geometry of individual base
steps or base-pairs that can be directly attributed to the presence
of the modified base. The m4C G base-pair itself adopts a
conformation extremely similar to that of the equivalent
unmodified base-pair in Z-DNA with the obvious exception of
the presence of the methyl group which adopts a conformation
trans to the N3 atom of the cytosine base and sticks out away
from the rest of the base pair into the solvent region on the major
groove side of the double helix. If the base-pairs adopt a similar
conformation in B-DNA this would allow for the easy recognition
of m4C by restriction and modification enzymes in thermophilic
bacterial DNA where it is known to occur as a replacement for
the less temperature stable base m5C.
ACKNOWLEDGEMENTS
We thank the Wellcome Trust, the Science and Energy Research
Council (UK) and the University of Manchester for financial
support.
Nucleic Acids Research, 1993, Vol. 21, No. 24 5629
REFERENCES
1. Ehrlich, M., Gama-Sosa, M.A., Carreira, L.H., Ljungdahl, L.G., Kuo, K.C
& Gehrke, C.W. (1985) Nucleic Acids Res. 13, 1399-1412.
2. Pohl, F.M. & Jovin, T.M. (1972) J. Mol. Biol. 67, 375-396.
3. Wang, A.H-J., Quigley, G.J., Kolpak, F.J., Crawford, J.L., van Boom,
J.H., van der Marel, G.A. & Rich, A. (1979) Nature 282, 680-686.
4. Behe, M. & Felsenfeld, G. (1981) Proc. Natl. Acad. Sci. USA. 78,
1619-1623.
5. Fujii, S., Wang, A.H-J., van der Marel, G.A., van Boom, J.H. & Rich,
A. (1982) Nucleic Acids Res. 10, 7879-7892.
6. Fazakerley, G.V., Kraszewski, A., Teoule, R. & Guschlbauer, W. (1987)
Nucleic Acids Res. 15, 2191-2201.
7. Kraszewski, A., Delort, A.M. & Teoule, R. (1986) Tetrahedron Lett. 27,
861-864.
8. McPherson, A. (1982) Preparation and Analysis of Protein Crystals. Wiley,
San Francisco.
9. North, A.C.T., Phillips, D.C. & Mathews, F.S. (1968) Acta Cryst A24,
580-584.
10. Molecular Structure Corporation (1985) TEXSAN. TEXRAY Structure Analysis
Package. MSC, 3200A Research Forest Drive, The Woodlands, TX 77381,
USA.
11. Gessner, R.G., Frederick, C.A., Quigley, G.J., Rich, A. & Wang, A.H-J.
(1989) J. Biol. Chem. 264, 7921-7935.
12. Abola, E.E., Bernstein, F.C., Bryant, S.H., Koetzle, T.F & Weng, J. (1987)
in Crystallographic Databases information content, software systems, scientific
applications. (Allen, F.H., Bergerhoff, G & Servers, R., eds) pp 107-132.
Data Commission of the International Union of Crystallography, Cambridge.
13. Sheldrick, G.M. (1976) SHELX: a program for refinement of crystal
structures. University of Cambridge.
14. Hendrickson, W.A. & Konnert, J.H. (1981) in Biomolecular Structure,
Conformation, Function and Evolution. (Srinivasan, R., ed) voll, pp 43—57,
Pergamon, Oxford.
15. Westhof, E., Dumas, P. & Moras, D. (1985) / Mol. Biol. 184, 119-145.
16. Jones, T.A. (1978) J. Appl. Cryst. 11, 268-272.
17. Briinger, A.T. (1990) X-PLOR (Version 2.2) Manual. Howard Hughes
Medical Institute, Yale University.
18. Brunger, A.T., Kuriyan, J. & Karplus, M. (1987) Science, 234, 458-460.
19. Egli, M., Williams, L.D., Gao, Q. & Rich, A. (1991) Biochemistry, 30,
11388-11402.
20. Schneider, B., Gimell, S.L., Jones, R., Gaffhey, B & Berman, H.M. (1992)
Biochemistry, 31, 9622-9628.
21. Shoup, R.R., Miles, T.H. & Becker, E.D. (1972) / Phys. Chem. 79, 64-70.
22. Engel, J.D. & von Hippel, P.H. (1974) Biochemistry, 13, 4143-4158.
23. Frederick, C.A., Quigley, G.J., van der Marel, G.A., van Boom, J.H.,
Wang, A.H-J. & Rich, A. (1988) / Biol. Chem. 263, 17872-17879.
24. Kumar, V.D., Harrison, R.W., Andrews, L.C. & Weber, I T . (1992)
Biochemistry 31, 1541-1550.
25. Saenger, W, (1984) in Principles of Nucleic Acid Structure. Springer-Verlag,
New York.
26. Leonard, G.A. & Hunter, W.N. (1993) / Mol. Biol. in press.
27. Dickerson, R.E., Bansal, M., Calladine, C.R., Diekmann, S., Hunter, W.N.,
Kennard, O., von Kitzing, E., Lavery, R., Nelson, H.C.M., Olson, W.K.,
Saenger, W., Shakked, Z., Sklenar, H., Soumpasis, D.M., Tang, C-S.,
Wang, A,H-J. and Khurkin, V.B. (1989). Nucleic Acids Res. 17, 1797-1803.