Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`BBRC
`
`www.elsevier.com/locate/ybbrc
`
`Chemical modification of siRNAs to improve serum
`stability without loss of efficacy
`
`Sorim Choung a, Young Joo Kim b, Seonhoe Kim a, Han-Oh Park a, Young-Chul Choi a,*
`
`a Gene2Drug Research Center, Bioneer Corporation, 49-3, Munpyeong-dong, Daedeok-gu, Daejeon 306-220, Republic of Korea
`b National Genome Information Center, KRIBB, 52 Eoeun-dong, Yuseong-gu, Daejeon 305-333, Republic of Korea
`
`Received 4 February 2006
`Available online 20 February 2006
`
`Abstract
`
`Development of RNA interference as a novel class of therapeutics requires improved pharmacokinetic properties of short interfering
`RNA (siRNA). To confer enhanced serum stability to Sur10058, a hyperfunctional siRNA which targets survivin mRNA, a systematic
`modification at the 20-sugar position and phosphodiester linkage was introduced into Sur10058. End modification of three terminal
`nucleotides by 20-OMe and phosphorothioate substitutions resulted in a modest increase in serum stability, with 30 end modification
`being more effective. Alternating modification by 20-OMe substitution significantly stabilized Sur10058, whereas phosphorothioate mod-
`ification was only marginally effective. Through various combinations of 20-OMe, 20-F and phosphorothioate modifications that were
`directed mainly at pyrimidine nucleotides, we have identified several remarkably stable as well as efficient forms of Sur10058. Thus,
`our results provide an effective means to stabilize siRNA in human serum without compromising the knockdown efficiency. This
`advancement will prove useful for augmenting the in vivo potency of RNA interference.
`Ó 2006 Elsevier Inc. All rights reserved.
`
`Keywords: Survivin; siRNA; Chemical modification; Stability
`
`Gene silencing by RNA interference (RNAi) has
`emerged as a powerful method that is useful for the
`study of functional genomics, drug target identification
`and validation, and therapeutic development
`[1,2]. In
`the process of RNAi, double stranded RNA is first
`cleaved into 21–25 nucleotide (nt) RNA duplexes by
`the RNase III enzyme, named dicer [3–5]. The resulting
`siRNAs are then associated with a protein complex
`which is called the RNA-induced silencing complex
`(RISC) [6]. The siRNA duplex is unwound by the puta-
`tive helicase in RISC through an ATP-dependent process
`[7] and the single stranded RNA subsequently guides the
`activated RISC to homologous mRNA targets. After
`binding to target mRNA, the Ago2 protein in the RISC
`acts as a slicer and functions to cleave target mRNA
`
`* Corresponding author. Fax: +82 42 930 8600.
`E-mail address: ycchoi@bioneer.co.kr (Y.-C. Choi).
`
`0006-291X/$ - see front matter Ó 2006 Elsevier Inc. All rights reserved.
`doi:10.1016/j.bbrc.2006.02.049
`
`[8–10]. Degradation of cleaved target RNA is accom-
`plished by ribonuclease activity as a result of the lack
`of protection by 50 caps or poly(A) tails.
`experiments
`The
`execution of
`successful RNAi
`depends upon multiple factors. Specifically it is impor-
`tant to: (1) design and identify effective and specific siR-
`NA sites,
`(2)
`enhance pharmacokinetic properties,
`including serum stability, and (3) to perform efficient
`and specific delivery of siRNA to the desired target cell
`types. As a first step toward developing siRNA therapeu-
`tics, we have screened 82 siRNAs which specifically tar-
`get survivin. This important protein has been previously
`implicated in both anti-apoptosis and the regulation of
`cytokinesis [11–15]. Screening of 82 siRNAs resulted in
`the identification of several highly effective siRNA inhib-
`itors including Sur10058. The identified inhibitors were
`hyperfunctional and were capable of silencing at
`the
`in vitro (IC50 values are below
`sub-nanomolar level
`100 pM). Due to the potential promise of Sur10058 as
`
`Alnylam Exh. 1072
`
`

`

`920
`
`S. Choung et al. / Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`an effective siRNA inhibitor, we were interested to study
`this particular siRNA in detail. In order to develop
`Sur10058 as an efficient anti-cancer agent, we subse-
`quently introduced various chemical modifications into
`Sur10058 and tested their effects on the serum stability
`of Sur10058. Since unmodified siRNA duplexes are high-
`ly unstable especially in serum,
`it is often desirable to
`enhance serum and intracellular
`stability of
`siRNA
`duplexes by chemical modification as an effective means
`to increase in vivo potency.
`In addition to modification at phosphodiester linkages
`by phosphorothioate (PS) substitution, chemical modifica-
`tions, such as 20-sugar modification by 20-OMe, 20-F, and
`locked nucleic acid (LNA) substitutions have been widely
`used to enhance serum and intracellular stability of siRNA
`[16–23]. These studies have shown that in principle, chem-
`ical modification is capable of enhancing the serum stabil-
`ity of siRNA without compromising and affecting the
`functionality of siRNA. Furthermore, it has been reported
`that when transfected into cells, modified siRNA was capa-
`ble of maintaining the knockdown effect for a longer period
`as compared to unmodified siRNA [17,21]. In the case of
`in vivo experiments, when a fully modified siRNA target-
`ing hepatitis B virus was introduced into mice by the
`hydrodynamic injection method, the modified siRNA was
`significantly more effective in reducing serum HBV DNA
`and hepatitis B surface antigen than the unmodified siRNA
`[24]. In addition, when modified siRNA was formulated
`into a stable nucleic-acid-lipid particle (SNALP) by using
`a liposome and was injected into mice intravenously, the
`modified siRNA provided a high improvement of in vivo
`stability as compared to unmodified siRNA. It should be
`particularly noted that unmodified siRNA caused an
`immune stimulation, whereas modified siRNA did not
`induce interferon a and inflammatory cytokines [25]. Tak-
`en together, these results have clearly demonstrated that
`the use of modified siRNA is capable of greatly improving
`pharmacokinetic properties.
`In this study, we have introduced various combinations
`of chemical modifications into Sur10058 in order to identi-
`fy the modified structures which can greatly enhance its
`serum stability without reducing its knockdown efficiency.
`In contrast to the unmodified Sur10058 which was rapidly
`degraded in human serum, we have identified several mod-
`ified structures of Sur10058 which exhibited remarkably
`enhanced serum stability. Collectively, our results present-
`ed here will serve as a useful guideline for the design of sta-
`ble siRNA.
`
`Materials and methods
`
`RNA and DNA oligonucleotides. 20-OMe-, 20-F- and phosphorothio-
`ate-modified siRNA duplexes were synthesized by Proligo (Boulder, CO)
`and IDT (Coralville, IA). All siRNA duplexes have the sense strand
`sequence of 50-AAG GAG AUC AAC AUU UUC A dTdT-30. The 19
`nucleotides represent the target sequence in survivin (NM_001168) and
`dTdT is a 2-nt 30 overhang. The antisense strand was composed of
`nucleotides that are complementary to the target sequence and the dTdT
`
`30 overhang sequence. All siRNAs were resuspended in DEPC-treated
`water to give the final concentration of 30 lM. DNA oligonucleotides
`were obtained from Bioneer (Daejeon, Republic of Korea).
`Cell culture. HeLa cells were maintained at 37 °C under 5% CO2 in
`RPMI 1640 medium (Gibco-BRL/Invitrogen, Carlsbad, CA) supple-
`mented with 10% fetal bovine serum and antibiotics (100 U of penicillin
`per ml and 100 lg of streptomycin per ml).
`Transfection. A single day prior to transfection, HeLa cells were plated
`in 6-well culture dishes at a density of 2.5 · 105 cells per well.Transfection
`was performed using Lipofectamine 2000 reagent (Invitrogen, Carlsbad,
`CA) with methods as recommended by the manufacturer. Briefly, cells
`were washed once with Opti-MEM (Gibco-BRL/Invitrogen, Carlsbad,
`CA) and 500 ll of Opti-MEM was added to each well. For each trans-
`fection, 3.5 ll of lipofectamine was mixed with 250 ll of Opti-MEM and
`incubated for 5 min at room temperature. In a separate tube, siRNA was
`added to 250 ll of Opti-MEM and the solution was added to the lipo-
`fectamine mixture. In order to allow complex formation, the siRNA
`mixture was incubated for an additional 20 min at room temperature. The
`solution was subsequently added to the cells in the 6-well plate, giving an
`end volume of 1 ml. Cells were incubated at 37 °C in the presence of the
`transfection solution for 6 h. The Opti-MEM medium containing the
`complexes was then replaced with 2 ml of standard growth media and
`cultured at 37 °C for 18 h. Twenty-four hours after transfection, total
`RNA was isolated from transfected cells by using the RNeasy mini kit
`(Qiagen, Hilden, Germany).
`Northern blot analysis. Northern blot analysis was performed as pre-
`viously described [26]. Briefly, total RNA (8 lg per lane) was fractionated
`on a 1% agarose gel containing formaldehyde. Resolved RNA was blotted
`onto Hybond-N+ membrane (Amersham Biosciences, Piscataway, NJ) via
`capillary transfer with 20· SSC. After UV cross-linking, blots were
`maintained at 65 °C in a prehybridization buffer (1% bovine serum albu-
`min, 2 mM EDTA, 0.5 M sodium phosphate, pH 7.2, and 7% NaDodSO4)
`for 1 h [27]. Blots were subsequently hybridized with a 32P-labeled cDNA
`probe for 18 h. The blot was washed twice at room temperature for 10 min
`in 1· SSC, 0.1% NaDodSO4, and subsequently, at 65 °C for 1 h in 0.5·
`SSC, 0.1% NaDodSO4. During the initial hybridization, the membranes
`were first probed with survivin cDNA and were subsequently rehybridized
`with positive control b-actin cDNA which effectively monitored the
`loading of total RNA in each lane. The probes were synthesized from
`832 bp within the survivin cDNA (Accession No. NM_001168; nucleotide
`756–1587) and 788 bp of the b-actin cDNA (Accession No. NM_001101;
`nucleotide 433–1220) using a random primer labeling kit (Stratagene, La
`Jolla, CA), and [a-32P]dCTP (Amersham Biosciences, Piscataway, NJ)
`[28].
`Real-time quantitative PCR analysis. Triplicate real-time PCR were
`performed in 384-well plates using an Applied Biosystems Prism 7900
`Sequence Detection System (Applied Biosystems). Each reaction mixture
`(20 ll) consisted of 10 ll of SYBR Green Master Mix (Applied Biosys-
`tems) and 0.4 ll of forward and reverse primers (10 pmol/ll) that were
`specific for survivin and GAPDH. PCR conditions were as follows; initial
`incubation at 50 °C for 2 min and 95 °C for 10 min followed by 40 cycles
`of 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 30 s. The sequences of the
`primers specific for survivin were: forward primer, 50-GCACCACTTC
`CAGGGTTTAT-30;
`reverse primer, 50-CTCTGGTGCCACTTTCAA
`GA-30. The sequence for the GAPDH primers were 50-TGCACCACCA
`ACTGCTTAGC-30 (forward primer), and 50-GGCATGGACTGTGGTC
`ATGAG-30 (reverse primer). Serial dilutions (1, 0.5, and 0.25) of control
`cDNA were included in order to generate relative standard curves for
`survivin and GAPDH. The amount of survivin transcript was divided by
`that of GAPDH as a means to obtain normalized expression levels.
`Serum stability. siRNA duplex (9 lg) was incubated at 37 °C in 90 ll of
`10% human serum. Aliquots of 12 ll were taken at various time points and
`immediately frozen at 70 °C. Samples (2.5 ll) were analyzed on 15%
`non-denaturing polyacrylamide gels in TBE buffer and were visualized on
`a UV-transilluminator subsequent to EtBr staining.
`Nuclear staining. HeLa cells (100,000 cells/well) were plated in 6-well
`culture dishes and the next day, cells were transfected with unmodified and
`modified Sur10058 siRNAs (2 nM). Four days after transfection, cells
`
`

`

`S. Choung et al. / Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`921
`
`were stained with DAPI (Roche Applied Science, Indianapolis, IN) at a
`concentration of 1 lg/ml dissolved in methanol for 15 min at 37 °C.
`Afterward, cells were washed once with methanol, 1 ml of PBS was added
`to each well, and DAPI-stained nuclei were observed by using a Nikon
`Eclipse TS100 epifluorescence microscope (Nikon, Tokyo, Japan) equip-
`ped with a ProgRes C10 digital camera (Jenoptik laser, Eching/Munich,
`Germany).
`Western blot analysis. HeLa cells (100,000 cells/well) were plated in 6-
`well culture dishes and the next day, cells were transfected with unmodified
`and modified Sur10058 siRNAs (2 nM). Seventy-two hours after trans-
`fection, cells were harvested and total protein extracts were prepared using
`PRO-PREP protein extraction solution (Intron Biotechnology, Seong-
`nam, Republic of Korea) following the manufacturer’s instruction. Total
`protein extracts (40 lg) were electrophoresed on 4–20% precast protein gel
`(Pierce, Rockford, IL) and transferred to Protran Nitrocellulose mem-
`brane (S&S, Dassel, Germany). Immunoblotting was carried out with an
`anti-survivin monoclonal antibody D-8 (Santa CruZ, CA) using the rabbit
`anti-mouse IgG/HRP (Pierce, Rockford, IL) as a secondary antibody.
`Peroxidase activity was detected by using the ECL plus Western blotting
`reagent (Amersham Biosciences, Piscataway, NJ).
`
`Results and discussion
`
`Effect on stability and RNAi activity of 50 and 30 end
`modification
`
`We were interested to identify specific chemical modifi-
`cations which affect the structure of Sur10058 and function
`to greatly enhance its serum stability. In our first line of
`investigation, we examined the effect of nucleases on the
`stability of siRNA in human serum by introducing 20-
`OMe or PS modification into three terminal nucleotides
`at 50 and 30 ends of Sur10058 (Fig. 1).
`The unmodified Sur10058 was very unstable in 10%
`human serum. As a result, unmodified Sur10058 was most-
`ly degraded after 1 h incubation at 37 °C (Fig. 2B). When
`three terminal nucleotides at the 50 end were modified with
`20-OMe or PS substitution, the stability of modified siRNA
`was not substantially increased (Fig. 2B, Sur10058-PS-5,
`and -Me-5). On the other hand, 30 modifications to the
`
`Fig. 1. Structure of modified RNAs used in this study. Modified moieties
`are boxed.
`
`end of the siRNA resulted in significant changes in stability.
`Specifically, the intact siRNA was retained at significant
`levels subsequent to 3 h incubation at 37 °C. Furthermore,
`intact siRNA was also detected even after 6 h incubation
`in serum (Fig. 2B, Sur10058-Me-3, and -Me-53). These
`the 30
`terminal modification is
`results suggested that
`more efficient
`for enhancing the serum stability of
`Sur10058 and supports the supposition that 30-specific exo-
`nuclease will act on the degradation of Sur10058 in human
`serum.
`
`Alternating modification significantly stabilizes siRNA
`
`Because the modification of three terminal nucleotides
`showed only some increase in the serum stability, addition-
`al nucleotides along the length of siRNA were modified
`with 20-OMe or PS substitution. In the previous study, it
`has been reported that alternating modification with 20-
`OMe could increase the serum stability of siRNA without
`reducing their knockdown efficiency [17]. Accordingly, we
`have examined the effect of alternate modification with
`20-OMe or PS substitutions on the serum stability of
`Sur10058.
`As shown in Fig. 3B (Sur10058-Me-33), when a stretch
`of three nucleotides was modified by 20-OMe alternately
`with two or three unmodified nucleotides, intact siRNA
`was present at a significant amount even after a 24 hr incu-
`bation period in 10% human serum. The modified
`Sur10058-Me-33 has the structure wherein 20-OMe-modi-
`fied nucleotides of sense and antisense strands face one
`another. In the case of Sur10058-Me-23, the antisense
`strand is identical to that of Sur10058-Me-33 but 20-
`OMe-modified nucleotides of the sense strand are present
`alternately with those of the antisense strand. As demon-
`strated in Fig. 3B, Sur10058-Me-23 was less stable than
`Sur10058-Me-33. This reduction in stability was most
`likely due to the small number of 20-OMe-modified nucle-
`otides of the sense strand in Sur10058-Me-23 (the number
`of 20-OMe-modified nucleotides was reduced from 12 in
`Sur10058-Me-33 to 7 in Sur10058-Me-23). Alternatively,
`this decrease in stability could have resulted from a differ-
`ence in RNA structures between them.
`Because Sur10058-Me-23 and Sur10058-Me-33 showed
`a significantly reduced knockdown efficiency as compared
`to unmodified Sur10058 (Figs. 3C and D), we next exam-
`ined the structure of siRNAs wherein modified nucleotides
`are alternately present at every other position along the
`length of siRNA. Sur10058-Me-AL1 to AL4 have struc-
`tures which differ in the position of nucleotide to initiate
`20-OMe modification and Sur10058-PS-AL1 contains alter-
`nating modification with PS substitution (Fig. 3A). As it
`can be seen in Fig. 3B, all four of the siRNAs that were
`alternately modified with 20-OMe showed a great improve-
`ment in stability. In contrast to these data, modifications
`with PS substitution to Sur10058-PS-AL1 did not exhibit
`the substantial enhancement in serum stability (Fig. 3B).
`Since it was reported that the extensive modification with
`
`

`

`922
`
`S. Choung et al. / Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`B
`(Hr)
`
`0 1 3 6 24
`
`A
`
`1) Sur10058
`5’
`aaggagaucaacauuuuca dTdT 3 ’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`2) Sur10058-PS-5
`5’
`asasgs
`asasgsgagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaa sasgsu 5 ’
`
`3) Sur10058-PS-3
`5’
`aaggagaucaacauuususcsa dTdT 3’
`3’ dTdT ususcscucuaguuguaaaagu 5 ’
`
`4) Sur10058-PS-53
`5’
`asasgsgagaucaacauuususcsa dTdT 3’
`3’ dTdT ususcscucuaguuguaaasasgsu 5 ’
`
`5) Sur10058-Me-5
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`
`6) Sur10058-Me-3
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`7) Sur10058-Me-53
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`
`D
`NC
`
`1 2 3 4 5 6 7
`
`Survivin
`
`Actin
`
`120
`100
`80
`60
`40
`20
`0
`
`C
`
`mRNA Expression %
`
`NC 1
`2
`3
`4
`5
`6
`7
`Fig. 2. Stability and knockdown efficiency of siRNAs modified at 50 and 30 terminal three nucleotides by 20-OMe and PS substitutions. (A) Name,
`sequence, and chemical modifications of unmodified and modified Sur10058. Lowercase letters = unmodified RNA, underlined bold letters = 20-OMe,
`small underlined s between nucleotides = phosphorothioate. (B) Serum stability of siRNAs. Unmodified and modified siRNAs were incubated in 10%
`human serum at 37 °C for 0, 1, 3, 6, or 24 h and aliquots were analyzed on 15% native polyacrylamide gels. (C). Inhibition of survivin mRNA expression in
`HeLa cells transfected with modified Sur10058 siRNAs (2 nM). Knockdown efficiency was quantitatively measured by real-time PCR analysis. Survivin
`mRNA expression levels were obtained subsequent to normalization with constitutively expressed GAPDH mRNA. (D) Northern blot analysis of total
`RNA extracts. Non-silencing siRNA (NC) and unmodified Sur10058 were transfected along with modified siRNAs to serve as negative and positive
`controls, respectively. Blots were first probed with survivin cDNA and were subsequently reprobed with b-actin cDNA for visualization of lane-to-lane
`variation.
`
`PS caused a cytotoxicity, it appears that PS modification
`should only be used with careful attention and the appro-
`priate precautionary measures [16,19].
`Interestingly, AL2 and AL4 showed an approximate 2-
`fold decrease in their knockdown efficiency, whereas the
`silencing efficiency of AL1 and AL3 was comparable to
`unmodified Sur10058. Consistent with a previous report,
`these data showed that the alternate modifications which
`initiate from the 50 terminal nucleotide of antisense strand
`RNA provide better knockdown efficiency than the alter-
`nate modifications which initiate from the second nucleo-
`tide at the 50 end of antisense strand [17].
`
`Enhanced stability of siRNAs containing fully modified sense
`strand
`
`Since it has been reported that modification to the sense
`strand is better tolerated than the modification of the anti-
`sense strand, we next examined the stability and efficiency
`of siRNAs containing the sense strand fully modified with
`
`20-OMe. As can be seen in Fig. 4B (Sur10058-Me-S), when
`only two 30 terminal nucleotides of the antisense strand
`were modified, the stability was not greatly increased in
`comparison to the alternating modification. Even in the
`case of Sur10058-Me-S-AS, wherein the antisense strand
`was modified with nine 20-OMe pyrimidine nucleotides,
`the stability in human serum was not greatly increased
`when compared to Sur10058-Me-S. In Sur10058-Me-S-
`AS, 9th and 10th uridine nucleotides from the 50 end of
`antisense strand were not modified since the cleavage of
`target mRNA occurs in the center of the antisense strand
`[29,30]. As a result, there is a possibility that 20-OMe mod-
`ification in that region may inhibit the target RNA cleav-
`age reaction.
`In addition to 20-OMe modification, 20-F and PS modi-
`fications were introduced into the antisense strand to give
`rise to Sur10058-Me-F and Sur10058-Me-SP. As it can be
`seen in Fig. 4B, the stability of both siRNA duplexes was
`greatly increased and significant fractions of intact siRNA
`remained even after 48 h incubation in human serum.
`
`

`

`S. Choung et al. / Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`923
`
`B
`
`(Hr)
`
`0 1 3 6 24
`
`1) Sur10058
`5’
`aaggagaucaacauuuuca dTdT 3 ’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`2) Sur10058-Me-33
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`3) Sur10058-Me-23
`aaggagaucaacauuuuca dTdT 3 ’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`
`4) Sur10058-Me-AL1
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`5) Sur10058-Me-AL2
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`6) Sur10058-Me-AL3
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`
`7) Sur10058-Me-AL4
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`8) Sur10058-PS-AL1
`asagsgasgasucsaascasuusuusca dTdT 3’
`5’
`3’ dTdT uusccsucsuasgusugsuasaasagsu 5 ’
`
`D
`NC
`
`1 2 3 4 5 6 7 8
`
`Survivin
`
`Actin
`
`A
`
`120
`100
`80
`60
`40
`20
`0
`
`C
`
`mRNA Expression %
`
`NC 1
`2
`3
`4
`5
`6
`7
`8
`Fig. 3. Enhanced stability of siRNAs modified by 20-OMe and PS substitutions at alternating nucleotide positions. (A) Name, sequence, and chemical
`modifications of unmodified and modified siRNAs. Lowercase letters = unmodified RNA, underlined bold letters = 20-OMe, small underlined s between
`nucleotides = phosphorothioate. (B) Serum stability of siRNAs. Unmodified and modified siRNAs were incubated in 10% human serum at 37 °C for 0, 1,
`3, 6, or 24 h. Aliquots were withdrawn at indicated time points and were analyzed by electrophoresis on 15% native polyacrylamide gels. (C) Knockdown
`efficiency was quantitatively measured by Real-time PCR analysis and survivin expression levels were normalized to those of GAPDH. (D) Potency of
`modified siRNAs targeting survivin. HeLa cells were transfected with modified siRNAs (2 nM), total RNA was extracted, and subjected to Northern blot
`analysis. Non-silencing siRNA (NC) and unmodified Sur10058 were included as negative and positive controls, respectively. A probe specific for b-actin
`was used to monitor equal loading between lanes.
`
`Furthermore, the knockdown efficiency was comparable to
`that of unmodified Sur10058 in both cases. On the other
`hand, the modification of 9 pyrimidine nucleotides of anti-
`sense strand with 20-OMe in Sur10058-Me-S-AS resulted in
`significant decrease in the gene-silencing efficiency (Figs.
`4C and D). These results contrast to the previous data
`which showed that the modification of either sense or anti-
`sense strand fully with 20-OMe substitution made siRNA
`non-functional [17,29]. Even in case where the sense strand
`was fully modified with 20-OMe, the knockdown efficiency
`was not reduced if the antisense strand was modified with
`20-F or PS substitutions. However, when both the sense
`strand and the antisense strand were extensively modified
`with 20-OMe, as in Sur10058-Me-S-AS, the knockdown
`efficiency was greatly reduced. Consistent with our results,
`it was recently reported that one out of four siRNAs tar-
`geting PTEN mRNA was as potent as unmodified siRNA
`duplex even when the sense strand was fully modified with
`20-OMe [31]. Interestingly, full modification of sense strand
`with 20-OMe was better tolerated in 20-bp blunt-end
`
`siRNA duplexes targeting the same four sites and in some
`cases, complete modification with PS in either or both
`strand was tolerated. Taken together, these results demon-
`strated that the extensive modification of the sense strand
`with 20-OMe is a viable option and is dependent upon
`the siRNA structure and the types of chemical modifica-
`tion that are directed towards the complementary antisense
`strand.
`
`Identification of the stable as well as efficient modification
`through various combinations of pyrimidine modifications
`
`Since RNA can be sufficiently stabilized by only modify-
`ing pyrimidine residues [20,21], we examined the stability
`and knockdown efficiency of siRNAs modified with 20-F-
`pyrimidine and 20-OMe-pyrimidine nucleotides at various
`positions along the entire length of siRNA duplexes.
`As shown in Fig. 5B, in the case where all pyrimidine
`nucleotides of the sense and antisense strands were substi-
`tuted with 20-OMe-pyrimidine,
`the serum stability of
`
`

`

`924
`
`S. Choung et al. / Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`A
`1) Sur10058
`5’
`aaggagaucaacauuuuca dTdT 3 ’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`2) Sur10058-Me-S
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`B
`
`(Hr)
`
`0 1 3 6 24
`
`36 48
`
`3) Sur10058-Me-SP
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT ususcscsuscsusasgsususgsusasasasasgsu 5’
`
`4) Sur10058-Me-S-AS
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucu aguuguaaaagu
`5’
`5) Sur10058-Me-F
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`
`D
`
`NC
`
`1 2 3 4 5
`
`Survivin
`
`Actin
`
`120
`100
`80
`60
`40
`20
`0
`
`C
`
`mRNA Expression %
`
`NC
`1
`2
`3
`4
`5
`Fig. 4. Enhanced stability of modified siRNAs containing a fully modified sense strand. Sense strand RNA, which was fully modified with 20-OMe
`substitution, was duplexed with 20-OMe-, 20-F-, and PS-modified antisense strand RNA. (A) Name, sequence, and chemical modifications of unmodified
`and modified siRNAs. Lower case letters = unmodified RNA, underlined bold letters = 20-OMe, italic letters = 20-F, small underlined s between
`nucleotides = phosphorothioates. (B) Serum stability. Unmodified and modified siRNAs were incubated in 10% human serum at 37 °C for 0, 1, 3, 6, 24,
`36, and 48 h. Aliquots were analyzed by electrophoresis on 15% native polyacrylamide gels. (C) Potency of modified siRNAs targeting survivin was
`analyzed by real-time PCR analysis. Survivin expression levels were normalized to those of GAPDH. (D) HeLa cells were transfected with siRNAs (2 nM).
`Twenty-four hours post-transfection, total RNA was isolated and subjected to Northern blot analysis. Non-silencing siRNA (NC) and unmodified
`Sur10058 served as negative and positive controls, respectively. A b-actin probe was used as a control to monitor equal loading of samples between lanes.
`
`Sur10058-Me-Py1 was greatly increased. It is important to
`note that even after a 48 h incubation in human serum,
`most siRNA duplexes remained intact. However, concom-
`itant to this remarkable increase in stability, a great reduc-
`tion in the knockdown efficiency was observed (Figs. 5C
`and D). These data were consistent
`to the case of
`Sur10058-Me-S-AS wherein the efficiency was greatly
`reduced when both the sense and antisense strands were
`extensively modified with 20-OMe.
`Since the modification of the antisense strand has a
`greater effect on the reduction of knockdown efficiency,
`the number of 20-OMe pyrimidine in the antisense strand
`was reduced from a total of 11 in Sur10058-Me-Py1 to 5
`in Sur10058-Me-Py2. In this particular case, the silencing
`efficiency was vastly improved, but at the same time, the
`stability of Sur10058-Me-Py2 was reduced. In this case,
`intact siRNA was not substantially evident subsequent to
`a 24 h incubation in serum (Fig. 5B). With the exception
`of 20-OMe pyrimidines, even when the remaining nucleo-
`tides were modified with PS substitution, there was no evi-
`dent increase in stability (Fig. 5B; Sur10058-Me-Py3).
`Since the cleavage of target mRNA occurs in the center
`portion of antisense strand, every pyrimidine residue except
`the 9th and 10th uridine of the antisense strand was mod-
`ified with 20-OMe in Sur10058-Me-Py4. In this case, the
`stability was enhanced but the knockdown efficiency was
`
`greatly reduced; bearing similarity to the other cases where
`both sense and antisense strands were extensively modified
`with 20-OMe.
`However, when the antisense strand was modified with
`20-F-pyrimidine in Sur10058-Me-Py5 and Sur10058-Me-
`Py6, the stability was greatly increased and these modifica-
`tions did not reduce knockdown efficiency. It is important
`to note that these modifications to the antisense strand did
`not have deleterious effects on knockdown ability like the
`case of Sur10058-Me-F which contained a completely mod-
`ified sense strand. Taken together, these results clearly indi-
`cated that 20-F and PS modifications were better tolerated
`in antisense strand than 20-OMe modification, especially
`when the sense strand was extensively modified with 20-
`OMe substitutions.
`
`Stabilized Sur10058 siRNAs are as potent as unmodified
`Sur10058
`
`To compare knockdown efficiencies more accurately,
`transfection was carried out with much lower concentra-
`tions of siRNA duplexes. Considering both increased sta-
`bility and maintenance of knockdown efficiency, we have
`chosen Sur10058-Me-SP, Sur10058-Me-F, Sur10058-Me-
`Py5, and Sur10058-Me-Py6 for titration experiments.
`Although Sur10058-Me-Py1 was the most stable structure,
`
`

`

`S. Choung et al. / Biochemical and Biophysical Research Communications 342 (2006) 919–927
`
`925
`
`A
`
`B
`(Hr)
`
`0 1 3 6 24
`
`36 48
`
`1) Sur10058
`5’
`aaggagaucaacauuuuca dTdT 3 ’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`2) Sur10058-Me-Py1
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucu aguuguaaaagu
`5’
`3) Sur10058-Me-Py2
`aaggagaucaacauuuucsa dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu 5 ’
`
`4) Sur10058-Me-Py3
`asasgsgsasgsaucaacauuuucsa dTdT 3’
`5’
`3’ dTdT uuccucusasgsususgsusasasasasgsu 5 ’
`
`5) Sur10058-Me-Py4
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucu aguuguaaaagu
`5’
`
`6) Sur10058-Me-Py5
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ dTdT uuccucuaguuguaaaagu
`5’
`
`7) Sur10058-Me-Py6
`aaggagaucaacauuuuca dTdT 3’
`5’
`3’ idTdT uuccucuaguuguaaaagu
`5’
`
`D
`
`NC
`
`1 2 3 4 5 6 7
`
`Survivin
`
`Actin
`
`120
`100
`80
`60
`40
`20
`0
`
`C
`
`mRNA Expression %
`
`NC
`1
`2
`3
`4
`5
`6
`7
`Fig. 5. Systematic analysis of pyrimidine-modified siRNAs. Various combinations of modifications including 20-OMe, 20-F, and PS substitutions at
`pyrimidine nucleotides were used to identify stable as well as efficient modifications of Sur10058. (A) Name, sequence, and chemical modifications of
`Sur10058. Lowercase letters = unmodified RNA, underlined bold letters = 20-OMe, italic letters = 20-F, small underlined s between nucleotides = phosp-
`horothioates, idT = inverted deoxythymidine in Sur10058-Me-Py6. (B) Serum stability. Unmodified and modified Sur10058 siRNAs were incubated in
`10% human serum at 37 °C for 0, 1, 3, 6, 24, 36, and 48 h. Aliquots were analyzed by electrophoresis on 15% native polyacrylamide gels. Lower bands
`appearing in Sur10058-Me-Py2, -Py3, -Py4, and -Py5 after hours of incubation in serum seem to be a degradation product. It was not a denatured RNA
`molecule because single stranded RNA moved much faster than the lower band under our condition. In the case of Sur10058-Me-Py6 containing inverted
`deoxythymidine on 30-end of antisense strand, a lower band appeared immediately after incubation in serum, indicating that it is not a degradation
`product nor a denatured single stranded RNA. It is likely that an inverted deoxythymidine residue may have something to do with altered mobility of
`Sur10058-Me-Py6 in the presence of serum since another modified Sur10058 siRNA containing inverted deoxythymidine residues on 30-end of both sense
`and antisense strands also exhibited faster migration patterns under same conditions (data not shown). (C) Knockdown efficiency of modified siRNAs
`targeting survivin was analyzed by Real-time PCR. HeLa cells were transfected with modified siRNAs (2 nM). Expression levels obtained for GAPDH
`were used for normalization of survivin expression. (D) Total RNA was isolated from HeLa cells transfected with siRNAs (2 nM). Eight lg of total RNA
`was then subjected to Northern blot analysis. Non-silencing siRNA (NC) and unmodified Sur10058 were negative and positive controls, respectively. A
`b-actin probe was used as a means to assess loading variations between lanes.
`
`it did not maintain knockdown efficiency as well as the
`aforementioned examples. As can be seen in Fig. 6, all four
`stabilized