Inverse phosphotriester DNA synthesis using photochemically-removable dimethoxybenzoin phosphate protecting groups

Article abstract of DOI:10.1021/jo951593c

A method has been developed to prepare short DNA sequences using light to deprotect a nucleoside 3′-phosphotriester, generating a phosphodiester useful for coupling with a free 5′-OH-nucleotide. The dimethoxybenzoin group is used as the photochemically-removable protecting group for the 3′-phosphate. Cyanoethyl is most effective as the second protecting group on the phosphodiester. Because the method is directed at the preparation and use of the DNA sequences while still bound to the support, allyl and allyloxycarbonyl protecting groups are used for the nitrogenous bases since, based on the work of Hayakawa and Noyori, they can be removed without cleaving the DNA from the support. Two simple trinucleotides have been prepared in solution using this method. It has been demonstrated that the photochemical deprotection conditions do not lead to the formation of cyclobutane dimers from adjacent T residues.

Full text of DOI:10.1021/jo951593c

J . Org. Chem. 1996, 61, 2129-2136
2129
In ver se P h osp h otr iester DNA Syn th esis Usin g
P h otoch em ica lly-Rem ova ble Dim eth oxyben zoin P h osp h a te
P r otectin g Gr ou p s
Michael C. Pirrung,*,1 Lara Fallon, David C. Lever, and Steven W. Shuey
Department of Chemistry, P. M. Gross Chemical Laboratory, Duke University,
Durham, North Carolina 27708-0346
Received August 31, 1995 (Revised Manuscript Received December 13, 1995X)
A method has been developed to prepare short DNA sequences using light to deprotect a nucleoside
3′-phosphotriester, generating a phosphodiester useful for coupling with a free 5′-OH-nucleotide.
The dimethoxybenzoin group is used as the photochemically-removable protecting group for the
3′-phosphate. Cyanoethyl is most effective as the second protecting group on the phosphodiester.
Because the method is directed at the preparation and use of the DNA sequences while still bound
to the support, allyl and allyloxycarbonyl protecting groups are used for the nitrogenous bases
since, based on the work of Hayakawa and Noyori, they can be removed without cleaving the DNA
from the support. Two simple trinucleotides have been prepared in solution using this method. It
has been demonstrated that the photochemical deprotection conditions do not lead to the formation
of cyclobutane dimers from adjacent T residues.
In tr od u ction
method10 that can be used for the preparation of DNA
arrays has recently been reported.
The collection of the complete sequence of human DNA
(Human Genome Project)2 and its use once it is available
will require powerful new techniques in genetic analysis.
One of the most widely-discussed technologies for this
purpose is based on “DNA chips”, miniaturized arrays
of diverse DNA sequences that can be used to detect
complementary sequences in unknown nucleic acid
samples. The preparation of DNA arrays can be ac-
complished by robotic immobilization of pre-synthesized
oligonucleotides3,4 or by in situ synthesis.5 The latter
provides an advantage in preparing arrays of thousands
of oligonucleotides because the principles of combinatorial
chemistry6 can be used to prepare sequences in parallel
in relatively few chemical steps. Miniaturization of DNA
arrays furthermore makes the manipulation of complete
libraries of thousands or millions of sequences physically
convenient and minimizes dilution of the DNA sample,
providing a high concentration and therefore more effec-
tive binding kinetics. The primary means to prepare
miniaturized arrays of polymeric molecules is light-
directed synthesis.7 Development of light-based ver-
sions8,9 of the traditional phosphoramidite DNA synthesis
The key to adapting a solid-phase synthesis method
to the light-directed format is a photochemically-remov-
able protecting group that is bound to the surface. When
this work was begun, effective photochemically-remov-
able alcohol protecting groups that could be used to
replace a conventional dimethoxytrityl (DMTr) 5′-protect-
ing group were not well-known, though we have subse-
quently developed several.11 We instead considered an
approach involving two significant deviations from stand-
ard practice in oligonucleotide synthesis. By attaching
the 5′-hydroxyl of the DNA chain to the solid support, a
photoremovable phosphate protecting group12 could be
used for a 3′-phosphate in the phosphotriester13 coupling
method (Figure 1). This format requires activation of a
3′-phosphodiester bound to the support, but still permits
the reaction of an incoming 5′-hydroxyl with a 3′-
activated phosphate and has been reported.14 This
inverse 5′ f 3′ synthesis permits a photoremovable group
for the phosphate rather than the hydroxyl to be used.
The tethering of the DNA to the support through the 5′
end also offers the opportunity for enzymatic reactions
at free 3′ hydroxyls of the resulting short oligonucleotides.
One concern with the use of oligonucleotide libraries
for genetic analysis is the low melting temperatures of
short oligonucleotides and possible mishybridization
because of the short recognition sequence and end-fraying
effects.15 Lehrach has addressed these issues through
X Abstract published in Advance ACS Abstracts, March 1, 1996.
(1) Fellow of the J ohn Simon Guggenheim Memorial Foundation,
1994-95.
(2) Yager, T. D.; Zewert, T. E.; Hood, L. E. Acc. Chem. Res. 1994,
27, 94-100.
(3) Livshits, M. A.; Ivanov, I. B.; Mirzabekov, A. D.; Florentiev, V.
A. Mol. Biol. (Moscow) 1992, 26, 1298-1313. Pevzner, P. A.; Lysov,
Yu. P.; Khrapko, K. R.; Belyavskii, A. V.; Florent’ev, V. L.; Mirzabekov,
A. D. J . Biomol. Struct. Dyn. 1991, 9, 399-410. Mirzabekov, A. D.
Bioorg. Khim. 1992, 18, 1361-74. Eggers, M. D.; Hogan, M. E.; Reich,
R. K.; Lamture, J . B.; Beattie, K. L.; Hollis, M. A.; Ehrlich, D. J .;
Kosicki, B. B.; Shumaker, J . M.; et al. Proc. SPIE-Int. Soc. Opt. Eng.
1993, 1891 (Proceedings of Advances in DNA Sequencing Technology,
1993), 113-26.
(7) Pirrung, M. C.; Read, J . L.; Fodor, S. P. A; Stryer, L. US Patent,
5,143,854, Sept 1, 1992. Fodor, S. P. A.; Read, J . L.; Pirrung, M. C.;
Stryer, L.; Liu, A. T.; Solas, D. Science 1991, 251, 767.
(8) Pirrung, M. C.; Bradley, J .-C. J . Org. Chem. 1995, 60, 6270.
(9) Pease, A. C.; Solas, D.; Sullivan, E. J .; Cronin, M. T.; Holmes,
C. P.; Fodor, S. P. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 5022-
5026.
(4) Shumaker, J . M.; Metspalu, A.; Caskey, C. T. Hum. Mutat. 1996,
in press.
(10) Caruthers, M. H. Science 1985, 230, 281.
(11) Pirrung, M. C.; Lee, Y. R. J . Org. Chem. 1993, 58, 6961. Pirrung,
M. C.; Bradley, J .-C. J . Org. Chem. 1995, 59, 1116.
(12) Pirrung, M. C.; Shuey, S. W. J . Org. Chem. 1994, 59, 3890 and
references cited therein.
(13) Gait, M. J . Oligonucleotide Synthesis; IRL Press: Washington,
DC, 1984.
(14) Pochet, S.; Huynh-Dinh, T.; Igolen, J . Tetrahedron 1987, 43,
3481-90. Gorn, V. V.; Zarytova, V. F.; Yarmolinskaya, E. V. Bioorg.
Khim. 1983, 9, 1511-7.
(5) Southern, E. M.; Maskos, U. Nucleic Acids Res. 1993, 21, 4663-
9. Southern, E. M.; Maskos, U. Nucleic Acids Res. 1993, 21, 2269-70.
Southern, E. M.; Maskos, U. Nucleic Acids Res. 1993, 21, 2267-8.
Southern, E. M.; Maskos, U. Nucleic Acids Res. 1992, 20, 1675-8.
Southern, E. M.; Maskos, U. Nucleic Acids Res. 1992, 20, 1679-84.
Matson, R. S.; Rampal, J . B.; Coassin, P. J . Anal. Biochem. 1994, 217,
306-10.
(6) Pirrung, M. C. Chemtracts: Org. Chem. 1995, 8, 3-8.
0022-3263/96/1961-2129$12.00/0 © 1996 American Chemical Society

2130 J . Org. Chem., Vol. 61, No. 6, 1996
Pirrung et al.
of deoxyguanine are also required (Chart 1). While
procedures to prepare these or very similar compounds
are available in the literature, difficulties were encoun-
tered in reproducing some of these methods. Modification
of the routes to include in most cases the protection of
the 5′ alcohol with tert-butyldiphenylsilyl (TBDPS) even-
tually provided pathways that were reproducible and
provided 5-10 g quantities of material. These routes
relied on the deprotection of the 5′-O-TBDPS group (and
other silyl groups) using our recently-described triethyl-
amine trihydrofluoride procedure.21 New protocols for
the preparation of protected A, G, and C are provided in
the Experimental Section. The synthesis of 5-meth-
yldeoxycytidine is based on a synthetic route to 4-sub-
stituted thymidine nucleosides through formation of a
4-triazolothymidine intermediate (eq 1).22 tert-Butyldi-
F igu r e 1.
F igu r e 2.
the use of 5-methylcytidine (5MC), which increases
discrimination between target and nontarget DNA16 in
probes used to map the Herpes simplex genome by
hybridization.17 The use of inosine as a “universal base”18
flanking the analyzing sequence of the probe can increase
hybrid stability19 without increasing the recognition
sequence, the length of the probes, and therefore the
length of the synthesis. It also places mismatches that
would otherwise be located at the end of a hybridization
probe in a duplex region where the thermodynamic cost
will be greater (Figure 2).
This report focuses on an inverse 5′ f 3′ photochemical
phosphotriester DNA synthesis method using our previ-
ously-developed dimethoxybenzoinphosphate protecting
group, including the four natural bases and the modified
bases inosine and 5-methylcytidine. It also exploits the
allyloxycarbonyl protecting group that has previously
been developed by Hayakawa and Noyori20 that can be
removed without cleaving the DNA from the support on
which it is prepared.
methylsilyl (TBDMS)-protected thymidine (1) is obtained
in 96% yield and converted to the 4-triazolo derivative
by addition to a suspension of 1,2,4-triazole, triethyl-
amine, and phosphorus oxychloride in acetonitrile. Con-
version to the desired bis-TBDMS-protected 5-methyl-
2′-deoxycytidine 2 can be effected in 94% yield (chro-
matographed) by stirring with 1:3 ammonium hydroxide/
1,4-dioxane for 4 h. Protection of the amino function can
be accomplished in 90% yield using the easily-prepared
hydroxybenzotriazole active carbonate AOC-OBt. De-
silylation gives 3, which can be selectively protected at
the 5′ alcohol with a TBDPS group.
The preparation of the 3′-O-(cyanoethyl dimethoxy-
benzoin phosphate) esters of five of these specially-
protected nucleosides can be accomplished by conversion
first to the 3′-O-(cyanoethyl phosphoramidite) and then
condensation with dimethoxybenzoin (eq 2). As expected,
Resu lts
The first need was a source of nucleoside monomers
deoxyadenosine, deoxycytidine, and 5-methyldeoxycyti-
dine with allyloxycarbonyl (AOC) protection on their
exocyclic amino groups. Allyl ether protection of O6
(required to prevent its activation by the condensing
agent) and AOC protection of the exocyclic amino group
(15) Wood, W. I.; Gitscher, L. A.; Lasky, L. A.; Lawn, R. M. Proc.
Natl. Acad. Sci. U.S.A. 1985, 82, 1585-1588.
(16) Bugg, C. E.; Thomas, J . M. Biopolymers 1971, 10, 175-219.
(17) Hoheisel, J . D.; Craig, A. G.; Lehrach, H. J . Biol. Chem. 1990,
265, 16656-60.
(18) Crick, F. H. C. J . Mol. Biol. 1966, 19, 548-555. Ohtsuka, E.;
Matsuki, S.; Ikehara, M.; Takahasi, Y.; Matsubara, K. J . Biol. Chem.
1985, 260, 2605-2608.
these reactions are exquisitely sensitive to moisture and
(19) Martin, F. H.; Castro, M. M.; Aboul-ela, F.; Tinoco, I. Nucl. Acids
Res. 1985, 24, 8927-8938.
(20) Hayakawa, Y.; Kato, H.; Uchiyama, M.; Kajino, H.; Noyori, R.
J . Org. Chem. 1986, 51, 2400-2. Hayakawa, Y.; Kato, H.; Noyori, R.
Tetrahedron Lett. 1985, 26, 6505-8. Hayakawa, Y.; Wakabayashi, S.;
Kato, H.; Noyori, R. J . Am. Chem. Soc. 1990, 112, 1691-1696.
(21) Pirrung, M. C.; Shuey, S. W.; Lever, D. C.; Fallon, L. Bioorg.
Med. Chem. Lett. 1994, 4, 1345.
(22) Reese, C. B.; Skone, P. A. J . Chem Soc. Perkin Trans. 1 1984,
1263-1271.

Inverse Phosphotriester DNA Synthesis
J . Org. Chem., Vol. 61, No. 6, 1996 2131
Ta ble 1.
Ch a r t 1
base
T
C
A
G
MeC
I
allyl/AOC
CE/AOC
71
74
67
65
61
68
60
89
65
50
58
40
work required an (allyloxy)chlorophosphine reagent that
is not available commercially. (Allyloxy)dichlorophos-
phine could be prepared by slow addition of freshly-
distilled allyl alcohol to neat, freshly-distilled PCl3.
Yields for this reaction were never very high, but it can
be successfully conducted on a 1.0 mol scale, so that
adequate quantities of the compound are available.
Preparation of a chlorophosphoramidite analogous to the
commercially-available cyanoethoxy(N,N-diisopropyl-
amino)chlorophosphine proved more difficult. Allyloxy-
(N,N-diisopropylamino)chlorophosphine could be pre-
pared by the addition of 2 equiv of diisopropylamine to
(allyloxy)dichlorophosphine, but filtration of the amine
hydrochloride was inevitably accompanied by large
amounts of decomposition, and the product also could not
be distilled. The removal of the amine hydrochloride
could be obviated by the use of 1 equiv of (trimethylsilyl)-
diisopropylamine instead of the free amine. The TMSCl
thus produced could be removed by vacuum distillation,
yielding material which, although not very pure, was
nevertheless usable in the preparation of allyl nucleoside
diisopropyl phosphoramidites. Yields were lower than
in the preparation of the corresponding cyanoethyl
derivatives. Another solution to this problem was to
switch to the corresponding N,N-diethylamino deriva-
tives. (Allyloxy)(N,N-diethylamino)chlorophosphine proved
distillable under high vacuum and could be isolated as a
clear, colorless liquid, pure by 1H and 31P NMR, and used
to derivatize nucleosides. The resulting allyl diethyl
phosphoramidites cannot be chromatographed, however,
as they are sufficiently more reactive than the corre-
sponding diisopropyl phosphoramidites in that they are
hydrolyzed despite attempts to prevent it by adding
triethylamine to the solvent or using triethylamine-
pretreated silica gel. These phosphoramidites could be
partially purified by precipitation from petroleum ether,
however, and if the 5′ protecting group were dimethoxy-
trityl, the resulting product would be a foam. If the 5′
protecting group were TBDPS, the phosphoramidites
would form thick gums rather than foams. These gums
were generally used in subsequent coupling reactions
without any purification, in the same flask in which they
were prepared, since attempts to precipitate these ma-
terials inevitably led to large losses. The yields in the
conversion of the cyanoethyl and allyl phosphoramidites
to the 3′-O-(dimethoxybenzoin phosphate) derivatives are
collected in Table 1.
require azeotropic drying of the tetrazole, phosphor-
amidite, and benzoin. The sixth 3′-O-(cyanoethyl
dimethoxybenzoin phosphate) ester, of T, is prepared
from the condensation of the commercially-available
T-phosphoramidite as was described in our earlier report
on dimethoxybenzoin phosphates. The choice was made
to use the racemic 3′,5′-dimethoxybenzoin based on the
difficulty in obtaining the optically-active material on
very large scale. Racemic 3′,5′-dimethoxybenzoin can be
prepared by a fairly simple, two-step process and is
crystalline, which facilitates its purification. The prepa-
ration is also amenable to scale up, and 25-30 g batches
of this material have been made repeatedly in our
laboratory. The only disadvantage in using racemic
benzoin is the formation of four diastereomers of the
desired mononucleotides instead of two (the phosphorus
center is also stereogenic and under no stereochemical
1
control). Since H NMR is virtually useless for charac-
terization and purity evaluation of these compounds even
when only two diastereomers are present, the benefits
of using chiral benzoin do not outweigh the disadvantages
presented by the difficulty of its preparation.
Four of these allyl/AOC-protected benzoin nucleotides
were characterized by UV spectroscopy, and the half-lives
for their photochemical deprotection (Rayonet photo-
chemical reactor, 350 nm phosphor lamps) to the corre-
sponding phosphate salts were measured. Table 2 lists
the UV maxima, extinction coefficients, and half-lives (at
the given concentration) for deprotection of the four
monomers with a free 5′ hydroxyl group, as well as the
5′ acetylated thymidine and deoxycytidine derivatives.
The values for the half-lives of deprotection are very
similar to those found in our earlier work. The two
deoxycytidine derivatives show the expected longer wave-
length UV absorption, but the absorption at 360 nm (the
wavelength for the deprotection step) is much less than
with the benzoyl-protected deoxycytidine studied earlier
Though they were not ultimately useful in coupling
reactions, 3′-O-(allyl dimethoxybenzoin phosphate) esters
of these six nucleosides were also prepared (eq 3). This

2132 J . Org. Chem., Vol. 61, No. 6, 1996
Pirrung et al.
Ta ble 2.
ment. (The excess of the phosphate salt could be removed
by base washing of the reaction mixture.) We then turned
to the improved phosphotriester coupling conditions
developed by Efimov utilizing ethoxypyridine N-oxide and
triisopropylbenzenesulfonyl chloride.25 These reactions
were generally successful with cyanoethyl diesters, but
do not couple the extremely recalcitrant allyl diesters.
In the initial example, the purified yield of G-T is 75%.
Repetition of an irradiation/coupling cycle with the
protected deoxycytidine mononucleotide gives the pro-
tected G-T-C trinucleotide in 69% yield.
concn
(mM)
half-
life, s
UV λmax
extinction
coeff
compound
thymidine
(nm)
4.65
3.50
109
105
276
263
316
278
274
279
262
316
20400
29100
10100
25600
25400
17700
19700
6960
cytidine
adenosine
guanine
5′-acetyl thymidine
5′-acetyl cytidine
3.17
2.68
2.66
2.75
95
81
82
71
To ensure that the photochemical deprotection reaction
conditions do not result in the degradation of the result-
ing DNA strand by formation of T-T cyclobutane dimers,
a T-T-T trinucleotide was prepared (eq 5), with 68%
and very close to zero. It is not thought that the longer
wavelength absorption affects the rate of deprotection.
The observed half-lives support this belief, since the
values for the deoxycytidine derivatives are in line with
the remaining nucleotides.
With these protected mononucleotides in hand, their
coupling in solution to form short, protected oligonucleo-
tides was investigated. In place of a 5′-link to the solid
support, a silyl group was used which was stable to the
conditions of deprotection and coupling. Dimethoxytrityl
is not stable, since the phosphodiester is acidic enough
to slowly remove it. The dimethoxybenzoin group was
removed by irradiation and the phosphodiester converted
to its triethylammonium salt (eq 4). The first coupling
coupling yield in the first step and 53% coupling yield in
the second step. The resulting trinucleotide was de-
graded in acid and analyzed for the presence of dimeric
heterocycle using the method of Greenberg.26 None was
seen.
Con clu sion
Monomers and coupling chemistry have been provided
for the photochemical preparation of DNA using phos-
photriester chemistry. The two syntheses described
reach the limit of oligomers that can be prepared using
solution-phase synthesis because higher oligomers and
intermediates leading to them demonstrate much-
reduced solubility. This work completes a first step
toward the development of a novel chemical combinato-
rial DNA synthesis methodology; solid-phase synthesis
of single sequences will lay the groundwork for prepara-
tion of permutational libraries27 of oligonucleotides on
solid phase. This method will permit the synthesis of
short oligonucleotide libraries attached to the support
through their 5′-ends, which will leave their 3′-ends free
for further manipulations.
Exp er im en ta l Section
Gen er a l. All starting materials were purchased from
Aldrich Chemical Co. except 5′-O-(dimethoxytrityl) nucleo-
sides, which were from Cruachem Inc. and were used without
further purification. Dichloromethane, acetonitrile, benzene,
and pyridine were freshly-distilled from calcium hydride. THF
and diethyl ether were distilled from sodium/benzophenone
ketyl. Triethylamine and diisopropylamine were distilled from
sodium and stored under argon. All reactions were carried
out under an atmosphere of argon in glassware which was
oven-dried and/or flame-dried. Chromatography was per-
formed using EM Reagents 0.042-0.063 mm grade silica gel
(Kieselgel 60). NMR spectra were recorded at 121 MHz for
phosphorus and 300 MHz for proton. Phosphorus shifts are
reported referenced to 85% phosphoric acid, and proton shifts
relative to internal CHCl3. Elemental analyses were con-
ducted by Atlantic Microlabs, Inc. High resolution mass
spectrometry was performed at the University of North
Carolina and the University of South Carolina.
reagent investigated was 1-(mesitylenesulfonyl)-3-nitro-
1,2,4-triazole/N-methylimidazole.23 While it generally
accomplishes phosphotriester couplings with 2-chloro-
phenyl esters in high yield, coupling is slow with cyano-
ethyl diesters.24 In one instance, we prepared the T-T
dinucleotide with cyanoethyl internucleotide protection
(vide infra) in >95% yield by using a 10-fold molar excess
of the phosphate salt over the alcohol, treating this
coupling like a conventional 3′ f 5′ solid-phase experi-
N4-(Allyloxyca r bon yl)-2′-d eoxycytid in e. To a slurry of
14.3 g of 2′-deoxycytidine (twice azeotroped from 50 mL of
pyridine) in 160 mL of pyridine, cooled in an ice bath, was
(23) Chandrasegaran, S.; Murakami, A.; Kan, L.-s. J . Org. Chem.
1984, 49, 4951-57. Zarytova, V. F.; Knorre, D. G. Nucleic Acids Res.
1984, 12, 2091-2110.
(24) 2-Chlorophenyl esters could not be used because of their lability.
They undergo an intramolecular cyclization from the benzoin carbonyl
group to form a cyclic phosphate that leads to hydrolysis of the triester.
(25) Efimov, A.; Chakhmakhcheva, O. G. Chem. Scr. 1985, 26.
(26) Greenberg, M. M.; Gilmore, J . L. J . Org. Chem. 1994, 59, 746.
(27) Pirrung, M. C. Chemtracts: Org. Chem. 1993, 6, 88-91.
Pirrung, M. C. Chemtracts: Org. Chem. 1995, 8, 5-12.

Inverse Phosphotriester DNA Synthesis
J . Org. Chem., Vol. 61, No. 6, 1996 2133
added 40 mL of TMSCl over 10 min After 15 min the cold bath
was removed and stirring was continued for 1.5 h. Allyl
1-benzotriazolyl carbonate (Aldrich) (27.5 g) was added por-
tionwise, and the residue was washed in with 40 mL of
pyridine. After stirring overnight the mixture was poured into
300 mL of saturated NaHCO3 and stirred for 2 h. Water (500
mL) was added, and the mixture was extracted with CH2Cl2
(150 mL) followed by 1:1 CH2Cl2/pyridine (4 × 250 mL). The
combined extracts were washed with brine, dried (Na2SO4),
and concentrated, finally under high vacuum, to give a cream-
colored, greasy solid, which was recrystallized from EtOAc to
give 10.8 g (55%) of the product as a white powder. This
material showed spectroscopic properties identical to those
reported by Noyori. 1H NMR (CDCl3): δ 2.04 (m, 1H), 2.29
(m, 1H), 3.61 (m, 2H), 4.24 (m, 1H), 5.05 (m, 1H), 5.22-5.39
(m, 2H), 5.98 (m, 1H), 6.17 (t, 1H, J ) 6.0), 7.05 (d, 1H, J )
8), 8.32 (d, 1H, J ) 8). Anal. Calcd for C13H17N3O6: C, 50.16;
H, 5.51; N, 13.50. Found: C, 50.16; H, 5.56; N, 13.50.
5′-O-(ter t-Bu tyld ip h en ylsilyl)-N4-(a llyloxyca r bon yl)-2′-
d eoxycytid in e. To a solution of the above yellow grease in
150 mL of pyridine was added 11.0 mL of TBDPSCl via syringe
over 10 min The reaction was stirred for 3 d, pyridine was
removed under vacuum, and the residue was triturated with
toluene to remove silanol. Flash chromatography on silica gel
with a step gradient from 2:98-5:95-10:90 EtOH/CH2Cl2 gave
8.52 g (55% based on 2′-deoxycytidine) of product as a white
foam. In addition ∼10 g (40%) of 3′,5′-bis-silylated material
was obtained as a yellow grease. 1H NMR (CDCl3): δ 1.07
(9H, s), 2.19 (1H, m), 2.58 (1H, m), 3.84-4.05 (2H, m), 4.08
(1H, m), 4.49 (1H, br), 4.67 (2H, d, J ) 5), 5.25-5.4 (2H, m),
5.92 (1H, m), 6.30 (1H, t, J ) 6.5), 7.35-7.45 (6H, m), 7.6-
7.75 (4H, m), 8.25 (1H, br). Anal. Calcd for C29H35N3O6Si:
C, 63.36; H, 6.42; N, 7.64. Found: C, 63.31; H, 6.38; N, 7.70.
3′,5′-Bis(ter t-Bu t yld im et h ylsilyl)-2′-d eoxy-5-m et h yl-
cytid in e (2). To a solution of 10 g of 2′-deoxythymidine in
DMF (80 mL) were added 12.37 g of imidazole and 13.69 g of
tert-butyldimethylsilyl chloride. After stirring overnight, the
solution was poured into ether (350 mL) and washed with
saturated NaHCO3 and brine and was dried (MgSO4). The
solution was concentrated to give 18.83 g of product as a white
solid (96%). This material was used without further purifica-
tion.
To a solution of 1,2,4-triazole (55.63 g) and triethylamine
(120 mL) in CH3CN (100 mL), cooled in an ice bath, was added
dropwise phosphorus oxychloride (16.2 mL). The slurry was
stirred for 30 min, and a solution of protected 2′-deoxythymi-
dine (10 g) in CH3CN was added over 15 min at 0 °C. After
stirring overnight at room temperature, the mixture was
poured into CH2Cl2, washed with saturated NaHCO3 and
brine, and dried (NaSO4). The solution was concentrated to
give the triazolyl intermediate as an orange solid. The orange
solid was immediately treated with 1:3 v/v NH4OH/dioxane
(400 mL) and was stirred at room temperature for 4 h. The
mixture was poured into ether, washed with saturated NaH-
CO3 and brine, and dried (MgSO4). Flash chromatography on
silica gel eluting with 5:95 EtOH:CH2Cl2 gave 9.4 g of the
product as a white solid (97% based on protected thymidine).
1H NMR (CDCl3): δ 0.035, 0.083 (12H, 2 × s), 0.857, 0.895
(18H, 2 × s), 1.89 (3H, s), 1.95 (1H, m), 2.34 (1H, m), 3.68-
3.89 (2H, m), 3.91 (1H, m), 4.34 (1H, m), 6.29 (1H, t, J ) 6.3),
7.5 (1H, s)). Anal. Calcd for C22H43N3O4Si2: C, 56.25; H, 9.23;
N, 8.94. Found: C, 56.18; H, 9.25; N, 8.90.
3.95 (2H, m), 3.96 (1H, m), 4.57 (1H, m), 4.63 (2H, d, J ) 3),
5.20-5.38 (2H, m), 5.97 (1H, m), 6.15 (1H, t, J ) 6.3). Anal.
Calcd for C14H19N3O6‚H2O: C, 48.98; H, 6.17; N, 12.24.
Found: C, 49.24; H, 6.69; N, 12.30.
5′-O-(ter t-Bu tyld ip h en ylsilyl)-N4-(a llyloxyca r bon yl)-2′-
d eoxy-5-m eth ylcytid in e. To a solution of N4-AOC-2′-deoxy-
5-methylcytidine (0.746 g) in 20 mL of pyridine was added tert-
butyldiphenylsilyl chloride (1.01 mL). After stirring overnight,
the pyridine was removed under vacuum and the resulting
oil was taken up in CH2Cl2, washed with saturated NaHCO3
and brine, and dried (NaSO4). Flash chromatography on silica
gel with a step gradient from 3:97-5:95 EtOH:CH2Cl2 gave
the 1.23 g of the product as a white foam (95%). 1H NMR
(CDCl3): δ 1.07 (9H, s), 1.69 (3H, s), 2.22 (1H, m), 2.43 (1H,
m), 3.83-3.94 (2H, m), 3.99 (1H, s), 4.53 (1H, m), 4.64 (2H, d,
J ) 5.7), 5.26-5.38 (2H, m), 5.97 (1H, m), 6.32 (1H, t, J )
6.0), 7.36-7.49 (6H, m), 7.64 (4H, d, J ) 7.5). Anal. Calcd
for C30H37N3O6Si‚1/2H2O: C, 62.91; H, 6.69; N, 7.33. Found:
C, 63.10; H, 6.68; N, 7.08.
N6-(Allyloxyca r bon yl)-2′-d eoxya d en osin e. To a solution
of 7.0 g of 2′-deoxyadenosine (twice azeotroped from 40 mL of
pyridine) in 100 mL of pyridine was added 10.0 g of 1,3-
dichloro-1,1,3,3-tetraisopropyldisiloxane. After stirring over-
night the solvent was removed under reduced pressure and
the residue dissolved in 300 mL of CH2Cl2 and washed with
saturated NaHCO3 (2 × 150 mL) and brine (150 mL). The
resulting solution was dried (Na2SO4) and concentrated to give
a white foam that was used without purification.
To a solution of 4.0 g of 1H-tetrazole in 100 mL of dry THF,
cooled in an ice bath, was added 8.75 mL of triethylamine,
followed by 6.66 mL of allyl chloroformate dropwise over 30
min. After 15 min the cooling bath was removed and the
reaction was stirred for 2 h. The resulting suspension was
filtered through a glass frit, washed with 60 and 30 mL
portions of THF, and concentrated to a volume of ∼25-30 mL.
This solution was added rapidly to a solution of the above
white foam in 100 mL of dry THF, and the reaction was heated
at 70 °C overnight. The solvent was removed under reduced
pressure, and the residue was dissolved in 300 mL of CH2Cl2
and washed with 200 mL portions of saturated NaHCO3 and
brine. The resulting solution was dried (Na2SO4) and concen-
trated to give a thick yellow oil, which was used directly.
To a solution of this oil in 300 mL of THF was added 18.2
mL of triethylamine trihydrofluoride, and the resulting solu-
tion was stirred for 2 d. The solvent was removed under
reduced pressure, and the resulting brown tarry material was
applied directly to a silica gel column and eluted with a step
gradient of 5:95-10:90-35:65 EtOH/CH2Cl2 to give 6.59 g
(71%) of the product as a white foam. This material showed
spectroscopic properties identical to those reported by Noyori.
1H NMR (CDCl3): δ 2.33 (1H, m), 2.75 (1H, m), 3.49-3.69 (2H,
m), 3.91 (1H, m), 4.42 (1H, m), 4.65 (2H, d, J ) 5.7), 5.26-
5.42 (2H, m), 5.9 (1H, m), 6.45 (1H, t, J ) 6.3), 8.61 (1H, s),
8.64 (1H, s). HRMS (FAB, MH+): Calcd for C14H18N5O5:
336.3291 Found: 336.1308.
5′-O-(ter t-Bu t yld ip h en ylsilyl)-N6-(a lloxyca r b on yl)-2′-
d eoxya d en osin e. To a solution of 3.0 g of N6-(allyloxycar-
bonyl)-2′-deoxyadenosine in 50 mL of pyridine was added 3.5
mL of TBDPSCl dropwise over 5 min. After stirring overnight
the solvent was removed under reduced pressure and the
residue was dissolved in 250 mL of CH2Cl2 and washed with
150 mL portions of saturated NaHCO3 and brine. The
resulting solution was dried (Na2SO4) and concentrated, and
the residue chromatographed on silica gel with a step gradient
of 5:95-10:90-20:80 EtOH/CH2Cl2 to give 3.9 g (76%) of
product as a white foam. 1H NMR (CDCl3): δ 1.05 (s, 9H),
2.55 (1H, m), 2.78 (1H, m), 3.81-3.96 (2H, m), 4.12 (1H, m),
4.75 (3H, br d), 5.21-5.42 (2H, m), 5.95 (1H, m), 6.48 (1H, t,
J ) 6.0), 7.3-7.45 (6H, m), 7.59-7.66 (4H, m), 8.15 (1H, s),
8.69 (1H, s). Anal. Calcd for C30H35N5O5Si: C, 62.80; H, 6.15;
N, 12.21. Found: C, 62.86; H, 6.22; N, 12.22.
N4-(Allyloxyca r bon yl)-2′-d eoxy-5-m eth ylcytid in e. To a
solution of 4 g of 2 in pyridine (50 mL), cooled in an ice bath,
was added allyl 1-benzotriazolyl carbonate (3.7 g). After
stirring overnight, the pyridine was removed under vacuum
and the resulting oil was taken up in CH2Cl2, washed with
saturated NaHCO3 and brine, and dried (NaSO4). The solution
was concentrated to give the product as a yellow oil which
could be used directly in the next reaction.
To a solution of the oil (5.3 g) in THF (50 mL) was added
6.9 mL of triethylamine trihydrofluoride. After stirring over-
night, the solution was concentrated. Flash chromatography
on silica gel eluting with 10:90 EtOH:CH2Cl2 gave 2.55 g of
the product as a white solid (92% based upon 2). 1H NMR
(acetone): δ 1.96 (s, 3H), 2.26 (1H, m), 2.37 (1H, m), 3.82-
5′-O-(ter t-Bu t yld ip h en ylsilyl)-N2-(a llyloxyca r b on yl)-
O6-a llyl-2′-d eoxygu a n osin e. Application of Noyori’s proce-
dure for protection of bis(tert-butyldimethylsilyl)deoxygua-

2134 J . Org. Chem., Vol. 61, No. 6, 1996
Pirrung et al.
nine28 on a 15 g scale gave 3′,5′-O-bis(tert-butyldimethylsilyl)-
N2-(allyloxycarbonyl)-O6-allyl-2′-deoxyguanine in 40% yield
after several column chromatographies. To a solution of 5.2
g of this compound in 100 mL of THF was added 5.5 mL of
triethylamine trihydrofluoride. The mixture was stirred
overnight, solvent was removed under reduced pressure, and
the residue was chromatographed on silica gel with 1:9 MeOH/
CHCl3 to give 2.9 g (88%) of slightly impure diol, which was
used directly. To a solution of this material in 60 mL of
pyridine was added 2.83 mL of TBDPSCl via syringe over 5
min. After stirring overnight, the solvent was removed under
reduced pressure, and the residue was dissolved in 150 mL of
EtOAc and washed with saturated NaHCO3 (3 × 50 mL),
saturated CuSO4 (3 × 50 mL), and brine (75 mL). The
resulting solution was dried (Na2SO4) and concentrated, and
the residue was chromatographed on silica gel with 3:97 EtOH/
CH2Cl2 to give 3.2 g (70%) of product as a white foam (61%
for both steps). 1H NMR: δ 1.03 (9H, m), 2.5-2.75 (2H, m),
3.79-3.92 (2H, m), 4.12 (m, 1H), 4.58 (2H, m), 4.8 (1H, br s),
5.05 (2H, m), 5.18-5.5 (4H, m), 5.9-6.0 (1H, m), 6.04-6.2 (1H,
m), 6.6 (1H, m), 7.2-7.4 (6H, m), 7.6 (4H, m), 8.02 (1H, s).
Anal. Calcd for C33H39N5O6Si: C, 62.93; H, 6.24; N, 11.12.
Found: C, 62.86; H, 6.25; N, 11.11.
5′-O-(ter t-Bu tyld ip h en ylsilyl)-2′-d eoxyin osin e. To a so-
lution of 1.0 g of 2′-deoxyinosine (Sigma) in 40 mL of pyridine
was added TBDPSCl (1.24 mL), and the suspension was stirred
overnight. Monitoring by showed TLC unreacted starting
material, so another 0.5 mL of TBDPSCl was added and the
solution was stirred for 4 h. The pyridine was removed under
vacuum, and the resulting oil was taken up in CH2Cl2 and
washed with saturated NaHCO3, causing the product to
precipitate as a white solid which was collected and washed
to give 1.67 g of the desired compound (86%). 1H NMR (d6-
DMSO): δ 0.95 (9H, s), 2.34 (1H, m), 2.72 (1H, m), 3.69-3.89
(2H, m), 3.96 (1H, m), 4.49 (1H, m), 6.33 (1H, t, J ) 6.6), 7.33-
7.45 (6H, m), 7.58 (4H, t, J ) 5.4), 7.98 (1H, s), 8.21 (1H, s).
Anal. Calcd for C26H30N4O4Si‚1/2H2O: C, 62.50; H, 6.25; N,
11.21. Found: C, 62.88; H, 6.25; N, 11.14.
Gen er a l P r oced u r e for th e P r ep a r a tion of P h osp h a te
Tr iester Bu ild in g Block s w ith Cya n oeth yl In ter n u cleo-
tid e P h osp h a te P r otection a n d Allyloxyca r bon yl a n d
Allyl Ba se P r otection . To a solution of the appropriately
protected nucleoside (1.0 equiv) in dichloromethane (∼30 mL/
gram of substrate) was added diisopropylethylamine (2.0
equiv), followed by a solution of cyanoethyl(N,N-diisopropyl-
amino)chlorophosphine (1.6 equiv) in dichloromethane (∼10
mL/g) via cannula. After stirring for 1.5 h a portion of MeOH
(∼1.5 mL) was added to quench excess phosphitylating reagent
and stirring was continued for 15 min. The reaction mixture
was then poured into 300 mL of EtOAc with 15 mL of Et3N
added (for 1-2 g scale reactions) and washed with 10% Na2CO3
(2 × 100 mL) and brine (2 × 100 mL). The resulting solution
was dried (Na2SO4) and concentrated to give the phosphor-
amidite in nearly quantitative yield, which was checked for
purity by 31P NMR and used directly.
treated with 10.0 equiv of triethylamine trihydrofluoride in
THF (∼50 mL/g of substrate) overnight, the solvent was
removed under reduced pressure, and the residue was chro-
matographed directly on silica gel using ∼500 mL of 1:1
EtOAc/CH2Cl2 followed by a step gradient of 2:98-5:95-10:
90 EtOH/CH2Cl2 to give the expected phosphate triesters in
yields ranging from 50-75%.
AOC-Nu cleosid e-5′-TBDP S-3′-cya n oet h yl-DMB-p h os-
p h a tes. 5′-O-TBDPS-O6-allyl-N-(allyloxycarbonyl)deoxygua-
nine (1.0 g, 1.59 mmol) was azeotroped 3× from 5 mL of
benzene and then dissolved in 10 mL of CH2Cl2. Diisopropyl-
ethylamine (0.60 mL, 3.32 mmol) was added, and the reaction
mixture was cooled to 0 °C. Cyanoethoxy-(N,N-diisopropyl-
amino)chlorophosphine (0.533 mL, 2.39 mmol) was added by
syringe, and the reaction mixture was stirred under nitrogen
for 1 h. It was poured into 50 mL of cold 5% bicarbonate
solution and extracted with CH2Cl2. The extracts were dried
over Na2SO4, and the solvent was removed to give the crude
1
product. The products were checked by 31P and H NMR and
taken on to the next step without further purification.
The phosphoramidite thus prepared and dimethoxybenzoin
(0.866 g, 3.18 mmol) were charged to a flask and azeotroped
3× from benzene. A stock solution of rigorously dried tetrazole
(17.84 mL, 6.36 mmol) was added by syringe, and the mixture
was stirred for 1 h. TLC (98:2 CH2Cl2:EtOH) showed the
reaction to be complete. The solvent was removed and the
residue taken up in CH2Cl2. Washing with 5% bicarbonate
and brine followed by drying and removal of the solvent left a
clear yellow residue which after chromatography on silica with
98:2 CH2Cl2:EtOH gave 1.41g (89%) of the title compound as
a white foam. The NMR was consistent with a mixture of four
1
diastereomers: 31P NMR showed only four lines; H NMR is
complex but shows the correct diagnostic resonances. Yields
for the other compounds are shown in Table 1. Thymidine
was done using DMTr as the 5′ protecting group. The
deprotection of the 5′-O-TBDPS-nucleotides was performed
using at least a 2-fold excess of Et3N‚3HF, with stirring for
3-12 h followed by basic aqueous workup and chromatography
using 95:5 CH2Cl2:EtOH.
N4-(Allyloxyca r b on yl)-2′-d eoxycyt id in e-3′-O-(cya n o-
eth yl 3′′,5′′-d im eth oxyben zoin p h osp h a te). 1H NMR δ
2.34 (1H, m), 2.62 (2H, m), 2.85 (1H, m), 3.76 (6H, s), 3.80-
4.20 (4H, m), 4.40 (1H, m), 4.70 (2H, m), 5.1 (1H, m), 5.26-
5.44 (2H, m), 5.9-6.10 (1H, m), 6.20 (1H, m), 6.45 (1H, m),
6.60 (2H, m), 7.39-7.60 (3H, m), 7.90 (1H, m). Anal. Calcd
for C32H35N4O12P: C, 55.01; H, 5.05; N, 8.02. Found: C, 55.36;
H, 5.25; N, 7.88.
N4-(Allyloxyca r bon yl)-2′-d eoxy-5-m eth ylcytid in e-3′-O-
(cya n oet h yl 3′′,5′′-d im et h oxyb en zoin p h osp h a t e). 1H
NMR (CDCl3): δ 1.99 (3H, br d), 2.38 (1H, m), 2.62 (2H, m),
2.88 (1H, m), 3.76 (6H, s), 3.9-4.12 (4H, m), 4.30 (1H, m), 4.66
(2H, br d), 5.22-5.41 (2H, m), 5.9-6.1 (1H, m), 6.44 (1H, br
s), 6.59 (2H, m), 7.39-7.46 (2H, m), 7.51-7.59 (1H, m), 7.89
(2H, br d). Anal. Calcd for C33H37N4O12P: C, 55.62; H, 5.23;
N, 7.86. Found: C, 55.76; H, 5.28; N, 7.78.
A mixture of the above phosphoramidite (1.0 equiv) and
3′,5′-dimethoxybenzoin (2.0 equiv) was thrice azeotroped from
40-50 mL of benzene. To the solid residue thus obtained was
added rapidly via syringe 5.0 equiv of a freshly-prepared
solution of 1H-tetrazole in acetonitrile (25 mg/mL) and the
reaction swirled to effect full solution and let stand for 1-2 h.
(The tetrazole solution was prepared by thrice azeotroping the
required amount of 1H-tetrazole from benzene and dissolving
in freshly distilled acetonitrile under Ar.) A stir bar and 1.8
equiv of a 3.0 M solution of tert-butyl hydroperoxide in
isooctane were added, and the reaction mixture was stirred
for 30-45 min and poured into 250 mL of EtOAc (for 1.5-3 g
scale reactions) and washed with 150 mL portions of saturated
NaHCO3, 10% Na2CO3, and brine (twice). The resulting
solution was dried (Na2SO4) and concentrated to give a yellow
foam, which was checked by 31P NMR and used in the next
step without further purification.
N6-(Allyloxyca r bon yl)-2′-d eoxya d en osin e-3′-O-(cya n o-
eth yl 3′′,5′′-d im eth oxyben zoin p h osp h a te). 1H NMR: δ
2.45 (1H, m), 2.62 (2H, m), 2.9 (1H, m), 3.76 (6H, m), 3.95-
4.15 (4H, m), 4.45 (1H, m), 4.77 (2H, m), 5.28-5.65 (3H, m),
6.0 (1H, m), 6.2 (1H, m), 6.35-6.50 (2H, m), 6.60 (2H, m),
7.38-7.6 (3H, m), 7.90 (2H, m), 8.7 (1H, s). Anal. Calcd for
C33H37N6O12P: C, 53.51; H, 5.04; N, 11.35. Found: C, 53.76;
H, 5.10; N, 11.30.
N2-)Allyloxyca r b on yl)-O6-a llyl-2′-d eoxygu a n in e-3′-O-
(cya n oet h yl 3′′,5′′-d im et h oxyb en zoin p h osp h a t e). 1H
NMR: δ 2.6 (3H, m), 2.89 (1H, m), 3.75 (6H, m), 3.9-4.25 (3H,
m), 4.45 (2H, m), 4.68 (2H, m), 5.09 (2H, m), 5.2-5.6 (5H, m),
5.95 (1H, m), 6.12 (2H, m), 6.29 (1H, m), 6.43 (1H, m), 6.60
(2H, m), 7.39-7.6 (3H, m), 7.89 (3H, m). Anal. HRMS (FAB,
MH+) Calcd for C36H39N6O12P: 778.7081. Found: 779.2442.
2′-Deoxyin osin e-3′-O-(cya n oeth yl 3′′,5′′-d im eth oxyben -
zoin p h osp h a te). 1H NMR: δ 2.62 (2H, m), 2.90 (1H, m),
3.38 (1H, m), 3.75 (6H, m), 3.92 (1H, m), 4.05-4.50 (4H, m),
5.45 (1H, m), 6.12 (1H, m), 6.30 (1H, m), 6.41 (1H, m), 6.61
(2H, m), 7.30-7.55 (3H, m), 7.8-8.0 (3H, m), 8.15 (1H, m).
The thymidine derivative was deprotected with 3% dichlo-
roacetic acid in CH2Cl2. The remaining compounds were
(28) Ogilvie, K. K. Can. J . Chem. 1973, 51, 3799-3807.

Inverse Phosphotriester DNA Synthesis
J . Org. Chem., Vol. 61, No. 6, 1996 2135
Anal. HRMS (FAB, MH+) Calcd for C29H30N5O10P: 639.6450.
Found: 640.1807.
90 EtOH/CH2Cl2 to give the expected phosphate triesters in
yields ranging from 50-75%.
N4-(Allyloxycar bon yl)-2′-deoxycytidin e-3′-O-(allyl 3′′,5′′-
d im eth oxyben zoin p h osp h a te). 1H NMR (CDCl3): δ 2.25
(m, 1H), 2.45 (m, 1H), 3.69 (s, 6H), 3.77-4.4 (m, 4H), 4.62 (m,
4H), 5.1-5.4 (m, 4H), 5.75 (m, 1H), 5.89 (m, 1H), 6.1 (m, 1H),
6.2 (m, 1H), 6.37 (m, 1H), 6.54 (m, 2H), 7.35 (m, 2H), 7.48 (m,
1H), 7.86 (m, 2H), 8.35 (m, 2H). HRMS (FAB, MH+) Calcd
for C32H36N3O12P: 685.6288. Found: 686.2101.
2′-Th ym id in e-3′-O-(cya n oet h yl 3′′,5′′-d im et h oxyb en -
zoin p h osp h a te). 1H NMR: δ 1.95 (3H, m), 2.4 (1H, m), 2.65
(2H, m), 2.95 (1H, m), 3.78 (6H, m), 3.98-4.2 (4H, m), 4.30
(1H, m), 4.50 (1H, m), 5.4 (1H, m), 6.1 (1H, m), 6.28 (1H, m),
6.5 (1H, m), 6.65 (2H, m), 7.4-7.65 (3H, m), 7.95 (2H, m), 8.5
(1H, br s). Anal. Calcd for C16H32N3O11P‚1/2H2O: C, 54.54;
H, 5.21; N, 6.58. Found: C, 54.44; H, 5.21; N, 6.58.
N4-(Allyloxyca r bon yl)-2′-d eoxy-5-m eth ylcytid in e-3′-O-
(a llyl 3′′,5′′-d im et h oxyb en zoin p h osp h a t e). 1H NMR
(CDCl3): δ 1.98 (m, 3H), 2.37 (m, 1H), 2.48 (m, 1H), 3.74 (m,
6H), 3.85-4.08 (m, 4H), 4.22-4.45 (m, 4H), 5.2-5.4 (m, 4H),
5.72-5.88 (m, 2H), 5.9-6.1 (m, 1H), 6.21 (m, 1H), 6.41 (m,
1H), 6.55-6.6 (m, 2H), 7.40 (m, 2H), 7.54 (m, 1H), 7.74 (s, 1H),
7.89 (m, 2H). HRMS (FAB, MH+) Calcd for C33H38N3O12P:
699.6532. Found: 700.2267.
(Allyloxy)d ich lor op h osp h in e. To 87 mL of freshly dis-
tilled phosphorus trichloride, cooled in an ice bath, was added
68 mL of freshly distilled allyl alcohol dropwise, with stirring,
over 8 h. After stirring overnight, house vacuum was applied
for 2 h to remove most of the HCl, and then Ar was bubbled
through the solution to remove the last vestiges of HCl. The
product was obtained by distillation through a short path still
at 0.1 torr into a dry ice-cooled receiver. In this way the
explosive decomposition reported to result from distillation at
higher pressures was avoided. Great care still must be taken
to discontinue the distillation as soon as the temperature
begins to rise, or as soon as the pot residue turns orange, in
order to avoid a vigorous exotherm. The yield is 95 g (60%) of
colorless liquid, with properties consistent with those reported
in the literature.29 31P NMR: δ 178.5
(Allyloxy)(N,N-d ieth yla m in o)ch lor op h osp h in e. To 32 g
of (allyloxy)dichlorophosphine dissolved in 150 mL of dry Et2O
was added dropwise 42 mL of diethylamine, with vigorous
stirring, over 3-4 h. The resulting slurry was filtered through
a Schlenck funnel under Ar and washed with 50 mL of
additional Et2O. Solvent was removed by distillation under
house vacuum, and the residue was distilled at 0.04 torr (bp
50-58 °C) to give 20.55 g (52%) of the product as a colorless
liquid. 31P NMR (CDCl3): δ 181.6. 1H NMR (CDCl3): δ 1.15
(6Η, m), 3.25-3.40 (4H, m), 4.35-4.45 (2H, m), 5.18-5.38 (2H,
m), 5.9-6.05 (1H, m).
Gen er a l P r oced u r e for th e P r ep a r a tion of P h osp h a te
Tr iester Bu ild in g Block s w ith Allyl In ter n u cleotid e
P h osp h a te P r otection a n d Allyloxyca r bon yl Ba se P r o-
tection . To a solution of protected nucleoside (1.0 equiv) in
dichloromethane (∼30 mL/g) was added diisopropylethylamine
(2.0 equiv), followed by a solution of (allyloxy)(N,N-diethyl-
amino)chlorophosphine (1.6 equiv) in dichloromethane (∼10
mL/g) via cannula. After stirring for 1.5 h a portion of MeOH
(∼1.5 mL) was added to quench excess phosphitylating reagent
and stirring was continued for 15 min. The reaction mixture
was then poured into 300 mL of EtOAc with 15 mL of Et3N
added (for 1-2 g scale reactions) and washed with 10% Na2CO3
(2 × 100 mL) and brine (2 × 100 mL). The resulting solution
was dried (Na2SO4) and concentrated to give in nearly quan-
titative yield the phosphoramidites, which were checked for
purity by 31P NMR and used directly.
A mixture of the above phosphoramidite (1.0 equiv) and
3′,5′-dimethoxybenzoin (2.0 equiv) was thrice azeotroped from
40-50 mL of benzene. To the solid residue thus obtained was
added rapidly via syringe 5.0 eq of a freshly prepared solution
of 1H-tetrazole in acetonitrile (25 mg/mL), and the reaction
swirled to effect full solution and allowed to stand for 1-2 h.
A stir bar and 1.8 equiv of a 3.0 M solution of tert-butyl-
hydroperoxide in isooctane were added, and the reaction
mixture was stirred for 30-45 min. It was poured into 250
mL of EtOAc (for 1.5-3 g scale reactions) and washed with
150 mL portions of saturated NaHCO3, 10% Na2CO3, and brine
(twice). The resulting solution was dried (Na2SO4) and
concentrated to give a yellow foam, which was checked by 31P
NMR and used in the next step without further purification.
The thymidine derivative was deprotected with 3% dichlo-
roacetic acid in CH2Cl2 as reported below for the compounds
with acyl base protection. The remaining compounds were
treated with 10.0 equiv of triethylamine trihydrofluoride in
THF (∼50 mL/g of substrate) overnight, the solvent was
removed under reduced pressure, and the residue was chro-
matographed directly on silica gel using ∼500 mL of 1:1
EtOAc/CH2Cl2 followed by a step gradient of 2:98-5:95-10:
N6-(Allyloxyca r b on yl)-2′-d eoxya d en osin e-3′-O-(a llyl
3′′,5′′-d im eth oxyben zoin p h osp h a te). 1H NMR (CDCl3): δ
2.35 (m, 1H), 2.6 (m, 1H), 3.81 (s, 6H), 3.76-4.7 (m, 4H), 4.75
(m, 4H), 5.1-5.47 (m, 4H), 5.8-6.03 (m, 2H), 6.12 (m, 1H),
6.29-6.4 (m, 2H), 6.58 (m, 2H), 7.35 (m, 2H), 7.48 (m, 1H),
7.87 (m, 2H), 8.08 (s, 1H), 8.68 (m, 1H). HRMS (FAB, MH+)
Calcd for C30H36N5O11P: 709.6671. Found 710.2217.
N2-(Allyloxyca r b on yl)-O6-a llyl-2′-d eoxygu a n in e-3′-O-
(a llyl 3′′,5′′-d im et h oxyb en zoin p h osp h a t e). 1H NMR
(CDCl3): δ 2.3 (m, 1H), 2.6 (m, 1H), 3.68 (s, 6H), 3.7-4.45 (m,
6H), 4.62 (m, 4H), 5.05 (m, 2H), 5.1-5.5 (m, 6H), 5.7-6.15 (m,
3H), 6.2 (m, 1H), 6.36 (m, 1H), 6.57 (m, 2H), 7.3-7.5 (m, 3H),
7.78-7.92 (m, 3H). HRMS (FAB, MH+) Calcd for
C36H40N5O12P: 765.7203. Found: 766.2520.
2′-Deoxyin osin e-3′-O-(a llyl 3′′,5′′-d im et h oxyb en zoin
p h osp h a te). 1H NMR (CDCl3): δ 2.42 (m, 1H), 2.8 (m, 1H),
3.74 (s, 6H), 4.45 (m, 3H), 4.7 (m, 2H), 5.1-5.35 (m, 2H), 5.45
(m, 1H), 5.75-6.05 (m, 1H), 6.1 (m, 1H), 6.3 (m, 1H), 6.42 (m,
1H), 6.63 (m, 2H), 7.4 (m, 2H), 7.52 (m, 1H), 7.91 (m, 2H), 8.24
(m, 2H). HRMS (FAB, MH+) Calcd for C29H31N4O10P: 626.5651.
Found: 627.1848.
Th ym id in e-3′-O-(a llyl 3′′,5′′-d im eth oxyben zoin p h os-
p h a te). 1H NMR (CDCl3): δ 2.04 (s, 3H), 2.32(m, 1H), 2.50
(m, 1H), 3.72 (s, 6H), 3.8-4.4 (m, 4H), 4.64 (m, 1H), 4.9 (m,
1H), 5.15-5.4 (m, 2H), 5.7-5.92 (m, 1H), 6.19 (m, 1H), 6.4 (s,
1H), 6.58 (m, 2H), 7.37 (m, 2H), 7.52 (m, 1H), 7.9 (m, 2H), 9.0
(br, 1H). HRMS (FAB, MH+) Calcd for C29H33N2O11P: 616.5667.
Found: 617.1911.
Cou p lin g Rea ction s To P r ep a r e Cya n oeth yl P h osp h o-
tr iester s. G-T Din u cleotid e. 5′-O-TBDPS-AOC-O6-allyl dG-
3′-CE DMB phosphate (300 mg, 0.295 mmol) was dissolved in
100 mL of benzene with 1 mL of pyridine. The solution was
degassed by bubbling with Ar and irradiated for 30 min in a
Rayonet photochemical reactor using 350 nm phosphor lamps.
The solvent was removed. Column chromatography on tri-
ethylamine-treated silica gel using 90:9:1 CH2Cl2:EtOH:Et3N
gave the desired deprotected product plus a great deal of
Et3N‚HCl, which was removed by precipitation from ethyl
acetate. 31P NMR showed one clean singlet, and this material
was taken to the next step. The dG salt, 5′-OH T-DMB-CE
phosphate (0.278 g, 0.443 mmol) and 4-ethoxypyridine N-oxide
(0.205 g, 1.48 mmol) were thrice azeotroped from pyridine and
taken up in 12 mL of dichloroethane. Triisopropylbenzene-
sulfonyl chloride was added (0.183 g, 0.590 mmol), and the
mixture was stirred under Ar overnight. TLC revealed three
spots, the 5′-sulfonate of the T monomer, the desired product,
and excess T monomer. The solvent was removed. Column
chromatography on silica gel using 95:5 CH2Cl2:EtOH gave
305 mg (75%) of a white foam. Its spectra were consistent
with the title compound. HRMS (FAB, MH+) Calcd for
C65H73N9O19P2Si: 1374.3800. Found: 1374.4346.
G-T-C Tr in u cleot id e. The above dinucleotide (240 mg,
0.175 mmol) was dissolved in 100 mL of benzene and 1 mL of
pyridine. The solution was degassed by bubbling with Ar and
irradiated for 30 min in a Rayonet photochemical reactor using
350 nm phosphor lamps. Triethylamine (1 mL) was added,
the solvent was removed, and this material was taken directly
to the next step without purification. This salt, 5′-OH-dC-
DMB-CE-phosphate (0.35 mmol) and 4-ethoxypyridine N-oxide
(29) J . Gen. Chem. (USSR) 1936, 6, 1198-1202.

2136 J . Org. Chem., Vol. 61, No. 6, 1996
Pirrung et al.
were azeotroped thrice from pyridine and then taken up in 10
mL of dichloroethane. Triisopropylbenzenesulfonyl chloride
(106 mg, 0.350 mmol) was added and the mixture stirred
overnight. Column chromatography on silica gel using 95:5
CH2Cl2:EtOH gave 219 mg (69%) of a white foam. HRMS
(FAB, MH+) Calcd for C65H77N13O24P3Si: 1545.4091. Found:
1546.4309.
-3.89 to -2.51. HRMS (FAB, MH+) Calcd for C71H82N9O25P3-
Si: 1582.4878. Found. 1581.4332.
Ack n ow led gm en t. Financial support by a Human
Genome Distinguished Postdoctoral Fellowship (to
D.C.L.), sponsored by the U. S. Department of Energy,
Office of Health and Environmental Research, and
administered by the Oak Ridge Institute for Science and
Education, and the Department of Energy Human
Genome Program (DE-FG05-92ER61388) is greatly ap-
preciated.
T-T-T Tr in u cleotid e. The procedure for the G-T dinucleo-
tide was applied to 5′-TBDPS T-DMB-CE-phosphate and 5′-
OH-T-DMB-CE-phosphate to give 262 mg (68%) of the T-T
dinucleotide. This compound (120 mg) was dissolved in 40 mL
of benzene and 1 mL of pyridine and degassed by bubbling
with Ar. The solution was irradiated for 40 min as described
above, followed by the addition of 1 mL of Et3N. The solution
was concentrated and 70 mg of the desired salt was isolated
by flash chromatography on Et3N-treated silica gel, eluting
with a step gradient of 11:89-20:80-50:50 EtOH:CH2Cl2. The
salt, 5′-OH-T-DMB-CE (62 mg), and 4-ethoxypyridine N-oxide
(45 mg) were azeotroped thrice from pyridine and then
dissolved in 10 mL of dichloroethane. Triisopropylbenzene-
sulfonyl chloride (40 mg) was added, and the reaction mixture
was stirred overnight. The solution was concentrated. Flash
chromatography on silica gel using a step gradient of 4:96-
6:94-10:90 EtOH:CH2Cl2 gave 50 mg of the desired trinucleo-
Su p p or tin g In for m a tion Ava ila ble: NMR spectra of
compounds characterized by HRMS: 3′-O-(CE-DMB-phosphate)-
2′-deoxy-I, 3′-O-(allyl-DMB-phosphate)-N-AOC-2′-deoxy-C, 3′-
O-(allyl-DMB-phosphate)-N-AOC-2′-deoxy-5-MC, 3′-O-(allyl-
DMB-phosphate)-N-AOC-2′-deoxy-A, 3′-O-(allyl-DMB-phos-
phate)-O-allyl-N-AOC-2′-deoxy-G, 3′-O-(allyl-DMB-phosphate)-
2′-deoxy-I, 3′-O-(allyl-DMB-phosphate)-thymidine (9 pages).
This material is contained in libraries on microfiche, im-
mediately 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.
tide (53% based upon dinucleotide). 31P NMR (CDCl3):
δ
J O951593C