UnyLinker dimer impurity characterization and process improvement

Article abstract of DOI:10.1016/j.tetlet.2017.01.101

Herein, we describe the isolation and characterization of a UnyLinker dimer which, if left uncontrolled, can become incorporated in oligonucleotide products. The dimer is formed as a result of an unusual intermolecular osmium-catalyzed etherification. We

Full text of DOI:10.1016/j.tetlet.2017.01.101

Accepted Manuscript
UnyLinker dimer impurity characterization and process improvement
Joshua L. Brooks, Phil Olsen, Lijian Chen, Andrew A. Rodriguez
PII:
S0040-4039(17)30146-6
DOI:
http://dx.doi.org/10.1016/j.tetlet.2017.01.101
TETL 48604
Reference:
Tetrahedron Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
10 December 2016
24 January 2017
30 January 2017
Please cite this article as: Brooks, J.L., Olsen, P., Chen, L., Rodriguez, A.A., UnyLinker dimer impurity
characterization and process improvement, Tetrahedron Letters (2017), doi: http://dx.doi.org/10.1016/j.tetlet.
2017.01.101
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Graphical Abstract
UnyLinker dimer impurity characterization
and process improvement
Leave this area blank for abstract info.
Joshua L. Brooks, Phil Olsen, Lijian Chen, and Andrew A. Rodriguez
UnyLinker dimer impurity characterization and process improvement
Joshua L. Brooksa,, Phil Olsena, Lijian Chena, and Andrew A. Rodrigueza
a Ionis Pharmaceuticals, 2855 Gazelle Court, Carlsbad, CA 92010
Highlights:





Isolation and characterization of UnyLinker dimer
The dimer was discovered as an impurity in oligonucleotide synthesis
Description mechanistic proposal for osmium catalyzed intermolecular dimerization
Process improvement that completely prevents dimer formation
New procedure scaled to 35 kg
———
 Corresponding author. Tel.: +1-760-603-2369; fax: +1-760-603-3862; e-mail: jbrooks@ionisph.com

2
ARTICLE INFO
ABSTRACT
Article history:
Received
Received in revised form
Accepted
Herein, we describe the isolation and characterization of a UnyLinker dimer which, if left
uncontrolled, can become incorporated in oligonucleotide products. The dimer is formed as a
result of an unusual intermolecular osmium-catalyzed etherification. We demonstrate that by
simply replacing H2O2 as the co-oxidant with NMO, none of the UnyLinker dimer impurity is
formed.
Available online
2009 Elsevier Ltd. All rights reserved.
Keywords:
Oligonucleotide
Dihydroxylation
UnyLinker
Dimerization

3
Synthetic oligonucleotides are a novel class of drug1 with the
potential to treat a wide range of illnesses including cancer,
cardiovascular and metabolic conditions, neurological disorders,
and ophthalmic diseases. Oligonucleotides designed to interact
with a variety of molecular targets including coding and non-
coding RNA and proteins are being investigated with 140 clinical
trials currently listed on ClinicalTrials.gov. Five oligonucleotide
drugs have reached the market.
During a recent analysis of an oligonucleotide, new
impurities were observed corresponding to mass increases of 458,
476, and 494 Daltons and totaling 0.64% of the product. The
masses were consistent with an oligonucleotide attached at the 3′-
end to a dimeric form of the UnyLinker, 7 (Figure 1). Unlike
UnyLinker, the dimer likely fails to cleave from the
oligonucleotide because the attacking hydroxyl group is blocked
as the ether linkage (Scheme 1).
As part of an ongoing investigation into oligonucleotide
impurities generated during the manufacturing process 2 we report
here the isolation, characterization, and elimination of
a
previously unknown impurity of the universal linker molecule
(UnyLinker®) that is commonly used in solid phase
oligonucleotide synthesis. The UnyLinker impurity leads to
multiple oligonucleotide impurities that are difficult to remove.
Therefore, its elimination results in higher purity oligonucleotide
products.
Our oligonucleotide manufacturing process is performed using
an optimized version of the phosphoramidite approach.2 We
utilize a solid-support with UnyLinker 1 attached to the polymer-
matrix (Scheme 1).3 The use of UnyLinker eliminates the need to
load the solid-support with the first nucleotide, thereby enabling
the synthesis of any oligonucleotide with a single solid-support,
regardless of the identity of its 3′-nucleotide. Upon completion of
synthesis, solid-support-bound oligonucleotide 2 is subjected to
ammonolysis to cleave UnyLinker from the resin, forming
intermediate 3. The newly unmasked alcohol attacks the 3′-
thiophosphate to release oligonucleotide 4 and UnyLinker cyclic
thiophosphate 5. Byproducts from synthesis, including
hydrolyzed Unylinker fragments, are removed during the
subsequent chromatographic purification step.
Figure 1: Proposed structures of UnyLinker dimer impurities showing
relative stereochemistry.
Examination of the batch of UnyLinker used to synthesize the
oligonucleotide revealed that it indeed contained a dimeric
species. It was determined that the dimer formed during the
osmium tetroxide dihydroxylation of olefin 8, resulting in desired
diol 9 and the dimeric ether 10 in 84:16 ratio and 94% yield
(Scheme 2).4 The dimer 10 was isolated from a crude reaction
mixture, fully characterized by NMR,5 and the structure
confirmed by X-ray crystal analysis of the 4-nitro-benzyl ester
analog (Figure 2).
Scheme 2: Synthesis of diol 9 and dimer 10
Figure 2: X-ray crystal structure of nitrobenzyl adduct
The mechanistic origin of this unusual dimer requires
additional explanation. One possibility is that olefin 8 undergoes
epoxidation to 11 and subsequent ring opening by attack of diol 9
(Scheme 3). This was ruled out however, as neither epoxide 11 or
cis-trans product 12, i.e. the product of attack of 9 on 11, were
observed. Furthermore, when epoxide 11 was prepared and
subjected to reaction conditions (from Scheme 2) with diol 9, no
dimer was formed.
Scheme 1: Solid phase oligonucleotide synthesis with
UnyLinker and UnyLinker dimer.

4
Tetrahedron
Table 1: Improved synthesis of UnyLinker diol 9a
Entry
OsO4
(mol%)
Reaction
time (h)
Co-oxidant
Product
Ratio (9:10)
% yield
Scheme 3: Proposed epoxide opening resulting in dimer
formation
1
1
8
H2O2
84:16
88:12
100:0
100:0
100:0
100:0
100:0
94
45
90
90
90
89
90d
2
1
8
tBuO2H
NMO
NMO
NMO
NMO
NMO
3
1
4
A second possibility is that 10 results from intermolecular
osmium-catalyzed dimerization. While several examples of
4b
5b
6b
7b,c
1
4
6
, 7
osmium-catalyzed intramolecular etherifications are known, to
the best of our knowledge, the intermolecular variant has not been
reported. In particular, we propose that osmate ester 11 (Scheme
4) can either be hydrolyzed to form the diol 9, or oxidized to the
Os(VIII) intermediate 12 by H2O2 (Path A). A second equivalent
of olefin 8 can add to the osmate ester 12 resulting in 13, which is
hydrolyzed to dimer 10 and OsO3. A second possibility (Path B)
is that Os(VI) intermediate 11 reacts directly with olefin 8
resulting in the Os(IV) oxidation state of 13,8 which is
subsequently hydrolyzed and reoxidized to give dimer 10 and
OsO2.
0.1
0.01
0.1
4
24
4
a. Reaction conditions: 10.0 g (41.5 mmol) olefin suspended in
100 mL acetone, and co-oxidant added (4.65 equiv) followed
by OsO4 as a 20 mM solution in tBuOH. The solution was
heated to reflux, behind a blast shield, until all of the starting
material was consumed (TLC), typically 8h. b. 2.5 equiv. NMO
added. c. The reaction was preformed starting from 35 kg
olefin. d. Isolated yield.
In conclusion, we have identified and characterized a new
oligonucleotide impurity resulting from a Unylinker dimer. The
dimer likely resulted from an unusual intermolecular osmium-
catalyzed etherification during the dihydroxylation step of
Unylinker synthesis and was formed as a result of an osmate ester
intermediate reacting with a second equivalent of olefin. We
demonstrated that by changing the co-oxidant used during the
dihydroxylation from H2O2 to NMO, we can prevent formation of
the dimer while reducing catalyst loading and increasing yield.
This new method is currently used to produce dimer-free
UnyLinker at large scale.
Acknowledgments
Scheme 4: Catalytic mechanism for formation of dimer 10
We sincerely thank Herbert Brown Pharmaceutical Research
& Laboratories (India), Daniel Capaldi and Andy McPherson
(Ionis) for helpful discussions, and Arnold Rheingold (UCSD) for
X-ray crystallography services.
To address the problem of dimer formation and to improve the
dihydroxylation, a brief screening of reaction conditions was
conducted (Table 1). As noted, the use of H2O2 as a co-oxidant
led to dimer formation (entry 1). Similar results were obtained
with tBuO2H (entry 2), albeit with diminished yield. By switching
the co-oxidant to 4-methylmorpholine N-oxide (NMO, entry 3),
no dimer product was observed, and the reaction rate was
increased (from 8 to 4h reaction time, entry 3)4. We were able to
reduce the amount of co-oxidant to 2.5 equivalents and reduce the
catalyst loading to 0.1 mol% without affecting the conversion or
reaction time (entries 4 and 5). However, when the catalyst
loading was reduced to 0.01 mol%, the reaction was too slow to
be practicable (24h, entry 6). Our best reaction conditions9 were
repeated at a 35 kg scale, and resulted in 90% isolated yield of
diol 9 without formation of dimer 10.

5
1
a) Aartsma-Rus, A. Molecular Therapy 2016, 24, 193. b)
Haussecker, D.; Kay, M. A. Science 2015, 347, 1069.
2
a) Rodriguez, A. A.; Cedillo, I.; McPherson, A. K. Bioorg. Med.
Chem. Lett. 2016, 26, 3468. b) Rodriguez, A. A.; Cedillo, I.;
Mowery, B. P.; Gaus, H. J.; Krishnamoorthy, S. S.; McPherson, A.
K. Bioorg. Med. Chem. Lett. 2014, 24, 3243. 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) Capaldi, D. C.;
Gaus, H. J.; Carty, R. L.; Moore, M. N.; Turney, B. J.; Decottignies,
S. D.; McArdle, J. V.; Scozzari, A. N.; Ravikumar, V. T.; Krotz, A.
H. Bioorg. Med. Chem. Lett. 2004, 14, 4683. e) 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.
3
For other examples of support-bound linkers see: a) Wang, Z.;
Olsen, P.; Ravikumar, V. T. Nucleosides, Nucleotides, and Nucleic
Acids 2007, 26, 259. b) Anderson, K. M.; Jaquinod, L.; Jensen, M.
A.; Ngo, N.; Davis, R. W. J. Org. Chem. 2007, 72, 9875. c) Ferreira,
F.; Meyer, A.; Vasseur, J.-J.; Morvan, F. J. Org. Chem. 2005, 70,
9198. d) Kumar, P.; Mahajan, S.; Gupta, K. C. J. Org. Chem. 2004,
69, 6482. e) Azhayev, A. V.; Antopolsky, M. L. Tetrahedron 2001,
57, 4977. f) Nelson, P. S.; Muthini, S.; Vierra, M.; Acosta, L.; Smith,
T. M. BioTechniques 1997, 22, 752.
4 For representative HPLC trace see supporting information.
5 1H NMR (300 MHz) δ 7.55-7.38 (m, 6H), 7.20 (d, J = 6.0 Hz, 4H),
4.92 (broad s, 2H), 4.78 (s, 2H), 4.40 (s, 2H), 4.07 (d, J = 7.0 Hz,
2H), 3.94 (d, J = 7.0 Hz, 2H), 3.15 (q, J = 74.5 Hz, 4H). 13C NMR
(75 MHz) δ 176.0, 132.2, 128.9, 126.7, 84.8, 82.0, 80.6, 73.0, 45.8.
6
For review see: a) Pilgram, B. S.; Donohoe, T. J. J. Org. Chem.
2013, 78, 2149. b) Piccialli, V. Synthesis 2007, 17, 2585
7 a) McDonald, F. E.; Towne, T. B.; Schultz, C. C. Pure Appl. Chem.
1998, 70, 355-358. b) Keinan, E.; Sinha, S. C. Pure Appl. Chem.
2002, 74, 93-105. c) Piccialli, V. Molecules 2014, 19(5), 6534-6582.
d) Morimoto, Y.; Kinoshita, T.; Iwai, T. Chirality 2002, 14, 578-586.
e) McDonald, F. E.; Towne, T. B. J. Am. Chem. Soc. 1994, 116,
7921-7922.
8
Donohoe, T. J.; Wheelhouse, K. M. P.; Lindsay-Scott, P. J.;
Glossop, P. A.; Nash, I. A.; Parker, J. S. Angew. Chem., Int. Ed.
2008, 47, 2872–2875.
9
The alkene 6 (10.00 g, 41.45 mmol) was suspended in 85 mL of
acetone, and a solution of 50% aqueous NMO (2.5 equiv., 21.49 mL,
103.6 mmol) was added resulting in a biphasic slurry of alkene. A
solution of OsO4 in tBuOH (0.02 M, 2.07 mL, 0.001 equiv.) was
added and the reaction mixture brought to reflux. As the solution
begins to reflux the layers coalesce and the alkene fully dissolves (~5
min.); the diol then precipitates from solution as it is formed. Reflux
is continued until the starting material is consumed at which point the
solution is cooled to room temperature, quenched with saturated
aqueous sodium thiosulfate (50 mL), and the diol collected by
filtration. The solid was washed with H2O (50 mL) and acetone (100
mL), collected, and dried under vacuum to afford diol 9 (10.22 g,
90% yield).
1H NMR (300 MHz) δ 7.55-7.38 (m, 3H), 7.25-7.18 (m, 2H), 5.03 (s,
2H), 4.41 (s, 2H), 3.97 (s, 2H), 3.17 (s, 2H). 13C NMR (75 MHz) δ
176.2, 132.2, 128.9, 128.4, 126.8, 84.1, 71.8, 45.5.