UnyLinker: An efficient and scaleable synthesis of oligonucleotides utilizing a universal linker molecule: A novel approach to enhance the purity of drugs

Article abstract of DOI:10.1021/op8000178

A novel universal linker (UnyLinker) molecule which has a conformationally rigid and chemically stable bridge head ring oxygen atom carrying a conventional 4,4-dimethoxytrityl (DMT) and succinyl groups locked in a syn orientation has been developed to car

Full text of DOI:10.1021/op8000178

Organic Process Research & Development 2008, 12, 399–410
UnyLinker: An Efficient and Scaleable Synthesis of Oligonucleotides Utilizing a
Universal Linker Molecule: A Novel Approach To Enhance the Purity of Drugs
Vasulinga T. Ravikumar,*,† R. Krishna Kumar,† Phil Olsen,† Max N. Moore,† Recaldo L. Carty,† Mark Andrade,†
Dennis Gorman,† Xuefeng Zhu,† Isaiah Cedillo,† Zhiwei Wang,† Lucio Mendez,† Anthony N. Scozzari,† Gerardo Aguirre,‡
Ratnasamy Somanathan,*,‡ and Sylvain Bernee`s§
Isis Pharmaceuticals, Inc. 2282 Faraday AVenue, Carlsbad, California 92008 U.S.A., Centro de Graduados e InVestigacio´n del
Instituto Tecnolo´gico de Tijuana, Apdo. Postal 1166, Tijuana, B.C., Me´xico, and DEP, Facultad de Ciencias Quı´micas, UANL,
Guerrero y Progreso S/N, Col. TreVin˜o, 64570, Monterrey, N.L., Me´xico
Abstract:
advancing rapidly in the clinic, and two drugs have already been
approved by the Food and Drug Administration and other
international regulatory agencies.1,2 Many more drugs are in
phase II trials, and few are in potentially pivotal phase III trials.3
These drugs are currently synthesized using solid-phase chem-
istry via the phosphoramidite approach4 where a solid support
containing the first nucleoside attached through a cleavable
linker is used as the polymer-matrix and ꢀ-cyanoethyl protected
phosphoramidites of various nucleosides are used as the building
blocks. The synthesis is performed in an automated DNA/RNA
synthesizer using various commercially available solid supports.
Thus, four solid supports are required for synthesis of oligode-
oxyribonucleotides. The advancement of antisense therapeutics
has resulted in the use of RNA-DNA-RNA chimeric mol-
ecules wherein the RNA wings consists of 2′-O-alkyl RNA
nucleosides (2′-O-methoxyethyl or 2′-O-methyl).5 In addition,
other modified nucleosides such as 4′-thio, Locked nucleic acid,
2′-R-fluoro nucleic acids are also being evaluated for various
applications. This has necessitated the need to have several
additional prederivatized solid supports.
A novel universal linker (UnyLinker) molecule which has a
conformationally rigid and chemically stable bridge head ring
oxygen atom carrying a conventional 4,4′-dimethoxytrityl (DMT)
and succinyl groups locked in a syn orientation has been developed
to carry out oligonucleotide synthesis efficiently and smoothly. The
geometry of the vicinal syn oxygen functionalized group allows
fast and clean cleavage under standard aqueous ammonia depro-
tection conditions to afford high-quality oligonucleotides. No base
modification is observed, based on the ion-pair HPLC-UV-MS
(IP-HPLC-UV-MS) method with detection limit of <0.1%. A
class of impurities formed by branching from the exocyclic amino
group of nucleosides loaded onto a solid support has been
eliminated by the use of this method. Examples demonstrating the
versatile nature of this molecule are shown by syntheses of different
chemistries such as 2′-deoxy, 2′-O-methyl, 2′-O-methoxyethyl, Lock
nucleic acids (LNA), 2′-r-fluoro nucleic acids (FANA), conjugates
such as 5′-phosphate monoester and biotin, and phosphate diester
and phosphorothioate backbone modifications. This molecule was
loaded onto several commercial solid supports and used in both
gas-sparged and packed-bed automated DNA/RNA synthesizers.
Large-scale syntheses (up to 700 mmol) of multiple phosphorothio-
ate first- and second-generation antisense drugs on GE-Amer-
sham’s OligoProcess synthesizer are demonstrated further, show-
ing that this chemistry could be used for efficient synthesis of
multiple oligonucleotide drugs using a single raw material, thereby
eliminating a difficult to characterize nucleoside-loaded polymer
matrix used as a starting material. A mechanism for deprotection
and cleavage of the linker molecule to liberate the free oligonucle-
otide is proposed. Characterization of the cyclic byproduct formed
during release of the oligonucleotide is presented. The exo-syn
configuration of the dihydroxy structure of the UnyLinker
molecule is conclusively established by X-ray crystallography
studies. A novel method to remove the last traces of osmium used
during the synthesis of the UnyLinker molecule to reach undetect-
able levels (<1 ppm) is also described.
(1) Vitravene and Macugen are two oligonucleotide-based drugs approved
by the FDA.
(2) (a) Antisense strategies: Crooke, S. T. Curr. Mol. Med. 2004, 4, 465.
(b) Progress in antisense technology: Crooke, S. T. Ann. ReV. Med.
2004, 55, 61. (c) Crooke, S. T. Progress in Antisense Technology. In
Drug DeliVery Systems in Cancer Therapy; Brown, D. M., Ed. Humana
Press: Totowa, NJ, 2004. (d) Crooke, S. T.; Twain, M. Antisense
Therapy. In Cytokine Handbook, 4th ed., Thomson, A. W., Lotze,
M. T., Eds.; Elsevier Sciences: London, U. K., 2003. (e) Dias, N.;
Stein, C. A. Antisense Oligonucleotides: Basic Concepts and Mech-
anisms. Mol. Cancer Ther. 2000, 1, 47. (f) Bennett, C. F. Efficiency
of Antisense Oligonucleotide Drug Discovery. Antisense Nucleic Acid
Drug DeV. 2002, 12, 215. (g) Kurreck, J. Eur. J. Biochem. 2003, 270,
1628.
(3) (a) A number of second-generation phosphorothioate oligonucleotides
are in various stages of preclinical and clinical trials against APOB-
100, PTP-1B, VLA4, TRPM2, survivin, STAT-3. eIF-eE, GCGR,
GCCR, SOD1, SGLT2, CRP, etc. for the treatment of a variety of
diseases such as cancer, psoriasis, diabetes, asthma, arthritis, multiple
sclerosis, etc. (b) About 40 drugs are being evaluated in the clinic.
Isis Pharmaceuticals, Inc. (www.isispharm.com) has the largest
portfolio of antisense oligonucleotide products in clinical trials. Visit
www.coleypharm.com for CpG-based oligonucleotides in clinical trials.
(4) (a) Beaucage, S. L.; Iyer, R. P. Tetrahedron 1992, 48, 2223. (b)
Beaucage, S. L.; Iyer, R. P. Tetrahedron 1993, 49, 1925. (c) Beaucage,
S. L. Oligonucleotide Synthesis. In DNA and Aspects of Molecular
Biology; Kool, E. T., Vol. Ed.; Comprehensive Natural Products
Chemistry; Barton, D. H. R.; Nakanishi, K., Eds.; Pergamon Press:
Elmsford, NY, 1999; Vol. 7.05, 105.
Introduction
Oligonucleotide-based drugs (antisense, aptamer, DNA
decoys, CpG-based immunomodulatory drugs, siRNA, etc.) are
* Corresponding author. Telephone: (760)-603-2412. Fax: (760)-603-4655.
E-mail: vravikumar@isisph.com.
(5) Altmann, K.-H.; Dean, N. M.; Fabbro, D.; Freier, S. M.; Geiger, T.;
Haner, R.; Husken, D.; Martin, P.; Monia, B. P.; Muller, M.; Natt, F.;
Nicklin, P.; Phillips, J.; Pieles, U.; Sasmor, H.; Moser, H. E. Chimia
1996, 50, 168.
† Isis Pharmaceuticals, Inc.
‡ Centro de Graduados e Investigacio´n del Instituto Tecnolo´gico de Tijuana.
§ DEP, Facultad de Ciencias Químicas, UANL.
10.1021/op8000178 CCC: $40.75
Published on Web 05/16/2008
2008 American Chemical Society
Vol. 12, No. 3, 2008 / Organic Process Research & Development
•
399

Scheme 1
An important aspect in the development of any drug is the
cost of manufacturing, and this assessment is even more
important for a non-small-molecule drug such as a phospho-
rothioate oligonucleotide where multiple drugs are advancing
in human clinical evaluation to treat large patient populations
suffering from chronic diseases such as diabetes and cholesterol.
This novel therapeutic approach may soon lead to a necessity
to manufacture several hundreds of kilograms of drug. Recently,
we have examined several aspects of oligomerization chemistry
with an aim to reduce cost, improve quality, and make the
chemistry more rugged, greener, and safe.6–9 Since the antisense
approach is a platform technology, it is a given that this
approach of drug discovery is faster, better, and cheaper for
testing against several diseases. Thus, in a relatively short period
of time multiple drugs have advanced to human clinical trials,
and for various reasons some of them may fail to become an
approved drug. Since not all drugs contain the same 3′-
nucleoside, a clinical failure will lead to waste of a potentially
large inventory of expensive solid support containing that
particular loaded nucleoside.
Figure 1. Pentacoordinated transition state to facilitate cleavage.
Figure 2. Molecules evaluated as universal linkers on solid
support.
FDA guidelines a starting material has to be well characterized,
and it is almost impossible with existing technology to
characterize the nucleoside-loaded polymer matrix and show
consistency in quality. Thus, it would be a real advantage if
we can eliminate this starting material and still make the drug
on a solid support. One solution is to use a solid support loaded
with a molecule which could act as a linker between the support
and the oligonucleotide to be synthesized.
For high-throughput screening of oligonucleotides, 396-well
plate synthesizers capable of synthesizing multiple oligonucle-
otides at the same time are becoming popular. This requires
placing appropriate nucleoside-loaded solid supports in the
wells, and enough chances exist for human error to cause failure
and discarding of potentially the complete plate. For synthesis
of novel oligonucleotides, a continuously growing number of
supports carrying other modified nucleosides and 3′-terminal
modifiers are required. Universal supports are also expected to
reduce side reactions that take place on the preattached
nucleoside of standard supports because universal supports lack
any nucleosides attached to them.9d,11
In addition, the nucleoside-loaded solid support is considered
as a starting material and not a raw material.10 According to
(6) (a) Cheruvallath, Z. S.; Wheeler, P. D.; Cole, D. L.; Ravikumar, V. T.
Nucleosides Nucleotides 1999, 18, 485. (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. (c) Krotz,
A. H.; Gorman, D.; Mataruse, P.; Foster, C.; Godbout, J. D.; Coffin,
C. C.; Scozzari, A. N. Org. Process Res. DeV. 2004, 8, 852. (d)
Ravikumar, V. T.; Andrade, M.; Carty, R. L.; Dan, A.; Barone, S.
Bioorg. Med. Chem. Lett. 2006, 16, 2513. (e) Krotz, A. H.; Hang, A.;
Gorman, D.; Scozzari, A. N. Nucleosides Nucleotides Nucleic Acids
2005, 24, 1293.
(7) (a) Krotz, A. H.; Cole, D. L.; Ravikumar, V. T. Bioorg. Med. Chem.
1999, 7, 435. (b) Krotz, A. H.; Cart, R. L.; Moore, M. N.; Scozzari,
A. N.; Cole, D. L.; Ravikumar, V. T. Green Chem. 1999, 277. (c)
Krotz, A. H.; Carty, R. L.; Scozzari, A. N.; Cole, D. L.; Ravikumar,
V. T. Org. Process Res. DeV. 2000, 4, 190.
(8) (a) Capaldi, D. C.; Scozzari, A. N.; Cole, D. L.; Ravikumar, V. T.
Org. Process Res. DeV. 1999, 3, 485. (b) Hayakawa, Y.; Hirata, A.;
Sugimoto, J.; Kawai, R.; Sakakura, A.; Kataoka, M. Tetrahedron 2001,
57, 8823.
To address all these issues, several research groups have
proposed the concept of a universal solid support using a cis-
3,4-dihydroxytetrahydrofuran or a suitably functionalized nu-
cleoside or other acyclic molecule.12 The oligonucleotides
synthesized using these universal solid supports could be used
either for research or for pharmaceutical applications where very
high standards of drug quality are needed. In our hands, when
evaluated for oligonucleotide synthesis and analyzed carefully
(9) (a) Capaldi, D. C.; Gaus, H.; Krotz, A. H.; Arnold, J.; Carty, R. L.;
Moore, M. N.; Scozzari, A. N.; Lowery, K.; Cole, D. L.; Ravikumar,
V. T. Org. Process Res. DeV. 2003, 7, 832. (b) Gaus, H.; Olsen, P.;
Van Sooy, K.; Rentel, C.; Turney, B.; Walker, K. L.; McArdle, J. V.;
Capaldi, D. C. Bioorg. Med. Chem. Lett. 2005, 15, 4118. (c) Rentel,
C.; Wang, X.; Batt, M.; Kurata, C.; Oliver, J.; Gaus, H.; Krotz, A. H.;
McArdle, J. V.; Capaldi, D. C. J. Org. Chem. 2005, 70, 7841. (d)
Kurata, C.; Bradley, K.; Gaus, H.; Luu, N.; Cedillo, I.; Ravikumar,
V. T.; Van Sooy, K.; McArdle, J. V.; Capaldi, D. C. Bioorg. Med.
Chem. Lett. 2006, 16, 607.
(10) Starting material is any chemical component that contributes an atom
to the drug (e.g., phosphoramidite), and a raw material (solvents,
activators such as 4,5-dicyanoimidazole) does not.
(11) Cazenave, C.; Bathany, K.; Rayner, B. Oligonucleotides 2006, 16,
181.
400
•
Vol. 12, No. 3, 2008 / Organic Process Research & Development

by ion-pair–liquid chromatography-electrospray mass spec-
troscopy (IP-LC-EMS) at least three of the reported universal
solid supports showed that the quality of drug produced is not
the same as the quality of drug produced with the standard
method and were contaminated with unacceptable levels of the
pendent linker molecule. In addition, there was a considerable
(ca. 10%) amount of oligonucleotide still attached to the solid
support, making it an inefficient approach. Addition of salt such
as lithium chloride or heating at high temperatures (75–80 °C)
or use of heavy metals or sulfides or use of strong nucleophiles
such as 40% aqueous methylamine solution is recommended
for release of oligonucleotides from some linker molecules.
These conditions may not be practical or may add additional
cost or may not be acceptable for large-scale synthesis or for
therapeutic applications. The above limitations have thus far
outweighed the added convenience of using one solid support
material for all sequences, and prederivatized supports are still
predominantly preferred. Clearly, while important strides have
been made in this area, additional research efforts are warranted.
For a universal solid support to be widely accepted, it has
to produce good yield of high-quality oligonucleotides and be
applicable for synthesizing different kinds of oligonucleotide
analogs. In addition, the universal linker molecule should be
easily scaleable and inexpensive at large scales. It should work
well across different solid supports, chemistries, and synthesiz-
ers. The molecule should be stable to oligomerization conditions
and cleaved under standard ammonium hydroxide incubation
conditions (55 °C) without use of any additives.
Figure 3. Exo and endo structures of olefin 9.
Figure 4. Exo and endo structures of diol 10.
The universal support described in this work has several
features that should enable greater acceptance by academics
for performing basic research and by industry for development
of various therapeutics: (a) the oligonucleotide is released from
the linker molecule quantitatively under standard deprotection
Figure 5. ORTEP diagrams (different projections) of diol 10.
(12) (a) Kumar, P.; Gupta, K. C. HelV. Chim. Acta 2003, 86, 59. (b) Kumar,
P.; Mahajan, S.; Gupta, K. C. J. Org. Chem. 2004, 69, 6482. (c) Kumar,
P.; Dhawan, R.; Chandra, R.; Gupta, K. C. Nucleic Acids Res. 2002,
30, e130. (d) Ghosh, P. K.; Kumar, P.; Gupta, K. C. Ind. J. Chem.
Soc. 1998, 75, 206. (e) Kumar, P.; Gupta, K. C. Nucleic Acids Res.
1990, 27, e2. (f) Kumar, P.; Gupta, K. C. React. Funct. Polym. 1999,
41, 197. (g) Lyttle, M. H.; Hudson, D.; Cook, R. M. Nucleic Acids
Res. 1996, 24, 2793. (h) Lyttle, M. H.; Dick, D. J.; Hudson, D.; Cook,
R. M. Nucleosides Nucleotides 1999, 18, 1809. (i) Gough, G. R.;
Brunden, M. J.; Gilham, P. T. Tetrahedron Lett. 1983, 24, 5321. (j)
deBear, J. S.; Hayes, J. A.; Koleck, M. P.; Gough, G. R. Nucleosides
Nucleotides 1987, 6, 821. (k) Schwartz, M. E.; Breaker, R. R.;
Asteriadis, G. T.; Gough, G. R. Tetrahedron Lett. 1995, 36, 27. (l)
Scheuer-Larsen, C.; Rosenbohm, C.; Joergensen, T. J. D.; Wengel, J.
Nucleosides Nucleotides 1997, 16, 67. (m) Nelson, P. S.; Muthini, S.;
Vierra, M.; Acosta, L.; Smith, T. H. Biotechniques 1997, 22, 752. (n)
Azhayev, A. V. Tetrahedron 1999, 55, 787. (o) Scheuer-Larson, C.;
Rosenbohm, C.; Jorgensen, T. J. D.; Wengel, J. Nucleosides Nucle-
otides 1997, 16, 67. (p) Nelson, P. S.; Muthini, S.; Kent, M. A.; Smith,
T. H. Nucleosides Nucleotides 1997, 16, 1951. (q) Azhayev, A. V.;
Antopolsky, M. L. Tetrahedron 2001, 57, 4977. (r) Zheng, X.; Gaffney,
B. L.; Jones, R. A. Nucleic Acids Res. 1997, 25, 3980. (s) Crea, R.;
Horn, T. Nucleic Acids Res. 1980, 8, 2331. (t) Scott, S.; Hardy, P.;
Sheppard, R. C.; McLean, M. J. In InnoVations and PerspectiVes in
Solid-Phase Synthesis; Epton, R., Ed.; Mayflower Worldwide: Bir-
mingham, U. K., 1994; pp 115. (u) Anderson, E.; Brown, T.; Picken,
D. Nucleosides Nucleotides 2003, 22, 1403. (v) Ferreira, F.; Meyer,
A.; Vasseur, J-J.; Morvan, F. J. Org. Chem. 2005, 70, 9198. (w)
Guzaev, A. P.; Manoharan, M. J. Am. Chem. Soc. 2003, 125, 2380.
(x) Krishna Kumar, R.; Guzaev, A. P.; Rental, C.; Ravikumar, V. T.
Tetrahedron 2006, 62, 4528. (y) Wang, Z.; Olsen, P.; Ravikumar, V. T.
Nucleosides Nucleotides and Nucleic Acids 2007, 26, 259. (z)
Anderson, K. M.; Jaquinod, L.; Jensen, M. A.; Ngo, N.; Davis, R. W.
J. Org. Chem. 2007, 72, 9875. (aa) Mackie, H. Glen Res. Rep. 2001,
141.
conditions; (b) the linker molecule is compatible with various
modified chemistries as shown by extensive examples of
syntheses and analyses; (c) it enables the synthesis of high-
quality oligonucleotides (as judged by LC-MS analysis)
without any detectable levels of base modification.
Results and Discussion
Design of UnyLinker Molecule. During our extensive
investigation during this project, we initially designed a
molecule 4 which on loading to a solid support and used led to
the formation of high-quality oligonucleotides with similar or
higher yields compared to those from nucleoside-loaded sup-
ports (Scheme 1). However, after completing all the laboratory-
scale research work, when we started to scale up, we realized
that this molecule failed to scale up efficiently. Isolated yields
of two intermediates, 2 and 3, were low, and it was not
considered economical if we were to scale up to quantities of
several hundreds of kilograms. In addition, there were some
drawbacks in the design of this molecule. During the osmium-
catalyzed dihydroxylation step to form 2, the reaction has to
be monitored by 1H NMR (using relatively expensive deuterated
pyridine) since there is no UV chromophore present in the
molecule. Second, the dicarboxylic acid 3 was highly water
soluble even in presence of a hydrophobic group such as 4,4′-
dimethoxytrityl. Isolation of the product was found to be difficult
Vol. 12, No. 3, 2008 / Organic Process Research & Development
•
401

Scheme 2. Synthesis of UnyLinker succinate 12
Scheme 3. Synthesis of pixyl-derivatized UnyLinker 14
and cumbersome. Third, the final molecule 4 is an anhydride,
and the purity was difficult to quantify by HPLC as it underwent
hydrolysis under the analysis conditions. In addition, loading
of 4 to yield solid support 5 is not straightforward. The cyclic
anhydride ring opens during loading, liberating a free carboxyl
group that needs to be capped. This protection is done by
reacting with an amine in the presence of an activator such as
HBTU. However, it will be difficult to ensure all carboxyl
groups are capped. The presence of a free carboxyl group was
found to hinder oligomerization yield particularly during
coupling when low excess (1.7 equiv) of amidites was used.
Because of these various hurdles in the use of this molecule
(particularly not being able to scale up economically) we were
forced to abandon this route. Subsequently, we started inves-
tigating the design of a new and better molecule that retained
all the key features of the above molecule while still solving
all the issues to make it a versatile chemistry for various
applications.
In the design of a universal linker molecule, both the angle
R and the distance d are important for efficient hydrolysis.
Optimal values are R equals 90° or 180° and d equals 3 Å. In
syn conformation of hydroxyl groups, d is reduced and R is
increased. Because of this stabilized transition state, more rapid
hydrolysis is expected (Figure 1).
We have not thoroughly investigated the necessity of two
carboxyl groups attached to the bicyclic ring system although
we know that the bridge head oxygen is needed for efficient
synthesis since linker molecules based on norbornylene (6),
5-norbornene-2-carboxylic acid (7), and 5-norbornene-2-endo-
3-exo-dicarboxylic acid (8) did not perform efficiently
(Figure 2).
and also possess a UV chromophore. Thus, we had to design
a new molecule that includes all of these needed features. We
report here that a vicinal diol-based linker molecule, UnyLinker
(12) has all the salient features of a good universal linker. We
hypothesized that locking the two vicinal C-O bonds of an
alkyl 2-hydroxyethyl phosphate in a syn-periplanar conforma-
tion could make the distance between the 2′-O and P(V) atoms
appropriate for efficient attack and release of the oligonucleotide.
We proposed 12 to be an optimally designed molecule for
efficient synthesis of oligonucleotides. Thus, UnyLinker 12 was
synthesized and loaded onto various solid supports and evalu-
ated for the synthesis of multiple modified oligonucleotides in
different designed automated synthesizers. The molecule has
been scaled up to multikilogram quantities without any hurdle
and successfully loaded onto polymeric supports on very large
scales.
Synthesis of UnyLinker (12). The synthesis starts with the
Diels–Alder reaction of furan and N-phenyl maleimide to afford
the adduct 9 as a colorless crystalline solid (95%). Bis-
hydroxylation of olefin 9 was carried out using hydrogen
peroxide and a catalytic amount of osmium tetroxide (85%).
Chemoselective protection of diol 10 with 4,4′-dimethoxytrityl
chloride in pyridine gave the product 11 as a colorless,
amorphous solid (80%). Treatment with succinic anhydride in
the presence of triethylamine in methylene chloride afforded
12 as a colorless to pale-yellow amorphous solid in high
(>60%) overall yield based on N-phenylmaleimide (Scheme
2).
Synthesis of Diels–Alder Adduct (9). Furan-based Diels-
Alder adducts have found extensive use in organic synthesis;
the resulting 7-oxanorbornene is often elaborated and/or ring-
opened, providing routes to a wealth of valuable synthetic
Assuming that the two carbonyl groups are desired, it would
be desirable if they are masked in the form of a cyclic imide
402
•
Vol. 12, No. 3, 2008 / Organic Process Research & Development

targets.13 Thus, Diels–Alder addition with N-phenylmaleimide
gave the adduct 9.14 Depending on the condition used, exo or
a mixture of endo and exo isomers could be obtained (Figure
3). Since we wanted to have a single thermodynamically more
stable exoisomer, we refluxed the reaction mixture in MeCN
as solvent. Similar results were obtained when toluene or xylene
X-ray Crystallography Studies of Diol 10. The structure
of diol 10 was confirmed to be exo,exo-cis-8,9-dihydroxy-4-
phenyl-10-oxa-4-aza-tricyclo[5,2,1,02,6]decane-3,5-dione based
on X-ray crystallography. Single crystals suitable for X-ray
structure analysis were obtained by slow evaporation from
a concentrated solution in dimethylformamide:methanol
at 298 K.
1
was used as solvent. As expected, H NMR of the product
Refinement. The H atoms bonded to O1 and O2 were found
in a difference Fourier map and refined with free coordinates
and isotropic U parameters. The C-bound H atoms were placed
in idealized positions and refined as riding on their parent C
atom, with the following constraints: 0.93 %A for aromatic CH,
0.98 %A for methine CH. Isotropic U parameters were fixed
showed a singlet for the C-5/C-6 protons (δ 3.0 ppm) corre-
sponding to the exo isomer, with the absence of a coupled signal
around δ 3.8 ppm attributable to the endo isomer. The
7-oxabicyclo[2.2.1]heptane ring system is conformationally rigid
with a dihedral angle (H-C5-C6-H) of almost 90° for the
exo isomer.14a
Synthesis of Dihydroxy Compound (10). The dihydroxy-
lation of N-phenylmaleimide-furan adduct has received little
attention although such reactions with other Diels–Alder adducts
and other olefins have been reported extensively in the
literature.15–17 The adduct 9 was cis-dihydroxylated using the
method described by Milas18 which involves oxidation using a
catalytic amount of osmium tetroxide in the presence of
hydrogen peroxide in tert-butyl alcohol. Excellent yields were
obtained by refluxing for 8 h to obtain the desired product as
colorless, amorphous powder.
Stereochemistry of cis-Glycol 10. It has been established
that hydroxylation with osmium tetroxide produces cis-glycol.19
However, in the present case two cis-glycols are possible, viz.
10a and 10b (Figure 4). A choice can be made by applying
Alder’s rule of exoaddition.20 Since the cis dihydroxylation
proceeds21 via the osmate ester of glycol, steric consideration
for this intermediate strongly suggests the dihydroxylated
compound has the configuration shown in 10a i.e. the exo-cis-
glycol.18 For the formation of this product, there is less steric
interference between osmium tetroxide and the oxygen bridge
than between the reagent and the two-carbon bridge.
˜
at U1so˜(H) ) 1.2U∼eq˜(carrier atom) for aromatic CH and
methine CH (Figure 5).
Monoprotection of Diol and Succinylation. Efficient
monoprotection of diol 10 with 4,4′-dimethoxytrityl chloride
depends on its quality. Generally the diol is coevaporated twice
with pyridine to ensure it is anhydrous. Otherwise, more than
1.2 equiv of DMT chloride is needed and renders the product
purification difficult. In addition to pyridine, other bases such
as lutidine and syn-collidine were found to perform equally
efficiently for chemoselective protection. Subsequent treatment
of 11 with succinic anhydride in dichloromethane afforded the
UnyLinker succinate 12 as its triethylammonium salt. Formation
of 12 was found to be slow when ethyl acetate was used as
solvent.
Synthesis of Pixyl-Derivatized UnyLinker (14). Earlier,
we observed that removing of DMT group from the secondary
hydroxyl of 5 with dichloroacetic acid in an organic solvent
such as toluene was relatively slow compared to removing the
(13) (a) Kappe, C. O.; Murphee, S. S.; Padwa, A. Tetrahedron 1997, 53,
14179. (b) Lautens, M.; Rovis, T. Tetrahedron 1998, 54, 1107.
(14) (a) Cooley, J. H.; Williams, R. V. J. Chem. Educ. 1997, 74, 582. (b)
Keller, K. A.; Guo, J.; Punna, S.; Finn, M. G. Tetrahedron Lett. 2005,
46, 1181. (c) Rulisek, L.; Sebek, P.; Havlas, Z.; Hrabal, R.; Capek,
P.; Svatos, A. J. Org. Chem. 2005, 70, 6295. (d) Sinner, F. M.;
Buchmeiser, M. R. Angew. Chem., Int. Ed. 2000, 39, 1433.
(15) (a) Kolb, H. C.; Van Nieuwenhze, M. S.; Sharpless, K. B. Chem. ReV.
1994, 94, 2483. (b) Schroder, M. Chem. ReV. 1980, 80, 187. (c)
Sharpless, K. B.; Akashi, K. J. Am. Chem. Soc. 1978, 98, 1986. (d)
Akashi, K.; Palermo, R. E.; Sharpless, K. B. J. Org. Chem. 1978, 43,
2063. (e) Kobayashi, S.; Ishida, T.; Akiyama, R. Org. Lett. 2001, 3,
2649. (f) Kobayashi, S.; Endo, M.; Nagayama, S. J. Am. Chem. Soc.
1999, 121, 11229.
Figure 6. UnyLinker succinate loaded onto various solid
supports.
(16) (a) Horiuchi, C. A.; Dan, G.; Sakamoto, M.; Suda, K.; Usui, S.;
Sakamoto, O.; Kitoh, S.; Watanabe, S.; Utsukihara, T.; Nozaki, S.
Synthesis 2005, 2861.
Figure 7. Noncritical impurities formed during synthesis of
UnyLinker succinate 12.
(17) (a) Fatialdi, A. J. Chem. ReV. 1980, 80, 85. and references cited therein.
(b) Bhushan, V.; Rathore, R.; Chandrasekaran, S. Synthesis 1984, 431.
(c) Taylor, E. C.; Janini, T. E.; Elmer, O. C. Org. Process Res. DeV.
1998, 2, 147. (d) Branytska, O.; Neumann, R. Synlett 2005, 2525. (e)
Horiuchi, C. A.; Dan, G.; Sakamoto, M.; Suda, K.; Usui, S.; Sakamoto,
O.; Kitoh, S.; Watanabe, S.; Utsukihara, T.; Nozaki, S. Synthesis 2005,
2861.
(18) Milas, N. A.; Sussman, S. J. Am. Chem. Soc. 1936, 58, 1302.
(19) (a) Daniels, R.; Fischer, J. L. J. Org. Chem. 1963, 28, 320. (b) Freire,
F.; Calderon, F.; Seco, J. M.; Fernandez-Mayoralas, A.; Quinoa, E.;
Riguera, R. J. Org. Chem. 2007, 72, 2297 and references cited therein.
(20) (a) Alder, K.; Stein, G. Ann. 1935, 515, 185. (b) Alder, K.; Wirtz, H.;
Koppelberg, H. Ann. 1956, 601, 138.
Figure 8. Influence of 3′-substitution on depurination of dA
nucleotide.
(21) (a) Criegee, R. Ann. 1936, 522, 75. (b) Criegee, R.; Marchand, B.;
Wannowius, H. Ann. 1942, 550, 99.
Vol. 12, No. 3, 2008 / Organic Process Research & Development
•
403

Scheme 4. Postulated mechanism for deprotection and
release of oligonucleotide
of a nucleoside succinate. The molecule was stable to at least
one year when stored at refrigerated temperatures. Third, the
water solubility problem has been completely eliminated.
Compound 12 is hydrophobic and easily soluble in many
organic solvents. Finally, during synthesis of 11, bis-DMT
protected compound 17 is also formed (∼5–10%) and is
removed easily by selective extractions. However, even if this
compound is present as a contaminant, it will not undergo
further reaction with succinic anhydride and will be inert
towards loading. Similarly, if compound 18 is formed (due to
succinylation of any unreacted diol 10 that is carried over), it
will get loaded to solid support but is incapable of chain
elongation and will be innocuous (Figure 7). Thus, UnyLinker
12 is optimally designed for evaluating the synthesis of various
oligonucleotides.
Synthesis of Oligonucleotides. Extensive evaluation of
UnyLinker was carried out by synthesizing DNA and several
modified oligonucleotides such as 2′-MOE, 2′-OMe, LNA,
OMe/RNA mixed backbone, conjugates such as biotin, 5′-
phosphate, 3′-phosphate, etc. Both phosphate and phospho-
rothioate diester backbones were synthesized. Syntheses of
multiple oligonucleotides at various scales on different auto-
mated synthesizers were successfully carried out.
Small-Scale Synthesis of Phosphorothioate Oligonucle-
otides. 20-Mer 2′-O-methoxyethyl gapmer (0.2 µmol scale) P
) S (5-10-5) [5′-[CTCTA]-GTTCCTCTCA-[ATGTC]] was
synthesized on MerMade plate synthesizer. UnyLinker loaded
on CPG (45 µmol/g loading) was used. Conditions used for
oligonucleotide synthsis were 4,5-dicyanoimidazole (0.5 M
solution in ACN) for coupling, Beaucage reagent (0.05 M
solution in ACN) for thiolation, 3% DCA/toluene for detrity-
lation. At the end of the synthesis, the solid support was
incubated with concentrated aqueous ammonium hydroxide at
55 °C for 8 h. Crude oligonucleotides were analyzed by IP-
LC-MS (Table 1).
Medium-Scale Synthesis of Phosphorothioate Oligo-
nucleotides. Phosphorothioate oligonucleotide 4 (Table 6) (5-
10-5 MOE gapmer) was chosen as an example; 1 mmol scale
on Akta 100 synthesizer was performed; 10% DCA/toluene for
detritylation, 0.2 M PADS for sulfurization, 1H-tetrazole for
coupling, 2.0 equiv of phosphoramidites for coupling, triethy-
lamine/MeCN treatment to remove the cyanoethyl group while
the oligo is still bound to the support, etc. were used; crude
weight was determined at end of synthesis; the support was
then incubated in a known amount of ammonia, diluted with a
known amount with water, and quantitated with RP HPLC. The
crude oligonucleotide was analyzed by IP-LC-MS; both
oligonucleotides were purified on a BioCad instrument; the
DMT-on oligonucleotide was detritylated and lyophilized to
constant weight (Table 2).
Evaluation of UnyLinker on OligoPrep. Application of
UnyLinker on OligoPrep was evaluated at 1 mmol scale by
synthesizing phosphorothioate oligonucleotide 4 (Table 6) and
was compared to MOE meC-nucleoside-loaded OligoPrep and
PS200 solid supports. Oligomerization conditions similar to
the above synthesis were followed for OligoPrep except that
twice the volume of deblock (2 × delivery time) was used
primary hydroxyl group; extended time (three times) was needed
for complete removal. If removal of DMT group is not complete
during oligomerization, it could be removed during subsequent
acid-treatment cycles, leading to an increased (n - 1)-mer
(deletionmers) in the final synthesized oligonucleotide. This
prompted us to replace DMT with a more labile group. In
connection with a different project aimed at reducing depuri-
nation, we were developing a DMT-like behaving but much
easier to remove under acidic conditions viz. trimethylpixyl
group as an alternative to the DMT group for oligonucleotide
synthesis. This group is about 8 times faster to remove compared
to DMT when treated under identical acidic conditions. Hence,
we wanted to investigate if this pixyl group could minimize
the time needed for complete removal. Treatment of diol 10
with 2,7-dimethyl-9-chloro-9-toluoylxanthene (pixyl chloride;
made in our laboratory) in the presence of pyridine gave the
monoprotected compound 13 in 73% isolated yield. Subsequent
treatment with succinic anhydride in dichloromethane afforded
the UnyLinker succinate 14 as its triethylammonium salt in
almost quantitative yield (Scheme 3).
Loading of UnyLinker to Solid Supports. Solid supports
are generally functionalized with amino or hydroxyl groups.
Depending on the functionality, loading conditions could vary
slightly. Hence, a study was undertaken to evaluate the influence
of activators, solvents, bases, and loading time. We found that
HBTU and HATU were the best activators, MeCN was the
best solvent, and ethyldiisopropylamine or triethylamine was
the preferred base. Controlled pore glass (CPG), HL30, PS200,
NittoPhase, OligoPrep, and few other polymeric supports were
evaluated as solid supports (Figure 6). These supports were
loaded in the range of 25–325 µmol/g as recommended by the
manufacturers.
Advantages of UnyLinker 12 over Compound 4. Many
of the hurdles in the design of molecule 4 were eliminated. First,
there is a UV chromophore handle to monitor the progress of
the reaction easily using TLC or HPLC. Second, the final
product is a triethylammonium salt of succinic acid of the
molecule. Stability of this compound should be similar to that
404
•
Vol. 12, No. 3, 2008 / Organic Process Research & Development

Table 1. Comparison of phosphorothioate oligonucleotide on 0.2 µmol scale using MOE meC (controls #1, #2; repetition of
same experiments) and UnyLinker (Syn #1, #2, #3; repetition of same experiments) loaded CPG solid supportsa
control #1
(%)
control #2
(%)
UnyLinker Syn #1
(%)
UnyLinker Syn #2
(%)
UnyLinker Syn #3
(%)
mass
full length
1783.5
1779.5
1709.1
1703.4
1685.1
82.26
10.26
2.86
1.05
3.57
100
83.69
8.71
2.64
1.1
3.85
100
89.61
5.05
0.89
1.21
3.24
100
88.10
5.48
1.23
0.84
4.36
100
88.96
5.52
1.13
0.81
3.57
100
PO
3′-TPT
n-T
n-MOE meC/MOE meU
total
a TPT ) 3′-terminal phosphorothioate monoester; PO ) phosphate diester.
Table 2. Comparison of phosphorothioate oligonucleotide on
1 mmol scale using MOE meC (0963-146)- and UnyLinker
(0963-148)-loaded solid supportsa
UnyLinker loaded solid supports. Comparison of a phospho-
rothioate oligonucleotide synthesized on OligoProcess using
UnyLinker- and MOE meC-loaded solid supports is shown in
Table 5.
Oligo 4
0963-146
0963-148
n
94.87
2.71
0.05
0.05
0.45
0.03
0.05
0.15
0.29
0.22
0.24
0.14
0.33
0.44
1.54
94.25
4.21
95.59
1.7
Synthesis of Chemically Modified Oligonucleotides. Apart
from 2′-MOE, several other nucleoside and base-modified
oligonucleotides were synthesized efficiently. These are shown
in Table 7. Both fully thiolated and PS/PO backbone-containing
oligonucleotides were also synthesized. Synthesis of 5′-
phosphate monoester, biotin-conjugated oligonucleotides, LNA,
2′-R-fluoro, etc. demonstrates the versatility of this approach.
Improvement in Oligonucleotide Quality. Benzoyl group
is routinely used to protect the exocyclic amino group of
deoxyadenosine, 5-methyldeoxycytidine, 2′-O-methoxyethy-
ladenosine, and 2′-O-methoxyethyl-5-methylcytidine. Even
though it is accepted as a good protecting group, its stability
during loading to solid support or upon storage of loaded support
at room temperature for extended time is not ideal. A portion
of this group falls off leading to unprotected amino group sites
capable of reacting with amidites. A class of compounds termed
(2n - 1)-mers are formed due to branching from these sites
and have been characterized recently.9d
Another process-related impurity termed 3′-terminal phos-
phorothioate monoester (3′-TPT) is formed due to depurination
of dA attached to solid support upon exposure to deblocking
conditions. It has been reported that 18a is more prone to
depurination when compared to nucleosides having phosphate
or phosphorothioate group attached at the 3′-end similar to
compounds such as 18b.22 Use of 18a for synthesis of Oligo 7
leads to formation of 3′-terminal phosphorothioate monoester
(3′-TPT) (Figure 8).
PO 1789
n-dG
n-MOEG
n-MOE C/U
n-dA
0.05
0.26
0.74
0.06
0.05
0.2
n-MOEA
n-T/meC
-A
0.3
0.2
-A + H2O
-A + MeOH
3′-TPT
0.16
0.04
0.31
0.33
1.64
95.82
2.54
CNET
n + 156
shortmer
full length
longmer
a TPT
CNET ) cyanoethyl adduct on N1 of thymine/uracil.
)
3′-terminal phosphorothioate monoester; PO
)
phosphate diester;
Table 3. Yield comparison of UnyLinker and MOE meC
nucleoside loaded on OligoPrep
crude yield crude yield purified yield
OD/µmol
support
mg/µmol
mg/µmol
PS200 MOE meC
113
116
105
109
126
123
7.3
8.0
8.8
7.5
8.4
8.7
3.8
4.2
3.6
3.8
4.6
3.4
OligoPrep MOE meC
OligoPrep UnyLinker
PS200 MOE meC
OligoPrep MOE meC
OligoPrep UnyLinker
On the other hand, high-quality oligonucleotide (Oligo 7)
was obtained by use of 15. Very low levels (<0.1%) are still
observed probably due to nonquantitative capping of solid
support. Similar use of MOE meC- and MOE A-loaded solid
supports leads to formation of (2n - 1)-mers. When UnyLinker-
loaded support was used, these impurities were either eliminated
or reduced substantially, leading to increased quality of oligo-
nucleotide drugs (Figure 9). We have consistently seen between
1 and 3% increase in overall improvement in quality by using
UnyLinker loaded solid supports (Table 8).
Evaluation of DMT vs Pixyl Group. Solid supports 15b
and 16 were compared for efficiency of removal of DMT and
pixyl groups on secondary hydroxyl group of the UnyLinker
molecule and its impact on (n - 1)-mer level in oligonucleotide
to remove DMT group. The following tables (Tables 3 and
4) show the performance of UnyLinker on OligoPrep.
Large-Scale Synthesis of Phosphorothioate Oligonucle-
otides. Next, we scaled up the evaluation of UnyLinker to
several hundred-fold (400–700 mmol scales) on OligoProcess
synthesizer. Multiple oligonucleotides (Table 6; oligo sequences
1-8) were performed using PS200 and NittoPhase solid
supports loaded with UnyLinker at approximately 200 µmol/
g. Optimized conditions similar to those of laboratory-scale Akta
100 syntheses were followed. Subsequently, the oligonucleotides
were purified using C-18 reversed phase HPLC; appropriate
fractions were collected, detritylated, precipitated, and lyoph-
ilized. An average of 2380 g of purified drug was obtained per
700 mmol synthesis. Extensive analysis was then performed
on the purified active pharmaceutical ingredient (API). Con-
sistently higher-quality oligonucleotides were obtained using
(22) (a) Septak, M. Nucleic Acids Res. 1996, 24, 3053. (b) Paul, C. H.;
Royappa, A. T. Nucleic Acids Res. 1996, 24, 3048.
Vol. 12, No. 3, 2008 / Organic Process Research & Development
•
405

Table 4. Quality comparison of UnyLinker and MOE meC nucleoside loaded on OligoPrep
UV purity (%)
mass purity (%)
support
shortmer
full length
longer
full length + PO
n - 1
depurination + TPT
PS200 MOE meC
4.0
7.4
6.6
4.9
6.7
6.7
91
89
90
90
90
90
4.9
3.8
3.7
4.8
3.7
3.8
96
96
96
97
97
96
1.3
1.1
2.1
1.0
1.1
2.0
1.7
1.9
1.6
1.7
1.9
1.6
OligoPrep MOE meC
OligoPrep UnyLinker
PS200 MOE meC
OligoPrep MOE meC
OligoPrep UnyLinker
Table 5. Comparison of phosphorothioate oligonucleotides synthesized on OligoProcess using UnyLinker- and MOE
meC-loaded solid supportsa
oligo 4
SYN000031 (UnyLinker) (%) SYN000032 (UnyLinker) (%) SYN000014 (MOE meC succinate) (%)
n
PO
93.5
2.2
1.9
1.5
94.2
1.7
1.6
1.2
94.9
1.8
0.8
-
DMT-on total n - 1
DMT-on n-(MOE meC +
MOE meU) (#8)
DMT-on total depurination
DMT-on TPT
DMT-on CNET
DMT-on EPD
DMT-on 2′-OMe
other DMT-on
combined area % of two
DMT-on peaks
0.16
0.11
0.3
0.1
0.33
0.10
0.03
0.87
0.48
0.46
0.06
0.09
0.68
0.76
0.2
0.2
0.7
0.9
76.6
78.8
70.2
a TPT ) 3′-terminal phosphorothioate monoester; PO ) phosphate diester; CNET ) cyanoethyl adduct on N1 of thymine/uracil.
Table 6. Phosphorothioate oligonucleotides synthesized on
oligonucleotide (Scheme 4). A cyclic phosphorothioate diester
is formed having a characteristic downfield shift in 31P NMR
(78.2, 74.5 ppm) (Figure 10) as a pair of isomers.
GE Amersham Akta 100 and OligoProcess synthesizers
oligo synthesized
sequence 5′–3′
Another interesting reaction happens at the other end of
UnyLinker molecule during release of the oligonucleotide.
Nucleophilic attack of ammonia on cyclic imide leads to
diamide 20 which then undergoes facile intramolecular cycliza-
tion to form 19 (Scheme 5). The identity of this byproduct was
confirmed by comparing it with an authentic sample prepared
synthetically according to Scheme 6.
oligo 1
2′-O-methoxyethyl modified phosphorothioate
20-nucleotide gapmer, ISIS 345794
[CTG]-AGTCTGTTT-[TCCATTCT]
[GCTCC]-TTCCACTGAT-[CCTGC]
[GCCTC]-AGTCTGCTTC-[GCACC]
[TGGAA]-AGGCTTATAC-[CCCTC]
[TCCCGC]-CTGTGACA-[TGCATT]
d[GCCCAAGCTGGCATCCGTCA]
[CAGC]-AGCAGAGTCTTCA-[TCAT]
oligo 2
oligo 3
oligo 4
oligo 5
oligo 6
oligo 7
oligo 8
Synthesis of Cyclic Byproduct 19. Nucleoside 2′,3′-O,O-
phosphorothioates are usually prepared via the reactions of 5′-
O,N-protected ribonucleosides with thiophosphoryl chloride,
cyclization of nucleoside 2′(3′)-phosphorothioate derivatives,
or using salicyclic chlorophosphite as a phosphitylating agent.23
However, recently, there was a report where 5′-protected
ribonucleosides were reacted with diphenyl H-phosphonate in
pyridine to afford the corresponding 2′,3′-cyclic H-phosphonate
which upon sulfurization gave the nucleoside 2′,3′-O,O-cyclo-
phosphorothioates.24
Following this protocol, treatment of diol 10 with com-
mercially available diphenyl H-phosphonate in pyridine fol-
lowed by sulfurization gave the cyclic phosphorothioate diester,
22. Incubation in aqueous ammonium hydroxide at 55 °C for
12 h gave 19, whose 31P NMR (DMSO-d6) showed two peaks
corresponding to two isomers.
synthesized. Phosphorothioate oligonucleotide 7 was synthe-
sized on Akta 100 synthesizer using 10% dichloroacetic acid
in toluene. Twice the normal volume and time was used for
removal of these acid-labile groups on the UnyLinker molecule
as compared to removal of the DMT group from the 5′-hydroxyl
of oligonucleotide. Analysis of the crude oligonucleotides by
IP-LC-MS showed no difference in level of total (n - 1)-mer
and n-dA (first base in sequence), indicating that no additional
benefit was observed in using a more labile pixyl group as
compared to the standard DMT group.
Mechanism of Deprotection. A reasonable pathway for
cleavage and release of oligonucleotide from UnyLinker is
proposed here. Treatment of support-bound, fully protected
phosphorothioate triester with triethylamine in MeCN for 2 h
removes all acrylonitrile formed due to ꢀ-elimination of
cyanoethyl group, leading to a phosphorothioate diester back-
bone. Subsequent incubation with aqueous ammonium hydrox-
ide cleaves the base-labile succinate ester linkage. The liberated
vicinal hydroxyl group attacks the neighboring phosphorous
center of the diester linkage, releasing the free 3′- hydroxyl
(23) (a) Gerlt, J. A.; Wan, W. H. Y. Biochemistry 1979, 18, 4630. (b)
Eckstein, F.; Ginll, H. Chem. Ber. 1968, 101, 1670. (c) Holy, A.; Kois,
P. Collect. Czech. Chem. Commun. 1980, 45, 2817. (d) Ludwig, J.;
Eckstein, F. J. Org. Chem. 1989, 54, 631.
(24) Jankowska, J.; Wenska, M.; Popenda, M.; Stawinski, J.; Kraszewski,
A. Tetrahedron Lett. 2000, 41, 2227.
406
•
Vol. 12, No. 3, 2008 / Organic Process Research & Development

Figure 9. Upper panel (A): HPLC-UV trace of the synthesized phosphorothioate oligonucleotide 4 using MOE meC nucleoside-
loaded solid support. Lower panel (B): HPLC-UV trace of the synthesized phosphorothioate oligonucleotide using UnyLinker-
loaded solid support.
Synthesis of Trans Configured Molecule 26. Even though
by design only the cis-diol-based UnyLinker molecule will
liberate free 3′-hydroxy oligonucleotide, we wanted to find out,
for therapeutic application purposes, that if 24 were to be present
as a contaminant, would it lead to any other species or even
give the right product or, as expected, have a fragment attached
to the 3′-end of the oligonucleotide. The synthesis of trans-
configured UnyLinker molecule was attempted according to
Scheme 7. Treatment of olefin 9 with m-chloroperbenzoic acid
gave the epoxide as a colorless solid in 85% yield. Ring opening
of the epoxide with perchloric acid or many other acids did
not afford the trans-diol 24.25 Epimerization with Raney nickel
also failed to give the desired product. No further attempts were
made to synthesize 24. This probably indicates that in this
bridge-head ring system, trans-diol is very difficult to be formed.
Therapeutic Application of UnyLinker Chemistry. It is
important to demonstrate the safe nature of any chemistry before
being accepted for therapeutic applications. Initially, extensive
testing (ICP-MS) of DMT-protected hydroxyl compound 11
as well as its succinate molecule 12 showed very low levels of
osmium (<20 ppm). Subsequently, treatment of 11 with
charcoal and extraction with solvent showed no detectable levels
of osmium (<1 ppm). During the development of this process
to eliminate residual osmium, an alternative approach was also
successfully carried out. Amino-derivatized polymeric solid
support (Wang resin) was capped with pentenoic anhydride
under standard capping conditions. A solution of compound
11 containing a higher level of osmium tetroxide (320 ppm;
intentionally spiked) was stirred with a 2-fold theoretical excess
of this capped support overnight at room temperature. Filtration
and concentration of solution afforded compound 11 containing
no detectable levels (<1 ppm) of osmium (Scheme 8). This
principle is similar to the dihydroxylation of an olefin except
that no oxidizing agent is present and the cyclic osmium
complex stays bound to the solid support and is removed after
(25) (a) Fringuelli, F.; Germani, R.; Pizzo, F.; Savelli, G. Synth. Commun.
1989, 19, 1939. (b) Taylor, E. C.; Janini, T. E.; Elmer, O. C. Org.
Process. Res. DeV 1998, 2, 147 and references cited therein. (c)
Fringuelli, F.; Germani, R.; Pizzo, F.; Savelli, G. Synth. Commun.
1989, 19, 1939. (d) Swern, D.; Billen, G. N.; Scanland, J. T. J. Am.
Chem. Soc. 1946, 68, 1504. (e) Owen, L. N.; Smith, P. N. J. Chem.
Soc. 1952, 4026.
Vol. 12, No. 3, 2008 / Organic Process Research & Development
•
407

Table 7. Oligonucleotides synthesized on GE Amersham
Table 8. IP-LC-MS analysis of Oligo 4 synthesized using
MOE meC-loaded and UnyLinker-loaded solid supports at 1
mm scale on Akta 100 synthesizera
Akta 10/100 and other automated DNA/RNA synthesizers
oligo
synthesized
sequence 5′–3′a
using MOE
meC-loaded
support (%)
using support
Oligo 9
5′-PO-d[AATGCATGTCACAGGCGGGA]
sequence: oligo 4
15e (%)
m
m
m
Oligo 10
mCmsTms CmoAmoGmo CmoAmo CmoA-
m
m
moTmo CmoTmoAmo CmoAmoAmoGmsAmsAmsA
n
94.87
2.71
0.05
0.05
0.45
0.03
0.05
0.15
0.29
0.22
0.24
0.14
0.33
0.44
1.54
94.25
4.21
95.59
1.7
m
m
m
m
Oligo 11
Oligo 12
Oligo 13
Oligo 14
Oligo 15
Oligo 16
Oligo 17
Oligo 18
Oligo 19
Oligo 20
Oligo 21
mCmsTms CmsAmsGms CmsAms CmsAmsTms CmsTms
P ) O
n-dG
n-MOE G
n-MOE meC/MOE meU
n-dA
n-MOE A
n-T/5-meC
-A
-A + H2O
-A + MeOH
3′-TPT
CNET
N + 156
early eluting shortmer
full length
late eluting longmer
m
Ams CmsAmsAmsGmsAmsAmsA
AroAeoGroCeoAroAeoCroGeoAro
0.05
0.26
0.74
0.06
0.05
0.2
-
GeoAroAeoGroCeoGroAeoUroAeoAr
m
m
5′-Aes CesAes CesAesAesAesTes
-
m
m
m
Tes CesGesGesTesTes CesTesAes CesAesGesGesGe
PO3-UroUroAroUroCroGmoCmoUroUroCmoUmoCroGro-
UroUro-GroCmoUmoUm
0.3
0.2
m
m
m
5′-mCesGes CesGesTesAes Ces Ces-
0.16
0.04
0.31
0.33
1.64
95.82
2.54
AesAesAesAesGesTesAesAesTesAesAesTesGe
m
mCesUfs CesAfsGesCfsAesCfsAes-
m
Ufs CesUfsAesCfsAesAfsGesAesAe
m
m
5′-Tes CesAes CesAesAesGesTesTes-
m
m
AesGesGesGesTes CesTes CesAesGesGesGesAe
m
m
5′-Ges CesTes CesAesTesTesTesAes-
m
m
m
GesTes CesTesGes Ces CesTesGesAesTe
mClsTlsGdsCdsTdsAdsGdsCdsCds-
TdsCdsTdsGdsGdsAdsTdsTdsTdsGlsAl
mCloTloGdsCdsTdsAdsGdsCdsCds-
TdsCdsTdsGdsGdsAdsTdsTdsTdoGloAl
a TPT
CNET ) cyanoethyl adduct on N1 of thymine/uracil.
)
3′-terminal phosphorothioate monoester; PO
)
phosphate diester;
m
m
m
m
solid supports. Phosphorothioate oligonucleotides manufactured
for therapeutic applications undergo rigorous analytical tests and
show no trace of osmium. Multitudes of modified oligonucle-
otides have been evaluated for pharmacological, toxicological,
pharmacokinetic, and other biological evaluations in various
animal models. Many phosphorothioate 2′-O-methoxyethyl-
modified oligonucleotide drugs manufactured using UnyLinker-
loaded solid supports are being evaluated in the human clinic.
These evidences clearly indicate the safe nature of using this
chemistry.
Economical Impact of Using UnyLinker Chemistry.
There are several direct and indirect savings to using this
UnyLinker-loaded solid supports. Even though it may be
difficult to calculate the indirect cost, it is obvious that any labor
involved in QC testing of several different nucleoside loaded
supports can be eliminated. Warehousing, stability testing, and
5′-Aes CesAesTesAes CesTes Ces Ces-
m
m
TesTesTes CesTes CesAesGes-
AesGesTes Ces CesAe
TlsTlsGlsTdsTdsCdsTdsGdsAdsAdsTdsGdsTds Cls ClsAl
5′-AesAes Ces CesTesAesTes Ces Ces-
m
m
m
m
Oligo 22
Oligo 23
m
m
m
m
m
TesGesAesAesTesTesAes CesTesTesGesAesAe
m
m
m
m
Oligo 24
5′-Ams CmsAmsAmsAms CmsAms Cms Cms-
m
m
AmsTmsTmsGmsTms CmsAms Cms-
m
m
m
Ams CmsTms Cms CmsAm
Oligo 25
Oligo 26
5′-PO-d[GAGGGGTATAAGCCTTTCCA]
m
m
m
Gls ClsTls ClsAdsTdsAds Cds-
m
m
m
Tds CdsGdsTdsAdsGdsGls Cls ClsAl
m
m
Oligo 27
Oligo 28
Oligo 29
Oligo 30
Oligo 31
Oligo 32
Oligo 33
Oligo 34
Oligo 35
Oligo 36
Oligo 37
5′-Ges CesAds CdsTdsTdsTdsGdsTds-
GdsGdsTdsGds Cds CdsAdsAdsGdsGesmCe
m
m
m
m
m
5′-mCes CdsAdsGds CdsAdsTds Cds-
TdsGds CdsTdsGes CesTesTesmCe
m
m
m
m
m
5′-mCesTds CesAdsGes CdsAes Cds-
m
m
AesTds CesTdsAes CdsAesAdsGesAdsAesAe
m
m
5′-biotin-GesAesAesGesTesAdsGds Cds Cds-
m
m
m
Ads Cds CdsAdsAds CdsTesGesTesGesmCe
5′-mCesTfsCfsAfsGfsCfsAfsCfsAfs-
TfsCfsTfsAfsCfsAfsAfsGfsAfsAesAe
5′-PO3-AroUroAroGroAroCmoUmoUro-
CroAmoUmoCroCroUroUroGroUmoUmoGm
5′-UroAroCroGroCroAmoAmoAroCroCmo
moUroGroAroUroGroUmoCmoCm
5′-PO3-AroAroAroUroGroUmoUmoCroCro-
moGmoCroCroCroAroGroGmoGmoCm
-
U
A
5′-GroGeoAroCeoAroUeoCroAeoAro-
GeoGroUeoUroUeoGroCeoGroUeoAr
5′-UroCroAroCroUroCroGroGro-
CroUroGroGroAroUroGroGroAroGroUrsTesT
5′-[CTCTA]-GTTCCTCTCA-[ATGTC](for
plate synthesis comparison)
a mC ) 5-methyl C; d ) 2′-deoxy; e ) 2′-O-methoxyethyl; m ) 2′-O-methyl; r
) 2′-OH; l ) LNA; f ) 2′-R-fluoro; o ) PO; s ) PS; PO3 ) phosphate
monoester.
filtration. This concept may have additional industrial applica-
tions such as in water purification, etc.
Figure 10. 31P NMR (CDCl3) of 19 from oligonucleotide
synthesis.
Several hundreds of oligonucleotides have been synthesized
in our laboratories so far using UnyLinker loaded onto various
408
•
Vol. 12, No. 3, 2008 / Organic Process Research & Development

Table 9. Synthesis parameters of cycle used on Akta 100 DNA/RNA synthesizera
step
reagent
volume (mL)
time (min)
detritylation
coupling
10% dichloroacetic acid/toluene
72 (144)
8.8, 8.8
3 (6)
3
phosphoramidite (0.2 M) and 1H-tetrazole (0.45 M) or DCI (0.7
M) + NMI (0.1 M) in MeCN
sulfurization
capping
PADS (0.2M) in 3-picoline/MeCN (1:1, v/v) or pyridine/MeCN
(1:1, v/v)
44
3
Ac2O/pyridine/MeCN,NMI/MeCN
30, 30
2.5
a Volume and time in parentheses indicate the conditions used for detritylation of UnyLinker molecule.
Scheme 5. Mechanism depicting formation of cyclic
by-product
fixed-bed synthesis column. Appropriate amounts of solid
supports were packed in the column for synthesis. Dichloro-
acetic acid (10%) in toluene was used for deblocking of the
dimethoxytrityl (DMT) groups from the 5′-hydroxyl group of
the nucleotide. Extended detritylation conditions (twice the
column volume and contact time as the normal cycle) were
used to remove the DMT group from the secondary hydroxyl
group of the UnyLinker molecule.
4,5-Dicyanoimidazole (0.7 M) in the presence of N-meth-
ylimidazole (0.1 M)27,28 or 1H-tetrazole (0.45 M) in MeCN was
used as activator during the coupling step. During the coupling
step 1.75 equiv of amidites (both deoxy and 2′-O-methoxyeth-
ylribonucleosides) and a ratio of 1:1 (v/v) of amidite to activator
solution were used. Amidite and activator solutions were
prepared using low-water MeCN (water content <30 ppm) and
were dried further by addition of activated 4Å molecular sieves
(∼50 g/L). Phosphorothioate linkages were introduced by
sulfurization with a 0.2 M solution of phenylacetyl disulfide in
MeCN in 3-picoline or pyridine (1:1 v/v) for a contact time of
2–3 min.6 Phosphate diester linkages were incorporated via
oxidation of phosphite triesters using a solution of iodine/
tetrahydrofuran/pyridine/water for 2 min. Capping reagents were
made to the manufacturer’s receipe: Cap A: N-methylimidazole/
MeCN (1:4 v/v), Cap B: acetic anhydride/pyridine/CH3CN (2:
3:5, v/v/v). Details of synthesis cycle are given in Table 9.
At the end of synthesis, the support-bound DMT-on oligo-
nucleotide was treated with a solution of triethylamine and
MeCN (1:1, v/v) for 2 h to remove acrylonitrile formed by
deprotection of cyanoethyl group from phosphorothioate
triester.9a Subsequently, the solid support containing oligonucle-
otide was incubated with concentrated aqueous ammonium
hydroxide at 55 °C for 13 h to complete cleavage from support,
elimination of UnyLinker molecule to liberate 3′-hydroxy group
of oligonucleotide and deprotection of base-protecting groups.
Removal of 2′-O-silyl group was effected using a solution of
N-methylpyrrolidinone/triethylamine/TEA·3HF (1.5:0.75:1.0,
v/v/v) at 65 °C for 3 h. The purification of RNA oligonucle-
otides was carried out by anion-exchange chromatography and
desalted to afford the oligonucleotide.
various documentations related to it that are involved in routine
GMP manufacturing of drugs can be avoided. Labor involved
in these indirect operations could be a significant cost-saving
factor. On the direct cost savings, as in any other field, cost
and quantity are directly related. Synthesis of this molecule has
been outsourced and scaled up to several kilogram quantities,
and loading has been done on 10-20-kg batch sizes. At this
scale it costs us few cents per micromole to synthesize and load
onto polymeric supports.
Conclusions
In summary, we have designed a linker molecule that has
all the features of being termed a universal linker. Various
oligonucleotide modifications have been synthesized to dem-
onstrate its utility. Large-scale syntheses of phosphorothioate
oligonucleotides for therapeutic applications have been carried
out, leading to improved quality in addition to eliminating one
crucial starting material in the supply chain. A mechanism for
cleavage and release of oligonucleotides is proposed and
supported by characterization of the byproduct. Here at Isis
Pharmaceuticals, Inc. we have switched over to this UnyLinker-
loaded solid support, both for research (plates and small-scale
syntheses) and large-scale (OligoProcess) routine manufacture
of antisense drugs.26
Experimental Section
Oligonucleotide Synthesis. Oligonucleotide syntheses were
performed either on an Akta 10/100 or OligoProcess or
MerMade 396-plate synthesizer by the phosphoramidite-
coupling method. Syntheses on Akta 10 and 100 (closely
resembling the production-scale synthesizer) and OligoProcess
DNA/RNA synthesizers were performed using a stainless steel
HPLC Analysis and Purification of Oligonucleotides.
Analysis and purification of oligonucleotides by reversed phase
high performance liquid chromatography (RP-HPLC) was
(27) (a) Wang, Z.; Siwkowski, A.; Lima, W. F.; Olsen, P.; Ravikumar,
V. T. Bioorg. Med. Chem. 2006, 14, 5049. (b) Vargeese, C.; Carter,
J.; Yegge, J.; Krivjansky, S.; Settle, A.; Kropp, E.; Peterson, K.; Pieken,
W. Nucleic Acids Res. 1998, 26, 1046.
(26) Other pharmaceutical companies such as Eli Lilly, Oncogenex, iCo
Therapeutics, Johnson & Johnson, Australian Therapeutics Limited
(ATL) have started using oligonucleotide drugs manufactured using
UnyLinker chemistry.
(28) (a) Hayakawa, Y.; Iwase, T.; Nurminen, E. J.; Tsukamoto, M.; Kataoka,
M. Tetrahedron 2005, 61, 2203. (b) Eleuteri, A.; Capaldi, D. C.; Krotz,
A. H.; Cole, D. L.; Ravikumar, V. T. Org. Process Res. DeV 2000, 4,
182 and references cited therein for many other activators.
Vol. 12, No. 3, 2008 / Organic Process Research & Development
•
409

Scheme 6. Synthesis of cyclic phosphorothioate diester 19
Scheme 7. Attempted synthesis of trans-UnyLinker-loaded
support 27
1.0 mL/min, tR(DMT-off) 10–11 min, tR(DMT-on) 14–16 min.
The DMT-on fraction was collected, evaporated in vacuum,
and redissolved in water, and the DMT group was removed as
described below.
Dedimethoxytritylation. An aliquot (30 µL) was transferred
into an Eppendorff tube (1.5 mL), and acetic acid (50%, 30
µL) was added. After 30 min at room temperature sodium
acetate (2.5 M, 20 µL) was added, followed by cold ethanol
(1.2 mL). The mixture was vortexed and cooled in dry ice for
20 min. The precipitate was spun down with a centrifuge, the
supernatant was discarded, and the precipitate was rinsed with
ethanol and dried under vacuum.
Acknowledgment
We sincerely thank Herbert Brown Pharmaceutical Research
& Laboratories (India), Nitto Denko Corporation (Japan),
Innovasynth Technologies (India), National Institutues of
Pharmaceutical Research & Development (China), and Dr.
Douglas L. Cole, Jeff Vicenzi, Herb Boswell, Rob Andrews
for their valuable discussions and help.
Scheme 8. Complete removal of residual osmium from
UnyLinker 11
Supporting Information Available
General methods and experimental details related to Uny-
Linker molecule, loading conditions to various solid supports,
1
oligonucleotide synthesis; copies of H NMR, 13C NMR of
various compounds; IP-UV-HPLC-MS, tables containing
characterization of synthesized oligonucleotides; tables contain-
ing crystallographic data. This material is available free of
charge via the Internet at http://pubs.acs.org.
performed on a Novapak C18 column (3.9 mm × 300 mm)
using a HPLC system (600E System Controller, 996 Photodiode
Array Detector, 717 Autosampler). For analysis an MeCN (A)/
0.1 M triethylammonium acetate gradient was used: 5% to 35%
A from 0 to 10 min, then 35% to 40% A from 10 to 20 min,
then 40% to 95% A from 20 to 25 min, flow rate ) 1.0 mL/
min/50% A from 8 to 9 min, 9 to 26 min at 50% flow rate )
Received for review January 29, 2008.
OP8000178
410
•
Vol. 12, No. 3, 2008 / Organic Process Research & Development