Synthesis and Characterization of Oligonucleotides Containing 2′-S,3′-O-Cyclic Phosphorothiolate Termini

Full text of DOI:10.1021/jo9903508

5700
J . Org. Chem. 1999, 64, 5700-5704
of 2′-deoxy-2′-mercaptouridine 3′-(p-nitrophenyl phos-
Syn th esis a n d Ch a r a cter iza tion of
Oligon u cleotid es Con ta in in g
phate), whereby the 2′-sulfur attacks the activated
neighboring phosphodiester and displaces p-nitrophe-
nol.12 Unfortunately, neither of these approaches is
readily adaptable to the installation of a 2′-S,3′-O-cyclic
phosphorothiolate at the 3′ terminus of oligonucleotides.
Although oligonucleotides containing a 2′-mercaptonu-
cleoside are synthetically accessible,9 they cannot serve
as precursors for cyclic phosphorothiolate formation via
the transphosphorylation pathway because of the weak
nucleophilicity of sulfur at nonactivated phosphate cen-
ters12 and the tendency of the 2′-sulfur to attack the 1′-
carbon and displace the heterocyclic base.9 However, Liu
et al. and Weinstein et al. showed that 3′-S-phospho-
rothiolate linkages are 2-3 orders of magnitude more
susceptible to nucleophilic attack by the adjacent 2′-
hydroxyl than the 3′-O-phosphate linkage in normal
RNA, presumably due to geometric factors.13,14 If a 2′-S-
phosphorothiolate linkage is similarly susceptible to
nucleophilic attack by its neighboring 3′-hydroxyl group,
then an oligonucleotide containing this kind of linkage
might be an effective precursor to a 2′-S,3′-O-cyclic
phosphorothiolate. Here we describe a general method
for the synthesis of oligonucleotides containing a 2′-S-
phosphorothiolate linkage and show that this linkage
reacts under alkaline conditions to form an oligonucle-
otide 2′-S,3′-O-cyclic phosphorothiolate.
The preparation of ribonucleoside phosphoramidite
building blocks for standard oligonucleotide synthesis
requires protection of the 5′- and 2′-oxygens as 4,4′-
dimethoxytrityl (DMTr) and tert-butyldimethylsilyl (TB-
DMS) ethers, respectively, and phosphitylation of the 3′-
oxygen. Using the same 5′ and 2′ protecting groups, Sun
et al. prepared 3′-thioribonulceoside 3′-S-phosphoramid-
ites and used them to install 3′-S-phosphothiolate link-
ages into oligonucleotides.15 Analogously, we approached
the installation of 2′-S-phosphorothiolate linkages into
oligonucleotides using 2′-mercaptonucleoside 2′-S-phos-
phoramidites, in which the 5′- and 3′-hydroxyl groups
were protected as DMTr and TBDMS ethers, respec-
tively. Attempts to protect the 3′-oxygen of 4-N-benzoyl-
5′-O-(dimethoxytrityl)- 2′-deoxy-2′-(tritylthio)cytidine as
a TBDMS ether failed even in the presence of strong base
catalysts presumably due to steric hindrance from the
large trityl group. Instead, we first converted the 2′-
tritythio moiety to a 2′-pyridyl disulfide moiety and
synthesized 4-N-benzoyl-5′-O-(dimethoxytrityl)-3′-O-(tert-
butyldimethylsilyl)-2′-deoxy-2′-thiocytidin-2′-yl â-cyano-
ethyl N,N-diisopropylphosphoramidite 6 according to
Scheme 1. Treatment of 4-N-benzoyl-2′-deoxy-2′-(tritylth-
io)cytidine (1)9 with AgNO3 followed by Aldrithiol-2 gave
the disulfide 2 in 83% yield (Scheme 1). Reaction of 2
with DMTrCl in pyridine using DMAP as a catalyst gave
3 in 85% yield. Treatment of 3 with a large excess of
2′-S,3′-O-Cyclic P h osp h or oth iola te Ter m in i
Michelle L. Hamm and J oseph A. Piccirilli*
Howard Hughes Medical Institute and Departments of
Biochemistry and Molecular Biology, and Chemistry,
The University of Chicago, 5841 South Maryland Avenue,
MC 1028, Chicago, Illinois 60637
Received February 25, 1999
Sulfur substitution of the oxygen atoms in nucleic acids
has been widely used to study nucleic acid structure and
function and the mechanisms of catalysis employed by
ribozymes and enzymes that modify nucleic acids. In the
case of metalloenzymes that catalyze phosphoryl transfer
reactions, sulfur substitution of the oxygen atoms of the
substrates and products is an especially powerful tool for
mapping ground state and transition state interactions
at the active site and for determining whether these
interactions are mediated by metal ions.1-8 A specific
oxygen atom is identified as a ligand for a metal ion when
sulfur substitution of that atom shifts the metal ion
specificity of the metalloenzyme to a more thiophilic
metal ion (for example, from Mg2+ to Mn2+).1 Accordingly,
for those enzymes and ribozymes that catalyze site
specific cleavage of RNA by facilitating attack of the 2′-
hydroxyl on its neighboring phosphodiester (e.g., RNases
and the hammerhead and HDV ribozymes), analogues
of the substrate and product, in which the 2′-oxygen at
the cleavage site is replaced by sulfur, should provide
valuable probes for studying metal interactions with the
nucleophile. Toward this goal, we recently described a
method to install 2′-mercaptonucleosides into oligonucle-
otides, allowing the synthesis of substrate analogues.9 To
complement these studies, we now report methods for the
synthesis of the corresponding product analoguess
oligonucleotides containing 2′-S,3′-O-cyclic phosphorothi-
olate termini.
There are two examples of mononucleoside 2′-S,3′-O-
cyclic phosphorothiolates reported in the literature. Patel
et al. prepared the cytidine analogue by treating 2,2′-
anhydro-1-â-D-arabinocytosine with P2S5 followed by
incubation in pyridine/water and treatment of the result-
ing product with iodine.10,11 Dantzman and Kiessling
prepared the uridine analogue by transphosphorylation
† Phone: (773) 702-9312. Fax: (773) 702-3611. e-mail: jpicciri@
midway.uchicago.edu.
(1) Piccirilli, J . A.; Vyle, J . S.; Caruthers, M. H.; Cech, T. R. Nature
1993, 361, 85.
(2) Kuimelis, R. G.; McLaughlin, L. W. J . Am. Chem. Soc. 1995, 117,
11019.
(3) Kuimelis, R. G.; McLaughlin, L. W. Biochemistry 1996, 35, 5308.
(4) Zhou, D.-M.; Usman, N.; Wincott, F. E.; Matulic-Adamic, J .;
Orita, M.; Zhang, L.-H.; Komiyama, M.; Kumar, P. K. R.; Taira, K. J .
Am. Chem. Soc. 1996, 118, 5862.
(5) Zhou, D.-M.; Zhang, L.-H.; Taira, K. Proc. Natl. Acad. Sci. U.S.A.
1997, 94, 14343.
(6) Sontheimer, E. J .; Sun, S.; Piccirilli, J . A. Nature 1997, 388, 801.
(7) Curley, J . F.; J oyce, C. M.; Piccirilli, J . A. J . Am. Chem. Soc.
1997, 119, 12691.
(8) Weinstein, L. B.; J ones, B. C. N. M.; Cosstick, R.; Cech, T. R.
Nature 1997, 388, 805.
(9) Hamm, M. L.; Piccirilli, J . A. J . Org. Chem. 1997, 62, 3415.
(10) Patel, A. D.; Schrier, W. H.; Nagyvary, J . J . Org. Chem. 1980,
45, 4830.
(11) Divikar et al. called into question the authenticity of the product
on the basis of NMR inconsistencies with a similar product they
synthesized by an independent route. See: Divikar, K. J .; Mottah, A.;
Reese, C. B.; Sanghvi, Y. S. J . Chem. Soc., Perkin. Trans. 1 1990, 969.
(12) Dantzman, C. L.; Kiessling, L. L. J . Am. Chem. Soc. 1996, 118,
11715.
(13) Liu, X.; Reese, C. B. Tetrahedron Lett. 1996, 37, 925.
(14) Weinstein, L.B.; Earnshaw, D. J .; Cosstick, R.; Cech, T. R. J .
Am. Chem. Soc. 1996, 118, 10341.
(15) Sun, S.; Yoshida, A.; Piccirilli, J . A. RNA 1997, 3, 1352.
10.1021/jo9903508 CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/02/1999

Notes
J . Org. Chem., Vol. 64, No. 15, 1999 5701
Sch em e 1
F igu r e 1. The alkaline cleavage reactions of oligonucleotides
containing 2′-O- (7a ; lane 2) and 2′-S-linked (7b; lane 11)
phosphodiesters. 7a was treated with 0.01 M NaOH for 2, 4,
8, 15, and 30 min at 90 °C (lanes 3-8), and 7b was treated at
30 °C (lanes 12-17). To compare reactivity under the same
conditions, both oligonucleotides were incubated with 0.1 M
NaOH at 50 °C for 10 min (lanes 9 and 18). Oligonucleotides
were tested for the presence of a phosphorothiolate linkage
by treatment with 20 mM AgNO3 for 20 min in the dark (lanes
1,7, 10, and 16).
imidazole and TBDMSCl in a small volume of pyridine
afforded 4 in 86% yield. The disulfide was reduced with
dithiothreitol (DTT) to generate the 2′-mercapto deriva-
tive 5 in 95% yield, which was phosphitylated to the
phosphoramidite 6 in 90% yield (Scheme 1).
10; Scheme 2). To determine the feasibility of using 2′-
S-phosphorothiolate-linked oligonucleotides to generate
2′-S,3′-O-cyclic phosphorothiolate termini, we examined
the alkaline cleavage reactions of 7a and 7b (Figure 1).
The reactivity of 2′-5′-linked RNA 7a under basic condi-
tions (Figure 1, lanes 3-8) is expected to be similar to
the well-documented pathway by which natural RNA
reacts: the neighboring hydroxyl group attacks the
phosphodiester to form a 2′,3′-O-cyclic phosphate 8a (the
slower migrating band) that then hydrolyzes to a mixture
of 2′- and 3′-monophosphates, 9a and 10a , respectively
(the faster migrating band).17-19 Consistent with these
assignments, the band corresponding to 8a decreases in
intensity relative to that for 9a /10a upon longer exposure
to alkaline conditions, suggesting that the former is
converted to the latter. As observed previously for the
reaction of 3′-S-phosphorothiolate linkages under basic
conditions,13-15 the 2′-S-phosphorothiolate linkage was
much more susceptible to alkaline cleavage than the
corresponding phosphate linkage. In 0.01 M NaOH, 7a
required heating at 90 °C (Figure 1, lanes 3-8) to show
a rate of cleavage comparable to that of 7b at 30 °C (lanes
12-17). Under identical conditions (0.1 M NaOH and 50
°C), 7b reacts completely within 10 min (Figure 1, lane
18) whereas 7a remains mostly unreacted after 10 min
(lane 9). The initial stages of the reaction of 7b at 30 °C
give rise predominately to a product that migrates in a
manner similar to that of the cyclic phosphate product
8a (derived from 7a ) and is presumably the 2′-S,3′-O-
cyclic phosphorothiolate 8b. As the reaction proceeds, two
additional radioactive bands form: one that migrates
faster than 8b and in a manner similar to that of 9a /
10a and one that migrates slower than 8b and in a
manner similar to that of 11. The former product
presumably arises from hydrolysis of the 2′-S,3′-O-cyclic
phosphorothiolate, which is expected to yield exclusively
the 2′-S-phosphorothiolate monoester 10b.20 The latter
The thiophosphoramidite was then used to install a 2′-
S-phosphorothiolate linkage within a DNA tetramer
(5′TC2′SPTU3′). We chose to incorporate 2′-deoxyribose
residues at positions flanking the modified linkage to
prevent cleavage at these sites during exposure of the
oligonucleotide to acid or base. As described previously
for the installation of 3′-thionucleosides into oligonucle-
otides, the thiophosphoramidite 6 was coupled manually
using p-nitrophenyltetrazole as an activator.15 As a
control oligonucleotide, we synthesized a tetramer con-
taining a 2′-O-phosphodiester linkage (5′TC2′OPTU3′) using
4-N-benzoyl-5′-O-(dimethoxytrityl)-3′-O-(tert-butyldi-
methylsilyl)cytidin-2′-yl â-cyanoethyl N,N-diisopropyl-
phosphoramidite. Both oligonucleotides were deprotected
using standard conditions: treatment with NH4OH:
EtOH (3:1) for 16 h at 55 °C to remove the phosphate
and heterocyclic protecting groups and treatment with
tetrabutylammonium fluoride (TBAF; 1 M in THF) for
36 h at room temperature to remove the 3′-silyl protecting
group.
The tetramers were incubated with polynucleotide
kinase and [γ-32P]-ATP to generate the corresponding 5′-
[32P]-oligonucleotides 7a and 7b, which were purified by
polyacrylamide gel electrophoresis (PAGE). Because the
2′-S-phosphorothiolate linkage is expected to be suscep-
tible to nucleophilic attack by the adjacent 3′-hydroxyl
group, we eluted 7a and 7b from the electrophoresis gel
into a low pH buffer (pH 5.5) at 4 °C and carried out
subsequent workup procedures at pH 5.5.
The presence of a bridging sulfur in tetramer 7b was
confirmed by treatment with Ag+, which specifically
cleaves bridging phosphorothiolates but leaves normal
phosphodiester linkages intact.16 As expected, silver
treatment did not affect 7a (Figure 1, lane 1) but
quantitatively converted 7b to a faster migrating species,
presumably the 2′-mercapto derivative 11 (Figure 1, lane
(17) Abrash, H. I.; Cheung, C.-C. S.; Davis, J . C. Biochemistry 1967,
6, 1298.
(18) Usher, D. A.; McHale, A. H. Proc. Natl. Acad. Sci. U.S.A. 1976,
73, 1149.
(19) The 2′- and 3′-monophosphates commonly migrate together in
PAGE.
(16) Cosstick, R.; Vyle, J . S. Nucl. Acids Res. 1990, 18, 829.

5702 J . Org. Chem., Vol. 64, No. 15, 1999
Notes
Sch em e 2
F igu r e 2. Biochemical characterization of the 2′-S,3′-O-cyclic
phosphorothiolate 8b. Standards were generated either by
incubation of 7b with 20 mM AgNO3 in the dark for >20 min
(lane 1) followed by treatment with 20 mM iodoacetamide for
>20 min at 30 °C (lane 2), or by alkaline treatment (A) with
0.01 M NaOH at 30 °C for 15 min (lane 3). 8b was incubated
with either 0.1 M HCl (lanes 7-9) or 0.1 M NaOH (lanes 10-
12) at 50 °C for 15 min before neutralization with NaOH or
HCl, respectively. These reactions, as well as untreated 8b
(lanes 4-6), were divided into thirds. One-third was loaded
directly on the gel (lanes 4, 7 and 9), one-third was reacted
with AgNO3 (lanes 5, 8, and 11), and one-third was reacted
with iodoacetamide (lanes 6, 8, and 12) as described above.
ucts were compared to standards generated by Ag+
cleavage (11; Figure 2, lane 1) and base cleavage (8b,
10b, 11; lane 3) of tetramer 7b. 11 can be further reacted
with iodoacetamide to a slower migrating species (12;
Figure 2, lane 2). The much faster mobility of 11 relative
to 12 (compare Figure 2, lanes 1 and 2) may be because
11 harbors an additional negative charge due to ioniza-
tion of the mercapto group under the electrophoresis
conditions.14
To test for the presence of a phosphorothiolate linkage
in the putative 2′-S,3′-O-cyclic phosphorothiolate 8b
(Figure 2, lane 4), we treated it with Ag+. As expected,
Ag+ converted 8b completely to a faster migrating
product (Figure 2, lane 5). This product migrates more
slowly than the 2′-S-monophosphorothiolate 10b (Figure
2, lane 3) and is converted to a slower migrating species
by iodoacetamide (see below; lane 9), consistent with the
structure 9b. Treatment of 8b with acid (0.1 M HCl, 50
°C, 15 min) results in two new products (Figure 2, lane
7), one of which comigrates with 9b from silver cleavage.
Neither of these two products is affected by Ag+ (Figure
2, lane 8), but both are converted to slower migrating
species by iodoacetamide (lane 9), indicating the presence
of a free mercapto group and suggesting that ring opening
at the cyclic phosphorothiolate under acidic conditions
occurs preferentially through scission of the S-P bond.22
One of the iodoacetamide-modified products comigrates
with 12, suggesting that acid treatment also gives rise
to the dephosphorylated species 11.
product may arise from dephosphorylation of 10b to give
11. Consistent with these assignments, when 15 min
aliquots of the cleavage reactions (Figure 1, lanes 6 and
15) were treated with Ag+ (lanes 7 and 16), only those
species that are expected to contain a sulfur-phosphorus
(S-P) bond (7b, 8b, 10b; Scheme 2) disappear, whereas
those species lacking an S-P bond (7a , 8a , 9a /10a , 11)
are unaffected.
Following base-catalyzed cleavage of 7b (0.01 M NaOH
at 30 °C; Figure 1, lane 13) for 4 min, we purified the
2′-S,3′-O-cyclic phosphorothiolate 8b by DTT-PAGE, and
in a manner similar to that of 7b, maintained the pH at
5.5 during elution from the gel and subsequent workup
steps. To obtain further evidence for our structural
assignments and the aforementioned reaction pathway,
we analyzed the reactivity of 8b under acidic and basic
conditions and probed the susceptibility of the resulting
products and 8b itself to modification by Ag+ and
iodoacetamide. The latter reagent converts mercaptans
to acetamide thioethers but does not modify alcohols.21
Figure 2 and Scheme 2 show the results of this biochemi-
cal analysis. The electrophoretic mobilities of the prod-
Finally, we studied the reactivity of 8b under basic
conditions (Figure 2, lane 10). Upon exposure to 0.1 M
(22) It is known that hydrolytic ring opening of inosine 2′-O,3′-S-
cyclic phosphorothiolate under acidic conditions also cleaves the 3′-
S-P bond.14 The observation that one of the acid cleavage products of
8b (Figure 2, lane 7) comigrates with the silver cleavage product (lane
5) suggests that acidic conditions also cleave the 2′-S-P bond of the
2′-S,3′-O-cyclic phosphorothiolate. The reasons for the different regi-
oselectivities resulting from base and acid cleavage are discussed in
refs 12 and 14 and references therein.
(20) An NMR study showed that 2′-S,3′-O-uridine cyclic phospho-
rothiolate yields exclusively the 2′-S-phosphorothiolate under basic
conditions.12
(21) Coleman, R. S.; Kesicki, E. A. J . Am. Chem. Soc. 1994, 116,
11636.

Notes
J . Org. Chem., Vol. 64, No. 15, 1999 5703
8.37 (d, 1H), 8.26 (d, 1H), 8.02 (d, 2H), 8.02 (d, 2H), 7.74 (t, 1H),
7.64 (t, 2H), 7.54 (t, 2H), 7.23 (d, 1H), 7.18 (t, 1H), 6.50 (d, 1H),
6.23 (d, 1H), 5.15 (s, 1H), 4.40 (s, 1H), 4.11 (s, 1H), 3.85 (m, 1H),
3.58 (m, 2H). 13C NMR (DMSO-d6) δ: 168.1, 164.0, 159.5, 155.8,
150.3, 146.2, 138.6, 133.9, 133.7, 129.4, 122.1, 120.1, 97.6, 89.5,
87.7, 73.0, 62.0, 61.8. HRMS (MH+) calcd 473.0953, found
473.0936.
4-N -Be n zoyl-5′-O-(d im e t h oxyt r it yl)-2′-d e oxy-2′-(2-p y-
r id yld ith io)cytid in e (3). The disulfide 2 (140 mg; 0.3 mmol)
was dissolved in dry pyridine (4 mL). Dimethoxytrityl chloride
(203 mg; 0.6 mmol) and (dimethylamino)pyridine (18 mg; 0.15
mmol) were added, and the solution was stirred for 6 h. The
solution was concentrated and purified by column chromatog-
raphy (eluting with a gradient of 0-1% MeOH in CHCl3) to
afford 3 as a yellow foam (195 mg; 85% yield). 1H NMR (CDCl3)
δ: 8.71 (s, 1H), 8.55 (d, 1H) 8.22 (s, 1H), 7.94 (s, 2H), 7.77 (t,
1H), 7.64 (t, 2H), 7.55 (t, 2H), 7.4 (m, 9H), 7.25 (d, 1H), 6.86
(dd, 4H), 6.42 (d, 1H), 4.47 (s, 1H), 4.34 (d, 1H), 3.83 (s, 6H),
3.78 (dd, 1H), 3.55 (m, 2H). 13C NMR (CDCl3) δ: 159.1, 158.1,
150.1, 144.5, 137.7, 135.8, 135.5, 133.6, 130.6, 130.5, 129.6, 129.5,
128.6, 128.5, 128.3, 128.2, 128.1, 127.6, 122.8, 122.3, 113.7, 113.6,
87.6, 85.3, 72.2, 64.0, 63.7, 55.7. HRMS (MNa+) calcd 797.2080,
found 797.2124.
4-N-Ben zoyl-5′-O-(d im et h oxyt r it yl)-3′-O-(ter t-b u t yld i-
m eth ylsilyl)-2′-d eoxy-2′-(2-p yr id yld ith io)cytid in e (4). Nu-
cleoside 3 (195 mg; 0.25 mmol) was dissolved in dry pyridine (3
mL) followed by addition of imidazole (858 mg; 12.6 mmol) and
tert-butyldimethylsilyl chloride (1.5 g; 10 mmol). The solution
turned clear immediately, and after a few minutes of stirring, a
white solid appeared. After stirring for 24 h, the reaction mixture
was concentrated and purified by column chromatography
(eluting with a gradient of 0-1% MeOH in CHCl3) to afford
nucleoside 4 (190 mg; 86% yield) as a white foam. 1H NMR
(CDCl3) δ: 8.45 (d, 1H), 7.64 (m, 6H), 7.39 (d, 2H), 7.3 (m, 10H),
7.21 (d, 2H), 7.02 (m, 1H), 6.85 (dd, 4H), 6.58 (d, 1H), 4.61 (d,
1H), 4.16 (s, 1H), 3.94 (dd, 1H), 3.84 (s, 6H), 3.52 (dd, 2H), 0.94
(s, 9H), 0.13 (s, 3H), 0.04 (s, 3H). 13C NMR (CDCl3) δ: 137.8,
135.5, 130.6, 130.3, 129.7, 128.5, 128.3, 128.1, 128.0, 127.4, 126.2,
124.8, 120.2, 116.8, 116.1, 113.6, 113.4, 55.8, 54.5, 50.3, 26.0,
0.3, -4.7. HRMS (MNa+) calcd 911.2944, found 911.2924.
4-N-Ben zoyl-5′-O-(d im et h oxyt r it yl)-3′-O-(ter t-b u t yld i-
m eth ylsilyl)-2′-d eoxy-2′-m er ca p tocytid in e (5). Nucleoside 4
(190 mg; 0.21 mmol) was dissolved in anhydrous, argon-
saturated THF (6 mL); anhydrous DTT (129 mg; 0.84 mmol) and
triethylamine (100 µL) were added. The reaction was stirred for
3 h, concentrated, and purified by column chromatography
(eluting with a gradient of 1-2% MeOH in CHCl3) to afford
nucleoside 5 (155 mg; 95% yield) as a white foam. 1H NMR
(CDCl3) δ: 8.43 (d, 1H), 8.07 (d, 1H), 7.89, (d, 2H), 7.61 (t, 1H),
7.53 (t, 2H), 7.42 (d, 2H), 7.36 (m, 7H), 6.90 (d, 4H), 6.27 (d,
1H), 4.27 (s, 1H), 3.81 (s, 6H), 3.71 (m, 2H), 3.60 (m, 1H), 3.41
(d, 1H), 0.86 (s, 9H), 0.02 (s, 3H), -0.03 (s, 3H). 13C NMR (CDCl3)
δ: 159.2, 145.1, 144.3, 135.6, 135.5, 133.6, 130.6, 129.5, 128.7,
128.5, 127.9, 127.7, 113.8, 97.0, 93.1, 87.6, 85.0, 71.6, 62.1, 55.7,
49.2, 30.1, 26.1, 18.5, 1.4, 0.4, -4.3, -4.5. HRMS (MNa+) calcd
802.2958, found 802.2977.
NaOH for 15 min, 8b disappeared completely, forming
products that comigrate with the 2′-S-phosphorothiolate
10b and the dephosphorylated product 11. The response
of the products to Ag+ (Figure 2, lane 11) or iodoaceta-
mide (lane 12) treatment is consistent with these assign-
ments: following Ag+ treatment 10b disappears, pre-
sumably being converted to 11, which consequently
increases in relative intensity compared to that of the
untreated lanes. Iodoacetamide treatment provides a new
band that comigrates with 12, presumably arising from
modification of 11, while 10b remains unaffected. These
data strongly argue that 8b contains a 2′-S,3′-O-cyclic
phosphorothiolate terminus, imply that no modifications
to the sulfur occurs during synthesis, isolation, or puri-
fication of the oligonucleotide, and provide a biochemical
“signature” for a cyclic phosphorothiolate terminus.
Taken together, our results show that the 2′-S-phos-
phorothiolate linkage is correctly installed into oligo-
nucleotides via phosphoramidite chemistry and is a
suitable precursor for generating oligonucleotides con-
taining 2′-S,3′-O-cyclic phosphorothiolate termini. Al-
though this study establishes the method using a short
DNA oligonucleotide, the approach is readily adapted to
longer oligonucleotides (data not shown). In cases where
PAGE does not readily separate the cyclic phosphorothi-
olate from its ring-opened hydrolysis product, exposure
of the 2′-S-phosphorothiolate precursor to base-catalyzed
cleavage conditions (0.01 M NaOH, 30 °C) must be
limited to 4 min so as to minimize hydrolysis. Because
standard ribonucleotide linkages undergo minimal base-
catalyzed cleavage under these conditions (data not
shown), it should be possible to generate RNA oligonucle-
otides containing 2′-S,3′-O-cyclic phosphorothiolate ter-
mini. The methods described herein provide access to new
probes with which to explore the biological structure and
function of macromolecules that interact with 2′,3′-cyclic
phosphates.
Exp er im en ta l Section
Gen er a l Meth od s. All reactions were preformed at room
temperature unless otherwise indicated. All reagents were from
Aldrich, and all solvents were from Fisher unless otherwise
indicated. Solutions of DTT (IDL) were stored in the dark and
discarded after 2 months of use. Merck silica gel, 8983 grade,
230-400 mesh, 60 Å, was used for column chromatography.
Oligonucleotides were synthesized on a 1 µmol scale using a
Millipore Expidite Nucleic Acid Synthesis System and standard
DNA and RNA protocols and phosphoramidites (Glenn) unless
otherwise indicated. Reverse phase high-pressure liquid chro-
matography (HPLC) was performed using a C18 column (10 ×
4-N-Ben zoyl-5′-O-(d im et h oxyt r it yl)-3′-O-(ter t-b u t yld i-
m eth ylsilyl)-2′-deoxy-2′-th iocytidin -2′-yl â-Cyan oeth yl N,N-
Diisop r op ylp h osp h or a m id ite (6). 2′-Mercapto nucleoside 5
(70 mg; 0.9 mmol) was added to dry CH2Cl2 (3 mL). Diisopro-
pylethylamine (78 µL; 0.45 mmol), redistilled methylimidazole
(3.5 µL; 0.045 mmol), and â-cyanoethyl N,N-diisopropylchloro-
phosphoramidite (40 µL; 0.18 mmol) were added, and the
reaction was stirred for 1 h under argon. The solution was
concentrated and purified by column chromatography (eluting
with a gradient of 2-6% acetone in CH2Cl2; both solvents were
filtered through basic alumina before mixing) to afford 6 (80 mg;
90% yield) as a white foam.31P NMR (CDCl3) δ: 166.17, 161.81.
HRMS (MNa+) calcd 1002.4037, found 1002.4065.
Syn th esis of Oligon u cleotid es. The oligonucleotide con-
taining the 2′-S-phosphodiester linkage (5′TC2′SPTU3′) was syn-
thesized using the 2′-S-phosphoramidite 6, which was coupled
in the presence of p-nitrophenyl tetrazole as described by Sun
et al. for 3′-S-phosphoramidites.15 The oligonucleotide containing
the 2′-O phosphodiester linkage was synthesized using N-ben-
zoyl-5′-O-(dimethoxytrityl)-3′-O-(tert-butyldimethylsilyl)-cytidin-
2′-yl â-cyanoethyl N,N-diisopropylphosphoramidite (Chem-
1
250 mm). H, 31P, and 13C spectra were obtained on a 500 MHz
NMR spectrometer. Mass spectral data were obtained from the
Mass Spectral Analysis Lab at the University of California,
Riverside, on a high-resolution mass spectrometer. PAGE was
performed using 20% polyacrylamide (Fisher; acrylamide: bis-
acrylamide 29:1) with 7 M urea. Gel loading solution contained
8 M urea (VWR), 50 mM EDTA (Fisher), 0.02% bromophenol
blue (EM Science), and 0.02% xylene cyanol FF (Kodak). DTT-
PAGE included 10 mM DTT in both the gel and the gel running
buffer. Gels were pre-electrophoresed for at least 5 h before use.
4-N-Ben zoyl-2′-deoxy-2′-(2-pyr idyldith io)cytidin e (2). The
benzoyl-protected cytidine derivative 19 (167 mg; 0.28 mmol) was
dissolved in chloroform (6 mL) and MeOH (3 mL). AgNO3 (51
mg; 0.3 mmol) and Aldrithiol-2 (132 mg; 0.6 mmol) were then
added, and after a few minutes of stirring, a white precipitate
formed. Stirring was continued for 2 h, and the reaction mixture
was concentrated and purified by column chromatography
(eluting with a gradient of 0-2% MeOH in CHCl3) to afford 2
1
as a white powder (110 mg; 83% yield). H NMR (DMSO-d6) δ:

5704 J . Org. Chem., Vol. 64, No. 15, 1999
Notes
Genes). Following synthesis, the oligonucleotides were deprotected
with concentrated NH4OH:EtOH (3:1, 2 mL) at 55 °C for 16 h.
Each solution was concentrated and then shaken in 0.1 M TBAF
in THF (0.5 mL) in the dark for 36 h. Then 0.1 M TEAAC (0.4
mL) was added, and the solutions were spun under vacuum for
3 h. The oligonucleotides were passed through a Sephadex G-25
column, followed by purification by reverse phase HPLC. The
HPLC column was heated at 40 °C to prevent blockage by the
thick TBAF solution.
Ra d iola belin g of Oligon u cleotid es. Oligonucleotides (20
pmol) were 5′- radiolabeled with [R-32P]ATP (NEN) and T4
Polynucleotide Kinase (USB) according to the manufacturer’s
(USB) protocol (10 µL scale). The reactions were quenched with
gel loading buffer (10 µL), and the oligonucleotides were
separated from failure sequences by PAGE and visualized by
autoradiography. The radiolabeled tetramers were excised and
eluted from the gel at 4 °C with triethylammoniumacetate
(TEAAC) buffer (pH 5.5) for 8-12 h followed by desalting on a
Sep-pak C18 cartridge (Waters). Before concentrating the frac-
tions containing the oligonucleotides, TEAAC was added to a
concentration of 1 mM to maintain a low pH. After drying, the
oligonucleotides were dissolved in 1 mM TEAAC to a concentra-
tion of 20 nM.
for over 20 min. Treatment with 20 mM iodoacetamide, a well-
known mercaptan modification reagent, at 30 °C for over 20 min
was used to assay for free mercapto groups.
Isola tion of th e 2′-S,3′-O-Cyclic P h osp h or oth iola te (8b).
7b (∼5 pmol; essentially 1/2 or 1/4 of the amount obtained from
the kinase reaction described above) was incubated with 0.01
M NaOH for 4 min at 30 °C in a 50 µL reaction. Then 0.1 M
HCl (5 µL) was added to neutralize the base, and the mixture
was dried and redissolved in water (10 µL). Gel loading buffer
(10 µL) was added, and the mixture was purified by DTT-PAGE.
The product was visualized by autoradiography and excised and
eluted from the gel with TEAAC (pH 5.5). The oligonucleotide
was purified by Sep-Pak C18 cartridge as described above. Before
drying the fractions containing the oligonucleotide, TEAAC
buffer (pH 5.5) was added to a final concentration of 1 mM. After
drying, the oligonucleotide was dissolved in 1 mM TEAAC (pH
5.5) to a concentration of 20 nM.
Ack n ow led gm en t. The authors thank Chris Barth
for helpful discussion and Qing Dai, Peter Gordon, and
J ason Schwans for critical reading of the manuscript.
J .A.P. is an assistant investigator of the Howard Hughes
Medical Institute.
Ch em ica l Clea va ge Rea ction s. Complete alkaline or acid
hydrolysis of the oligonucleotides was carried out at 50 °C for
10 or 15 min with 0.1 M NaOH or 0.1 M HCl, respectively, before
neutralizing with HCl or NaOH, respectively. Partial alkaline
cleavage of 7a and 7b was achieved by incubation with 0.01 M
NaOH at 90 °C or 30 °C, respectively, for various times (Figure
1), before quenching with HCl. Specific cleavage of the S-P bond
was attained through incubation with 20 mM AgNO3 in the dark
Su p p or tin g In for m a tion Ava ila ble: 1H NMR spectra of
compounds 2-5 and 31P NMR spectrum of compound 6. This
material is available free of charge via the Internet at
http://pubs.acs.org.
J O9903508