Synthesis of High-Quality Antisense Drugs. Addition of Acrylonitrile to Phosphorothioate Oligonucleotides: Adduct Characterization and Avoidance

Article abstract of DOI:10.1021/op020090n

It is demonstrated that the acrylonitrile (AN) generated during the ammonolysis step of oligonucleotide manufacture selectively adds to thymine residue present in ISIS 2302 to give a full-length oligonucleotide in which thymine is replaced by an N³-cyanoethylthymine residue. Treatment of support-bound ISIS 2302 with a solution of triethylamine in CH₃CN before ammonolysis is sufficient to prevent formation of this class of impurity.

Full text of DOI:10.1021/op020090n

Organic Process Research & Development 2003, 7, 832−838
Synthesis of High-Quality Antisense Drugs. Addition of Acrylonitrile to
Phosphorothioate Oligonucleotides: Adduct Characterization and Avoidance
Daniel C. Capaldi, Hans Gaus, Achim H. Krotz, Jim Arnold, Ricaldo L. Carty, Max N. Moore, Anthony N. Scozzari,
Kirsten Lowery, Douglas L. Cole, and Vasulinga T. Ravikumar*
Isis Pharmaceuticals, Inc., 2292 Faraday AVenue,Carlsbad, California 92008, U.S.A.
Abstract:
try.4 In brief, oligomerization proceeds by detritylation of a
support-bound nucleoside 1 and coupling of a cyanoethyl-
protected phosphoramidite to give an intermediate phosphite
triester 2 (Scheme 1).5 Subsequent sulfurization and capping
of unreacted hydroxy functionalities completes a synthesis
cycle and gives the phosphorothioate triester 3.
It is demonstrated that the acrylonitrile (AN) generated during
the ammonolysis step of oligonucleotide manufacture selectively
adds to thymine residue present in ISIS 2302 to give a full-
length oligonucleotide in which thymine is replaced by an N3-
cyanoethylthymine residue. Treatment of support-bound ISIS
2302 with a solution of triethylamine in CH3CN before am-
monolysis is sufficient to prevent formation of this class of
impurity.
Upon completion of the chain-assembly steps the oligo-
nucleotide is released from the support and deprotected by
incubation with ammonium hydroxide. The finished product
is obtained after purification (usually by reversed-phase
HPLC) and a final detritylation step. Typically, the quality
of the oligonucleotide is ascertained by a combination of
techniques including capillary gel electrophoresis (CGE),6
strong anion-exchange (SAX) chromatography,7 31P NMR
spectroscopy, and electrospray mass spectrometry (ES-MS).8
ISIS 2302, d(GCC-CAA-GCT-GGC-ATC-CGT-CA),is a
20-mer phosphorothioate oligodeoxynucleotide currently in
a pivotal quality trial for treatment of Crohn’s disease. During
the course of our studies on optimization of the ISIS 2302
manufacturing process and of solid-phase oligonucleotide
synthesis in general, we consistently observed the occurrence
of 2-3% of an impurity with characteristic mass 53 amu
more than ISIS 2302. Observations made on a number of
sequences indicated that impurity levels varied in a sequence-
specific manner and were present to a higher degree in
Phosphorothioate (PS) oligonucleotides demonstrate the
capacity to specifically block protein synthesis in vitro and
in vivo via an antisense mechanism.1 Because of their
specificity of binding, their ability to support RNase-H
mediated cleavage of hybridized mRNA target sequences,
and their enzymatic stability and low toxicity, phospho-
rothioate oligonucleotides targeted at a variety of human
conditions including cancer, psoriasis, and Crohn’s disease
are currently under clinical investigation.2 Approval of the
antisense oligonucleotide fomivirsen sodium (Vitravene) for
the treatment of CMV retinitis in AIDS patients by the FDA
promises to herald a new dawn in the treatment of human
disease.3 Further commercial realization of the promise of
antisense technology depends largely on the ability to
synthesize substantial quantities (hundreds of kilograms) of
high-quality oligonucleotide, a need which is met at present
by solid-phase synthesis utilizing phosphoramidite chemis-
(4) (a) Org. Process Res. DeV. 2000, 4, 167-225. (b) Krotz, A. H.; Cole, D.
L.; Ravikumar, V. T. Bioorg. Med. Chem. 1999, 7, 435. (c) Krotz, A. H.;
Carty, R. L.; Moore, M. N.; Scozzari, A. N.; Cole, D. L.; Ravikumar, V. T.
Green Chem. 1999, 277. (d) Ravikumar, V. T.; Krotz, A. H.; Cole, D. L.
Tetrahedron Lett. 1995, 36, 6587. (e) Turney, B. J., Cheruvallath, Z. S.,
Andrade, M.; Cole, D. L.; Ravikumar, V. T. Nucleosides Nucleotides 1999,
18, 89. (f) Cheruvallath, Z. S.; Wheeler, P. D.; Cole, D. L.; Ravikumar, V.
T. Nucleosides Nucleotides 1999, 18, 485. (g) Eleuteri, A.; Capaldi, D. C.;
Cole, D. L.; Ravikumar, V. T. Nucleosides Nucleotides 1999, 18, 475. (h)
Ravikumar, V. T.; Andrade, M.; Wyrzykiewicz, T. K.; Scozzari, A. N.;
Cole, D. L. Nucleosides Nucleotides 1995, 14, 1219. (i) Krotz, A. H.;
Klopchin, P.; Cole, D. L.; Ravikumar, V. T. Bioorg. Med. Chem. Lett. 1997,
7, 73. (j) Krotz, A. H.; Klopchin, P.; Cole, D. L.; Ravikumar, V. T.
Nucleosides Nucleotides 1997, 16, 1637. (k) Capaldi, D. C.; Scozzari, A.
N.; Cole, D. L.; Ravikumar, V. T. Org. Process Res. DeV. 1999, 3, 485. (l)
Eleuteri, A.; Capaldi, D. C.; Krotz, A. H.; Cole, D. L.; Ravikumar, V. T.
Org. Process Res. DeV. 2000, 4, 182. (m) Eleuteri, A.; Cheruvallath, Z. S.;
Capaldi, D. C.; Cole, D. L.; Ravikumar, V. T. Nucleosides Nucleotides 1999,
18, 1803.
(5) For basic explanation on nomenclature, structure of nucleosides, protecting
groups used, synthesis of oligonucleotides: Gait, M. J. Oligonucleotide
Synthesis: A Practical Approach; IRL Press: NY, 1984.
(6) (a) Paulus, A.; Ohms, J. I. J Chromatogr. 1990, 507, 113-123. (b) Guttman,
A., Cohen, A. S.; Heiger, D. N.; Karger, B. L. Anal. Chem. 1990, 62, 137-
141. (c) Warren, W.; J.; Vella, G. Biotechniques 1993, 14, 293-301. (d)
Srivatsa, G. S.; Batt, M.; Schuette, J.; Carlson, R. H.; Fitchett, J.; Lee, C.;
Cole D. L. J. Chromatogr., A 1994, 680, 469-477.
* To whom correspondence should be addressed. E-mail: vravikumar@
isisph.com.
(1) (a) Crooke, S. T. Antisense Ther. Biotechnol. Genet. Eng. ReV. 1998, 15,
121-157. (b) Crooke, S. T. Antisense Nucleic Acid Drug DeV. 1998, 8,
115-122. (c) Crooke, S. T. Basic Principles of Antisense Therapeutics. In
Handbook of Experimental Pharmacology: Antisense Research & Applica-
tion; Crooke, S. T., Ed.; Springer Verlag: Berlin, 1998. (d) Monia, B. P.,
Sasmor, H.; Johnston, J. F.; Freier, S. M.; Lesnik, E. A.; Muller, M.; Geiger,
T.; Altmann, K. H.; Moser, H.; Fabbro, D. Proc. Natl. Acad. Sci. U.S.A.
1996, 93, 15481-15484. (e) Monia, B. P.; Johnston, J. F.; Muller, M.;
Geiger, T.; Moser, H.; Fabbro, D. Nat. Med. 1996, 2, 668-675. (f) Crooke,
S. T. In Therapeutic Applications of Oligonucleotides; Crooke, S. T., Ed.;
R. G. Landes Co.: Austin, 1995 and references therein. (g) Crooke, S. T.
In Burgers Medicinal Chemistry and Drug DiscoVery; Wolff, M. E., Ed.;
John Wiley and Sons: New York, 1995; Vol. 1, pp 863-878.
(2) ISIS 2302 is in a pivotal Phase III trial for Crohn’s disease and in Phase II
clinical trials for renal transplant rejection, ulcerative colitis, and psoriasis.
Affinitak (ISIS 3521) is in pivotal Phase III clinical trials as an anticancer
agent. Currently using Amersham Biosciences’ OligoProcess, we routinely
synthesize phosphorothioate oligodeoxyribonucleotides at scales between
300 and 600 mmol, yielding more than 2 kg of purified drug in a synthesis
at the high end of synthesis (viz. 600 mmol scale). On the basis of IP-LC-
MS analysis, drug of very good quality is being manufactured here at Isis
Pharmaceuticals, Inc.
(3) More than 25 oligonucleotides (both first- and second-generation combined)
are being evaluated here at Isis Pharmaceuticals as well as in other places
for the treatment of various diseases.
(7) Bergot, J., B.; Egan, W. J. Chromatogr. 1992, 599, 35.
(8) Apffel, A.; Chakel, J. A.; Fischer, S.; Lichtenwalter, K.; Hancock, W. S.
Anal. Chem. 1997, 69, 1320.
832
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development
10.1021/op020090n CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/25/2003

Scheme 1. Synthesis of phosphorothioate oligonucleotidesa
a Detritylation ) removal of DMT groups, coupling ) reaction of phosphoramidite in presence of an activator such as 1H-tetrazole, sulphurization ) oxidizing
phosphite triester to phosphorothioate triester, capping ) blocking of unreacted hydroxyl groups.
thymidine-rich oligonucleotides (ca. 0.5%/thymine residue).
The observation that the impurity was associated with
thymine residues, coupled with the mass spectroscopy data,
led to the tentative postulate that the impurity arose by the
adventitious addition of acrylonitrile (AN) to thymine
residues present in the oligonucleotide.
nitrogen atom.12 We now report isolation and characterization
of the +53 amu impurity from a variety of phosphorothioate
oligonucleotides including ISIS 2302. In addition, a simple
and effective method to avoid its formation is described.
Results and Discussion
During the course of the ammonium hydroxide depro-
tection it is evident that the cyanoethyl groups protecting
the internucleotide linkages are removed by a â-elimination
mechanism to generate AN. It seemed possible that the thus
formed AN could be captured in a Michael-type9 fashion
by the N3 atom of thymidine. Alkylation of thymine residues
during the course of oligonucleotide synthesis has been
described previously. Gait et al noted that thymine residues
underwent transmethylation when methyl-protected oligo-
nucleotide triesters were treated with 1,8-diazabicyclo-
[5.4.0]-undec-7-ene (DBU).10 More recently Eritja11 de-
scribed similar transalkylation problems associated with DBU
treatment of cyanoethyl-protected oligonucleotide triesters
although no structural proof was provided. AN is known to
alkylate guanine and thymine residues in nucleic acids, giving
N7-2-cyanoethylguanine and N3-2-cyanoethylthymine resi-
dues, respectively, as the main products after acid-induced
depurination.12 Interestingly, reaction with cytosine and
adenine residues leads directly to the N3 and N1 (and in the
Part 1: Isolation and Characterization. Initial experi-
ments aimed at the isolation and characterization of the
impurity were performed on phosphorothioate oligothymidy-
lates. Nonadecathymidyloctadecaphosphorothioic acid (TPS)18T
was synthesised on a ca. 160 µmol scale under above-
described oligonucleotide synthesis conditions. Treatment of
the support-bound oligonucleotide with ammonium hydrox-
ide to effect succinate cleavage and phosphate deprotection
was followed by detritylation with AcOH-H2O (1:4 v/v).
The crude nonadecamer was analyzed by LC-MS. Selective
ion monitoring was used to expand the region encompassing
the -4 charge state. Figure 1 shows a main quadruply
charged ion with m/z ) 1500.6 corresponding to (TPS)18T.
In addition to the monophosphate diester (strand containing
one phosphate group by replacement of a sulfur atom by
oxygen) of (TPS)18T (m/z ) 1496.8) and the monosodium
adduct (m/z ) 1506.2), a molecular ion with m/z ) 1514
was also detected. This mass increase (+53 amu) is
consistent with the addition of acrylonitrile to a single
thymine residue in (TPS)18T. Integration of the relative peak
intensities showed the monocyanoethyl adduct to be present
to the extent of ca. 15%.
case of adenine, N6 via a Dimroth rearrangement13,14
)
carboxyethyl derivatives (an addition of +73 amu) due to
assisted hydrolysis of the nitrile by the proximal exocyclic
The presence of N3-cyanoethylthymine residues was
confirmed by desulphurization and digestion of (TPS)18T with
snake venom phosphodiesterase (SVPD) and alkaline phos-
phatase (AP). RP-HPLC analysis of the products showed that,
in addition to thymidine (RT ) 15 min), a later-eluting
product (RT ) 25 min) was also present (Figure 2). Mass
spectroscopic analysis of the latter material and co-injection
(9) Bergman, E. D.; Ginsberg, D.; Pappo, R. Org. React. 1959, 10, 179.
(10) Lehmann, C.; Xu, Y.-Z.; Christodoulou, C.; Tan, Z.-K.; Gait, M. J. Nucleic
Acids Res. 1989, 17, 2379.
(11) Avino, A.; Garcia, R. G.; Diaz, A.; Albericio, F.; Eritja, R. Nucleosides
Nucleotides 1996, 15, 1871.
(12) Solomon, J. J.; Cote, I. L.; Wortman, M.; Decker, K.; Segal, A. Chem.-
Biol. Interact. 1984, 51, 167.
(13) Grenner, G.; Schmidt, H. L. Chem. Ber. 1977, 110, 373.
(14) Segal, A.; Mate, U.; Solomon, J. J. Chem.-Biol. Interact. 1979, 28, 333.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
833

Table 1 shows that low levels of an impurity having a
mass of 53 amu greater than full length oligonucleotide were
also present in samples of d[(APS)18A] and d[(CPS)18C] that
were unblocked under standard conditions (row 1, Table 1).
Addition of AN to the deprotection mixtures increased these
levels by a factor of 2. Interestingly, if d[(CPS)18C] was treated
with NH4OH for 5 h at 60 °C to remove the N4-benzoyl
protecting groups (a standard group used to protect exocyclic
amino group of deoxycytidine, deoxyadenosine) prior to
addition of AN, the amount of adduct was seen to increase
to ca. 12%. No + 53 amu impurity was seen in samples of
d[(GPS)18G] unblocked under any of the above conditions,
and no adducts having a mass of 73 amu greater than full-
length oligonucleotide were ever observed. Under normal
unblocking conditions low levels of adduct were observed
in d[(APS)18A] and d[(CPS)18C]. In both cases levels were
much lower than those observed in (TPS)18T (row 1, Table
1). In addition, LC-MS analysis of desulfurized and
enzymatically digested d[(CPS)18C] and d[(TPSC)9T] indicated
the presence of low levels of a nucleoside having a mass 53
amu greater than that of 2′-deoxycytidine.5 Replacement of
10 out of the 19 cytosine residues with thymine in d[(TPSC)9T]
increased adduct level by a factor of ca. 7.
It should be remembered that AN has very limited stability
in NH4OH (half-life is ca. 10 min at 25 °C in ND4OD15)
and that for base modification to occur, the base must react
effectively with AN before its destruction by NH4OH.
Apparently, under these conditions, only thymine residues
react quickly enough to undergo cyanoethylation to any great
extent. These rate differences add credence to the idea that
in a mixed-base sequence (e.g. ISIS 2302) AN will add
selectively to thymine residues. It appears that this selectivity
is, at least in part, due to the fact that cytosine residues are
protected by a benzoyl group on N4 during the early stages
of ammonolysis. Table 1 shows that while addition of AN
to ammoniacal solutions of d[(CPS)18C] increased the amount
of adduct moderately (from 1.3 to 2.5%), addition of AN
after removal of the benzoyl protecting groups resulted in
ca. 10-fold increase in the amount of adduct formed. Further
selectivity no doubt arises because of the acidity of thymine
residues (pKa ) 9.79, dC, pKa ) 4.25; dA, pKa ) 1.75; dG,
pKa ) 1.6)16 with the consequence that under the ammonoly-
sis they would be expected to be at least partially deproto-
nated.
The finding of +53 amu adducts in homoadenylates and
homocytidylates is contradictory to reports in the literature
which state that only carboxyethyl adducts (+73 amu) of
adenine and cytosine residues are ever encountered.12
Although the adenine and cytosine adducts above have not
been characterized, the results indicate (at least for cytosine
residues) that AN adds to the base and not to, for example,
the free 3′-hydroxyl. The dramatic increases in adduct levels
upon preincubation of d[(CPS)18C] with NH4OH before AN
addition, coupled with the fact that d[(APS)18A] does not show
a similar increase, also support addition of AN to the base
moiety rather than a sugar hydroxyl. The fact that the adduct
has a molecular weight of 53 amu and not, as reported,12 73
amu more than that of full-length can be interpreted in one
Figure 1. MS of phosphorothioate oligodeoxyribonucleotide
nonadecamer (TPS)18T.
with an authentic sample confirmed its identity as N3-
cyanoethylthymidine.
The stability of N3-cyanoethylthymidine towards NH4OH
was investigated by incubating an ammoniacal solution at
60 °C for 16 h. RP-HPLC analysis of the products (Figure
3) indicated that ca. 12 area % of the starting material was
converted back to thymidine, while a further 8 area % was
converted into an another compound whose identity was not
established (possibly the corresponding amide or acid). These
results indicate that N3-cyanoethylthymidine17 and, by anal-
ogy, N3-cyanoethylthymine residues present in oligonucle-
otides are relatively stable toward the unblocking conditions.
Attempts were also made to look for products arising from
the addition of AN to adenine, cytidine, and guanine residues.
The extent of reaction was markedly less for these species.
Results obtained from LC-MS analysis of the products
arising from incubation of ammonium hydroxide solutions
of d[(APS)18A], d[(CPS)18C], d[(GPS)18G], and d[(TPSC)9T]
under a variety of conditions including elevated tempera-
ture are contrasted with results obtained from (TPS)18T in
Table 1.
(15) Capaldi, D. Unpublished observations.
(16) (a) Saenger, W. Principles of Nucleic Acid Structure; Springer-Verlag: New
York, 1984; p 108. (b) Albert, A. Ionization Constants of Pyrimidines and
Purines. In Synthetic Procedures in Nucleic Acid Chemistry, Zorbach, W.
W., Tipson, R. S., Eds.; Physical and Physicochemical Aids in Characteriza-
tion and in Determination of Structure, Vol. 2; Wiley-Interscience: New
York 1973; pp 1-46.
(17) (a) Mag, M.; Engels, J. W. Nucleic Acids Res. 1988, 16, 3525. (b) Andrzej,
W.; Grajkowski, A.; Phillips, L. R.; Beaucage, S. L. J. Org. Chem. 1999,
64, 7515. (c) Solomon, J.; Segal, A. EnViron. Health Perspect. 1985, 62,
227. (d) Solomon, J.; Cote, I. L.; Wortman, M.; Decker, K.; Segal, A. Chem.-
Biol. Interact. 1984, 51, 167. (e) Ogilvie, K. K.; Beaucage, S. L. Nucleic
Acids Res. 1979, 7, 805.
834
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 2. RP-HPLC of enzymatic digest of phosphorothioate oligodeoxyribonucleotide nonadecamer (TPS)18T.
Part II: Avoiding Formation of Cyanoethyl Adducts.
With the identity of the impurity confirmed it became of
interest to devise a method to avoid its formation. Initial
efforts were centered on two main strategies. First, depro-
tection of the oligonucleotide in the presence of a reagent
reactive towards acrylonitrile (i.e., a scavenger) would be
expected to reduce the percentage of thymidine residues that
undergo alkylation. In a second scenario, deprotection of the
internucleotide linkages could be performed under conditions
under which the liberated acrylonitrile did not react with
thymidine residues.
Addition of acrylonitrile scavengers to the deprotection
mixture raised several concerns. First, one must select a
reagent that adds exclusively to acrylonitrile and not to any
other functionality. Second, it must be demonstrated that
adduct can be completely removed from the final product.
Despite some initial success with some scavengers such as
thymine, we felt that these concerns would be difficult to
resolve and so concomitantly turned to the second strategy.
In essence, the second strategy involved treatment of the
support-bound oligonucleotide with a solution of a base in
a suitable solvent. The appropriate base would cleave
cyanoethyl groups at a reasonable rate while not attacking
the succinate ester tethering the oligonucleotide to the solid
support. Initial screening experiments performed (with
amines such as triethylamine, tributylamine, diethylamine,
and diisopropylamine and in solvents such as acetonitrile,
toluene, and dioxane) in solution at the dimer level led us to
believe that a 1:1 (v/v) mixture of triethylamine (TEA) in
acetonitrile would constitute a suitable reagent system.
The scavenging condition was extended to deprotection
of nonadecathymidyloctadecaphosphorothioic acid (TPS)18T.
Following complete oligomerization, the support-bound
oligonucleotide was washed with a solution of TEA-CH3-
CN (1:1 v/v) at a linear flow rate of 0.8 cm min-1 for 4 h.
The eluate was collected as four consecutive 1-h fractions
and analyzed for the presence of acrylonitrile by gas
Figure 3. RP-HPLC explainaing the stability of N3-cyanoet-
hylthymidine toward ammonium hydroxide.
of two ways. First it may be that, under the specific
conditions studied, cyanoethyl adducts of adenine and
cytosine are stable. Alternatively, one could postulate that
the initially formed cyanoethyl adducts do in fact hydrolyze
to the corresponding carboxyethyl adducts as reported,12 but
that these adducts are themselves unstable and cyclize to form
the lactam structures 4 and 5 (Figure 4).
The finding that AN reacts substantially faster with
thymine residues than with the three remaining bases
indicates that the 2-3 area % of adduct peak detected in
ISIS 2302 is composed mainly of N3-cyanoethylthymine
containing full-length oligonucleotide. It seems reasonable
to treat each of the three thymine residues in ISIS 2302 as
equivalent and postulate that the monoalkylated impurity
peak is an equimolar mixture of three compounds.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
835

Table 1
% of +53 amu impuritya
deprotection
conditions
d[(APS)18A]
d[(CPS)18C]
d[(GPS)18G]
d[(TPSC)9T]
(TPS)18T
NH4OHb
1.5
2.5
2.6
1.3
2.5
12
N.D.e
N.D.
N.D.
10
20
-
15
-
-
NH4OH, ANc
NH4OH then
NH4OH, ANd
a Per cent of +53 amu impurity relative to full-length oligonucleotide as judged by LC-MS. b 50 mg support-bound oligo treated with NH4OH (1 mL) for 16 h at
60 °C. c 50 mg support-bound oligo treated with NH4OH (10 µL) for 16 h at 60 °C. d 50 mg support-bound oligo treated with NH4OH (1 mL) for 5 h at 60 °C then
with NH4OH (1 mL) and AN (10 µL) for 16 h at 60 °C°. e N.D. ) not detected.
Figure 4. Cyclic adducts of bases.
chromatography (GC). Approximately 90% of the theoretical
amount of acrylonitrile was contained within the first fraction,
and the reaction was essentially complete after 4 h. After
washing the support-bound material briefly with CH3CN the
oligonucleotide was released from the solid support by
treatment with ammonium hydroxide in the normal fashion.
Detritylation followed by LC-MS analysis revealed the
crude product to be free of cyanoethylated thymine residues
(Figure 5).
Large-Scale Synthesis of Phosphorothioate Oligonucle-
otides. With a suitable reagent system in hand, attention was
turned to the ISIS 2302 sequence. Synthesis was performed
at the 150 mmol scale on an OligoProcess DNA/RNA
synthesizer. Following complete chain assembly, the support-
bound oligonucleotide was again washed with a v/v solution
of TEA in CH3CN at a linear flow rate of 0.8 cm min-1 for
1 h. After being allowed to remain in contact with above
solution for an additional period of 12 h, the oligonucleotide
was cleaved from support and deprotected with ammonium
hydroxide in the usual manner. The crude synthetic oligo-
nucleotide was purified and detritylated in the normal fashion
and then analyzed for the presence of cyanoethyl adducts
by LC-MS. Parts a and b of Figure 6 show the mass spectra
obtained from untreated and TEA-treated ISIS 2302, respec-
tively. In the untreated control the +53 amu peak is clearly
seen at m/z ) 1604, while inspection of the TEA treated
material shows this peak to be absent.
Multiple phosphorothioate oligonucleotides of 20-mer in
length are routinely being synthesized at Isis Pharmaceuticals
at 300-600 mmol scale using the above reagent system to
generate high-quality antisense drugs for clinical evaluations
as well as potentially for the market.
Figure 5. MS of TEA treated DMT (TPS)18T.
autosampler. UV spectra were recorded on a HP 84524A
diode array spectrophotometer. Gas chromatography was
performed on an HP 5590 gas chromatograph using a J&W
DB1 Megabore column (30 mm × 0.53 mm, 5 µ) heated
from 40 to 220 °C at a rate of 10 °C/min. The injector and
detector temperatures were held at 220 and 280 °C, respec-
tively. HPLC-MS analyses were performed using a YMC
ODS AQ column (2 mm × 250 mm), eluted under a gradient
of CH3CN in triethylammonium acetate (TEAA) (20 mM,
pH 7), attached to an HP MSD 1100 spectrometer fitted with
an electrospray source. Spectra were recorded in the negative
ionization mode using a mass window of ca. 200 amu
centered about the M-4 charge state.
5′-O-(4,4′-Dimethoxytrityl)-N3-(2-cyanoethyl)thymi-
dine (6). 5′-O-(4,4′-Dimethoxytrityl)thymidine17 (10.5 g, 19.4
mmol) was dissolved in a mixture of THF and CH3CN (3:5
v/v, 80 mL), and acrylonitrile (3.1 g, 58 mmol) and tetra-
n-butylammonium hydroxide (0.3 mL of a 40% w/w solution
in water) were added; the mixture was heated at 60 °C for
Experimental Section
General. Reversed-phase high-performance liquid chro-
matography (RP-HPLC) was performed using a Waters 600E
controller set at 254 nm equipped with a Waters 717
836
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

and activated with a 0.45 M solution of tetrazole in CH3CN
(amidite:tetrazole ) 1:3 v/v). The coupling/recycling time
was set to 5 min. Sulfurization was effected using a 0.2 M
solution of phenylacetyldisulphide (PADS, Acharya Chemi-
cals, Mumbai, India) in picoline-CH3CN (1:1 v/v, 1.25
column volumes for a contact time of 2 min).18 Capping
reagents were made to recommended Pharmacia composi-
tion: Cap A: N-methylimidazole-CH3CN (1:4 v/v); Cap
B: acetic anhydride-pyridine-CH3CN (2:3:5, v/v/v).
Base Treatment and Ammonolysis of Crude Support-
bound Oligonucleotides. Cyanoethyl protecting groups were
removed prior to ammonolysis by treating support-bound
oligonucleotide (ca. 50 mg) with triethylamine-acetonitrile
(TEA- CH3CN) (1:1 v/v, 1 mL) for 12-16 h at 25 °C.
Alternatively, a solution of the same mixture was passed
through the synthesis column at a linear flow rate of 0.8
cm/min using a small peristaltic pump. After 1 h the pump
was turned off and the column allowed to remain in contact
with the solution for an additional 12-16 h. In both cases
the support-bound oligonucleotide was then washed with
CH3CN and dried under vacuum. Ammonolysis was carried
out by suspending the partially protected, support-bound
oligonucleotide (50 mg) in concentrated aqueous ammonium
hydroxide (28%, NH4OH; d 0.88, 1 mL) and heating the
mixture at 60 °C for 12-16 h. The cooled products were
then filtered, and the support was washed with EtOH-water
(1:1 v/v, 1 mL). The combined filtrate and washings were
concentrated in a SpeedVac.
Figure 6. MS of TEA untreated (a) and treated (b) ISIS 2302.
Rate of Removal of Cyanoethyl Groups. The rate of
removal of cyanoethyl groups from phosphorothioate oli-
gonucleotides was estimated by passing a solution of TEA-
CH3CN (1/1 v/v) through support-bound phosphorothioate
homothymidine octadecamer (TPS)18T contained within the
synthesis column at a flow rate of 0.8 cm/min. Four fractions
of equal volume were collected over a period of 4 h and the
fractions analyzed by GC for the presence of acrylonitrile.
Oligonucleotide Purification and Final Detritylation.
Crude DMT protected (DMT-on) oligonucleotide (ca. 2000
Figure 7. N3-Cyanoethylthymidine.
16 h. The cooled products were concentrated under vacuum;
a solution of the residue in CHCl3 (200 mL) was washed
with water (2 × 100 mL), dried (Na2SO4), and the CHCl3
was evaporated. The residual oil was purified by chroma-
tography on silica gel, and concentration of the fractions
eluted with CHCl3-MeOH (49:1 v/v) gave the title com-
pound (6) (3.5 g, 6 mmol, 30%) as a colorless glass17 (Figure
7).
N3-(2-Cyanoethyl)thymidine. 5′-O-(4,4′-Dimethoxytri-
tyl)-N3-(2-cyanoethyl)thymidine (0.5 g, 0.8 mmol) was
dissolved in AcOH-MeOH (1:1 v/v, 20 mL). After 2 h the
products were concentrated, and the residue was purified by
chromatography on silica gel. Evaporation of the fractions
eluted with CHCl3-MeOH (19:1 to 9:1 v/v) gave the title
compound (0.21 g, 0.7 mmol, 90%) as a colorless gum which
slowly crystallized upon standing.17 FAB-HRMS: calcd for
C13H17N3O5 (M + H)+ 296.1247, found 296.1230.
A260 units) was dissolved in water and purified by RP-HPLC
using a Waters NovaPak C18 column (3.9 mm × 300 mm)
eluted under a gradient of CH3CN in 0.1 M TEAA (pH 7).
The solution of DMT-on material (eluted with CH3CN-0.1
M TEAA, 1:1 v/v) was collected and concentrated. The
residue was dissolved in water to a concentration of ca. 1000
A260 units/mL, and 3.5 volumes of NaOAc (pH 3.0, 0.1 M)
was added. The products were incubated at room temperature
for 1 h, and then 2.5 M NaOAc (pH 7.5) was added to a
final concentration of ca. 0.35 M. The oligonucleotide was
precipitated by addition of 5 volumes of cold (-20 °C) EtOH
and collected by centrifugation.19
Oligonucleotide Synthesis. Oligonucleotide synthesis4a
was carried out on an Amersham Biosciences OligoPilot II
DNA/RNA synthesizer. All sequences were synthesized on
a ca. 165 µmol scale using Primer support (ca. 90 µmol/g,
Amersham Biosciences). Detritylation was achieved using
a 3% v/v solution of dichloroacetic acid in toluene for a
contact time of 5 min (flow rate ) 12.5 mL/min). Com-
mercial phosphoramidites, (2 equivalents per coupling)
purchased from Pierce (www.piercebc.com), were dissolved
to a nominal concentration of 0.2 M in anhydrous CH3CN
Desulfurization and Enzymatic Digestion. RP-HPLC-
purified ISIS 2302 (10 A260 units) in water (100 µL) was
added to 1 mL of a solution of iodine (0.2 g) and
(18) (a) Cheruvallath, Z. S.; Wheeler, P. D.; Cole, D. L.; Ravikumar, V. T.
Nucleosides Nucleotides 1999, 18, 484. (b) Cheruvallath, Z. S.; Carty, R.
L.; Moore, M. N.; Capaldi, D. C.; Krotz, A. H.; Wheeler, P. D.; Turney, B.
J.; Craig, S. R.; Gaus, H. J.; Scozzari, A. N.; Cole, D. L.; Ravikumar, V. T.
Org. Process Res. DeV. 2000, 4, 199.
(19) Krotz, A. H.; McElroy, B.; Scozzari, A. N.; Cole, D. L.; Ravikumar, V. T.
Org. Process Res. DeV. 2003, 7, 47.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
837

N-methylimidazole (1 ml) in water (3 mL) and THF (16 mL).
The products were incubated at 37 °C for 3 h and then
concentrated to a volume of ca. 100 µL. Water (1 mL) was
added and the mixture centrifuged. The supernatant was
removed and concentrated to a volume of ca. 100 µL. An
aqueous NaOAc solution (2.5 M, pH 7.5, 20 µL) was added
and the DNA precipitated by addition of ethanol (1 mL).
The DNA was washed with ethanol (1 mL) and dried. Snake
venom phosphodiesterase (SVPDE) [5 µL of a standard
solution in 100 mM trisHCl, 10 mM with respect to MgCl2,
pH 8.3 (2 units mL-1)] and alkaline phosphatase (AP) [5
µL of a standard solution in the same buffer (100 units
mL-1)] were added to a solution of the oligonucleotide in
100 mM tris-HCl (tris(hydroxymethyl)aminomethane hy-
drochloride), 10 mM with respect to MgCl2, pH 8.3 (200
µL), and the mixture was incubated at 37 °C for 16 h. The
products were analyzed by RP-HPLC using a Phenomenex
Luna C18 (5 µ) column eluted under a linear gradient of
methanol in water (7 to 100% over 30 min).
cyanoethyl adduct. This selectivity can be explained in part
by the instability of AN in NH4OH, by the relatively high
acidity of thymine residues, and at least in the case of cytidine
residues, by the fact that at the onset of ammonolysis these
residues are protected by a benzoyl group. In the case of
thymine the adduct has been identified as N3-cyanoethylth-
ymine and shown to be quite stable towards the ammonolysis
conditions. The observation of +53 amu adducts in synthetic
homopolymers of 2′-deoxyadenosine and 2′-deoxycytosine
under standard conditions was unexpected; the structures of
these adducts have not yet been determined. No evidence of
addition to guanine residues in terms either of adduct
formation or increased levels of depurination was observed.
The inherent rate differences between the bases mean that
in ISIS 2302 the vast majority of the ca. 2-3% of +53
adduct observed is composed of three compounds, and in
each compound one of the three thymine residues is replaced
by a N3-cyanoethylthymine. A simple method to avoid
cyanoethyl adduct formation has been developed. It was
shown that removal of the cyanoethyl protecting groups with
TEA in CH3CN at room temperature prior to heating in
ammonium hydroxide results in the production of synthetic
oligonucleotides free from cyanoethylated thymine residues.
This simple and cost-effective methodology has become part
of our routine, large-scale oligonucleotide manufacturing
process.
Conclusions
Acrylonitrile (AN) adds to nucleobases of DNA under
forcing conditions. The present study revealed that AN
generated under the ammonolysis conditions employed for
deprotection of synthetic phosphorothioate oligonucleotides
is also capable of adduct formation with DNA bases. Studies
on a variety of sequences indicated that, while AN is capable
of adding to adenine, cytidine, and thymine residues, only
the latter react quickly enough to form high levels of
Received for review October 30, 2002.
OP020090N
838
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development