Small interfering RNAs containing full 2¢-O-
`methylribonucleotide-modified sense strands display
`Argonaute2/eIF2C2-dependent activity
`
`BRYAN A. KRAYNACK and BRENDA F. BAKER
`ISIS Pharmaceuticals, Carlsbad Research Center, Carlsbad, California 92008, USA
`
`ABSTRACT
`
`RNA interference (RNAi) is a process by which short interfering RNAs (siRNAs) direct the degradation of complementary single-
`strand RNAs. In this study, we investigated the effects of full-strand phosphorothioate (PS) backbone and 2¢-O-methyl (2¢-OMe)
`sugar modifications on RNAi-mediated silencing. In contrast to previous reports, we have identified active siRNA duplexes
`containing full 2¢-OMe-modified sense strands that display comparable activity to the unmodified analog of similar sequence.
`The structure of these modified siRNAs is the predominant determinant of their activity, with sequence and backbone
`composition being secondary. We further show, by using biotin-tagged siRNAs and affinity-tagged hAgo2/eIF2C2, that activity
`of siRNA duplexes containing full 2¢-OMe substitutions in the sense strand is mediated by the RNA-induced silencing complex
`(RISC) and that strand-specific loading (or binding) to hAgo2 may be modulated through selective incorporation of these
`modifications.
`Keywords: 2¢-OMe RNA; anti-sense; RNA; small interfering
`
`INTRODUCTION
`
`Double-stranded short interfering RNAs (siRNAs) direct the
`degradation of complementary target RNAs in a process
`known as RNA interference (RNAi) (Fire et al. 1998;
`Hammond et al. 2001b; Hannon 2002; McManus and
`Sharp 2002). Identification of siRNAs as intermediates of
`the RNAi pathway has led to the demonstration that chemi-
`cally synthesized siRNAs are sufficient substrates for reduc-
`tion of target mRNA in mammalian systems (Caplen et al.
`2001; Elbashir et al. 2001a). Further application of medicinal
`chemistry to these synthetic oligoribonucleotides has in turn
`enabled usage and identification of modifications that can be
`incorporated into siRNAs for further mechanistic analysis
`(Manoharan 2004), as well as development of siRNAs with
`improved pharmacokinetic and pharmacodynamic proper-
`
`Reprint requests to: Brenda F. Baker, ISIS Pharmaceuticals, Carlsbad
`Research Center, 1896 Rutherford Road, Carlsbad, CA 92008, USA; e-mail:
`bbaker@isisph.com; fax: (760) 268-5035.
`Abbreviations: RNAi, RNA interference; siRNAs, short interfering RNAs;
`PS, phosphorothioate; 2¢-OMe, 2¢-O-methyl; 2¢-MOE, 2¢-O-2-methox-
`yethyl; RISC, RNA-induced silencing complex; ASO, anti-sense oligonu-
`cleotides; SAR, structure activity relationship; CDS, coding sequence;
`PAGE, polyacrylamide gel electrophoresis.
`Article published online ahead of print. Article and publication date are
`at http://www.rnajournal.org/cgi/doi/10.1261/rna.2150806.
`
`ties for in vivo applications (Layzer et al. 2004; Soutschek
`et al. 2004; de Fougerolles et al. 2005).
`Two modifications that have been extensively evaluated,
`for both anti-sense oligonucleotides (ASO) and siRNAs, are
`the phosphorothioate (PS) backbone modification (Eckstein
`2000) and the 2¢-O-methyl (2¢-OMe) sugar modification
`(Lubini et al. 1994; Braasch and Corey 2002; Grunweller
`et al. 2003; Harborth et al. 2003). The replacement of a
`nonbridging oxygen with a sulfur atom creates a PS linkage
`that may afford an increased resistance to nuclease degrada-
`tion and raise the in vivo bioavailability of the modified
`siRNA, as has been demonstrated previously for single-
`stranded ASOs that operate via an RNase H-based mecha-
`nism (Crooke 1999; Eckstein 2000; Geary et al. 2001). Incor-
`poration of the PS modification, however, also lowers ther-
`mal duplex stability for complementary sequences, promotes
`nonspecific protein binding, and—upon extensive incorpora-
`tion—may result in cytotoxicity (Amarzguioui et al. 2003;
`Harborth et al. 2003).
`The 2¢-OMe sugar modification, although more bulky
`than the 2¢-OH, preserves the ribose sugar pucker and
`retains the canonical right-handed A-form helical geometry
`(Lubini et al. 1994; Cummins et al. 1995), which is required
`for siRNA activity (Chiu and Rana 2003). As with the PS
`modification, this modification has also been shown to
`
`RNA (2006), 12:163–176. Published by Cold Spring Harbor Laboratory Press. Copyright ª 2006 RNA Society.
`
`163
`
`Alnylam Exh. 1060
`
`

`

`Kraynack and Baker
`
`increase the nuclease resistance of oligoribonucleotides
`(Lubini et al. 1994; Cummins et al. 1995) and siRNA
`duplexes (Chiu and Rana 2003; Czauderna et al. 2003).
`The effect of the PS and 2¢-OMe modifications on siRNA
`activity has been found to be dependent on both position
`and extent of incorporation. With respect to the latter, a
`significant number of reports in the literature have demon-
`strated that siRNA constructs containing full 2¢-OMe sub-
`stitutions in either sense, anti-sense, or both strands display
`little to no RNAi activity (Elbashir et al. 2001b; Amarzguioui
`et al. 2003; Braasch et al. 2003; Chiu and Rana 2003; Czau-
`derna et al. 2003). For example, an siRNA that had either or
`both strands 2¢-OMe modified was shown to completely lose
`RNAi activity against a luciferase mRNA in Drosophila cell
`lysates (Elbashir et al. 2001b). In agreement with this,
`Braasch and coworkers (Braasch et al. 2003) failed to observe
`inhibition of hCAV expression in HeLa cells with fully sub-
`stituted 2¢-OMe RNA in one or both strands of the duplex.
`In further studies using 19-bp 3¢-dTdT duplexes in HeLa
`cells, Chiu and Rana (2003) found that 2¢-OMe-modified
`siRNA greatly diminished EGFP silencing, achieving 16%–
`25% RNAi activity when either the anti-sense or the sense
`strand was fully modified, and no activity was seen when
`both strands were modified, where 93% activity was achieved
`with the respective unmodified siRNA. Czauderna and co-
`workers also demonstrated that blunt-end 21-bp duplexes
`targeting human PTEN with either one or both strands fully
`substituted with 2¢-OMe were relatively inactive compared
`to the unmodified parent construct in their mammalian cell
`culture system (Czauderna et al. 2003).
`In contrast to prior observations, this report demon-
`strates that siRNA duplexes containing full 2¢-O-methyl-
`modified sense strands, targeted to sites within the PTEN
`mRNA, mediate mRNA target reduction with similar effi-
`cacy and potency as the unmodified parent siRNA in HeLa
`cells. To determine the basis of the dif-
`ferences between our observations and
`those of other investigators, we evalu-
`ated the effects of construct design, tar-
`get site sequence, and backbone com-
`position on 2¢-OMe-modified siRNA
`activity in cell culture.
`We also sought to determine (or con-
`firm)
`the underlying mechanism of
`target mRNA reduction by the 2¢-
`OMe-modified siRNA complexes. Arg-
`onaute2/eIF2C2 (hAgo2) has been iden-
`tified as an endonuclease activity as-
`sociated with the RNA-induced silen-
`cing complex (RISC) (Hammond et al.
`2000, 2001a), which provides the mini-
`mal scaffold for catalytic RISC cleavage
`(Liu et al. 2004; Meister et al. 2004;
`Rand et al. 2004; Song et al. 2004;
`Rivas et al. 2005). We found by using
`
`Site
`
`I
`IB
`II
`IIB
`III
`IIIB
`IV
`IVB
`
`biotin-tagged siRNAs, an affinity-tagged hAgo2 overexpres-
`sion vector, an RNase H-based ASO knock-down assay, and
`an in vitro hAgo2-based RNA cleavage assay that hAgo2 is a
`mediator of target mRNA knock-down in cell culture and
`RNA substrate cleavage in solution for this class of modified
`siRNAs.
`
`RESULTS
`
`Target site selection and siRNA design
`
`We selected four active siRNA sites (Table 1, Sites I–IV),
`identified previously from a screen of 36 siRNAs that
`target the endogenous tumor suppressor PTEN mRNA
`(Vickers et al. 2003), to examine the effects of full-strand
`sugar-phosphate backbone modifications on siRNA act-
`ivity. Full-strand PS and 2¢-OMe modifications were
`evaluated in a combinatorial
`fashion at each site,
`in
`conjunction with two types of siRNA construct designs,
`the canonical 19-bp duplex with 3¢-dTdT overhangs (19-
`bp 3¢-dTdT) and a 20-bp duplex with blunt ends (blunt-
`end 20 bp).
`
`Activity of 19-bp 3¢-dTdT siRNA duplexes with full
`2¢-O-methyl sense strands is sequence dependent
`In our initial analysis, we evaluated combinations of a PS-
`modified anti-sense strand, a 2¢-OMe sense strand, and the
`respective unmodified strands in the context of the canon-
`ical 19-bp 3¢-dTdT overhang construct. Each of the unmod-
`ified siRNA molecules (I_1, II_1, III_1, and IV_1) directed
`against the four PTEN target sites displayed a dose-depen-
`dent reduction in PTEN mRNA expression levels (Fig. 1).
`The siRNA duplex targeting Site III (III_1) was the most
`potent with an IC50 of 3 nM. Constructs targeting Sites I
`
`Start
`
`1505
`1504
`1541
`1541
`1855
`1855
`2097
`2096
`
`Sequence 5¢–3¢ (sense mRNA)
`
`A.G.U.A.A.G.G.A.C.C.A.G.A.G.A.C.A.A.A
`A.A.G.U.A.A.G.G.A.C.C.A.G.A.G.A.C.A.A.A
`C.A.G.U.C.A.G.A.G.G.C.G.C.U.A.U.G.U.G
`C.A.G.U.C.A.G.A.G.G.C.G.C.U.A.U.G.U.G.U
`G.G.G.U.A.A.A.U.A.C.A.U.U.C.U.U.C.A.U
`G.G.G.U.A.A.A.U.A.C.A.U.U.C.U.U.C.A.U.A
`C.A.A.A.U.C.C.A.G.A.G.G.C.U.A.G.C.A.G
`U.C.A.A.A.U.C.C.A.G.A.G.G.C.U.A.G.C.A.G
`
`TABLE 1. Location of PTEN target sites
`
`A comparative analysis of 36 siRNA sequences identified as active siRNA sites within
`human phosphatase and tensin homolog PTEN (Accession No. U92436) message (Vickers
`et al. 2003). The location and sequence of the active target sites used in this study are shown
`above. The PTEN coding sequence (CDS) starts at 1035 and terminates at 2246. The sites
`shown in gray were used for generating 19-mers with 2-nt deoxythymidine overhangs at the
`3¢–end. The sites shown in white were used for the generation of 20-mer blunt siRNAs. The
`sequence of the antisense strand is an exact complement to the sense strand shown in the
`table.
`
`164
`
`RNA, Vol. 12, No. 1
`
`

`

`2¢-OMe-mediated RNAi
`
`FIGURE 1. Activity of 19-bp 3¢-dTdT siRNA duplexes with full 2¢-O-methyl sense strands is sequence-dependent. Dose response analyses of 19-
`bp 3¢-dTdT siRNAs targeting Site I (A), Site II (B), Site III (C), and Site IV (D) of the PTEN mRNA in HeLa cells. The sequence and structure of
`siRNAs are shown to the left of each graph and are unmodified oligoribonucleotides unless indicated otherwise. ( ) A phosphorothioate (PS)
`modification, (underline) 2¢-OMe-modified nucleotides, (dT) 2¢-deoxythymidine. HeLa cells were treated with siRNAs at 0.6, 3, 15, and 75 nM, as
`detailed in Materials and Methods. PTEN mRNA levels were determined relative to c-raf mRNA levels by real-time RT-PCR (TaqMan). Results
`shown represent the percentage of untreated control expression (UTC). Each bar represents the mean of three biological replicates (6standard
`deviation).
`
`www.rnajournal.org
`
`165
`
`

`

`Kraynack and Baker
`
`and II (I_1 and II_1) were less potent, yet displayed sig-
`nificant target reduction with IC50s between 9 and 10 nM.
`The least potent siRNA was directed against the Site IV
`(IV_1) with an IC50 of 22 nM. The efficacy of all four
`sites was similar, achieving 70%–80% reduction of target
`in each case.
`In comparison, the 19-bp 3¢-dTdT duplexes containing
`a fully modified 2¢-OMe sense strand designed against Site
`I (Fig. 1A, siRNAs I_3 and I_4), and Site II (Fig. 1B,
`siRNAs II_3 and II_4) showed no reduction in PTEN
`mRNA levels when tested in combination with either a
`PO or PS anti-sense strand. At Site III, however, the
`duplex consisting of a PO anti-sense strand duplexed to
`a fully modified PO 2¢-OMe sense strand showed efficient
`reduction of PTEN mRNA expression levels with an IC50
`of 9 nM (Fig. 1C, III_3). The corresponding PO:PS_2¢-
`OMe duplex was slightly more potent with an IC50 of 6
`nM (data not shown). In addition, we noted that if a
`modified PS anti-sense strand was inactive (Fig. 1C,
`III_2), activity could not be restored by incorporating a
`fully modified 2¢-OMe sense strand into the duplex (Fig.
`1C, III_4). Constructs containing a full 2¢-OMe sense
`strand that targeted Site IV displayed very limited activity
`when duplexed to the PS anti-sense strand (Fig. 1D, cf.
`IV_2 and IV_4).
`Taken together, these results suggest that a fully modified
`2¢-OMe sense strand can elicit an effective siRNA response
`with similar potency and efficacy as the unmodified parent
`duplex. However, target mRNA reduction with fully mod-
`ified 2¢-OMe sense strands is impacted by strong sequence-
`dependent effects as only one active site out of the four
`
`examined was able to reduce target RNA in the context of
`the 19-bp 3¢-dTdT construct design (Fig. 1).
`
`Activity of siRNAs with full 2¢-O-methyl sense strands
`is dependent on construct design
`
`We next investigated the effects of construct design on
`siRNAs with full 2¢-OMe-modified sense
`activity of
`strands. Blunt-end 20-bp constructs were designed
`against the PTEN Sites I, II, III, and IV (Table 1), and
`their activity was compared to the analogous 19-bp 3¢-
`dTdT constructs by dose response analysis in HeLa cells.
`In contrast to the 19-bp 3¢-dTdT overhang constructs,
`target mRNA reduction was observed at all four sites
`with the blunt-end 20-bp duplexes containing a full 2¢-
`OMe sense strand (Fig. 2). Constructs IB_2, IIB_2, and
`IIIB_2 achieved 70%–75% PTEN mRNA reduction at
`the 75 nM dose with IC50s of 10, 5, and 14 nM,
`respectively. Activity of the blunt-end construct at Site
`IV (IVB_2) displayed 50% reduction at the highest
`dose tested (75 nM). This relatively low level of activity
`is in accordance with the activity of the unmodified
`siRNA at Site IV, which was also the least active in its
`class. The 19-bp 3¢-dTdT construct corresponding to
`Site III remained the most potent and efficacious design
`with an IC50 of 3 nM and 78% total reduction of target
`RNA.
`These data suggest that activity of siRNAs with fully
`modified 2¢-OMe sense strands is dependent on construct
`design. Furthermore, the blunt-end 20-bp design appears to
`be more broadly applicable when incorporating a full 2¢-
`
`FIGURE 2. Activity of siRNAs with full 2¢-O-methyl sense strands is dependent on construct design. Dose response analysis of two types of siRNA
`constructs (19-bp 3¢-dTdT vs. blunt-end 20 bp) targeting each of the four PTEN mRNA target sites in HeLa cells. The construct structure and
`sequence are shown to the left of each graph and are unmodified oligoribonucleotides unless indicated otherwise. (Underline) 2¢-OMe-modified
`nucleotides, (dT) 2¢-deoxythymidine residues. HeLa cells were dosed at 0.6–75 nM with siRNA, as described in Materials and Methods. The results
`shown represent the percentage of untreated control expression (UTC). Each bar represents the mean of three biological replicates (6standard
`deviation).
`
`166
`
`RNA, Vol. 12, No. 1
`
`

`

`OMe sense strand, as it has a broader range of activity
`across the four sites in comparison to the 19-bp 3¢-dTdT
`overhang constructs.
`
`Evaluation of blunt-end 20-bp siRNA duplexes with full
`2¢-O-methyl and PS backbone modifications
`Having established that blunt-end 20-bp constructs may be
`the optimal design for siRNAs with 2¢-OMe-modified sense
`strands, we next evaluated the effects of the full 2¢-OMe and
`PS backbone modifications on RNAi activity in cell culture
`based on this construct design (Fig. 3).
`Anti-sense strands were designed against the PTEN Sites
`I and III as either an unmodified PO strand or fully mod-
`ified PS strand. Sense strands were duplexed to the anti-
`sense strands as an unmodified PO, a PO with full 2¢-OMe
`substitutions, or a fully modified PS with full 2¢-OMe
`modifications (Fig. 3A,B). At PTEN Site I, the parent
`(IB_1) achieved 75% reduction of PTEN mRNA with an
`IC50 of 11 nM. Modified duplexes directed to this site
`achieved 60%–65% reduction of PTEN mRNA when either
`a PO-2¢-OMe or a PS-2¢-OMe sense strand was duplexed to
`a PO anti-sense strand (IB_2 and IB_3, respectively). siR-
`NAs containing either a PO-2¢-OMe (77% reduction) or a
`PS-2¢-OMe (79% reduction) sense strand duplexed to a PS-
`modified anti-sense strand were as efficacious as the parent
`compound (IB_4, 75%). The greatest potency was observed
`with the PS-2¢-OMe sense strand duplexed to a PO anti-
`sense strand (IB_3; IC50 of 7 nM) or a PS anti-sense strand
`(IB_6; IC50 of 10 nM).
`The unmodified blunt-end 20-bp duplex directed against
`Site III (IIIB_1) achieved a similar efficacy (75% reduction)
`as seen with the unmodified parent at Site I (IB_1) but was
`more potent with an IC50 of 3 nM (Fig. 3B). The modified
`duplexes directed against Site III showed a 65%–75%
`reduction of PTEN mRNA when a PO anti-sense strand
`was duplexed to either a PO-2¢-OMe (IIIB_2) or a PS-2¢-
`OMe sense strand (IIIB_3). Modified duplexes comprised
`of a PS anti-sense strand and PO sense strand (IIIB_4)
`showed only a modest 25% reduction in target mRNA.
`Activity increased to 50% and 60% reduction at the 75
`nM dose when the PS-modified anti-sense strand was
`duplexed to either a PO-2¢-OMe (IIIB_5) or a PS-2¢-OMe
`sense strand (IIIB_6), respectively. Overall, siRNAs contain-
`ing a PS anti-sense strand consistently displayed lower
`activity at Site III in comparison to Site II, independent of
`construct design.
`To complete evaluation of these modifications in the
`context of the blunt-end 20-bp duplex, we also evaluated
`constructs to Site I with full 2¢-OMe modifications in the
`anti-sense strand, or both the anti-sense strand and sense
`strand (Fig. 3C). Consistent with results reported elsewhere
`(Elbashir et al. 2001b; Amarzguioui et al. 2003; Braasch et
`al. 2003; Chiu and Rana 2003; Czauderna et al. 2003),
`duplexes containing either a PO-2¢-OMe anti-sense strand
`
`2¢-OMe-mediated RNAi
`
`(Ias_1, Ias_2, Ias_5, and Ias_6) or a PS-2¢-OMe anti-sense
`strand (Ias_3, Ias_4, Ias_7, and Ias_8) were inactive and
`failed to reduce PTEN mRNA.
`Our results indicate that 20-bp blunt-end duplexes com-
`prised of fully modified 2¢-OMe sense strands can mediate
`target mRNA reduction, and that complete substitutions
`with phosphorothioates in either or both strands is toler-
`ated. As expected, duplexes containing a full 2¢-OMe-mod-
`ified sense strand displayed improved resistance against
`serum nucleases (Czauderna et al. 2003), as did PS-mod-
`ified strands, when compared to unmodified duplexes (data
`not shown).
`Taken together, these data suggest that activity of the
`modified siRNAs with a full 2¢-OMe sense strand is pre-
`dominantly dependent on construct design as suggested by
`the activity seen with the blunt-end 20 bp across the four
`PTEN sites (cf. Figs. 2 and 3). Sequence and backbone com-
`position appear to be secondary determinants of activity.
`
`Modified siRNA activity is attenuated by ASO-mediated
`knock-down of hAgo2
`
`We next sought to address whether the fully modified
`siRNA duplexes containing 2¢-OMe sense strands function
`through the RISC complex. As a first measure, we took
`advantage of the RNase H-based anti-sense oligodeoxynu-
`cleotide (ASO) technology to ascertain hAgo2’s role in 2¢-
`OMe-modified siRNA-mediated target mRNA reduction
`(Bennett and Cowsert 1999; Ravichandran et al. 2004). In
`this assay, mRNA expression levels of hAgo2 were first
`reduced by treatment of cells with a complementary
`RNase H-based ASO (ISIS 136764). ASO-treated cells
`were incubated for 48 h to allow for hAgo2 protein to be
`eliminated through normal degradative processes, after
`which cells were transfected with siRNAs to determine the
`effect of hAgo2 knock-down on their activity.
`The hAgo2 knock-down assay was validated by perform-
`ing a dose response analysis of the unmodified 19-bp 3¢-
`dTdT siRNA to Site I (I_1). HeLa cells were treated with
`either an ASO targeting hAgo2 (ISIS 136764) or a random
`control ASO (ISIS 129686), prior to siRNA administration,
`as described in Materials and Methods. The unmodified
`siRNA (I_1) to Site I displayed a dose-dependent reduction
`of target mRNA achieving 80%–90% knock-down follow-
`ing treatment with the random control ASO (ISIS 129686),
`while hAgo2 levels remained unchanged (Fig. 4A, lanes 2–
`5). Treatment with an ASO targeting hAgo2 (ISIS 136764)
`resulted in 65%–70% knock-down of hAgo2 mRNA and
`a concomitant attenuation of siRNA (I_1) activity (Fig. 4A,
`lanes 7–10). Attenuation of siRNA activity was less pro-
`nounced at the higher dose points, suggesting that, in the
`presence of excess siRNA and incomplete hAgo2 knock-
`down, sufficient RISC complexes form to result in target
`PTEN mRNA reduction (70%, lane 10), but to a lesser
`extent
`than that observed in the control
`treated cells
`
`www.rnajournal.org
`
`167
`
`

`

`Kraynack and Baker
`
`FIGURE 3. Blunt-end 20-bp siRNAs with full 2¢-O-methyl sense strands and full PS backbone modifications mediate target mRNA reduction. Dose
`response analyses of modified blunt-end 20-bp siRNAs comprised of full 2¢-OMe-modified sense strands to Site I (A), full 2¢-OMe-modified sense
`strands to Site III (B), and full 2¢-OMe-modified anti-sense strands to Site I (C). The siRNA constructs are shown to the left of each graph and are
`unmodified oligoribonucleotides unless indicated otherwise. ( ) A PS modification, (underline) 2¢-OMe-modified nucleotides. The results shown
`represent the percentage of untreated control expression (UTC). Each bar represents the mean of three biological replicates (6standard deviation).
`
`168
`
`RNA, Vol. 12, No. 1
`
`

`

`2¢-OMe-mediated RNAi
`
`FIGURE 4. Modified siRNA activity is attenuated by knock-down of hAgo2. (A) Validation of ASO knock-down assay in HeLa cells. siRNA (I_1)
`was transfected into HeLa cells at a dose of 0.6–75 nM post-treatment with either the control ASO (ISIS 129686) or the ASO specific for hAgo2
`(ISIS 136764). (B) Effect of hAgo2 knock-down on activity of siRNAs containing a full 2¢-OMe-modified sense strand. Modified and unmodified
`siRNAs targeting PTEN Site III were transfected (75 nM) into HeLa cells previously treated with ISIS 129686 or ISIS 136764. PTEN and hAgo2
`mRNA levels were determined relative to the cRAF mRNA level by real-time RT-PCR (TaqMan). The results shown represent the percentage of
`untreated control expression (UTC). Each bar represents the mean of three biological replicates (6standard deviation). The control ASO (ISIS
`129686) is a random sequence that has no significant homology to the human genome.
`
`(92%, lane 5). These data indicate that hAgo2 is a rate-
`determining step for siRNA-mediated target mRNA reduc-
`tion, and that the RNase H-mediated hAgo2 knock-down
`assay provides a useful tool for examining the mechanistic
`basis of activity for modified siRNA constructs.
`We next asked whether full 2¢-OMe-modified siRNA
`constructs require hAgo2 for target mRNA reduction.
`Four constructs targeted to Site III (III_1, III_3, IIIB_1,
`and IIIB_2) were transfected into HeLa cells following
`treatment with either the random control ASO (ISIS
`129686) or the hAgo2-specific ASO (ISIS 136764). As
`shown in Figure 4B, both the unmodified 19-bp 3¢-dTdT
`(III_1) and the 20-bp blunt-end (IIIB_1) constructs dis-
`played a loss of activity upon knock-down of hAgo2 levels.
`These data are in agreement with recent reports indicating
`that hAgo2 is the catalytic engine responsible for RNAi-
`mediated activity (Liu et al. 2004; Meister et al. 2004; Rivas
`et al. 2005).
`Loss of activity was also observed with the fully modified
`2¢-OMe counterparts (III_3 and IIIB_2) with ASO-mediated
`knock-down of hAgo2 (Fig. 4B). This result suggests that
`fully modified 2¢-OMe constructs can enter the RNAi path-
`way and, furthermore, reduction of target mRNA is depen-
`dent on the abundance of hAgo2. The effect of hAgo2 knock-
`down on siRNA activity was not dependent on siRNA
`sequence (cf. Fig. 4, A and B) or construct design (Fig. 4B).
`
`Interaction of modified siRNAs with overexpressed
`hAgo2
`
`Continuing our mechanistic analysis of siRNAs containing
`fully modified strands, we next examined the binding prop-
`erties of these duplexes with the RISC complex, specifically
`hAgo2. For this purpose, we generated a construct that
`expressed hAgo2 with an N-terminal HA epitope (pE2-
`N_HA). Duplexed siRNAs that contained a biotin tag on
`the 3¢-terminus of either the anti-sense or the sense strand
`were then transfected into HeLa cells transiently transfected
`with the pE2-N_HA construct. The biotinylated strand was
`captured from total cell lysates under nondenaturing con-
`ditions by affinity purification, and then probed for the
`presence of HA-hAgo2 by Western blot. a-HA monoclonal
`antibodies recognized a single protein in crude HeLa
`extracts following transfection with the plasmid pE2-
`N_HA (Fig. 5A).
`In the first analysis, we compared the binding ability of
`unmodified siRNAs to those containing a full 2¢-OMe-mod-
`ified sense strand (Fig. 5B,C). HA-hAgo2 was not isolated from
`cells transfected with siRNAs lacking a biotin residue (Fig. 5B,C,
`lanes 2,7) or the vector (pE2-N_HA) alone (data not shown).
`HA-hAgo2 was isolated, however, when the anti-sense strand
`of the duplex contained a 3¢ biotin moiety, regardless of the
`sense strand makeup. In other words, the strand containing
`
`www.rnajournal.org
`
`169
`
`

`

`Kraynack and Baker
`
`FIGURE 5. Modified siRNAs interact with overexpressed hAgo2. (A) Characterization of pE2-N_HA. Total HeLa lysates prepared from untreated
`cells (lane 1) and cells expressing HA-hAgo2 (lane 2) were separated by SDS-PAGE and immunoblotted with anti-HA antibodies. The migration of
`the molecular mass standards is shown on the left. (B) Western blot analysis of biotin-siRNA samples isolated from HeLa cells expressing HA-
`hAgo2. Biotin-tagged strands were isolated from total cell lysates, as described in Materials and Methods. Isolates were then subjected to SDS-
`PAGE and Western blotted with anti-HA antibody. (C) Structure and sequence of siRNA constructs included in analysis. ( ) A PS modification,
`(underline) 2¢-OMe-modified nucleotides, (Bi) biotin residue.
`
`unmodified ribose residues was bound to hAgo2, indicating
`that an anti-sense strand duplexed to a full 2¢-OMe sense
`strand can be loaded into RISC. Surprisingly, we found
`substantially less hAgo2 isolated when the biotin moiety
`was located on the 3¢-end of either unmodified or modified
`sense strands. In addition, we found that a PS-modified AS
`strand bound more hAgo2 than the corresponding PO
`strand. This is likely a reflection of the enhanced protein-
`binding properties of a PS modification (Eckstein 2000).
`Recent reports in the literature have shown that which-
`ever strand is most easily unwound in a 5¢–3¢ direction will
`be preferentially loaded into RISC (Khvorova et al. 2003;
`Schwarz et al. 2003). The thermodynamic profile of the
`siRNA targeting Site I is similar for the 5¢-end of the anti-
`sense strand and the 5¢-end of the sense strand, as each end
`contains four A:U bp and 1 G:C bp. This would have
`predicted that either strand could load equally well into
`the RISC complex. However, as shown in Figure 5B, the
`anti-sense strand is preferentially loaded. It is of interest to
`note that the sense strand is extremely purine rich (80%),
`and by default the anti-sense strand is pyrimidine rich.
`In a second analysis, we investigated siRNAs that con-
`tained a full 2¢-OMe anti-sense strand to determine if the
`lack of activity observed in cell culture results from a
`decrease in binding to hAgo2 (Fig. 6). As demonstrated in
`the previous analysis, HA-hAgo2 was not isolated with
`siRNAs lacking a biotin residue (lanes 1,4), but was isolated
`when the RNA anti-sense strand (either PO or PS) of the
`duplex contained a 3¢ biotin moiety (lanes 2,5). HA-hAgo2
`was also isolated when a full 2¢-OMe anti-sense strand
`
`170
`
`RNA, Vol. 12, No. 1
`
`(either PO or PS) of the duplex contained a 3¢ biotin moiety,
`although in substantially lower amounts compared to the
`respective parent siRNA (cf. lanes 2 and 3 and lanes 5 and
`
`FIGURE 6. Reduced association of full 2¢-OMe anti-sense strands
`with overexpressed hAgo2. (A) Western blot analysis of biotin-
`siRNA isolated from HeLa cells expressing HA-hAgo2. Samples were
`subjected to SDS-PAGE and Western blotted with anti-HA antibody,
`as described in Materials and Methods. (B) Structure and sequence of
`siRNA constructs included in analysis. ( ) A PS modification, (under-
`line) 2¢-OMe-modified nucleotides, (Bi) biotin residue.
`
`

`

`6). Therefore, as observed with the full 2¢-OMe sense strand,
`the full 2¢-OMe anti-sense strand is not effectively loaded
`and/or is not as tightly associated with the HA-hAgo2/RISC
`complex. In other words, the bulky methyl group may
`interfere with loading or binding to hAgo2.
`
`same sequence as the expected product (Marker_23, lane
`2). Similarly, incubation of the 5¢-end-labeled RNA substrate
`with HA-hAgo2 RISC complexes purified from IB_4- and
`IB_5-treated cells also yielded the expected 17-nt cleavage
`product (lanes 6,7), as confirmed by a 17-nt synthetic RNA
`
`2¢-OMe-mediated RNAi
`
`Modified siRNAs enter into the
`hAgo2/RISC complex to mediate
`target RNA cleavage
`
`We used an hAgo2-based in vitro RNA
`cleavage assay to confirm that initial cleav-
`age of the targeted RNA by the modified
`2¢-OMe duplexes is a direct consequence
`of hAgo2/RISC activity (Martinez et al.
`2002; Liu et al. 2004; Song et al. 2004;
`Rivas et al. 2005). In brief, HeLa cells
`were first transfected with pE2-N_HA,
`to allow for expression of HA-hAgo2,
`and then 24 h later were transfected
`with the siRNAs. RISC complexes were
`allowed to assemble endogenously for an
`additional 24 h, after which cells were
`lysed and the HA-hAgo2 RISC com-
`plexes were immunoprecipitated with
`the anti-HA antibody. Following isola-
`tion of the hAgo2 RISC complex, an
`RNA cleavage assay was performed
`using a 40-nt 32P-5¢-labeled RNA sub-
`strate corresponding to the region con-
`taining Site I (Fig. 7A).
`Constructs containing a PS anti-sense
`and a PO sense strand (IB_4) or a PS
`and a PO_2¢-OMe
`antisense
`sense
`strand (IB_5) were selected for the anal-
`ysis (Fig. 7B). To confirm the sequence
`specificity of RNA substrate cleavage, we
`introduced a pair of 19-bp 3¢-dTdT con-
`trol siRNAs that were complementary to
`a different region of the RNA substrate
`and consequently expected to elicit cleav-
`age at a different site.
`As shown by PAGE analysis (Fig. 7C),
`incubation of the 5¢-end-labeled RNA
`substrate with the HA-hAgo2 RISC com-
`plex isolated from cells treated with
`either the unmodified control siRNA
`(Con_1, lane 3) or control siRNA con-
`taining a fully modified 2¢-OMe sense
`strand (Con_2,
`lane 4) yielded the
`expected 23-nt product, where cleavage
`occurred between nucleotides 10 and 11
`from the 5¢-end of the control anti-sense
`strand. This assignment was verified with
`a 23-nt synthetic RNA marker of the
`
`FIGURE 7. Modified siRNAs enter into the hAgo2/RISC complex to mediate target RNA
`cleavage. (A) A 5¢-[P32] end-labeled 40-mer substrate containing Site I was incubated with
`immunopurified RISC complexes, as described in Materials and Methods. The sequence of
`siRNA constructs and lane assignments are shown in B. The sequence of each RNA
`marker corresponds to the expected cleavage product of 23 nt (control site) or 17 nt
`(Site I). ( ) A phosphorothioate modification, (underline) 2¢-OMe-modified nucleotides.
`(C) Cleavage products were resolved on a 12% polyacrylamide 7 M urea gel. The 5¢
`cleavage products are indicated by arrows. Sequence identity was assigned according to the
`control digestions of the substrate by RNase T1, base (OH) hydrolysis, and synthetic RNA
`markers.
`
`www.rnajournal.org
`
`171
`
`

`

`Kraynack and Baker
`
`marker of equivalent sequence (Marker_17, lane 5). In this
`assay, we also noted that the HA-hAgo2 RISC complex
`isolated from cells treated with siRNA duplexes containing
`a full 2¢-OMe anti-sense strand (Ias_3) were inactive, show-
`ing no cleavage products (lane 8). These results indicate that
`modified siRNAs with a full 2¢-OMe sense strand are capable
`of cleaving mRNA in a sequence-dependent manner and that
`this activity is mediated by hAgo2.
`
`DISCUSSION
`
`Using a combination of cell culture and biochemical assays,
`we investigated the sequence and structure activity relation-
`ships of siRNAs containing full-strand PS backbones and
`2¢-OMe sugar modifications in two types of constructs, the
`canonical 19-bp RNA duplex with 3¢-dTdT overhangs and a
`20-bp RNA duplex with blunt ends. In contrast to previous
`reports, we found that siRNAs with a fully modified 2¢-
`OMe sense strand can reduce target mRNA levels with
`similar efficacy and potency as the unmodified parent
`duplexes.
`To better understand the mechanism of activity for si-
`RNAs containing fully modified 2¢-OMe sense strands, we
`sought to determine if the mRNA knock-down seen with
`modified siRNAs was dependent on the RISC complex, as
`would be expected for an siRNA-mediated mechanism. We,
`and others, have found that hAgo2 is required for siRNA-
`mediated knock-down (Liu et al. 2004; Meister et al. 2004;
`Rivas et al. 2005), as inhibition of hAgo2 expression, by
`ASO/RNase H-mediated targeting, attenuated siRNA activ-
`ity. The requirement for endogenous hAgo2 was not depen-
`dent