Efficient RNase H-directed cleavage of RNA promoted by antisense DNA or 2′F-ANA constructs containing acyclic nucleotide inserts

Article abstract of DOI:10.1021/ja025557o

The ability of modified antisense oligonucleotides (AONs) containing acyclic interresidue units to support RNase H-promoted cleavage of complementary RNA is described. Manipulation of the backbone and sugar geometries in these conformationally labile mono

Full text of DOI:10.1021/ja025557o

Published on Web 12/20/2002
Efficient RNase H-Directed Cleavage of RNA Promoted by
Antisense DNA or 2′F-ANA Constructs Containing Acyclic
Nucleotide Inserts
Maria M. Mangos, Kyung-Lyum Min, Ekaterina Viazovkina, Annie Galarneau,
Mohamed I. Elzagheid, Michael A. Parniak,† and Masad J. Damha*
Contribution from McGill UniVersity, Departments of Chemistry and Experimental Medicine,
Montreal, QC, Canada
Received January 10, 2002 ; E-mail: masad.damha@mcgill.ca
Abstract: The ability of modified antisense oligonucleotides (AONs) containing acyclic interresidue units
to support RNase H-promoted cleavage of complementary RNA is described. Manipulation of the backbone
and sugar geometries in these conformationally labile monomers shows great benefits in the enzymatic
recognition of the nucleic acid hybrids, while highlighting the importance of local strand conformation on
the hydrolytic efficiency of the enzyme more conclusively. Our results demonstrate that the duplexes support
remarkably high levels of enzymatic degradation when treated with human RNase HII, making them efficient
mimics of the native substrates. Furthermore, interesting linker-dependent modulation of enzymatic activity
is observed during in vitro assays, suggesting a potential role for this AON class in an RNase H-dependent
pathway of controlling RNA expression. Additionally, the butyl-modified 2′F-ANA AONs described in this
work constitute the first examples of a nucleic acid species capable of eliciting high RNase H activity while
possessing a highly flexible molecular architecture at predetermined sites along the AON.
of the latter.4 Thus, a single enzyme-active AON can ideally
induce the specific destruction of multiple copies of an RNA
Introduction
Antisense oligonucleotides (AONs) represent an attractive and
promising class of clinically useful compounds that are rationally
designed to alter gene expression patterns.1 As more sophisti-
cated methods evolve for their construction, their widespread
use in functional genomics, drug target validation, and as human
therapeutics may soon be realized.2 Good exo- and endonucleo-
lytic robustness, efficient cell entry, and tight and specific
binding with their intended genetic targets in the presence of
cellular proteins2c are few of the criteria important for the
chemotherapeutic potential of AONs.
transcript (i.e., AON recycling) and exhibit greater biological
activity than those that do not activate RNase H.1a To date,
however, most AONs are deficient in one or more of these
requisite traits.
Indeed, oligonucleotide “gapmer” modifications that combine
intervening enzyme-active AON segments (e.g., DNA) with
conformationally restricted residues (e.g., 2′-O-alkyl RNA) at
the periphery may still effect cleavage of their intended targets.5
However, these usually focus on preorganizing the antisense
strand to adopt a more compatible structure that affords tighter
target binding, while compromising high RNase H induction.6
Our previous discovery7 that arabinose-derived AON (e.g.,
ANA, 2′F-ANA) can elicit RNase H degradation of target RNA
has established that an O4′-endo conformation8 in the antisense
sugar residues is of prime importance;8a,9 yet, other carbohydrate-
modified AONs with this same antipodal bias have been unable
Additionally, a desirable mechanism which could provide a
catalytic mode of supressing gene activity in eukaryotic cells
relies upon the activity of ribonuclease H (RNase H), a
ubiquitous enzyme of many vital intracellular functions.3 These
activities ultimately culminate in the destruction of the RNA
component when duplexed with DNA or a suitable AON variant
* To whom correspondence should be addressed. Otto Maass Chemistry
Bldg., McGill University, 801 Sherbrooke St. W., Montreal, QC, Canada
H3A 2K6.
(4) (a) Nakamura, H.; Oda, Y.; Iwai, S.; Inoue, H.; Ohtsuka, E.; Kanaya, S.;
Kimura, S.; Katsuda, C.; Katayanagi, K.; Morikawa, K.; Miyashiro, H.;
Ikehara, M. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 11535-11539. (b) Oda,
Y.; Iwai, S.; Ohtsuka, E.; Ishikawa, M.; Ikehara, M.; Nakamura, H. Nucleic
Acids Res. 1993, 21, 4690-4695. (c) Uchiyama, Y.; Miura, Y.; Inoue, H.;
Ohtsuka, E.; Ueno, Y.; Ikehara, M. J. Mol. Biol. 1994, 4, 782-791.
(5) (a) Baker, B. F.; Lot, S. S.; Condon, T. P.; Cheng-Flournoy, S.; Lesnik, E.
A.; Sasmor, H. M.; Bennett, C. F. J. Biol. Chem. 1997, 272, 11994-12000.
(b) Freier, S. M.; Altmann, K.-H. Nucleic Acids Res. 1997, 25, 4429-
4443. (c) Nishizaki, T.; Iwai, S.; Ohtsuka, E.; Nakamura, H. Biochemistry
1997, 36, 2577-2585.
† Current address: Department of Medicine, University of Pittsburgh,
Scaife Hall, 3550 Terrace St., Pittsburgh, PA 15261.
(1) (a) Uhlmann, E.; Peyman, A. Chem. ReV. 1990, 90, 543-584. (b) Crooke,
S. T.; Bennett, C. F. Annu. ReV. Pharmacol. Toxicol. 1996, 36, 107-129.
(2) (a) Eckstein, F. Annu. ReV. Biochem. 1985, 54, 367-402. (b) Akhtar, S.;
Agrawal, S. Trends Pharmacol. Sci. 1997, 18, 12-18. (c) See, for
example: Lebedeva, I.; Stein, C. Annu. ReV. Pharmacol. Toxicol. 2001,
41, 403-419 and references therein.
(3) (a) Walder, R. Y.; Walder, J. A. Proc. Natl. Acad. Sci. U.S.A. 1978, 75,
5011-5015. (b) Itaya, M.; Kondo, K. Nucleic Acids Res. 1991, 19, 4443-
4449. (c) Monia, B. P.; Lesnik, E. A.; Gonzalez, C.; Lima, W. F.; McGee,
D.; Guinosso, C. J.; Kawasaki, A. M.; Cook, P. D.; Freier, S. M. J. Biol.
Chem. 1993, 268, 14514-14522.
(6) (a) Shen, L. X.; Kandimalla, E. R.; Agrawal, S. Bioorg. Med. Chem. 1998,
6, 1695-1705. (b) Christensen, N. K.; Peterssen, M.; Nielsen, P.; Jacobsen,
J. P.; Olsen, C. E.; Wengel, J. J. Am. Chem. Soc. 1998, 120, 5458-5463.
(c) Nielsen, K. E.; Singh, S. K.; Wengel, J.; Jacobsen, J. P. Bioconjugate
Chem. 2000, 11, 228-238.
9
654
J. AM. CHEM. SOC. 2003, 125, 654-661
10.1021/ja025557o CCC: $25.00 © 2003 American Chemical Society

Oligonucleotide Flexibility and RNase H
A R T I C L E S
NaIO3 salts and followed by in situ reduction of the dialdehyde via
treatment with NaBH4 (0.378 g, 10 mmol, 1.0 equiv) for 10-20 min
at room temperature. The reaction mixture was quenched with acetone,
neutralized with 20% acetic acid, and concentrated to an oil under
reduced pressure. The residue was then diluted with CH2Cl2 (200 mL)
and washed with H2O. The aqueous layer was back-extracted, and the
combined organic layers were dried using anhydrous Na2SO4, filtered,
and evaporated to give the product as a pure white foam in 98% isolated
yield (5.08 g; 9.8 mmol). Rf (CH2Cl2:MeOH, 9:1) 0.18; FAB-MS (NBA)
519.6; Calcd 518.57.
Figure 1. AON constructs containing acyclic internucleotide units.
(B) 5′-O-MMT-2′-O-tert-butyldimethysilyl-2′,3′-secouridine (3b)
and 5′-O-MMT-3′-O-tert-butyldimethysilyl-2′,3′-secouridine (3c).
Monoprotection of either of the free hydroxyl functions of 3a was
achieved nonselectively by adding tert-butyldimethylsilyl chloride (0.81
g, 5.4 mmol, 1.1 equiv) to a stirred 0.1 M solution of 3a (2.55 g, 4.9
mmol) in dry THF at 0 °C containing a suspension of AgNO3 (0.92 g,
5.39 mmol, 1.1 equiv). The reaction temperature was returned to room
temperature after 20 min and maintained as such for 24 h. The workup
was initiated by filtering the mixture directly into an aqueous solution
of 5% NaHCO3 (50 mL), followed by extraction of the aqueous layer
twice with CH2Cl2. The combined organic layers were dried (anhydrous
Na2SO4), filtered, and evaporated under reduced pressure to give the
crude product as a yellow oil. The residue was purified by flash silica
gel column chromatography using a gradient of 0-25% acetone in
CH2Cl2 to recover both monosilyl isomers as pure white foams. Isolated
yields for 3b and 3c were 22% and 14%, respectively. Rf (CH2Cl2:
Et2O, 3:1) 3b, 0.18; 3c, 0.05. FAB-MS (NBA) 633.4; Calcd 632.83.
The regioisomers are distinguished on the basis of COSY NMR
cross-peak correlations that are used to demonstrate the connectivity
of the protons in the acyclosugar. In both spectra, the H1′ protons are
split by the nonequivalent H2′ and H2′′ protons into a doublet of
doublets which suggests a certain degree of structural rigidity around
the C1′-C2′ bond. More significantly, a single well-resolved hydroxyl
peak is observed for both 3b and 3c in DMSO-d6 which negates rapid
chemical exchange of these moieties. As a result, the effect of the
protons at C2′ of 3c is transmitted to the 2′-hydroxyl proton which in
turn appears as an overlapping doublet of doublets. In 3b, splitting of
the hydroxyl resonance is also observed; however, it shows correlations
with H3′ and H3′′ and therefore rules out the presence of a silyl group
at the 3′-position. Taken together, these data confirm the assignment
of 3b and 3c as the 2′- and 3′-monosilylated isomers, respectively.
(C) 5′-O-MMT-3′-O-tert-butyldimethysilyl-2′,3′-secouridine-2′-O-
[N,N-diisopropylamino-(2-cyanoethyl)]phosphoramidite (3d). To a
nitrogen-purged solution of 4-(dimethylamino)pyridine (DMAP; 12 mg,
0.10 mmol, 0.1 equiv), N,N-diisopropylethylamine (DIPEA; 0.68 mL,
3.9 mmol, 4 equiv) and 3c (620 mg, 0.98 mmol) in THF (0.2 M) at 0
°C was added N,N-diisopropylamino-â-cyanoethylphosphonamidic
chloride (0.24 mL, 1.1 mmol, 1.1 equiv) dropwise over 5 min. The
immediate appearance of a white precipitate due to the rapid formation
of diisopropylethylammonium hydrochloride signified sufficiently
anhydrous conditions, and the reaction was allowed to warm to room
temperature, whereupon it was stirred for 2.5 h prior to the reaction
workup. Briefly, the reaction mixture was diluted with EtOAc (50 mL,
prewashed with 5% NaHCO3) and washed with saturated brine (2 ×
20 mL). The recovered organic layer was dried (anhydrous Na2SO4)
and filtered, and the solvent was removed via reduced pressure, yielding
a crude yellow oil. Co-evaporation of the crude product with Et2O
afforded a pale yellow foam. Purification of the product by flash silica
gel column chromatography using a CH2Cl2:hexanes:TEA gradient
system (25:74:1 adjusted to 50:49:1) afforded a white foam in 99%
isolated yield. Rf (EtOAc:Tol, 4:1) 0.77, 0.65. FAB-MS (NBA) 833.3;
Calcd 833.05.
to sustain the same activity (e.g., [3.3.0]bc-ANA).6b,10 Thus, we
hypothesize that the inherent flexibility of these sugar conforma-
tions impacts on the ability of the enzyme7e,9a to bind to and
recognize hybrids of ANA-derived AON and RNA.
We wished to explore the significance of imparting con-
formational variability in 2′F-ANA (2′-deoxy-2′-fluoro-â-D-
arabinonucleic acid, 2) to determine whether the insertion of
an acyclic residue in a known RNase H-active AON could
accelerate the enzymatic degradation. This was accomplished
by the systematic introduction of acyclic nucleotides consisting
of a 2′,3′-secouridine synthon 3 or a butanediol linker 4 (Figure
1). We surmised that this relatively unconstrained molecular
architecture would improve RNase H induction properties as
well as provide some insight toward elucidating the structural
factors that provide the “optimal” AON/RNA substrate.
Materials and Methods
Synthesis of Acyclic Monomers and AONs. The synthesis of 2′F-
ANA monomers of 2 and of mixed backbone AON comprising 2′F-
ANA and DNA has already been extensively described elsewhere.7a,b,f,g
Secouridine 2′-phosphoramidites of 3 were prepared by variations of
published protocols.11 Briefly, the acyclic nucleoside residues consist
of a 1-[1,5-dihydroxy-4(S)-hydroxymethyl-3-oxapent-2(R)-yl]-uracil
unit which has been appropriately protected (Scheme 1) and function-
alized for oligonucleotide incorporation as described below.
(A) 5′-O-MMT-2′,3′-seco-â-D-uridine (3a). To a 0.1 M solution
of 5′-monomethoxytrityluridine12 (5′-MMT-rU, 5.16 g, 10 mmol) in
dioxane was added a saturated solution of NaIO4 in H2O (2.26 g, 10.6
mmol, 1.06 equiv) and the reaction allowed to proceed at room
temperature for 2-3 h until complete conversion to the dialdehyde
was observed by TLC visualization (Rf 0.52 in CH2Cl2:MeOH, 9:1).
The reaction was diluted with dioxane (100 mL), filtered to remove
(7) (a) Damha, M. J.; Wilds, C. J.; Noronha, A.; Brukner, I.; Borkow, G.;
Arion, D.; Parniak, M. A. J. Am. Chem. Soc. 1998, 120, 12976-12977.
(b) Wilds, C. J.; Damha, M. J. Nucleic Acids Res. 2000, 28, 3625-3635.
(c) Noronha, A. M.; Wilds, C. J.; Lok, C.-N.; Viazovkina, K.; Arion, D.;
Parniak, M. A.; Damha, M. J. Biochemistry 2000, 39, 7050-7062. (d)
Damha, M. J.; Noronha, A. M.; Wilds, C. J.; Trempe, J.-F.; Denisov, A.;
Gehring, K. Nucleosides Nucleotides Nucleic Acids 2001, 20, 429-440.
(e) Lok, C.-N.; Viazovkina, E.; Min, K.-L.; Nagy, E.; Wilds, C. J.; Damha,
M. J.; Parniak, M. A. Biochemistry 2002, 41, 3457-3467. (f) Elzagheid,
M. I.; Viazovkina, E.; Damha, M. J. In Current Protocols in Nucleic Acid
Chemistry, Unit 1.7; Beaucage, S. L., Bergstrom, D. E., Gli, G. D., Eds.;
2002. (g) Viazovkina, E.; Mangos, M. M.; Elzagheid, M. I.; Damha, M. J.
In Current Protocols in Nucleic Acid Chemistry, Unit 4.15; Beaucage, S.
L., Bergstrom, D. E., Gli, G. D., Eds.; 2002.
(8) (a) Berger, I.; Tereshko, V.; Ikeda, H.; Marquez, V. E.; Egli, M. Nucleic
Acids Res. 1998, 26, 2473-2480. (b) Ikeda, H.; Fernandez, R.; Wilk, A.;
Barchi, J. J.; Huang, X.; Marquez, V. E. Nucleic Acids Res. 1998, 26, 2237-
2244. (c) Venkateswarlu, D.; Ferguson, D. M. J. Am. Chem. Soc. 1999,
121, 5609-5610.
(9) (a) Denisov, A. Y.; Noronha, A. M.; Wilds, C. J.; Trempe, J.-F.; Pon, R.
T.; Gehring, K.; Damha, M. J. Nucleic Acids Res. 2001, 29, 4284-4293.
(b) Trempe, J. F.; Wilds, C. J.; Denisov, A. Y.; Pon, R. T.; Damha, M. J.;
Gehring, K. J. Am. Chem. Soc. 2001, 123, 4896-4903.
(10) Minasov, G.; Teplova, M.; Wengel, J.; Egli, M. Biochemistry 2000, 39,
3525-3532.
The
dimethoxytrityl-butanediol-O-N,N-diisopropylamino-O-(2-
(11) (a) Mikhailov, S. N.; Pfleiderer, W. Tetrahedron Lett. 1985, 26, 2059-
2062. (b) Nielsen, P.; Dreiøe, L. H.; Wengel, J. Bioorg. Med. Chem. 1995,
3, 19-28.
cyanoethyl)-phosphoramidite precursor 4a was purchased from Chem-
Genes Corp. (Ashland, MA) and was used as received. DNA or 2′F-
ANA oligomers containing 3 or 4 were prepared as described
(12) Wu, T.; Ogilvie, K. K.; Pon, R. T. Nucleic Acids Res. 1989, 17, 3501-
3517.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 655

A R T I C L E S
Mangos et al.
Scheme 1 a
a Reactants: (a) NaIO4, dioxane, H2O, 2 h (quantitative); (b) NaBH4, dioxane, H2O, 15 min (quantitative); (c) t-BDMS-Cl, AgNO3, THF, 24 h (3b, 22%;
3c, 14%); (d) Cl-P(OCE)N(i-Pr)2, DMAP, EtN(i-Pr)2, THF, 2.5 h (99%).
Table 1. Melting Temperatures (Tm) and RNase H-Mediated Hydrolysis Profiles for the AON/RNA Heteroduplexesa
relative rates (k )
rel
entry
sequence typeb,c (5′f3′)
(i) DNA
Tm (°C)
of enzyme cleavaged
I
II
ttt ttt ttt ttt ttt ttt
ttt ttt ttt Btt ttt ttt
39
33
1
2.7
III
IV
V
tta tat ttt ttc ttt ccc
tta tat ttt Btc ttt ccc
tta tat ttt ctc ttt ccc
tta tat ttt B ttc ttt ccc
53
48
40
48
1
3.4
0.7
2.5
VI
(ii) 2′F-ANA
VII
VIII
IX
X
XI
TTT TTT TTT TTT TTT TTT
TTT TBT TTT TTT TTT TTT
TTT TTT TTT BTT TTT TTT
TTT TTT TTT TTT BTT TTT
TTT TTT TTT STT TTT TTT
TTT TTT TTS STT TTT TTT
53
49
47
49
47
42
1
0.6
7.9
5.1
1.6
2.8
XII
XIII
XIV
XV
XVI
XVII
TTA TAT TTT TTC TTT CCC
TTA TAT TTT BTC TTT CCC
TTA TAT TTT CTC TTT CCC
TTA TAT TTT tTC TTT CCC
TTA TAT TTT B TTC TTT CCC
64
55
55.5
63
57
1
3.5
0.9
1.6
2.3
(iii) Ha-ras AONe
XVIII
XIX
XX
XXI
XXII
att ccg tca tcg ctc ctc
att ccg tca Bcg ctc ctc
att ccg tca ccg ctc ctc
ATT CCG TCA TCG CTC CTC
ATT CCG TCA BCG CTC CTC
70
58
63
82
72
33.8
31.6
31.9
1
23.3
a Aqueous solutions of 2.38 × 10-6 M of each oligonucleotide, 140 mM KCl, 1 mM MgCl2, 5 mM Na2HPO4 buffer (pH 7.2); uncertainty in Tm is (0.5
°C. b Target RNA sequences correspond to rA18, or 5′-r(GGGAAAGAAAAAAUAUAA)-3′. c Upper case letters, 2′F-ANA nucleotides; lower case letters,
deoxynucleotides; C, arabinofluorocytidine or c, deoxycytidine mismatch residue; B, butanediol linker; S, 2′-secouridine insert. d Human enzyme; rates
shown have been obtained at 22 °C and are normalized according to the parent strand of each series except within Ha-ras sequences in which data have been
obtained at 37 °C and are normalized to all-2′F-ANA AON (entry XXI). e Target RNA (40mer) sequence: 5′-r(CGCAGGCCCCUGAGGAGCGAUGACG-
GAAUAUAAGCUGGUG)-3′; underlined residues denote the region in the RNA to which the AON binds.
previously,7a,b and were characterized by analytical PAGE, anion
exchange HPLC, and MALDI-TOF mass spectrometry. Sequences of
the oligonucleotides used are provided in Table 1.
RNase H Assays. RNase H assays were carried out at 14-15 °C or
room temperature (ca. 22 °C), with the exception of Ha-ras derived
sequences in which assays were conducted at 37 °C. Nucleic acid duplex
substrates were prepared by mixing the AON (3 pmol) with 5′-32P-
labeled complementary target RNA (1 pmol) in 10 µL of 60 mM Tris-
HCl (pH 7.8) containing 60 mM KCl and 2.5 mM MgCl2, followed by
heating at 90 °C for 2 min and slow cooling to room temperature.
Duplex substrate solutions were allowed to stand at room temperature
for at least 1 h prior to use. Reactions were initiated by the addition of
human RNase HII or E. coli RNase HI, and aliquots were removed at
various times and quenched by the addition of an equal volume of
98% deionized formamide containing 10 mM EDTA, 1 mg/mL
UV Thermal Denaturation Measurements. AON and comple-
mentary target RNA oligonucleotides were mixed in equimolar ratios
in 140 mM KCl, 1 mM MgCl2, and 5 mM Na2HPO4 buffer, pH 7.2, to
provide a total duplex concentration of ca. 5 µM. Samples were heated
to 90 °C for 15 min and then cooled slowly to room temperature, and
stored at 4 °C overnight before measurements. The AON/RNA duplex
solution was then exposed to increasing temperature at 1-min intervals
(0.5 °C/measurement), and the UV absorbance at 260 nm was
determined after temperature equilibration. Tm values were calculated
using the baseline method and have an uncertainty of (0.5 °C.
Purification of RNase H. Escherichia coli RNase HI was purified
as described previously.13 Human RNase HII was overexpressed and
purified according to the method of Wu et al.14
(13) Yang, W.; Hendrickson, W. A.; Kalman, E. T.; Crouch, R. J. J. Biol. Chem.
1990, 265, 13553-13559.
(14) Wu, H.; Lima, W. F.; Crooke, S. T. J. Biol. Chem. 1999, 274, 28270-
28278.
9
656 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Oligonucleotide Flexibility and RNase H
A R T I C L E S
Figure 2. (A) PAGE results showing RNase H-mediated cleavage of duplexes. Assays comprised 1 pmol of target 5′-[32P]-rA18 and 3 pmol of test AON
in buffer (10 µL) consisting of 60 mM Tris-HCl (pH 7.8), 2 mM dithiothreitol, 60 mM KCl, and 2.5 mM MgCl2. Reactions were started by the addition of
human RNase HII (22 °C; lanes: 0, 5, 10, 20, 30 min); arrow indicates coalesced 5′-scission products obtained from multiple cleavage within the same target
strand. (B) Quantification of remaining full-length [32P]-RNA obtained by UN-SCAN-IT software (Silk Scientific; Orem, UT).
bromophenol blue and 1 mg/mL xylene cyanol. After heating at 100
°C for 5 min, reaction products were resolved by electrophoresis on
16% polyacrylamide sequencing gels containing 7 M urea, visualized
by autoradiography, and product formation was quantified by densi-
tometry.
18-nt oligomer enhances RNase H-assisted sense strand hy-
drolysis (Figure 2). Moreover, a second contiguous insert
amplifies the effect, whereas the incorporation of a single
butanediol insert in the middle of the same sequence potentiates
the targeted destruction of the AON-chimera:substrate duplex
even further, as much as ca. 8-fold relative to the uniformly
modified control AON (Table 1, e.g., VII and IX). This
methodology is entirely applicable in DNA backbones as well,
in which the relative rate enhancements easily approach a 3-fold
improvement in target degradation over the native strands (e.g.,
I vs II; III vs IV). The enhancement in rate, coupled with a
distinct cleavage pattern upstream of the site of the inserts in
the AON, is also apparent in mixed-sequence analogues (see
Supporting Information), suggesting the generality of the
structural effects on the conformational discretion of the human
enzyme.
Results and Discussion
Duplexation Studies. In assessing the thermal behavior of
the AONs, we have noted without exception that a drop in
stability is characteristic of all strands with single acyclic inserts
(Table 1). While this was expected for butyl-containing AONs,
we were somewhat surprised that 3 offered no improvement in
binding affinity relative to the former. The effect was ap-
proximately additive upon incorporating a second secouridine
residue, suggesting minimal base pairing or interresidue base
stacking. The extent to which the linkers were destabilizing was
also contingent upon base sequence, that is, mixed 2′-fluoro-
arabino sequences with the butanediol linker 4 were ca. 3 °C
more destabilizing than their homopolymeric counterparts (IX
and XIV). Single substitutions were also less tolerated at the
center (IX; ∆Tm ) -6 °C) as opposed to the ends of the
sequences (VIII and X; ∆Tm ) -4 °C), which is an expected
and recurrent phenomenon11b,15 that is manifest from end-fraying
of the duplex.
Interestingly, placement of a C-A mismatch in either of the
mixed-sequence AON backbones significantly destabilizes their
hybrids with RNA, with the deoxy sequences proving less
competent toward retaining adequate target avidity as compared
to the 2′F-ANA complements. Furthermore, only modest
destabilization occurs upon insertion of one deoxynucleotide
in the all-2′F-ANA mixed sequences, which is consistent with
the notion that the arabinofluoro residues enhance thermal
stability via strand preorganization over the native residues yet
do not perturb the stereoregular structure of the AON to any
appreciable degree.7a,8,9
A possible, yet partial, explanation for the elevated RNA
degradation may owe to an increased turnover efficiency of the
enzyme for the linker-modified duplexes, which may arise from
an enhanced rate of dissociation for these hybrids over their
thermostable counterparts (e.g., ∆Tm ) -9 °C for XIV vs XIII).
For instance, 2′F-ANA, which forms stronger duplexes, exhibits
reduced turnover relative to DNA and 2′F-ANA-B10 hybrids,
and therefore would seem to accommodate such a trend (see
Supporting Information). On the other hand, the activity is
diminished in mismatched DNA and 2′F-ANA AON mixed
sequences, where a respective deoxy- or arabinofluorocytidine
mismatch replaces the butyl linker (Table 1, V and XV).
Because the mismatch also induces an equivalent (2′F-ANA)
or larger (DNA) drop in duplex thermal stability relative to the
butyl insertion, it appears that increased turnover is not the sole
basis for preferential enzyme discrimination toward the flexible
AONs in these examples. We further speculate that residual
activity is retained in mismatched sequences by virtue of this
characteristic alone (i.e., increased turnover), which now
becomes operative in segments of partial complementarity.
RNase H Assays. Fortunately, the destabilization introduced
into the AON by the acyclic linkers is largely outweighed by
the high stabilization7,9 afforded by the rigid 2′F-ANA backbone.
Indeed, when linked to a proven RNase H-competent analogue,
for example, 2′F-ANA,7-9 an acyclic insert in the middle of an
Linker Position. To examine the relevance of enzyme
turnover efficiency and of the actual cleavage event in other
contexts, we next prepared AONs with linker 4 at various
positions to determine whether optimal activity is dependent
upon the site of the insertion. Moreover, this strategy should
lend further credibility to the precise pattern and rate of cleavage
(15) Vandendriessche, F.; Augustyns, K.; Aerschot, A. V.; Busson, R.; Hoog-
martens, J.; Herdewijn, P. Tetrahedron 1993, 49, 7223-7238.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 657

A R T I C L E S
Mangos et al.
Figure 3. Gel (PAGE) showing linker-dependent human RNase HII-mediated hydrolysis patterns of duplexes. Assays (10 µL final volume) comprised 1
pmol of 5′-[32P]-target RNA and 3 pmol of test oligonucleotide in 60 mM Tris-HCl (pH 7.8, containing 2 mM dithiothreitol, 60 mM KCl, and 10 mM
MgCl2. Reactions were started by the addition of RNase H and carried out at 14-15 °C for 20 min. Lengths of the RNA fragments generated via enzyme
scission and corresponding position along the AON are indicated.
Table 2. Relative Rates of RNA Targeta Degradation among
Mixed-Sequence DNA and 2′F-ANA AON Constructs
that accompanies the movement of the linker along the 2′F-
ANA backbone (Table 1 and Supporting Information).
sequence typeb
k
rel (human)c
k
rel (E. coli)c
Close inspection of the autoradiogram in Figure 2 reveals
that the exact location of RNase H-executed primary cuts on
the RNA strand is difficult to measure under ambient temper-
ature for the homopolymers, which are already known to be
good substrates for the enzyme. As we were now interested to
see where the first cuts were occurring, this information was
instead extracted from assays conducted at the lower temper-
ature, under which enzyme activity is retarded just enough to
enable a qualitative comparison on the preferred cleavage modes
toward each substrate (Figure 3). At higher temperatures, the
pattern becomes less interpretable as it results from the
superimposition of multiple cleavages on a single target by the
enzyme (Figure 2).
We observe an overall enhancement in target degradation in
the order of 2′F-ANA-B10 (IX) > 2′F-ANA-B13 (X) > 2′F-
ANA (VII) > 2′F-ANA-B5 (VIII). Furthermore, all of the
butyl-containing AONs induce additional primary cuts at the
3′-end of the RNA except for 2′F-ANA-B5, which coinciden-
tally, is the only AON superseded in rate by all-2′F-ANA. As
such, the B5 and B13 homopolymeric substrates show large
differences in activation potencies, although their sequences are
virtually identical and equally thermostable, yet with opposite
directionalities with respect to the butyl site in the AON. Indeed,
the different activities of these two compositions contrast their
thermal and structural similarities, and suggest a minorsif not
absentsrole for the turnover effect. The diminished rate
enhancement in the AON with butyl at position 5 might
alternatively reflect the remote positioning of RNase H along
the substrate, which is known to bind near the 3′-end of the
AON in the hybrid duplex13 and so may be unaffected by the
butyl insertion. This does not explain, however, why there is a
reduction relative to 2′F-ANA, although an adverse effect on
entry
III
V
XIII
XIV
XV
XVI
DNA
DNA-c10
2′F-ANA
2′F-ANA-B10
2′F-ANA-C10
2′F-ANA-t10
4.6
3.3
1.0
3.5
0.9
1.6
8.8
7.1
1.0
1.4
1.3
2.6
a Target RNA: 5′-r(GGG AAA GAA AAA AUA UAA)-3′. b AON
sequences are abbreviated as follows: c10, C10 B10 and t10 denote
deoxycytidine, arabinofluorocytidine, butyl, or deoxythymidine residue in
place of the 10th nucleotide in the AON, respectively; complete description
is given in Table 1. c Assays were conducted at 22 °C; rates are normalized
according to the all-2′F-ANA AON (XIII).
the catalytic activity of the enzymeswhich initiates cleavage
upstream with respect to the binding along the hybridsmay be
accountable.
2′F-ANA vs DNA in Human and E. coli Isotypes. These
findings prompted us to explore the activating potential of our
purely artificial 2′F-ANA constructs directly to those of oligo-
nucleotides mainly containing deoxyribose residues in the
antisense strand. While a mismatching residue in either 2′F-
ANA or DNA mixed-base sequences imposes a negligible effect
on the overall enzyme hydrolytic efficiency as compared to the
parent AONs, a substantial acceleration in degradation rate is
readily achieved by sequences incorporating a modest degree
of flexibility in the antisense strand (Table 2). For example,
substitution of a central arabinofluoro residue in 2′F-ANA XIII
with a more flexible deoxy sugar (XVI) elevates the activity of
both RNase H isotypes. In the human system, an immediate
and pronounced increase in target degradation is likewise
apparent upon interchange of a deoxythymidine in the rigid 2′F-
ANA antisense for a butyl linker (AON XIV), which closes
9
658 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Oligonucleotide Flexibility and RNase H
A R T I C L E S
the efficiency gap between 2′F-ANA (krel 3.5) and DNA-derived
(krel 4.6) AON considerably.
In either case, we speculated that forcing the linker out of
the helix in this manner could potentially disrupt some of the
interactions between RNase H and the two strands by reducing
the number of stable contacts between the enzyme and the
duplex minor groove. Remarkably, enzyme assays with both
2′F-ANA-B-Lp and the analogous DNA-derived oligonucleo-
tides suggest that such a protrusion is only associated with a
minor energy penalty for enzyme docking and subsequent
processive activities. In fact, both the DNA and 2′F-ANA
sequences still considerably enhance RNA degradation (compare
oligomers XIII vs XVII and III vs VI, Table 1), which suggests
that flexibility in the antisense strand is important for effective
RNase H induction, irrespectiVe of whether the flexible linker
resides directly within or extends away from the helix axis.
The converse is true when a mismatching residue occupies a
central position within the antisense strand. For instance, the
mismatched nucleotide in DNA-c10 (V) and 2′F-ANA-C10
(XV) oligonucleotides should likewise occupy an external
position along the hybrid duplex (similar to “B-Lp” AONs VI
and XVII), as steric prohibition with the opposing RNA
nucleotide likely occurs to “bulge” the mismatch away from
the helix. However, if proper enzyme interaction with these
substrates necessitates subtle repositioning of the strands within
the AON:RNA complex, a high energy penalty associated with
this event could reduce the enzyme’s overall capacity to bind
to and cleave the mismatched substrates. This presumably
disables extensive processing in these duplexes as opposed to
an inherently flexible alternative. In the butyl-modified duplexes
where the linker loops out of the helix, enzyme activation is
conceivably retained as steric interactions are relieved via an
induced spatial rearrangement of the linker by the enzyme upon
substrate docking. It is especially noteworthy that a possible
transient delay in catalysis due to such repositioning of the
looping linker during the enzyme interaction may be responsible
for the reduced rates relative to DNA- (IV) and 2′F-ANA-B10
(XIV) substrates in which the linker resides parallel with the
helix and likely requires minimal distortion.
Ha-ras RNA. As the extent and rate of RNA cleavage should
also depend on the structure, length, and base composition of
the target polyribonucleotide, we next evaluated whether the
flexible AONs could accelerate cleavage in a highly structured,
physiologically relevant RNA species.
We particularly examined the specific reduction of target
RNA levels using an in vitro Ha-ras model system, especially
since hyperactivated forms of both point mutant and nonmutated
mammalian c-ras protooncogenes are frequently implicated in
various human cancers,16 thus making this a desirable therapeutic
target. As such, the linker-modified antisense oligonucleotides
were directed to hybridize with an internal 18-nucleotide region
in the c-ras mRNA encompassing the translation initiation site
of the ras gene and corresponding to residues -19 to +21 of
the full length transcript.
Even more intriguing is the finding that the preference for a
more flexible duplex substrate is a property displayed by the
bacterial (E. coli) enzyme as well, such that target strand
hydrolysis is accelerated in the presence of a single deoxy or
butyl insert within the parent AON (Table 2). It is noteworthy,
however, that the discriminatory properties of the two isotypes
display some subtle variations. Specifically, the bacterial
homologue shows greater tolerance for deoxythymidine-contain-
ing constructs over the butyl-containing 2′F-ANA strands, in
contrast to that observed with its human counterpart. The reasons
for this disparity are presently unclear but could be related to
variations in the placement of catalytically active residues
between the two isotypes. These may ultimately result in subtle
differences in the way these homologues interact with a
particular substrate type. Unfortunately, the lack of definitive
structural data regarding the physical dimensions of the class
II enzyme’s binding and catalytic cavities render this hypothesis
merely presumptive at present.
Effect of “Looping” Butyl Linker. As enzyme efficiency
may be related to the energetic cost of distorting the duplex
into a more catalytically reactive conformation, placing the
linker outside of the helix should enable closer scrutiny into
the dynamic nature of the RNase H-AON complex interaction.
As such, when the nonnucleotidic linker is shifted outside of
the helix by inserting a nucleotide complementary to the RNA
single-gap region (e.g., VI and XVII, Table 1), this gives an
oligonucleotide of 19 units in length with two nonameric
nucleotide segments that are capable of forming perfect
complexes with the RNA.
In this regard, the 2′F-ANA-B-Lp sequence (XVII, Table 1)
contains unifying elements of both the 2′F-ANA (XIII) and 2′F-
ANA-B10 (XIV) mixed-sequence oligonucleotides. These con-
sist of a localized flexible site in the center of the sequence
(similar to 2′F-ANA-B10) as well as the ability to fully hybridize
with the target RNA (similar to 2′F-ANA). For this sequence,
AON binding to the RNA is likely accompanied by a local
compression of the AON backbone at the extrahelical site to
project or “loop” the linker away from the duplex core. Such a
compensatory distortion of the helix may further be reconciled
through kinking in the overall conformation of the hybrid which
probably also occurs to maximize the number of residues that
form base pairs between this oligonucleotide and the RNA. This
is reflected in a modest gain in the thermal stability of this
sequence relative to that of 2′F-ANA-B10 (∆Tm ) 2 °C), which
although not as high as that of all-2′F-ANA, indicates that some
linker-induced motility of the strands surrounding the junction
may be necessary to relieve local intrastrand steric effects.
In contrast, DNA with a looping butyl linker (DNA-B-Lp,
VI) is incapable of promoting stable base pairing between butyl-
anchoring residues and the opposing RNA strand, as this
sequence exhibits the same thermal stability as when the linker
resides completely parallel with the helix axis (e.g., AON IV).
This is not entirely unexpected however, as the deoxynucleotides
themselves are considerably more flexible than 2′F-ANA
residues and so would be more likely to experience greater
intrahelical disorder around the backbone region flanking the
linker insertion.
Possible secondary structures of the c-ras target RNA were
extracted using the MFOLD program17 which estimates ∆G°
37
values for self-complementary duplex formation (available at
(16) (a) Chakraborty, A. K.; Cichutek, K.; Duesberg, P. H. Proc. Natl. Acad.
Sci. U.S.A. 1991, 88, 2217-2221. (b) Hua, V. Y.; Wang, W. K.; Duesberg,
P. H. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 9614-9619.
(17) (a) Zuker, M.; Mathews, D. H.; Turner, D. H. In RNA Biochemistry and
Biotechnology 11-43; Barciszewski, J., Clark, B. F. C., Eds.; Kluwer
Academic Publishers: Norwell, MA, 1999. (b) Mathews, D. H.; Sabina,
J.; Zuker, M.; Turner, D. H. J. Mol. Biol. 1999, 288, 911-940.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 659

A R T I C L E S
Mangos et al.
Figure 4. (A) Probable secondary structure adopted by 40mer Ha-ras RNA, predicted using the Turner-Zuker algorithm.17 Residues in blue denote the
region to which the AON binds. (B) Denaturing PAGE illustrating human RNase HII-mediated cleavage of 5′-32P-labeled Ha-ras RNA (1 pmol) duplexed
with mixed base AON sequences (5 pmol) at 37 °C. Lanes 1 and 2: 2′F-ANA-B10-XXII, 10 and 20 min; lanes 3 and 4: 2′F-ANA-XXI, 10 and 20 min;
lanes 5 and 6: DNA-B10-XIX, 10 and 20 min; lanes 7 and 8: DNA-XVIII, 10 and 20 min; lanes 9 and 10: DNA-c-10-XX, 10 and 20 min. The slowly
migrating band in lanes 3 and 4 is the 2′F-ANA:RNA duplex (Tm > 80 °C) which withstands the denaturing conditions of the gel electrophoresis and the
heating of the sample prior to being loaded onto the gel; see Table 1 for complete list of sequences. (C) Lengths of the RNA fragments generated via enzyme
scission and corresponding position along the AON. Bold arrows represent the major cleavage sites; the highlighted A residue in the RNA sequence denotes
the residue opposite the insertion position of the linker in the AON. Kinetic data (krel) of RNA cleavage is provided in Table 1.
http://bioinfo.math.rpi.edu/∼zukerm/rna/). Although the program
employed is based on parameters generated from the folding
of RNA under high ionic solution conditions (i.e., 1 M NaCl),
these calculations argue for the existence of a single predominant
species in which most of the target site (ca. 75%) is sequestered
in a stable 25-base hairpin structure (Figure 4A). This prediction
is supported by analysis of the UV thermal profile of the 40mer,
which indicates biphasic melting behavior likely due to structural
changes induced by the unfolding of the RNA or denaturation
of self-associated (i.e., intermolecular) RNA complexes. That
all of these characteristics are closely reproduced in a longer
ras-derived target spanning residues -24 to +26 of the full
length transcript, further corroborates these observations (data
not shown).
For DNA-based AONs within this series, an increase in
human RNase HII-mediated target degradation is not apparent
upon interchanging a deoxynucleotide residue in the parent
strand for a butanediol linker (i.e., XVIII and XIX; Table 1
and Figure 4) and appears to be consistent with prior work by
other groups.18
consequently a lesser amount of substrate duplex presented to
the enzyme for cleavage. Duplex heterogeneity seems even more
likely given that hybrid formation with DNA-B10-XIX produces
the least intrinsically stable duplex within the series (Table 1).
Consequently, greater competition for intramolecular folding
of the target as opposed to intermolecular duplexation could
conceivably mask any observable differences in cleavage rates.
However, this is not the case for the 2′F-ANA constructs.
As demonstrated throughout this work, substitution of the
arabinofluoronucleoside residue in 2′F-ANA with a more
flexible butanediol linker elevates the activity of RNase HII
considerably, ca. 23-fold over the parent compounds and
approaching the level of enzyme-induced degradation attainable
with the native oligodeoxynucleotides. Once again, these
properties are consistently reproduced when the 2′F-ANA-based
constructs are directed against a slightly longer c-ras 50mer
RNA (data not shown), indicating that a significant enhancement
(18) (a) Vorobjev, P. E.; Pyshnaya, I. A.; Pyshnyi, D. V.; Repkova, M. N.;
Venyaminova, A. G.; Zenkova, M. A.; Ivanova, E. M.; Scalfi-Happ, C.;
Seliger, H.; Bonora, G.; Zarytova, V. F. Bioorg. Khim. 2000, 26, 844-
851. (b) Vorobjev, P. E.; Pyshnaya, I. A.; Pyshnyi, D. V.; Venyaminova,
A. G.; Ivanova, E. M.; Zarytova, V. F.; Bonora, G.; Seliger, H. Antisense
Nucleic Acid Drug DeV. 2001, 11, 77-85.
Alternatively, this effect may be explained in part by a
decreased propensity for this AON to associate with the RNA,
resulting in a nonhomogeneous AON:RNA population and
(19) Toulme´, J.-J.; Frank, P.; Crouch, R. J. In Ribonucleases H; Crouch, R.,
Toulme´, J.-J., Eds.; INSERM: Paris, France 1998; p 147.
9
660 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Oligonucleotide Flexibility and RNase H
A R T I C L E S
in target hydrolysis may readily be achieved by the butyl-
modified AON, even when targeting a highly structured RNA
species.
We are presently undertaking further studies with these and
other acyclic analogues as well as extending our knowledge to
other RNase H isozymes. These endeavors should afford
valuable insight into the substrate requirements of the enzyme
and provide a versatile method in generating a new repertoire
of more potent antisense compounds.
Conclusions
As such, a balance between flexibility and rigidity in the AON
appears to be essential for enhancing targeted destruction of
the hybrid duplex by RNase H, provided that the geometrical
elements around the actual cleavage site still retain the necessary
requirements that have thus far been demonstratedsthat is,
Eastern sugar conformation and B-form7d,8c,9 character in the
AON strand. Indeed, 2′F-ANA-B constructs (e.g., XIV) repre-
sent the first examples of modified AON devoid of deoxyribose
sugars that elicit RNase H activity with comparable efficiency
to the native (DNA) systems. We speculate that the malleability
of the duplex backbone region surrounding the acyclic modi-
fications likely better accommodates the enzyme into the minor
groove4a,b,19 and may aid in more efficient hydrolysis during
the catalytic event, thereby giving rise to the observed rate
enhancements.
Acknowledgment. This work is dedicated to Professor Kelvin
K. Ogilvie on the occasion of his 60th birthday. We thank
R. T. Pon and S. Carriero for synthesis and purification of the
40mer Ha-ras RNA, respectively. M.J.D. and M.A.P. gratefully
acknowledge support from CIHR (Canada) and Anagenis, Inc.
Supporting Information Available: A table of calculated
extinction coefficients and mass spectral data (MALDI-TOF);
gels of DNA-B10 oligothymidylates, mixed-sequence 2′F-ANA-
B10; and relative turnovers of 2′F-ANA, dT18, and 2′F-ANA-
B10 oligothymidylates (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
JA025557O
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 661