 1996 Oxford University Press
2868–2876 Nucleic Acids Research, 1996, Vol. 24, No. 15
Phosphorus 31 solid state NMR characterization of
oligonucleotides covalently bound to a solid support
P. M. Macdonald*, M. J. Damha1, K. Ganeshan, R. Braich1 and S. V. Zabarylo
Department of Chemistry and Erindale College, University of Toronto, Toronto, Ontario M5S 1A2, Canada and
1Department of Chemistry, McGill University, Montreal, Quebec H3A 2K6, Canada
Received June 3, 1996; Accepted June 22, 1996
ABSTRACT
31P
cross polarization (CP) magic angle spinning
(MAS) nuclear magnetic resonance (NMR) spectra
were acquired for various linear and branched di- and
tri-nucleotides attached to a controlled pore glass (CPG)
solid support. The technique readily distinguishes the
oxidation state of the phosphorus atom (phosphate
versus phosphite), the presence or absence of a
protecting group attached directly to phosphorus
(cyanoethyl), and other large changes in the phosphorus chemistry (phosphate versus phosphorothioate). However, differences in configurational
details remote from the phosphorus atom, such as the
attachment position of the ribose sugar (2′5′ versus
3′5′), or the particulars of the nucleotide bases (adenine
versus uridine versus thymine), could not be resolved.
When different stages of the oligonucleotide synthetic
cycle were examined, 31P CPMAS NMR revealed that
the cyanoethyl protecting group is removed during the
course of chain assembly.
INTRODUCTION
Solid-phase synthesis has become the method of choice for
producing oligonucleotides of defined sequence (1,2). The most
popular solid-phase supports consist of long-chain alkylaminemodified controlled-pore glass (CPG) (3,4) and highly crossedlinked polystyrene (5). In the case of CPG, the long-chain alkyl
spacer renders the terminal nucleoside highly accessible to
coupling reagents while the pore size favours the synthesis of very
long DNA and RNA oligomers (6). A typical synthesis involves
successive rounds of coupling of an activated protected nucleoside
precursor to the terminal pentose of the growing oligonucleotide,
followed by capping of non-coupled sites and oxidation of phosphite
triester linkages in preparation for the next round of coupling (1).
When the nascent oligonucleotide has been completely assembled,
it is deprotected and cleaved from the solid support.
It is undoubtedly true that such synthetic strategies as described
above have been enormously successful, and that their commercialization has placed oligonucleotide synthesis within the reach
of the non-specialist. Multiple gram solid-phase synthesis of
DNA and RNA and analogs for therapeutic-based investigations
(7), and for X-ray crystallography, NMR and other physical
studies, has recently become a reality (8,9). It is equally true that,
* To
whom correspondence should be addressed
because the chemistry involved in the individual steps is difficult
to study in situ, the ultimate success for any particular oligonucleotide sequence can only be determined upon its release from
the solid support. Perhaps the most common means for monitoring
synthesis efficiency is spectroscopic quantification of trityl
cations released during chain elongation. The major limitation of
this technique is that, other than 5′-detritylation, it does not
provide any information about structural/chemical modifications
occurring at other positions in the oligonucleotide chain.
Solid state NMR spectroscopy affords the capability of
characterizing chemical structure, conformation and dynamics in
solid-phase supported syntheses. It has been used, for instance, to
analyze surface modifications of silica particles in a variety of
situations (10,11). In this report we describe the first cross-polarization (CP) magic angle spinning (MAS) phosphorus-31 (31P)
NMR investigations of oligonucleotides covalently bound to a
solid support. 31P was chosen as the nucleus of interest because,
first, it is 100% naturally abundant and has good sensitivity in the
NMR experiment, so that no isotopic enrichment is necessary.
Second, since many of the chemical manipulations performed
during the course of a typical solid phase oligonucleotide
synthesis result in changes directly affecting the phosphorus
atom, the 31P NMR spectrum should provide a sensitive monitor
of the course of the synthesis. In this study we explore the
capabilities and limitations of 31P CPMAS NMR for reporting on
the detailed chemistry of the phosphorus atom in oligonucleotides
bound to controlled-pore glass, and for resolving different
oligonucleotide structural motifs.
EXPERIMENTAL
Synthetic procedures
Synthesis of ApU 1a and 1b. Dinucleoside monophosphate ApU 1a,
shown in Figure 1A, was synthesized by the phosphodichloridite
procedure (12,13). To a cold solution (0C) of 2,4,6-collidine
(3.63 mmol, 490 µl) in dry THF (0.8 ml) was added 2-cyanoethylphosphodichloridite (CNEOPCl2, 0.55 mmol, 68 µl). A THF
solution of N6-benzoyl-5′-O-monomethoxytrityl-2′-O-t-butyldimethylsilyladenosine (0.50 mmol, 379 mg, 1.2 ml THF) was
added dropwise over a period of 15 min. After 30 min of stirring
at room temperature, a THF solution of 2′,3′-O-di-t-butyldimethylsilyluridine (0.40 mmol, 189 mg, 1.5 ml THF) was added and the
reaction mixture stirred for 1 h at room temperature. Finally, a
solution of iodine (0.1 M) in THF/water (3:1.5; without any
2869
Nucleic Acids
Acids Research,
Research,1994,
1996,Vol.
Vol.22,
24,No.
No.115
Nucleic
2869
pyridine) was added to oxidize the phosphite triester intermediate.
After 5 min, the reaction mixture was dissolved in chloroform
(50 ml) and washed with 5% sodium bisulfite aqueous solution
(5 ml). The organic phase was washed with water (3 × 50 ml) to
remove collidine, dried (anhydrous sodium sulfate), filtered and
concentrated. The crude product was purified on a silica gel
column by elution with ethyl acetate/chloroform, gradient 25–50%
ethyl acetate. The pure product ApU 1a was obtained in 84% yield
(454 mg). ApU 1a was stirred with triethylamine/acetonitrile
(4:6 v/v) for 90 min at room temperature to afford, after evaporation
under reduced pressure, ApU 1b (the reaction was monitored by
tlc which indicated quantitative conversion of 1a into the more
polar product 1b) (14).
Physical properties of 1a. TLC, ethyl acetate, Rf = 0.80;
methanol/chloroform 0.5:9.5 (two developments), Rf = 0.35. UV
(EtOH, nm): λ max, 236, 262sh, 278; λ min, 248 31P NMR
(CDCl3): –1.15, –1.28 p.p.m. (upfield relative to 85% H3PO4,
external capillary). 1H NMR (CDCl3, 500 MHz, δ p.p.m.),
anomeric protons: 6.095 (isomer 1), 6.128 (isomer 2), 1:3 ratio.
Physical properties of 1b. TLC, methanol/methylene chloride
1:9, Rf = 0.15. UV (EtOH, nm): λ max, 231,275; λ min, 228,249.
31P NMR (CDCl ): –0.43 p.p.m. (upfield relative to 85% H PO ,
3
3
4
external capillary). 1H NMR (CDCl3, 500 MHz, δ p.p.m.),
anomeric proton: 6.220.
Synthesis of ApU 2, ApUpU 3 and branched A(pU)pU 4 attached
to CPG. The structures of 2, 3 and 4 are illustrated in Figure 1A.
The required 5′-MMT-3′-silylated uridine-CPG support was
prepared as previously described (15). Syntheses of linear and
branched oligomers were conducted via the silyl-phosphoramidite method as described (16,17). Ancillary reagents for automated synthesis (Applied Biosystem DNA synthesizer, Model
381A) were prepared as follows: (a) coupling reagent: 0.5 M
solution of tetrazole in acetonitrile; (b) capping: cap A, 10% acetic
anhydride/10% 2,6-lutidine/THF and cap B, 16% N-methylimidazole/THF; (c) oxidation: 0.1 M iodine in THF/pyridine/water
(75:20:2 v/v/v); de-methoxytritylation: 5% trichloroacetic acid/
dichloromethane. Syntheses were conducted on a 1 µmol scale
and repeated (4×) to obtain ∼150 mg of each solid support (total:
4 µmol of bound oligomer). Average coupling efficiencies of
ribo-U and ribo-A3′- amidites were circa 98.3% and 95.1%,
respectively. Coupling of ribo-A2′,3′-bisphosphoramidite solution (0.03 M) onto CPG-ribo-U (30 µmol/g) yielded ∼70% of the
expected branched trinucleoside diphosphate A(pU)pU, 4 (18). A
small portion of each support was deprotected (16,17) and
analyzed by HPLC and gel electrophoresis (UV shadowing) to
confirm the efficiency of synthesis and the purity (∼90% in each
case) (HPLC conditions: reverse-phase Whatman Partisil ODS-2
column; solvent A: 20 mM KH2PO4, with a linear gradient 0–50%
solvent B: methanol, over 25 min, 1.5 ml/min flow, room
temperature).
Synthesis of TpoTpoT 5, TpsTpoT 6 and TpsTpsT 7 attached to
CPG. These oligomers, the structures of which are illustrated in
Figure 1B, were synthesized on an Applied Biosystem DNA
synthesizer (Model 381A) via the phosphoramidite method (see
above). Coupling yields were 96–98% (trityl assay method). For
the preparation of TpsTpsT, the sulfurizing reagent tetraethylthiuram disulfide (TETD) (19), obtained from Applied Biosystems,
was used in place of the iodine oxidizing reagent. The CPG beads
were allowed to react with TEDT for 10 min at room temperature
and then washed extensively with acetonitrile. In the case of
Figure 1. Structures of the various linear and branched, di- and trinucleotides
synthesized specifically for this study.
TpsTpoT, the iodine solution was replaced by TETD reagent after
the introduction of the first internucleotide (‘po’) linkage. Four
micromoles (∼150 mg) of oligomer-bound CPG were obtained in
each case for 31P NMR analysis. HPLC and gel electrophoresis
of the products cleaved from the support showed that, as expected,
2870 Nucleic Acids Research, 1996, Vol. 24, No. 15
the deprotected, crude phosphorothioate samples were composed
of a mixture of diastereomers, of ∼90% purity.
Synthesis of TpoT 8 via t-butyl hydroperoxide oxidation. TpoTCPG cyanoethylphosphite triester was prepared by coupling DMT-T
cyanoethylphosphoramidite reagent/tetrazole in acetonitrile to
T-CPG (5 × 1 µmol). The solid support was then capped with
acetic anhydride/2,6-lutidine/N-Me imidazole/THF (5 min, room
temperature). A solution containing t-butyl hydroperoxide (0.5 M,
6 ml) was drawn into a synthesis column containing TpT-CPG
(5 µmol) phosphite triester (20). After 3 min the solution was
expelled and the solid washed with dichloromethane followed by
ether, and dried under vacuum. A small portion of the support was
detritylated to determine the extent of coupling (∼100%).
NMR spectroscopy
Solid state NMR spectra were acquired using a Chemagnetics
CMX300 spectrometer equipped with a Chemagnetics MAS probe
doubly tuned to the resonance frequencies of phosphorus-31
(121.25 MHz) and hydrogen-1 (299.53 MHz). Samples consisting
of 100–125 mg of controlled-pore glass bearing 4–5 µmol
oligomer were spun in a 7.5 mm o.d. zirconium rotor at a spinning
rate of 6000 Hz. A single contact cross polarization technique was
employed using a contact time of 2 ms with proton decoupling
during the acquisition period. The proton radio field strength was
50 000 Hz. The magic angle was set using the resonance signal
from zinc(II) bis(O,O′-diethyldithiophosphate) (21). Spectra
were acquired using a sweep width of 100 kHz, a data size of 2 kHz,
and a recycle delay of 2 s. Adequate signal-to-noise was generally
achieved upon averaging ∼32 000 transients (∼18 h acquisition
time). All chemical shifts were referenced to external 85% H3PO4.
RESULTS AND DISCUSSION
Solid state 31P NMR of free oligonucleotides
of 31P
CPMAS NMR studies of
Before we describe the results
nucleotides bound to CPG, it is informative to examine the
spectra of similar compounds free of CPG, but likewise in the
solid state. Since large quantities of such nucleotides are readily
available, the spinning rotor of the MAS probe may be filled to
contain upwards of a millimole of phosphorus, provided no CPG
is present. Such compounds serve, therefore, the dual functions
of acting as convenient references for the optimization of
cross-polarization and signal acquisition conditions, and providing
a perspective on the 31P NMR spectra of nucleotides bound to
CPG. The cross-polarization contact time (2 ms) and the relaxation
delay (2 s) used throughout these measurements were chosen by
optimizing such spectra.
Figure 2 shows cross-polarization 31P NMR spectra of a dry
powder of the dinucleotides ApU 1a and ApU 1b acquired under
both static and MAS conditions. In this instance the 3′ position of
the ribose sugar of adenosine is linked via a phosphate bridge to
the 5′ position of the ribose sugar of uridine. Results are shown
for the phosphate with (1a) and without (1b) its cyanoethyl
(CNE) protecting group. During the course of a solid-phase
oligonucleotide synthesis, the CNE-protecting group blocks the
non-bridging phosphate oxygen from undergoing chemical
modification, and is removed just prior to the release of the
completed oligonucleotide from the solid support.
The broad 31P NMR spectra obtained for dry powdered
samples of 1a and 1b under static conditions arise from the fact
Figure 2. 31P CP NMR spectra of a dry powder of the dinucleotide ApU 1 not
attached to controlled pore glass. Top two spectra: CNE-protecting group
attached 1a, static and MAS conditions. Bottom two spectra: CNE-protecting
group removed 1b, static and MAS conditions. The static spectra were acquired
using cross-polarization as described in the text, but with a spin echo (tau = 30 µs)
technique to avoid ring-down problems. The principal components of the
chemical shift tensor, σ11 , σ22 and σ33 , are indicated. The MAS spectra
(ωr = 4500 Hz) were also acquired using cross-polarization, but without any
spin echo refocusing. The spinning side bands (SSBs) are indicated with
asterisks while the isotropic resonances are indicated as δ0.
that the chemical shielding experienced by a nucleus in a magnetic
field is a function of both isotropic (i.e. orientation-independent)
and anisotropic (i.e. orientation-dependent) interactions between the
nuclear spin magnetic moment and the magnetic field. In liquids
the rapid isotropic molecular motions average the anisotropic
interactions to zero, leaving only the isotropic interactions.
Consequently, in the NMR spectrum of a liquid sample the
resonance lines are narrow and the frequency at which a resonance
line is observed is a function of the isotropic chemical shift only.
In solids both the isotropic and the anisotropic interactions come
into play. Consequently, in the NMR spectrum of a solid sample
the frequency at which a resonance line is observed is a function of
the orientation of that nucleus with respect to the magnetic field,
according to
ω = ω0 (1 – δ0) + ω0 (1/2)(σ33 – δ0)(3 cos2 θ – 1 – ηsin2 θ cos2 φ ) 1
2871
Nucleic Acids
Acids Research,
Research,1994,
1996,Vol.
Vol.22,
24,No.
No.115
Nucleic
where ω = 2′πν is the observed frequency, ω0 is the Larmor
frequency, and δ0 is the isotropic chemical shielding. The anisotropic
chemical shielding is described in terms of a second-order tensor
having principal components σ11, σ22 and σ33. The coordinate
system for this tensor is fixed with respect to the molecular
geometry. The angles θ and φ are the colatitude and azimuthal polar
angles, respectively, describing the orientation of the chemical
shift tensor with respect to the magnetic field direction. The
asymmetry parameter η is defined such that η = (σ22 – σ11) /σ33.
The anisotropic chemical shift tensor elements are defined such
that |σ33| |σ22| |σ11|.
In a solid powder sample all orientations of the chemical shift
tensor with respect to the magnetic field are possible and the
NMR spectrum consist of a superposition of resonance lines from
individual spins at individual orientations. The result is a
so-called powder pattern lineshape in which the values of the
principal components of the chemical shift tensor appear as
inflection points. For instance, in Figure 2 the static 31P NMR
spectrum (i.e. in the absence of MAS but employing CP) for both
1a and 1b consists of a powder pattern lineshape from which the
values of the principal chemical shift tensor components are readily
measured, as shown by the arrows indicating the frequencies of
σ11, σ22 and σ33. Table 1 lists the values of the principal chemical
2871
shift tensor components measured for these and all other compounds
studied here. The overall shape of such powder spectra is defined
in terms of two quantities: the chemical shift anisotropy and the
asymmetry parameter. The chemical shift anisotropy corresponds
to the frequency spread in the spectrum, |σ33 – σ11|, and the values
measured for compounds 1a and 1b are typical of phosphate diand triesters in the solid state (22,23). Note that removing the
CNE-protecting group has little effect on the size of the chemical
shift anisotropy. The asymmetry parameter, η, was defined above
and one observes that a major effect of removing the CNE-protecting group is to increase the asymmetry parameter, as is readily
apparent if one compares the values of σ11 versus σ22 in the static
31P NMR spectra of 1a versus 1b in Figure 2. Converting a
neutral CNE-protected phosphotriester to an anionic phosphodiester decreases the symmetry of the electronic charge distribution
about the phosphorus nucleus. This will cause the asymmetry
parameter to increase. Any deeper interpretation of the spectral
changes observed upon removal of the CNE-protecting group
requires information regarding the orientation of the principal
tensor components relative to the molecular geometry. This
information cannot be obtained from powder NMR spectra alone.
Instead, one needs to investigate the orientation dependence of the
NMR spectra of single crystals of the species of interest.
Table 1. 31P NMR isotropic chemical shifts and anisotropic chemical shift tensor elements for free oligonucleotides and oligonucleotides bound to CPG
Oligonucleotide
δ0
CPG Free
1 ApU static (1a)
+TEA (1b)
–2.0
–2.0
ApU MAS (1a)
+TEA (1b)
σ11
σ22
σ33
|σ33 – σ11|
η
71
86
52
26
–122
–112
193
198
0.16
0.53
–2.0
–2.0
71
86
53
25
–122
–112
193
198
0.16
0.53
CPG Bound
2 ApU
+TEA
–2.0
–2.2
84
85
26
25
–110
–111
194
196
0.53
0.54
3 ApUpU
+TEA
–2.0
–2.1
90
88
25
25
–115
–114
205
202
0.56
0.55
4 A(2′pU)3′pU
+TEA
–2.0
–2.2
89
88
27
26
–116
–116
205
204
0.53
0.53
5 TpoTpoT
+TEA
–2.0
–2.1
93
87
24
25
–117
–113
210
200
0.59
0.55
6 TpsTpsT (i)
(ii)
+TEA
66.0
56.0
56.0
102
95
92
54
38
38
–157
–132
–130
259
227.0
222
0.31
0.43
0.41
7 TpsTpoT (po)
+TEA
(ps)
+TEA
–2.0
–2.0
66.0
56.0
85
90
104
91
30
25
55
32
–115
–116
–160
–123
200
206
264
214
0.48
0.56
0.30
0.48
8 TpT I2 oxidized
t-BuOOH oxidized
–2.0
–2.0
80
81
27
30
–106
–110
186
191
0.50
0.46
All chemical shifts are in parts per million (p.p.m.) referenced to 85% H3PO4. Anisotropic chemical shift tensor elements are defined such that |σ33| |σ22| |σ11|
where σii = (δii – δ0 ) and δ0 and δii are the observed isotropic and anisotropic chemical shifts. The asymmetry parameter is defined as η = (σ22 – σ11)/σ33.
2872 Nucleic Acids Research, 1996, Vol. 24, No. 15
When such solid powder samples are spun rapidly at the
so-called magic angle, the powder pattern spectrum breaks up
into a series of spinning side bands (SSBs) plus the isotropic
chemical shift resonance line. The SSBs are separated from the
isotropic resonance line by multiples of the spinning rate ωr. If the
spinning rate exceeds the size of the interaction (σ33 – δ0) then
only the isotropic chemical shift resonance line remains visible.
MAS produces a large enhancement in signal-to-noise because
the entire spectral intensity is now focussed under a few narrow
resonance lines rather than spread out over a range of frequencies.
In the 31P NMR spectra of solid powders of the same two
dinucleotides, 1a and 1b, again in Figure 2, under MAS
conditions (ωr = 4500 Hz) one observes an isotropic resonance
line at δ0 = –2.0 p.p.m. accompanied by a series of SSBs. The
isotropic resonance line is identified using the fact that its
frequency remains constant at different sample spinning rates,
while the positions of the SSBs (indicated by asterisks in the
figure) are directly a function of the rate of sample spinning.
The isotropic chemical shifts, δ0 , for 1a and 1b (see Table 1 for
details) are typical of phosphate di- and triesters in general and
oligonucleotides in particular (22–24). The presence or absence
of the CNE-protecting group has little impact on δ0, consistent
with the observed ∆δ0 in solution of ∼0.6–0.8 p.p.m. (see
Experimental). The isotropic chemical shift of phosphorus is
known to be sensitive to the phosphorus-oxygen bond angles, and
distortions occurring in the solid state can lead to differences in
the observed isotropic chemical shift positions for the identical
compound in the solid versus the solution state 31P NMR
spectrum (25). Since the isotropic chemical shifts observed here
in the solid state are similar to those reported for identical
compounds in solution, we conclude that any bond angle
distortions in the solid are slight.
It is also possible to extract the anisotropic chemical shift
information (i.e. the values of the individual phosphorus chemical
shift tensor components) from such 31P CPMAS NMR spectra
using the graphical method of Herzfeld and Berger (26). This
method relies on measuring the intensities and frequencies of the
SSBs versus the isotropic resonance line. For example, in the 31P
CPMAS NMR spectra of 1a versus 1b in Figure 2, it is evident
that the ratio of the intensity of the (n +1) SSB intensity to that of
the isotropic resonance line is very different in the two
compounds. (The n +1 SSB is immediately downfield of the
isotropic resonance line.) This is a manifestation of the different
asymmetry parameters characterizing the chemical shift tensor in
the two cases. When the intensities of all the SSBs are taken into
account for the 31P CPMAS NMR spectra of ApU 1a and 1b, and
the analysis of Herzfeld and Berger (26) is applied, one obtains
an excellent agreement between the values of the chemical shift
tensor components obtained under static versus MAS conditions
(see Table 1). Therefore, both isotropic and anisotropic chemical
shift parameters can be obtained from 31P CPMAS NMR spectra.
This is an important point because, as will be described below, for
di- and tri-nucleotides attached to a solid support it is not possible
to acquire static 31P NMR spectra in a reasonable length of time.
Thus, one has no choice but to employ MAS.
Solid state 31P NMR of oligonucleotides bound to CPG
When attempting to observe the 31P NMR spectrum of short
oligonucleotides bound to CPG, the bulk of the sample volume
consists of the solid-phase support itself. Consequently, the
Figure 3. 31P CPMAS NMR spectra of linear and branched di- and
trinucleotides attached to a CPG solid-phase support. Top spectrum: ApU 2.
Middle spectrum: ApUpU 3. Bottom spectrum: A(2′pU)3′pU 4. All cases
ωr = 6200 Hz. The SSBs are indicated with asterisks while the isotropic
resonances are indicated with arrows.
maximum phosphorus content of any one sample is only 5 µmol
for a monophosphate dinucleoside, and MAS becomes an absolute
necessity for obtaining reasonable signal-to-noise spectra in reasonable lengths of time. Figure 3 shows representative 31P CPMAS
NMR spectra of selected di- and tri-nucleotides bound to CPG,
after removal of the 5′-trityl protecting group. The top spectrum
was obtained with the dinucleotide ApU 2. As with compound 1a,
the internucleotide linkage is 3′-5′, but the 2′ position of the
uridine ribose sugar is now linked via a long-chain alkylamine
spacer to the surface of the CPG support (Fig. 1). The middle and
bottom spectra in Figure 3 were obtained with the linear
trinucleoside diphosphate (‘trinucleotide’) ApUpU 3 and the
branched ‘trinucleotide’ A(2′pU)3′pU 4, respectively, both
attached to CPG via a long-chain alkylamine linkage to the 2′
position of the ribose ring of uridine. In the linear trinucleotide,
both phosphodiesters are involved in 3′-5′ links to the adjoining
ribose sugars. In the branched trinucleotide, one phosphate bridge
links the 3′ position of the adenosine ribose ring to the 5′ position
of the CPG-attached uridine ribose ring, while another links the
2′ position of the adenosine ribose to the 5′ position of a second
uridine ribose (see Fig. 1 for structural details).
2873
Nucleic Acids
Acids Research,
Research,1994,
1996,Vol.
Vol.22,
24,No.
No.115
Nucleic
In Figure 3 the 31P CPMAS NMR spectra of 2, 3 and 4 each
consist of an isotropic phosphorus resonance line (δ0 = –2.0 p.p.m.)
plus a series of SSBs (ωr = 6200 Hz). The isotropic resonances are
indicated with an arrow while the SSBs are indicated with
asterisks. The chemical shift of the isotropic resonance line in all
three cases is typical of a phosphate phosphorus (Table 1), and is
virtually identical to that of the dinucleotides ApU 1a or 1b not
bound to CPG. It follows that the oxidation step in the
oligonucleotide synthetic cycle, from the phosphite to the
phosphate triester, occurs to completion in situ, since a phosphite
triester is expected to manifest an isotropic chemical shift in the
region of 140 p.p.m. (22,23). Moreover, it appears that the presence
(1a and 1b) or absence (2, 3 and 4) of the monomethoxytrityl
protecting group on the ribose ring has no influence on the
observed isotropic chemical shift.
In each case in Figure 3, only a single isotropic 31P NMR
resonance line is resolved, so that it is not possible to differentiate
between di- and trinucleotides, nor between linear and branched
trinucleotides, on the basis of their 31P NMR isotropic chemical
shifts when bound to CPG. Any difference in chemical shift
between the two 3′-5′ linked phosphodiesters in the linear
trinucleotide is expected to be quite small (24). As for the
branched trinucleotide, it is known from 31P NMR measurements
in solution that the isotropic chemical shift of 2′-5′ linked
phosphodiesters occurs about 0.94 p.p.m. towards higher field
than that of 3′-5′ linked phosphodiesters (24). The width-at-halfheight (∆ν1/2) of the isotropic resonance lines in the solid state
spectra in Figure 3 is typically ∼75 Hz. This rather broad
linewidth probably reflects a distribution of bond angles amongst
the individual phosphoesters—a distribution which cannot average
out due to the large energy barriers to rotation in the solid state.
When the differences in two isotropic chemical shifts are small
relative to the linewidth, they will not be resolved. It may also
occur that bond angle distortions in the solid state, relative to the
solution state, render the isotropic chemical shifts of 2′5′ and 3′5′
phosphodiesters into similar values. In either case, it would
appear that 31P CPMAS NMR spectroscopy has limited utility for
differentiating such structural details in the solid state.
The chemical shift anisotropy and the asymmetry parameters
for the dinucleotide, 2, and the linear, 3, and branched, 4,
trinucleotides (Table 1) all correspond closely with the values
measured for ApU lacking its CNE-protecting group (i.e. 1b). In
fact, no change in spectral features was observed after the
supports were treated with triethylamine solution for 90 min,
conditions known to selectively cleave CNE groups (14). Note
that the principal components of the chemical shift tensor, and the
asymmetry parameter in particular, are quite sensitive to the
presence or absence of the CNE-protecting group, as shown in
Figure 2 and detailed in Table 1. Thus, 31P CPMAS NMR
indicates that CNE-protecting groups are removed from the
phosphate during the course of chain assembly.
The size of the anisotropic chemical shifts indicate further that
the phosphate group experiences no large amplitude motions
when the nucleotides are bound to CPG and there is no solvent
present. As for any difference in the mobilities of the two
phosphate diesters in ApUpU 3, proximal and distal to the CPG,
these would have to be rather large before they could be
differentiated, given the MAS technique employed and the levels
of phosphorus available in these samples. The fact is that the
isotropic chemical shifts of these two phosphorus atoms are
similar, and the isotropic chemical shift will not be affected by
2873
motional averaging. The chemical shift anisotropy would be
reduced by motional averaging, and this would affect the
intensities of the SSBs, but not their frequencies. Small differences
in motional averaging will affect primarily the intensities of those
SSBs furthest in frequency from the isotropic chemical shift
position. However, these very intensities are the lowest to begin
with and have the highest uncertainty in quantification.
The spectra in Figure 4 demonstrate that 31P CPMAS NMR
spectroscopy is well suited for differentiating large changes in
phosphorus chemistry, such as the substitution of a phosphorothioate for a phosphate moiety. Here we compare three linear
thymidine trinucleotides: one containing only phosphate linkages
TpoTpoT, 5, a second containing only phosphorothioate linkages
TpsTpsT, 6, and a third containing both phosphate and phosphorothioate linkages TpsTpoT, 7. In the preparation of TpsTpsT, 6, the
sulfurizing reagent tetraethylthiuram disulfide (TETD/acetonitrile)
(19) was used in place of the iodine oxidizing reagent (see
Experimental). In the synthesis of TpsTpoT, 7, the iodine solution
was replaced by TETD reagent after the introduction of the first
phosphate ‘po’ linkage. Spectra were recorded before (Fig. 4A)
and after (Fig. 4B) treating the supports with triethylamine.
The spectrum of the trinucleotide containing only phosphate
linkages, TpoTpoT, 5, is virtually identical before and after
treatment with triethylamine (top spectra in parts A and B of Fig. 4).
In both cases one observes a single isotropic resonance line
(denoted by an arrow in the figure) accompanied by SSBs
(denoted by asterisks in the figure). The isotropic resonance line
displays a chemical shift consistent with phosphate phosphorus
(22,23). Analysis of the relative intensities and positions of the
isotropic resonance and the SSBs by the Herzfeld and Berger
method (26) reveals that the asymmetry parameter of TpoTpoT,
5, both before and after treatment of triethylamine, is rather
similar to that of the dinucleotide ApU, 1b, lacking a CNE group
as discussed above. This indicates again that the CNE-protecting
group can be removed prior to any treatment with triethylamine.
The fact that the 31P isotropic chemical shifts of the thymidine
nucleotides are nearly identical to those of the adenosine and
uridine nucleotides above, is a further indication that in the solid
state chemical details at points distant from the phosphorus atom
itself are not manifested in the spectrum.
The spectrum of the trinucleotide containing only phosphorothioate linkages, TpsTpsT, 6, prior to treatment with triethylamine
(middle spectrum in Fig. 4A), consists of two distinct isotropic
resonance lines (66.0 and 56.0 p.p.m., respectively) and two
associated patterns of SSBs. Summing the intensities of the
isotropic resonance line plus SSBs yields an intensity ratio of
approximately 3:2 for the 66.0 p.p.m. versus 56.0 p.p.m.
component. Herzfeld and Berger analysis (26), shows that the
66.0 p.p.m. component has an asymmetry parameter of 0.31
while the 56.00 p.p.m. component has an asymmetry parameter
of 0.43. Again, since no resonance line is observed in the
phosphite region of the spectrum, one concludes that the
oxidation step, this time with tetraethylthiuram disulfide, has
occurred to completion. When one treats the CPG-bound
TpsTpsT, 6, with triethylamine one obtains the middle spectrum
in Figure 4B, which shows only a single isotropic resonance line
at 56.0 p.p.m. and an asymmetry parameter of 0.41.
We suggest that the two components in the 31P CPMAS NMR
spectrum of TpsTpsT, 6, prior to triethylamine treatment
correspond to populations of 6 with and without the CNEprotecting group. First, evidence in the literature indicates that
2874 Nucleic Acids Research, 1996, Vol. 24, No. 15
Figure 4. 31P CPMAS NMR spectra of phosphate and phosphorothioate thymidine trinucleotides attached to a CPG solid-phase support. (A) Before TEA treatment,
(B) after TEA treatment. Top spectrum: TpoTpoT 5. Middle spectrum: TpsTpsT 6. Bottom spectrum: TpsTpoT 7. All cases ω r = 6400 Hz, except 7 after TEA treatment
where ω r = 5200 Hz. The SSBs are indicated with asterisks while the isotropic resonance lines are indicated with arrows.
converting a triester to a diester results in a large upfield shift
in the isotropic chemical shift of phosphorothioates, but has
little effect on phosphates. For instance, (EtO)3P(S) has an
isotropic chemical shift of 68.1 p.p.m. (27), but loss of one ethyl
ester to produce (EtO)2P(S)O– yields an isotropic chemical
shift of 48.2 p.p.m. (28). Other examples abound and have been
compiled by Tebby (23). In contrast, the isotropic chemical
shift of the phosphotriester (EtO)3P(O) equals –1.2 p.p.m. (29)
which shifts to only –1.8 p.p.m. upon loss of one ethyl ester to
produce (EtO)2P(O)O– (30). Second, the asymmetry parameter
for the 56.0 p.p.m. isotropic chemical shift phosphorothioate
component is significantly larger than that of the 66.0 p.p.m.
isotropic chemical shift phosphorothioate component. The same
trend of an increase in the asymmetry parameter of the chemical
shift tensor components was observed upon removal of the
CNE-protecting group for the case of 1a versus 1b as discussed
above. Third, upon treatment of TpsTpsT, 6, with triethylamine
to effect CNE-protecting group removal, the only spectral
component remaining was that with the isotropic chemical shift of
56.0 p.p.m., while its asymmetry parameter was nearly identical
to that of the 56.0 p.p.m. component present prior to triethylamine treatment.
In the spectrum of the mixed phosphate-phosphorothioate
trinucleotide TpsTpoT, 7, prior to treatment with triethylamine,
as shown at the bottom of Figure 4A, one observes two isotropic
resonance lines and two overlapping sets of SSBs. One isotropic
resonance line occurs at –2.0 p.p.m. and Herzfeld and Berger
analysis of its SSBs (26) yields an asymmetry parameter equal to
0.48, consistent with a phosphate phosphorus lacking its CNEprotecting group. The second isotropic resonance occurs at
66.0 p.p.m., and Herzfeld and Berger analysis of its SSBs (26)
yields an asymmetry parameter equal to 0.30, consistent with a
phosphorothioate phosphorus having its CNE-protecting group
intact. Thus, oxidation of the first internucleotide linker to ‘po’
with the iodine reagent apparently removes CNE-protecting
groups, while oxidation of the second internucleotide linker to
‘ps’ with TETD reagent leaves the CNE-protecting groups at least
partially intact.
In the spectra of TpsTpoT 7 prior to TEA treatment (bottom
spectrum in Fig. 4A), it appears that the total summed intensity
2875
Nucleic Acids
Acids Research,
Research,1994,
1996,Vol.
Vol.22,
24,No.
No.115
Nucleic
of the isotropic chemical shift line plus SSBs is lower for the
phosphorothioate versus phosphate spectral component. We note
that caution must be exercised before relying on intensities from
CPMAS NMR spectra for quantification. First, the cross-polarization conditions employed here were optimized for phosphate
esters, and are not necessarily optimal for phosphorothioate
esters. Thus, any disagreement between the relative intensities of
phosphate versus phosphorothioate resonances might be purely
fortuitous. That being said, the similarities in the proton
populations surrounding the two phosphorus types provides little
reason to anticipate large differences in their rates of cross-polarization, or relaxation, or intensity enhancements under cross-polarization conditions. Second, the rate of sample spinning in this
particular instance places the first spinning side band from the
phosphate resonance at ∼56.0 p.p.m., i.e. virtually overlapping
the expected resonance frequency of any phosphorothioates
lacking a CNE-protecting group. Third, it is likely that a portion
of the phosphorothioates have lost their CNE-protecting groups,
as above for TpsTpsT, 6, prior to TEA treatment. This will split
each resonance into two components and complicate quantification
of intensities. In summary, the phosphorothioate versus phosphate
levels cannot be stated unequivocally from this particular spectrum.
The bottom spectrum of Figure 4B show the results of
triethylamine treatment of the trinucleotide TpsTpoT, 7 on its 31P
CPMAS NMR spectrum. After TEA treatment, two spectral
components are again evident. One displays an isotropic chemical
shift equal to –2.0 p.p.m. and an asymmetry parameter of 0.56,
consistent with a phosphate phosphorus lacking any CNE-protecting
group. The second displays an isotropic chemical shift equal to
56.0 p.p.m. and an asymmetry parameter of 0.48, consistent with
a phosphorothioate phosphorus lacking any CNE-protecting
group. Note that in this spectrum the spinning frequency was
slowed to 5200 Hz (versus 6400 Hz for the others in this figure)
in order to avoid overlap of SSBs from the phosphate and
phosphorothioate powder patterns. The sum of the integrated
areas of the SSBs plus the isotropic resonance line is now
approximately the same for the phosphate and phosphorothioate
resonances. This suggests, but does not prove (see qualifying
statements above), that virtually all available sites (5′-hydroxyls)
were coupled during the second round of synthesis which added
the terminal (Tps) residue. This conclusion is supported by the
coupling yield efficiencies (∼98%) measured by the trityl assay
method (see Experimental).
Concerning removal of the CNE-protecting group
The 31P CPMAS results obtained here indicate that this technique
is particularly sensitive to the presence or absence of the
CNE-protecting group. For phosphate phosphorus, removal of
the CNE-protecting group produces a large increase in the
spectral asymmetry parameter η, but only small changes in the
isotropic chemical shift. For phosphorothioate phosphorus,
removal of the CNE-protecting group produces a relatively small
increase in the value of η, but a large upfield shift in the isotropic
chemical shift. The results suggest further that removal of
CNE-protecting groups can occur during the course of oligonucleotide synthesis prior to their deliberate removal. This is
somewhat surprising since it has generally been assumed that
oligomers prepared on CPG supports retain CNE-protecting
groups prior to deprotection with a base (usually concentrated
2875
aqueous ammonia). In principle, CNE groups can be removed in
the course of the capping reaction, or subsequent to capping
during the iodine/pyridine/water oxidation reaction (31). We
conducted model experiments in solution which show that ApU
phosphate-triester (1a) is stable to the oxidizing iodine/pyridine/
water reagent (24 h) and to the capping acetic anhydride/Nmethylimidazole solution (30 min) under conditions used in the
solid-phase synthesis. Also, a sample of TpU at the phosphite
triester stage was stable, in solution, to the capping reagent as
judged by liquid state 31P NMR (δ 140.1 and 140.3 p.p.m.,
CDCl3). Addition of iodine/pyridine/water to this sample produced
a TpU phosphate triester, demonstrating that, when free in
solution, a nucleoside cyanoethyl phosphite or phosphate triester
is resistant to the reagents present in the capping or oxidation
solutions. This suggests that some surface property of the support
itself contributes to the removal of CNE groups.
Letsinger and co-workers (20) observed that partial loss of a
P-O methyl protecting group occurred during iodine oxidation of
an oligonucleotide phosphite bond to CPG, and that the loss of the
protecting group could be avoided by substituting tertiary butyl
peroxide (t-BuOOH) as the oxidizing agent. We tested whether
this strategy could be employed to avoid removal of the
CNE-protecting group in our syntheses by comparing the 31P
CPMAS NMR spectra for TpT bound to CPG 8 for the case of
iodine versus t-BuOOH oxidation. The results of the spectral
analysis are shown in Table 1 and they indicate that the
CNE-protecting group is lost during the course of the solid phase
synthesis regardless of the method of oxidation. Again, this
suggests that some property of the CPG surface environment
encourages removal of the CNE-protecting group. Further
investigations of these effects are clearly warranted. It would be
interesting to monitor the 13C CPMAS NMR spectrum of
13C-labelled cyanoethyl versus methyl protecting groups to
directly observe the point at which they are removed, rather than
indirectly through 31P NMR.
SUMMARY AND CONCLUSIONS
Solid-state CPMAS 31P NMR spectroscopy of oligonucleotides
attached to a solid-phase support provides specific information on
the chemistry of the phosphorus atom located within the
oligonucleotide backbone. The phosphorus oxidation state is
readily differentiated, but the detailed configuration of the
phosphorus-sugar linkages (2′5′ versus 3′5′) are beyond the
present limits of resolution. Other large changes in the phosphorus
chemistry, such as removal of the CNE-protecting group during
chain assembly, or replacement of a phosphate with a phosphorothioate moiety, are readily accessible to characterization. It seems
likely that, in the future, one will be able to employ CPMAS 31P
NMR to monitor individual steps during a solid phase oligonucleotide synthetic cycle in situ, and to assess the efficiency of
sulfuration, and other modifications at phosphorus, such as those
that are relevant in therapeutic (antisense) applications [for a
recent review see (32)].
ACKNOWLEDGEMENTS
This work was supported by operating grants from the National
Science and Engineering Research Council (NSERC) of Canada
to PMM and MJD.
2876 Nucleic Acids Research, 1996, Vol. 24, No. 15
REFERENCES
1 Gait,M.J. (ed.) (1984) Oligonucleotide Synthesis: A Practical Approach.
IRL Press, Oxford.
2 Caruthers,M.H. (1987) In Narang,S. (ed.), Synthesis and Applications of
DNA and RNA. Academic Press, Orlando, FL, USA, 20, 47–94.
3 Atkinson,T. and Smith,M. (1984) Oligonucleotide Synthesis: A Practical
Approach. IRL Press, Oxford, pp. 35–81.
4 Pon,R.T. (1993) In Agrawal,S. (ed.), Methods in Molecular Biology—
Protocols for Oligonucleotides and Analogs. Humana Press, Totowa, NJ,
USA, 20, 81–114.
5 McCollum,C. and Andrus,A. (1991) Tetrahedron Lett., 48, 4069–4072.
6 Adams,S.P., Kavka,K.S., Wykes,E.J., Holder,S.B. and Gallupi,G.R. (1983)
J. Am. Chem. Soc., 105, 661–663.
7 Crooke,S., Lebleu,B. (eds) (1993) Antisense Research and Applications.
CRC Press, Boca Raton, FL, USA.
8 Sproat,B., Colonna,F., Mullah,B., Tsou,D. andrus,A., Hampel,A. and
Vinayak, R. (1995) Nucleosides Nucleotides, 14, 255–273.
9 Sinha,N.D. and Fry,S. (1994) In Sanghvi,Y.S. and Cook,D.P. (eds)
Carbohydrate Modifications in Antisense Research. ACS Symposium
Series, 580, 184–198.
10 Caravajal,G.S., Leyden,D.E., Quinting,G.R. and Maciel,G.E. (1988) Anal.
Chem., 60, 1776–1786.
11 Kallury,K.R., Macdonald,P.M. and Thompson,M. (1994) Langmuir, 10,
492–499.
12 Letsinger,R.L. and Lunsford,W.B. (1976) J. Am. Chem. Soc., 98,
3655–3661.
13 Ogilvie,K.K., Theriault,N.Y., Seifert,J.M., Pon,R.T. and Nemer,M.J. (1980)
Can.J. Chem., 58, 2686–2693.
14 Efimov,V.A., Reverdatto,S.V., and Chakhmakhcheva,O.G. (1982)
Tetrahedron Lett., 23, 961–964.
15 Damha,M.J., Giannaris,P.A. and Zabarylo,S.V. (1990) Nucleic Acids Res.,
18, 3813–3821.
16 Usman,N., Ogilvie,K.K., Jiang,M.-Y. and Cedergren,R.J. (1987) J. Am.
Chem. Soc., 109, 7845–7854.
17 Damha,M.J. and Ogilvie,K.K. (1993) In Agrawal,S. (ed.) Methods in
Molecular Biology—Protocols for Oligonucleotides and Analogs. Humana
Press, Totowa, NJ, USA, 20, 81–114.
18 Damha,M.J., Ganeshan,K., Hudson,R.H.E. and Zabarylo,S.V. (1992)
Nucleic Acids Res., 20, 6565–6573.
19 Vu,H. and Hirschbein,B.L. (1991) Tetrahedron Lett., 32, 3005–3008.
20 Alul,R.H., Singman,N., Zhang,G. and Letsinger,R.L. (1991) Nucleic Acids
Res., 19, 1527–1532.
21 Kubo,A. and McDowell,C.M. (1991) J. Magnet. Res., 92, 409–412.
22 Gorenstein,D.G. (1984) Phosphorus-31 NMR: Principles and
Applications. Academic Press, Orlando, FL, USA.
23 Tebby,J.C. (1991) Handbook of Phosphorus-31 Nuclear Magnetic
Resonance. CRC Press, Boca Raton, FL, USA.
24 Damha,M.J. and Ogilvie,K.K. (1988) Biochemistry, 27, 6407–6416.
25 Verkade,J.G. and Quinn,L.D. (1984) Phosphorus-31 NMR Spectroscopy in
Stereochemical Analysis. VCH Publishers, Dearfield Beach, FL, USA.
26 Herzfeld,J. and Berger,A.E. (1980) J. Chem. Phys., 73, 6021–6030.
27 Moedritzer,L., Maier,L. and Groenwegh,L.C.D. (1962) J. Chem. Eng.
Data, 7, 307.
28 Cooper,D.B., Harrison,J.M., Inch,T.D. and Lewis,G.J. (1974) J. Chem.
Soc. Perkin Trans., 1, 1049.
29 Brown,B., Boudreau,J.A., Hewitson,B. and Hudson,R.F. (1976) J. Chem.
Soc. Perkin Trans., 2, 888.
30 Satterthwait,A.C. and Westheimer,F.H. (1980) J. Am. Chem. Soc., 102,
4464.
31 Alul,R.H., Singmand,C.N., Zhang,G. and Letsinger,R.L. (1991) Nucleic
Acids Res., 19, 1527–1532.
32 Sanghvi,Y.S. and Cook,D.P. (eds) (1994) Carbohydrate Modifications in
Antisense Research. ACS Symposium Series 580.