Using solid-state 31P{19F} REDOR NMR to measure distances between a trifluoromethyl group and a phosphodiester in nucleic acids

Article abstract of DOI:10.1016/j.jmr.2005.06.018

REDOR is a solid-state NMR technique frequently applied to biological structure problems. Through incorporation of phosphorothioate groups in the nucleic acid backbone and mono-fluorinated nucleotides, ³¹P{ ¹⁹F} REDOR has been used t

Full text of DOI:10.1016/j.jmr.2005.06.018

Journal of Magnetic Resonance 178 (2006) 11–24
www.elsevier.com/locate/jmr
Using solid-state P{19F} REDOR NMR to measure distances
31
between a trifluoromethyl group and a phosphodiester in nucleic acids
Elizabeth A. Louie, Panadda Chirakul, Vinodhkumar Raghunathan,
Snorri Th. Sigurdsson 1, Gary P. Drobny *
University of Washington, Chemistry Department, Campus Box 351700, Seattle, WA 98195-1700, USA
Received 6 March 2005; revised 25 June 2005
Available online 4 October 2005
Abstract
REDOR is a solid-state NMR technique frequently applied to biological structure problems. Through incorporation of phosp-
horothioate groups in the nucleic acid backbone and mono-fluorinated nucleotides, 31P{19F} REDOR has been used to study the
˚
binding of DNA to drugs and RNA to proteins through the detection of internuclear distances as large as 13–14 A. In this work,
31P{19F} REDOR is further refined for use in nucleic acids by the combined use of selective placement of phosphorothioate groups
and the introduction of nucleotides containing trifluoromethyl (–CF3) groups. To ascertain the REDOR-detectable distance limit
between an unique phosphorous spin and a trifluoromethyl group and to assess interference from intermolecular couplings, a series
of model compounds and DNA dodecamers were synthesized each containing a unique phosphorous label and trifluoromethyl
group or a single 19F nucleus. The dipolar coupling constants of the various 31P and 19F or –CF3 containing compounds were com-
pared using experimental and theoretical dephasing curves involving several models for intermolecular interactions.
Ó 2005 Published by Elsevier Inc.
Keywords: Solid-state nuclear magnetic resonance; 31P{19F} REDOR; Rotational-echo double resonance; Dipolar recoupling; Internuclear distance;
Trifluoromethyl group
1. Introduction
directly observed using REDOR [1,2]. For example,
31P{19F} REDOR has been implemented to measure di-
rect changes in long-range protein structure upon ligand
binding [3], and changes in minor groove width of DNA
oligomers upon binding distamycin [4,5] with reported
31P to 19F internuclear distances of approximately 16
Solid-state nuclear magnetic resonance (NMR) spec-
troscopy is widely used to study the structure of biomol-
ecules and biological complexes that are not easily
crystallized or dissolved, and thus cannot be convenient-
ly studied by X-ray crystallography and solution-state
NMR, respectively. The distance range accessible by
REDOR generally exceeds that of NOE or residual
dipolar coupling measurements, therefore structural fea-
tures that must be derived indirectly from short range
distance measurements using NOE data can often be
˚
and 13.5 A, respectively. These are among the longest
internuclear distances measured by NMR methods.
The heteronuclear dipolar coupling constant DIS is
directly related to the gyromagnetic ratio (cI, cS) of the
two spins (I,S) of interest and inversely proportional
to the cube of the internuclear distance rIꢀS : D / cI cS=
r3IꢀS. Therefore, nuclear spins with larger magnetic
moments, and thus the larger gyromagnetic ratios, will
result in larger dipolar coupling constants for a given
distance than nuclear spins with smaller gyromagnetic
ratios. The 13C nucleus is a mainstay for solid-state
*
Corresponding author. Fax: +1 206 685 8665.
E-mail addresses: snorrisi@hi.is (S.Th. Sigurdsson), drobny@
chem.washington.edu (G.P. Drobny).
1
Present address: University of Iceland, Science Institute, Dunhaga3,
IS-107 Reykjavik, Iceland.
1090-7807/$ - see front matter Ó 2005 Published by Elsevier Inc.
doi:10.1016/j.jmr.2005.06.018

12
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
31 19
surement of P– F dipolar couplings in the 12–16 A
distance measurements, with the longest 13C distances
measured by dipolar recoupling in the neighborhood
˚
range is feasible, in some cases REDOR measurements
are susceptible to interference from intermolecular cou-
plings. One way to extend the range of 31P REDOR is
to dephase 31P spins with the fluorines of a trifluoromethyl
(CF3) group. For instance, if an isolated 31P–19F spin pair
˚
of 6 A [3], corresponding to a dipolar coupling of about
35 Hz. In contrast, 19F has a gyromagnetic ratio over 3.5
times the size of 13C and two 19F spins separated by 6 A
˚
have a dipolar coupling of over 490 Hz, while two 19F
˚
˚
spins separated by 14.5 A have a dipolar coupling of
is separated by 16 A, the corresponding dipolar coupling
about 35 Hz. But in practice, 19F–19F homonuclear
constant is about 11 Hz. The corresponding amount of
REDOR dephasing is 11% at 30 ms. If instead the 31P spin
is coupled to the three 19F spins of a rapidly rotating triflu-
recoupling seems limited, at least in nucleic acids, to a
˚
maximum distance range of 9–10 A [6]; recent work by
McDermott and co-workers [7] using linear polymers
with distally attached trifluoromethyl groups advanced
oromethyl group, REDOR data roughly equivalent to the
31 19
˚
˚
16 A P– F pair are achieved for a distance of 19 A be-
tween the 31P spin and the ‘‘pseudo-atom’’ center of the
trifluoromethyl group. Recently, Schaefer and co-work-
ers [8] have determined the conformation of a bound tri-
˚
that limit to about 12 A. A complication that may arise
in 19F–19F recoupling that is at least partly responsible
for the attenuated upper limit of distance measurement
is incomplete decoupling of 19F–1H dipolar interactions.
For the purpose of measuring internuclear distances
fluoromethylketal
substituted
shikimate-based
bisubstrate inhibitor (CF3-SBBI) to EPSP using
greater than 12 A, 31P{19F} REDOR has shown consid-
˚
P{ F } REDOR and measured a distance of 8.3 A.
({19F3} is used to designate the dephasing of the three flu-
orines from a rotating –CF3 group.) However, a distance
31 19
3
˚
erable promise. A 31P spin, with a somewhat smaller
magnetic moment, separated by 11 A from a 19F spin
˚
˚
has a dipolar coupling of about 35 Hz. Due to the small-
er magnetic moment and because most biologically rele-
vant 31P spins are in phosphate groups, 31P–1H
decoupling during 31P{19F} REDOR is not as problem-
atic as 19F–1H decoupling during 19F–19F recoupling.
The long T2 of 31P under conditions of the REDOR
experiment enables the measurement of 31P–19F distanc-
of less than 12 A is believed to be somewhat moderate for
recoupling 31P to the 19F spin of a rotating –CF3 group
due to its high gyromagnetic ratio. Thus, investigating
the limitations of dipolar recoupling of 31P and 19F of a
rotating –CF3 group is worth exploring when structural
information is desired at the molecular level for large bio-
molecular systems.
˚
es of up to 16 A (DIS = 11 Hz) as demonstrated by
The basic principles of REDOR dephasing by multiple
spins in the presence of molecular motion have been dis-
cussed by Goetz and Schaefer [9], with specific reference
to 13C spins dephased by –CF3 groups. A primary feature
of this type of REDOR is that even though the 19F spins
are strongly dipolar coupled, the equality of the averaged
31P–19F dipolar couplings which results from the rota-
tion of the trifluoromethyl group removes the effect of
the 19F–19F couplings from the REDOR data, at least
to second order. Therefore, 31P{19F3} REDOR has
excellent prospects for applications in nucleic acid prob-
lems where a long-range distance measurement is
desired.
In this work, we accomplish two objectives. First, we
demonstrate the degree of perturbation introduced by
trifluoromethyl groups by comparative measurements
of DNA helix stability and by direct comparison of dis-
tances measured in DNA by 31P{19F3} REDOR. Sec-
ond, we use model compound studies and simulations
to assess the degree of interference in REDOR measure-
ments from intermolecular 31P–19F3 dipolar couplings.
Schaefer and co-workers [3] in a study of a complex of
the enzyme 5-enolpyruvylshikimate-3-phosphate (EPSP)
and shikimate-3-phosphate (S3P).
In nucleic acids, and especially in DNA, the redun-
dancy of the phosphodiesters in the backbone results
in a poorly resolved 31P NMR spectrum [4]. To enable
site-specific detection of 31P–19F distances during RE-
DOR, a single phosphate group in DNA or RNA is re-
placed by a phosphorothioate group. The 31P in a
phosphorothioate group is shifted approximately
55 ppm from the 31P phosphate group signal in DNA
and RNA [4]. Selective placement of phosphorothioate
groups in combination with 19F-substituted nucleotides
allows detection of 31P–19F distances site specifically.
31P–19F distances spanning the minor groove of a
DNA oligomer have been measured as a function of
distamycin–DNA binding stoichiometry [5], and for
˚
2:1 drug:DNA binding stoichiometry
a
13.5 A
(DIS = 19 Hz) distance was determined [5].
Even the 16 A limit achieved by 31P{19F} may be insuf-
ficient for some applications, and the acquisition time nec-
essary to observe the effect of weak 31P–19F dipolar
couplings in a REDOR experiment may prove impracti-
cal in some cases. For example, it may prove desirable
to measure changes in the major groove width of DNA,
or changes in RNA tertiary structure as a result of protein
binding. Such experiments may require measurement of
˚
2. Materials and methods
2.1. General
The first four model compounds (1–4) shown in
Fig. 1 were synthesized from 4-fluorophenol (1a),
31 19
˚
P– F distances of 16 A or more. Also, although mea-

E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
13
A except 95% A for initial, 15 min linear gradient to
30% A, isocratic for 10 min, then a 5 min linear gradient
to initial conditions.
2.2. Synthesis and purification of fluorinated DNA
oligomers (5 and 6) and their complementary sequence
The DNA, 50-d(CGCATT-(pS)-TT-5-FU-GCG)-30 (5),
50d(CGCATT-(pS)-TT-5-CF U-GCG)-3 (6), and 5 -(dC
GCAAAAATGCG)-30, were synthesized on an Applied
Biosystems Model 394 synthesizer. DNA incorporated
with 5-fluoro-20deoxyuridine was achieved by using 50-
(4,40-dimethoxytrityl)-30-(cyanoethyl)phosphoramidite-
20deoxy-5-(fluoromethyl)uridine (Glen Research). The
usual acetyl (Ac) protected dT, dG, dA, and dC CE phos-
phoramidites (Glen Research) were used. The DNA
incorporated with the –CF3 group was synthesized with
50-(4,40-dimethoxytrityl)-30-(cyanoethyl)phosphorami-
dite-20deoxy-5-(trifluoromethyl)uridine [10–12], while
using PAC-phosphoramidites and mild deprotection [12].
Incorporation of phosphorothioates into DNAs was
done at the oxidation step in which a sulfurizing agent
(Glen Research) is used in place of iodine solution.
DNAs were deprotected and cleaved from the resin by
using concentrated aqueous ammonia at 55 °C for
16 h. DNA were purified by HPLC (gradient and sol-
vent: 10 lmol DNA was brought up to 8 mL by
100 mM Et3NHOAc prep-HPLC: buffer A, 100 mM
Et3NHOAc; buffer B, acetonitrile, 10 mL/min, k 260,
2 ml loop: 0–13 min at 2% B, 13–25 min at 2–15% B,
25–35 min at 15–35% B, 35–48 min at 35% B, 48–
53 min at 35–2% B, and 53–62 min at 2% B).
0
0
3
Fig. 1. Compounds and molecules (1–6) with unique phosphorous and
–F or –CF3 group, used in this study where the distance between the
phosphorous atom and –F or –CF3 were measured by 31P{19F}
REDOR.
a,a,a-trifluoro-p-cresol (2a), 4-fluoro-4-hydroxybenz-
ophenone (3a), and 4-4-(trifluoromethyl)phenoxyphenol
(4a), which were purchased from Aldrich. Ethylthiotet-
razole and 2-cynanoethyl diisopropylchlorophosph-
oramidite were purchased from Glen Research and
Chem Genes, respectively. Acetonitrile and dichloro-
methane were dried over calcium hydride followed by
distillation under argon. Fig. 2 shows the general scheme
for the synthesis of compounds 1–4. In general, all reac-
tions were run under argon atmosphere unless otherwise
noted. Flash column chromatography was carried out
with EM type 60 (230–400 mesh) silica gel. Cation ex-
change chromatography was performed on MP-50 resin,
grade 100–20 mesh, hydrogen form which was pre-
treated with water, 1.00 N NaOH, then neutralized with
water until pH 7.0. Preparative RP-HPLC separations
were performed using a Dynamax C18 column (Rainin,
˚
25 cm, 4.6 mm ID, 5 lm, 300 A) and the solvent gradi-
ents for preparative separation, RP-HPLC were run at
10 mL/min as follows: Gradient A: solvent A, 100 mM
Et3NHOAc (pH 7.0); solvent B, CH3CN; 92% A for ini-
tial, 12 min linear gradient to 61% A, 8 min linear gradi-
ent to 30% A, isocratic for 3 min, 3 min linear gradient
to initial conditions. Gradient B: the same as gradient
2.3. Sample preparation for NMR studies
2.3.1. Model compounds
To dilute the spins, 5–10 mg of 1–4 was dissolved in
deionized water and diluted with solutions of KBr and/
or benzoic acid, lyophilized, and packed in a 5 mm rotor.
Fig. 2. Synthetic pathway of compounds 1–4.

14
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
Sixty microliters of H2O was added to 5 mg (3) and the li-
quid sample was then placed in a 5 mm rotor with tight
end caps.
31P magnetization by CP from the protons, the 31Pꢀs are
subjected to a series of 180° refocusing pulses. This full-
echo spectrum is designated So. In the second half of the
experiment, in addition to the 31P refocusing pulses,19F
180° pulses are added at the middle of the rotor period.
These fluorine dephasing pulses reintroduce 31P–19F
dipolar couplings averaged by magic angle spinning
(MAS), and cause a dephasing of the observed 31P signal.
The second half of the spectrum is referred to as S, and the
ratio of S/So provides a dephasing curve that can be fit to
theoretical equations [1,2] which directly yields the dipo-
lar coupling, DPF, between the 31P and 19F spins. Given
the DPF, the internuclear distance, rPF, can be easily calcu-
2.3.2. DNA samples
The DNAs were salted with 10% (w/w) NaCl and
10% (w/w) Na2EDTA, annealed at 85 °C for 10 min,
lyophilized, and packed in 5 mm ceramic rotors, and
placed in an 85% hydration chamber to yield at least
11 waters per nucleotide by weight which ensures that
the DNA is in B-form [13].
2.3.3. NMR spectroscopy
All 31P{19F} REDOR experiments were performed on
a 4.7 T (200 MHz proton Larmor frequency) magnet,
homebuilt MAS triple resonance (1H, 19F, and 31P) probe
[6,14], and homebuilt console (J. Gladden and G.P. Dro-
bny, Unpublished work). The 1H p/2 pulse was set at 5 ls
and the 31P and 19F p pulses were 10 and 7 ls, respectively.
The 1H and 31P cross-polarization (CP) time was 2 ms. 1H
decoupling was set at 100–120 kHz during evolution and
80 kHz during acquisition [15–17]. Hydrated DNA or
samples diluted in water were run at low temperature
which ranged from ꢀ16 to ꢀ27 °C with a spinning speed
of 5988 Hz. The model compounds 1–4 were spun at 4505,
5000 or 5988 Hz. The pulse sequence employed was XY8-
REDOR [18] as shown in Fig. 3. Following generation of
lated from rPF = [(cPcFh)/4p2DPF 1/3 where cP and cF are
]
the gyromagnetic ratios of the P and F spins, respectively,
h is Planckꢀs constant, and D is the dipolar coupling in
Hertz.
19F observe 31P dephase REDOR cannot be exploited
due to short T2 relaxation of the 19F nucleus[19]. Pan re-
ports that after eight-rotor cycles of dephasing the fluo-
rine signal is nearly completely dephased on
a
fluoroapatite crystal [19]. Dephasing on the 31P nucleus
would only be feasible for the small model compounds
and not the dsDNA due to the dephasing contribution
of 31P nuclei in the phosphate backbone, and thus, the
presence of multiple spin dephasers.
Goetz and Schaefer [9] presented closed form numer-
ical algorithms for REDOR dephasing by multiple spins
in the presence of molecular motion. The calculation as-
sumes that the 19F–19F homonuclear interaction does
not affect REDOR dephasing. For a trifluoromethyl
group, fast rotation about the threefold symmetry axis
of the CF3 group ensures that the three fluorine nuclei
have identical, motionally averaged chemical shifts,
and identical motionally averaged dipolar couplings to
the 31P nucleus. In principle, REDOR data will be
dependent on the angle between the axis of threefold
symmetry of the CF3 group and the vector pointing
from the 31P nuclear coordinate to the ‘‘pseudo-atom’’
center of the CF3 group. The Schaefer group provided
simulations of 31P{19F3} REDOR data in this present
work include angles of 0° and 90°, the extreme cases.
2.4. Determination of error bars in experimental data and
uncertainty in distance measurements
The experiments were repeated a minimum of 2–3
times, averaged, and the standard deviation calculated.
The error bars in the S/So REDOR dephasing plots
were determined by reproducibility of the experiment.
The uncertainty of the distances measured by 31P{19F
or 19F3} REDOR for samples 1–6 were calculated by
determining the v2 and calculating the variation of v2
near a minimum as defined in Bevington [20]. Briefly,
the minimum of v2 was determined by fitting three
points that straddle the minimum of v2 vs. distance to
Fig. 3. The XY8-REDOR pulse sequence for measuring 31P–19
F
dipolar couplings. (A) The So (full echo) experiment shows that
1
following CP from H to 31P, the phosphorus signal is refocused by a
series of p pulses following the XY8 phase cycling. (B) On alternate
scans, the S (reduced echo) experiment is collected. The 19F p pulses
are used to recouple the fluorine to the phosphorous nuclei. N is the
number of XY8 cycles.

E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
15
a quadratic equation, and the uncertainty of the distance
was measured by taking the second derivative of v2 with
respect to the distance [20].
3. Results and discussion
3.1. Orientation and theoretical dephasing curves of –CF3
group
The orientation of the fluorines from a –CF3 group
with respect to the phosphorous atom can range from
0° to 90°. Fig. 4A depicts the 0° orientation where the
31P atom lies along the C3 axis of the –CF3 group while
the 90° orientation shown in Fig. 4B is where the 31P
atom is in the plane of the three fluorines. In both orien-
tations, the distance simulated is from the center of mass
of the three fluorines to the 31P atom.
Fig. 5. 31P{19F} REDOR dephasing curves for P–F and P–CF3
simulations at 6, 9, and 12 A. Dashed lines P–CF3 90°; dotted line
P–CF3 0°; and solid line P–F. The dephasing curve is the ratio of the S
and So experiments as a function of NcTr which is the number of rotor
cycles (Nc) multiplied by the rotor period (Tr).
˚
The simulated 31P–19F REDOR dephasing curves for
P–F and P–CF3 are compared at several distances 6, 9,
˚
and 12 A in Fig. 5. For each distance, there are three
simulated curves, where the solid line represents the
dephasing for a given distance between a 31P atom and
a single 19F atom. The short dashed and dashed lines de-
scribe the P–CF3 simulations where the C3 axis of the –
CF3 group is oriented at its extreme of either 0° or 90°
˚
at 6 A, the 90° orientation dephases faster than the 0°
orientation. At short distances, the difference in distance
from the 31P atom to the center of mass of the three flu-
orines compared to the distance from the 31P atom to
any one of the fluorines becomes significant, and thus
the orientation of the –CF3 group is relevant as com-
pared to the distances, where the orientation is irrele-
vant. At longer distances (>12 A), the three fluorines
of the –CF3 group collectively can thus be considered
as a ‘‘pseudo-atom.’’
31
˚
with respect to the P atom, respectively. At 6 A, the
P–F and P–CF3 dephasing curves are similar to one
˚
another. For the 9 and 12 A dephasing curves, the P–
˚
CF3 curves show greater dephasing dephased than the
P–F curves.
˚
At distances greater than 12 A, the REDOR dephas-
ing curve shows a decreasing dependence upon the ori-
entation of the –CF3 group to the point that the
apparent dephasing is nearly the same. However, at
3.2. Model compounds
˚
Model compounds were needed to check experimental
parameters and to ensure the accuracy of the distances
measured by 31P{19F} REDOR. Several model com-
pounds were synthesized and investigated in this study.
Fig. 1 shows the model compounds (1–4) that were syn-
thesized, each contained either a unique fluorine atom
or –CF3 group and phosphorous atom. The synthetic
pathway of compounds 1–4 is shown in Fig. 2. Com-
pounds 1 and 2 were specifically designed such that the
internuclear distance between 31P and 19F does not vary
due to bond rotations. Compounds 3 and 4 were synthe-
sized to measure a longer distance than 1 and 2, however,
short distances (ꢁ6 A), the dephasing is more sensitive
to the orientation of the CF3 group. Fig. 5 shows that
˚
the distance for both 3 and 4 have an approximate 1.5 A
range due to possible rotation of the C–O bond attached
to the phosphate group. The choice of model compounds
was based on structural similarity such that the dephasing
profile of a compound with one fluorine could be com-
pared to a similar compound where a –CF3 group replaces
a single –F spin.
Fig. 4. The orientation of the –CF3 group with respect to the
phosphorous atom. (A) The P–CF3 0° orientation is defined as the
phosphorous atom being along the C3 axis of the –CF3 group. (B)
The P–CF3 90° orientation is defined as the phosphorous atom
being in the plane of the three fluorines.
Two separate sources were used to determine the the-
oretical 31P to 19F distances of the model compounds in

16
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
the study. The first method involves searching the Cam-
bridge Structure Database (CSD) for compounds or
fragments of the compounds similar to 1–4. The distanc-
es based on CSD data for structurally similar fragments
were compared with the measured distances of the mod-
el compounds (1–4). Table 1 shows the structures of the
Table 1
Comparison of distances measured of 1–6 by X-ray crystal structures and 31P{19F} REDOR
a
˚
˚
Sample from
Fig. 1
Measured distance rP–F (A)
X-ray crystal structure, code, and distance rP–F (A)
1
6.0 0.9 (1:100 mol/mol KBr)
6.2 1.0 (1:5 mol/mol benzoic acid)
6.0 1.0 (1:20 mol/mol benzoic acid)
HOGDOWc
6.66
2
3
6.0 0.8 (1:100 mol/mol KBr)
FIPXILd
7.08
9.0 0.9 (undiluted)
9.5 1.1 (10 mg/30 lL)
9.3 1.8 (10 mg/60 lL)
WADBEIe
10.31b
4
5
8.3 3.7 (undiluted)
9.3 2.7 (1.40 mol/mol benzoic acid)
IBOMUHf
11.14b
12 1(W = 14)
1BDNg
11.89
6
13 2(W = 14)
1BDNg
11.16
The distance between 31P–19F/19F3 of compounds 1–4 and molecules 5 and 6 were measured using 31P–19F REDOR without and with dilution.
Fragments of crystal structures found in the Cambridge Structural Database (CSD) to be similar to compounds 1–4 are shown with their corre-
sponding CSD code. The crystal structure for molecules 5 and 6 was found in the Protein Data Bank (PDB). The dots in the structures represent the
placement of the 31P and 19F or center of mass of the CF3 group. The distance from the crystal structures were tabulated. The DNA samples were
hydrated and in B-form. W represents the hydration amount and is equivalent to the number of water molecules per nucleotide.
a
Distance measured using 31P{19F} REDOR.
Distance was averaged due to phosphorous atom could be in either position.
Ref. [29]
Ref. [30].
Ref. [31].
Ref. [32].
Ref. [25].
b
c
d
e
f
g

E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
17
crystal fragment with the CSD codes, as well as the dis-
tance measured between the phosphorous and fluorine
or –CF3 group labels, which are identified as dots on
the structure. The REDOR results are discussed below
whereby the change in dephasing between the single
–F and –CF3 analogues are compared. The second
method was a structural minimization using a semi-em-
pirical AM1 closed shell calculation with software pro-
vided by Cambridge Soft Corporation. The calculated
measurements were used to approximate intermolecular
couplings in the model compounds because the frag-
ments of the crystal structures found were not adequate
in determination of the next nearest neighbor.
REDOR So (full echo) spectra with eight rotor cycles
of evolution are shown in Figs. 7A and B for compound
1 undiluted and diluted in benzoic acid (1:40 mol/mol),
respectively. The REDOR dephasing curves for the
model compounds (1–4) are shown in Figs. 8A and B.
In Fig. 8A, the experimental data are shown for com-
pounds 1 and 2 (both diluted in KBr 1:100), along with
˚
the 6 A calculated dephasing curves for P–F and P–CF3
simulations at 0° and 90°. The X-ray crystal structure
˚
distances for compounds 1 and 2 are 6.66 and 7.08 A,
respectively. For compound 1, the data points follow
˚
the 6 A P–F dephasing curve. The experimental data
for 2 fits the calculated dephasing curve of P–CF3 at
90° up to about 4 ms, and then at longer dephasing
˚
3.3. Dephasing differences between –F and –CF3
times, the data points no longer follow the 6 A CF3
90° dephasing curve and begin to dephase more slowly.
In a 31P–19F spin system, REDOR recouples the di-
rect dipolar coupling between the two nuclei. This cou-
pling is quantified as a dephasing of spin coherences of
the observed nucleus (e.g., 31P) caused by the coupled
spin (e.g., 19F). The amount of dephasing observed is
directly related to the internuclear distance of the two
spins by the heteronuclear dipolar coupling. Fig. 6, for
example, shows a plot of P–F and P–CF3 distances for
a range of coupling constants. At a dipolar coupling
of 3 Hz, the difference in distance measurable by P–F
˚
and P–CF3 methods can be as great as 5 A. Due to lim-
itations of sample size or rotor size as in solid-state
NMR experiments, evolution times of about 20 ms are
needed for complete dephasing of a P–F sample at a
˚
modest distance of 9 A. By replacing a single fluorine
spin with three fluorine spins of a trifluoromethyl group,
the dephasing is increased, and thus a longer internucle-
ar distance can be measured at a shorter evolution time.
Fig. 6. Comparison of P–F and P–CF3 distances for a range of
coupling constants. As the coupling constant decreases, the difference
in distance measurable between the P–F (solid triangles) and P–CF3
(solid squares) increases to a maximum of 5 A. Solid diamond shows
difference in distance between P–F and P–CF3.
Fig. 7. (A) REDOR So (full echo) spectrum with 8 rotor cycles of
evolution for compound 4 undiluted spinning at 4 kHz, 2048 scans. (B)
REDOR So (full echo) spectrum with 8 rotor cycles of evolution for
compound 4 spinning at 5 kHz diluted 1:40 mol/mol benzoic acid.
˚

18
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
Fig. 8. Overlay of 31P{19F} REDOR dephasing curves of samples with
–F and –CF3 and simulations. (A) Experimental data for molecules 1
(filled circles) and 2 (filled squares). Both samples were diluted with
KBr (1:100). The lines are simulated REDOR curves where the solid
Fig. 9. (A) Experimental data for compound 3 undiluted (solid
squares) and diluted (solid-circles) in distilled water at 10 mg/60 lL.
The lines are simulated REDOR curves for P–F at 7, 8, and 9 A. (B)
˚
˚
˚
line is P–F at 6 A, short dashed lines are P–CF3 at 6 A 0°, long dashed
Experimental data for compound 4 undiluted (solid triangles) and
˚
lines are P–CF3 at 6 A 90°. (B) Experimental data for molecules 3
diluted (solid-diamonds) in 1:40 mol/mol benzoic acid. The lines are
˚
(filled circles) diluted 1:40 in benzoic acid and 4 (filled squares) diluted
10 mg/60 lL in H2O. The lines are simulated REDOR curves, solid
simulated REDOR curves for P–CF3 90° at 8, 9, and 10A.
˚
˚
line is P–F at 9 A, short dashed lines are P–CF3 at 9A 90°, and long
˚
ter-label distances were measured, there was a larger dis-
parity between the crystal structure fragment
measurement and to the 31P{19F} REDOR distances.
In addition, the experimental uncertainty for molecules
dashed lines are P–CF3 at 8 A 90°.
In addition, 1 was also diluted in benzoic acid at ratios
of 1:5 and 1:20 mol/mol. Regardless of the dilution
method, the average distances measured for 1 and 2
were shorter than the X-ray crystal distance as shown
in Table 1. The crystal structure distance was within
the experimental uncertainty of compound 1 but not
within the uncertainty for compound 2.
˚
3 and 4 in some cases was greater than 1 A, and proba-
bly due to heterogeneity in conformation of the sample
from rotation of the phosphate group and perhaps the
multiple spin couplings from intermolecular effects.
The rotation of the C–O bond of the phosphate group
˚
for 3 and 4 can possibly contribute a 1.2–1.5 A range
Compounds 3 and 4 were diluted in deionized water
(10 mg/30 lL and 10 mg/60 lL) and 1:40 mol/mol ben-
zoic acid, respectively. Fig. 8B shows the dephasing
curve for 3 diluted in deionized water at a concentration
of 10 mg/60 lL and 4 diluted with 1:40 mol/mol benzoic
in the observed distance.
McDermott and co-workers [7] have used model
fluorinated compounds to measure 19F–19F internucle-
ar distances via RFDR. Their solution for avoiding
intermolecular dipolar couplings from neighboring
fluorinated compounds was the dilution of fluorinated
compounds by co-crystallization with nonfluorinated
analogues that were structurally similar. In our study,
various dilution methods were attempted. For model
compound 1, the dilution methods used were diluting
˚
˚
acid. The experimental data for 3 best fits the 9 A P–F
dephasing curve and 4 fits between the 8 and 9 A 90°
P–CF3 dephasing curves. Similar to compounds1 and
2, 3 and 4 also show a shorter measured distance of
˚
about 9.3 A compared to the crystal structure fragment
˚
distance of 10.31 and 11.14 A, respectively. As longer in-

E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
19
1:100 mol/mol in KBr, 1:5 and 1:20 mol/mol in benzo-
ic acid. None of the dilution methods showed a signif-
icant change in distance measured as reported in
Table 1. Thus, for compound 2, the only dilution
method used was 1:100 mol/mol KBr. Another dilu-
tion method was tried for compound 3. In Fig. 9A,
the REDOR dephasing curves are shown for com-
pound 3 undiluted and diluted in a solution of deionized
water at a concentration of 10 mg/60 lL. Comparison
of the undiluted case and the diluted cases of compound
For compound 2, the distances obtained from a semi-
empirical AM1 closed shell calculation are 7.26, 7.36,
˚
and 7.64 A for the three fluorine atoms, respectively,
from the phosphorus atom. The simulated result is co-
plotted with compound 1 in Fig. 10B. The best fit simu-
˚
lation considered the center of the CF3 to be 5.34 A as
˚
opposed to the expected 7.42 A distance from the
AM1 calculation. This discrepancy is mostly due to
aggregation leading to spin diffusion. Most recent work
shows that in cases where multiple dephasers are in-
volved simulation of data requires the exact molecular
geometry [28]. Despite the experimental distances being
shorter than the calculated distances, the dephasing of
compound 2 was greater than compound 1, as expected
due to the –CF3 group in compound 2.
˚
3 demonstrates a distance change from 9.0 to 9.3 A,
respectively. The dephasing curves in Fig. 9A show that
there is a visible dephasing difference and that dilution
of the compound in deionized water has some effect in
isolating the phosphorous and fluorine spins. Likewise,
Fig. 9B shows the dephasing curves of compound 4
undiluted and diluted in 1:40 mol/mol benzoic acids,
and also illustrates that dilution of 4 in 1:40 benzoic acid
isolates the phosphorous and fluorine spins to some
extent.
Attempts were made to dilute the samples further
such that intermolecular couplings might be reduced.
However, as dilution of sample increased, the signal-
to-noise diminished and experimental points were more
scattered at long dephasing times. It has been noted that
small fluorinated compounds tend to aggregate and do
not make good model compounds (M.E. Merritt per-
sonal communication) [14]. McDermott et al. [7] also
encountered inhomogeneous or inadequate dilution of
samples with their fluorinated model samples. In addi-
tion, halogenated organic crystals are also known to
aggregate [21,22]. Thus, it is suspected that these appar-
ent intermolecular couplings may be due to aggregation
of the fluorines and use of an insufficient molecular mix-
ing method.
The AM1 minimized structure for compound 3 sug-
gested that the P–F distance could range between 9.8
˚
and 11.5 A depending upon the torsion angle of the phos-
phate group. The 31P{19F} REDOR data of the undiluted
sample of 3 could not be fit to a single distance. For com-
pound 3, when diluted to 10 mg/30 ll, the REDOR
˚
dephasing could be fit to a single distance of 9.2 A. This
shorter distance obtained from the REDOR experiment
is probably due to the presence of intermolecular
31P–19F couplings. For example, the curve could be fit
to a linear combination of two REDOR curves, one due
˚
to a single spin at 9.8 A distance contributing 67% and
˚
another due to two spins at distances 9.8 and 9.0 A,
˚
respectively, with the 9.0 A being the distance between
the phosphorus and a fluorine from another molecule.
These results are shown in Fig. 10C.
˚
For 4, the expected distance of 10.9 A from the AM1
structure minimization correlated with three P–F cou-
plings of 34, 39, and 49 Hz. The experimental data for
4 diluted with benzoic acid (1:40) was best fit to three
P–F couplings of 89, 93, and 122 Hz. Again, in this case
the fitted distances were much shorter than the expected
distance as determined from the AM1 calculation.
3.4. Multiple spin simulations and fits
To address the issue of the experimental distances
being shorter than the calculated distances for the com-
pounds 1–4, simulations using SIMPSON [23] were per-
formed to describe possible packing models of the
compounds where the intermolecular effects give rise
to a distribution of distances and thus, multi-spin effects.
For compound 1, the energy minimized structure ob-
tained from the AM1 calculation was put in a fictitious
3.5. Fluorinated and trifluoromethylated DNA samples
The sequences and labeling scheme of the fluorinated
DNA (5 and 6) used in this study are shown in Fig. 1.
The structure of the DNA sequence was selected because
it was not self-complementary and the X-ray crystal
structure was published in the Protein Data Bank
(PDB) [24] as code 1BDN [25]. A self-complementary se-
quence would complicate the analysis due to inter- and
intrastrand interactions. The DNA were synthetically
modified to accommodate a fluorine or –CF3 group
and phosphorothioate for the 31P{19F} REDOR
dephasing comparison studies. Thermodynamic data
for DNA duplexes with and without a trifluoromethy-
lated label were compared and previously reported by
Markley et al. [12]. Optical melting experiments were
performed and it was found that DNA duplex stability
˚
cubic crystal cell (Cell constant 10 A) in the Weblab
Viewerlite Visualization program by Accelerys. This cal-
culation showed the possibility of an intermolecular
31 19
˚
P– F distance of 3.5 A that can be seen in Fig. 10A.
Assuming partial aggregation causing intermolecular ef-
fects, the experimental data were fit to a linear combina-
tion of two spin and three spin simulations. The best fit
19
occurred at a 90% of a 31P– F distance of 6 A and 10%
˚
31 19
˚
of P– F distances of 6 and 3.5 A as shown in
Fig. 10B.

20
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
Fig. 10. Multi-spin model, simulations, and dephasing curves. (A) Model for multi-spin simulation of 1 where the 19F atom is coupled to 31P at 6.0
˚
and 3.5 A. (B) Experimental data for 1 (solid square) diluted with KBr (1:100). The dashed line represents the best fit to a linear combination of two
˚
˚
spin (90% 6 A) and three spin simulations (10% of 6 and 3.5 A). Experimental data for 2 (solid triangle) diluted with KBr (1:100). The best fit
˚
simulation considered a –CF3 at a distance of 5.34 A. (C) Experimental data for 3 (solid circle) diluted 10 mg/30 lL in H2O. The curve was fit to a
linear combination of two REDOR effects, one due to a single spin at 9.8 A distance contributing 67% and another due to two spins at distances 9.8
˚
˚
and 9.0 A, respectively, with the 9.0 A being the distance between the phosphorus and a fluorine from another molecule. Experimental data for 4
(solid diamond) diluted with benzoic acid (1:40). The dashed line represents the best fit to a four spin simulation (89, 93, and 122 Hz).
˚
at pH 6.9 was lowered slightly (a Tm difference of 3 °C)
due to the incorporation of the trifluoromethyl group
[12].
experimental points were between the 10 and 14 A P–
CF3 90° simulations as shown in Fig. 12B. The v2 is
shown in the inset of Fig. 12B, and the minimum v2
and standard deviation was determined to be
Based on the crystal structure data, the phosphoro-
thioate and –F or –CF3 labels in the duplex DNA were
˚
13 2 A, which is also in agreement with the X-ray
designed to be 12 A apart. Fig. 11 is the 31P{19F} RE-
˚
crystal structure. In addition, by co-plotting the experi-
mental data points of 5 and 6, as in Fig. 12C, the
difference in dephasing behavior between the single –F
and –CF3 dsDNA is demonstrated, especially at 20 ms.
The intermolecular coupling effects were not ob-
served in the dsDNA samples, likely due to the oligo-
mers self-diluting. The ratio of distance measured to
length of the molecule is smaller than the model com-
pounds 1–4. The P–F and P–CF3 labels were strategi-
cally placed so that the distance of the labels would
be shorter than the length of the duplex 12 mer
DNA. This may have contributed to a more accurate
distance measurement by preventing intermolecular
couplings and better isolating the P–F and P–CF3 spin
pairs. The distance of the P–F or P–CF3 labels in 1–4
were essentially the length of the molecule, and there-
fore contributed to intermolecular effects creating an
apparent faster dephasing.
DOR So spectrum of dsDNA 5 showing the phosphoro-
thioate peak chemically shifted upfield 55 ppm from the
phosphate backbone. The sample was spun at 5988 Hz
in order to allow for adequate resolution between the
spinning sideband peak of the backbone phosphates
and the phosphorothioate peak. For the dsDNA sam-
ples, 5 and 6, the dephasing curves are shown in Figs.
12A and B, respectively. The data points for the dsDNA
˚
sample, 5, fall between the 11 and 13 A P–F dephasing
simulations. A v2 plot vs. distance is shown in the inset
of Fig. 12A where the equation of a parabola was fit to
pass through three points that straddle the minimum.
From the parabolic fit, the v2 minimum and standard
deviation can be determined [20]. The best fit for 5
˚
was 12 1 A, which is in good agreement with the crys-
˚
tal structure distance of 11.89 A. In the case of the
dsDNA with the –CF3 group, 6, the 31P{19F} REDOR

E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
21
Fig. 11. REDOR So (full echo) spectrum with 8 rotor cycles of evolution for 50d(CGCATT-(pS)-TT5-FUGCG)Æd(CGCAAAAATGCG) where 5-F
U
denotes 20deoxy-5-fluorouridine and (pS) denotes a phosphorothioate (SPO3ꢀ) in place of a phosphate(OPO3ꢀ) group. Magic-angle spinning was
performed at 5988 2 Hz. Three sets of 6144 scans were acquired. The spinning sidebands are labeled ssb. Inset shows phosphorothioate peak and
spinning sideband.
3.6. Advantages and disadvantages of –CF3
porating a –CF3 group when the molecule of interest is
inadequately diluted and/or the distance measured is
large compared to the size of the molecule due to
intermolecular effect. Thus, the –F label is best used
when measuring short distances in order to avoid
intermolecular effects. Despite the inadequate molecu-
lar mixing causing aggregation or poor dilution of
the spins, for compounds 1–4, it was shown that
replacing a –F with a –CF3 group allowed the dephas-
ing behavior to increase. This trend of increased
dephasing was also seen with the dsDNA samples of
5 and 6. A greater apparent dephasing due to the
CF3 substitution will allow for longer distance mea-
surement especially when working with biological sam-
ples in lmol scale quantities.
In deciding when to include a –CF3 group or –F in a
sample, the choice depends on the information desired,
the system of interest, and placement of the label.
Besides the ability to measure a long-range distance with
a –CF3 group, there is an advantage of a –CF3 group at
short distances, as well. For distances not more than
˚
ꢁ6 A, dephasing is sensitive to the orientation of the
–CF3 group relative to the phosphorous nucleus, if the
error in the dephasing data points is low. Fig. 8A illus-
˚
trates that the 6 A 0° and 90° dephasing curves are suf-
ficiently distinguishable to allow determination of the
orientation of the –CF3 group relative to the phospho-
rous nucleus. The first three experimental points of the
dephasing curve of 2, indicated by solid squares, are in
˚
good agreement with the theoretical P–F3 6 A 90°
3.7. Limit of P–CF3 coupling
dephasing curve (long dashes). Due to the C–O–P
arrangement as in molecule 2, the –CF3 group is situated
90° relative to the phosphate atom, which is in agree-
ment with the experimental result. As a result of inter-
molecular couplings, at longer dephasing times
(greater than 5 ms), the experimental results no longer
Table 2 is a theoretical comparison of the amount of
dephasing between P–F and P–F3 and shows that a dis-
tance of 20 A has 8% dephasing after 30 ms [26,27] evo-
˚
lution for a P–F3 system, which is an estimate of the
upper limit of the distance measurable using a –CF3
group in place of a –F. The Schaefer group [3] has mea-
˚
follow the 6 A 90° dephasing curve. At distances greater
˚
˚
than 6 A, the theoretical P–F3 0° and 90° dephasing
sured a maximum distance of 16 A (12 Hz) using P–F
curves begin to converge and the orientation of the
–CF3 group can no longer be distinguished. Thus, as
seen in the results, there can be a disadvantage to incor-
REDOR. According to Table 2, P–CF3 coupling of
˚
12 Hz correlates to a 19 A. This implies that it is possible
˚
to measure a 19 A distance with P–CF3 REDOR.

22
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
Table 2
A comparison of the percent dephasing of P–F and P–CF3 at 30 ms
evolution
˚
Distance (A)
Percent dephasing
P–F
P–CF3
6
9
100
100
73
50
15
100
100
80
60
35
26
15
8
11
13
15
16
18
20
23
27
30
5.5
3
1.2
0.5
03
4
1.5
0.8
3.8. Effective gamma for ‘‘Super’’ fluorine in –CF3 group
The same dipolar coupling for P–CF3 and P–F yield
different distances due to their difference in dephasing
behaviors. Due to the proportional relationship of the
gyromagnetic ratios and the dipolar coupling, an average
effective gyromagnetic ratio for the fluorines in the –CF3
group was estimated using cP–CF ¼ cP–F½ðrP–CF Þ3=
3
3
ðrP–FÞ ꢂ. The 19F gyromagnetic ratio is 25,182 1/(G s)
3
and for the same dipolar coupling, a ratio of the distances
3
3
ðrP–CF Þ =ðrP–F
Þ
can be determined. The effective
3
gyromagnetic ratio for the fluorines in a –CF3 group for
several dipolar couplings was averaged to be 42,556
2000 1/(G s). Thus, the gyromagnetic ratios of the CF3
to F is 1.69:1 which is about a 39% increase.
4. Conclusion
31P{19F} and 31P{19F3} REDOR experiments were
performed on small model compounds and nucleic
acid oligomers. With replacement of a –CF3 group
for –F, the behavior of 31P–19F dephasing was studied.
The experimental distances ascertained by 31P{19F}
REDOR for the model compounds, 1–4, ranged from
half to one Angstrom less than the distance calculated
from the minimized structure. There was inadequate
dilution of the spins and the hindrance was most
likely due to the nature of the size of the compound
in relation to the distance being measured causing
intermolecular effects as well as possible aggregation
of the fluorine atoms. The moderate distances mea-
sured in the double-stranded 12-mer DNA molecules
were self-diluting and thus, the experimental distances
were in good agreement with the X-ray crystal
structure.
Fig. 12. 31P{19F} REDOR dephasing curves for dsDNA. (A) Dots
are experimental data for dsDNA with single fluorine (5). The lines
are simulated 31P–19F dephasing curves: dots 11 A (34 Hz), solid line
˚
2
˚
˚
12 A (26 Hz), and dashed lines 13 A (21 Hz). Inset is of v vs.
˚
distance. (B) short dashed lines are P–CF3 at 12 A 0°, long dashed
2
˚
lines are P–CF3 at 12 A 90°. Inset is of v vs. distance. (C) Co-plot of
31P{19F} REDOR dephasing curves for dsDNA 5 (solid circles) and
6 (solid squares) where the DNA with the –CF3 (6) incorporated
shows more dephasing than the DNA with a single fluorine (5)
especially the data points at 20 ms shows the increase in dephasing
due to the CF3 group. CF3 dephased 60%, CF dephased 35% at
20 ms.
Regardless of the intermolecular effects, in all sets of
molecules, samples with a –CF3 group consistently
exhibited more dephasing than the –F sample counter-
part. This increase in dephasing indicates that a –CF3

E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
23
[8] L.M. McDowell, D.R. Studelska, B. Poliks, R.D. OꢀConnor, J.
Schaefer, Characterization of the complex of a trifluoromethyl-
substituted shikimate-based bisubstrate inhibitor and 5-eno-
lpyruvylshikimate-3-phosphate synthase by REDOR NMR,
Biochemistry 43 (2004) 6606–6611.
group is a more sensitive measure than a single fluorine,
and more accurate distance measurements can be as-
sessed at a shorter dephasing time for distances greater
˚
than 12 A, which can be important when studying large
biomolecules or biological complexes.
[9] J.M. Goetz, J. Schaefer, REDOR dephasing by multiple spins in the
presence of molecular motion, J. Magn. Reson. 127 (1997) 147–
154.
[10] W.H. Gmeiner, R.T. Pon, J.W. Lown, Synthesis, annealing
properties, F-19 NMR characterization, and detection limits of
a trifluorothymidine-labeled antisense oligodeoxyribonucleotide
21-Mer, J. Org. Chem. 56 (1991) 3602–3608.
In summary, long-distance constraints of at most
20 A are theoretically measurable with the 31P{19F} RE-
˚
DOR technique in nucleic acids by incorporation of a
–CF3 group and slight modification of the phosphate
backbone with one phosphorothioate.
[11] P.D. Cook, Y.S. Sanghvi. US Patent 5,614,617, 1992; Chem.
Abstr. 1992, 117, 111996., 1992.
[12] J.C. Markley, P. Chirakul, D. Sologub, S.T. Sigurdsson, Incorpo-
ration of 20-deoxy-5-(trifluoromethyl)uridine and 5-cyano-20-deox-
yuridine into DNA, Bioorg. Med. Chem. Lett. 11 (2001) 2453–
2455.
[13] T.M. Alam, G.P. Drobny, Solid-state Nmr-studies of Dna-
structure and dynamics, Chem. Rev. 91 (1991) 1545–1590.
[14] J.A. Stringer, The Design and Analysis of Solid State Nuclear
Magnetic Resonance Probes for the Determination of Biomolec-
ular Structure, University of Washington, 1998.
[15] J.R. Garbow, T. Gullion, Improvements in Redor NMR-spectros-
copy—minimizing resonance-offset effects, J. Magn. Reson. 95
(1991) 442–445.
[16] A.K. Mehta, D.J. Hirsh, N. Oyler, G.P. Drobny, J. Schaefer,
Carbon–proton dipolar decoupling in REDOR, J. Magn. Reson.
145 (2000) 156–158.
Acknowledgments
We thank M.E. Merritt for helpful discussions; J.
Schaefer and his group members for P–CF3 REDOR
simulations, helpful discussions, and comments; T.E.
Edwards, J. D. Wellhausen, J. C. Markley, and D. Solo-
gub for synthesizing and purifying the DNA; N.A. Oyler
for the use of his Matlab code for the NMR data work-
up and the C++ code for the P–F REDOR simulations;
and James Gibson for reading repeated revisions of the
manuscript. This research was supported by the Nation-
al Institute of Health. Grants (RO1 EB03152-O5) and
(RO1 DE-12554-08) and by NSF Grant (DMR
0110505).
[17] F.G. Vogt, J.M. Gibson, S.M. Mattingly, K.T. Mueller, Deter-
mination of molecular geometry in solid-state NMR: rotational-
echo double resonance of three-spin systems, J. Phys. Chem. B 107
(2003) 1272–1283.
[18] T. Gullion, D.B. Baker, M.S. Conradi, New compensated Carr–
Purcell sequences, J. Magn. Reson. 89 (1990) 479–484.
[19] Y. Pan, P-31-F-19 rotational-echo, double-resonance nuclear-
magnetic-resonance experiment on fluoridated hydroxyapatite,
Solid State Nucl. Magn. Reson. 5 (1995) 263–268.
Appendix A. Supplementary data
Supplementary data associated with this article can
be found, in the online version, at doi:10.1016/
j.jmr.2005.06.018.
[20] P.R. Bevington, D.K. Robinson, Data Reduction and Error
Analysis for the Physical Sciences, (1992).
[21] O.V. Grineva, P.M. Zorkii, Aggregation of halogen atoms in
haloorganic crystals with low halogen contents, Russ. J. Phys.
Chem. 74 (2000) 1758–1764.
[22] O.V. Grineva, P.M. Zorkii, Aggregation of halogen atoms in
crystalline isomers, J. Struct. Chem. 43 (2002) 995–1005.
[23] M. Bak, J.T. Rasmussen, N.C. Nielsen, SIMPSON: a general
simulation program for solid-state NMR spectroscopy, J. Magn.
Reson. 147 (2000) 296–330.
[24] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H.
Weissig, I.N. Shindyalov, P.E. Bourne, The Protein Data Bank,
Nucleic Acids Res. 28 (2000) 235–242.
[25] A.D. Digabriele, M.R. Sanderson, T.A. Steitz, Crystal-lattice
packing is important in determining the bend of a Dna dodecamer
containing an adenine tract, Proc. Natl. Acad. Sci. USA 86 (1989)
1816–1820.
References
[1] T. Gullion, J. Schaefer, Rotational-echo double-resonance NMR,
J. Magn. Reson. 81 (1989) 196–200.
[2] T. Gullion, J. Schaefer, Detection of weak heteronuclear dipolar
coupling by rotational-echo double-resonance nuclear magnetic
resonance, Adv. Magn. Reson. 13 (1989) 57–83.
[3] D.R. Studelska, C.A. Klug, D.D. Beusen, L.M. McDowell, J.
Schaefer, Long-range distance measurements of protein binding
sites by rotational-echo double-resonance NMR, J. Am. Chem.
Soc. 118 (1996) 5476–5477.
[4] M.E. Merritt, S.T. Sigurdsson, G.P. Drobny, Long-range distance
measurements to the phosphodiester backbone of solid nucleic
acids using P-31-F-19 REDOR NMR, J. Am. Chem. Soc. 121
(1999) 6070–6071.
[26] A.K. Mehta, J. Schaefer, Rotational-echo double resonance of
uniformly labeled C-13 clusters, J. Magn. Reson. 163 (2003) 188–
191.
[5] G.L. Olsen, E.A. Louie, G.P. Drobny, S.T. Sigurdsson, Determina-
tionof DNA minorgroovewidthin distamycin–DNA complexesby
solid-state NMR, Nucleic Acids Res. 31 (2003) 5084–5089.
[6] J.A. Stringer, G.P. Drobny, Methods for the analysis and design
of a solid state nuclear magnetic resonance probe, Rev. Sci.
Instrum. 69 (1998) 3384–3391.
[7] M.L. Gilchrist, K. Monde, Y. Tomita, T. Iwashita, K. Nakanishi,
A.E. McDermott, Measurement of interfluorine distances in
solids, J. Magn. Reson. 152 (2001) 1–6.
[27] L.M. McDowell, A. Schmidt, E.R. Cohen, D.R. Studelska, J.
Schaefer, Structural constraints on the ternary complex of 5-
enolpyruvylshikimate-3-phosphate synthase from rotational-echo
double-resonance NMR, J. Mol. Biol. 256 (1996) 160–171.
[28] G. Goobes, V. Raghunathan, E.A. Louie, J.M. Gibson, G.L.
Olsen, G.P. Drobny, REDOR study of diammonium hydrogen
phosphate: A model for distance measurements from adsorbed
molecules to surfaces. Solid state Nucl. Magn. Reson. (2005) in
press.

24
E.A. Louie et al. / Journal of Magnetic Resonance 178 (2006) 11–24
[29] P. Sood, A. Chandrasekharan, R.O. Day, R.R. Holmes, Struc-
tural displacement of phosphites, phosphates, and pentaoxyphos-
phoranes to higher coordinate geometries by sulfur and oxygen
donor action, Inorg. Chem. 37 (1998) 6329.
[30] C.H.L. Kennard, O. Smith, T. Hari, The crystal structure of
acifluorfen [5-[2-chloro-4-(trifluoromethyl)phenoxy]-2-nitrobenzo-
ic acid], Aust. J. Chem. 40 (1987) 1131.
[31] S. Guo, G. Su, F. Pan, Y. He, The crystal structure of acifluorfen
[5-[2-chloro-4-(trifluoromethyl)phenoxy]-2-nitrobenzoic acid], Z.
Kristallogr. 202 (1992) 296.
[32] T.M. Klapotke, B. Krumm, P. Meyer, H. Piotrowski, Synthesis
and chemitry of CF3C6F4OC6F4NH2 (RFNH2) and pseudohalo-
gen derivatives; Crystal structures of RFX (X = COOH, COCl,
CN, N3, NCS, NHCOCONHRF), Z. Anorg. Allg. 627 (2001) 1983.