| May
1, 2005 |
| By:
David
V. Morrissey, Shawn
P. Zinnen, Brent
A. Dickinson, Kristi
Jensen, James
A. McSwiggen, Chandra
Vargeese, Barry
Polisky |
| Pharmaceutical
Discovery |
|
Introduction The use of siRNAs
to specifically silence gene expression in cell culture has become a
powerful and widely-used research tool (1). However, the development of
siRNAs as therapeutic agents requires improvements in both their
inherent physical stability and the efficiency and specificity of
tissue-targeted delivery in vivo.
It is thought that the RNAi mechanism evolved to defend the cell
against foreign genetic elements that are presented as, or proceed
through, a double-stranded RNA (dsRNA) intermediate (2). While many of
the specific details of the RNAi mechanism have yet to be elucidated,
the general steps in the process and a number of the required proteins
have been identified. Intracellular dsRNA is processed by the nuclease
Dicer, in an ATP-dependent reaction, into defined duplexes of
approximately 21 base pairs (3-5), termed siRNAs (4). siRNAs have 5'
terminal phosphates (6-8), 3' terminal hydroxyls and two or three
nucleotide 3' terminal overhangs (5, 9, 10). siRNAs apparently undergo
an ATP-dependent unwinding step and are then incorporated into the
RNA-induced silencing complex (RISC) (6). RISC isolated from cell
extracts contains only one of the two siRNA strands (11). However,
psoralen cross-linked strands retain activity, indicating that total
unwinding is not essential for activity (7). The loaded RISC then is
competent to degrade target RNAs having sequence complementarity to the
siRNA strand it is carrying. Activated RISCs have been shown to be
capable of multiple turnovers (12), and silencing activity appears to
last three to five days in many cell culture systems.
Chemical modifications in synthetic siRNAs for the purpose of
stabilization not only must provide resistance to nuclease degradation
but also permit proper recognition and function of the siRNA. We
designed and synthesized a series of chemically-modified siRNAs to
assess the degree of modification required to improve resistance to
nuclei and determine the extent and type of modification tolerated by
the RNAi mechanism. Resistance to nuclease degradation was assessed in
human or mouse serum and in human or mouse liver extracts. The silencing
activity of modified siRNAs targeted to hepatitis B virus RNA was
evaluated in a cell culture system. We observed that modified siRNAs
completely lacking 2' OH residues demonstrated increased human serum
stability (t1/2=39-408 h) and potent levels of silencing
activity.
Design of Modified siRNAs and the
Effect of Modifications on Stability Selection of chemical
modifications. A number of studies have examined the effects of
various modifications on silencing activity (9, 13-16). Phosphorothioate
linkages are tolerated in siRNA (16-18) although total P-S substitution
can reduce siRNA efficacy, compared to P-O linkages (19). The presence
of 2'-fluoro residues at pyrimidine positions has no apparent negative
effect on silencing in a number of mammalian cell-based systems (17, 18,
20, 21). Deoxyribose substitutions are partially tolerated, but tend to
reduce overall activity (9, 16, 18). DNA/RNA heteroduplexes have been
observed to be inactive, regardless of which strand carried the
deoxyribose substitutions (9, 16), while others have demonstrated
measurable activity of DNA/RNA and 2'-OMe DNA/RNA heteroduplexes only
when the antisense strand was all RNA. Activity has been demonstrated in
HeLa cells with siRNAs with up to four positions as locked nucleic acid
(LNA) substitutions.
We evaluated the effects of chemical modifications on synthetic siRNA
stability and function using 2'-fluoro, 2'-OMe and 2'-deoxy sugars and
terminus capping chemistries. Five modified strands, designated A
through E (Table I), were tested in three sets of duplexes that
demonstrated a range of stability and activity. These duplexes,
designated A:B, C:D and C:E, contained differentially modified sense and
antisense strands. All modified sense strands contained terminal 5' and
3' inverted abasic caps, while antisense strands had a single 3'
terminal TsT linkage. Modified sense and antisense strands had 2'-fluoro
substitutions at all pyrimidine positions. Purine positions were 2'-OH,
2'-H or 2'-OMe, as detailed in Table I.
Figure 1. Human liver and serum
stability time courses. The fraction of the full-length
radiolabeled antisense strand in unmodified or C:E duplexes
present in human liver or serum is shown as a function of time.
Time courses were fit to a first order exponential. The t1/2s
and correlation coefficients follow respectively: Open triangle,
E strand in the CE duplex in human serum, t1/2 = 3.3
days; R = 0.98. Open square, E strand in the CE duplex in human
liver homogenate, t1/2 = 36 days; R = 0.60. Open
diamond, antisense strand in the unmodified duplex in human
serum, t1/2 = 1.02 min; R = 0.999. Open circle,
antisense strand in the unmodified duplex in human liver
homogenate, t1/2 = 2.5 h; R = 0.996.
|
Effect of siRNA modification on
stability. Modified and unmodified siRNA were assessed for their
resistance to degradation in human and mouse serum and liver extracts.
We examined the stability of modified siRNAs in 90% serum, rather than
much lower levels typical of cell culture conditions (19), to more
closely emulate in vivo conditions. Following gel
electrophoresis, full-length material was quantitated by phosphorimaging.
Representative curve fits allowed the derivation of half-lives, as shown
in Figure 1 and Table II.
Table I. Sequence and modification
description
|
The A:B duplex contains 2'-fluoro
substitution on all pyrimidine positions. This modification provides
significant stability in human and mouse serum (t1/2s range
from 10-408 h) and human liver extract (t1/2s range from
28-43 h). In human serum, the A strand in the context of the A:B duplex,
possesses greater stability than the B strand (t1/2 = 408 vs.
39 h). While this difference could be due to the sequence differences
between the sense and antisense strands, we believe it is a consequence
of the impact of terminal modifications on stability. We have observed
greater than 400-fold increases in stability comparing inverted abasic
terminal capping to unmodified termini, in the context of 2'-fluoro
pyrimidine modifications (data not shown).
Table II. Duplex half-lives in
human and mouse serum and liver extracts
|
Next , we generated a fully-modified
(no 2'-OH residues) C:D duplex by substituting the ribose sugars in all
purine positions with deoxyribose. The human serum stability of the
antisense strand was increased, but in other test conditions the
deoxyribose purines provided less stabilization than ribose purines
(Table II). The second fully-modified construct, C:E, replaced all
purine positions in the antisense strand with 2'-OMe ribose. This
construct proved to be the most stable antisense strand observed, with t1/2
= 816 h in human liver extract.
Effect of Chemical Stabilization on
siRNA Activity HBV RNA as a
target for siRNA. To assess the activity of normal and modified
siRNAs against HBV, a replication-competent HBV cDNA, derived from the
psHBV-1 vector, was co-transfected along with duplexed siRNA into human
Hep G2 cells. The transfected, circularized cDNA forms the replicative
intermediate in the nucleus, thus bypassing the infection step and
initiating the HBV life cycle (23).
Table III. The activity of
modified and unmodified siRNAs targeted to HBV site 1580 in
reduction of HBsAg levels in transfected HepG2 cells
|
Silencing activities of modified anti-HBV
siRNA compared to the all-ribose molecule in Hep G2 cells are shown in
Table III, in which the activity (percent inhibition compared to matched
inverted control) of the all-ribose siRNA is considered to be 100%. At
all three siRNA concentrations tested (100, 50 and 25 nM), the
partially-modified A:B siRNAs display essentially the same level of
silencing activity as the all-RNA molecule. The fully-modified C:D and
C:E duplexes had very similar levels of activity, which ranged from
80-90% of that of the unmodified siRNA.
Conclusions Earlier
studies with potential nucleic acid-based therapeutics, such as
antisense, aptamers and ribozymes, have established the necessity of
chemical modification to increase nuclease resistance (24-26). The need
for chemical stabilization of siRNA is demonstrated by the instability
of all-ribose siRNA in serum and liver extracts. The series of chemical
modifications described here have improved stability while retaining
high levels of silencing activity.
In the HBV cell culture system,
partially-modified siRNA (duplex A:B) displayed silencing activity
equivalent to that of unmodified siRNA. The levels of activity of
fully-modified siRNAs (duplexes C:D and C:E) were 10-20% less than those
of the unmodified or partially-modified duplexes. Similar results have
been obtained with siRNAs directed against a number of non-viral
endogenous RNA targets in cell culture (data not shown).
This report describes chemically
modified siRNAs that possess ex vivo stabilities required of
systemically delivered drugs. Taken together, these modifications
represent an important step towards the introduction of siRNA into
validated animal models and, ultimately, the clinic.
Acknowledgements
The authors thank Chris Shafer and Keith Bowman for their significant
contributions to this work.
References
1. T. Tuschl and A. Borkhardt, Molecular Interventions 2, 158-167
(2002).
2. P.D. Zamore, Science 296,
1265-1269 (2002).
3. S.M. Hammond, E. Bernstein, D.
Beach and G.J. Hannon, Nature 404, 293-296 (2000).
4. P.D. Zamore, T. Tuschl, P.A.
Sharp and D.P. Bartel, Cell 101, 25-33 (2000).
5. S.M. Elbashir, J. Harborth, W.
Lendeckel et al., Nature 411, 494-498 (2001).
6. A. Nykanen, B. Haley and P.D.
Zamore, Cell 107, 309-321 (2001).
7. Y-L. Chiu and T.M. Rana, Molecular
Cell 10, 549-561 (2002).
8. D.S. Schwarz, G. Hutvagner, B.
Haley and P.D. Zamore, Molecular Cell 10, 537-548 (2002).
9. S.M. Elbashir, J. Martinez, A.
Patkaniowska et al., The EMBO Journal 20, 6877-6888
(2001).
10. S.M. Elbashir, W. Lendeckel and
T. Tuschl, Genes & Development 15, 188-200 (2001).
11. J. Martinez, A. Patkaniowska,
H. Urlaub et al., Cell 110, 563 (2002).
12. G. Hutvagner and P.D. Zamore, Science
297, 2056-2060 (2002).
13. H. Hohjoh, Febs Letters 521,
195-199 (2002).
14. M. Amarzguioui, T. Holen, E.
Babaie and H. Prydz, Nucleic Acids Res. 31, 589-595
(2003).
15. Y-L. Chiu and T.M. Rana, Molecular
Cell 10, 549-561 (2002).
16. M. Manoharan, Curr. Opinion
Chem. Biol. 8, 570-579 (2004).
17. J. Harborth, S.M. Elbashir, K.
Vandenburgh et al., Antisense & Nucleic Acid Drug Development
13, 83-105 (2003).
18. Y-L. Chiu, and T.M. Rana, RNA
9, 1034-1048 (2003).
19. D. Braasch, S. Jensen, Y. Liu
et al., Biochemistry 42, 7967-7975 (2003).
20. J. Capodici, K. Kariko and D.
Weissman, Journal of Immunology 169, 5196-5201 (2002).
21. J. Meis and J. Decker, Epicentre
Forum 10, 4-5 (2003).
22. F. Wincott, A. DiRenzo et al., Nucleic
Acids Res. 23, 2677-2684 (1995).
23. M.A. Sells, M.L. Chen and G.
Acs, Proc. Natl. Acad. Sci. USA 84, 1005-1009 (1987).
24. L. Beigelman, J.A. McSwiggen,
K.G. Draper et al., J Biol Chem. 270, 25702-25708 (1995).
25. S.P. Zinnen, K. Domenico, M.
Wilson et al., RNA 8, 214-228 (2002).
26. S. Agrawal and E.R. Kandimalla,
Curr. Cancer Drug Targets 1, 197-209 (2001).
David V. Morrissey is
associate director of biology, Shawn P. Zinnen is associate
director of biochemistry, Brent A. Dickinson and Kristi Jensen
are research associates, James A. McSwiggen is senior scientist, Chandra
Vargeese is vice president of chemistry and Barry Polisky is
senior vice president and chief scientific officer at Sirna
Therapeutics. Barry Polisky can be reached at 2950 Wilderness Place,
Boulder, CO 80301 USA; e-mail poliskyb@sirna.com.
|
