Solid-phase synthesis and on-column deprotection of RNA from 2′- (and 3′-) O-Levulinated (Lv) ribonucleoside monomers

Article abstract of DOI:10.1021/ol0629521

(Chemical Equation Presented) The solid-phase synthesis of oligoribonucleotides derived from ribonucleosides esterified at the 2′ (or 3′-) position with the levulinyl (Lv) group is described. The oligomers can be released from the solid support as 2′-O-Lv

Full text of DOI:10.1021/ol0629521

ORGANIC
LETTERS
2007
Vol. 9, No. 5
789-792
Solid-Phase Synthesis and On-Column
Deprotection of RNA from 2 - (and 3 -)
′
′
O-Levulinated (Lv) Ribonucleoside
Monomers
Jeremy G. Lackey, David Sabatino, and Masad J. Damha*
Department of Chemistry, McGill UniVersity, Montreal, QC, Canada
masad.damha@mcgill.ca
Received December 6, 2006
ABSTRACT
The solid-phase synthesis of oligoribonucleotides derived from ribonucleosides esterified at the 2
′
- (or 3
′
-) position with the levulinyl (Lv)
group is described. The oligomers can be released from the solid support as 2′-O-Lv ester derivatives or fully deprotected while still attached
to the solid support.
The advent of the RNA interference (RNAi)1 methodology
and its application to therapeutics has created an urgent and
growing need for the synthesis of large quantities of native
and chemically modified short interfering RNA (siRNA) for
animal and human studies. For many years, RNA synthesis
has been regarded as far more difficult than DNA synthesis
because of the difficulty in finding a compatible 2′-protecting
group that is stable throughout chain assembly and can be
removed selectively at the end of synthesis without phos-
phodiester bond isomerization or degradation.2 In fact,
finding a satisfactory 2′-protecting group is a research
problem that has spanned several decades.3 The application
of the 2′-O-t-butyldimethylsilyl group (TBDMS) from Ogil-
vie and co-workers has proved sufficiently robust to allow
assembly of RNA chains at lengths suitable for RNA
structure-function studies.4
Two other widely used RNA synthesis strategies (2′-O-
ACE5 and 2′-O-TOM6) allow oligoribonucleotides to be
produced somewhat routinely and, together with the 2′-O-
(3) (a) Brown, D. M.; Magrath, D. I.; Nielson, A. H.; Todd, A. R. Nature
1956, 177, 1124-1125. (b) Pathak, T.; Chattopadhyaya, J. Acta Chem.
Scand. 1985, B39, 799-806. (c) Beaucage, S. L.; Iyer, P. Tetrahedron 1992,
48, 2223-2311. (d) Reese, C. B. Tetrahedron 1978, 34, 3143-3179. (e)
Herdewijn, P. Oligonucleotide Synthesis: Methods and Applications;
Humana Press: New Jersey, 2005; Vol. 288.
(4) (a) Ogilvie, K. K.; Usman, N.; Nicoghosian, K.; Cedergren, R. J.
Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 5764-5768. (b) Ogilvie, K. K.;
Sadana, K. L.; Thompson, E. A.; Quilliam, M. A.; Westmore, J. B.
Tetrahedron Lett. 1974, 15, 2861-2867.
(5) Scaringe, S. A.; Wincott, F. E.; Caruthers, M. H. J. Am. Chem. Soc.
1998, 120, 11820-11821.
(1) (a) Fire, A.; Xu, S.; Montgomery, S. K.; Kostas, S. A.; Driver, S. E.;
Mello, C. C. Nature 1998, 391, 806-811. (b) Zimmermann, T. S. et al.
Nature 2006, 441, 111-114.
(2) Reese, C. B. Org. Biomol. Chem. 2005, 3, 3851-3868.
10.1021/ol0629521 CCC: $37.00
© 2007 American Chemical Society
Published on Web 02/06/2007

TBDMS strategy, satisfy the need for high-throughput RNA
synthesis. A variety of other 2′-hydroxyl protecting groups
have appeared in the literature.7-11 The TBDMS, ACE, TOM,
and most of the previously described 2′-protecting groups
all share the same requirements for a manual solution-phase
2′-deprotection. The need for manual 2′-deprotection is time
and labor intensive, particularly for large-scale synthesis, and
a potential source for material losses and ribonuclease
contamination.
Scheme 1. Synthesis of 2′/3′-Lv Monomers
In this report, we have re-evaluated 2′- (and 3′-) O-acyl
ribonucleosides, as possible synthons for RNA synthesis. At
the onset, we recognized that 2′-O-acyl protecting groups
(e.g., benzoyl) in RNA synthesis had been largely unsuc-
cessful due to ease of 2′-3′ migration9 and lack of specificity
and compatibility with other groups. In fact, for these reasons,
2′-O-acyl protecting groups have only very rarely been used
in oligoribonucleotide synthesis.9,10 The levulinyl (Lv) group
has been previously described by van Boom and others as a
“transient” 5′-protecting group for ribonucleosides.12 Ogilvie
was the first to utilize the Lv group for transient 3′-protection
in conjunction with the procedures their group developed
for block coupling of 3′,5′-oligoribonucleotides.13 The present
study builds upon Ogilvie’s work as we examine, for the
first time, the application of 2′- (and 3′-) O-Lv ribonucleo-
sides as direct synthons for RNA synthesis. As documented
in an earlier study with 2′/3′-O-(2-chlorobenzoyl) ribonucleo-
sides,10b we reasoned that O-phosphitylation of a mixture of
interconverting 2′-O- and 3′-O-Lv ribonucleoside isomers
would provide separable, regioisomerically stable, phos-
phoramidite monomers that would be suitable for RNA (and
2′,5′-RNA) synthesis.
[2.2.2]octane (DABCO) (Scheme 1). Because N-levulination
of guanine proceeded sluggishly (20%),13b the dimethyl
formamidine (dmf) group14 was used instead as it has been
found that it can be efficiently introduced (>95%) and
removed under the same conditions as the N-Lv and O-Lv
groups.15 Uridine was subjected to the same steps, except
that it did not require base protection (Scheme 1). At this
stage, we obtained, as expected, inseparable mixtures of 2′/
3′-O-Lv regioisomers in favor of the 2′-O-Lv isomers (ca.
2:1 ratio for U, C, and A; 3:1 for G, after column
The required adenosine and cytosine monomers 3b/c and
4b/c were synthesized by sequential N-levulination of the
exocyclic amines (87-99%),13b 5′-dimethoxytritylation (DMTr;
88-90%), and 2′/3′-O-levulination (70-78%) using 2-chlo-
romethyl-pyridinium iodide (CMPI) and 1,4-diazabicyclo-
1
chromatography as determined by H NMR). The mixture
of 3 + 4 was then carried on to the final phosphitylation
step to afford mixtures of 5a-d and 6a-d in 77-85% yields
(% 6a-d > % 5a-d). Separation of the 2′-O- and 3′-O-
phosphoramidite regioisomers was possible by flash silica
gel column chromatography yielding 5a (19%), 6a (42%),
5b (10%), 6b (7%), 5c (29%), 6c (35%), 5d (22%), and 6d
(35%) in isomerically pure forms. These isolated yields are
generally lower than those obtained with TBDMS nucleo-
sides, particularly for the adenosine derivatives, and reflect
the very similar (tlc) chromatographic properties of each 5/6
pair (∆Rf ∼ 0.1). Nevertheless, we consider these results and
those described below to be most encouraging. The identities
of 5a-d and 6a-d were confirmed with ESI-MS in
conjunction with 2D NMR (31P-1H) experiments (Support-
ing Information).
(6) Pitsch, S.; Weiss, P. A.; Wu, X.; Ackermann, D.; Honegger, T. HelV.
Chim. Acta 1999, 82, 1753-1761.
(7) (a) Umemoto, T.; Wada, T. Tetrahedron Lett. 2004, 45, 9529-9531.
(b) Karwowski, B.; Seio, K.; Sekine, M. Nucleosides Nucleotides 2005,
24, 1111-1114. (c) Ohgi, T.; Matsutomi, Y.; Ishiyama, K.; Kitagawa, H.;
Shiba, Y.; Yano, J. Org. Lett. 2005, 7, 3447-3480. (d) Semenyuk, A.;
Foldesi, A.; Johansson, T.; Estmer-Nilsson, C.; Blomgren, P.; Brannvall,
M.; Kirseborn, L. A.; Kwiatkowski, M. J. Am. Chem. Soc. 2006, 128,
12356-12357. (e) Zhou, C.; Honcharenko, D.; Chattopadhyaya, J. Org.
Biomol. Chem. 2007, 5, 333-343.
(8) (a) Hayes, J. A.; Brunden, M. J.; Gilham, P. T.; Gough, G. R.
Tetrahedron Lett. 1985, 26, 2407-2410. (b) Ohtsuka, E.; Tanaka, S.;
Ikehara, M. J. Am. Chem. Soc. 1978, 100, 8210-8213. (c) Schwartz, M.
E.; Breaker, R. R.; Asteriadis, G. T.; deBear, J. S.; Gough, G. R. Bioorg.
Med. Chem. Lett. 1992, 2, 1019-1024.
(9) Griffin, B. E.; Reese, C. B. Tetrahedron Lett. 1964, 2925-2931.
(10) (a) Kempe, T.; Chow, F.; Sundquist, W. I.; Nardi, T. J.; Paulson,
B.; Peterson, S. M. Nucleic Acids Res. 1982, 10, 6695-6714. (b) Rozners,
E.; Renhofa, R.; Petrova, R.; Popelis, J.; Kumpins, V.; Bizdena, E.
Nucleosides Nucleotides 1992, 11, 1579-1593. (c) Reese, C. B.; Trentham,
D. R. Tetrahedron Lett. 1965, 2467-2472.
(11) Parley, N.; Baraguey, C.; Vasseur, J. J.; Debart, F. Org. Lett. 2006,
8, 3869-3872.
(12) (a) van Boom, J. H.; Burgers, P. M. J. Tetrahedron Lett. 1976,
4875-4878. (b) Ohtsuka, E.; Iwari, S. Nucleic Acids Res. 1988, 16, 9443-
9456.
With phosphoramidite monomers in hand, a 3′,5′-linked
RNA sequence (8) was synthesized on a 1 µmol scale using
monomers 5a-d (0.15 M in MeCN), 0.25 M 5-ethylthiotet-
razole (ETT) as an activator, and 14 min coupling cycles
(Table 1).
(13) (a) Ogilvie, K. K.; Nemer, M. J. Can. J. Chem. 1980, 58, 1389-
1397. (b) Ogilvie, K. K.; Nemer, M. J.; Hakimelahi, G. H.; Proba, Z. A.;
Lucas, M. Tetrahedron Lett. 1982, 23, 2615-2618. (c) Nemer, M. Ph.D.
Thesis, McGill University, Montreal, 1982 (Supervisor: K.K. Ogilvie).
(14) McBride, L.; Kierzek, R.; Beaucage, S.; Caruthers, M. J. Am. Chem.
Soc. 1986, 108, 2040-2048.
(15) Saneyoshi, H.; Seio, K.; Sekine, M. J. Org. Chem. 2005, 70, 10453-
10460.
790
Org. Lett., Vol. 9, No. 5, 2007

of: (1) treatment with anhydrous 2:3 v/v NEt3/CH3CN (rt,
60 min) to deblock the phosphates’ â-cyanoethyl groups;
(2) washing the solid support (acetonitrile, 5 min); (3)
hydrazinolysis (0.5 M hydrazine hydrate in 3:2 v/v pyridine/
acetic acid, 25 min, rt) to simultaneously deprotect bases
and hydroxyl positions; and (4) washing the solid support
to fully remove hydrazine. At this stage, the deprotected RNA
strand is attached to the CPG support through the Q linker.
A final fluoride treatment (1 M TBAF, rt, 15 min) releases
the RNA chain now ready for purification (HPLC). For
comparison, an RNA oligomer with the same sequence was
prepared under the same conditions, except that 2′-TBDMS
monomers and the conventional succinyl LCAA-CPG were
used (entry 9, Table 1). The two oligomers were identical,
as shown by PAGE, HPLC, and MALDI-TOF (calcd 6616,
found 6617 [M + H+]). Tm values of duplexes formed by
the hybridization of 8 and 9 with a complementary RNA
strand synthesized via TBDMS chemistry were 51.7 and
49 °C, respectively.
Table 1. Data for Synthesized Oligoribonucleotides
av
coup
yield
CT
b
no.
5'-sequence-3'
PG mina
%
8
9
GCUUGAAGUCUUUAAUUAAtt
GCUUGAAGUCUUUAAUUAAtt
Lv
Si
Lv
14 72.5 98.5
14 72.8 98.5
14 72.5 97.8
10 tttttttCUCUCtttc
11
12
13
14
U
U
U
U
10tt
10tt
10tt
10tt
Lv
Lv
Si
Si
1
69.6 97.0
10 71.2 97.2
10 90.1 99.1
1
10 66.8 98.1
10 71.5 98.4
10 ndd ndd
10 ndd ndd
60.7 95.9
15 U(2′OLv)19tt
16 19tt
Lv
U
Lv
OLv)Utt Si
Si
17 U(3′OLv)U3U(3′OLv)U
18 UU3UU13UUttc
U
(3′
12
a Coupling time. b% yield of oligomer in crude material (HPLC). cC
d
and U ) 2′,5′-rC and rU, respectively. t ) dT. nd ) not determined.
The solid support used was controlled pore glass (500 Å
CPG) with a 5′-O-DMTr-thymidine unit (45 µmol/g) ap-
pended through a hydroquinone-O,O′-diacetic acid linker (Q
linker).16 This linker, unlike the standard succinyl linker,
makes it possible to release an oligonucleotide chain with
fluoride ions under conditions that do not lead to internucle-
otide cleavage (1 M TBAF in THF; rt, 15 min; 92% recovery;
see below). Oxidation of the phosphite triester intermediates
was achieved using the standard 0.1 M iodine/pyridine/water
treatment. These conditions resulted in an average stepwise
coupling yield (98.5%) that was comparable to that obtained
by 2′-O-TBDMS chemistry (Figure 1).
To further confirm the integrity of the phosphodiester
linkages during RNA synthesis/deprotection, dTTTT-[3′,5′-
rU]-dTTTT and dTTTT-[2′,5′-rU]-dTTTT were prepared
using monomers 5a and 6a and commercial dT amidite.
These oligomers were separable by HPLC (rt, 42.8 and 44.2
min, respectively) and were free from their their isomeric
oligomer (detection limit <1%; Supporting Information).
We also evaluated the activity of siRNAs synthesized with
monomers 5a-d in an RNAi assay that targets luciferase
mRNA.17 The siRNA duplex [8/complement] had the same
gene silencing activity [IC50 ∼ 0.08 nM] as the reference
siRNA duplex prepared via TBDMS chemistry [9/comple-
ment], further confirming the integrity of the synthesized
RNA strand (Figure 2).
Figure 2. Silencing of luciferase mRNA expression by siRNA
duplexes (light units are relative to a scrambled siRNA control).
The antisense strands were synthesized via TBDMS and Lv
chemistries (sequences 8 and 9, respectively, Table 1), whereas
the sense strand was obtained via TBDMS chemistry.
Figure 1. Ion-exchange HPLC analysis of 5′,3′-linked siRNA (a)
8 prepared via 2′-O-Lv monomers (after purification), (b) 8 prepared
via 2′-O-Lv monomers (crude), and (c) 9 prepared via 2′-O-TBDMS
monomers (crude).
A number of 2′,5′-linked RNA sequences of moderate
length were also successfully prepared by this approach (e.g.,
Following chain assembly, the fully deprotected oligomers
were obtained following a deprotection scheme composed
(17) Dowler, T.; Bergeron, D.; Tedeschi, A.-L.; Paquet, L.; Ferrari, N.;
Damha, M. J. Nucleic Acids Res. 2006, 34, 1669-1675.
(16) Yu, S.; Pon, R. Nucleic Acids Res. 1997, 18, 3629-3635.
Org. Lett., Vol. 9, No. 5, 2007
791

entry 10, Table 1; and data not shown). The fully levulinated
sequence, 5′-(U2′OLv)19dT2-3′ (entry 15, Table 1), was also
synthesized. In this case, the solid support bound oligomer
was treated with (a) anhydrous NEt3/CH3CN (2:3 v/v; rt, 60
min) to deblock the phosphate’s â-cyanoethyl groups and
(b) 1 M TBAF (15 min) to release the protected 2′-O-Lv RNA
oligomer from the Q-CPG solid support. The average step-
wise coupling efficiency was 98.1% based on the purity of
the crude product isolated (Supporting Information). A small
amount of this material was deprotected in solution by hydra-
zinolysis (25 min; rt). After quenching the excess hydrazine
with acetone, the fully deprotected product was analyzed by
HPLC. The purity of U19tt matched that of the 2′-O-Lv
precursor (15), indicating that delevulination proceeds very
cleanly (see Supporting Information, Figure S4). Finally, the
Lv and TBDMS chemistries were combined to produce
partially esterified RNA strands (oligomer 17, Table 1; calcd
6657, found 6659 [M + H+]). Deprotection of 17 involved
on-column decyanoethylation with NEt3/CH3CN (2:3 v/v;
rt, 60 min) followed by fluoride treatment to desilylate and
release the oligomer from the Q-CPG solid support.
Partially or fully esterified RNAs may find applications
in RNAi gene silencing by virtue of their enhanced stability
against serum nucleases and, possibly, uptake by cells or
tissues. For instance, monomers 5 and 6 may be used in
combination with a N-Lv ribonucleoside containing a dif-
ferent 2′-ester group (e.g., one derived from linoleic acid)
to produce, after removal of all N/O-Lv groups, a partially
esterified siRNA.18 To study the impact of esters on the
nuclease and thermal stability of duplexes, we compared the
rate of degradation of oligomer 15 (t1/2 ∼ 2.5 h) in fetal
bovine serum to that of the native deprotected sequence 16
(t1/2 ∼15 min; Supporting Information). To assess the effect
of the O-Lv groups on the thermal stability of RNA duplexes,
the Tm values of modified hybrids 15/rA21 and 17/rA21 were
compared to those of the controls 18/rA21 (34.7 °C) and 16/
rA21 (35.7 °C). The duplex 17/rA21 showed a single coopera-
tive transition at 34.5 °C, whereas a 1:1 mixture of 15 and
rA21 showed a much weaker and broader transition that is
derived from the rA21 alone. Thus, provided that the number
of ester groups remains small, the impact of interdispersed
3′-ester groups on an RNA duplex appears to be small.
After completion of the above studies, we found that a 1
min coupling time is sufficient to provide good yields of
sequences 11 and 12 , which resulted in 70% of desired
products (HPLC analysis). The use of the bulkier 2′-TBDMS
required a 10 min coupling time to achieve excellent results
(entries 13 and 14; Table 1).
In conclusion, we have shown that the Lv group is a
suitable protecting group for the 2′-hydroxy functions of
ribonucleoside building blocks. Its major advantage over
other 2′-protecting groups is in the on-column unblocking
step at the end of the synthesis which greatly simplifies and
speeds up postsynthesis processing. With regard to the
introduction of the Lv group, the reagent required, namely,
4-oxopentanoic acid, is inexpensive and readily available
(prepared in >70% yield by treating starch or cellulose with
acid).19 Although this provides a clear cost advantage over
some current protection schemes, the arduous separation of
derivatives such as 5/6 must be taken into account. The
orthogonal deprotection conditions of 2′-TBDMS and a
transient 2′-Lv group lend themselves to the synthesis of
branched RNA.20 Furthermore, the Lv strategy should make
it possible to synthesize (and deprotect) RNA directly on
microarrays.
Acknowledgment. This paper is dedicated to Dr. Ricard
T. Pon on the occasion of his 50th birthday. We acknowledge
financial support from the Natural Science and Engineering
Research Council of Canada. We are grateful to Dr. R. T.
Pon for critically reading this manuscript and many helpful
discussions. We also acknowledge Dr. F. Robert and Dr. J.
Pelletier for conducting RNAi assays; Dr. Z. Xia for
assistance with 2D NMR; and Dr. A. Hlil for MS analyses.
M.J.D. is a James McGill Professor and the recepient of the
2007 Bernard Belleau Award (CSC).
Supporting Information Available: Experimental pro-
1
cedures, compound characterizations, and selected H and
31P NMR spectra. This material is available free of charge
via the Internet at http://pubs.acs.org.
OL0629521
(19) Ritter, S. Chem. Eng. News 2006, 84, 47.
(20) Sabatino, D.; Lackey, J. G.; Damha, M. J. Unpublished results.
(18) Lackey, J.G.; Damha, M. J. Unpublished results.
792
Org. Lett., Vol. 9, No. 5, 2007