The crystal structure analysis of d(CGCGAASSCGCG)
2
, a synthetic DNA dodecamer duplex containing four 4
'
-thio-2
'-deoxythymidine nucleotides
The crystal structure analysis of d(CGCGAASSCGCG) 2 , a synthetic DNA dodecamer duplex containing four 4 ' -thio-2 '-deoxythymidine nucleotides
Titus J.
Boggon
,
E. Louise
Hancox
1
,
Katherine E.
McAuley-Hecht
2
,
Bernard A.
Connolly
3
,
William N.
Hunter
,
Tom
Brown
2,+
,
Richard T.
Walker
1
and
Gordon A.
Leonard*
Department of Chemistry, University of Manchester, Oxford Road,
Manchester
M13 9PL,
UK
,
1
School of Chemistry, University of Birmingham,
Birmingham
B15 2TT,
UK
,
2
Department of Chemistry, University of Edinburgh, King's Buildings, West Mains
Road,
Edinburgh
EH9 3JJ,
UK
and
3
Department of Biochemistry and Genetics, University of Newcastle-upon-Tyne,
Newcastle-upon-Tyne
NE2 4HH,
UK
Received September 19, 1995;
Revised and Accepted December 15, 1995
ABSTRACT
The crystal structure refinement of the synthetic dodecamer d(CGCGAASSCGCG),
where S = 4
'
-thio- 2
'
-deoxythymidine, has converged at
R
= 0.201 for 2605 reflections with
F
> 2
[sigma]
(
F
) in the resolution range 8.0-2.4 Å for a model consisting of the dodecamer duplex and 66 water
molecules. A comparison of its structure with that of the native dodecamer d(CGCGAATTCGCG) has revealed that the major differences between the two
structures is a change in the conformation of the sugar-phosphate backbone in the regions at and adjacent to the positions of the modified nucleosides.
Examination of the fine structural parameters for each of the structures reveals that the thiosugars adopt a C3
'
-
exo
conformation in d(CGCGAASSCGCG), rather than the approximate C1
'
-
exo
conformation found for the analogous sugars in the structure of
d(CGCGAATTCGCG). The observed differences in structure between the two duplexes
may help to explain the enhanced resistance to nuclease digestion of synthetic
oligonucleotides containing 4
'
-thio-2
'
-deoxynucleotides.
INTRODUCTION
The biological properties of 4'-thionucleosides (
1
-
5
), which are sulphur-containing isoelectronic analogues of natural nucleosides (Fig.
1
), are not always apparent from a comparison with those of the corresponding
oxysugar-containing deoxynucleoside. Hence, while (
E
)-5-(2-bromovinyl)-2'-deoxyuridine and its sulphur-containing analogue both show potent
and selective inhibition of HSV-1 and varicella zoster virus (VZV), the strong antiviral properties of 5-isopropyl- and 5-cyclopropyl-4'-thio-2'-deoxyuridine are markedly
different to those of the corresponding natural nucleosides, which show no antiviral activity at all. Similarly, 5-vinyl-2'-deoxyuridine is very toxic, whereas the corresponding 4'-thio analogue shows no toxicity, but is
active against HSV-1.
Some of the most interesting properties of pyrimidine 4'-thio-2'-deoxynucleotides are that they are significantly
more lipophilic than oxysugar-containing deoxynucleotides and have a much longer serum half-life than normal nucleotides, the latter property arising because
they are stable to phosphorolysis. This led to the suggestion that
oligonucleotides containing pyrimidine 4'-thio-2'-deoxynucleosides might be useful in the
antisense field, especially as structure determination had shown the modified
nucleosides to adopt similar conformations to those of the natural nucleosides
(
6
). Additionally, circular dichroism and ultraviolet melting studies (
7
) of thiosugar nucleoside-containing oligonucleotides had shown them to adopt the same overall
conformation as the analogous unmodified oligonucleotides and that the duplexes
were only slightly less thermodynamically stable.
Experiments involving the synthesis of oligonucleotides containing the
Eco
RV recognition site revealed that when a 4'-thiothymidine residue was incorporated on the 5'-side of the scissile bond the DNA duplex was resistant
to endonuclease attack (
7
). The reasons for this were not at all clear, as a crystal structure of the complex between
Eco
RV and the modified duplex (Kostrewa,D., Hancox,E.L., Walker,R.T., Connolly,B.A.
and Winkler,F.K., unpublished results;
8
) showed that the conformation of the DNA had not significantly changed.
However, The Mg
2+
ion required for enzyme activity was not present in the structure.
MATERIALS AND METHODS
Chemical synthesis
The oligonucleotide was synthesized (10 [mu]M scale) using an Applied Biosystems 381A DNA synthesizer. The modified
deoxynucleoside phosphoramidite (
7
) was introduced at the fifth position of the synthesis as previously described
(
7
). The synthesis and purification of the oligodeoxynucleotide was achieved under
standard conditions.
Crystallization, X-ray data collection and structure refinement
Thin, colourless, needle-shaped crystals were grown at ambient temperature in sitting drops which contained the dodecanucleotide (0.5 mM), sodium cacodylate buffer (14 mM, pH 6.0), magnesium chloride (17 mM), spermine tetrahydrochloride (1 mM) and 2-methyl-2,4-pentanediol (MPD) (17% v/v), which were diffused against external reservoirs containing 50% aqueous MPD.
Two such crystals were then mounted in thin-walled glass capillaries for X-ray data collection. Data from the first of these, which had dimensions of 1.0 * 0.05 * 0.05 mm, was collected at [lambda] = 0.882 Å on station PX9.6 at the Synchrotron
Radiation Source, Daresbury Laboratory, to a resolution limit of 2.35 Å using a 18 cm MAR Research image plate detector and 17 4o oscillation frames, each exposed for 120 s. This data was later
supplemented by a set collected from a bigger crystal (1.5 * 0.15 * 0.15 mm) to a resolution limit of 2.5 Å on a Rigaku RU200 rotating anode generator ([lambda] = 1.5418 Å) operating at 50 kV, 140 mA and equipped
with a RAXIS IIc image plate system. This latter data set was collected on 39 3o oscillation frames, each of which were exposed to X-rays for 1 h. All the data was processed using the program MOSFLM
(version 5.20) (
10
), then scaled, merged and reduced using programs from the CCP4 suite of
programs (
11
) to yield a total of 2820 unique reflections with
R
sym
= 0.065 to a resolution of 2.4 Å and a multiplicity of 6.3.
Orthorhombic unit cell dimensions of a = 24.99 Å, b = 40.36 Å, c = 65.99 Å and space group P2
1
2
1
2
1
suggested that the structure of d(CGCGAASSCGCG) is isomorphous with that of the
native dodecamer d(CGCGAATTCGCG) (
9
). The initial stage of the structure refinement therefore consisted of rigid
body minimization, against the observed data, of the structure of the native dodecamer stripped of
solvent molecules and with sulphur atoms replacing oxygen atoms where
necessary. This procedure, as with the rest of the refinement, was carried out
using the program X-PLOR (
12
,
13
) and converged with
R
= 0.333 for the 2605 reflections with
F
> 2[sigma](
F
) in the resolution range 8.0-2.4 Å. The refinement then continued, using the same data, with a round of simulated
annealing-based minimization. This consisted of initial energy minimization of the
model followed by a slow cooling procedure (initial temperature 1000 K, final
temperature 300 K, temperature decrement 25 K, time step 0.5 fs) and one round
(120 cycles) of Powell minimization. Throughout this procedure and for
subsequent rounds of refinement the dictionary of restraints used was that
provided with X-PLOR and modified to include C-S bond length and C-S-C bond angle values for 4'-thio-2'-deoxy-thymidine,
which were taken from Dyson
et al.
(
3
). It should be noted that the values used for dihedral angles and improper
angles in the modified nucleoside were those already in place for 2'-deoxythymidine.
Figure 3
.
A least squares superposition of the structure of d(CGCGAASSCGCG) (black
sticks) and the 1BNA structure of d(CGCGAATTCGCG) (red sticks). Note that the
positions of the bases remains almost constant and that major differences
between the two structures are confined to the conformation of the sugar-phosphate backbone in the vicinity of the modified nucleosides.
At this stage
R
had been reduced to 0.258. Both electron density (2
F
o
-
F
c
, [alpha]
calc
) and difference density maps (
F
o
-
F
c
, [alpha]
calc
) were then calculated using the CCP4 suite of programs (
11
) and displayed and examined on a graphics workstation using the program O (
14
). Modifications to the model were then made in order to improve its fit to the
electron density and a number of potential solvent molecules were added to the
model. Solvent molecules were added only if they appeared with approximately
spherical density in both electron density and difference density maps and had
good hydrogen bonding geometry with other atoms already included in the model.
Discrimination of potential peaks on the difference maps was greatly aided by
the use of the program WATERSHED (Bond,C.S., personal communication). The
refinement then consisted of several rounds, each of which consisted of 40
cycles of positional refinement, 20 cycles of temperature factor (individual,
restrained isotropic) minimization, map calculation and examination, model
manipulation and the addition of new solvent molecules coupled with the
checking of those already incorporated into the model.
The refinement procedure was deemed to have converged when no further solvent
molecules could be added to the model and no further manual manipulation of the
model was required. At this point the
R
factor was 0.201 for the 2605 reflections with
F
> 2[sigma](
F
) in the resolution range 8.0-2.4 Å for a model consisting of the dodecanucleotide (486 atoms) and 66
solvent molecules, each of which was modelled as an oxygen atom. The data used
in the refinement represents 89% of the total theoretical data available
between the resolution limits stated, ~50% for the range 2.5-2.4 Å. The final model has root mean square (r.m.s.) deviations
from ideality of 0.018 Å for bond lengths and 3.53o for bond angles. The average temperature factors for the phosphate groups, deoxyribose groups, bases and solvent molecules are 44.2 Å
2
, 35.0 Å
2
, 24.4 Å
2
and 61.4 Å
2
respectively. The fit of the model to the final calculated electron density is
generally excellent (Fig.
2
), although there are occasional discontinuities in the sugar-phosphate backbone. Both observed structure factor data and the final
refined coordinates have been deposited with the Brookhaven Protein Databank
(PDB) (
15
) and have been assigned the accession code 233d.
RESULTS AND DISCUSSION
In the structure of the modified dodecamer the nucleotides are labelled C1-G12 on strand 1 and C13-G24 on strand 2, both in the 5' -> 3' direction. The 66 solvent molecules are
labelled HOH25-HOH91.
As indicated by nucleotide sequence, unit cell parameters and space group, the structure of d(CGCGAASSCGCG) is isomorphous with that of the native dodecamer (
9
). The parameters that define global double helical conformation do not differ
significantly between the two structures. A least squares superposition of the
two double helices (Fig.
3
) shows the r.m.s. difference in atomic positions between the two structures to
be 0.49 Å and also suggests that the global conformation of the two structures is
extremely similar. However, a closer inspection of the two superposed
structures reveals that incorporation of the four 4'-thiosugar moieties into the sequence has led to some significant differences in the local structure of the two duplexes. This view
is supported by an examination of Figure
4
. Figure
4
a shows the difference in position for every atom in the two duplexes. As can be seen, on each of the two strands the deviation in atomic position remains
fairly constant until one reaches a point in the vicinity of the 4'-thio-2'-deoxythymidine residues, when the deviations
increase to a maximum before falling away to background level. Although the magnitudes of the positional differences are somewhat greater on
strand 1 than strand 2, in both cases the distribution is rather symmetrical,
which indicates that these deviations are real and not artefacts of the
refinement procedure. Figure
4
b shows that for an average deviation per nucleotide the differences in position
reach their peak for residues S8 and S20, which are the second of the two
modified nucleotides on strands 1 and 2 respectively.
The above analysis refers to the superposition of the modified dodecamer onto
the coordinate set 1BNA (
9
) as deposited in the PDB. There have been several other determinations of the structure of d(CGCGAATTCGCG), either at low temperature (deposition 2BNA in the PDB;
17
) or using different refinement techniques (depositions 7BNA, 9BNA in the PDB;
18
,
19
). The refined structure of d(CGCGAASSCGCG) has been superimposed on each of these
coordinate sets (Figure
4
a), yielding similar results to those found when the structure of
d(CGCGAASSCGCG) is superimposed in the original 1BNA coordinate set. This is further evidence
that the differences we observe between the structure of d(CGCGAASSCGCG) and
that of the native doecamer are real. Moreover, when the coordinate sets 2BNA, 7BNA and 9BNA are
superimposed on those of 1BNA (Fig.
4
c) there is no significant increase in the r.m.s. deviations of the sugar atoms
in nucleotides 7, 8, 19 and 20 when compared with the rest of the atoms,
although, inexplicably, for the 1BNA versus 9BNA superposition anomalously large deviations occur for the C6 and M5 atoms of nucleotides 7, 8, 19 and 20.
Incorporation of the 4'-thiosugars into the sequence does not greatly affect the positions
of the bases when both the modified and native structures are compared (Fig.
3
), but rather they induce a change in the conformation of the sugar-phosphate backbone in the vicinity of the modified sugars. Indeed, the
greatest deviations in the positions of the atoms in the modified and native
(1BNA) dodecamers occurs in the phosphodiester linkages on the 5'-side of the thiosugars themselves (Fig.
5
)
Figure 4
.
(
a
) A plot of the r.m.s. differences in the positions of the atoms in
d(CGCGAASSCGCG) and d(CGCGAATTCGCG) resulting from a superposition of the
modified and the 1BNA, 2BNA, 7BNA and 9BNA native structures. The atoms on
strand 1 are numbered from 1 to 243, those on strand 2 from 244 to 486. (
b
) A plot showing the average r.m.s. deviation per residue resulting from the
least squares superposition of the modified and the 1BNA native structures. In
this figure the residues are numbered from 1 to 12 for both strands in the 5' -> 3' direction. (
c
) A plot of the r.m.s. differences in atomic position following the
superposition of the 2BNA, 7BNA and 9BNA coordinate sets of d(CGCGAATTCGCG) on
the 1BNA coordinate set. In both this figure and Figure 7, 233d refers to the
refined coordinates of d(CGCGAASSCGCG) as deposited in the PDB.
Figure 5
.
A schematic representation of the sugar-phosphate backbone of d(CGCGAASCGCG) in the region of the 4'-thio-2'-deoxythymidine moieties showing the
average deviation (in Å) between each atom and its counterpart in the 1BNA structure of the
native d(CGCGAATTCGCG).
There are few other differences in structure between d(CGCGAASSCGCG) and its native analogue. However, there is an increase in helical rise
at base step 6 (Fig.
6
a) in the modified duplex which is coupled with a decrease in helical twist at the same base step (Fig.
6
b). As the base step in question is the 5'-ApS-3' step it would appear that this local change in
conformation of the duplex is a direct result of the incorporation of sulphur
into each strand.
Examination (Table
1
) of cross-strand phosphorus-phosphorus distances in d(CGCGAASSCGCG) appears to suggest that
compared with d(CGCGAATTCGCG) the width of the minor groove increases in the
vicinity of the 4'-thionucleotides. This is sensible, in that in order to accommodate the larger 4'-thio'-2'- deoxyribose groups cross-strand phosphate
groups will tend to be further apart. However, if we define minor groove width
by measuring cross-strand O4'-O4', O4'-S4' or S4'-S4' distances (Table
2
) we can see that in d(CGCGAASSCGCG) there appears to be a slight
narrowing
of the minor groove with respect to that in d(CGCGATTCGCG). In the region of the 4-thiosugars it is tempting to say that this apparent narrowing of the groove is due to
the increased van der Waals radius of sulphur compared with oxygen. However, as
the narrowing is observed at all points at which the minor groove is measured,
it may be that it is a result of the refinement procedure used. Given that the
average difference in minor groove width is only 0.5 Å, it is probable that incorporation of 4-thiosugars into this oligonucleotide has not significantly modified
the accessibility of the floor of the minor groove and that potential protein-DNA interactions in this area are unlikely to be affected.
Table
3
details the values of the sugar-phosphate backbone torsion angles in the structures of both the modified
and native dodecamers. It shows that there are a number of differences in the
torsion angle values for the two structures, particularly in the regions
containing the 4'-thiosugars (residues S7, S8, S19 and S20). Although there is no
real consistency in the differences over both strands, there is a suggestion of
an increase in the value of the torsion angle [delta] for the residues which have 4'-thiosugars. This is indicative of a change in the sugar
conformations of the modified residues on going from the native to the modified
duplex. This indication is supported if one examines the deoxyribose
pseudorotation phase angles (
P
) in the structure of d(CGCGAASSCGCG) (Table
1
).
For all the modified sugars the average value of
P
is 197o, which is close to those values (178o and 180o) found in the crystal structures of two 4'-thio-2'-deoxynucleosides (
3
), but much higher than the average value of
P
(116o) found for the analogous residues in the crystal structure of d(CGCGAATTCGCG) (
9
). The values of
P
suggest (
19
) that the 4'-thiosugars have a conformation close to C
3'
-
exo
, rather than the approximate C
1'
-
exo
found for the analogous sugars in the native structure.
This observation is supported if we compare the pseudorotaion angles in
d(CGCGAASSCGCG) with those in the 2BNA (
16
), 7BNA (
17
) and 9DNA (
18
) native structures (Fig.
7
a). In each case the pseudorotation angle for the modified sugars is greater
than those of the analagous sugars in the native dodecamer, regardless of the
refinement technique used or the temperature at which the native structure was
determined. Moreover, the values of the pseudorotation angles in the 2BNA, 7BNA
and 9BNA native structures closely mimic those of the 1BNA structure (Fig.
7
a).
A surprising observation in the structure of d(CGCGAASSCGCG) is that the sugar
puckering amplitude ([tau]) of the 4'-thio-2'-deoxyribose groups appears to be generally
lower than those for the analagous sugars in the structures of the native oligonucleotide
(Fig.
7
b). The cause of our suprise is that one might reasonably expect that in 4'-thiosugars the increase in the length of the C-S bonds with respect to the C-O bonds in unmodified sugars would lead to an
increase in the sugar puckering amplitude. Indeed, such an increase is observed
if one compares [tau] in the structures of 4'-thiothymidine (47.9o) and (
E
)-5-(2-bromovinyl)-2'-deoxy-4'- thiouridine (48.6o) (
3
) with that of free thymidine (37.8o) (
20
).
Figure 6
.
Comparison of individual values of (
a
) helical rise and (
b
) helical twist in the structures of d(CGCGAASSCGCG) ([diamonds]) and the 1BNA structure of d(CGCGAATTCGCG) (+).
A comparison of cross-strand O4'-O4', O4'-S4' and S4'-S4' distances
(Å) which define minor groove width in d(CGCGAASSCGCG) and the 1BNA
structure of d(CGCGAATTCGCG)
d(CGCGAATTCGCG)
d(CGCGAASSCGCG)
O4'(4)-O4'(24)
7.0
6.8
O4'(5)-O4'(23)
6.9
6.0
O4'(6)-O4'(22)
5.9
6.1
S4'(7)-O4'(21)
a
4.6
4.1
S4'(8)-S4'(20)
a
4.3
3.7
O4'(9)-S4'(19)
a
3.7
3.2
O4'(10)-O4'(18)
4.7
4.4
O4'(11)-O4'(17)
4.5
3.8
O4'(12)-O4'(16)
5.9
6.3
In order to take into account the van der Waals radii of the atoms the distances
quoted are those measured less 2.8, 3.25 and 3.7 Å respectively.
a
In d(CGCGAATTCGCG) the S4' atoms are, in fact, O4' atoms
Figure 7
.
Comparisons of the values of (
a
) pseudorotation angle and (
b
) sugar puckering amplitude ([tau]) found in the structure of d(CGCGAASSCGCG) and in the 1BNA, 2BNA, 7BNA and
9BNA structures of d(CGCGAATTCGCG).
Do the observed differences between the native and the modified oligonucleotide
structures help to explain the resistance to digestion by restriction
endonucleases of oligonucleotides containing 4'-thiosugars? The extremely high specificity of restriction enzymes is due to two major types of interaction between the
enzyme and its cognate DNA sequence (
19
). These are: (i)
direct readout
of the base sequence of the cognate DNA resulting from interactions between the
enzyme and the major groove functional groups of the DNA; (ii)
indirect readout
of the base sequence of the cognate DNA which occurs via interactions between
the endonuclease and the sugar-phosphate backbone of the DNA. In addition to these modes of inducing sequence specificity there are also
non-sequence-specific interactions
between the protein and the sugar-phosphate backbone of the DNA which usually help to increase the overall
binding affinity of the DNA for the protein.
In some cases these interactions can only occur if there is a distortion of the
DNA sequence from the canonical B-form and it had been suggested (
7
) that the presence of thiosugars in oligonucleotides would `rigidify' them and
mean that the sequences could not undergo the distortion required for them to
be good substrates for their restriction endonuclease. However, analysis of the
structure of
Eco
RV endonuclease complexed with a thiosugar-modified cognate DNA sequence (Kostrewa,D., Hancox,E.L., Walker,R.T.,
Connolly,B.A. and Winkler,F.K., unpublished results) has shown that the DNA
adopts the same kinked structure observed in the complex of the endonuclease
with an unmodified cognate DNA sequence (
8
), but, crucially, that the Mg
2+
ion, which is an absolute requirement for endonuclease activity, is not
present.
.
Sugar-phosphate backbone and glycosyl torsion angles, pseudorotation phase
angles (
P
) and sugar puckering amplitudes ([tau]) for the structures of d(CGCGAASSCGCG) and d(CGCGAATTCGCG) (1BNA)
Residue
[alpha]
[beta]
[gamma]
[delta]
[epsilon]
[zeta]
[chi]
P
[tau]
C1
154
172
213
200
272
181
50
174
157
218
216
255
161
60
G2
312
152
36
144
207
221
248
147
43
295
170
40
128
174
262
249
140
43
C3
312
134
58
83
199
260
215
42
46
298
172
59
98
183
272
225
93
39
G4
312
163
49
159
199
215
268
175
42
297
180
57
155
205
208
268
167
52
A5
304
164
49
147
198
248
250
166
37
317
143
53
119
180
268
234
129
48
A6
286
171
56
119
186
267
247
133
34
287
180
66
121
174
272
238
127
52
S7
273
171
66
138
261
175
288
178
27
303
181
52
99
174
274
232
101
49
S8
69
229
195
151
205
214
262
218
30
301
174
64
109
170
271
234
116
51
C9
56
132
318
159
198
256
266
181
36
302
180
61
129
203
266
241
141
49
G10
292
183
40
156
265
145
281
179
42
293
169
47
143
257
150
270
146
58
C11
286
145
45
147
200
258
251
166
42
286
139
56
136
198
271
235
148
50
G12
287
177
51
97
262
102
35
279
176
57
111
248
113
54
C13
57
156
251
193
238
163
43
56
137
201
234
232
153
45
G14
298
137
43
145
203
218
248
141
46
309
164
49
122
178
267
244
128
47
C15
316
129
62
82
200
262
226
48
43
297
169
61
86
175
262
226
68
45
G16
298
157
73
123
151
283
248
146
27
291
171
73
136
174
262
245
149
42
A17
286
198
58
151
195
213
252
158
42
303
190
54
146
177
263
254
169
45
A18
56
181
282
193
225
200
254
205
44
303
186
48
130
174
259
251
146
45
S19
297
132
48
141
238
217
286
194
37
302
173
60
109
179
272
228
116
49
S20
285
129
68
134
197
265
267
199
24
301
179
55
122
179
266
240
129
52
C21
278
196
53
126
191
264
258
139
34
301
185
45
110
183
274
246
114
44
G22
292
180
41
156
240
172
288
176
42
293
179
50
150
260
171
271
156
54
C23
304
161
28
133
189
245
260
176
32
289
138
45
113
186
263
235
117
46
G24
310
143
55
73
223
24
44
295
171
47
79
225
35
48
Values for d(CGCGAASSCGCG) are given on the first line and for d(CGCGAATTCGCG)
on the second line in italic. All angles are in degrees. All parameters were
calculated using the NEWHEL92 program distrubuted by R.E.Dickerson through the
Brookhaven Protein Data Bank (15).
This observation, coupled with the fact that our analysis of the structure of
d(CGCGAASSCGCG) reveals that the major effect of incorporation of 4'-thio-2'-deoxyribose groups into a DNA duplex is to
change its backbone conformation, suggests the following reasons for the
resistance to nuclease digestion of oligonucleotides containing 4'-thiosugars. (i) Changes in the backbone conformation result in the
enzyme being able to bind less well to the modified DNA than to its unmodified
cognate sequence, thus causing a reduction in activity of the endonuclease;
(ii) changes in backbone conformation resulting from DNA modification cause a
reduction in Mg
2+
ion binding, which again would result in a reduction in the activity of the
enzyme. The decision as to which, if either, of these two hypotheses is correct
would require further analysis of the structures of 4'-thiosugar-modified DNAs, both in isolation and, more importantly,
complexed to their specific restriction endonucleases.
We conclude with the observation that the inclusion of 4'-deoxy-2'-thiothymidine into DNA has a small, but
detectable, effect on the backbone conformation of a DNA duplex. On occasions
this may appear to be of little significance, but should it be associated with
the conferring of increased nuclease resistance or increased lipophicity it may
make such oligodeoxynucleotides candidates for antisense molecules.
ACKNOWLEDGEMENTS
This work was supported by the United Kingdom Engineering and Physical Sciences
Research Council, the Biotechnology and Biology Science Research Council and the Wellcome Trust. We are grateful to the
Wellcome Foundation for financial support (to ELH). We thank Drs Miroslav Papiz
and Pierre Rizkallah of the Synchrotron Radiation Source, Daresbury Laboratory, for the allocation of
contingency beam time, part of which was used to collect the data described
here. TB wishes to thank the Royal Society of Edinburgh for a Caledonian
Research Fellowship.
REFERENCES
1 Dyson,M.R., Coe,P.L. and Walker,R.T. (1991) Carbohydrate Res., 216, 237-248.
2 Secrist,J.A., Tiwari,K.N., Riordan,J.M. and Montgomery,J.A. (1991) J. Med. Chem., 34, 2361-2366.
3 Dyson,M.R., Coe,P.L. and Walker,R.T. (1991) J. Med. Chem., 34, 2782-2786.
15 Abola,E.E., Bernstein,F.C., Bryant,S.H., Koetzle,T.F. and Weng,J. (1987) In Allen,F.H., Bergerhoff,G. and Seivers,R. (eds), Crystallographic Databases-Information Content, Software Systems, Scientific Applications. Data Commission of the International Union of Crystallography, Cambridge, UK.
16 Drew,H.R., Samson,S. and Dickerson,R.E. (1982) Proc. Natl. Acad. Sci. USA, 79, 4040-4044.MEDLINE Abstract
17 Holbrook,S.R., Dickerson,R.E. and Kim,S.-H. (1985) Acta Crystallogr., B41, 255-262.
18 Westhof,E. (1987) J. Biomol. Struct. Dyn., 5, 581-600.
19 Saenger,W. (1984) Principles of Nucleic Acid Structure. Springer-Verlag, New York, NY.
20 Young,D.W., Tollin,P. and Wilson,H.R. (1969) Acta Crystallogr., B25, 1423.