OLIGONUCLEOTIDES 18:187–200 (2008)
`© Mary Ann Liebert, Inc.
`DOI: 10.1089/oli.2008.0123
`
`Chemical Modification Patterns Compatible with High Potency
`Dicer-Substrate Small Interfering RNAs
`
`Michael A. Collingwood,1 Scott D. Rose,1 Lingyan Huang,1 Chris Hillier,1 Mohammad Amarzguioui,2,3
`Merete T. Wiiger,2 Harris S. Soifer,4 John J. Rossi,5 and Mark A. Behlke1
`
`Dicer-substrate small interfering RNAs (DsiRNAs) are synthetic RNA duplexes that are processed by Dicer into
`21-mer species and show improved potency as triggers of RNA interference, particularly when used at low dose.
`Chemical modification patterns that are compatible with high potency 21-mer small interfering RNAs have been
`reported by several groups. However, modification patterns have not been studied for Dicer-substrate duplexes.
`We therefore synthesized a series of chemically modified 27-mer DsiRNAs and correlated modification pat-
`terns with functional potency. Some modification patterns profoundly reduced function although other patterns
`maintained high potency. Effects of sequence context were observed, where the relative potency of modification
`patterns varied between sites. A modification pattern involving alternating 2′-O-methyl RNA bases was devel-
`oped that generally retains high potency when tested in different sites in different genes, evades activation of the
`innate immune system, and improves stability in serum.
`
`Introduction
`
`RNA interference (RNAi) is a natural pathway in
`
`mammalian cells where double-stranded RNAs (dsR-
`NAs) suppress expression of genes with complementary
`sequence (Hannon and Rossi, 2004; Meister and Tuschl,
`2004). Long dsRNAs are processed by the RNase-III class
`nuclease Dicer into 21–23 bp products having 2-base 3′-over-
`hangs and 5′-phosphates (Bernstein et al., 2001). These spe-
`cies, called small interfering RNAs (siRNAs), are the actual
`molecular triggers of RNAi and direct target specificity of
`the RNA-induced silencing complex (RISC) (Elbashir et al.,
`2001). Synthetic 21-mer RNA duplexes can be introduced
`into cells and will function as mimics of the natural siRNAs
`that result from Dicer processing. Different RNA structures
`can also be used to suppress gene expression via RNAi.
`For example, short hairpin RNAs that have a 19–21 base
`duplex domain and a 4–9 base loop can be effective trig-
`gers of RNAi. Another approach is to use longer synthetic
`linear RNA duplexes that are substrates for Dicer. These
`longer duplexes (typically 25–30 bp length) are processed
`by Dicer into 21-mer siRNAs, enter RISC, and trigger RNAi.
`Synthetic Dicer-substrate RNAs can have increased potency
`
`when compared with 21-mer duplexes, possibly owing to
`effects relating to linkage of Dicer processing and RISC
`loading (Kim et al., 2005).
`SiRNAs are routinely used today as an experimental tool
`to suppress gene expression in vitro and many thousands
`of research reports have been published using RNAi in cell
`culture. SiRNAs are also being used in vivo for investiga-
`tional purposes and are under development as therapeu-
`tics (Behlke, 2006). To realize the full potential of siRNAs
`in vivo, it may be necessary to chemically modify the RNA
`duplexes to improve nuclease stability, alter pharmacokinet-
`ics, and prevent recognition by the innate immune system.
`Even though dsRNAs are more stable than single-stranded
`RNAs (ssRNAs), siRNAs are still rapidly degraded in serum
`and can benefit from chemical modification. A wide variety
`of novel chemical modifications have been developed and
`tested for use in nucleic acids applications such as antisense
`(AS) and ribozyme technologies (Kurreck, 2003). Many
`of these modifications may have similar utility for use in
`siRNAs.
`Although extensive studies have been published relating
`to chemical modification of 21-mer siRNAs, similar studies
`
`1
`
`2
`
` Integrated DNA Technologies, Inc. 1710 Commercial Park, Coralville, IA.
` The Biotechnology Centre of Oslo, University of Oslo, Oslo, Norway.
` siRNAsense AS, Oslo, Norway.
`4
` Division of Molecular Biology and 5Graduate School of Biological Sciences, Beckman Research Institute of the City of Hope,
`Duarte, CA.
`
`3
`
`187
`
`Alnylam Exh. 1038
`
`

`

`188
`
`COLLINGWOOD ET AL.
`
`exploring the modification patterns compatible with func-
`tion of Dicer-substrate RNAs have not been reported. We
`previously described optimization of unmodified RNA
`substrates for human Dicer (Rose et al., 2005). A series of
`chemically modified Dicer-substrate small interfering RNAs
`(DsiRNAs) starting from this basic design were synthesized
`using combinations of RNA, 2′-O-methyl (2′OMe) RNA,
`2′-fluoro (2′-F) RNA, and DNA bases and were compared
`for functional potency in knockdown of endogenous tar-
`gets or reporter constructs in vitro. Some of the modification
`patterns retained similar high potency compared with the
`unmodified parent DsiRNA, whereas other patterns showed
`reduced potency or were wholly inactive. The functional
`behavior of chemically modified siRNAs can vary with
`sequence context, so some of the DsiRNA modification pat-
`terns were tested against different endogenous genes as well
`as reporter constructs to assess to what extent correlations
`between potency and specific modification patterns might
`be sequence specific. A modification pattern that includes
`use of 2′OMe modified bases was identified that generally
`had minimal impact on potency at multiple sites tested. The
`DsiRNAs made with this modification pattern did not trig-
`ger immune responses in human immune cells in vitro and
`showed improved serum stability.
`
`Materials and Methods
`
`Chemically synthesized siRNAs
`
`All RNA oligonucleotides described in this study were
`synthesized using t-Butyl-dimethylsilyl (TBDMS) chemistry
`and purified using HPLC (Integrated DNA Technologies,
`Coralville, IA). All oligonucleotides were QC tested by
`electrospray-ionization mass spectrometry (ESI-MS) and
`were within ±0.02% predicted mass. Duplexes were also
`QC tested by analytical HPLC and were >90% pure. Final
`duplexes were prepared as sodium salts.
`
`In vitro dicing assays
`
`RNA duplexes (100 pmol) were incubated in 20 µL of 20
`mM Tris pH 8.0, 200 mM NaCl, 2.5 mM MgCl2 with or with-
`out 1 unit of recombinant human Dicer (Stratagene, La Jolla,
`CA) at 37°C for 18–24 hours. Samples were desalted using a
`Performa SR 96-well plate (Edge Biosystems, Gaithersburg,
`MD). Electrospray-ionization liquid chromatography mass
`spectroscopy (ESI-LCMS) of duplex RNAs pre- and post-
`treatment with Dicer was done using an Oligo HTCS system
`(Novatia, Princeton, NJ; Hail et al., 2004), which consisted of
`a ThermoFinnigan TSQ7000, Xcalibur data system, ProMass
`data processing software and Paradigm MS4 HPLC
`(Michrom BioResources, Auburn, CA). All dicing experi-
`ments were performed at least twice.
`
`HeLa cell culture, transfections, qRT-PCR, and
`western blots
`
`HeLa cells were split into 24-well plates at 40% conflu-
`ency and were transfected the next day with TriFECTin
`(Integrated DNA Technologies, Coralville, IA) using 1 µL
`per 100 µL OptiMEM I (Invitrogen, Carlsbad, CA) with RNA
`
`duplexes at the indicated concentrations. All transfections
`were performed three times or more. RNA was prepared 24
`hours after transfection using the SV96 Total RNA Isolation
`Kit (Promega, Madison, WI). RNA quality was verified
`using a Bioanalyzer 2100 (Agilent, Palo Alto, CA) and cDNA
`was synthesized using 150 ng total RNA with SuperScript-II
`Reverse Transcriptase (Invitrogen, Carlsbad, CA) per the
`manufacturer’s instructions using both random hexamer
`and oligo-dT priming.
`Quantitative real-time PCR assays were done using
`10 ng cDNA per 10 µL reaction, Immolase DNA poly-
`merase (Bioline, Randolph, MA), 200 nM primers, and
`200 nM probe. Hypoxanthine phosphoribosyltransferase
`(HPRT1) (NM_000194) specific primers were HPRT-For
`5′ GACTTTGCTTTCCTTGGTCAGGCA, HPRT-Rev
`5′
`GGCTTATATCCAACACTTCGTGGG and probe HPRT-P 5′
`FAM-ATGGTCAAGGTCGCAAGCTTGCTGGT-IowaBlackFQ
`(IBFQ). STAT1 (NM_007315) specific primers were STAT1-
`For 5′ AACCGCATGGAAGTCAGGTT, STAT1-Rev 5′
`ATGGGCTTCATCAGCAAGGA, and probe STAT1-P 5′ FAM-
`AGCTCTCACTGAACCGCAGCAGGAA-IBFQ. Cycling condi-
`tions employed were 95°C for 10 minutes followed by 40 cycles
`of 2-step PCR with 95°C for 15 seconds and 60°C for 1 minute.
`The PCR and fluorescence measurements were done using an
`ABI Prism 7900 Sequence Detector (Applied Biosystems Inc.,
`Foster City, CA). All data points were performed in triplicate.
`Expression data were normalized to levels of an internal con-
`trol gene, human acidic ribosomal phosphoprotein P0 (RPLP0;
`NM_001002), which was measured in separate wells in parallel
`using primers RPLP0-For 5′ GGCGACCTGGAAGTCCAACT,
`RPLP0-Rev 5′ CCATCAGCACCACAGCCTTC, and probe
`RPLP0-P 5′ FAM-ATCTGCTGCATCTGCTTGGAGCCCA-
`IBFQ (Bieche et al., 2000). Copy number standards were run in
`parallel using linearized cloned amplicons for both the HPRT
`and RPLP0 assays. Unknowns were extrapolated against
`standards to establish absolute quantitative measurements.
`Total protein was extracted from HeLa cells using
`Radioimmunoprecipitation buffer (RIPA) buffer (50 mM
`Tris–HCl pH 7.4, 1% NP-40, 0.25% sodium deoxycholate, 150
`mM NaCl, 1 mM EDTA, protease Inhibitor Cocktail [Sigma-
`Aldrich, #P8340]), and centrifuged. Aliquots of 15µg of protein
`for each sample were run on a 10% polyacrylamide gel with
`0.1% sodium dodecyl sulfate (SDS–PAGE) and transferred to
`a PVDF membrane (Bio-Rad, Hercules, CA). After blocking
`with 5% powdered milk, the membrane was incubated with
`the primary antibody (1 hour), washed with TBST (10 mM Tris
`pH 8, 150 mM NaCl, 0.1% Tween-20), incubated with the sec-
`ondary horseradish-peroxidase-conjugated antibody (1 hour)
`and washed again with TBST. The detection reaction was
`prepared with SuperSignal West Dura Extended Duration
`Substrate (Pierce, Rockford, IL), and the protein bands were
`visualized using a Kodak Image Station 440CF (Kodak,
`Rochester, NY). For quantification of band intensities, Kodak
`1D 3.6 software was used. Band densities were averaged from
`three separate experiments (three separate transfection wells
`run on separate lanes) and normalized to internal β-actin
`controls. The antibodies used were anti-HPRT (ab10479,
`Abcam, 1:1000, 1 hour incubation) (Abcam, Cambridge, MA),
`anti-β-actin (ab8226, Abcam, 1:2000, 1 hour incubation), goat
`antirabbit HRP conjugated (12–348, Upstate, 1:5000, 1 hour
`
`

`

`CHEMICALLY MODIFIED DICER-SUBSTRATE siRNA
`
`incubation) and goat antimouse (12–349, Upstate, 1:3000, 1
`hour incubation) (Upstate, Millipore, Billerica, MA).
`
`Cotransfection of HCT116 cells with psiCheck
`luciferase reporter plasmids and DsiRNA
`
`The pLuc-EGFP vector, in which a portion of enhanced
`green fluorescent protein (EGFP) was cloned into the 3′-UTR
`of the Renilla expression cassette in the psiCHECK-2 vec-
`tor (Promega, Madison, WI) has been described previously
`(Rose et al., 2005). The HCT116 colon carcinoma cell line was
`grown in Dulbecco’s modified Eagle’s medium (DMEM)
`supplemented with 10% fetal bovine serum, 20 U/mL peni-
`cillin/streptomycin, and 2 mM L-Gln (DMEM-complete).
`Cells were grown in a humidified, 5% CO2
` incubator at 37°C.
`HCT116 cells were seeded in 48-well plates and transfected
`24 hours later (~70% confluency) using 0.5 µL of Lipofectamine
`2000 (Invitrogen, Carlsbad, CA) per well. Each well received
`100 ng of psi-GFP and DsiRNAs at the indicated concentra-
`tions (100 pM, 1 nM, and 10 nM). All samples were trans-
`fected in duplicate or triplicate, and the experiment was
`performed three times. Luciferase assays were performed
`24 hours after transfection using the Dual Luciferase Assay
`system according to the manufacturer (Promega, Madison,
`WI). Renilla luciferase activities in relative light units (RLUs)
`are reported as normalized values (i.e., Renilla luciferase
`activity units/Firefly luciferase activity units).
`For studies involving mouse Tissue Factor target (F3),
`HCT116 cells were cotransfected in suspension in triplicate
`wells in 96-well plates using 25 ng reporter DNA, 0.15 µL
`Lipofectamine 2000 and 10–250 pM siRNA, for a 100 µL
`final transfection volume per well. Triplicate batch compl-
`exations were performed as follows: 3 µL DsiRNA at 100×
`final concentration was diluted with 15 µL 5 ng/µL Opti-
`MEM-diluted reporter DNA (prepared in one large batch
`for all samples) and mixed with an equal volume of a batch
`dilution of Lipofectamine 2000 in Opti-MEM (for 10 sam-
`ples: 145 µL Opti-MEM + 4.5 µL LF2000). Complexes were
`allowed to form at room temperature for 30 minutes. During
`complexation, near-confluent HCT116 cells were detached
`by trypsinization, resuspended in full media, centrifuged
`to remove residual trypsin, and resuspended at a cell den-
`sity sufficient to achieve approximately 80%–90% conflu-
`ency equivalence upon plating. Complexes were mixed with
`270 µL cell suspension, and 3 × 100 µL of the complexes/
`cell mixtures were plated in triplicate wells of a 96-well
`plate. The day after transfection, the medium was removed,
`cells lysed with 40 µL Passive Lysis Buffer (Promega), and
`dual luciferase assays performed on a Veritas Luminometer
`(Turner Biosystems, Sunnyvale, CA), using 10 µL lysate and
`50 µL of each substrate reagent (Promega). Target-specific
`Renilla reporter expression was standardized to internal
`nontarget Firefly luciferase expression, and normalized to
`values for samples treated with a scrambled control siRNA.
`For construction of the mouse Tissue Factor reporter,
`a fragment of the F3 gene (NM_010171) spanning bases
`154–1591 was cloned into SpeI sites in the 3′-UTR of the
`Renilla luciferase gene in the psiCHECK-2 vector (Promega)
`5′-gatgattctagactcgagGAGAC
`using primers mTF-for
`CTCGCCTCCAGCC-3′ and mTF-rev 5′-gatgatgtcgactag
`
`189
`
`tATCACAAAGATGCCCCAAGC-3′ (uppercase letters are
`complementary to the F3 target gene). All siRNAs studied
`were located between bases 211 and 1371 and were included
`within the subcloned fragment.
`
`Serum stability assay
`
`Aliquots of 2 nmol of each RNA duplex were incubated
`at 37°C in 200 µL buffer (PBS, pH 7.4, with 2 mM MgCl2)
`containing 50% (v/v) of fetal bovine serum (Invitrogen,
`Carlsbad, CA), resulting in a concentration of 10 µM siRNA
`duplex. The serum was not heat inactivated. The reactions
`were stopped with addition of 100 µL of phenol/chloro-
`form/isoamyl alcohol (25/24/1 v/v/v) and the mixture was
`stored at –80°C. When all incubations were done, samples
`were extracted and ethanol precipitated. For gel analysis,
`20 pmol (approximately 0.0067 OD260) of each sample was
`loaded onto 20% polyacrylamide nondenaturing gels in
`TBE buffer. Following electrophoresis, the gels were stained
`with 1× GelStar (Lonza, Rockland, ME) and visualized on a
`UV-transilluminator.
`
`Detection of human interferon-α (IFN-α) in PBMC
`culture supernatants
`
`Human peripheral blood was obtained from normal healthy
`donors (Ullevaal University Hospital Blood Center, Oslo,
`Norway), and peripheral blood mononuclear cells (PBMCs)
`were isolated by standard Isopaque-Ficoll (Lymphoprep;
`Nycomed) gradient centrifugation. Cells were counted and
`resuspended at 3.0 × 106 per ml. DsiRNAs were complexed
`with Lipofectamine 2000 for triplicate suspension transfec-
`tions at 500 µL 100 nM siRNA in 24-well plates. The suspension
`of 300 pmole siRNA (5.1 µg) was diluted with 200 µL serum-
`free RPMI medium (SFM; Gibco, Invitrogen, Carlsbad, CA)
`and mixed with 200 µL of SFM-diluted Lipofectamine 2000
`(10.5 µL). Following 30 minutes incubation at room tempera-
`ture, the complexes were mixed by resuspension with 1.1 mL
`of 3.0 × 106/mL PBMC and 3 × 0.50 mL of the cell/complexes
`mixture seeded into triplicate wells of 24-well plates. Medium
`was collected from samples 20 hours posttransfection, trans-
`ferred to 1 mL tubes on ice, centrifuged for 5 minutes at 400 ×
`g at 4°C, and diluted with one part dilution buffer provided in
`the ELISA kit (PBL Biomedical Laboratories, Piscataway, NJ).
`All samples were assayed in duplicate. The amount of IFN-α
`was quantified from a standard curve according to the manu-
`facturer’s high sensitivity protocol.
`
`Cytokine assays in T98G cells
`
`T98G cells were obtained from ATCC and maintained
`under standard culture conditions in high glucose DMEM
`containing 10% fetal calf serum and 1% pen/strep. Cells
`were plated at 3 × 104 per well in a 24-well plate the day
`before transfection in 500 µL complete media without antibi-
`otics. Transfections were conducted using 0.6 µL per well of
`siLentFect (Bio-Rad, Hercules, CA) and appropriate amount
`of siRNA duplex to give either 10 nM or 100 nM final con-
`centration. Media was not changed to remove siRNA com-
`plexes. Following incubation at 37°C, cell supernatants were
`
`

`

`190
`
`COLLINGWOOD ET AL.
`
`collected and placed in tubes and stored at –20°C until
`cytokine assays were run as a group.
`Secreted cytokine levels were measured 24 hours post-
`transfection using Luminex xMAP technology with Beadlyte
`reagents (IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, TNF-α, IFN-γ,
`and GM-CSF) from Upstate (Millipore, Billerica, MA) or IFN-α
`Biosource (Invitrogen, Carlsbad, CA) on a Luminex 100 sys-
`tem (Luminex, Austin, TX). Standard curves for the Beadlyte
`assays were generated using Beadlyte Human Multi-Cytokine
`Standard 3 (5000–6.9 pg/mL) from Upstate. IFN-α standard
`curves were generated using the Hu Cytokine II Standard
`(7100–9.7 pg/mL) from Biosource. Mean Fluorescence Intensity
`(MFI) was measured using gate settings of 7500–13,500, sam-
`ple size 75 µL and 75 detection events per bead set. Sample con-
`centrations (pg/mL) were determined by extrapolation from
`the standards using a 5-PL curve.
`
`Results
`
`Testing and optimization of modification patterns
`
`We previously described optimization of synthetic RNA
`substrates for human Dicer. The final preferred structure
`was an asymmetric duplex with a 25-mer sense (S) strand
`and a 27-mer AS strand having a single 2-base 3′-overhang
`positioned on the AS strand and two DNA residues at the
`3′-end of the S strand (Rose et al., 2005). This design results
`in a single, predictable product from Dicer processing and
`improves relative loading of the AS strand into RISC. A
`series of chemically modified DsiRNAs were synthesized at a
`known effective site in the human STAT1 gene (NM_007315)
`employing combinations of RNA, 2′OMe RNA, 2′-F RNA,
`and DNA bases (Table 1). Site selection was performed using
`accepted design criteria (Peek, 2007; Peek and Behlke, 2007).
`Previously, end modification of blunt 27-mer RNA duplexes
`with a bulky, hydrophobic dye (fluorescein) was found to
`interfere with Dicer processing and significantly reduced
`functional potency (Kim et al., 2005). We therefore avoided
`chemical modifications adjacent to the predicted site of Dicer
`processing (Fig. 1) so as to not interfere with cleavage.
`The series of STAT1 chemically modified DsiRNAs were
`transfected into HeLa cells at 10 nM, 1 nM, and 0.1 nM con-
`centrations and relative knockdown of STAT1 expression was
`assessed by qRT-PCR at 24 hours (Fig. 2). This site showed
`high potency and the unmodified construct suppressed
`gene expression by more than 85% at 0.1 nM concentration
`(Sequence 1 in Table 1 and Fig. 2). Several of the modification
`patterns showed potency nearly identical to the unmodified
`version at high (10 nM) and medium (1 nM) doses (such as
`Sequences 2, 4, 6, 8, 10, and 11), whereas other modification
`patterns showed reduced activity (Sequences 9, 14, 16, and
`17). At low dose (0.1 nM), alternating 2′OMe bases gener-
`ally showed the best potency; however none of the modified
`duplexes was as potent as the original unmodified duplex.
`These results were not surprising; many chemical modifica-
`tion patterns have been described for 21-mer siRNAs that
`can similarly reduce effectiveness of the RNA duplex as a
`trigger of RNAi. These effects can vary with sequence con-
`text, making it important to examine the behavior of modifi-
`cation patterns at different sites in different gene targets.
`
`Since the alternating 2′OMe RNA modification pattern
`appeared promising in the STAT1 series, modified DsiRNA
`using this pattern were synthesized specific to a second target,
`the reporter gene EGFP (Fig. 3a). These duplexes were cotrans-
`fected into HCT116 cells with a luciferase expression plasmid
`containing EGFP sequences cloned into the 3′-UTR of the firefly
`luciferase gene, which was used as a reporter to assess relative
`efficacy of knockdown (Fig. 3b). As before, duplexes having an
`alternating 2′OMe pattern in either the S or AS strands showed
`high potency (in this case identical to the unmodified version).
`However, the duplex having modifications present on both S
`and AS strands was largely inactive, highlighting the poten-
`tial for a modification pattern to perform well in one sequence
`context (Fig. 2, Duplexes 3 and 5) but not in a different con-
`text (Fig. 3b, Duplex 4). For general research applications, a
`modification pattern that is generally effective irrespective of
`sequence context is desirable. For DsiRNAs, selective modi-
`fication of a single strand (either S or AS) may therefore be
`preferable to designs having both strands modified. Of note,
`in the EGFP series modification pattern #EGFP-5 included
`placement of 2′OMe bases at the 5′-end of the AS strand so that
`modifications flanked the expected Dicer cut site. This pattern
`showed potency similar to the unmodified and other modi-
`fied duplexes that did not span the cut site.
`Since a modification pattern that could be generally
`employed at any site was desired, the utility of the alter-
`nating 2′OMe pattern was tested in the sequence context
`of additional sites on two new target genes. A site in the
`human HPRT1 gene (NM_000194) was studied comparing
`an unmodified DsiRNA with duplexes having the anti-
`sense-strand modified (ASm) or both strands modified (Sm/
`ASm; Fig. 4a). Duplexes were transfected into HeLa cells at
`10 nM, 1 nM, and 0.1 nM concentrations and the relative
`level of HPRT mRNA levels was assessed by qRT-PCR at
`24 hours (Fig. 4b). The ASm duplex had similar potency to
`the unmodified duplex and, in this case, the duplex hav-
`ing both strands modified was nearly as potent. Active sites
`in the mouse F3 gene (Mus musculus coagulation factor III,
`NM_010171) were identified by screening a series of candi-
`date RNA duplexes (data not shown). The three most potent
`sites were studied in greater detail, comparing the func-
`tional potency of unmodified and ASm duplexes (Fig. 5a).
`HCT116 cells were cotransfected with a luciferase expres-
`sion plasmid containing F3 sequence cloned into the 3′-UTR
`of the firefly luciferase gene, which was used as a reporter
`to assess relative efficacy of knockdown. Functional studies
`for two duplexes are shown (Fig. 5b). The DsiRNAs studied
`were extremely potent, having IC50 concentrations of 10–20
`pM. For these hyperfunctional sites, the ASm duplexes were
`equally potent to the unmodified duplexes.
`
`Dicer processing, stability, and kinetics
`
`Ideally, the chemical modification patterns employed
`should not alter processing of the RNA duplexes as Dicer sub-
`strates (Fig. 1). To verify cleavage patterns, mass spectrom-
`etry analyses of the products of in vitro dicing reactions were
`performed (Rose et al., 2005) to identify the actual species
`produced by cleavage of each of the HPRT-specific DsiRNAs
`(Fig. 4a) when incubated with recombinant human Dicer.
`
`

`

`CHEMICALLY MODIFIED DICER-SUBSTRATE siRNA
`
`191
`
`Table 1. STAT1 27-mer Modified DsiRNAs
`
`Sequence
`
`pCUUCCUCUCUUUCUCUCCCUUGUga
`5'
`3' AGGAAGGAGAGAAAGAGAGGGAACACU
`
`pGCACCAGAGCCAAUGGAACUUGAtg
`5'
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`pGCACCAGAGCCAAUGGAACUUGAtg
`5'
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`----------
`5.
`pGCACCAGAGCCAAUGGAACUUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`---------
`
`pGCACCAGAGCCAAUGGAACUUGAtg
`5'
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`----------
`5' pGCACCAGAGCCAAUGGAACUUGAtg
`3. UUCGUGGUCUCGGUUACCUUGAACUAC
`----------
`5.
`pGCACCAGAGCCAAUGGAACUUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`5. p~ C~ CCAGAGCCAAUGGAACUUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`5.
`pGCACCAGAGCCAAUGGAACUUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`-- -
`-
`5' p~ C~ CCAGAGCCAAUGGAACUUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`-- -
`-
`5' pGc AccAGAGccAAuGGAAcuUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`5.
`pGCACCAGAGCCAAUGGAACUUGAtg
`3' UucGuGGucucGGuuAccuuGAACUAC
`
`5' pGc AccAGAGccAAuGGAAcuUGAtg
`3' UucGuGGucucGGuuAccuuGAACUAC
`
`5' pGc AccAGAGccAAuGGAAcuUGAtg
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`5. p~ C~ CCAGAGCCAAUGGAACUUGAtg
`3' UucGuGGucucGGuuAccuuGAACUAC
`
`5.
`pGCACCAGAGCCAAUGGAACUUGAtg
`3' UucGuGGucucGGuuAccuuGAACUAC
`-- -
`-
`5' p~ C~ CCAGAGCCAAUGGAACUUGAtg
`3' UucGuGGucucGGuuAccuuGAACUAC
`-- -
`-
`5' pGc AccAGAGccAAuGGAAcuUGAtg
`3' UucGuGGucucGGuuAccuuGAACUAC
`-- -
`-
`
`Name
`
`Control (CJ
`
`STATl-1
`
`STATl-2
`
`STATl-3
`
`STATl-4
`
`STATl-5
`
`STATl-6
`
`STATl-7
`
`STATl-8
`
`STATl-9
`
`STATl-10
`
`STATl-11
`
`STATl-12
`
`STATl-13
`
`STATl-14
`
`STATl-15
`
`STATl-16
`
`STATl-17
`
`RNA = AGCU
`AGCU
`2'0Me
`
`2'F
`DNA
`
`cu
`agct
`
`p
`
`phos
`
`Results are summarized in Table 2. DsiRNAs are designed to
`offer the PAZ domain of Dicer a single 3′-overhang and help
`orient the substrate so that cleavage occurs at a predictable site
`21–22 bases from the 3′-end of the overhang (Rose et al., 2005).
`A mix of 21-mer and 22-mer products is often seen. For the
`unmodified HPRT DsiRNA duplex, mass analysis of cleav-
`age products was consistent with the 27-mer substrate being
`cleaved by Dicer at the expected position, resulting in a single
`21-mer RNA species with symmetric 2-base 3′-overhangs and
`
`5′-phosphate groups at each end (i.e., a mature siRNA). A sim-
`ilar processing pattern was observed for the 2′OMe modified
`ASm and Sm/ASm duplexes except that a mix of both 21-mer
`and 22-mer species was produced. Therefore, limited chemi-
`cal modification as used here does not significantly interfere
`with Dicer processing of short substrate RNAs.
`The 2′OMe modification has been used to improve the
`stability of 21-mer siRNAs in the presence of serum nucle-
`ases. The modification pattern employed here was designed
`
`

`

`192
`
`COLLINGWOOD ET AL.
`
`Avoid modilication in this
`region so as to not interfere
`with Dicer processing
`
`A
`
`Sequence
`
`5 '
`3 '
`
`pGCCAGACUUUGUUGGAUUUGAAAtt
`UUCGGUCUGAAACAACCUAAACUUUAA
`
`Name
`
`Hprt
`
`GCACCAGAGCCAAUGGAACUUGAtg
`5 '
`3' UUCGUGGUCUCGGUUACCUUGAACUAC
`
`l
`
`f
`
`Dicer Cleavage l
`
`GCACCAGAGCCAAUGGAACUU
`5'
`3 ' UUCGUGGUCUCGGUUACCUUG
`
`21 mer product siRNA
`
`FIG. 1. Scheme for designing chemically modified Dicer-
`substrate siRNAs (DsiRNAs). The region of the RNA duplex
`surrounding the expected cleavage site (Rose et al., 2005)
`was left as unmodified RNA. Sequence 5′- to the cleavage
`site were modified in the sense strand and sequence 3′- to
`the cleavage site were modified in the antisense strand. In
`all cases, two DNA residues were placed at the 3′-end of the
`sense strand.
`
`to leave an unprotected domain so that the RNA duplexes
`could be cleaved by the endoribonuclease Dicer. This
`modification pattern might or might not confer stability to
`other nucleases. Unmodified and 2′OMe modified duplexes
`(Fig. 6a) were incubated in fetal calf serum (which was
`not heat inactivated), and the degradation products were
`
`25/27mer DsiRNA
`
`5 '
`3 '
`
`pACCCUGAAGUUCAUCUGCACCACcg
`ACUGGGACUUCAAGUAGACGUGGUGGC
`
`EGFP-1
`
`5 '
`3 '
`
`pACCQU~N\,GYUQAYCYGQACCACcg
`ACUGGGACUUCAAGUAGACGUGGUGGC
`
`EGFP- 2
`
`5 '
`3 '
`
`pACCCUGAAGUUCAUCUGCACCACcg
`ACUGGGACUUCAAGOAGACGUGGUGGC
`
`EGFP-3
`
`5 '
`3 '
`
`pACCQUG~GYUCAYCQGQACCACcg
`ACUGGG~CYUQN\,GQAGACGUGGUGGC
`
`EGFP-4
`
`5 '
`3 '
`
`pACCCUGAAGUUCAUCUGCACCACcg
`ACUGGG~CQUQN\,GQAGAQGUGGUGGC
`
`EGFP- 5
`
`B
`
`RNA
`DNA
`
`AGCO
`acgt
`
`2 ' 0Me = ~
`p = phos
`
`g 1.2 ~ - - - - - - - - - - - - - - - - -~ mo.1 nM
`...J u.
`a.
`u.
`IB o.s
`+ " j 0.6
`
`0:::
`~ 0.4
`
`.!:! 1 0.2
`0 z
`
`0
`
`(cid:127) r (cid:127) :a,e ._
`- lli(cid:127)
`-=-~ --=•=-=-=-• -
`--ktl ~
`...Abtk-
`-.,,...-=e"a (cid:127) --
`ia=r.:a=(cid:127)
`
`- ---------- ----
`
`, • • a'9,.,."9 (cid:127)(cid:127)
`.
`~~,-~
`
`-
`
`W
`
`-
`---
`--=------------
`-
`-
`·--
`
`~ z
`a::
`u
`i,::
`T5
`Cl)
`a.
`(/)
`.....
`:§
`(/)
`
`C
`
`1:li-
`2--=--
`
`3 -
`
`I I
`
`I
`
`6 ..,..
`7 -...+--
`• -..J
`
`9
`
`l----' .-.- =-
`10 --11 -
`
`12 -
`
`13 -
`
`14
`
`-
`
`15 .........
`
`16
`
`17
`
`I
`
`I
`
`'
`
`I
`
`(cid:127) 1~
`C 1nM
`~ .1nM
`
`y,.'?~-<,,
`
`~v-'1-
`t-<>
`
`~v-">
`t-<>
`Duplex
`
`I
`
`I
`
`I
`
`i
`i
`
`FIG. 3. Relative potency of 2′OMe modification patterns
`for EGFP DsiRNAs. (A) Sequence of the EGFP-specific and
`control (HPRT) DsiRNAs are shown. Black uppercase letters
`are RNA, bold uppercase underlined letters are 2′OMe RNA,
`and lowercase italic letters are DNA bases. (B) DsiRNA
`duplexes were cotransfected into HCT116 cells at the con-
`centrations indicated with a psiCHECK-2 vector containing
`EGFP sequence in the 3′-UTR of firefly luciferase. Luciferase
`assays were performed 24 hours posttransfection and rela-
`tive FLuc signal was normalized against an internal Renilla
`luciferase control.
`
`resolved using polyacrylamide gel electrophoresis (PAGE;
`Fig. 6b). No intact 21-mer siRNA is seen after 6 hours incuba-
`tion in fetal calf serum. Consistent with earlier reports, the
`unmodified 27-mer duplexes were somewhat more nucle-
`ase resistant than 21-mer duplexes at the same site (Kubo
`et al., 2007), and small amounts of the unmodified 27-mer
`DsiRNA were still present at 6 hours. In contrast, an appre-
`ciable amount of the 2′OMe ASm modified 27-mer DsiRNA
`remained intact after 24 hours incubation in serum. Thus
`modification with 2′OMe bases in even the limited ASm pat-
`tern improved nuclease stability of the DsiRNA. Greater sta-
`bilization was seen when the alternating pattern of 2′OMe
`bases was extended through the full length of the duplex
`
`- RNA
`- DNA
`
`- 2'-0-Me RNA
`- 2'-FRNA
`
`0%
`
`20% 40% 60% 80% 100% 120% 140%
`% Rel. Expression
`
`FIG. 2. Relative potency of various chemical modification
`patterns for human STAT1 DsiRNAs. RNA duplexes were
`transfected into HeLa cells at 10, 1, and 0.1 nM concentra-
`tion and relative STAT1 mRNA levels were assessed at 24
`hours using qRT-PCR. The chemical modification patterns
`employed are illustrated schematically on the left where
`black represents RNA, Blue represents 2′OMe RNA, red rep-
`resents 2′-F RNA, and purple represents DNA bases.
`
`(cid:127)
`

`

`CHEMICALLY MODIFIED DICER-SUBSTRATE siRNA
`
`A
`
`Sequence
`
`Name
`
`A
`
`Sequence
`
`193
`
`Name
`
`Con
`
`pCUUCCUCUCUUUCUCUCCCUUGUga
`5 '
`3' AGGAAGGAGAGAAAGAGAGGGAACACU
`
`pGCCAGUAAUUCAGCAGUUUGAACaa
`5 '
`3 ' GUCGGUCAUUAAGUCGUCAAACUUGUU
`
`mF3-l4
`
`pGCCAGUAAUUCAGCAGUUUGAACaa
`5 '
`3 ' GUCG~UQAYU~gU~G2CAAACUUGUU
`
`mFJ- 14 ASm
`
`pCACCAAUGAAUUCUCGAUUGAUGtg
`5 '
`3 ' UUGUGGUUACUUAAGAGCUAACUACAC
`
`mF3-l7
`
`pCACCAAUGAAUUCUCGAUlJGAIJG tg
`5 '
`3 ' UUGUGGUUACUUAAGA!,CUAACUACAC
`
`mFJ- 17 ASm
`
`pGCAUUCCAGAGAAAGCGUUUAAUtt
`5'
`3' UCCGUAAGGUCUCUUUCGCAAAUUAAA
`
`mF3- 2l
`
`pGCAUUCCAGAGAAAGCGUUUAAUtt
`5 '
`3 ' ~GQ~-l,G~UQUQUQUQGCAAAUUAAA
`
`mFJ-21 ASm
`
`pCUUCCUCUCUUUCUCUCCCUUGUga Con
`5 '
`3 ' AGGAAGGAGAGAAAGAGAGGGAACACU
`
`pGCCAGACUUUGUUGGAUUUGAAAtt HPRT unmod
`5 '
`3 ' UUCGGUCUGAAACAACCUAAACUUUAA
`
`pGCCAGACUUUGUUGGAUUUGAAAtt HPRT ASm
`5 '
`3 ' ~G~UQU~AAAQAACQUA.o.ACUUUAA
`
`pGCCAGACUU2G{lUGGAU2UGAAAtt HPRT Sm/ASm
`5 '
`3 ' !JUCGGUCUGAAACAACCUAAACUUUAA
`
`RNA
`ONA
`
`AGCU
`agct
`
`2'0Me = AGCO
`p = phos
`
`B
`
`0
`:;;J
`
`2'0Me = AGCU
`AGCU
`RNA
`p = phos
`DNA
`agct
`1·2 , - - - - - - - - - - - - - - - __ r :(cid:143) :-· :-:10:-p-:M::-i
`12150pM
`(cid:127) 250pM
`
`..J u. -LL.
`'e 0.8
`+
`0 _3 0.6
`er
`~ 0.4
`.!:!
`iii E 0.2
`0 z
`
`Con
`
`Unmod
`
`ASm
`
`Sm/ASm
`
`FIG. 4. Relative potency of 2′OMe modification patterns for
`human HPRT DsiRNAs. (A) Sequence of the HPRT-specific
`and control DsiRNAs are shown. Black uppercase letters are
`RNA, bold uppercase underlined letters are 2′OMe RNA,
`and lowercase italic letters are DNA bases. (B) DsiRNA
`duplexes were cotransfected into HeLa cells at the concen-
`trations indicated. RNA was prepared at 24 hours posttrans-
`fection and qRT-PCR was performed to assess expression of
`HPRT mRNA.
`
`(data not shown); however, this extensive modification also
`blocked Dicer processing and was not studied further.
`We next examined the initial rate of target knockdown and
`duration of silencing using the HPRT DsiRNAs. The unmod-
`ified HPRT DsiRNA was transfected into HeLa cells at 10 nM
`concentration and RNA was prepared at 3, 6, 12, 24, and 48
`hours posttransfection. HPRT mRNA levels were assessed
`using qRT-PCR (Fig. 7a). Degradation of HPRT mRNA was
`rapid, showing a 56% reduction at 3 hours, 90% reduction at
`6 hours,