Nucleoside H-phosphonates. 17. Synthetic and 31P NMR studies on the preparation of dinucleoside H-phosphonothioates

Article abstract of DOI:10.1021/jo960810m

Formation of H-phosphonothioate diesters via condensation of H-phosphonate monoesters with a hydroxylic component in the presence of various coupling agents and possible side reactions that may accompany this process were studied using ³¹P NMR spectroscopy. Optimal reaction conditions, which eliminate or significantly suppress the side reactions, have been designed and assessed in syntheses of dinucleoside H-phosphonothioate diesters.

Full text of DOI:10.1021/jo960810m

J . Org. Chem. 1996, 61, 6617-6622
6617
Nu cleosid e H-P h osp h on a tes. 17. Syn th etic a n d 31P NMR Stu d ies
on th e P r ep a r a tion of Din u cleosid e H-P h osp h on oth ioa tes
Rula Zain and J acek Stawin´ski*
Department of Organic Chemistry, Arrhenius Laboratory, Stockholm University,
S-106 91 Stockholm, Sweden
Received May 3, 1996X
Formation of H-phosphonothioate diesters via condensation of H-phosphonate monoesters with a
hydroxylic component in the presence of various coupling agents and possible side reactions that
may accompany this process were studied using 31P NMR spectroscopy. Optimal reaction conditions,
which eliminate or significantly suppress the side reactions, have been designed and assessed in
syntheses of dinucleoside H-phosphonothioate diesters.
In tr od u ction
access to this class of phosphonate analogues, we have
developed two efficient methods for the preparation of
nucleoside H-phosphonothioate monoesters16,19 and showed
that these compounds can be converted to the appropriate
H-phosphonothioate diesters9 via condensation with suit-
ably protected nucleosides.
H-Phosphonothioate diesters are versatile synthetic
intermediates, for example, they may serve as precursors
to both phosphorothioate9,20 and phosphorodithioate9,20
diesters or to other analogues21-23 that are rather difficult
to obtain by other routes. Moreover, the possibility to
control the stereochemistry at the phosphorus center
during oxidation of H-phosphonothioate diesters with 3H-
2,1-benzoxathiol-3-one 1-oxide20 or with iodine21 can
provide a convenient entry to both diastereomers of
P-chiral phosphate analogues, depending on the oxidation
procedure chosen.
Substitution of one oxygen in H-phosphonate esters by
sulfur has severe stereochemical and chemical conse-
quences. At the monoester level,16,19 it introduces chiral-
ity at the phosphorus and this may potentially be
exploited in stereospecific synthesis of phosphorothioates
and other chiral phosphate analogues. The presence of
sulfur, however, makes two different nucleophilic centers
(O and S) available for activation by a condensing agent.
Lack of chemoselectivity in this process may cause a
partial elimination of sulfur during the course of the
reaction which can lead to formation of side products. At
the diester level,9,17,18 substitution of oxygen by sulfur at
the phosphorus center causes the P-H bond in H-
phosphonothioates to become noticeably more reactive
than that in the oxygen congener. This, in principle, also
is a potential source of side reactions.
The antisense and the antigene approaches to modula-
tion of gene expression rely on formation of stable
duplexes or triplexes between a target RNA or DNA
sequence and a synthetic oligonucleotide. These powerful
experimental techniques, which hold considerable me-
dicinal promise for therapeutic intervention in many
human genetic diseases,1,2 require oligonucleotide ana-
logues with good hybridization properties under physi-
ological conditions (high affinity and high specificity) and
with higher stability than their natural congener against
degradation by cellular and serum nucleases. Among
synthetic DNA and RNA, an important class of sur-
rogates with modified internucleotidic linkages consti-
tutes analogues in which one or two nonbridging oxygen
atoms at the phosphorus center have been replaced by
sulfur (oligonucleoside phosphorothioates3 and oligo-
nucleoside phosphorodithioates,4 respectively). These
oligonucleotide analogues are most conveniently acces-
sible via sulfurization of the appropriate P(III)
precursors,5-11 however, approaches based on P(V) de-
rivatives have also been developed.12,13
As a part of our research in H-phosphonate chemis-
try14,15 directed toward development of new synthetic
methods for biologically important phosphate esters and
their analogues, we have recently begun synthetic studies
on H-phosphonothioate esters.9,16-18 To provide an easy
X Abstract published in Advance ACS Abstracts, September 1, 1996.
(1) Wagner, R. W. Nature 1994, 372, 333-335.
(2) Stein, C. A.; Cheng, Y. C. Science 1993, 261, 1004-1012.
(3) Stein, C. A.; Cohen, J . S. In Oligodeoxynucleotides-Antisense
Inhibitors of Gene Expression; Cohen, J . S., Ed.; The Macmillan Press
Ltd.: New York, 1989; Vol. 12; pp 97-117.
(4) Marshall, W. S.; Caruthers, M. H. Science 1993, 259, 1564-1569.
(5) Burgers, P. M. J .; Eckstein, F. Tetrahedron Lett. 1978, 3835-
3838.
Having in mind possible applications of H-phosphono-
thioates as precursor to various oligonucleotide ana-
logues, we have undertaken systematic studies on this
class of compounds. In this paper we address the
(6) Stec, W. J .; Zon, G.; Egan, W.; Stec, B. J . Am. Chem. Soc. 1984,
106, 6077-6079.
(7) Nielsen, J .; Brill, W. K.-D.; Caruthers, M. Tetrahedron Lett. 1988,
29, 2911-2914.
(16) Stawinski, J .; Thelin, M.; Westman, E.; Zain, R. J . Org. Chem.
1990, 55, 3503-3506.
(17) Stawinski, J .; Stro¨mberg, R.; Szabo´, T.; Thelin, M.; Westman,
E.; Zain, R. Nucleic Acids Symp. Ser. 1991, 24, 95-98.
(18) Stawinski, J .; Stro¨mberg, R.; Thelin, M. Nucleosides Nucleotides
1991, 10, 511-514.
(19) Zain, R.; Stro¨mberg, R.; Stawinski, J . J . Org. Chem. 1995, 60,
8241-8244.
(20) Stawinski, J .; Thelin, M. Tetrahedron Lett. 1992, 33, 3189-
3192.
(21) Stawinski, J .; Stro¨mberg, R.; Zain, R. Tetrahedron Lett. 1992,
33, 3185-3188.
(22) Stawinski, J .; Thelin, M. J . Org. Chem. 1994, 59, 130-136.
(23) Bollmark, M.; Zain, R.; Stawinski, J . Tetrahedron Lett. 1996,
37, 3537-3540.
(8) Porritt, G. M.; Reese, C. B. Tetrahedron Lett. 1989, 30, 4713-
4716.
(9) Stawinski, J .; Thelin, M.; Zain, R. Tetrahedron Lett. 1989, 30,
2157-2160.
(10) Agrawal, S.; Tang, J .-Y. Tetrahedron Lett. 1990, 31, 7541-7544.
(11) Almer, H.; Stawinski, J .; Stro¨mberg, R.; Thelin, M. J . Org.
Chem. 1992, 57, 6163-6169.
(12) Eldrup, A. B.; Bjergårde, K.; Felding, J .; Kehler, J .; Dahl, O.
Nucleic Acids Res. 1994, 22, 1797-1804.
(13) Dahl, O. Sulfur Rep. 1991, 11, 167-192.
(14) Stawinski, J . In Handbook of Organophosphorus Chemistry;
Engel, R., Ed.; Marcel Dekker: New York, 1992; pp 377-434.
(15) Stawinski, J .; Stro¨mberg, R. Trends Org. Chem. 1993, 4, 31-
67.
S0022-3263(96)00810-9 CCC: $12.00 © 1996 American Chemical Society

6618 J . Org. Chem., Vol. 61, No. 19, 1996
Zain and Stawin´ski
Sch em e 1
problem of possible side reactions which may occur
during preparation of H-phosphonothioate diesters via
condensation of suitably protected nucleoside 3′-H-
phosphonothioate monoesters with nucleosides. Various
coupling agents and different reaction conditions for the
condensation have been evaluated. These studies en-
abled us to develop an optimal synthetic protocol which
eliminated or significantly suppressed side reactions
during condensations to 3. Dinucleoside H-phosphono-
thioate diesters 3a -d containing different heterocyclic
bases have been synthesised in 80-95% yields using the
newly developed conditions.
Resu lts a n d Discu ssion
Dinucleoside H-phosphonothioates 3 can be produced
in a condensation reaction as in Scheme 1 using nucleo-
side H-phosphonothioates 1 as a nucleotidic material or,
alternatively, by employing H-phosphonodithioate8 mo-
noesters for this purpose. Preliminary studies18 showed,
however, that possible advantages of using the latter
derivatives are outnumbered by serious drawbacks, e.g.,
lower reactivity of H-phosphonodithioates in condensa-
tions, formation of significant amounts of side products
during coupling with nucleosides, and the limited number
of coupling agents which may be used for their activation.
For these reasons we decided to pursue our original
approach,9 which is based on H-phosphonothioates of type
1, and to investigate it in more detail.
The high rates of condensations of H-phosphonothioate
monoesters with nucleosides have the consequence that,
similar to H-phosphonate diester syntheses, competing
reactions of condensing agents with the hydroxylic
component during the course of coupling can practically
be neglected. The situation is, however, more compli-
cated than in the synthesis of H-phosphonate diesters.
Due to higher reactivity of H-phosphonothioate diesters,
the condensation products are prone to subsequent
reactions at the phosphorus center. In addition, the
starting monoesters may react with a condensing agent
in two different ways (O vs S activation) producing
different products. For these reasons, the choice of a
condensing agent in H-phosphonothioate diester synthe-
sis is more critical than for H-phosphonates.
Different classes of condensing agents, i.e., acyl chlo-
rides, chlorophosphates, arylsulfonyl chlorides, and car-
bodiimides, have been evaluated from points of view of
their (i) efficiency to promote formation of H-phospho-
nothioate diesters, (ii) chemoselectivity during the activa-
tion of H-phosphonothioate monoesters, and (iii) reactiv-
ity toward the P-H bond in H-phosphonothioate diesters.
To have uniform set of data, all condensations have
been carried out in pyridine with variable amounts of a
condensing agent, and progress of the reactions was
monitored by 31P NMR spectroscopy.
P iva loyl Ch lor id e a s a Con d en sin g Agen t. Since
pivaloyl chloride (PV-Cl) is commonly used as the cou-
pling reagent in the automated synthesis of DNA and
RNA fragments via the H-phosphonate approach,15 it was
our initial choice for the H-phosphonothioate diester
formation (Scheme 1). Unfortunately, although the
reagent promotes fast coupling, it may also react with
the condensation product 3a (δP ∼71.3 and 72.7 ppm)
leading to side products. These have been identified as
the P-acylated dimer 5 (δP ∼65.5 ppm, two singlets;
isolation and 1H NMR analysis) and the phosphite
triester 6 (δP ∼141.5 ppm; comparison with an authentic
sample) (Scheme 2). The amounts of 5 and 6 varied,
depending on the ratio of the reagents used. Condensa-
tion of equimolar amounts of H-phosphonothioate 1a and
the nucleosidic component 2, in the presence of 3 equiv
of PV-Cl, produced less than 2% of 6 and ca. 8% of the
P-acylated product 5 (after 3 min). The signals related
to 5 increased with time and amounted to ca. 25% of all
nucleotidic material after 70 min. Under analogous
conditions but with 2 equiv of nucleoside 2, acylphos-
phonate 5 and the triester 6 were formed in ca. 5 and
8%, respectively (after 3 min). These amounts did not
increase significantly upon standing, probably due to a
complete consumption of PV-Cl in the acylation reaction
of 2. With the equimolar amounts of 1a and 2 and 6-fold
excess of PV-Cl, the H-phosphonothioate 3a was com-
pletely P-acylated within 40 min, while the amount of
phosphite 6 was ca. 5%.24
It is worth mentioning that in PV-Cl-promoted con-
densations to 3 we did not observe any formation of the
H-phosphonate 4. This indicates that activation of 1 with
PV-Cl is chemoselective and occurs exclusively at the
oxygen atom.

Nucleoside H-Phosphonates
Sch em e 2a
J . Org. Chem., Vol. 61, No. 19, 1996 6619
clean formation of the H-phosphonothioates 3 when used
in a 3 equiv excess over the starting material 1. In the
reactions when more than equimolar amounts of a
hydroxylic component was used, some 5′-phosphorylation
of 2 by DPCP was observed. Possible side products,
which might have resulted from the subsequent reaction
of 3a with the condensing agents (phosphite triester 6
or hypophosphates29 ), were not observed. All the inves-
tigated chlorophosphates showed also a complete chemo-
selectivity in the activation process of 1, as judged from
the absence of H-phosphonate diesters of type 4 in the
reaction mixtures.
The coupling reactions promoted by DPCP or DECP
were complete in less than 5 min (time necessary to
record the first 31P NMR spectrum), while the two other
chlorophosphates, NEP and OXP, required ∼15 and ∼40
min, respectively, to drive the reactions to completion.
As we already observed in the PV-Cl-promoted condensa-
tion, the resonances due to the product 3a were in all
instances accompanied by two very small singlets at
higher and lower fields. The intensities of these signals,
compared to those of the diesters 3, were stronger in the
NEP- and OXP-promoted condensations than in the
couplings where DPCP or DECP was used.
TP S-Cl-P r om oted Con d en sa tion s. When a mixture
of 1a and 2 (1.5 equiv) in pyridine was treated with
incremental amounts of 2,4,6-triisopropylbenzenesulfonyl
chloride30 (TPS-Cl) (Scheme 1) and the 31P NMR spec-
trum monitored, two singlets at ∼55.5 ppm due to the
starting material 1a gradually disappeared. New ab-
sorptions for the product 3a , the H-phosphonate 4a (δP
∼7.9 and 9.7 ppm), the desulfurized starting material (δP
∼3.7 ppm), and for at least three additional compounds
without P-H bonds (singlet at ∼116.5 ppm, and two pair
of singlets at ∼68 and ∼59 ppm) emerged simultaneously.
The signals due to the product 3a and the desulfurized
starting material completely disappeared within 20 min
while intensities of the other resonances increased. In
an analogous reaction when 3 equiv of TPS-Cl was added
in one portion, a similar pattern of 31P NMR signals was
observed and the desired product of the reaction, the
H-phosphonothioate diester 3a , completely disappeared
within 15 min. A higher excess of the condensing agent
(10 equiv) made the reaction proceed more rapidly
(presence of 3a could not be detected in the first 31P NMR
spectrum), and the final reaction mixture showed only
the presence of the H-phosphonate 4a and a compound
resonating at ∼69 ppm (two singlets).
a
R ) 5′-O-(dimethoxytrityl)thymidin-3′-yl. For other abbrevia-
tions, see Scheme 1.
Another interesting observation was that in all inves-
tigated reactions the 31P NMR resonances due to 3a were
always accompanied by two very small singlets at higher
and lower fields. These signals (δP ∼70.3 and 73.6 ppm),
whose intensities slowly increased with time, totally
amounted to ca. 5% of the nucleotidic material after 1 h.
Since they were present also in the reaction mixtures
when other coupling agents were used, they could not
be related to side products contained as an integral part
a condensing agent moiety (see later in the text).
Ch lor op h osp h a tes a s Con d en sin g Agen ts. Due to
low reactivity of S-nucleophiles toward the phosphorus
center,25 we expected that the use of chlorophosphates
to produce H-phosphonothioate diesters should alleviate
problems connected with subsequent reactions of the
condensation products 3 with a condensing agent. To
this end we carried out condensations of H-phosphono-
thioate 1a with a nucleosidic component (2) in the
presence of diphenyl phosphorochloridate26 (DPCP), 2-chlo-
ro-5,5-dimethyl-2-oxo-1,3,2-dioxaphosphinane27 (NEP),
diethyl phosphorochloridate27 (DECP), and bis(2-oxo-3-
oxazolidinyl)phosphinic chloride27,28 (OXP) (Scheme 1).
All these reagents, which have been previously evaluated
in the synthesis of H-phosphonate diesters, promoted
In light of our previous studies on using TPS-Cl as a
condensing agent in the synthesis of H-phosphonate
diesters,31 the above results can be interpreted as follows.
TPS-Cl promoted rather fast formation of the H-phospho-
nothioate 3a , but this compound, due to its higher
susceptibility to oxidation17,18 than that of the corre-
sponding H-phosphonate diester, rapidly underwent
subsequent reactions with the condensing agent produc-
ing compounds which lacked P-H bonds. No attempt
was made to identify these products, but on the basis of
the 31P NMR data, it seems that they are analogous to
those produced from H-phosphonate diesters and aryl-
sulfonyl derivatives.31,38 Since H-phosphonate diesters
(24) Extent of P-acylation and phosphite triester formation strongly
depend on reaction conditions. For example, the PV-Cl-promoted
condensation of 1a + 2 was significantly cleaner in acetonitrile-2,6-
lutidine (4:1, v/v) than in neat pyridine (P-acylation less than 5% and
no phosphite triesters formation; 31P NMR experiments). See also later
in the text.
(25) Dostrovsky, I.; Halmann, M. J . Chem. Soc. 1953, 502-507.
(26) Hall, R. H.; Todd, A.; Webb, R. F. J . Chem. Soc. 1957, 3291-
3296.
(27) Stro¨mberg, R.; Stawinski, J . Nucleic Acids Symp. Ser. 1987,
18, 185-188.
(28) Cabre-Castellvi, J .; Palomo-Coll, A.; Palomo-Coll, A. L. Syn-
thesis 1981, 616-620.
(29) Stec, W.; Zwierzak, A.; Michalski, J . Bull. Acad. Pol. Sci. Ser.
Chim. 1969, 17, 587-594.
(30) Lohrman, R.; Khorana, H. G. J . Am. Chem. Soc. 1966, 88, 829-
833.
(31) Regberg, T.; Stawinski, J .; Stro¨mberg, R. Nucleosides Nucle-
otides 1988, 7, 23-35.

6620 J . Org. Chem., Vol. 61, No. 19, 1996
Zain and Stawin´ski
4 were formed under these reaction conditions in signifi-
cant amount (35-45%), we could conclude that the
H-phosphonothioate monoesters 1 showed only a slight
preference for the formation of O-activated species in the
reactions with TPS-Cl.
Oth er Cou p lin g Agen ts. Utility of carbodiimides as
coupling agents for H-phosphonothioate monoesters was
assessed in the diisopropylcarbodiimide (DICD) promoted
condensation. To this end 3 equiv of DICD was added
to a mixture of 1a and 2 in pyridine containing 2 equiv
of pyridinium hydrochloride as an activator (Scheme 1).
This resulted in an immediate disappearance of 1a (first
31P NMR spectrum, ∼5 min) and formation of two
nucleotidic compounds without sulfur at the phosphorus
center, i.e., 4 and the nucleoside H-phosphonate mono-
ester (from desulfurization of 1a ) in ∼1:1 ratio. The
amount of the H-phosphonothioate diester 3a did not
exceed 5% and remained unchanged during the course
of the reaction (∼40 min). This indicated that the
presence of the H-phosphonate 4 was not due to rapid
formation of 3a followed by its desulfurization but rather
resulted, as expected,32 from the predominant S-activa-
tion of the H-phosphonothioate 1a by DICD. The nucleo-
side H-phosphonate monoester observed at the initial
stages of the condensation was most likely formed by a
partial hydrolysis of the S-activated species by spurious
water, and it was gradually disappearing with time to
form 4.
We have also evaluated utility of bis(pentafluoro-
phenyl) carbonate (PFPC), a coupling agent recently
proposed for the synthesis of H-phosphonate diesters.33
The formation of 3a in a coupling reaction between 1a
and 2 promoted by this reagent (3 equiv) in pyridine
(Scheme 1) was rather slow (85% completion after ∼40
min). Though the activation of 1 with this reagent was
completely chemoselective (O-activation; no H-phospho-
nate derivatives present), nevertheless the reaction was
accompanied by formation of unidentified side products,
as judged from the presence of several small resonances
at δP ∼50-60 ppm in the 31P NMR spectrum.
Sid e Rea ction s Not Rela ted to th e Use of a
Con d en sin g Agen t. The presence of two singlets (δP
∼70.3 and 73.6 ppm), placed almost symmetrically
around the resonances due to 3a in the 31P NMR spectra
of all investigated condensation reaction mixtures, was
puzzling. Although low intensities of these signals
(Figure 1a) prevented detailed NMR analysis, we nev-
F igu r e 1. 31P NMR spectra of the reaction between 1a and 2
promoted by DPCP in pyridine: (a) {1H}-decoupled spectrum
after ∼5 min (the highest signals due to 3a ); (b) {1H}-decoupled
spectrum after 24 h; (c) 31P-1H coupled spectrum after 24 h.
1
ertheless noticed that both of them possessed large J PH
constants (∼670 Hz) indicative of a P-H bond. Since
their chemical shifts were close to those of the diaster-
eomers of 3a , we assumed that atoms directly connected
to the phosphorus centers should be the same in both
instances. These eliminated, in principle, a vast number
of possible structures, which might have resulted from
reactions of 3 with condensing agents, and prompted us
to consider a slow, spontaneous isomerization of the
H-phosphonothioate diester 3a as a source of this side
product formation.
This assumption was substantiated by the observation
that intensities of the two singlets became comparable
to those of the product 3a when the reaction mixture,
resulting from the coupling of equimolar amounts of 1a
and 2 in the presence of 3 equiv of PV-Cl in pyridine,
was left standing overnight. In analogous condensations,
when the nucleosidic component 2 was used in excess (2
equiv), the intensities of the singlets were ∼7:1, in favor
of that resonating at the lower field (δP ∼73.6 ppm).
Inasmuch as the reaction mixtures were heavily con-
taminated by a variety of other side products, we carried
out similar experiments but with chlorophosphates as
condensing agents to be able to extract more detailed 31P
NMR data. As anticipated, when the reaction mixtures
from the DPCP-promoted condensations were left for 24
h, the whole nucleotidic material could be assigned only
to these two singlets and the two resonances due to 3a
(Figure 1b). Since intensities of the two unassigned
singlets varied, depending on the excess of the nucleosidic
component used for the condensation, we assumed that
they belonged to two compounds having two identical
substituents at the phosphorus center rather than arising
(32) Mikolajczyk, M. Chem. Ber. 1966, 99, 2083-2090.
(33) Efimov, V. A.; Kalinkina, A. L.; Chakhmakhcheva, O. G. Nucleic
Acids Res. 1993, 21, 5337-5344.

Nucleoside H-Phosphonates
J . Org. Chem., Vol. 61, No. 19, 1996 6621
from a one-spin system of phosphorus diastereomers (e.g.,
as in 3a ). The multiplicity of these signals in the 31P-
1H coupled spectrum (Figure 1c) taken in conjunction
In separate experiments we also assessed the composition
of the of crude reaction mixture resulting from the DPCP-
promoted condensation of the H-phosphonothioate 1d
with 2 (the
1
31P NMR spectra). No indication of phos-
with the chemical shift values [δP ∼73.60, J PH ) 669.0
3
1
Hz (d), J PH ) 9.8 Hz (quintet); δP ∼70.33, J PH ) 676.3
Hz (d), 3J PH ) 11.6 Hz (t)] indicated that these resonances
can be assigned to two symmetrical dinucleoside H-
phosphonothioate diesters 7 and 8, respectively (Scheme
2).
phorylation of the heteroaromatic lactam system of
guanine31 was observed under the reaction conditions.
Exp er im en ta l Section
The mechanism of formation of 7 and 8 is unknown.
We can only speculate that these compounds are formed
in a pyridine-mediated ligand-exchanged process, which
leads to scrambling of substituents at the phosphorus
center of the H-phosphonothioate diester 3a . Detailed
studies of this phenomenon are subjects of separate
investigations in this laboratory. However, in connection
with the present synthetic studies we made a preliminary
attempt to suppress this ligand exchange process. To this
end we investigated stability of the H-phosphonothioate
3a in acetonitrile containing various amounts of 2,6-
lutidine. In neat acetonitrile no ligand exchange could
be detected by 31P NMR spectroscopy, and in the presence
of 20% 2,6-lutidine, it did not exceed ∼3% after 24 h.
P r ep a r a tion of Din u cleosid e H-P h osp h on oth io-
a te Diester s 3a -d . On the basis of the above 31P NMR
studies we concluded that among the various condensing
agents investigated only chlorophosphates secured clean
formation of H-phosphonothioate diesters in a coupling
reaction between nucleoside H-phosphonothioate mo-
noesters 1 and the appropriate hydroxylic component 2
(Scheme 1). To minimize the ligand exchange process
observed for H-phosphonothioate diesters 3 in pyridine,
the condensation time should be as short as possible and
neat pyridine should be replaced by a mixture of solvents
containing only a limited amount of pyridine or other
base. To fulfil these requirements we selected for further
synthetic studies chlorophosphates that performed cou-
plings most rapidly and cleanly (i.e., DPCP and DECP)
and as a solvent mixture, acetonitrile-pyridine (4:1, v/v).
Under these reaction conditions, using either DPCP or
DECP, the condensation of H-phosphonothioate mono-
esters 1 to form the corresponding diesters 3 could be
effected within 5 min (31P NMR) without the formation
of detectable amounts of side products. Condensations
to 3 in acetonitrile containing 10 equiv of 2,6-lutidine also
proceeded rapidly with DPCP, but they were rather
sluggish when DECP was used as a coupling agent
(∼30% completion after 15 min). No scrambling of
ligands in 3 during the course of the reaction could be
detected under these conditions (31P NMR spectroscopy).
For a prospective solid phase synthesis of oligonucleo-
side H-phosphonothioates, the latter reaction conditions
(DPCP or DECP in acetonitrile with variable amounts
of 2,6-lutidine) may constitute a reasonable starting point
to formulate a working synthetic protocol. However, for
the preparation of H-phosphonothioate diesters using
“solution chemistry”, we found it more convenient to use
a limited amount of pyridine as a base and as a cosolvent.
These alleviated the problem of a nuisance interference
from 2,6-lutidine during chromatography, while keeping
the ligand exchange process in 3 below the detection level
of 31P NMR spectroscopy. All preparative syntheses of
the dinucleoside H-phosphonothioate diesters 3a -d were
thus carried out in acetonitrile-pyridine (4:1, v/v) and
afforded the desired products in 80-95% yields. Es-
sentially, no differences in yields were observed when,
instead of DECP, DPCP was used as a condensing agent.
Ma t er ia ls a n d Met h od s. Pyridine, 2,6-lutidine, aceto-
nitrile, and triethylamine (TEA) were refluxed with CaH2 and
then distilled and stored over 4 Å molecular sieves or CaH2
(TEA). Pivaloyl chloride, diethyl phosphorochloridate, diphen-
yl phosphorochloridate, 2,4,6-triisopropylbenzenesulfonyl chlo-
ride, bis(2-oxo-3-oxazolidinyl)phosphinic chloride, and diiso-
propylcarbodiimide were commercial grade (Aldrich).
The nucleoside H-phosphonothioate monoesters 1a -d were
prepared19 via nonoxidative thiation of the corresponding
H-phosphonates. Introduction of the dimethoxytrityl group
was done by standard methods.34 Protecting groups for the
heterocyclic bases are the same as those proposed previously35
for RNA synthesis via the H-phosphonate approach. 2-Chloro-
5,5-dimethyl-2-oxo-1,3,2-dioxaphosphinane36 (NEP) and bis-
(pentafluorophenyl) carbonate33 (PFPC) were prepared accord-
ing to published procedures. TLC analyses were carried out
on Merck silica gel 60 F254 precoated plates using the following
eluents: toluene-methanol (9:1 v/v; system A); chloroform-
methanol (9:1 v/v; system B).
The 31P NMR experiments concerning the formation of
H-phosphonothioate diesters 3 from 1 and 2 were carried out
in 10-mm tubes using 0.05 mmol of phosphorus-containing
compounds (1) in 2 mL of a solvent. H3PO4 (2%) in D2O was
used as external standard (coaxial inner tube). Amounts of
the hydroxylic component (2) and a coupling agent and the
solvent composition are as indicated in the text. The values
of the chemical shifts for the intermediates produced in situ
in some experiments varied ((1 ppm) depending on the
reaction conditions.
Gen er a l P r oced u r e for Syn th esis of Din u cleosid e
H-P h osp h on oth ioa te Diester s 3. A suitably protected
nucleoside 3′-H-phosphonothioate (1a -d , triethylammonium
salt, 0.5 mmol) and the nucleosidic component 2 (0.6 mmol)
were rendered anhydrous by evaporation of added pyridine.
The residue was dissolved in acetonitrile-pyridine (4:1 v/v,
15 mL) and treated with diethyl phosphorochloridate (DECP,
3 equiv) (for 1a -d ) or diphenyl phosphorochloridate (DPCP,
3 equiv) (for 1a and 1d ) during ca. 5 min (TLC analysis).
The reaction was quenched with saturated sodium chloride
(1 mL) and partitioned between toluene (2 × 50 mL) and brine
(30 mL). The organic phase was evaporated, and the residue
was purified on a silica gel column using ethyl acetate-toluene
(1:1, v/v) containing 0.02% triethylamine. This chromato-
graphic system37 was also suitable for separation of the
diastereomers of 3a -d . Fractions containing the desired
product were pooled, concentrated under reduced pressure, and
dried overnight on a vacuum line.
5′-O-(Dim et h oxyt r it yl)t h ym id in -3′-yl 3′-O-(Dim et h -
oxyt r it yl)t h ym id in -5′-yl H -P h osp h on ot h ioa t e (3a ).
(34) Schaller, H.; Weimann, G.; Lerch, B.; Khorana, H. G. J . Am.
Chem. Soc. 1963, 85, 3821-3827.
(35) Westman, E.; Stawinski, J .; Stro¨mberg, R. Collect. Czech. Chem.
Commun. (Special Issue) 1993, 58, 236-237.
(36) McConnell, R. L.; Coover, H. W. J . Org. Chem. 1959, 24, 630-
635.
(37) One should exercise some caution during chromatography of
the reaction mixtures resulting from the DECP-promoted condensa-
tions, since tetraethyl pyrophosphate has a chromatographic mobility
similar to that of the dimers 3. In cases where difficulties arise in
getting rid of this contamination, a purification using reversed phase
chromatography on the silanized silica gel is recommended.
(38) Singlet at 116.5 ppm is most likely due to the corresponding
dinucleoside S-aryl phosphorodithioate, two singlet at ∼69 ppm due
to the dinucleoside phosphorothiochloridate, and two singlets at ∼59
ppm, due to the dinucleoside phosphorothioate.

6622 J . Org. Chem., Vol. 61, No. 19, 1996
Zain and Stawin´ski
Ta ble 1. 1H NMR Ch em ica l Sh ift Va lu es (in p p m ) of Som e Dia gn ostic Sign a ls of Din u cleosid e H-P h osp h on oth ioa tes 3a
P-H (1J PH
)
1′-H
3′-H
2-H
5-Me
6-H
8-H
3a (RP)
3a (SP)
3b (RP)
3b (SP)
3c (RP)
3c (SP)
3d (RP)
3d (SP)
7.64 (668.2 Hz), d
7.69 (664.6 Hz), d
7.62 (671.5 Hz), d
7.69 (662.7 Hz), d
7.65 (667.0 Hz), d
7.70 (667.1 Hz), d
7.71 (668.2 Hz), d
7.72 (664.6 Hz), d
6.35
6.40
6.23
6.30
6.35
6.38
6.10
6.15
5.47
5.45
5.33
5.34
5.50
5.50
5.45
5.45
1.50, 1.85c
1.47, 1.88c
1.85
1.88
1.88
1.90
1.86
1.89
7.59
7.57
8.09
8.06
b
b
8.57
8.62
7.74
7.83
8.10
8.13
a
b
Signals from the methoxy grops of the dmt moieties appeared at ∼3.7 ppm. Signals 5-H buried in multiplets of the aromatic protons.
c Protons in the thymidin-5′-yl unit of the dimer.
Yield: 95%. 3a (RP)39 (faster moving isomer): Rf ) 0.20
(system A); 31P NMR (EtOAc, δ in ppm) 71.38 (1J PH ) 673.9
data, see Table 1. 3d (SP) (slower moving isomer): Rf ) 0.22
(system A) 31P NMR: (EtOAc, δ in ppm) 72.11 (1J PH ) 675.1
3
3
Hz, J PH ) 9.8 Hz, dq). 3a (SP) (slower moving isomer): Rf )
Hz, J PH ) 9.2 Hz, dq). For 1H NMR data, see Table 1.
0.18 (system A); 31P NMR (EtOAc, δ in ppm) 72.32 (1J PH
)
HRMS(FAB) (mixture of diastereomers): calcd for
C70H67O15N7PS (M - H+) 1308.4154, found (M - H+) 1308.4187.
Syn th esis of 5′-O-(Dim eth oxytr ityl)th ym id in -3′-yl 3′-
1
670.8 Hz, 3J PH ) 10.0 Hz, dq). For H NMR data, see Table 1.
HRMS(FAB) (mixture of diastereomers): calcd for C62H62O14N4-
PS (M - H+) 1149.3721, found (M - H+) 1149.3748.
O-(Dim et h oxyt r it yl)t h ym id in -5′-yl
2,2,2-Tr im et h yl-
5′-O-(Dim e t h oxyt r it yl)-N 4-p r op ion yld e oxycyt id in -
3′-yl 3′-O-(Dim eth oxytr ityl)th ym id in -5′-yl H-P h osp h on o-
th ioa te (3b). Yield: 85%. 3b (RP) (faster moving isomer):
Rf ) 0.22 (system A); 31P NMR (EtOAc, δ in ppm) 71.15 (1J PH
) 674.6 Hz, 3J PH ) 9.9 Hz, dq) For the 1H NMR data, see Table
1. 3b (SP) (slower moving isomer): Rf ) 0.20 (system A); 31P
a cetylp h osp h on oth ioa te (5). 5′-O-(Dimethoxytrityl)thymi-
dine 3′-H-phosphonothioate 1a (triethylammonium salt, 0.3
mmol) and 3′-O-(dimethoxytrityl)thymidine 2 (0.33 mmol) were
rendered anhydrous by evaporation of added pyridine. The
residue was dissolved in pyridine (2 mL) and treated with
pivaloyl chloride (6 equiv) during 1 h. The reaction was
quenched with brine, the mixture was extracted with meth-
ylene chloride, and the P-acylated product 5 was isolated by
silica gel chromatography using ethyl acetate-toluene (1:1,
v/v) containing 0.02% TEA as eluent. 31P NMR parameters
of 5 were found to be identical with those of the postulated
side product observed by 31P NMR spectroscopy during the
coupling reaction of 1a with 2 promoted by PV-Cl (see the text).
5: Rf 0.23 and 0.19 (system A) or 0.61 and 0.58 (system B);
31P NMR (EtOAc, δ in ppm) 65.88 (3J PH ) 8.6 Hz, q), 65.56
(3J PH ) 7.4 Hz, q); 1H NMR (CDCl3, δ in ppm, mixture of
diastereomers, selected signals) 7.05 (1H, 6-H), 6.32 (2H, m,
1′-H and 1′-H*), 5.46 (1H, m, 3′-H), 1.84 (3H, s, 5-Me*) 1.46
(3H, s, 5-Me), 1.27 and 1.22 (9H, 2s, Me3C). HRMS(FAB)
(mixture of diastereomers): calcd for C65H66O13N4PS (M - H+)
1173.4085, found (M - H+) 1173.4028.
3
NMR (EtOAc, δ in ppm) 72.19 (1J PH ) 668.9 Hz, J PH ) 9.8
1
Hz, dq). For H NMR data, see Table 1. HRMS(FAB) (mixture
of diastereomers): calcd for C64H65O14N5PS (M - H+) 1190.3987,
found (M - H+) 1190.4012.
5′-O-(Dim e t h ox y t r it y l)-N 6-b u t y r y ld e o xy a d e n o sin -
3′-yl 3′-O-(Dim eth oxytr ityl)th ym id in -5′-yl H-P h osp h on o-
th ioa te (3c). Yield: 81%. 3c (RP) (faster moving isomer): Rf
) 0.22 (system A); 31P NMR (EtOAc, δ in ppm) 71.13 (1J PH
)
3
672.6 Hz, J PH ) 9.8 Hz, dq). For 1H NMR data, see Table 1.
3c (SP) (slower moving isomer): Rf ) 0.19 (system A); 31P NMR
3
(EtOAc, δ in ppm) 72.00 (1J PH ) 670.8 Hz, J PH ) 10.4 Hz,
1
dq). For H NMR data, see Table 1. HRMS(FAB) (mixture of
diastereomers): calcd for C66H67O13N7PS (M - H+) 1228.4255,
found (M - H+) 1228.4229.
5′-O -(D im e t h o x y t r it y l)-N 2-(p h e n o x y a c e t y l)d e o x y -
gu a n osin -3′-yl 3′-O-(Dim eth oxytr ityl)th ym id in -5′-yl H-
P h osp h on oth ioa te (3d ). Yield: 82%. 3d (RP) (faster moving
isomer): Rf ) 0.24 (system A) 31P NMR (EtOAc, δ in ppm)
Ack n ow led gm en t. We are indebted to Prof. Per J .
Garegg for his interest and Dr. Tomas Szabo´ for
preparation of bis(pentafluorophenyl) carbonate. Finan-
cial support from the Swedish Research Council for
Engineering Sciences and the Swedish Natural Science
Research Council is gratefully acknowledged
3
71.58 (1J PH ) 674.63 Hz, J PH ) 10.7 Hz, dq). For 1H NMR
(39) A tentative assignment of the configurations at the phosphorus
center was done on the basis of the relative mobilities of the isomers
on silica gel and the relative positions of their resonances in 31P NMR
spectra.11,20,21
J O960810M