Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`www.elsevier.com/locate/ybbrc
`
`A structure–activity relationship study of siRNAs
`with structural variations
`
`Chan Il Chang a, Sun Woo Hong a, Soyoun Kim b, Dong-ki Lee a,*
`
`a Department of Chemistry and BK School of Molecular Science, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea
`b Department of Chemistry, Dongguk University, Seoul 100-715, Republic of Korea
`
`Received 26 May 2007
`Available online 11 June 2007
`
`Abstract
`
`Specific knock-down of cellular gene expression using small interfering RNAs (siRNAs) is a powerful gene silencing technique in
`mammalian cells. Early siRNAs were double stranded, and 19–21 bp in length, but several variations in siRNA structure have been intro-
`duced to achieve better silencing efficiency. In addition, siRNA modules have been incorporated into higher-order RNA structures to
`generate multi-functional RNA molecules. The effects of such structural variations on the activities of siRNAs have not been thoroughly
`studied. Here, we present a structure–activity relationship study of siRNA structural variants. Specifically, we focus on the effect on
`silencing efficiency of the attachment of extra, target-unrelated sequences to the conventional short duplex siRNA structure. Interest-
`ingly, while some siRNA structural variants efficiently silence target gene expression, others show a reduction in or a complete lack
`of silencing activity. Off-target effects and innate immune responses triggered by siRNA structural variants were also measured. In vitro
`Dicer cleavage reactions show that all siRNA structural variants are substrates of Dicer, but digestion patterns vary. To our knowledge,
`this is the first systematic structure–activity relationship analysis of siRNAs bearing structural variations. Our results provide useful
`guidelines for the design of siRNA structural variants and for the construction of complex RNA molecules bearing functional siRNA
`modules.
`Ó 2007 Elsevier Inc. All rights reserved.
`
`Keywords: RNA interference (RNAi); Small interfering RNA (siRNA); Lamin; Structure–activity relationship; Off-target gene silencing; Innate immune
`response; OAS2; Dicer
`
`Small interfering RNAs (siRNAs) are double-stranded
`RNAs (dsRNAs) 19- to 21-bp in length that are able to tar-
`get and degrade cellular mRNAs with complementary
`nucleotide sequences [1]. siRNA, along with its close rela-
`tive, short hairpin RNA (shRNA), utilizes the naturally
`occurring RNA interference (RNAi) pathway [2]. Because
`of high efficiency and specificity, siRNA-based gene silenc-
`ing has become a powerful molecular tool in target valida-
`tion and a promising future therapeutic approach [3].
`Initially, siRNA was designed as 19-bp double-stranded
`RNA with 2-nt 30 overhangs (termed the 19+2 structure) to
`mimic the Dicer-cleaved product of a long dsRNA [1].
`
`* Corresponding author.
`E-mail address: dlee@postech.edu (D. Lee).
`
`0006-291X/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved.
`doi:10.1016/j.bbrc.2007.06.004
`
`Since then, several variations in the siRNA structure have
`been introduced to achieve better silencing efficiency. The
`first example of structural variation in the canonical
`19+2 siRNA was the development of 27 bp-long siRNAs
`[4,5] and 29 bp-long shRNAs [6]. These longer RNA
`duplexes provided better silencing efficiency than did
`19+2 siRNAs. In addition, siRNA modules were imple-
`mented within a higher-order RNA structure to obtain
`multi-functional RNA molecules. McNamara et al.
`[7]
`recently developed an aptamer–siRNA fusion RNA mole-
`cule to achieve aptamer-driven cell type-specific delivery of
`siRNAs. In another study, pRNAs were developed. These
`were based on bacteriophage phi29 RNA sequence and
`included a number of functional RNA modules, including
`siRNA modules, producing multi-functional RNA nano-
`particles [8]. Also, siRNAs with extended sense strands
`
`Alnylam Exh. 1088
`
`

`

`998
`
`C.I. Chang et al. / Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`have been developed to facilitate quantification of cellular
`siRNA concentrations [9].
`Although many siRNA structural variants exist, to date
`no systematic structure–activity relationship analyses have
`been performed on siRNA structures deviating from the
`canonical 19+2 structure. This is especially true of siRNAs
`fused to sequences unrelated to target mRNAs. An under-
`standing of the effect of appending extra sequences to siR-
`NAs on silencing efficiency and specificity is important for
`designing structurally modified siRNAs with optimal gene
`silencing efficiency. Also, several studies have shown that
`short dsRNAs such as siRNAs can be recognized by sev-
`eral cellular dsRNA sensors [10–12] to elicit innate immune
`responses, resulting in the non-specific induction of inter-
`feron-regulated genes and inflammatory cytokines. An
`understanding of the potential of siRNA structural vari-
`ants to trigger non-specific innate immune responses is
`important in the design of functional RNA molecules with
`few side-effects.
`In the present study, we designed and synthesized a
`number of siRNA structural variants and studied their
`structure–activity relationships, focusing on silencing effi-
`ciency, off-target gene silencing, and the potential to trigger
`innate immune responses. Specifically, we measured the
`effect of appending extra target-unrelated sequences to
`the conventional short duplex siRNA structure. We also
`examined the cleavage patterns of these siRNAs by the
`recombinant Dicer protein in vitro.
`
`Materials and methods
`
`Cell culture and transfection. HeLa cells were cultured in Dulbecco’s
`modified Eagle’s medium (Hyclone) supplemented with 10% (v/v) fetal
`bovine serum, 100 U/ml penicillin, and 100 lg/ml streptomycin. Cells were
`plated 24 h before transfection at 50% confluence in complete medium
`without antibiotics, in 12-well plates. All transfection assays were per-
`formed using oligofectamine,
`following the manufacturer’s protocol
`(Invitrogen) at the indicated concentrations, and cells were harvested 24 h
`after transfection.
`siRNAs. Chemically synthesized RNAs were purchased from Bioneer
`and annealed according to the manufacturer’s protocol. The sense and
`antisense strands of siRNAs were:
`
`19+0
`
`19+2
`
`27+0
`
`30LAS
`
`50LAS
`
`50-UGUUCUUCUGGAAGUCCAG-30 (antisense)
`30-ACAAGAAGACCUUCAGGUC-50 (sense)
`
`50-UGUUCUUCUGGAAGUCCAGdTdT-30 (antisense)
`30-dTdTACAAGAAGACCUUCAGGUC-50 (sense)
`
`50-UGUUCUUCUGGAAGUCCAGUUCCUCCU-30
`(antisense)
`30-ACAAGAAGACCUUCAGGUCAAGGAGGA-50
`(sense)
`
`50-UGUUCUUCUGGAAGUCCAGCGCUCGCCCGG-30
`(antisense)
`30-ACAAGAAGACCUUCAGGUC-50 (sense)
`
`50-CCGGGCGAGCGUGUUCUUCUGGAAGUCCAG-30
`(antisense)
`30-ACAAGAAGACCUUCAGGUC-50 (sense)
`
`30LS
`
`50LS
`
`50-UGUUCUUCUGGAAGUCCAG-30 (antisense)
`30-GGCCCGCUCGCACAAGAAGACCUUCAGGUC-50
`(sense)
`
`50-UGUUCUUCUGGAAGUCCAG-30 (antisense)
`30-ACAAGAAGACCUUCAGGUCGCGAGCGGGCC-50
`(sense)
`
`30LS30LAS 50-UGUUCUUCUGGAAGUCCAGCGCUCGCCCGG-30
`(antisense)
`30-GGCCCGCUCGCACAAGAAGACCUUCAGGUC-50
`(sense)
`
`50LS50LAS 50-CCGGGCGAGCGUGUUCUUCUGGAAGUCCAG-30
`(antisense)
`30-ACAAGAAGACCUUCAGGUCGCGAGCGGGCC-50
`(sense)
`
`30LS50LAS 50-CCGGGCGAGCGUGUUCUUCUGGAAGUCCAG-30
`(antisense)
`30-GGCCCGCUCGCACAAGAAGACCUUCAGGUC-50
`(sense)
`
`50LS30LAS 50-UGUUCUUCUGGAAGUCCAGCGCUCGCCCGG-30
`(antisense)
`30-ACAAGAAGACCUUCAGGUCGCGAGCGGGCC-50
`(sense)
`
`Quantitative real-time PCR. Total RNA was extracted from cell lysates
`using the Tri-reagent kit (Ambion). One microgram of total RNA was
`used as a template for cDNA synthesis performed with the ImProm-IIä
`Reverse Transcription System (Promega), according to the manufacturer’s
`protocol. Aliquots (1/20) of the cDNA reactions were analyzed by
`quantitative real-time PCR on a Rotor-Gene 3000 PCR machine (Corbett
`Research) according to the manufacturer’s protocol. Gene-specific primers
`for GAPDH, lamin, and OAS2 genes were mixed separately with Power
`SYBR Green PCR Master Mix (Applied Biosystems) containing cDNAs
`to be analyzed. Samples were run in duplicate and data obtained were
`analyzed with Rotor-Gene6 software
`(Corbett Research). Primer
`sequences for each gene were:
`
`Lamin—forward
`Lamin—reverse
`
`GAPDH—forward
`GAPDH—reverse
`
`OAS2—forward
`OAS2—reverse
`
`50-CCGAGTCTGAAGAGGTGGTC-30
`50-AGGTCACCCTCCTTCTTGGT-30
`50-GAGTCAACGGATTTGGTCGT-30
`50-GACAAGCTTCCCGTTCTCAG-30
`50-TCAGAAGAGAAGCCAACGTGA-30
`50-CGGAGACAGCGAGGGTAAAT-30
`
`In vitro Dicer cleavage assay. One hundred picomoles each siRNA
`duplex was cleaved using the Turbo Dicer siRNA Generation Kit (Gen-
`lantis) for 8 h, according to the manufacturer’s protocol. The concentra-
`tions of Dicer used varied from 0.1 to 1 U, as indicated. A 2-ll aliquot of
`each reaction (20 pmol) was separated in a 15% (w/v) non-denaturing
`polyacrylamide
`gel,
`stained with EtBr
`and visualized by UV
`transillumination.
`
`Results and discussion
`
`Structure–activity relationships of siRNAs with single strand
`extensions on one strand
`
`We first synthesized an siRNA targeting lamin A/C
`mRNA as described by Elbashir et al. [1] (Fig. 1A and
`
`

`

`C.I. Chang et al. / Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`999
`
`A
`
`19+0
`
`19+2
`
`27+0
`
`3’LAS
`
`5’LAS
`
`3’LS
`AS 5’
`
`5’LS
`
`19+0 27+0 3’LAS 5’LAS 3’LS 5’LS
`
`B
`
`C
`
`1nM
`10nM
`100nM
`
`5'LS
`3'LS
`5'LAS
`3'LAS
`27+0
`19+2
`19+0
`0nM
`
`0
`
`20
`
`80
`60
`40
`lamin level
`
`100
`
`120
`
`Fig. 1. Structures and activities of siRNAs with single strand extensions on one strand. (A) Structure of control siRNAs and siRNAs with single strand
`extensions on one strand. Extended sequences are underlined. (B) Non-denaturing PAGE analysis of the siRNAs in (A). (C) Lamin gene silencing activity
`of siRNAs in (A). Lamin mRNA level was measured 24 h after transfection. Data were calculated as ratios of lamin mRNA level/GAPDH mRNA level.
`Transfection with oligofectamine alone (0 nM) was the control and the lamin/GAPDH ratio was set at 100%. All data in the graph show means ± SD
`values from three independent experiments.
`
`B). The conventional 19+2 siRNA was transfected into
`HeLa cells using oligofectamine. As expected, good silenc-
`ing efficiency was observed at all concentrations tested
`(Fig. 1C). siRNAs with a 19-bp duplex with blunt ends
`(19+0), and a 27-bp duplex with blunt ends (27+0), also
`showed effective gene silencing (Fig. 1A–C). We used these
`siRNAs as control siRNAs in comparisons with siRNA
`structural variants.
`To study the structure–activity relationship of siRNA
`structural variants, we first designed and synthesized lamin
`siRNA variants with 11-nt extensions of target-unrelated
`sequences at either the 30 or the 50 ends of the antisense
`or sense strands, to generate the 30LAS, 50LAS, 30LS,
`and 50LS structures (Fig. 1A and B). These siRNA struc-
`tural variants were transfected into HeLa cells, and the
`lamin gene silencing efficiency was measured at three con-
`centrations of the siRNAs (100, 10, and 1 nM) and com-
`pared with the gene silencing efficiency of control
`siRNAs (19+0, 19+2, and 27+0). At 100 and 10 nM con-
`centrations, all four variant siRNAs showed good silencing
`activity (Fig. 1C). However, at the lowest siRNA concen-
`tration (1 nM), the silencing activities of the 50LS and
`50LAS, and the 30LAS siRNAs, which contained sequence
`extensions at the 50 end of the sense strand, the 50 end of
`the antisense strand or the 30 end of the antisense strand,
`respectively (Fig. 1A), were somewhat reduced (40–50%
`silencing)
`compared with control
`siRNAs
`(60–70%
`
`silencing) (Fig. 1C). However, 30LS, which has a sequence
`extension at the 30 end of the sense strand (Fig. 1A),
`showed gene silencing efficiency (70% silencing) similar
`to that of the control siRNAs (Fig. 1C). Single strand
`extension of extra sequences unrelated to the target mRNA
`thus has different effects on the gene silencing activity of
`siRNAs, depending upon the position of extension.
`
`Structure–activity relationships of siRNAs with single strand
`extensions on both strands
`
`synthesized siRNAs with both strands
`Next, we
`extended in opposite directions, resulting in bidirectional
`single-strand extension structures (Fig. 2A and B). The
`30LS30LAS siRNA had extensions at the 30 end of both
`strands, while the 50LS50LAS siRNA had sequence exten-
`sions at the 50 end of both strands (Fig. 2A). We transfec-
`ted these variant siRNAs and measured their gene silencing
`efficiencies at different siRNA concentrations (Fig. 2C).
`Interestingly, the gene silencing efficiencies of the two var-
`iant siRNAs were dramatically different. We found that
`even at the highest concentration tested (100 nM), the
`silencing efficiency of 50LS50LAS was reduced (50%
`silencing), compared with control siRNAs. On the other
`hand, 30LS30LAS siRNA effectively silenced lamin gene
`expression at this concentration. At 1 nM concentration,
`the 50LS50LAS siRNA completely lost gene silencing
`
`

`

`1000
`
`C.I. Chang et al. / Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`A
`
`C
`
`3’LS + 3’LAS
`
`5’LS + 5’LAS
`
`5'LS 5'LAS
`
`3'LS 3'LAS
`
`27+0
`
`19+2
`
`19+0
`
`0nM
`
`B
`
`19+0 27+0
`
`3’LS 5’LS
`3’LAS 5’LAS
`
`1nM
`10nM
`100nM
`
`D
`
`5'LS 5'LAS
`
`3'LS 3'LAS
`
`19+0
`
`0
`
`2
`
`8
`6
`4
`OAS2 level
`
`10
`
`12
`
`0
`
`20
`
`80
`60
`40
`lamin level
`
`100
`
`120
`
`Fig. 2. Structures and activities of, and antiviral responses to, siRNAs with single strand extensions on both strands. (A) Structures of siRNAs with single
`strand extensions on both strands. Extended sequences are underlined. (B) Non-denaturing PAGE analysis of the siRNAs in (A). (C) Lamin gene silencing
`activity of the siRNAs in (A). See the legend of Fig. 1C for details. (D) Fold-induction of OAS2 mRNA levels by siRNAs (100 nM) with single strand
`extensions on both strands. Data were calculated as ratios of OAS2 mRNA level/GAPDH mRNA level. OAS2 induction by 19+0 siRNA was set at unity.
`All data in the graphs show means ± SD values from two independent experiments.
`
`activity (Fig. 2C) but 30LS30LAS siRNA still maintained
`some silencing activity, although silencing was somewhat
`lowered (40% silencing). The low silencing activity of
`the 50LS50LAS siRNA was not due to a non-specific gen-
`eral decrease in the mRNA level, which might have been
`triggered by a dsRNA-mediated antiviral response, because
`the GAPDH mRNA level was similar in different RNA
`samples (data not shown). These results again demonstrate
`that extensions of siRNA strands affect gene silencing dif-
`ferently, depending on the end to which the extra sequence
`is attached.
`
`Structure–activity relationships of siRNAs with double
`strand extensions
`
`We also tested the silencing efficiency of long duplex
`siRNAs generated by adding 11-bp duplexes of target-
`unrelated sequences to either end of the 19+0 lamin siRNA
`(Fig. 3A and B). The 50LS30LAS siRNA, in which the extra
`duplex sequence was appended to the 30 end of the anti-
`sense strand, showed silencing efficiency comparable to
`that of the control siRNAs at all concentrations tested
`(Fig. 3C). However, the 30LS50LAS structure, in which
`the extra duplex sequence was attached to the 50 antisense
`end, showed a different gene silencing profile. While the
`silencing efficiency of this siRNA at 100 nM concentration
`
`was similar to that of the control siRNAs, silencing effi-
`ciency was reduced at 10 nM, and was further reduced
`(20% silencing) at 1 nM (Fig. 3C). As with other siRNA
`structural variants studied, we thus observed strong posi-
`tion effects when extra double-strand sequences were
`appended to gene silencing siRNAs.
`
`Off-target gene silencing triggered by siRNA structural
`variants
`
`To test whether the attachment of extra, target-unre-
`lated sequence generates off-target gene silencing, that is,
`to trigger silencing of genes other than lamin A/C [13],
`we performed DNA microarray experiments using micro-
`arrays containing about 8000 human cDNAs [10]. We
`transfected HeLa cells individually with either 19+0 blunt
`siRNA as a control or siRNA structural variants such as
`30LS, 30LS30LAS, and 50LS30LAS siRNAs. Microarray
`analysis showed that when HeLa cells were transfected
`with 19+0 siRNA, only a few genes in addition to lamin
`A/C gene were down-regulated more than twofold (Supple-
`mentary Fig. 1). Similarly, when cells were transfected with
`each siRNA structural variants, only one or two genes
`other than lamin A/C gene were down-regulated more than
`twofold (Supplementary Fig. 1). Moreover, the fold reduc-
`tion of these down-regulated genes was only slightly more
`
`

`

`C.I. Chang et al. / Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`1001
`
`A
`3’LS + 5’LAS
`
`5’LS + 3’LAS
`
`C
`5'LS 3'LAS
`
`3'LS 5'LAS
`
`27+0
`
`19+2
`
`19+0
`
`0nM
`
`B
`
`19+0 27+0
`
`3’LS 5’LS
`5’LAS 3’LAS
`
`1nM
`10nM
`100nM
`
`0
`
`20
`
`40
`
`80
`60
`lamin level
`
`100
`
`120
`
`Fig. 3. Structures and activities of long duplex siRNAs. (A) Structures of long duplex siRNAs. Extended sequences are underlined. (B) Non-denaturing
`PAGE analysis of the siRNAs in (A). (C) Lamin gene silencing activity of siRNAs in (A). See the legend of Fig. 1C for details.
`
`than twofold. From these results, we conclude that the
`structural variation of siRNA by the attachment of extra
`target-unrelated sequences does not trigger off-target gene
`silencing.
`
`Innate immune responses triggered by siRNA structural
`variants
`
`Several dsRNA sensors exist in cells to detect foreign
`dsRNAs and to then elicit innate immune responses [14].
`OAS2 is one of the key genes located downstream of
`dsRNA-induced innate immune responses. The gene can
`be induced by long dsRNAs such as poly(I:C) [15]. We
`examined if any of the siRNA structural variants used in
`this work might significantly induce OAS2 gene expression.
`While the absolute level of OAS2 mRNA was somewhat
`variable from experiment to experiment, consistent with a
`previous report [15], we found that most of the siRNA
`variants we tested did not increase the OAS2 mRNA level
`when compared with the OAS2 mRNA level induced by
`19+0 control siRNA, as measured by quantitative real-
`time PCR (data not shown). Intriguingly, even 30-bp
`duplex siRNAs, such as 50LS30LAS and 30LS50LAS, did
`not induce OAS2 gene expression above the level induced
`by the control 19+0 siRNAs. On the other hand, at
`100 nM concentrations, siRNAs with single strand exten-
`sions on both strands, the 50LS50LAS and 30LS30LAS siR-
`NAs,
`induced OAS2 mRNA levels severalfold higher
`than that induced by 19+0 siRNA (Fig. 2D). The level of
`
`induction was much lower, however, than the OAS2 induc-
`tion seen with poly(I:C), and while poly(I:C) triggered non-
`specific reduction of the global mRNA level and cell death,
`the 50LS50LAS or 30LS30LAS siRNAs did not show any of
`these effects (data not shown). Also, at lower concentra-
`tions (10 or 1 nM), even 50LS50LAS and 30LS30LAS
`siRNAs did not trigger the induction of OAS2 mRNA
`level. These results suggest that the siRNA structural vari-
`ants used in this study do not trigger a strong dsRNA-med-
`iated antiviral innate immune response.
`Reynolds et al. [16] have shown that long (27 bp) siR-
`NAs could trigger induction of a number of genes involved
`in the antiviral innate immune response, in a cell type-spe-
`cific manner. It is to note that the HeLa cell line we have
`used in this study was identified as one of the cell lines
`which is insensitive to the treatment of long siRNAs [16].
`Therefore, the relative lack of innate immune response trig-
`gered by siRNA structural variants in our study is consis-
`tent with their observation. In fact,
`in our microarray
`data, none of the genes induced by the long siRNA in
`the study of Reynolds et al. [16] was significantly induced
`by siRNA variants tested, except a moderate increase
`(2.5-fold) in the IL-8 mRNA level by the 30LS siRNA
`(Supplementary Figs. 2 and 3).
`
`Dicer-mediated cleavage of siRNA structural variants
`
`Some of the siRNA structural variants we designed
`showed reduced or negligible silencing efficiencies. We
`
`

`

`1002
`
`C.I. Chang et al. / Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`wondered if defective gene silencing might be due to the
`inefficient recognition and cleavage of such siRNA struc-
`tural variants by Dicer. To test
`this, we performed
`in vitro Dicer-mediated cleavage reactions. First, each siR-
`NA variant was digested for 8 h with 1 U recombinant
`Dicer enzyme, and the cleavage products were analyzed
`by running a non-denaturing PAGE gel (Fig. 4A). Dicer
`cleaves dsRNA substrates
`to generate 19-
`to 21-bp
`duplexes [17]. As expected, the 27+0 duplex was cleaved
`by Dicer to generate products of a size similar to that of
`the 19+0 siRNA duplex (Fig. 4A). To our surprise,
`Dicer-mediated cleavage of the siRNA structural variants
`generated products of various electrophoretic mobilities
`(Fig. 4A), and most of the cleavage products migrated
`differently from the 19+0 duplex siRNA (Fig. 4A). While
`further characterization of each cleavage product
`is
`required, this result strongly suggests that Dicer-mediated
`cleavage of siRNA structural variants generates products
`whose structures are different from that of the 19-bp
`duplex.
`Each siRNA with a single strand extension on one
`strand showed a unique cleavage pattern (Fig. 4A). The
`cleavage product of 30LAS was similar in size to the
`19+0 siRNA and to the cleavage product of 27+0 siRNA.
`However, 30LS and 50LS yielded Dicer-mediated cleavage
`products of more rapid electrophoretic mobility than
`19+0 siRNA and 50LAS yielded the cleavage product of
`highest mobility (Fig. 4A and B). On the other hand, siR-
`NAs with single strand extensions at both strands, the
`30LS30LAS and 50LS50LAS siRNAs, produced cleavage
`products with slower electrophoretic mobilities than that
`of 19+0 siRNA (Fig. 4A and Supplementary Fig. 4A),
`and the cleavage patterns of the two siRNAs were different.
`The
`cleavage pattern of
`the
`long duplex siRNAs,
`30LS50LAS and 50LS30LAS, were also intriguing. While
`
`these two siRNAs share an identical structure (duplexes
`of 30 bp), and differ in only the site of appended extensions,
`they yielded Dicer-cleaved products of different sizes
`(Fig. 4A and Supplementary Fig. 4B). 30LS50LAS gave
`cleavage products migrating faster than 19+0, whereas
`50LS30LAS yielded species migrating slower than 19+0
`(Fig. 4A and Supplementary Fig. 4B). This result suggests
`that Dicer-mediated cleavage patterns may depend not
`only on the structures but also on the sequences of siRNAs.
`We then tested whether the efficiency of gene silencing
`by each siRNA variant correlated with the efficiency of
`cleavage by Dicer, by using varying concentrations of the
`Dicer enzyme (Fig. 4B and Supplementary Fig. 4). Within
`the range of Dicer concentrations used, we did not observe
`significant differences in cleavage efficiencies between siR-
`NA variants (Fig. 4B and Supplementary Fig. 4). From
`these results, we hypothesize that different gene silencing
`efficiencies shown by different siRNA variants may not
`be because of differences in efficiencies of Dicer-mediated
`cleavage, but rather because of unique patterns and/or
`sequences of cleaved products generated from each siRNA
`variant.
`Although there have been a number of studies regarding
`the effect of chemical modification on the activity of siR-
`NAs [3], there have been no reports to date on the effect
`of siRNA structural modification on silencing activity,
`off-target gene silencing, and cellular innate immune
`responses. To our knowledge, this study is the first system-
`atic analysis of structure–activity relationships of siRNAs
`with structural variations.
`Jiang et al. [9] designed an siRNA structural variant that
`had a single strand DNA extension at the 30 end of the
`sense strand. This variant siRNA structure was efficient
`in target gene silencing, and was utilized as a PCR primer
`to quantitatively analyze siRNA concentrations inside
`
`A
`
`19+0 27+0 3’LS 5’LS 3’LAS 5’LAS 19+0 27+0
`—
`—
`—
`—
`—
`—
`
`Dicer
`
`3’LS 5’LS 3’LS 5’LS
`5’LAS 3’LAS 3’LAS 5’LAS
`—
`—
`—
`—
`
`B
`
`19+0 27+0 3’LS 5’LS
`(Unit) : 0 1 1 0.3 0.1 0 1 0.3 0.1 0
`
`Dicer
`
`Fig. 4. In vitro cleavage of siRNAs by recombinant Dicer. siRNAs before and after treatment with Dicer were analyzed by 15% (w/v) non-denaturing
`PAGE. (A) Cleavage of siRNAs by 1 U Dicer enzyme for 8 h. (B) Cleavage of 30LS and 50LS siRNAs with various concentrations of Dicer enzyme.
`
`

`

`C.I. Chang et al. / Biochemical and Biophysical Research Communications 359 (2007) 997–1003
`
`1003
`
`cells. We point out that 30 sense strand extension was the
`only single strand extension modification in our study that
`did not affect siRNA gene silencing efficiency (Fig. 1C). In
`addition, McNamara et al. [7] attached aptamer sequences
`to the 50 end of siRNA sense strands, and observed gene
`silencing. While the site of single strand extension used in
`their experiment was not optimal, based on our data, they
`used very high concentrations of siRNA (400 nM) to
`observe silencing. Indeed, at 100 nM concentration, all
`four siRNAs with single strand extensions showed good
`silencing efficiency (Fig. 1C). At the high concentration
`of aptamer–siRNA complex, efficient gene silencing should
`therefore have been observed. We predict that aptamer
`fusion to the 30 end of the sense strand siRNA might work
`better in gene silencing, at lower concentrations of the
`RNA complex, provided that the site of aptamer attach-
`ment does not affect aptamer structure.
`In summary, our results provide three important rules
`for the design of modified siRNA structures. First, the 30
`end of the sense strand of an siRNA is the site of choice
`for a strand extension containing extra sequences to main-
`tain maximal gene silencing. Second, extensions of both
`strands of an siRNA should be 30, not 50, to maintain gene
`silencing activity. Finally, when duplex extension is
`designed, addition of extra duplex sequences to the 30
`end of the antisense strand of an siRNA is preferred over
`extension on the 50 end of the antisense strand, to minimize
`the loss of gene silencing efficiency. We believe our findings
`will provide useful guidelines in the design of siRNAs with
`novel structures or in the synthesis of complex RNA mol-
`ecules which harbor siRNA as one of several functional
`modules. Toward the comprehensive understanding of
`the action mechanism of siRNA structural variants, future
`studies will focus on the characterization of the Dicer
`cleavage products generated from each siRNA variants,
`and the cleavage efficiency of each siRNA by Dicer–TRBP
`complex [18].
`
`Acknowledgments
`
`D.-K.L. was supported by grants from the National
`R&D Program for Cancer Control, Ministry of Health
`and Welfare, Republic of Korea (Grant 0520200-2), the
`Basic Research Program of KOSEF (Grant R01-2005-
`000-10266-0), and the SRC/ERC program of MOST/KO-
`SEF (Grant R11-2000-070-080010). S.K. was supported
`by a National Research Laboratory grant from MOST/
`KOSEF.
`
`Appendix A. Supplementary data
`
`Supplementary data associated with this article can be
`found,
`in the online version, at doi:10.1016/j.bbrc.
`2007.06.004.
`
`References
`
`[1] S.M. Elbashir, J. Harborth, W. Lendeckel, A. Yalcin, K. Weber, T.
`Tuschl, Duplexes of 21-nucleotide RNAs mediate RNA interference
`in cultured mammalian cells, Nature 411 (2001) 494–498.
`[2] G.J. Hannon, RNA interference, Nature 418 (2002) 244–251.
`[3] Y. Dorsett, T. Tuschl, siRNAs: applications in functional genomics
`and potential as therapeutics, Nat. Rev. Drug Discov. 3 (2004) 318–
`329.
`[4] D.H. Kim, M.A. Behlke, S.D. Rose, M.S. Chang, S. Choi, J.J. Rossi,
`Synthetic dsRNA Dicer substrates enhance RNAi potency and
`efficacy, Nat. Biotechnol. 23 (2005) 222–226.
`[5] S.D. Rose, D.H. Kim, M. Amarzguioui, J.D. Heidel, M.A. Colling-
`wood, M.E. Davis, J.J. Rossi, M.A. Behlke, Functional polarity is
`introduced by Dicer processing of short substrate RNAs, Nucleic
`Acids Res. 33 (2005) 4140–4156.
`[6] D. Siolas, C. Lerner, J. Burchard, W. Ge, P.S. Linsley, P.J. Paddison,
`G.J. Hannon, M.A. Cleary, Synthetic shRNAs as potent RNAi
`triggers, Nat. Biotechnol. 23 (2005) 227–231.
`[7] J.O. McNamara 2nd, E.R. Andrechek, Y. Wang, K.D. Viles, R.E.
`Rempel, E. Gilboa, B.A. Sullenger, P.H. Giangrande, Cell type-
`specific delivery of siRNAs with aptamer–siRNA chimeras, Nat.
`Biotechnol. 24 (2006) 1005–1015.
`[8] A. Khaled, S. Guo, F. Li, P. Guo, Controllable self-assembly of
`nanoparticles for specific delivery of multiple therapeutic molecules to
`cancer cells using RNA nanotechnology, Nano Lett. 5 (2005) 1797–
`1808.
`[9] M. Jiang, A.A. Arzumanov, M.J. Gait, J. Milner, A bi-functional
`siRNA construct induces RNA interference and also primes PCR
`amplification for its own quantification, Nucleic Acids Res. 33 (2005)
`e151.
`[10] J.W. Yoo, S.W. Hong, S. Kim, D.K. Lee, Inflammatory cytokine
`induction by siRNAs is cell type- and transfection reagent-specific,
`Biochem. Biophys. Res. Commun. 347 (2006) 1053–1058.
`[11] J.T. Marques, T. Devosse, D. Wang, M. Zamanian-Daryoush, P.
`Serbinowski, R. Hartmann, T. Fujita, M.A. Behlke, B.R. Williams, A
`structural basis for discriminating between self and nonself double-
`stranded RNAs in mammalian cells, Nat. Biotechnol. 24 (2006) 559–
`565.
`inflammatory cytokines and interferon
`[12] M. Sioud, Induction of
`responses by double-stranded and single-stranded siRNAs
`is
`sequence-dependent and requires endosomal
`localization, J. Mol.
`Biol. 348 (2005) 1079–1090.
`[13] A.L. Jackson, S.R. Bartz, J. Schelter, S.V. Kobayashi, J. Burchard,
`M. Mao, B. Li, G. Cavet, P.S. Linsley, Expression profiling reveals
`off-target gene regulation by RNAi, Nat. Biotechnol. 21 (2003) 635–
`637.
`[14] E. Meylan, J. Tschopp, Toll-like receptors and RNA helicases: two
`parallel ways to trigger antiviral responses, Mol. Cell 22 (2006) 561–
`569.
`[15] C. Dahlgren, C. Wahlestedt, H. Thonberg, No induction of anti-viral
`responses in human cell lines HeLa and MCF-7 when transfecting
`with siRNA or siLNA, Biochem. Biophys. Res. Commun. 341 (2006)
`1211–1217.
`[16] A. Reynolds, E.M. Anderson, A. Vermeulen, Y. Fedorov, K.
`Robinson, D. Leake, J. Karpilow, W.S. Marshall, A. Khvorova,
`Induction of the interferon response by siRNA is cell type- and
`duplex length-dependent, RNA 12 (2006) 988–993.
`[17] E. Bernstein, A.A. Caudy, S.M. Hammond, G.J. Hannon, Role for a
`bidentate ribonuclease in the initiation step of RNA interference,
`Nature 409 (2001) 363–366.
`[18] T.P. Chendrimada, R.I. Gregory, E. Kumaraswamy, J. Norman, N.
`Cooch, K. Nishikura, R. Shiekhattar, TRBP recruits the Dicer
`complex to Ago2 for microRNA processing and gene silencing,
`Nature 436 (2005) 740–744.
`
`