The solution structure of the DNA dodecamer d(CGCGAATCCCG)(2) has been stud
ied in an aqueous liquid crystalline medium containing 5% w/v bicelles. The
se phospholipid particles impose a small degree of orientation on the DNA d
uplex molecules with respect to the magnetic field and permit the measureme
nt of dipolar interactions. Experiments were carried out on several samples
with different isotopic labeling patterns, including two complementary sam
ples, in which half of the nucleotides were uniformly enriched with C-13 an
d deuterated at the H2 " and H5' positions. From this, 198 C-13-H-1 and 10
N-15-H-1 one-bond dipolar coupling restraints were derived, in addition to
200 approximate H-1-H-1 dipolar coupling and 162 structurally meaningful NO
E restraints. Although loose empirical restraints for the phosphodiester ba
ckbone torsion angles were essential for obtaining structures that satisfy
all experimental data, they do not contribute to the energetic penalty func
tion of the final minimized structures. Except for additional regular Watso
n-Crick hydrogen bond restraints and standard van der Waals and electrostat
ic terms used in the molecular dynamics-based structure calculation, the st
ructure is determined primarily by the dipolar couplings. The final structu
re is highly regular, without any significant bending or kinks, and with C2
'-endo/C1'-exo sugar puckers corresponding to regular a-form DNA. Most loca
l parameters, including sugar puckers, glycosyl torsion angles, and propell
er twists, are also tightly determined by the NMR data. The precision of th
e determined structures is limited primarily by the uncertainty in the exac
t magnitude and rhombicity of the alignment tensor. This causes considerabl
e spread in parameters such as the degree of base-pair opening and the widt
h of the minor groove, which are relatively sensitive to the alignment tens
or values used.