Synthesis of phosphorodithioate DNA via sulfur-linked, base-labile protecting groups

Article abstract of DOI:10.1021/jo960274y

Phosphorodithioate DNA, a new and potentially useful DNA analog with a deoxynucleoside-OPS₂O-deoxynucleoside internucleotide linkage, was synthesized from deoxynucleoside 3'-phosphorothioamidites having a variety of thioesters and thiocarbonates as base-labile phosphorus protecting groups. The major challenge in the synthesis of this DNA analog was to derive a reaction pathway whereby activation of deoxynucleoside 3'-phosphorothioamidites occurred rapidly and in high yield under conditions that minimize Arbuzov rearrangements, exchange reactions, unwanted oxidation to phosphorothioates, and several other side reactions. Of the various phosphorus protecting groups examined for this purpose, a thorough evaluation of these parameters led to the conclusion that β-(benzoylmercapto)ethyl was preferred. Synthesis of phosphorodithioate DNA began by preparing deoxynucleoside 3'-phosphorothioamidites from the appropriately protected deoxynucleoside, tris(pyrrolidino)phosphine, and ethanedithiol monobenzoate via a one-flask synthesis procedure. These synthons were activated with tetrazole and condensed with a deoxynucleoside on a polymer support to yield the deoxynucleoside thiophosphite. Subsequent steps involved oxidation with sulfur to generate the completely protected phosphorodithioate triester, acylation of unreacted deoxynucleoside, and removal of the 5'-protecting group. Yields per cycle were usually 97-98% with 2-5% phosphorothioate contamination as determined by ³¹P NMR. By using deoxynucleoside 3'-phosphorothioamidites and deoxynucleoside 3'-phosphoroamidites, deoxyoligonucleotides having phosphorodithioate and the natural phosphate internucleotide linkages in any predetermined order can also be synthesized.

Full text of DOI:10.1021/jo960274y

4272
J . Org. Chem. 1996, 61, 4272-4281
Syn th esis of P h osp h or od ith ioa te DNA via Su lfu r -Lin k ed ,
Ba se-La bile P r otectin g Gr ou p s1
William T. Wiesler† and Marvin H. Caruthers*
Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-0215
Received February 9, 1996X
Phosphorodithioate DNA, a new and potentially useful DNA analog with a deoxynucleoside-OPS2O-
deoxynucleoside internucleotide linkage, was synthesized from deoxynucleoside 3′-phosphorothio-
amidites having a variety of thioesters and thiocarbonates as base-labile phosphorus protecting
groups. The major challenge in the synthesis of this DNA analog was to derive a reaction pathway
whereby activation of deoxynucleoside 3′-phosphorothioamidites occurred rapidly and in high yield
under conditions that minimize Arbuzov rearrangements, exchange reactions, unwanted oxidation
to phosphorothioates, and several other side reactions. Of the various phosphorus protecting groups
examined for this purpose, a thorough evaluation of these parameters led to the conclusion that
â-(benzoylmercapto)ethyl was preferred. Synthesis of phosphorodithioate DNA began by preparing
deoxynucleoside 3′-phosphorothioamidites from the appropriately protected deoxynucleoside, tris-
(pyrrolidino)phosphine, and ethanedithiol monobenzoate via a one-flask synthesis procedure. These
synthons were activated with tetrazole and condensed with a deoxynucleoside on a polymer support
to yield the deoxynucleoside thiophosphite. Subsequent steps involved oxidation with sulfur to
generate the completely protected phosphorodithioate triester, acylation of unreacted deoxynucleo-
side, and removal of the 5′-protecting group. Yields per cycle were usually 97-98% with 2-5%
phosphorothioate contamination as determined by 31P NMR. By using deoxynucleoside 3′-
phosphorothioamidites and deoxynucleoside 3′-phosphoroamidites, deoxyoligonucleotides having
phosphorodithioate and the natural phosphate internucleotide linkages in any predetermined order
can also be synthesized.
In tr od u ction
possible solution to this problem,8,9 another lies in the
synthesis of modifications which are achiral at phospho-
rus.
Deoxyoligonucleotide analogs bearing modified phos-
phodiester linkages have been the focus of considerable
interest in the antisense field because of their nuclease
resistance and thus prolonged biological half-lives.2-7
Methyl phosphonates and phosphorothioates are the
most extensively studied variations to date. The substi-
tution of a single nonbridging oxygen atom with either a
methyl group or a sulfur atom renders the internucleotide
linkage nuclease resistant, but unlike natural DNA, the
modified phosphorus center is chiral which leads to a
mixture of unresolvable diastereomeric oligomers having
variable biochemical, biophysical, and biological proper-
ties. While stereocontrolled synthesis of P-chiral phos-
phorothioates or methylphosphonates represents one
The substitution of both nonbridging oxygen atoms
with sulfur gives rise to a phosphorodithioate linkage,
which, like natural DNA, is achiral at phosphorus. This
linkage has been found to be highly stable toward both
chemical and enzymatic hydrolysis.10 Recent studies
have found that phosphorodithioate DNA oligomers have
biophysical and biological properties which make them
ideally suited for antisense applications.11,12 For ex-
ample, of the many analogs synthesized to date, only two,
the phosphorodithioate and phosphorothioate derivatives,
have been shown to activate RNase H which appears to
be crucial for demonstrating antisense activity in vivo.7b
Phosphorodithioate deoxyoligonucleotides have also been
found to be potent inhibitors of HIV-1 reverse tran-
† Present address: Cadus Pharmaceutical Corporation, 777 Old Saw
Mill River Road., Tarrytown, NY 10591.
(8) Stec, W. J .; Grajkowski, A.; Koziolkiewicz, M.; Uznanski, B.
Nucleic Acids Res. 1991, 19, 5883-5888.
(9) Stec, W. J .; Lesnikowski, Z. J . in Methods in Molecular Biology,
Protocols for Oligonucleotides and Their Analogs, Synthesis and
Properties; Agarwal, S., Ed.; Humana Press: Clifton, NJ , 1993; pp
285-313.
(10) (a) Nielsen, J .; Brill, W. K.-D.; Caruthers, M. H. Tetrahedron
Lett. 1988, 29, 2911-2914. (b) Grandas, A.; Marshall, W. S.; Nielsen,
J .; Caruthers, M. H. Tetrahedron Lett. 1989, 30, 543-546. (c) Porritt,
G. M.; Reese, C. B. Tetrahedron Lett. 1989, 30, 4713-4716. (d)
Okruszek, A.; Sierzchala, A.; Feason, K. L.; Stec, W. J . J . Org. Chem.
1995, 60, 6998-7005.
(11) (a) Caruthers, M. H.; Beaton, G.; Cummins, L.; Dellinger, D.;
Graff, D. Ma, Y.-X.; Marshall, W. S.; Sasmor, H.; Shankland, P.; Wu,
J . V.; Yau, E. K. Nucleosides Nucleotides 1991, 10, 47-59. (b) Marshall,
W. S.; Beaton, G.; Stein, C. A.; Matsukura, M.; Caruthers, M. H. Proc.
Nat. Acad. Sci. U.S.., 1992, 89, 6265-6269. (c) Marshall, W. S.;
Caruthers, M. H. Science 1993, 259, 1564-1570.
X Abstract published in Advance ACS Abstracts, J une 15, 1996.
(1) This is paper 42 in a series on Nucleotide Chemistry. For paper
41, see: Fischer, R. W.; Caruthers, M. Tetrahedron Lett. 1995, 36, 6807.
This research was supported by the National Institutes of Health
(Grant GM25680) and a Postdoctoral Fellowship from the American
Cancer Society (PF-3465) to W.T.W. Abbreviations: DMT, 4,4′-
dimethoxytrityl; CPG, controlled pore glass; T, thymine; CBz, 4-N-
benzoylcytosine; C-Isob, 4-N-isobutyrylcytosine; ABz, 6-N-benzoyl-
adenine; G-Isob, 2-N-isobutyrylguanine; G-Pac, 2-N-phenylacetylguanine;
Ph, phenyl; Ac, acetyl; Prop, propyl; PAGE, polyacrylamide gel
electrophoresis; TCA, trichloroacetic acid; Nuc, 2′-deoxynucleoside;
B(prot), N-protected 2′-deoxynucleoside pyrimidine or purine base;
FMOC, 9-fluorenylmethyloxycarbonyl; DMAP, 4-(N,N-dimethylamino)-
pyridine; TEA, triethylamine.
(2) Uhlmann, E.; Peyman, A. Chem. Rev. 1990, 90, 543-584.
(3) Stein, C. A.; Tonkinson, J . L.; Yakubov, L. Pharmacol. Ther.
1991, 52, 365-371.
(4) Agrawal, S. TIBTECH 1992, 10, 152-155.
(12) (a) Dahl, B. H.; Bjergårde, K.; Sommer, V. B.; Dahl, O. Acta
Chem. Scand. 1989, 43, 896-901. (b) Bjergårde, K. Thesis, University
of Copenhagen, 1990. (c) Dahl, B. H.; Bjergårde, K.; Nielsen, J .; Dahl,
O. Tetrahedron Lett. 1990, 31, 3489-3492. (d) Bjergårde, K.; Dahl, O.
Nucleic Acids Res. 1991, 19, 5843-5850. (e) Bjergårde, K. Oligonucleo-
side Phosphorodithioates, Cand. Scient. Thesis, Kobenhauns Univer-
sitet, 1990, pp 61-63.
(5) Goodchild, J . Bioconjugate Chem. 1990, 2, 165-186.
(6) Cohen, J . S., Ed. Oligonucleotides: Antisense Inhibitors of Gene
Expression; Topics in Molecular and Structural Biology; MacMillan
Press: London, 1989; Vol. 12.
(7) (a) Stein, C. A.; Cohen, J . S. Cancer Res. 1988, 48, 2659-2668.
(b) Stein, C. A.; Cheng, Y.-C. Science 1993, 261, 1004-1012.
S0022-3263(96)00274-5 CCC: $12.00 © 1996 American Chemical Society

Synthesis of Phosphorodithioate DNA
J . Org. Chem., Vol. 61, No. 13, 1996 4273
scriptase and of HIV-1 viral replication and thus repre-
sent an additional class of potential therapeutic agents
against acquired immunodeficiency syndrome.11b,c
synthons 1b containing this group are relatively unstable
to oxidation which necessitates changing solutions of 1b
on the DNA synthesizer every few hours.
Since the initial reports on the solution10a and solid-
phase13 synthesis of phosphorodithioate DNA, there have
been numerous synthetic efforts directed toward the
development of this chemistry.10b-d,12,14 The solid-phase
synthesis of phosphorodithioate DNA oligomers is similar
to conventional methods for synthesizing deoxyoligo-
nucleotides,15 but there are several essential modifica-
tions since two sulfur atoms are introduced into the
phosphate linkage. The first sulfur is incorporated
through phosphorothioamidite synthon 1 and the second
via sulfur oxidation to give phosphorodithiotriester 2.
Here we describe investigations with deoxynucleoside
3′-phosphorothioamidites that have a series of sulfur pro-
tecting groups. These studies include an analysis of reac-
tion pathways and a characterization of side products
resulting from the synthesis of the deoxynucleoside 3′-
phosphorothioamidites, sulfurization of protected thio-
phosphite triesters, and deprotection of dithioate oligo-
mers. These studies led to the development of a novel
base-labile â-(benzoylmercapto)ethyl protecting group, its
incorporation into deoxynucleoside 3′-phosphorothioam-
idites (1c), and the use of this synthon in the preparation
of phosphorodithioate DNA. This protecting group com-
bines the best features of the two previously employed
for phosphorodithioate synthesis as it has a stability
toward oxidation similar to phosphorothioamidites con-
taining a â-cyanoethyl group and its removal can be con-
veniently carried out during a simple base-deprotection
treatment to give phosphorodithioate deoxyoligonucle-
otides with relatively low levels of phosphorothioate
contamination.
Resu lts a n d Discu ssion
Mod el Stu d ies. Previous studies had shown that
protection of sulfur with the 2,4-dichlorobenzyl group
gave rise to lower levels of phosphorothioate impurity
than those attained using â-cyanoethyl protection.14f
While a variety of mechanisms have been proposed to
account for the generation of the phosphorothioate
impurity,12d,e selectivity of deprotection would appear to
be an important factor. In particular, nucleophilic dis-
placement with thiophenolate of the 2,4-dichlorobenzyl
protecting group should be a more selective process than
removal of the â-cyanoethyl group with ammonium
hydroxide.
Upon completion of automated synthesis, the protecting
groups on sulfur and the deoxynucleoside bases are
removed and the phosphorodithioate oligomers are de-
tached from the support to yield 3. The successful
synthesis of this analog requires a thorough investigation
of various deoxynucleoside phosphorothioamidite syn-
thons, sulfurization parameters, and the conditions as-
sociated with removal of protecting groups.
As a consequence of these considerations, a protecting
group for sulfur was sought which could be eliminated
via intramolecular nucleophilic displacement. Use of the
â-mercaptoethyl group in deoxyoligonucleotide synthesis
has been previously reported. In these studies, the thiol
is masked as either a trityl thioether17 or a disulfide.18,19
Release of the free thiol and treatment with base results
in elimination of ethylene sulfide to give the desired
phosphate product. We adapted this “protected” protect-
ing group for blocking sulfur in phosphorodithioates. The
strategy involved masking the â-mercaptoethyl group
with a variety of thioesters or thiocarbonates that could
be readily eliminated under ammonia conditions.
Protection of phosphorodithioates with â-(acylmercap-
to)ethyl groups was initially evaluated on O,O-diethyl
S-(â-(acylmercapto)ethyl) phosphorodithioate model com-
pounds 4a -d where the acyl groups were varied from
acetate (4a ) to pivaloate (4d ). These compounds were
easily obtained from two independent routes: alkylation
of O,O-diethyl dithiophosphoric acid (6) with esters of
bromoethanethiol or acylation of O,O-diethyl S-(â-mer-
captoethyl) phosphorodithioate (5). Deprotection under
basic conditions to the desired phosphorodithioate (6) was
Two different sulfur protecting groups, the â-cyano-
ethyl and 2,4-dichlorobenzyl, have been used in previous
approaches with good but limited success. The â-cyano-
ethyl group can be removed under the standard am-
monium hydroxide treatment that is used to deblock the
base protecting groups; however, high levels (8-9%) of
contaminating phosphoromonothioate are incorporated
into the deoxyoligonucleotide products.12 This contami-
nation is reduced to more acceptable levels (2-4%) when
the 2,4-dichlorobenzyl group is used to block sulfur.14g
Deprotection of the resulting phosphorodithiotriester
linkage 2b can be cleanly carried out by a post-synthesis
treatment with either thiophenolate or an odorless
substitute.16 Unfortunately, the phosphorothioamidite
(13) Brill, W. K.-D.; Tang, J .-Y.; Ma, Y.-X.; Caruthers, M. H. J . Am.
Chem. Soc. 1989, 111, 2321-2322.
(14) (a) Farschtschi, N.; Gorenstein, D. G. Tetrahedron Lett. 1988,
29, 6843-6846. (b) Brill, W. K.-D.; Yau, E. K.; Caruthers, M. H.
Tetrahedron Lett. 1989, 30, 6621-6624. (c) Stawinski, J .; Thelin, M.;
Zain, R. Tetrahedron Lett. 1989, 30, 2157-2160. (d) Yau, E. K.; Ma,
Y.-X.; Caruthers, M. H. Tetrahedron Lett. 1990, 31, 1953-1956. (e)
Porritt, G. M.; Reese, C. B. Tetrahedron Lett. 1990, 31, 1319-1322. (f)
Beaton, G.; Brill, W. K.-D.; Grandas, A.; Ma, Y.-X.; Nielsen, J .; Yau,
E.; Caruthers, M. H. Tetrahedron 1991, 47, 2377-2388. (g) Beaton,
G.; Dellinger, D.; Marshall, W. S.; Caruthers, M. H. In Oligonucleotides
and Analogues: A Practical Approach;, Eckstein, F., Ed.; IRL Press:
1991; pp 109-135. (h) Piotto, M. E.; Granger, J . N.; Cho, Y.; Farschts-
chi, N.; Gorenstein, D. G. Tetrahedron 1991, 47, 2449-2461. (i)
Caruthers, M. H. Beaton, G.; Wu, J . V.; Wiesler, W. Methods Enzymol.
1992, 211, 3-20.
(17) Conolly, B. A. Tetrahedron Lett. 1987, 28, 463-466.
(18) (a) Asseline, U.; Bonfils, E.; Kurfu¨rst, R.; Chassignol, M.; Roig,
V.; Thuong, N. T. Tetrahedron 1992, 48, 1233-1254. (b) Asseline, U.;
Thuong, N. T. Tetrahedron Lett. 1989, 30, 2521-2524.
(19) (a) Kumar, P.; Bose, N. K.; Gupta, K. C. Tetrahedron Lett. 1991,
32, 967-970. (b) Gupta, K. C.; Sharma, P.; Kumar, P.; Sathya-
narayana, S. Nucleic Acids Res. 1991, 19, 3019-3025.
(15) Caruthers, M. H. Science 1985, 230, 281-285.
(16) Dahl, B. H.; Bjergårde, K.; Henriksen, L.; Dahl, O. Acta Chem.
Scand. 1990, 44, 639-641.

4274 J . Org. Chem., Vol. 61, No. 13, 1996
Wiesler and Caruthers
thesis without further purification even though a number
confirmed by 31P NMR. The rates of deprotection were
dependent upon the lability of the acyl group as well as
the amine base used. A variety of primary and secondary
amines were examined in addition to ammonia with
pyrrolidine giving the fastest deprotection (seconds with
acetate 4a , minutes with benzoate 4b). While the acetate
and benzoate compounds were fully deprotected within
an hour regardless of the amine used, deprotection of the
pivaloyl derivative 4d required several hours. The model
compounds were found to be stable to tetrazole, iodine/
water oxidation, sulfur oxidation, and acetic anhydride
capping conditions which indicated that the acyl-protect-
ed â-mercaptoethyl group was compatible with DNA
synthesis protocols.
of penta- and trivalent phosphorus impurities were
detected at low levels. Two of these, the oxidation (30
ppm) and hydrolysis (13 ppm) products, each typically
at the 2-4% level, were not considered to adversely affect
deoxyoligonucleotide synthesis, as these pentavalent
phosphorus compounds were unreactive under coupling
conditions. However, the participation of trivalent phos-
phorus impurities in side reactions could negatively affect
the quality of deoxyoligonucleotide products. One of
these reactive impurities, the bis(pyrrolidino) phospho-
rodiamidite 8 (133 ppm), was initially observed at the
level of 1%, but was found to be eliminated by increasing
the reaction time.
Syn th esis of S-(â-(Acylm er ca p to)eth yl) P h osp h o-
r oth ioa m id ites. Preliminary studies focused on the
preparation of various thymidine phosphorothioamidites
for subsequent comparison in automated phosphorodithio-
ate oligothymidylate synthesis on a solid support. These
synthons, in which the sulfur was protected with various
â-(acylmercapto)ethyl groups, were prepared via a modi-
fied version of previously reported syntheses.12c,13 The
5′-dimethoxytrityl protected thymidine 7 was first phos-
phitylated with bis(pyrrolidinyl)chlorophosphine to give
thymidine bis(pyrrolidino) diamidite 8. In the same pot,
Three other trivalent phosphorus impurities were not
as easily minimized. Two of these were by-products from
the undesired deacylation of the â-mercaptoethyl group
during phosphorothioamidite synthesis. One equivalent
of pyrrolidine was released upon conversion of phospho-
rodiamidite 8 to phosphorothioamidite 10. Pyrrolidine
was found to be a very effective sulfur deprotection agent,
particularly with the more labile (acetyl and benzoyl)
thioesters.20 Thus during phosphorothioamidite synthe-
sis, the released pyrrolidine caused formation of deacy-
lated product 12, which upon intramolecular cyclization
gave 2-6% dithiophosphite byproduct 13 (149 ppm). To
a lesser extent, the deacylated intermediate 12 partici-
pated in an intermolecular condensation with the phos-
phorodiamidite 8 to yield the phosphorothioamidite dimer
14 (157.6, 157.9, 160.9 and 161.2 ppm). Reaction of the
phosphorodiamidite 8 with ethanedithiol gave these two
products exclusively (13 and 14), which confirmed that
this intermediate was converted to the desired thymidine
phosphorothioamidites 10 upon addition of ethanedithiol
monoacylates 9. Following an aqueous workup and
drying, the products were isolated by precipitation from
heptane to give stable powders. Purity of the phospho-
rothioamidites (159, 163 ppm) as analyzed by 31P NMR
was typically 85-90%.
Attempts to purify the phosphorothioamidites resulted
in degradation and rearrangement, most notably to the
Arbuzov-type product 11 (89 ppm). Thus the phospho-
rothioamidites were used in deoxyoligonucleotide syn-
(20) Previous work had shown that it was essential to retain in the
deoxynucleoside phosphorothioamidite an amine with properties simi-
lar to pyrrolidine if these synthons were to be reactive in DNA
synthesis.21

Synthesis of Phosphorodithioate DNA
J . Org. Chem., Vol. 61, No. 13, 1996 4275
they were indeed deprotection byproducts. While phos-
phorothioamidite dimer 14 was expected to result in
unwanted side reactions during deoxyoligonucleotide
synthesis, the major byproduct, dithiophosphite 13, was
found to be unreactive under standard coupling condi-
tions.
that an alternative phosphitylation procedure using tris-
(pyrrolidino)phosphine under tetrazole catalysis did not
generate this byproduct. While this approach offered a
Synthesis of phosphorothioamidites with carbonate
protection of the â-mercaptoethyl group was plagued by
formation of an additional deprotection byproduct. In the
presence of released pyrrolidine, phenyl thiocarbonates
9h and 9i degraded to give dithiocarbonate 15 and
phenols 16a ,b. The addition of tetrazole to the reaction
(to increase the rate and buffer released pyrolidine)
reduced but did not eliminate this problem. Condensa-
tion of phenols with phosphorodiamidite 8 gave rise to
phosphoroamidites 17 (140, 141 ppm). These byproducts
means of eliminating the dinucleoside phosphoroamidite
contamination, it gave rise to increased levels of depro-
tection byproducts 13 and 14 because large amounts of
pyrrolidine were generated. Not only were 2 equiv
formed during the reaction sequence but additional
quantities of pyrrolidine were released from reaction of
excess phosphine 19 and thiol to yield phosphine adducts
20 and 21. This adverse effect of pyrrolidine was
(17) were more reactive than phosphorothioamidites
under deoxyoligonucleotide coupling conditions and form
phosphorothioates upon oxidation of this internucleotide
phosphite triester with sulfur. Thus the quality of
phosphorodithioate deoxyoligonucleotides made from phos-
phorothioamidites having carbonate protection was ad-
versely affected due to the increase in phosphorothioate
contamination.
Phosphorothioamidites prepared via phosphitylation
with bis(pyrrolidinyl)chlorophosphine were routinely con-
taminated with 1% dithymidine phosphoroamidite 18
(142 ppm). It presumably formed as the phosphitylation
neared completion, at which point the triethylamine
hydrochloride salt catalyzed condensation of 8 with
unreacted starting material. The resulting dinucleoside
countered in two ways. First, the amount of tetrazole
present in the reaction mixture was increased to 3 equiv
upon addition of thiol, as it buffered the pyrrolidine
produced and increased the rate of reaction. Reducing
the excess phosphine (1.3 to 1.0 equiv) and thiol (2 to 1.3
equiv) resulted in a further decrease of the deprotection
byproducts. However, without using excess phosphine,
the dinucleoside phosphoroamidite contaminant 18 reap-
peared at levels of 1-2%, presumably due to underphos-
phitylation of deoxynucleoside. Unreacted deoxynucle-
oside subsequently condensed with bis(pyrrolidino)
phosphorodiamidite 8 upon addition of the large tetrazole
excess. This problem was solved by addition of 0.1 equiv
of (trimethylsilyl)imidazole to convert unphosphitylated
deoxynucleoside to its silyl ether. In this manner,
phosphorothioamidites 22-25 were prepared completely
free from contaminating dinucleoside phosphoroamidite
18 and contained very low levels (1-4%) of deprotection
byproducts 13 and 14. All four bases with a variety of
protecting groups were conveniently prepared in multi-
gram quantities by this approach.
phosphoroamidite was more reactive using tetrazole-
catalyzed coupling conditions than 10 and thus gave rise
to greater than 1% side reactions under deoxyoligonucle-
otide synthesis conditions where the thioamidite synthon
was used in excess. The resulting adducts introduced
branching points in deoxyoligonucleotides and yielded
unwanted higher molecular weight products (see below).
Efforts to prevent formation of the reactive contami-
nant 18 by varying base, solvent, temperature, and order
of addition were not successful. However, it was found
Deoxyoligon u cleotid e Syn th esis. The synthesis of
phosphorodithioate containing deoxyoligonucleotides from
phosphorothioamidite synthons 1, 10, and 22-25 fol-
lowed previously reported procedures.14i After conven-
tional detritylation of a deoxynucleoside linked to CPG,
the phosphorothioamidites were coupled to the support-
linked, growing deoxyoligonucleotide via conventional

4276 J . Org. Chem., Vol. 61, No. 13, 1996
Wiesler and Caruthers
Ta ble 1. Syn th esis of Oligoth ym id yla te
P h osp h or od ith ioa tesa
deoxyoligo- phosphoro-
entry thymidylate thioamidite
phosphoro-
thioate, %
deprotectionb
NH4OH
pyrrolidine
NH4OH
NH4OH
pyrrolidine/NH4OH
NH4OH
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7
T(sT)7(pT)7
T(pT)7(sT)7
T(pT)7(sT)7
T(pT)7(sT)7
10a
10b
10c
10d
10d
10e
10e
10e
10e
10e
10e
10e
10e
10e
11.5
9.5c
9.2
8.0
5.2d
4.8
pyrrolidine/NH4OH
NH3‚MeOH/NH4OH
NH3‚EtOH, 24 h, rt
NH3‚MeOH, 24 h, rt
NH3‚MeOH, 24 h, rt
NH3‚MeOH, 24 h, rt
NH3‚MeOH
3.7d
3.8e
3.7e
4.0e
3.4e
1.5e
2.0e
2.3
NH4OH/EtOH (1:1)
(same oligo as entry 14 re-treated with
4.9
NH4OH/EtOH 15 h, 55 °C)
a
Percentage of phosphorothioate was determined by 31P NMR
(see Experimental Section). s ) phosphorodithioate; p ) phos-
b
phate diester. Unless otherwise indicated, deprotection was
completed at 55 °C in a sealed vial for 15 h. c 500 µL pyrrolidine,
11 h at 55 °C; workup by addition of 500 µL of water, removal of
d
liquid, and then washing beads with 500 µL of water. 500 µL
pyrrolidine, 55 °C for 1 h, addition of 500 µL of concd ammonium
hydroxide, proceed at 55 °C, 15 h. e Ammoniacal methanol and
ethanol were prepared by bubbling anhydrous ammonia through
these anhydrous alcohols.
tetrazole activation in two 45 s pulses. The resulting
thiophosphite triester 26 was oxidized with elemental
sulfur to the desired phosphorodithioate triester. Unre-
acted CPG-bound deoxynucleoside was then capped with
acetic anhydride and the cycle continued.
ethyl,12 2,4-dichlorobenzyl,21a or â-(benzoylmercapto)-
ethyl22 protection on sulfur. Thus the higher level of
phosphorothioate observed when preparing deoxyoligo-
nucleotides on a support was peculiar to this mode of
synthesis.
Initial 31P NMR studies to evaluate the phospho-
rothioamidites having varied acyl protection of the
â-mercaptoethyl group focused upon side-by-side com-
parisons of oligothymidylate products. In the first study,
the level of phosphorothioate impurity (55 ppm) in
octathymidylates having phosphorodithioates at all seven
linkages was found to be directly related to the lability
of the thioester group. Pivaloyl protection gave rise to a
high level of thioate (11.5%, Table 1, entry 1), which was
reduced using the more labile acetate (8.0%, entry 4) or
benzoate (4.8%, entry 6). These differences can be
attributed to the relative rates of two competing pro-
cesses: deacylation and subsequent elimination of eth-
ylene sulfide to yield the desired phosphorodithioate 3
versus hydroxide attack at phosphorus to displace thiol
and generate the phosphorothioate impurity 28. By
treating the support with pyrrolidine prior to the stan-
dard ammonia deprotection, the levels of phosphorothio-
ate were further reduced (entries 5 and 7), thus support-
ing the rationale that phosphorothioate derives largely
from hydroxide attack at phosphorus. However, as has
been shown previously, the phosphorothioate impurity
was only 1% in model experiments on the synthesis of
phosphorodithioate dimers in solution with â-cyano-
Because previous studies had not implicated deprotec-
tion procedures as a major cause of phosphorothioate
impurities,12 alternative sources of this contaminant were
also investigated. Variation of sulfurization conditions
and reaction time gave essentially the same levels of
phosphorothioate. While incomplete sulfurization of
thiophosphite triester 26 can lead to phosphorothioate
formation (via oxidation of 26 upon subsequent exposure
to TCA), it did not appear to be significant. Doubling
the coupling time, sulfurization time, or both had no
affect upon the level of phosphorothioate impurity.
In a previous study using â-cyanoethyl phospho-
rothioamidites, the high levels (8-9%) of phosphorothio-
ate impurity were attributed in part to the tetrazole-
catalyzed dismutation of phosphorothioamidites to
bisamidites of type 8. Under synthesis conditions this
(21) (a) Brill, W. K.-D.; Nielsen, J .; Caruthers, M. H. J . Am. Chem.
Soc. 1991, 113, 3972-3980. (b) Beaton, G.; Caruthers, M. H., unpub-
lished results.
(22) Goodman, B.; Wiesler, W. T.; Caruthers, M. H., unpublished
results.

Synthesis of Phosphorodithioate DNA
J . Org. Chem., Vol. 61, No. 13, 1996 4277
reactive impurity couples to hydroxyls of two growing,
support-linked oligomers to give the branched thiotriester
linkage 29a , which during deprotection slowly hydrolyzes
to phosphorothioate 28 and its 5′-5′ isomer.12e The 31P
the other via â-mercaptoethyl dithiotriester 27. While
contamination of phosphorothioamidites with 2% of
byproduct 14 could give rise to an additional 1% phos-
phorothioate contamination in deoxyoligonucleotides, 14
is sterically hindered relative to phosphorothioamidites
10 and is likely to be less reactive.
An additional minor contaminant observed in the 31P
NMR of phosphorodithioate deoxyoligonucleotides was
identified as thiotriester 31 (65 ppm). The level of this
contamination (typically 1% in initial studies) was di-
rectly correlated with the level of dinucleoside amidite
18 in the phosphorothioamidite preparations. Introduc-
tion of this branched triester linkage was also correlated
to the presence of higher molecular weight bands in crude
deoxyoligonucleotides as observed by PAGE. While hy-
NMR of model compound 29b (66.6 ppm) confirmed that
the thiotriester 29a (64.9, 66.4 ppm) was a minor
impurity in phosphorodithioate deoxyoligonucleotides as
synthesized on CPG. Experiments on supports22 where
phosphorothioamidites were spiked with high levels of
bisamidite 8 demonstrated that this impurity could be
the causative agent of thiotriester 29a and the resulting
increased level of phosphorothioate contamination in
phosphorodithioate deoxyoligonucleotides.
However, the extent to which dismutation contributed
to phosphorothioate formation was unclear. Coupling is
carried out with two deliveries of fresh phosphorothio-
amidite and tetrazole to the column, each followed by a
40 s wait, during which time dismutation may have
occurred. Extension of this wait period (and thus delay-
ing the delivery of fresh phosphorothioamidite) did not
increase the level of phosphorothioate impurity which
suggested that dismutation to bisamidite 8 was not a
significant source of phosphorothioate. Thus, in our work
on supports we attributed the presence of 29a to bis-
amidite 8 which was present as an impurity in some of
our deoxynucleoside phosphorothioamidite preparations.
Other reactive trivalent phosphorus impurities present
in phosphorothioamidites may also have contributed to
phosphorothioate formation. One of these impurities,
deprotection byproduct 14, is unique to â-(acylmercapto)-
ethyl-protected phosphorothioamidites. Coupling of this
dimer byproduct to one hydroxyl on the support would
likely be followed by a second coupling to give 30, in
which two parallel deoxyoligonucleotides are linked.
drolysis of triester 31 could contribute to phosphorothio-
ate 32, introduction of this branched linkage has the
potential of diverting significant quantities of deoxyoli-
gonucleotide product to high molecular weight byproducts
as 18 is more reactive than 10. Thus elimination of this
impurity was given priority, and an alternative route was
later developed for this purpose (see above).
Because initial studies indicated that nonspecific depro-
tection was the major source of phosphorothioate impu-
rity, additional variations were examined for protecting
the â-mercaptoethyl group. Phosphorothioamidites were
prepared with more labile benzoyl (10f,g) and carbonate
(10h -l) groups. Low levels (3.6-5.6%) of phosphorothio-
ate were observed with phenyl carbonate and â-cyano-
ethyl carbonate protection; however, the deoxyoligonu-
cleotides contained an additional thiotriester contam-
ination (61-63 ppm) owing to the presence of phospho-
roamidite impurities of type 17 in the phosphorothio-
amidites 10h -l. Interestingly, phosphoroamidites bear-
ing the most labile protecting groups (difluorobenzoyl,
10g; FMOC, 10l) gave rise to the highest levels of
phosphorothioate observed (>15%). These groups were
the most hydrophobic of all the variations examined,
which suggests that poor solubility of the protected
deoxyoligonucleotides in ammonia may increase the
extent of nonspecific hydrolysis to phosphorothioate.
In order to improve the solubility of phosphorodithioate
deoxyoligonucleotides during deprotection, aqueous am-
monia was replaced with various methanolic or ethanolic
ammonia mixtures. Clear reductions in the levels of the
phosphorothioate impurity were observed (Table 1, en-
Hydrolysis during ammonia deprotection would introduce
phosphorothioate impurity 28 into one deoxyoligonucle-
otide and the desired phosphorodithioate linkage 3 into

4278 J . Org. Chem., Vol. 61, No. 13, 1996
Ta ble 2. Syn th esis of Deoxyoligon u cleotid e P h osp h or od ith ioa tesa
Wiesler and Caruthers
phosphoro-
thioate, %
entry
deoxyoligonucleotide
deprotectionb,c
pyrrolidine/NH4OH
NH3‚EtOH/benzene
NH3‚MeOH/NH4OH
NH3‚EtOH/benzene
NH3‚EtOH/benzene
NH3‚EtOH/benzene (24 h, 4 °C)
NH3‚EtOH
NH3‚EtOH
NH3‚EtOH/benzene
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
C(sC)13
C(sC)13
C(sC)20
C(sC)20
C(sC)26
C(sC)27
7.0
2.6
5.0
3.2d
3.0
3.0e
5.0f
4.2
2.9
3.6g
3.6g
5.0
4.1
3.6
3.9
6.0
3.7
1.0
1.0
1.3
1.0
CsTsGsTsTsCsGsGsGsCsGsCsCsA
CsTsGsTsTsCsGsGsGsCsGsCsCsA
CsTsGsTsTsCsGsGsGsCsGsCsCsA
CsTsGsTsTsCsGsGsGsCsGsCsCsA
NH3‚EtOH/benzene (12 h, rt)
NH4OH/pyridine/EtOH (1:1:2) 3 d, 4 °C
NH3‚EtOH/benzene
CsTsGsTsTsCsGsGsGsCsGsCsCsA
GsAsCsAsGsAsTsGsCsAsCsCsAsTsTsCsTsG
GsAsCsApGsApTsGpCsApCsCpAsTpTsCsTsG
GsAsCsApGpAsTpGpCsApCpCsApTpTsCsTsG
GsApCsApGsApTsGpCsApCsCpAsTpTsCpTsG
NH3‚EtOH/benzene
NH3‚EtOH/benzene
NH3‚EtOH/benzene
GsGsCsAsGsGsAsAsGsAsAsGsCsGsGsAsGsAsCsAsGsCsGsAsCsGsA NH3‚EtOH/benzene
TsCsGsTsCsTsTsGsTsCsCsCsGsTsCsTsCsGsTsTsGsCsCsCsCsTsC
CmCmAmApAsApGsGpCsCpGsApGsApAsGpCmGmAmT
CmCmAmApApAsGpGpCsCpGpAsGpApAsGpCmGmAmT
TmCmGmTmCpGsCpTsGpTsCpTsCpCsGpCmTmTmCmT
TmCmGmTmCpGpCsTpGpTsCpTpCsCpGpCmTmTmCmT
NH3‚EtOH/benzene
NH4OH/pyridine/EtOH (1:1:2) 3 d, 4 °C
NH4OH/pyridine/EtOH (1:1:2) 3 d, 4 °C
NH4OH/pyridine/EtOH (1:1:2) 3 d, 4 °C
NH4OH/pyridine/EtOH (1:1:2) 3 d, 4 °C
a
Percentage of phosphorothioate was determined by 31P NMR (see Experimental Section). s ) phosphorodithioate; p ) phosphodiester;
b
m ) methylphosphonate. Unless otherwise indicated, deprotection was completed at 55 °C in a sealed vial for 15 h and deoxynucleoside
phosphorothioamidites 22a , 23a , 24a , and 25a were used to prepare the phosphorodithioate deoxyoligonucleotides. c Ammoniacal methanol
and ethanol were prepared by bubbling anhydrous ammonia through these anhydrous alcohols. Benzene (150 µL) and these ammoniacal
solutions (850 µL) were then mixed in order to generate appropriate deprotection reagents. Prepared from thioamidite 23c. e Prepared
d
from thioamidite 23d . f Prepared from thioamidites 22b, 23b, 24b, and 25b. Prepared from thioamidites 22a , 23d , and 25c.
g
tries 8-15) where the lowest level (1.5%, entry 12) was
with a 15-mer having seven phosphodiester linkages at
the 5′-end. The analogous 15-mer with seven phosphodi-
ester linkages at the 3′-end did not show any reduction
in phosphorothioate (entry 11). It is difficult to explain
these observations without more extensive studies. Be-
cause deprotection of normal internucleotide â-cyanoethyl
phosphate triesters is very rapid (deoxynucleoside 3′-(â-
cyanoethyl phosphoroamidites) are used for synthesizing
normal DNA internucleotide linkages) perhaps the oli-
gomer having this type of linkage at the 5′-end would be
more accessible to solvent as the diester following depro-
tection. Solubilization in this manner might therefore
lead to a lower phosphorothioate contamination in a
manner similar to model studies in solution.
Solubility of protected phosphorodithioate deoxyoligo-
nucleotides having no phosphate internucleotide linkages
was enhanced by the addition of benzene to ethanolic
ammonia which further reduced the level of phospho-
rothioate impurity during deprotection. For example,
with polydithioate deoxyoligocytidine 14-mers (Table 2,
entries 1 and 2) and heterodeoxyoligomers (Table 2,
entries 7-9), considerable reduction of phosphorothioate
content was observed. In longer deoxyoligonucleotides,
however, the level of phosphorothioate contamination
was higher with an increasing number of purines. For
example, the level of phosphorothioate in a purine rich
27-mer (6.0%, entry 16) was much higher than that for
a pyrimidine rich oligomer of the same length (3.7%,
entry 17). Again solubilization during deprotection most
likely accounted for the difference. Relative to this
interpretation of the results, it was interesting to note
that the fully deprotected purine rich 27-mer (entry 16)
showed limited solubility in water (pyrimidine rich
polydithioates were very soluble in water).
2.3% of phosphorothioate impurity was subjected to a
second 15 h treatment with base, the level of phospho-
rothioate increased to 4.9% (Table 1, entry 15). Thus as
much as 2-3% of the phosphorothioate contamination
may be introduced via desulfurization under standard
basic deprotection conditions. This observation quite
possibly puts a limit on the extent to which the level of
phosphorothioate may be reduced under these conditions.
These methods for synthesizing phosphorodithioate
DNA were fully compatible with the procedures previ-
ously developed for preparing unmodified as well as
phosphorothioate and methylphosphonate DNA. Rela-
tive to many biochemical and biological studies that
utilize antisense deoxyoligonucleotides, several deoxy-
oligonucleotides were prepared that contained the phos-
phorodithioate analog in combination with other types
of internucleotide linkages (Table 2, entries 13-15 and
18-21). The lowest levels of phosphorothioate impurity
(1%) were observed in a series of 20-mer deoxyoligonucle-
otides containing four methylphosphonate linkages at
each termini and 3-5 phosphorodithioate linkages in the
interior of the oligomers (Table 2, entries 18-21). The
mild deprotection conditions used for these oligomers
(ammonium hydroxide-pyridine-ethanol, 1:1:2, v/v/v for
3 days) were initially considered to be the reason for this
low level of contamination. In order to evaluate this
possibility, phosphorothioamidites 23d and 25c were
prepared from deoxynucleosides having more labile base-
protecting groups.23
However, deoxyoligonucleotides pre-
pared from these synthons had the usual levels of
phosphorothioate impurity (3.6%) using two different
deprotection procedures that were significantly milder
than the standard conditions (Table 2, entries 10 and 11).
These results suggested that the low levels of phospho-
rothioate observed in certain examples were more a
function of oligomer solubility than the temperature or
duration of the deprotection treatment.
Low levels of base-catalyzed desulfurization of phos-
phorodithioate DNA can also contribute to the generation
of phosphorothioate impurity. When a fully deprotected
phosphorodithioate deoxyoligonucleotide containing only
(23) Schulhof, J . C.; Molko, D.; Teoule, R. Nuc. Acids Res. 1987, 15,
397-416.

Synthesis of Phosphorodithioate DNA
J . Org. Chem., Vol. 61, No. 13, 1996 4279
mmHg (lit.27 135-137 °C/2.5 mmHg). 1H NMR (CDCl3): 4.15
(m, 4H), 3.15 (m, 2H), 2.98 (m, 2H), 2.30 (s, 3H, Me) 1.30 (t,
6H, Me’s). 31P NMR: 93.7 ppm. FAB+ MS: 288 (M).
O,O-Dieth yl S-(â-(Ben zoylm er ca p to)eth yl) P h osp h o-
r od ith ioa te (4b). The titled compound was prepared from
1-bromoethanethiol benzoate26 and diethyl dithiophosphate
following the procedure for 4a with filtration of pyridine
hydrobromide salt from the reaction mixture prior to aqueous
workup. The product could not be distilled but was purified
by flash chromatography (Rf ) 0.23; ethyl acetate-hexane, 1:9,
v/v). This compound was also prepared by treatment of O,O-
diethyl S-(â-mercaptoethyl) phosphorodithioate (5) (3.3 g, 13.4
mmol) with benzoyl chloride (2.3 g, 16.3 mmol) in pyridine (1
h). Following the same aqueous workups, chromatography,
and decolorization with activated charcoal, the intended
product was obtained in 77% yield (3.7 g, 10.6 mmol). 1H NMR
(CDCl3): 7.95 (m, 2H), 7.56 (m, 1H), 7.44 (m, 2H), 4.13 (m,
4H), 3.30 (dd, 2H), 3.10 (m, 2H), 1.34 (t, 6H, Me’s). 31P NMR:
92.6 ppm. FAB+ MS: 351 (M + 1), 245 (M - (benzoyl)).
O,O-Dieth yl S-(â-(Isobu tylm er ca p to)eth yl) P h osp h o-
r od ith ioa te (4c). O,O-Diethyl S-(â-mercaptoethyl) phospho-
rodithioate (5) (2.46 g, 10.0 mmol) was treated with isobutyrl
chloride (1.3 g, 12.2 mmol) in the presence of Ag2CO3 (1.5 g)
in dichloromethane (15 mL) to yield the desired product after
5 min when analyzed by TLC. The reaction mixture was
diluted with diethyl ether (50 mL), treated with activated
charcoal, and filtered through a small bed of Celite and silica,
and excess acid chloride was removed in vacuo. The desired
product was obtained without further purification. 1H NMR
(CDCl3): 4.13 (m, 4H), 3.09 (m, 2H), 2.95 (m, 2H), 2.65 (m,
1H), 1.31 (t, 6H, Me’s), 1.17 (m, 6H, Me’s). 31P NMR: 93.7
ppm.
O,O-Dieth yl S-(â-(P iva loylm er ca p to)eth yl) P h osp h o-
r od ith ioa te (4d ). O,O-Diethyl S-(â-mercaptoethyl) phospho-
rodithioate (5) (2.8 g, 11.3 mmol) was treated with pivaloyl
chloride (2.1 g, 17.4 mmol) in the presence of Ag2CO3 (3.0 g)
in dichloromethane (15 mL). TLC showed the reaction was
complete after 5 min. The workup procedure was the same
as for 4c. 1H NMR (CDCl3): 4.13 (m, 4H), 3.08 (m, 2H), 2.95
(m, 2H), 1.31 (t, 6H, Me’s), 1.18 (s, 9H, Me’s). 31P NMR: 93.7
ppm. FAB+ MS: 331 (M + 1), FAB- MS: 301 (M - (ethyl)).
O,O-Dieth yl S-(â-Mer ca p toeth yl) P h osp h or od ith ioa te
(5). Diethyl dithiophosphate (22.5 mL, 135 mmol) and eth-
ylene sulfide (8.5 g, 142 mmol) were refluxed without solvent
overnight. The mixture was distilled directly to yield the
desired product which was collected at 80-90 °C/0.2 mmHg
(lit.26 110-112 °C/1.5 mmHg) (13.7 g, 41%). 1H NMR
(CDCl3): 4.10 (m, 4H), 3.00 (m, 2H), 2.73 (m, 2H), 1.64 (t, 1H,
SH), 1.28 (t, 6H, Me’s). 31P NMR: 94.2 ppm.
The hydrophobic properties of phosphorodithioate DNA
were revealed not only by solubilization during depro-
tection but also by a variety of purification procedures.24
Purification of this analog by HPLC can best be carried
out using a divinylbenzene-polystyrene co-polymer re-
verse phase column (Hamilton PRP-1) where phospho-
rodithioate DNA oligomers were retained to a greater
extent than either their unmodified or phosphorothioate
counterparts. Another notable difference was the mobil-
ity of phosphorodithioate DNA by PAGE. Under dena-
turing conditions, phosphorodithioate DNA migrated
with a mobility similar to unmodified DNA; however, in
nondenaturing gels this analog as a single-stranded,
noncomplementary oligomer migrated like natural DNA
duplexes of the same length.25 This observation suggests
that single-stranded phosphorodithioate DNA may have
some defined rodlike structure. It is possible that the
phosphorodithioate internucleotide linkages allow for a
greater degree of base stacking than is observed in
natural DNA. This could account for the solubility
differences observed between deoxyoligopyrimidines and
various purine rich phosphorodithioate deoxyoligonucle-
otides.
Con clu sion
A major challenge in the synthesis of phosphorodithio-
ate DNA was to minimize contamination with phospho-
rothioate internucleotide linkages. A related challenge
was to develop a deoxynucloside phosphorothioamidite
synthon having chemical properties that lead to high
yields of this analog while at the same time producing
very few side products. Of the various synthons tested,
these objectives were best achieved with the deoxy-
nucleoside 3′-phosphorothioamidites having a â-(ben-
zoylmercapto)ethyl protecting group on sulfur. Exami-
nation of the deprotection conditions further demonstrated
that ethanolic ammonium hydroxide in benzene mini-
mizes the generation of the phosphorothioate side prod-
uct.
Exp er im en ta l Section
Gen er a l Meth od s. Protected deoxyribonucleosides were
obtained from Cruachem (Sterling, VA). Suppliers for other
reagents and solvents were as previously reported.21a 1H-
Tetrazole was sublimed before use. 31P NMR spectra were
recorded in either acetonitrile (all phosphorothioamidites) or
D2O (deoxyoligonucleotides). Integrated percentage of phos-
phorothioate impurity in deoxyoligonucleotides was reproduc-
ible to within 0.2%. Elemental analysis was performed by
Galbraith Laboratories (Knoxville, TN). DNA synthesis was
performed on an Applied Biosystems 380A automated DNA
synthesizer using Applied Biosystems 1 µmol CPG columns.
O,O-Diet h yl S-(â-(Acet ylm er ca p t o)et h yl) P h osp h o-
r od ith ioa te (4a ). Diethyl dithiophosphate (3.02 g, 16.2 mmol)
and 1-bromoethanethiol acetate26 (2.97 g, 16.2 mmol) were
refluxed in 10 mL of pyridine-dichloromethane (1:1, v/v) for
24 h, after which time starting material was converted to a
product of lower Rf (ethyl acetate-hexane, 3:97, v/v). The
reaction mixture was diluted with 50 mL of diethyl ether and
washed with 4% HCl (50 mL), 5% aqueous sodium bicarbonate
(50 mL), and brine (50 mL). After the solution was dried over
Na2SO4, the diethyl ether layer was concentrated to 4.6 g of a
yellow oil and distilled to give the product at 110-120 °C/0.15
Eth a n ed ith iol Mon op iva loa te (9a ). Ethanedithiol (5.7
g, 60.3 mmol) and pivaloyl chloride (7.3 g) were combined in
pyridine-dichloromethane (1:3, v/v) and stirred overnight. The
reaction mixture was filtered, concentrated in vacuum, and
distilled to give the desired product at 120 °C/15-17 mmHg
(lit.27 104-105 °C/12 mmHg). 1H NMR (CDCl3): 2.97 (m, 2H),
2.62 (m, 2H), 1.56 (t, 1H, SH), 1.17 (s, 9H, Me’s).
Eth a n ed ith iol Mon oisobu tyr a te (9b). The title com-
pound was prepared in the same manner as for 9a . The
product was collected in fractions 96-110 °C/10-12 mmHg.
1H NMR (CDCl3): 3.04 (m, 2H), 2.60-2.75 (m, 3H), 1.59 (t,
1H, SH), 1.17 (d, 6H, Me’s).
Eth a n ed ith iol Mon op r op r ion a te (9c). The title com-
pound was prepared in the same manner as for 9a . The
product was collected in fractions 90-110 °C/12 mmHg. 1H
NMR (CDCl3): 3.06 (m, 2H), 2.68 (m, 2H), 2.56 (q, 2H), 1.60
(t, 1H, SH), 1.15 (t, 3H, Me).
Eth a n ed ith iol Mon oa ceta te (9d ). Ethanedithiol (7.0 mL,
83.5 mmol) and acetic anhydride (8.0 mL, 83.5 mmol) were
stirred overnight in pyridine-dichloromethane (1:1, v/v). The
reaction mixture was concentrated and distilled to give the
product at 75 °C/10 mmHg (lit.26 92 °C/17 mmHg). 1H NMR
(CDCl3): 3.04 (m, 2H), 2.64 (m, 2H), 2.30 (s, 3H, Me), 1.58 (t,
1H, SH).
(24) Wiesler, W. T.; Marshall, W. S.; Caruthers, M. H. In Methods
in Molecular Biology: Oligonucleotide Synthesis Protocols; Agarwal,
S., Ed.; Humana Press: Clifton, NJ , 1993; pp 191-206.
(25) Marshall, W. S.; Caruthers, M. H., unpublished results.
(26) Hansen, B. Acta Chem. Scand. 1957, 11, 537-540.
(27) Mastriukova, T. A.; Odnoralova, V. N.; Kabachnik, M. I. Zh.
Org. Khim. USSR 1957, 28, 1563 (1613-1617, Engl. ed.).

4280 J . Org. Chem., Vol. 61, No. 13, 1996
Wiesler and Caruthers
Eth a n ed ith iol Mon oben zoa te (9e). Ethanedithiol (100
g, 1.06 mol) was placed in a 2 L three-neck flask with 400 mL
of anhydrous diethyl ether and 100 mL of anhydrous pyridine.
The solution was cooled to 0 °C, and benzoyl chloride (123 mL,
1.06 mmol) was added dropwise from a dropping funnel over
2 h with mechanical stirring. The reaction was allowed to sit
for an additional hour. The solution was then filtered to
remove the large mass of pyridine hydrochloride salt which
was washed with diethyl ether. The combined filtrates were
concentrated to an oil, which was taken up in 500 mL of hot
methanol and allowed to sit in a -5 °C freezer overnight. The
white crystals of bisbenzoate byproduct were removed by
filtration and washed with cold methanol. The combined
filtrates were concentrated to approximately 150 mL of a
yellow oil. Excess ethanedithiol starting material was re-
moved under high vacuum by heating to 60 °C and collected
in a -78 °C trap. The remaining oil was distilled with a
microdistillation apparatus. The desired product was collected
at 0.2 mmHg and 110 °C employing an oil bath at 160 °C.
Typical yields were approximately 80 g (40%). The absence
of any ethanedithiol contamination in the product was con-
firmed by 1H NMR (CDCl3): 7.95 ppm (d, 2H); 7.57 (t, 1H);
7.44 (m, 2H); 3.30 (t, 2H); 2.78 (m, 2H); 1.70 (t, 1H).
Eth a n ed ith iol Mon o(2,4-d iflu or oben zoa te) (9g). The
titled compound was prepared in the same manner as for 9a .
The product was collected at 110-120 °C/0.4 mmHg. 1H NMR
(CDCl3): 7.85 ppm (m, 1H); 6.88 (m, 2H); 7.44 (m, 2H); 3.23
(dd, 2H); 2.76 (m, 2H); 1.66 (t, 1H).
Eth a n ed ith iol Mon o(p h en yl ca r bon a te) (9h ). The titled
compound was prepared in the same manner as for 9a from
phenyl chloroformate. The product was collected at 100 °C/
0.3 mmHg. 1H NMR (CDCl3): 7.69 ppm (m, 2H); 7.51 (m, 2H);
7.31 (m, 4H); 4.43 (d, 2H); 4.20 (t, 1H); 3.00 (t, 2H); 2.64 (m,
2H); 1.56 (t, 1H, SH).
monoacylate was added (2 equiv). After 11 min, the reaction
was diluted with dichloromethane (20 mL) and ethyl acetate
(3 mL), washed with saturated bicarbonate (2 × 25 mL), 10%
Na2
CO3 (25 mL), and brine (25 mL), and dried over Na2SO4.
The organic layer was concentrated to a syrup, dissolved in
toluene (10 mL) with triethylamine (1 mL), and added drop-
wise to 400 mL of vigorously stirred heptane to precipitate
the white product. The precipitate was collected by filtration
through a medium frit glass funnel and dried in vacuo. Typical
yield was 80%.
Gen er a l P r oced u r e II (10d -l). After the identical phos-
phitylation procedure as described above, the protected thiol
(0.5 mL) and tetrazole (150 mg) were added together in a
solution of anhydrous acetonitrile (7 mL). The reaction was
shaken for 10-15 s and poured into dichloromethane (100 mL).
The solution was washed with bicarbonate (2 × 100 mL), 10%
Na2CO3 (100 mL), and brine (100 mL) and dried over Na2SO4.
The product was precipitated from heptane (10d -g,l), hex-
ane-cyclohexane (2:1, v/v) (10h ,i), or cyclohexane-toluene (9:
1, v/v) (10j,k ) as described above.
Tr is(p yr r olid in o)p h osp h in e (19). Phosphorous trichlo-
ride (50.4 g, 37 mmol) was placed in a 1 L flask with 400 mL
of anhydrous ether. The flask was fitted with a 250 mL
addition funnel containing 173.5 g of (trimethylsilyl)pyrrolidine
(3.3 equiv). The flask was cooled to -10 °C (ice/brine) under
argon with magnetic stirring, after which the TMS-pyrrolidine
was added dropwise over the course of 1.5 h. Stirring was
maintained for an additional 1 h, after which it was stopped
to allow the small amount of precipitates to settle. The
reaction mixture was filtered through a medium frit sintered
glass funnel, and the salts were washed with an additional
50 mL of anhydrous ether. Ether and trimethylsilyl chloride
were removed from the filtrate by rotary evaporation to give
approximately 100 mL of crude phosphine. Distillation at
reduced pressure (0.25 mmHg) afforded an initial fraction (40-
103 °C, 5-10 g) which was discarded and the product (104-
110 °C, 76.3 g) obtained in 86% yield as a clear colorless oil.
31P NMR: 104.6 ppm (THF), 102.8 ppm (CDCl3).
Eth a n ed ith iol Mon o(tr ich lor oeth yl ca r bon a te) (9j).
The titled compound was prepared in the same manner as for
9a from trichloroethyl chloroformate. The product was col-
lected at 100 °C/0.3 mmHg. 1H NMR (CDCl3): 4.86 (s, 2H),
3.17 (dd, 2H), 2.83 (m, 2H), 1.69 (t, 1H, SH).
Deoxyn u cleosid-3′-yl S-(â-(Ben zoylm er capto)eth yl) P yr -
r olid in o Th iop h osp h or oa m id ites (22-25). Gen er a l P r o-
ced u r e. 5′-O-(Dimethoxytrityl)deoxyribonucleoside (2 mmol)
was dissolved in 30 mL of anhydrous dichloromethane to which
a spatula of 3 Å molecular sieves was added. Tris(pyrrolidino)-
phosphine (485 mg, 470 µL) was added via syringe followed
by seven 0.1 mmol aliquots of tetrazole (7 × 0.2 mL of a 0.5 M
solution in anhydrous acetonitrile stored over sieves) at 2 min
intervals. (Trimethylsilyl)imidazole (30 µL, 0.1 equiv) was
then added to the reaction. After 5 min, tetrazole (10.8 mL of
a 0.5 M solution in anhydrous acetonitrile) was added, im-
mediately followed by the addition of ethanedithiol monoben-
zoate (520 mg, 440 µL). The reaction was allowed to proceed
for 105 s (T), 120 s (dC, dA), or 150 s (dG). The reaction was
quenched by pouring the solution into 75 mL of dichlo-
romethane containing 5 mL of triethylamine. The mixture was
immediately washed with saturated sodium bicarbonate (100
mL) followed by 10% sodium carbonate (2 × 100 mL) and
saturated brine (100 mL). The organic layer was dried over
Na2SO4. After 10-15 min the drying agent was removed by
filtration. Triethylamine (5 mL) was added to the solution
which was concentrated using a rotary evaporator to a syrup.
The syrup was dissolved in toluene (25 mL) and triethylamine
(5 mL), and this solution was pipetted into 900 mL of
vigorously stirred heptane or pentane (22a only) to precipitate
the fluffy white product. After most of the heptane was
decanted, the white precipitate was collected by filtration
through a medium sintered glass funnel and subsequently
dried under vacuum to give the desired product in ap-
proximately 75-80% yield.
E t h a n ed it h iol Mon o(â-cya n oet h yl ca r b on a t e) (9k ).
Triphosgene (24.5 g, 83 mmol) was dissolved in anhydrous
toluene (150 mL) and cooled to 0 °C. Dropwise addition of
â-cyanoethanol (17.6 g, 248 mmol) was followed by dropwise
addition of pyridine (20 mL) at 0 °C. The reaction mixture
was then filtered to remove pyridine hydrochloride salt and
concentrated to an oil. The crude â-cyanoethyl chlorocarbonate
was dissolved in 50 mL of anhydrous ether, placed in a
dropping funnel, and added dropwise to ethanedithiol (28 mL,
334 mmol) in anhydrous diethyl ether (200 mL) and pyridine
(25 mL). The reaction was filtered, and the filtrate was
washed with saturated bicarbonate (3 × 100 mL) and brine
(100 mL) and dried over Na2SO4. The diethyl ether solution
was concentrated to an oil and distilled. The product was
collected at 140-150 °C/0.4 mmHg. 1H NMR (CDCl3): 4.43
(dd, 2H), 3.13 (dd, 2H), 2.79 (m, 4H), 1.70 (t, 1H, SH).
Eth an edith iol Mon o(9-flu or en ylm eth yl car bon ate) (9l).
9-Fluorenylmethyl chloroformate (15 g, 58 mmol) in anhydrous
ether was added dropwise to an excess of ethanedithiol (15
mL, 178 mmol) in diethyl ether (50 mL) containing pyridine
(6 ml). After 1 h, salts were removed by filtering and the
filtrate was washed with bicarbonate (2 × 50 mL) and brine
(50 mL). The ether solution was concentrated to a thick syrup
under high vacuum, and the product was purified by flash
chromatography (ethyl acetate-hexane, gradient from 1:99 to
1:9). Rf ) 0.4 (ethyl acetate-hexane, 1:9, v/v). 1H NMR
(CDCl3): 7.69 ppm (m, 2H); 7.51 (m, 2H); 7.31 (m, 4H); 4.43
(d, 2H); 4.20 (t, 1H); 3.00 (t, 2H); 2.64 (m, 2H); 1.56 (t, 1H,
SH).
O-(5′-O-(Dim et h oxyt r it yl)t h ym id in -3′-yl) S-(â-Acyl-
m er ca p t o)et h yl) P yr r olid in o Th iop h osp h or oa m id it es
(10a -e). Gen er a l P r oced u r e I. 5′-O-(Dimethoxytrityl)-
deoxyribonucleoside (1 mmol) was dissolved in 2.5 mL of
anhydrous THF and 0.5 mL of anhydrous triethylamine. Bis-
(pyrrolidino)chlorophosphine21a (220 µL) was added via syringe
and allowed to stir for 5 min. The reaction mixture was diluted
with anhydrous dichloromethane (7 mL), and the ethanedithiol
O-(5′-O-(Dim et h oxyt r it yl)t h ym id in -3′-yl) S-(â-(Ben -
zoylm er ca p to)eth yl) P yr r olid in o Th iop h osp h or oa m id ite
(22a ). FAB+ MS: 841 (M), 857 (M + O), 527 (M - (OP-
(NC4H8)SCH2CH2C(O)Ph), 303 (DMT cation). Anal. Calcd for
C44H48N3O8PS2: C, 62.76; H, 5.75; N, 5.00; P, 3.68; S, 7.61.
Found: C, 62.48; H, 6.04; N, 4.95; P, 3.57; S, 7.42. 31P NMR:
162.7, 159.2 ppm. 1H NMR (CD3CN): 9.50 ppm (bs, 1H, NH),

Synthesis of Phosphorodithioate DNA
J . Org. Chem., Vol. 61, No. 13, 1996 4281
7.87 (m, 2H), 7.60 (m, 1H), 7.43 (m, 4H), 7.3-7.2 (m, 8H), 6.81
(bd, 4H), 6.10 (m, 1H, H1′), 4.77 (m, 1H, H3′), 4.03 (m, 1H),
3.74 (s, 6H, OMe’s), 3.2-3.0 (9H), 2.78 (m, 2H), 2.38 (m, 2H),
1.67 (m, 4H, CH2’s), 1.45 (bs, 3H, Me).
P h osp h or od ith ioa te DNA Dep r otection a n d P u r ifica -
tion . After automated solid-phase synthesis,14i the CPG
column was removed from the synthesizer and dried with
argon. The CPG support was removed and placed in a dry, 1
dram vial to which ethanolic ammonia (850 µL) and benzene
(150 µL) were added. The vial was sealed with a Teflon-lined
screw cap, allowed to remain at rt for 1 h, and then placed in
a 55 °C oven for 15 h. After the vial was removed from the
oven and cooled to rt, toluene (2 mL) was added and the
combined solution carefully removed from the CPG support
by pipetting. This solution contained traces of DNA and nearly
all of the benzamide. Water (1 mL) was added to the vial
containing the CPG in order to extract the phosphorodithioate
DNA from the support. In the case of purine-rich sequences,
water/acetonitrile (30%) was used.
O-(5′-O-(Dim eth oxytr ityl)-2′-d eoxycytid in -3′-yl) S-(â-
(Ben zoylm er ca p to)eth yl) P yr r olid in o Th iop h osp h or o-
a m id ite (23a ). FAB+ MS (2,4-di-tert-butylphenol matrix):
931 (M + 1), 303 (DMT cation). Anal. Calcd for C50H51N4O8-
PS2: C, 64.50; H, 5.52; N, 6.02; P, 3.33; S, 6.89. Found: C,
64.04; H, 5.69; N, 5.99; P, 3.16; S, 6.92. 31P NMR: 162.1, 159.2
ppm. 1H NMR (CD3CN): 9.40 ppm (bs, 1H, NH), 8.22 (m, 1H),
7.85 (m, 4H), 7.6-7.4 (m, 8H), 7.33-7.1 (m, 8H), 6.82 (bd, 4H),
6.09 (m, 1H, H1′), 4.74 (m, 1H, H3′), 4.12 (m, 1H), 3.73 (s, 6H,
OMe’s), 3.18 (m, 2H), 3.13-2.95 (5H), 2.78 (m, 2H), 2.58 (m,
1H), 2.15 (m, 1H), 1.68 (m, 4H, CH2’s).
O-(5′-O-(Dim eth oxytr ityl)-2′-d eoxya d en osin -3′-yl) S-(â-
(Ben zoylm er ca p to)eth yl) P yr r olid in o Th iop h osp h or o-
a m id ite (24a ). FAB+ MS: 955 (M + 1), 886 (M - pyrroli-
dinyl), 303 (DMT cation). Anal. Calcd for C51H51N6O7PS2: C,
64.13; H, 5.38; N, 8.80; P, 3.24; S, 6.71. Found: C, 63.75 H,
5.37; N, 8.93; P, 3.28; S, 6.29. 31P NMR: 160.3, 159.2 ppm.
1H NMR (CD3CN): 9.58 ppm (bs, 1H, NH), 8.51 (s, 1H), 8.23
(s, 1H), 7.96 (d, 2H), 7.88 (d, 2H), 7.58 (dd, 2H), 7.46 (m, 4H),
7.33 (d, 2H), 7.2-7.1 (6H), 6.74 (5H), 6.40 (m, 1H, H1′), 4.98
(m, 1H, H3′), 4.19 (m, 1H), 3.68 (s, 6H, OMe’s), 3.3-3.1 (9H),
2.80 (dd, 2H), 2.58 (m, 1H), 1.71 (m, 4H, CH2’s).
O-(5′-O-(Dim eth oxytr ityl)-2′-d eoxygu a n osin -3′-yl) S-(â-
(Ben zoylm er ca p to)eth yl) P yr r olid in o Th iop h osp h or o-
a m id ite (25a ). Anal. Calcd for C48H53N6O8PS2: C, 61.52; H,
5.70; N, 8.97; P, 3.31; S, 6.84. Found: C, 61.17; H, 5.83; N,
8.85; P, 3.31; S, 6.60. 31P NMR: 159.7, 158.6 ppm. 1H NMR
(CD3CN): 10.80 ppm (bs, 1H, NH), 7.85 (m, 3H), 7.60 (m, 1H),
7.44 (m, 2H), 7.36 (m, 2H), 7.20 (m, 7H), 6.72 (m, 4H), 6.19
(m, 1H, H1′), 4.78 (m, 1H, H3′), 4.15 (m, 1H), 3.70 (s, 6H,
OMe’s), 3.3-3.0 (8H), 2.9-2.5 (5H), 1.69 (m, 4H, CH2’s), 1.12
(d. 6H, Me’s).
The resulting deoxyoligonucleotides were purified according
to standard procedures.28 For deoxyoligonucleotides with
multiple phosphorodithioate internucleotide linkages (greater
than 50%), the recommended approach was reverse-phase
column chromatography on a PRP-1 column (Hamilton, Reno,
NV) TRITYL-ON, removal of the dimethoxytrityl group with
acid, and then purification by PAGE under denaturing
conditions.14i
Su p p or tin g In for m a tion Ava ila ble: Expanded versions
of 31P-NMR spectra and Tables 1 and 2 with data on 69
deoxyoligonucleotides (6 pages). This material is contained
in libraries on microfiche, immediately follows this article in
the microfilm version of the journal, and can be ordered from
the ACS; see any current masthead page for ordering
information.
J O960274Y
(28) Caruthers, M. H.; Barone, A. D.; Beaucage, S. L.; Dodds, D. R.;
Fisher, E. F.; McBride, L. J .; Matteucci, M.; Stabinsky, Z.; Tang, J .-Y.
Methods Enzymol. 1987, 154, 287-313.