NMR Spectroscopic Determination of the Solution Structure of a Branched Nucleic Acid from Residual Dipolar Couplings by Using Isotopically Labeled Nucleotides

Article abstract of DOI:10.1002/anie.200351632

Only a small set of magnetic-field-induced residual dipolar couplings is required to determine the global structure of branched nucleic acids. This is demonstrated for the example of the Holliday junction (shown schematically) after ¹³C labelin

Full text of DOI:10.1002/anie.200351632

Angewandte
Chemie
Branched Nucleic Acids
NMR Spectroscopic Determination of the
Solution Structure of a Branched Nucleic Acid
from Residual Dipolar Couplings by Using
Isotopically Labeled Nucleotides**
Bernd N. M. van Buuren, Jürgen Schleucher,
Valentin Wittmann, Christian Griesinger,
Harald Schwalbe, and Sybren S. Wijmenga*
Branched nucleic acids play important biological roles, and
well-known examples such as the Hammerhead ribozyme, the
all-RNA enzyme, a three-way junction (3H according to
IUPAC nomenclature[1a]), and the Holliday junctions[1b] or
DNA four-way junctions (J6, Figure 1a; 4H according to
IUPAC nomenclature[1a]) have recently been reviewed.[2a–c]
The Holliday junctions form the central intermediates in both
homologous and site-specific recombination. In solution, in
the presence of multivalent ions and thus under physiological
salt conditions, their structure consists of two quasi-continu-
ously stacked helices in an essentially canonical conformation
(Figure 1c).[2b,c] Four recent crystal structures of 4H[2d–g]
confirm these structural features, but no coherence is
observed in the value of the interhelix angle y (defined in
Figure 1b); three structures show right-handed interhelix
[*] Prof. Dr. S. S. Wijmenga
Laboratory of Physical Chemistry-Biophysical Chemistry
University of Nijmegen
Toernooiveld 1, 6225ED Nijmegen (The Netherlands)
Fax: (+31)24-3652112
E-mail: sybrenw@sci.kun.nl
Dr. B. N. M. van Buuren
Unilever Research & Development
Olivier van Noortlaan 120, 3133AT Vlaardingen (The Netherlands)
Dr. J. Schleucher
Department of Medical Biochemistry and Biophysics
Umeå University, S-90187 Umeå (Sweden)
Dr. V. Wittmann
Johann Wolfgang Goethe-Universität Frankfurt/M.
Institut für Organische Chemie und Chemische Biologie
Marie-Curie-Strasse 11, 60439 Frankfurt/M. (Germany)
Prof. Dr. C. Griesinger
Max-Planck-Institut für biophysikalische Chemie
Am Fassberg 11, 37077 Göttingen (Germany)
Prof. Dr. H. Schwalbe
Johann Wolfgang Goethe-Universität Frankfurt/M.
Zentrum für Biomolekulare Magnetische Resonanz
Institut für Organische Chemie und Chemische Biologie
Marie-Curie-Strasse 11, 60439 Frankfurt/M. (Germany)
[**] This work was supported by grants from the Kempe Minne
Foundation (BVB), the Swedish National Research Council (SW),
Bioteknik Medel Umeå University (SW), the DFG (GR1211/2-4,
En111/11-3) (C.G., H.S.), the Karl Winnacker foundation (H.S.),
and the European LSF for Biomolecular NMR (ER-BCT950034). We
thank Janny Hof for critical reading of the manuscript.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author. Included are
the fit results, the original JCH values as a function of magnetic field,
and the derived mRDCs.
Angew. Chem. Int. Ed. 2004, 43, 187 –192
DOI: 10.1002/anie.200351632
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
187

Communications
example, the anisotropy of the magnetic susceptibility,[8,9]
a
pronounced feature of double-stranded oligonucleotides,
induces a partial alignment of a molecule in strong homoge-
neous magnetic fields and thereby gives rise to magnetic-
field-induced RDCs (mRDCs).[5a–d,j–n,9]
The anisotropy of the molecular magnetic susceptibility
tensor (cmol) is caused by the magnetic susceptibility of the
individual nucleobases (cb). The latter are dominated by base
ring currents with the long axis (z) of cb perpendicular to the
base plane. Their principal components (cb,xx, cb,yy, cb,zz) can be
written as (0,0,ꢀ13) ” 10ꢀ34 m3 in SI units. The average value
of cb,zz per base can be calculated on the basis of the Cotton–
Mouton effect[8] or from quantum mechanics[10] with good
accuracy (Tables 1 and 2). cmol can be calculated as the tensor
Table 1: Estimate of the base c values.
CG [ꢀ10ꢀ34 m3]
AT [ꢀ10ꢀ34 m3]
ElOPt[a]
JB[b]
ꢀ25.4
ꢀ20.4
ꢀ23.8
ꢀ26
ꢀ19.9
ꢀ26.2
ꢀ30.4
ꢀ36
HM[b]
GP[b]
Average all
Average base
Figure 1. a) Sequence and arm designation for J6. The 13C-labeled thy-
ꢀ26ꢂ5
mines in two samples are indicated as boxed residues (solid/dashed
outline). b) The interhelix angle y, strand numbering, and conforma-
tion designations following 4H nomenclature. The parallel and
antiparallel orientations of the AD and BC helices are defined by the
corresponding drawings. Left-handed and right-handed indicate the
orientation of the AD helix relative to the BC helix. The grey region was
excluded for steric reasons. c) The chosen molecular axis frame and
designation of the coaxially stacked helices.
ꢀ13ꢂ2.5
[a] Value of c from reference [8]; [b] JB: c from ring currents with ring
currents obtained using Johnson–Bovey equations from reference [10a];
HM: c from ring currents with ring currents obtained using Haig–
Mallion equations from reference [10a]; GP: c from ring currents with
ring currents obtained from reference [10b]. For a CG base pair the ring
currents of C and G were added; an analogous method was used for an
AT base pair. The values for the ring currents are relative to those for
benzene and proportional to the magnetic susceptibility tensor. The c
values of the CG and AT base pairs were obtained by multiplying the ring
current with c for benzene (ꢀ12.7”10ꢀ34 m3[9]). The GP ring currents
were multiplied with a factor 1.7 to take into account the fact that GP
used lower ring currents values, because part of the base susceptibility is
attributed to bond magnetic susceptibility. The c tensor is oriented
perpendicular to the base. Consequently, its principal components can
be written as (0,0,czz), where czz indicates the z principal component. The
average of all different base pair estimates leads to an average value czz
per base of 13ꢂ2.5”10ꢀ34 m3.
angles,[2d–f] while the angle in the fourth structure is left-
handed.[2g] In the solution structure[3] of the J6 model for a
Holliday junction, based on NMR spectroscopy using NOE
and J coupling constants, the quasi-continuously stacked
helices AD and BC[3a] have a left-handed interhelix angle
(Figure 1b). The determination of the relative helix orienta-
tions is the central question concerning the structure of
branched nucleic acids.
Herein we describe how a relatively small set of long-
range NMR restraints can be used to determine the global
structure of branched nucleic acids. This is shown with the
example of the Holliday junction, and the previous assign-
ment of a left-handed interhelix angle is confirmed. The
approach relies on the measurement of a set of residual
dipolar couplings (RDCs) in a DNA sample with thymidine
residues that are 13C labeled in the deoxyribose moiety of the
nucleotide[4] (Figure 1a). This chemical labeling strategy
allows rapid determination of the overall conformation of
branched nucleic acids.
Table 2: Ring currents calculated for the four nucleic acid bases G, A, T,
and C.[a]
GP
JB
HM
G
A
T
0.94(0.62)
1.56(1.00)
0.11(0.07)
0.28(0.18)
1.30(0.73)
1.78(1.00)
0.28(0.16)
0.31(0.17)
1.51 (0.74)
2.04 (1.00)
0.35 (0.17)
0.37 (0.18)
C
[a] GP, JB, and HM are as defined in Table 1; the values in parentheses
indicate the ring current values relative to that for adenine.
In NMR spectra of liquid samples, RDCs contain
information about long-range orientation and complement
the local NMR restraints, NOEs, and J coupling constants.
The RDCs are observed for molecules for which isotropic
rotational tumbling is partially restricted.[5–7] This restriction
can be induced by non-isotropic media such as liquid crystals
or phages.[5a,e–h] However, partial alignment can also be
conferred by the intrinsic properties of a molecule. For
sum of the individual cb values as long as the structure of a
nucleic acid molecule is known [Eq. (1)].
N
X
!T
!
cmol
¼
ð n b;i ꢁ n b;iÞcb;zz;i
ð1Þ
i¼1
188
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2004, 43, 187 –192

Angewandte
Chemie
Here, nb is a unit row vector perpendicular to the base
plane i, N the number of bases in the molecule, and the
symbol ꢁ represents the direct matrix product of the column
and row vector.
can be derived. Thus, for the first set of two solutions (set 1),
the 4H conformation is defined by the angles wAD1 and wBC1
with the interhelix angle equal to y1 and y1 + 1808. The
second set of two solutions (set 2) is defined by the angles
wAD1 and wBC1 + 1808 with the interhelix angle again equal to
y1 and y1 + 1808. Set 2 is sterically disallowed, because
strands 1 and 3(Figure 1b) cross from the AD to the BC
helix. Furthermore, wAD1 + 1808 leads only to equivalent
solutions involving rotation of the whole molecule. The two
solutions of set 1 are reduced to one unique solution when
ꢀ608 ꢃ y < ꢀ1808, because the crossing strands intermingle
for 608 ꢃ y < 1808[12] and thus exclude this region (Figure 1b).
As shown below, we found a value close to ꢀ908 and thus one
unique solution was obtained.
When non-isotropic media[5a,e–h,6,7] are used to align
molecules, the size (axial and rhombic component) and
orientation (three Euler angles) of the alignment tensor in a
molecular frame have to be derived from the experimental
RDCs; this requires a minimum of five RDCs. The relative
domain orientation in an n domain molecule with known
domain structure is defined by three Euler angles for each
pair of domains, so that in addition a minimum of three
parameters (RDCs) are needed to derive the relative
orientation of each domain.[5a,d,7] In the procedure proposed
by Dosset et al.,[7f] for each domain an alignment tensor is
first derived and the relative domain orientations are
subsequently obtained by reorienting each domain such that
all alignment tensors have the same common orientation. This
procedure can be simplified in the case of magnetic-field-
induced alignment, because the size and orientation of the
alignment tensor can be calculated from each trial structure.
Such reduction in complexity is also obtained if the size and
shape of the alignment tensor can be predicted from the shape
and/or charge distribution of the molecule.[7d,e]
For each conformation cmol was calculated and subse-
calcd
CH
quently D
determined. The optimal conformation was
calcd
CH
exp
found by comparing D
with D as the minimum root
CH
mean square difference (rmsd). In total 28 mRDCs were
derived from least-squares fits of 1JCH coupling constants and
measured[13] at four different magnetic fields (400–800 MHz;
see Supporting Information). They are given in Table 3as
Table 3: Experimental dipolar couplings D of the model Holliday
junction J6, as determined from measurements at four magnetic fields.[a]
1
Residual dipolar couplings DCH have been measured in
Arm
Residue
D(C1’,H1’)
D(C3’,H3’)
D(C4’,H4’)
the sugar moiety of ten thymine residues of the J6 model. The
[Hz]
[Hz]
[Hz]
ꢀ
residual dipolar coupling DCH,i of a dipolar C H vector i can
A
A
A
T5
T35
T40
ꢀ1.7[b]
1.2
ꢀ0.2
0.0
0.1
0.6
1.4
be calculated in a chosen molecular reference frame by direct
!
0.6
matrix multiplications of cmol and n CH,i, where the latter is the
0.6
ꢀ
unit row vector pointing along the C H vector i expressed in
the chosen reference frame [Eq. (2)].
D
D
T29
T32
ꢀ1.2
ꢀ3.0
0.7
ꢀ0.3
2.1
ꢀ0.6
ꢀ
ꢁꢀꢂ
ꢃ
!
! ! !
n CH;i c mol n TCH;i
h
1
1
3
DCH;i ¼ ꢀS
gC gH rꢀ3
B2
2p2
CH 4p 15 k T o 4
B
B
B
T9
T12
T13
0.5
–
0.5
ꢀ0.7
0.2
1.1
3.1
1.8
2.3
ð2Þ
ꢂ
ꢃꢁ
!
!
1
ꢀ
Trace c mol
3
C
C
T20
T21
1.6
1.0
–
1.6
1.8
2.3
Here, h is Planck's constant, gC and gH are the gyromag-
netic ratios of the 13C and 1H nuclear spins, rCH is the distance
between the directly bonded C and H nuclei, kT is the
thermal energy, and Bo is the magnetic field strength in Tesla.
The order parameter S[5a,d,11] takes into account the reduction
in the size of DCH,i by intramolecular dynamics and has been
assumed to be 1 here.
[a] The dipolar couplings were determined from a least-squares fit of the
magnetic field dependence of the 1JCH coupling constants of H1’-C1’, H3’-
C3’, and H4’-C4’ at 400, 500, 600, and 800 MHz. The measured 1JCH value
can be written as 1JCH =aBo2 + 1JCHo, where 1JCHo represents the 1JCH value
at a magnetic field of zero and the slope a contains the DCH information
(a=DCH/B2o). For the least-squares fit the above equation was used. The
D
CH at a given field is then equal to (aB2o). The DCH values were calculated
The relative orientation of the two semi-continuously
stacked helices of 4H (Figure 1c) is defined by three
parameters. In any axes frame, the AD helix orientation is
specified by its three Euler angles, the polar angles, qAD and
from the slope as the difference DDCH between values at 800 and
400 MHz. Thus, they correspond to DCH measured at an effective
magnetic field of (64–16)0.5 ”100 MHz=693 MHz. Two samples were
used with 13C labeling in the ribose; sample 1 was labeled at T5, T9, T12,
T20, T32, and T35; and sample 2 at T13, T12, T29, and T40. [b] The
uncertainty in the DCH values is estimated to be ꢂ0.5 Hz.
f
AD, and the rotation around the helix axis wAD (BC helix
orientation: qBC, fBC, and wBC). We defined a right-handed
reference frame in which the AD helix lies along the z axis
(qAD = 0, fAD = 0) and the x axis is perpendicular to the plane
spanned by the AD and BC helix axes. The BC helix then
rotates by definition in the yz plane (fBC = 90) and qBC equals
the interhelix angle y (qAD = 0). In this molecular axis frame,
the conformation of 4H is defined by the interhelix angle y,
and the angles wAD and wBC indicate the rotation about their
own axes. From a given set of RDCs (minimum of three
required), two sets of solutions for the conformation of 4H
D1JCH between 400–800 MHz and denoted as DCH. A variety
of optimizations were performed with model helices or
helices derived from two NOE-based experimental structures
(Figure 2 and Supporting Information). Most striking is that y
always has a well-defined minimum (ꢀ888 < yopt < ꢀ988,
depending on the optimization, Figure 2a). This is more
negative than the value of approximately ꢀ708 in the NOE-
Angew. Chem. Int. Ed. 2004, 43, 187 –192
www.angewandte.org
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
189

Communications
mRDCs (see Supporting Information). Variation in sugar
puckering and orientation may affect the orientation of these
vectors (but not H1’-C1’). Note that in DNA the energy well
for the S-puckered deoxyribose is relatively wide, and sugars
are often not found to be 100% S-puckered even in regular
helices. Furthermore, the ribose rings are oriented approx-
imately parallel to the helix axis, so that the sugar pucker
motion, repuckering, and reorientation will mainly affect the
w angles of these vectors. Therefore, we operationally
included small additional w rotations, where the H3’-C3’
and H4’-C4’ mRDCs were treated as separate groups in the A,
B, C, and D helices, and found indeed improved fit results (see
Supporting Information). However, the number of adjustable
parameters (ꢃ 11) remains much smaller than the number of
experimental mRDCs (28).
Model helices gave similar results and the best fits
(Figure 2a, red curve), and the mRDC-optimized 4H struc-
tures nearly overlap (Figure 2c, red structure). Compared
with the NOE-based structure, the main change, apart from
the small change in y, is that the BC helix has rotated by
about 408 around its own axis (Figure 2c, yellow versus red
structure), leading to a net opening of the cavity formed by
the facing major grooves. In the mRDC-optimized NOE-
based structure (Figure 2c, blue structure), the BC helix has
rotated approximately 408 further. Optimization of the other
experimental structure leads to a smaller cavity opening, and
the resulting structure is close to the yellow structure shown in
Figure 2c. Generally, the fits using experimental helices are
not as good (Figure 2a, red versus blue and black structures),
thus leading to the larger uncertainty in the value of wBC. This
can be attributed to the difficulty of using NOEs to define
global features even within the regular helix domains.
We have assumed a single rigid structure for the calcu-
lation of cmol and DCH [Eqs. (1) and (2)]. It is of interest to
consider potential hinge motion, that is, motion of the BC
helix with respect to the AD helix. Internal motion is usually
taken into account by an order parameter S [Eq. (2)], which
simply scales the mRDCs, and the optimized angles should be
viewed as effective angles.[11] Assuming that the potential
hinge motion can be accounted for in this way a rough
estimate of its amplitude can be obtained. The errors in cmol
and DCH are about 10% and 15%, respectively, which
correspond to an overall error of around 18% in the tensor
and allows for an overall S value of about 0.82. If we further
assume that the hinge motion can be described by axial
wobbling in a cone according to 12 cose(1 + cose) = S, the
amplitude of this motion would imply an opening half-angle
of about 288. Thus, this would allow replacement of the model
with a fixed orientation of the two helices with an interhelical
angle of about ꢀ908 by a model in which the structure has an
effective angle of ꢀ908 with a spread of around ꢂ 288. It
should be noted, however, that the effect of hinge motion is
more complex than a simple scaling of the dipolar couplings,
because the hinge motion differently affects the axial and
rhombic components of cmol as well as its orientation with
exp
CH
calcd
CH
Figure 2. a) Plot of the rmsd of D and D
versus y. The AD and
BC helices were modeled as B-DNA helices capped with a known
stable CTTG loop[14] (red, r) or taken from the NOE-based structure set
(blue, black, z: two structures close to average). Model helices (r): fit
with 3 (r1) or 3+7 parameters (r2) or as r2 with base-specific cb,zz (r3;
see Supporting Information). Experimental helices: fit with 3 (b1, z1)
or 3+8 parameters (b2). b) Monte Carlo estimate of the spread in y of
fit r1 (see the text). The fitting was repeated for 100 samples of the 28
mRDCs with random variation in their values according to normal dis-
tribution with s=0.5 Hz. c) Stereo views of the mRDC-optimized 4H
(corresponding to fits r1–r3 and b2) and the NOE-based 4H (yellow,
optimized is b2) with chosen axes frame (see the text). cmol essentially
coincides with this chosen axes frame (see Supporting Information).
based structure. The Monte Carlo (MC) spread in y due to
the 0.5-Hz experimental error on DCH is small (rmsd 3–68,
Figure 2b and Supporting Information). The angles wAD and
wBC also have well-defined minima, but their MC errors are
somewhat larger (rmsd 6–158; see Supporting Information).
The use of base-specific values instead of one average value
for cb,zz in the calculation of cmol hardly affects the fit results
(Figure 2a). Similarly, increasing or decreasing the average
cb,zz value by one standard deviation does not change the fit
results (see Supporting Information). To verify these results,
cross validation was carried out on the three-parameter fit
M(3)_r1 (Figure 2) by randomly removing six of the
28 mRDCs in 100 different samples. For each sample the
fitting was carried out and the MC spread determined to yield
average values and deviations (rmsd) for yopt, wAD,opt, and
wBC,opt of ꢀ93(3) 8, ꢀ1598(68) and ꢀ1828 (68), respectively.
The similarity of these values with those found for the other
fit results (see Supporting Information) further confirms the
conclusions.
ꢀ
respect to the AD and BC helices (for certain C H vectors
Although the fits with three adjustable parameters gave
well-defined minima, the correlation between D and D
showed a systematic offset for the H3’-C3’ and H4’-C4’
hinge motion may even lead to an increase in DCH). An
accurate estimation of the effect of hinge motion would
require simulation of these motions.
exp
calcd
CH
CH
190
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2004, 43, 187 –192

Angewandte
Chemie
In summary, we have shown with the
example of the Holliday junction that the
global orientation of branched nucleic acids
can be derived from a small set of mRDCs
by taking advantage of the reduction in
required parameters, thus extending
approaches based on phage-induced
RDCs.[15] The same type of analysis can be
extended to other branched nucleic acids in
a relatively straightforward fashion. For
instance, under physiological salt condi-
tions, RNA and DNA 3H show coaxial
stacking of two of the three arms. The global
conformation is then defined by three
angles that describe the relative helix ori-
entations and would require the same mini-
mum number of mRDCs as for the 4H
discussed here. Similarly, for more strongly
branched nucleic acids, such as 5H and 6H
sometimes found in RNAs, the global con-
formation is defined by the relative orien-
tation of three helices when coaxial stacking
is present and would require a minimum of
six mRDCs (six Euler angles define the
relative helix orientations). The mRDC size,
although sufficiently large here, can easily
be increased by using larger magnetic fields
or larger nucleic acids (e.g. extension of
Scheme 1. Synthesis of 6. The asterisks indicate the 13C-labeled positions. AIBN=azobis-
isobutyronitrile, Bz=benzyl, DIEA=N,N-diisopropylethylamine, DMAP=4-dimethylami-
nopyridine, DMTr=4,4’-dimethoxytriphenylmethyl, OTf=trifluoromethanesulfonate,
TBAF=tetrabutylammonium fluoride, TIPDS=tetraisopropyldisilyl, TMS=trimethylsilyl.
each of the 4H helices by four base pairs
max
[1] a) D. M. J. Lilley, R. M. Clegg, S. Diekmann, N. C. Seeman, E.
leads to DD
of 16 Hz for 900 vs. 400 MHz). Also, with
CH
von Kitzing, P. J. Hagerman, Nucleic Acids Res. 1995, 23, 3363 –
present cryoprobe technology, magnetic-field-induced dipolar
couplings can be determined at natural abundance.
3364; b) R. Holliday, Genet. Res. 1964, 5, 282 – 304.
[2] a) A. Kuzminov, Microbiol. Mol. Biol. Rev. 1999, 63, 751 – 813;
b) S. C. West, Annu. Rev. Genet. 1997, 31, 213– 244; c) D. M. J.
Lilley, Q. Rev. Biophys. 2000, 33, 109 – 159; d) J. Nowakowski,
P. J. Shim, G. S. Prasas, C. D. Stout, G. F. Joyce, Nat. Struct. Biol.
1999, 6, 151 – 156; e) B. F. Eichman, J. M. Vargason, B. H. M.
Mooers, P. S. Ho, Proc. Natl. Acad. Sci. USA 2000, 97, 3971 –
3976; f) M. Ortiz-Lombardía, A. Gonzµlez, R. Eritja, J. Aymamí,
F. Azorín, M. Coll, Nat. Struct. Biol. 2000, 6, 913– 917; g) J.
Nowakowski, P. J. Shim, C. D. Stout, G. F. Joyce, J. Mol. Biol.
2000, 300, 93– 102.
[3] a) B. N. M. van Buuren, J. Schleucher, S. S. Wijmenga, J. Biomol.
Struct. Dyn. 2000, 11, 237 – 243; b) B. N. M. van Buuren, T.
Herman, S. S. Wijmenga, E. Westhof, Nucleic Acids Res. 2002,
30, 507 – 514.
[4] S. Quant, R. W. Wechselberger, M. A. Wolter, K. Wörner, P.
Schell, J. W. Engels, C. Griesinger, H. Schwalbe, Tetrahedron
Lett. 1994, 35, 6649 – 6652.
[5] a) J. H. Prestegard, H. M. Al-Hashimi, J. R. Tolman, Q. Rev.
Biophys. 2000, 33, 371 – 424; b) A. Bax, N. Tjandra, Nat. Struct.
Biol. 1997, 4, 254 – 256; c) N. Tjandra, J. G. Omichinski, A. M.
Gronenborn, G. M. Clore, A. Bax, Nat. Struct. Biol. 1997, 4, 73 2 –
738; d) J. H. Prestegard, Nat. Struct. Biol. 1998, 5, 517 – 522;
e) A. Bax, N. Tjandra, J. Biomol. NMR 1997, 10, 289 – 292;
f) M. R. Hansen, L. Mueller, A. Pardi, Nat. Struct. Biol. 1998, 5,
1065 – 1074; g) G. M. Clore, R. M. Starich, A. M. Gronenborn, J.
Am. Chem. Soc. 1998, 120, 10571 – 10572; h) G. M. Clore, A. M.;
Gronenborn, N. Tjandra, J. Magn. Reson. 1998, 131, 159 – 162;
i) G. Cornilescu, L. J. Marquardt, M. Ottiger, A. Bax, J. Am.
Chem. Soc. 1998, 120, 6836 – 6837; j) H. C. Kung, K. Y. Wang, I.
Goljer, P. H. Bolton, J. Magn. Reson. Ser. B 1995, 109, 323 – 325;
k) H. M. Al-Hashimi, A. Majundar, A. Gorin, A. Kettani, E.
Experimental Section
[13C5]-Thymidine phosphoramidite (6): Two J6 samples (Figure 1a)
were synthesized according to Scheme 1 with thymidine residues 13C-
labeled in the deoxyribose moiety of the nucleotide[4] (see Figure 1a).
[13C6]Glucose was converted into 1,2-di-O-acetyl-3,5-di-O-benzoyl
ribofuranose (1) and glycosylated to give nucleoside 2 according to a
published procedure.[4] 2’-Deoxygenation was carried out following a
procedure described by Robins et al. for the synthesis of 2’-
deoxyuridine.[16] Briefly, 2 was deprotected and treated with the
Markiewicz reagent to give 3’,5’-protected nucleoside 3, which was
further converted into its 2’-O-phenoxythiocarbonyl derivative.
Reductive deoxygenation with tri-n-butyltin hydride and the free-
radical initiator AIBN in warm toluene provided the thymidine
derivative 4. Fluoride-induced removal of the silyl protecting group
and treatment with dimethoxytrityl chloride gave 5, which was
phosphitylated to yield phosphoramidite 6. The product was ready for
use in automated DNA synthesis.
Received: April 10, 2003
Revised: September 30, 2003 [Z51632]
Keywords: Holliday junctions · isotopic labeling · NMR
.
spectroscopy · nucleic acids · structure elucidation
Angew. Chem. Int. Ed. 2004, 43, 187 –192
www.angewandte.org
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
191

Communications
Skripkin, D. J. Patel, J. Am. Chem. Soc. 2001, 123, 633 – 640;
l) H. M. Al-Hashimi, A. Gorin, A. Majumdar, D. J. Patel, J. Am.
Chem. Soc. 2001, 123, 3179 – 3180; m) H. M. Al-Hashimi, Y.
Gosser, A. Gorin, W. Hu, A. Majumdar, D. J. Patel, J. Mol. Biol.
2002, 315, 95 – 102; n) H. M. Al-Hashimi, J. R. Tolman, A.
Majumdar, A. Gorin, D. J. Patel, J. Am. Chem. Soc. 2001, 123,
5806 – 5807; p) N. Sibille, A. Pardi, J. P. Simorre, M. Blackledge,
J. Am. Chem. Soc. 2001, 123, 12135 – 12146.
[6] a) N. Tjandra, S. I. Tate, A. Ono, M. Kainosho, A. Bax, J. Am.
Chem. Soc. 2000, 122, 6190 – 6200; b) A. Vermeulen, H. Zhou, A.
Pardi, J. Am. Chem. Soc. 2000, 122, 9638 9647.
[7] a) M. W. F. Fischer, J. A. Losonczi, J. L. Weaver, J. H. Preste-
gard, Biochemistry 1999, 38, 9013– 9022; b) N. R. Skrynnikov,
N. K. Goto, D. Yang, W. Choy, J. R. Tolman, G. A. Mueller, L. E.
Kay, J. Mol. Biol. 2000, 295, 1265 – 1273; c) E. T. Mollova, R.
Hansen, A. Pardi, J. Am. Chem. Soc. 2000, 122, 11561 – 11562;
d) C. A. Bewley, G. M. Clore, J. Am. Chem. Soc. 2000, 122,
6009 – 6016; e) M. Zweckstetter, A. Bax, J. Am. Chem. Soc. 2000,
122, 3791 – 3792; f) P. Dosset, J. C. Hus, D. Marion, M. Black-
ledge, J. Biomol. NMR 2001, 20, 223– 231.
[8] G. Weill in Molecular Electro-Optics (Ed.: S. Krause), Plenum,
New York, 1981, pp. 473– 483.
[9] A. A. Bothner-By in Encyclopedia of Nuclear Magnetic Reso-
nance (Eds.: D. M. Grant, R. K. Harris), Wiley, Chicester, 1995,
pp. 2932 – 2938.
[10] a) D. A. Case, J. Biomol. NMR 1995, 6, 341 – 346; b) C. Giessner
Prettre, B. Pullman, Q. Rev. Biophys. 1987, 20, 113– 172.
[11] J. Meiler, J. J. Prompers, W. Peti, C. Griesinger, R. Brꢀschweiler,
J. Am. Chem. Soc. 2001, 123, 6098 – 6107.
[12] E. von Kitzing, D. M. Lilley, S. Diekmann, Nucleic Acids Res.
1990, 18, 2671 – 2683.
[13] N. Tjandra, A. Bax, J. Magn. Reson. 1997, 124, 512 – 515.
[14] J. H. Ippel, H. Van den Elst, G. A. Van der Marel, J. H. Van
Boom, C. Altona, Biopolymers 1998, 46, 375 – 393.
[15] K. Bondensgaard, E. T. Mollova, A. Pardi, Biochemistry 2002,
41, 11532 – 11542.
[16] M. J. Robins, J. S. Wilson, F. Hansske, J. Am. Chem. Soc. 1983,
105, 4059 – 4065.
192
ꢀ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2004, 43, 187 –192