Design and Application of BispecificSplice-Switching Oligonucleotides

Burcu Bestas,1,* Graham McClorey,2,* Ulf Tedebark,1,3 Pedro M.D. Moreno,1,{ Thomas C. Roberts,2,4

Suzan M. Hammond,2 C.I. Edvard Smith,1 Matthew J.A. Wood,2,* and Samir EL Andaloussi1,2,*

Targeting of pre-mRNA by short splice-switching oligonucleotides (SSOs) is increasingly being used as atherapeutic modality, one rationale being to disrupt splicing so as to remove exons containing prematuretermination codons, or to restore the translation reading frame around out-of-frame deletion mutations. The aimof this study was to investigate the effect of chemically linking individual SSOs so as to ascertain equimolarcellular uptake that would provide for more defined drug formulations. In contrast to conventional bispecificSSOs generated by conjugation in solution, here we describe a protocol for synthesis of bispecific SSOs on solidphase. These SSOs comprised of either a non-cleavable hydrocarbon linker or disulfide-based cleavable linkers.To assess the efficacy of these SSOs we have utilized splice switching to bypass a disease-causing mutation inthe DMD gene concurrent with disruption of the reading frame of the myostatin gene (Mstn). The premise ofthis approach is that disruption of myostatin expression is known to induce muscle hypertrophy and so forDuchenne muscular dystrophy (DMD) could be expected to have a better outcome than dystrophin restorationalone. All tested SSOs mediated simultaneous robust exon removal from mature Dmd and Mstn transcripts inmyotubes. Our results also demonstrate that using cleavable SSOs is preferred over the non-cleavable coun-terparts and that these are equally efficient at inducing exon skipping as cocktails of monospecific versions. Inconclusion, we have developed a protocol for solid-phase synthesis of single molecule cleavable bispecificSSOs that can be efficiently exploited for targeting of multiple RNA transcripts.

Introduction

Antisense oligonucleotides (AONs) have raised tre-mendous interest as a result of their inherent ability to

sequence-specifically bind to complementary RNAs andbeing amenable for targeting of virtually any disease (ELAndaloussi et al., 2012; Kole et al., 2012). One AON-basedplatform that has gained increasing attention is the use ofsplice-switching oligonucleotides (SSOs) to modulate splic-ing patterns through binding to pre-messenger (m)RNA(Bauman et al., 2009; Wood, 2010). SSOs have been devel-oped to address out-of-frame deletions, cryptic splice sites,and premature termination codons that result in nonfunc-tional protein products. It has been estimated that 20%–30%of all disease-causing mutations affect pre-mRNA splicing,giving rise to diseases such as b-thalassemia, cystic fibrosis,various types of cancers, spinal muscular atrophy (SMA) andDuchenne muscular dystrophy (DMD) (Bauman et al., 2009;

Muntoni et al., 2011). Importantly, for many of these dis-eases, in particular SMA and DMD, efficacious treatmentsthat target the genetic defect are currently lacking.

DMD is a common, fatal muscle degenerative disorder af-fecting approximately 1 in 3,500 newborn boys. The diseaseis caused by mutations in the DMD gene that typically disruptthe open reading frame of the dystrophin mRNA leading toabsence of functional protein (Hoffman et al., 1987). The ex-istence of mildly affected Becker muscular dystrophy indi-viduals with in-frame DMD gene deletions suggests thatmodification of dystrophin pre-mRNA splicing by forcedskipping of appropriate exons, using SSOs, could induce theproduction of partly functional dystrophin in DMD patients(Aartsma-Rus et al., 2003). Indeed, the use of SSOs to mediatesuch splice switching therapy has been demonstrated numer-ous times in vivo using the dystrophin-deficient mdx mousemodel (Lu et al., 2003, 2005; Alter et al., 2006; Wu et al., 2008,2010, 2012; Heemskerk et al., 2009; Yin et al., 2010). The

1Department of Laboratory Medicine, Karolinska Institutet, Huddinge, Sweden.2Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, United Kingdom.3GE Healthcare Bio-Sciences AB, Uppsala, Sweden.4Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, California.*These authors contributed equally to this work.{Current affiliation: Universidade do Porto, Instituto de Engenharia Biomedica, Porto, Portugal.

NUCLEIC ACID THERAPEUTICSVolume 24, Number 1, 2014ª Mary Ann Liebert, Inc.DOI: 10.1089/nat.2013.0462

13

therapeutic potential of this strategy has recently been shownin two independent systemic clinical trials. In one of thestudies the authors showed dose-dependent SSO efficacy,using the AON analog 2¢ -O-methyl phosphorothioate RNA (2¢-OMePS RNA), in patients with DMD (Goemans et al., 2011).Similarly, a phase 2 systemic trial using phosphorodiamidatemorpholino oligomer (PMO) chemistry has also demonstrateddose dependent restoration of dystrophin, albeit at low levels(Cirak et al., 2011).

Although this shows that SSOs have therapeutic potential,targeting single mutations in DMD with SSOs can onlybenefit a fraction of patients suffering from the disease sincedifferent mutations require different SSOs. Extensive effortshave therefore been made to develop tools to skip multipleexons (i.e., Ex45–55) in DMD patients that would allow asingle treatment to address almost 75% of DMD patientssuffering from deletion mutations (van Vliet et al., 2008).Furthermore, several other proteins could contribute to thepathology of DMD and targeting several pathways simulta-neously may therefore be beneficial. One important consid-eration has been to devise an efficient means of assuringequal delivery of multiple SSOs into the same cells. Re-searchers have attempted to induce multi-exon skipping ofthe large deletion mutation ‘‘hotspot’’ regions of the DMDgene (exons 45–55) using cocktails of SSOs with limitedsuccess (van Vliet et al., 2008), although it should be notedthat skipping with two SSOs to correct DMD mutations hasbeen successfully demonstrated (Aartsma-Rus et al., 2004;McClorey et al., 2006). Apart from DMD, others have de-veloped so-called bispecific AONs to simultaneously targetdifferent mRNAs with the same AON (Rubenstein et al.,2012). Most of these bispecifics have been constructed tocontain two antisense sequences spaced by nucleotide hair-pins and can therefore only target either one of two sites. In2010, the first bispecific short interfering RNA (siRNA) wasreported in which two siRNAs with different cellular targetswere connected together by a cleavable disulfide linker (Moket al., 2010). The authors demonstrated that such a singlemolecule siRNA was cleaved inside the reductive environ-ment of cells and could mediate robust RNA interference(RNAi) responses for both target genes. However, these bi-specific siRNAs were generated by mixing separate siRNAsharboring free thiol groups which resulted in mixtures ofmonospecific, bispecific and multimeric siRNAs (and werethus not of a defined composition). Follow-up work from thesame group showed that defined cleavable bispecific siRNAscould be generated by covalent conjugation in solution(Chung et al., 2011). However, the yields of such reactionsare typically relatively low.

Here, we sought to develop single molecule bispecificSSOs that are synthesized with internal cleavable linkers (i.e.,disulfide bridges) on solid phase (Fig. 1) and that potentiallycan be extended with further antisense sequences for multipletargeting. In the context of DMD, studies have primarilybeen based on targeting removal of a single exon within Dmdpre-mRNA. However, other proteins and small regulatorymicroRNAs (miRNAs) have recently been shown to be in-volved in the pathology of DMD (McPherron et al., 1997;Cacchiarelli et al., 2011; Roberts et al., 2012b) and aretherefore potential therapeutic targets. Of particular interestis myostatin (Mstn), a secreted transforming growth factorbeta (TGFb) superfamily member, which has a role in muscle

homeostasis and is thought to play a major role in variousmuscle wasting disorders (McPherron et al., 1997). In mousemodels of DMD, the simultaneous blockade of the myostatinpathway leads to muscle hypertrophy and concurrent withdystrophin restoration has been suggested to have a greatertherapeutic outcome than dystrophin restoration alone(Bogdanovich et al., 2002).

Thus, we here synthesized bispecific SSOs where one armtargets exon 23 of the mouse Dmd locus and the other targetsexon 2 of the Mstn gene to disrupt the reading frame, aspreviously reported by others (Kang et al., 2011b; Kemala-dewi et al., 2011; Malerba et al., 2012). These bispecificoligonucleotides were either synthesized having internaldisulfides, or with non-cleavable C18 spacers and thencompared with the use of singular SSOs. Our results dem-onstrate that bispecific SSOs based on 2¢-OMePS RNAchemistry can be readily synthesized on solid phase with highyields and purity. Importantly, such SSOs could effectivelypromote exon skipping of both transcripts in differentiatedH2k mdx myotubes. As expected, the cleavable bispecificSSOs displayed greater activity than the non-cleavable ver-sion, being as potent as, or even better, than a cocktail ofmonospecific SSOs.

Materials and Methods

Synthesis of oligonucleotides

All oligonucleotide derivatives were synthesized on anAKTA� oligopilot� 10 plus system at GE Healthcare Bio-Sciences in Uppsala, Sweden. Since priority was given topurity and that these oligonucleotide derivatives were *45–50 nucleotides in length, loading was lowered and an extraspacer was used to facilitate synthesis performance. Standardtemplate 2¢ -OMe RNA synthesis methods were used exceptthiolation which was carried out with a 1:1 mixture of ace-tonitrile (ACN) and PADS [(bis(diphenylacetyl)disulfide),ISIS Pharmaceuticals, 0.2 M in ACN/3-picoline 1:1(v/v)].Introduction of spacer (C18) and linker (SS or XS) werecarried out by double couplings (0.1 M amidite, 4 equiv.),2 · 10 minutes followed by standard oxidation [50 mM I2,Pyr/H2O 9:1 (v/v) in a 1:1 mixture of ACN]. Dilution ofoxidation/thiolation reagent is to avoid reductive cleavage ofthe disulfide. Other ancillary reagents (EMD Chemicals)were used as recommended by GE Healthcare Bio-Sciences.The C18 (spacer phosphoramidite 18 No. 10-1918-xx), SS(thiol modifier C6 Phosphoramidite # 10-1936-xx), and XS(dithiol phosphoramidite, DTPA, # 10-1937-xx) were pur-chased from Glen Research and used as recommended bythe supplier. Cy5 amidite (GE Healthcare No. 28904249) wasused as recommended by supplier. Final detritylation anddietylamine treatment was done (to remove beta-cyanoethylgroups) prior to cleavage and deprotection overnight in 25%aqueous ammonium hydroxide (Merck) at 55�C, releasingthe crude oligonucleotide derivative into solution ready forpurification.

Purification and desalting of SSOs

All purification was carried out using an AKTAexplorer�100 system equipped with Capto� Q ImPres columns foranion-exchange chromatography (AEC) (buffer A: 10%ACN, 10 mM NaClO4, 50 mM TRIS, pH 7.5; 1mM EDTA;

14 BESTAS ET AL.

buffer B: 10% ACN, 500 mM NaClO4, 50 mM TRIS, pH 7.5;1 mM EDTA). The AKTAexplorer system comprises a P-900high-performance liquid chromatography (HPLC) pump, aP-960 sample pump, a model UV-900 monitor model pH/C-900 pH/conductivity monitor and the system is controlledwith UNICORN� software.

After AEC purification, the SSOs were in a solution con-taining sodium perchlorate. The salt was removed from thesamples by gel filtration with an isocratic flow (3 mL/minute)of Milli-Q� water. Five HiTrap� desalting columns weremounted in serial on the AKTAexplorer system and en-abled desalting of one complete 12 mL fraction in about15 minutes.

Analysis and characterization of SSOs

Analytical HPLC was used with 15 mM triethylamine/400mM hexafluoroisopropanol in water as buffer A andmethanol as buffer B. Liquid chromatography–mass spec-trometry (LC-MS) was performed using an Ettan� LC (GEHealthcare Bio-Sciences) connected to a Finnigan� LCQ�Deca XP plus mass spectrometer (ThermoFischer Scientific),for the characterization of individual fractions (e.g., Dys-SS-Myo) after purification and desalting.

Xevo� G2 QTof together with an AQUITY� UPLC H-class (Waters Sweden AB) equipped with a ACQUITYUPLC OST C18, 1.7 mm, 2.1 · 50 mm column controlled byMassLynx� was used for purity analysis and characteriza-tion of most oligonucleotide derivatives (e.g., Myo-XS-Dys)together with Maxent1 software for molecular weight cal-culation (Waters Sweden AB).

Cell culture and transfection of cells

Mouse H2k mdx (Morgan et al., 1994) muscle cells weregrown on gelatinized plates at 33�C, 10% CO2 in Dulbecco’smodified Eagle’s medium (DMEM) with glutamax supple-mented with 20% fetal bovine serum, 0.5% chick embryoextract, 200 U/mL penicillin and 200 mg/mL of streptomycin(Invitrogen). H2k mdx myotubes were generated in gelatin-coated 24-well plates by seeding 30,000 cells per well,leaving them for 2 days to reach 90% confluency beforechanging media to starvation media (DMEM supplemented

with 5% horse serum) and transferring them to 37�C, 5% CO2

incubator for another 4 days. Cells were transfected us-ing Lipofectamine2000 (LF2000; Invitrogen) according tomanufacturer’s protocol. Briefly, 2.2 mL of LF2000 was usedper mg of AON. Complexes were formed in 50 mL OptiMEM(Invitrogen) and added to cells grown in 450 mL full growthmedia. Cells were processed 48 hours later in all transfectionexperiments.

Animal experiments

Experiments were carried out in the Biomedical SciencesUnit, University of Oxford according to procedures autho-rized by the UK Home Office. Intramuscular injections ofAONs into the tibialis anterior were carried out on 7–9 weekold C57BL/10ScSn-Dmdmdx/J mdx mice under isofluraneanesthesia. Briefly, 10 mg of AON was formulated with In-Vivo jetPEI (PolyPlus Transfection) at an N/P ratio of 10 in10% glucose, according to manufacturer’s instructions. Micewere sacrificed seven days later and muscles were snap-frozen in liquid nitrogen and stored at –80�C.

RNA extraction and reverse transcription–polymerasechain reaction

RNA was extracted from either H2k mdx monolayercultures or 50 · 8 mm frozen muscle sections using Trizol(Invitrogen) according to manufacturer’s instructions. In or-der to assess the degree of exon skipping, 400 ng of total RNAwas used as a template in a 50 mL reverse transcription–polymerase chain reaction (RT-PCR) using the GeneAmpRNA PCR kit (Applied Biosystems). Nested RT-PCR anal-ysis of the Dmd transcript between exons 20 to 26 was per-formed under the following conditions; 95�C for 20 seconds,58�C for 60 seconds, and 72�C for 120 seconds for 30 cycles,with 2 mL of this reaction used as a template for nestedamplification using Amplitaq Gold (Applied Biosystems)under the following conditions; 95�C for 20 seconds, 58�Cfor 60 seconds, and 72�C for 120 seconds for 25 cycles. Formyostatin transcript analysis, nested amplification of exons1–3 was performed under the following conditions; 95�C for20 seconds, 56�C for 30 seconds and 72�C for 60 secondsfor 35 cycles with 2 mL of this reaction used as a template for

FIG. 1. Principle of single molecule bis-pecific short splice-switching oligonucleo-tides (SSOs) with internal cleavabledisulfide linkers. Two SSO sequences (Aand B) are delivered into cells as a singlechemical entity. Upon internalization to thecytoplasm, the reductive environment me-diates disulfide cleavage and releases thetwo SSO sequences that subsequently cantarget two different transcripts. In the case ofBI-S, sequence A will have a free C6-SH inthe 3¢ end and sequence B would have a C6-SH in the 5¢ end. If the used linker is BI-XS,then both A and B SSO will have no mod-ification (i.e., both will have a free –OH inthe 3¢ and 5¢ end, respectively).

BISPECIFIC SPLICE SWITCHING OLIGONUCLEOTIDES 15

nested amplification; 95�C for 20 seconds, 56�C for 30 sec-onds, and 72�C for 60 seconds for 25 cycles. PCR productswere analyzed on 2% agarose gels and densitometry per-formed by QuantityOne software (BioRad). Exon 23 skip-ping levels are expressed in the text as the percentage of theof exon 23 skipped PCR product intensity relative to the sumof the intensities of the exon 23 skipped and full-lengthproducts. Similarly, Mstn exon 2 skipping is expressed as thepercentage of the of exon 2 skipped product intensity relativeto the sum of the intensities of full-length and exon 2 skippedproducts. Additionally, to quantify levels of Mstn exon 2skipping in vivo, 1 mg of RNA was reverse transcribed usingthe High Capacity cDNA RT Kit (Applied Biosystems) ac-cording to manufacturer’s instructions. Quantitative poly-merase chain reaction (RT-qPCR) analysis was performedusing 25 ng cDNA template and amplified with TaqManGene Expression Master Mix (Applied Biosystems) on aStepOne Plus Thermocycler (Applied Biosystems). Levelsof Mstn exon 2 skipping were determined by RT-qPCRof exon 2–3 (Assay Mm.PT.56a.13573446, IntegratedDNA Technologies), normalized to Cyclophilin B (AssayMm00478295, Applied Biosystems). Primer sequences areavailable upon request.

Wst-1 cell viability assay

Cell viability was assessed by the Wst-1 proliferation assayaccording to manufacturer’s protocol (Roche DiagnosticsScandinavia AB). Briefly, cells were grown and treated asdescribed above but in 96-well format (i.e., 8,000 cells see-ded and treated in 100-mL volume). Two days after trans-fection at 200 nM SSOs complexed with LF2000, Wst-1 wasadded to each well. Wst-1 measures the activity of mito-chondrial dehydrogenases to convert tetrazolium salts toformazan, and cell proliferation is directly correlated to theamount of formazan product that is formed. Absorbance wasmeasured on SpectraMax Absorbance Microplate reader(Molecular Devices).

Fluorescence-activated cell sorting analysis

H2k mdx myotubes were transfected with LF2000 (In-vitrogen, Sweden) according to manufacturer’s protocol.After 4 hours, the cells were trypsinized for 10 minutes. Aftertrypsinization the cells were washed twice with DPBS(Dulbecco’s phosphate-buffered saline). Before the analysiswith flow cytometry, the cells were resuspended in DPBS

containing 1 mL/mL propidium iodide (Life Technologies).Fluorescence-activated flow cytometry (FACS) was per-formed with BD FACS Calibur Flow Cytometry System(Center for Hematology and Regenerative Medicine [HERM],Karolinska Institutet). The analysis of the data was done withFlowJo (Department of HERM, Karolinska Institutet).

DTT cleavage assay

Dried AONs were resuspended in 100 mM 1,4-dithio-threitol (DTT) solution in sodium phosphate buffer (pH 8.3–8.5). The tubes were kept in 37�C overnight. The sampleswere run on 15% Tris/Borate/EDTA (TBE) urea gel (LifeTechnologies). After 10 minutes incubation with SYBR GoldNucleic Acid Gel Stain (Invitrogen), the detection of the gelwas performed by VersaDoc system (BioRad) and analysisdone using the QuantityOne software (BioRad).

Results

Evaluation of SSOs for myostatin corruptiveexon skipping

Before initiating the synthesis of novel bispecific SSOstargeting different transcripts, we first set out to screen forpotent SSO sequences targeting the Mstn transcript. Here, wedesigned SSOs that promoted skipping of exon 2 from theMstn pre-mRNA to disrupt the mRNA reading frame so as toreduce levels of myostatin protein. Four 25mer AONs weredesigned to target either the consensus splice sites of exon 2(AON 1 and AON 4) or putative exon splicing enhancer re-gions (AON 2 and AON 3) identified using the enhancerprediction algorithms ESEFinder 3.0 (Smith et al., 2006) andRescueESE (Fairbrother et al., 2002) (Fig. 2). Each AON wastransfected at 100 nM into H2k mdx cells along with a pre-viously published AON as a positive ( + ve) control targettingMstn (Kang et al., 2011a) ( + veAON) or a 25mer AON tar-geting exon 23 of the Dmd gene (DYS AON) (Fig. 2). AllAONs targeting Mstn effectively spliced exon 2 from themature transcript, with further titration studies demonstratingAON3 to have the greatest efficacy (data not shown). Basedon the efficacy of AON3 being as good, if not better than apreviously published AON sequence used to successfullyinduce a functional change in vivo, (Kang et al., 2011a) weconsidered this appropriately effective for Mstn targeting.This AON sequence (5¢-CCAUUCUCAUCCAAAGCUUUGAUUU-3¢), that targets positions 304–328 of exon 2, wastherefore chosen as the basis for bispecific SSO design.

FIG. 2. Development of SSOs for Mstn exon2 skipping. Four SSOs were developed to targetconsensus splice sites and exon splicing en-hancer regions of exon 2 of the myostatin (Mstn)gene as depicted in schematic. Exon skippingefficacy was determined by nested reversetranscription–polymerase chain reaction (RT-PCR) with 158-bp product representing removalof exon 2 from the mature Mstn transcript. AllSSOs successfully removed exon 2, at a similarefficacy to a previously published SSO ( + ve).SSOs targeting Dmd gene (exon 23) and un-treated cells were used as negative controls.AON, antisense oligonucleotide.

16 BESTAS ET AL.

Design of bispecific SSOs

Given the success with cleavable bispecific siRNAs andthe limited success with nucleotide-spaced bispecifics formultiexon skipping in DMD (Aartsma-Rus, personal com-munication), we set out to develop a system to generate singleAONs internally spaced with cleavable (i.e., disulfide) link-ers or non-cleavable C18 spacers by solid phase synthesis inAKTA oligopilot 10 plus system. The synthesis scheme isdepicted in Fig. 3. Briefly, standard template 2¢-OMe RNAsynthesis methods were used, except for thiolation/oxidationconditions (diluted with 50% ACN-as recommended by thesupplier (Glen Research) to avoid reductive cleavage of thedisulfide linkage). In the case of C18, we used the samesynthesis conditions (as for XS and SS) to have a relevantreference synthesis. SSO sequences and nomenclature arelisted in Table 1.

Bispecific SSOs promote exon skipping of dystrophinand myostatin

After confirmation that we could generate the correctSSO products with high purity (an example is depicted inSupplementary Fig. S1; Supplementary Data are availableonline at www.liebertpub.com/nat), we assessed the exonskipping efficacy of either (a) cotransfecting Dmd targetingEx23D ( + 7-18) SSO (which has previously been identifiedas being highly effective; Mann et al., 2002) together withMstn targeting Ex2A( + 304 + 328) (M + D); (b) a non-cleavable bispecific SSO comprising both sequencesspaced by C18 (BI-C18); or (c) a cleavable version inter-nally spaced by a conventional disulfide (BI-S), in H2kmdx myotubes. The efficacy of target exon removal wasdetermined by gel densitometry of nested RT-PCR prod-ucts as described in Materials and Methods section. Asseen in Fig. 4a, all SSOs induced Dmd exon 23 skipping ina dose-dependent manner as determined by RT-PCRanalysis. Interestingly, the activity of BI-S was similar orslightly higher as compared to transfection with the sepa-rate SSOs and exceeded the activity of the non-cleavableBI-C18. Additionally, all SSOs simultaneously mediatedexon 2 removal from the mature Mstn transcript. As weobserved near complete exon skipping of the Mstn tran-script at 100 nM concentration (Fig. 2), we chose to onlyassess exon skipping at a lower SSO concentration (50nM). In contrast to dystrophin, no major differences wereobserved between the different bispecifics or monospecificcocktails. A summary of independent transfection experi-ments are provided in Fig. 4b (Dmd) and Fig. 4c (Mstn).These results are in accordance with previous reports usingmultimeric siRNAs; a single SSO with more than onetarget site can target several transcripts simultaneously withequal or better efficiency than the separate SSOs (Moket al., 2010). Unsurprisingly, our results also suggest that acleavable linker within the bispecific SSOs is preferableover a non-cleavable linker.

We next validated whether such cleavable bispecific SSOscould induce exon skipping in vivo. Eight week-old mdx micewere injected intramuscularly into the tibialis anterior witheither monospecific SSOs (5 mg of each) or 10mg BI-Scomplexed with InVivo-JetPEI at an N/P ratio of 10. Tissuewas harvested 7 days later and RNA analyzed for Dmd Ex23skipping by RT-PCR and Mstn exon 2 skipping by RT-qPCR.

Efficacy of Mstn exon 2 skipping was indirectly assessed byquantitative measurement of Mstn transcripts containingexon 2 (using primers across exon 2–3 boundary) after nor-malization to housekeeping gene. At these dosage levels, wewere unable to detect any dystrophin exon skipping with any

FIG. 3. Generation of bispecific SSOs. (a) Synthesisscheme for generation of bispecifics where each SSO se-quences is covalently bound with either a noncleavable C18spacer or a cleavable disulfide linker (S-S). All SSOs are 2¢-OMe phosphorothioate RNAs P(S), whereas C18/disulfide is(PO). (b) Solid-phase synthesis of a bispecific antisenseoligonucleotide (AON), (AON-1-XS-AON-2) using stan-dard 2¢-O-methyl RNA (2¢-OMe RNA) conditions exceptmodified thiolation/oxidation parameters (see Materials andMethods). The XS linker is considered a more stable linkerwith an internal cleavable disulfide. DMT, dimethoxytrityl.

BISPECIFIC SPLICE SWITCHING OLIGONUCLEOTIDES 17

of the SSOs (data not shown), although a robust reduction inMstn transcript levels containing exon 2 could be observed(Fig. 4d). Similar to in vitro transfection results, no signifi-cant difference could be observed between BI-S SSO treat-ments compared with monospecific versions.

The uptake and toxicity profile of bispecific SSOsis similar to the monospecific versions

We have previously observed that the transfection effi-ciency of nucleic acids using LF2000 may depend onthe molecular weight of the cargo (unpublished data). To

Table 1. Sequences of Oligonucleotides

Myostatin-Ex2A ( + 304 + 328)-(M) 5¢ CCAUUCUCAUCCAAAGCUUUGAUUUDystrophin-Ex23D ( + 7-18)-(D) 5¢ GGCCAAACCUCGGCUUACCUGAAAUDystrophin shorter (used for BI-XS) 5¢ GGCCAAACCUCGGCUUACCUDystrophin-SS-myostatin (BI-S) 5¢ GGCCAAACCUCGGCUUACCUGAAAU-

5ThiolC6-CCAUUCUCAUCCAAAGCUUUGAUUUMyostatin-SS-dystrophin (in BI-S) 5¢ CCAUUCUCAUCCAAAGCUUUGAUUU-5ThiolC6-

GGCCAAACCUCGGCUUACCUGAAAUMyostatin-XS-dystrophin (BI-XS) 5¢ CCAUUCUCAUCCAAAGCUUUGAUUU-Dtpa-

GGCCAAACCUCGGCUUACCU

FIG. 4. Bispecific SSOs efficiently induce exon skipping of both Dmd and Mstn. (a) H2k mdx myotubes were transfectedwith indicated concentrations of the different SSOs using LF2000. Two days after transfection, RNA was extracted andreverse transcription-polymerase chain reaction (RT-PCR) performed to assess the levels of Dmd Exon 23 (Ex23) skipping.PCR products were run on a 2% agarose gel. The gel corresponds to one representative experiment. (b) Quantification ofEx23 skipping from three independent experiments in duplicate performed as indicated in (a). (c) Same treatments as in (a)but at 50 nM SSO concentration and analyzing Ex2 skipping of the Mstn transcript. The graph comprises a summary of threeindependent experiments. Error bars indicate the mean – standard deviation (SD). (d) Mstn exon skipping following in-tramuscular delivery: relative levels of exon 2–containing Mstn transcripts. Ten micrograms of BI-S or mix of 5 mg ofmonospecifics were complexed with In Vivo-JetPEI at N/P ratio 10 and injected to tibialis anterior (TA). At day 7, musclewas harvested and RNA extracted and myostatin silencing assessed by quantitative real-time PCR (RT-qPCR). The graphcorresponds to two mice per group with two parts of the muscle analyzed. Ex2 skipping was normalized to mock(phosphate-buffered saline) injected TA. U, untreated.

18 BESTAS ET AL.

investigate if the larger bispecific AONs had differentialuptake to monospecific AONs, we assessed cellular uptakeusing Cy5-labeled compounds. Cells were transfected at100 nM of Cy5 BI-XS or Cy5 exon 23 ( + 7-18) monospecificand analyzed at 4 hours post transfection. No notable dif-ference in cellular uptake could be observed under fluores-cence microscopy (data not shown). Thus, to more accuratelyquantify the level of Cy5-SSO uptake, FACS analysis wasperformed. As seen in Fig. 5a, the histograms display somedifferences in cell membrane association/uptake for BI-XS ascompared to a monospecific SSO. Overall, both SSOs displaynear 100% transfection efficiencies but the Cy5 signal in-tensity appears higher for the monospecific SSO (Fig. 5a,right panel). This increased signal could be explained byslightly higher fluorescence in the stock solution of the Cy5exon 23 monospecific SSO (data not shown).

Next, we validated whether the high activity of BI-S wasassociated with any cellular toxicity. Cells were transfectedat 200 nM SSOs and assessed for viability 48 hours laterusing the Wst-1 assay. None of the treatments showed anytoxicity compared to untreated cells whereas the positivecontrol (0.1% Triton in PBS) reduced the viability dramati-cally (Fig. 5b).

Evaluation of different types of bispecifics

After confirming the activity of BI-S, we next set out to testwhether the internal orientation of the antisense sequences

(Myo-Dys or Dys-Myo) altered the splice-switching activity.For this, we synthesized yet another version (depicted invBI-S) that contains the Mstn antisense sequence in the 5¢ end ofthe bispecific SSO. Cells were transfected at 50 nM of bothbispecific versions and exon skipping assessed 48 hours later.As seen in Fig. 6a, both SSOs mediate exon 23 skipping fromthe Dmd transcript with similar efficacy. Although invBI-Sappeared slightly more active for exon 2 skipping of the Mstntranscript, large variation between experiments make it dif-ficult to conclude if there was a significant difference and atmost we can say that internal orientation did not significantlyimpact on overall activity (Fig. 6a).

Over the course of experiments we observed greater var-iations in activity for the BI-S and invBI-S SSOs, whereassplice-switching efficacy was relatively consistent for themonospecific ONs and the non-cleavable BI-C18 (data notshown). We hypothesize that these variations and slight lossof activity could be a result of cleavage of the disulfide linker,following several freeze/thaw cycles of aliquots. To assessthis we exposed AONs to DTT overnight and ran them ondenaturing polyacrylamide gels to assess their resistance tocleavage. For the BI-S SSO a large proportion of the parentAON was reduced to its monospecific components (Fig. 6b).In contrast, BI-C18 appeared to have complete resistance toDTT-mediated cleavage, as might be anticipated. If wesubjected the BI-S compound to several freeze/thaw cycles,gel analysis demonstrated that the compound was alreadycleaved, even in the absence of DTT treatment (data not

FIG. 5. Uptake and toxicityprofiles of the different SSOs.(a) Left panel, histogramshowing the fluorescence-activated flow cytometry up-take profile of 100 nM mono-specific or bispecific SSO at4 hours post transfection intoH2k mdx myotubes. The y-axison the graph represents thenormalized number of eventscalculated using flowjo. Theright panel shows the quantityof SSOs being internalized(MFI, mean fluorescence in-tensity). (b) Cell viability48 hours post treatment with200 nM of indicated SSOs intomyotubes.

BISPECIFIC SPLICE SWITCHING OLIGONUCLEOTIDES 19

shown). Thus, while BI-S appears potent, it does seem to beless stable. To address this potential stability issue weexploited another approach to generate an internal disulfide(BI-XS) that produces monospecific oligonucleotides with-out 5¢ or 3¢ thiol groups upon intracellular cleavage andwould be expected to have reduced rate of cleavage (Fig. 6c).In contrast to the previously used bispecific SSOs that arecomprised of two 25mer targeting sequences, the BI-XS wasdesigned to contain a 20mer sequence targeting the Dmd pre-mRNA, the rational being that a shorter sequence would beeasier to synthesize in larger scale. Dose-response experi-ments comparing the 25mer and the 20mer Dmd exon 23targeting sequence did not show any significant differencesbetween the two SSOs, and so was thought not to affectoverall efficacy (data not shown). Before assessing the ac-tivity of BI-XS, we validated that the XS linker, with an

internal disulfide, could be cleaved in a reductive environ-ment (i.e., using DTT). As can be seen in Fig. 6d, the dis-ulfide was nearly completely cleaved in the presence ofDTT, resulting in the release of two nonmodified separateSSOs. We next conducted dose-response experiments com-paring BI-XS with the previously tested SSOs. All SSOspromoted dose-dependent exon skipping of exon 23 from theDmd transcript (Fig. 6e). Similarly, using the lower 50 nMSSO concentration, all versions mediated myostatin skip-ping (Fig. 6f, g). Encouragingly, the use of an internal dis-ulfide linker only partially reduced Dmd exon 23 skippingactivity compared to the BI-S bispecific and monospecificSSOs, and was equal, if not better than the branched C18linker SSO. For Mstn skipping, efficacy of the BI-XS wasalso reduced but only marginally compared to other AONchemistries assessed.

FIG. 6. Assessment ofexon skipping activity and sta-bility of different cleavablebispecifics. (a) Exon skippingof Dmd (left) and Mtsn (right)in mdx myotubes at 48 hoursfollowing transfection at 50 nMBI-S or invBI-S. The graph is asummary of two independentexperiments performed in du-plicate. (b) Polyacrylamide gelelectrophoresis analysis of BI-S and BI-C18 with or withoutpresence of 1,4-dithiothreitol(DTT). (c) Schematic illustra-tion of BI-XS and its cleavageto nonmodified monospecificversions. (d) Gel image of BI-XS without DTT (–) and in thepresence of DTT ( + ) com-pared with the non-cleavableBI-C18 SSO. (e) RT-PCRanalysis of cells treated as de-scribed in Fig. 4 using theBI-XS SSO at indicated con-centrations (gel image) andquantitative comparison ofDmd Ex23 skipping with theother SSOs (graph). The levelsof skipping were quantifiedfrom three independent exper-iments performed in duplicate.(f) Quantitative comparison ofMstn Ex2 skipping after treat-ment with 50 nM SSOs. Theskipping levels were quantifiedfrom three independent exper-iments in duplicate. (g) RT-PCR analysis of cells treatedwith SSOs as described in Fig.4 at 50nM to assess the levelsof Mstn Ex2 skipping. Tworepresentative experiments areshown for each SSO. Error barsindicate the Mean + SD. U,untreated.

20 BESTAS ET AL.

Discussion

The use of SSOs to modulate splicing has opened a newavenue for potential treatment of genetic diseases includingSMA and DMD. For DMD, current clinical trials have fo-cused on targeted skipping of exon 51, which as a single SSOwould directly address the largest number of patient muta-tions (approximately 13% of all DMD patients). Similarly,single SSO targeting of exons 45, 44, and 53 would directlybenefit 8%, 6%, and 8% of DMD patients respectively. Bytargeting pre-mRNA splicing therefore, an exon skippingapproach has the advantage that it can address many differentgenomic DMD mutations. However, some mutations requiremore than a single exon to be targeted such as when a pre-mature termination codon lies within an out-of-frame exon orwhen more than one exon is required to be skipped to restorethe reading frame around a deletion mutation. Additionally,some DMD exons are particularly difficult to remove fromthe mature DMD transcript, and in these situations a cocktailof two or three SSOs targeting the same exon may be pref-erable (Aartsma-Rus et al., 2006; Adams et al., 2007). Forthese reasons, a multiple SSO approach may be preferable orindeed necessary. Indeed, taking this concept further, at-tempts have been made to develop a ‘‘one cocktail fits all’’approach to address the major deletion ‘‘hotspot’’ within thecentral region of the DMD gene where it has been estimatedthat removal of the in-frame region of exons 45–55 couldaddress 30%–60% of all DMD patients (Aartsma-Rus, 2012).Extensive efforts have been made to devise efficient means toinduce multiexon skipping. To date, cocktails of SSOs tar-geting different exons have been used but the collectiveskipping levels of the entire locus remain very low (van Vlietet al., 2008). This is perhaps unsurprising given that approachis made more difficult by the necessity not only for eachindividual AON to reach single nuclei, but also for thecocktail of AONs to bind simultaneously to the pre-mRNAprior to splicing reaction. Recently, Akoi et al. demonstratedfor the first time relatively efficient multiexon skipping in themdx52 mouse model using a cocktail of PMOs conjugated toa branched delivery vector composed of cationic guanidi-nium groups (Aoki et al., 2012). Although promising, toxicityconcerns related to these types of delivery modalities remaina major obstacle.

In addition to the desire to simultaneously splice multipleexons from the DMD transcript, there is increasing evidencethat targeting other genes involved in modulating the diseasephenotype could be beneficial. DMD is associated with se-vere muscle wasting, fibrosis, altered muscle regenerationand, importantly, muscle degeneration whereby the musclestem cell pool becomes exhausted. Therefore, additionaltherapies have been proposed to overcome these issues. Forexample, signaling induced by myostatin via the activin re-ceptor (AcvR2b) is known to negatively regulate musclegrowth by targeting down-stream myogenic regulatory fac-tors in order to suppress proliferation (Thomas et al., 2000;Langley et al., 2002; McCroskery et al., 2003). This suggeststhat simultaneously restoring dystrophin function and re-ducing myostatin signaling would be advantageous. Anumber of approaches have been utilized to modulate theactivity of the myostatin pathway, including myostatin neu-tralizing antibodies (Wagner et al., 2008), endogenousmyostatin antagonists [myostatin propeptide (Bogdanovich

et al., 2005), follistatin (Nakatani et al., 2008) and solubleAcvr2b (the myostatin receptor; Lee et al., 2001)], RNAi(Magee et al., 2006) and transcriptional gene silencing (Ro-berts et al., 2012a). Indeed, Dumonceaux and colleaguesdemonstrated that combination of myostatin suppression viaRNAi and dystrophin rescue using U7-expressing adeno-associated viruses improved muscle function in mdx mice(Dumonceaux et al., 2010). Similarly, increased muscle sizewas observed when PMOs conjugated to a cell-penetratingpeptide were administered into mdx mice as cocktails(Malerba et al., 2012).

To address the need for targeting multiple exons within theDMD gene, and to help further explore the potential of tar-geting genes involved in disease modulation, we sought todevelop AONs that could incorporate dual targeting se-quences. Dual-targeting SSOs have been developed previ-ously to induce large multi-exon skipping of exons 45–51within the DMD gene through use of a bispecific SSO with a10-bp uracil spacing linker, compared to a cocktail ofmonospecific SSOs (Aartsma-Rus et al., 2004). Whilst bothapproaches could induce skipping, efficacy was low and re-sults inconsistent such that no firm conclusions could bemade as to which approach was more effective. BispecificAONs have also been developed for suppressing gene ex-pression of unrelated genes to target cancer (Rubensteinet al., 2012), but again, these bispecific AONs were inter-linked by non-cleavable nucleotide hairpins. From a thera-peutic efficacy point of view, we hypothesize that it would bepreferable to incorporate a cleavable disulfide linker thatwould be reduced inside cells, leaving the two antisense se-quences to bind their cognate RNA sequence (as depicted inFig. 1). Indeed, Chung and coworkers have demonstrated theadvantage of using disulfide-linked multimeric siRNAs forgene silencing applications (Chung et al., 2011). However,such dimeric and multimeric siRNAs are generated by con-jugation in solution, leading to multiple products and purifi-cation steps that reduce the yields.

Based on this, we here developed a new method togenerate bispecific SSOs internally spaced either by a non-cleavable hydrocarbon linker (BI-C18) or cleavabledisulfide-based linkers (BI-S and BI-XS). In contrast to ex-isting literature using bispecific AONs and siRNAs, we es-tablished a synthesis scheme that allows for synthesis of suchAONs on solid-phase (Fig. 3). As seen in SupplementaryFigs. S1 and S2, the methodology described here allowedbispecific SSOs to be efficiently synthesized with high purity.To verify the proof-of-principle of successful synthesis ofthese AONs, we demonstrate that these bispecifics couldinduce exon skipping for both Dmd and Mstn pre-mRNAtranscripts (Fig. 4). We would hypothesize that a cocktail ofmonospecific SSOs that could localize to their individual pre-mRNA targets would have greater efficacy than a non-cleavable linker and this was indeed the case. These resultsare in accordance with results from similar bispecific SSOsspaced by uracil nucleotide linkers (Aartsma-Rus, personalcommunication) and is likely due to the fact that each bis-pecific can only bind to either of the two target pre-mRNAs ata given time. In contrast to the carbon linker, when using theBI-S version that contains a cleavable linker, skipping levelswere dramatically increased, with efficacy equivalent ormarginally exceeding the levels observed for cocktails ofmonospecifics. Due to the generally low levels of Dmd and

BISPECIFIC SPLICE SWITCHING OLIGONUCLEOTIDES 21

Mstn transcripts in vitro, nested PCR analysis was required toassess levels of exon skipping. As such, measurement ofskipping efficiency can only be deemed semiquantitative andwe recognize the limitation of this approach. Nevertheless, itis clear that under the same assay conditions these compar-ative results emphasize the importance of intracellularcleavage of the two antisense components. To confirm thatthese bispecific AONs are cleaved under reducing conditions,each bispecific AON was exposed to DTT overnight andrelease of monospecific products assessed by gel electro-phoresis. Demonstration that the C18 linker was not de-graded, whereas as the BI-S and BI-XS were, is furtherevidence to support that cleavage in a reducing environment(such as in the cytosol) contributes to the enhanced efficacy.

Although the emphasis of this study was to develop amethod of synthesizing dual-targeting AONs, simultaneoustargeting of dystrophin to bypass disease-causing mutations,and myostatin to promote muscle hypertrophy, is a logicalapproach to address DMD. To this end, we sought to dem-onstrate a proof-of-principle and assess potency of bispecificversus monospecific AONs in vivo. At 7 days post intra-muscular injection into the mdx mouse model, we observedconsiderable reduction in the levels of Mstn transcript con-taining exon 2 by RT-qPCR analysis in SSO treated musclescompared to untreated controls. As myostatin is a circulatingsignaling factor, we would not anticipate that intramuscularinjection into a localized area would have a functional effect;however, this does provide proof-of-principle that a bispe-cific AON can have some function in vivo. Compared to thecocktail of individual SSOs, Mstn skipping activity wassimilar for the BI-S SSO indicating that at least for thisparticular gene target, that efficacy is not reduced when ap-plying this strategy. Despite this success in targeting Mstn,we were unable to detect any exon 23 skipping of the Dmdpre-mRNA under any of the treatment conditions. As skip-ping could be observed for Mstn targeting we would antici-pate that both the individual SSO and bispecific SSO wouldbe taken up into the nuclei of muscle fibers. We can onlyhypothesize that the lack of skipping in vivo for exon 23 isdue to insufficient nuclear concentration of AON to inducesufficient Dmd skipping to be detected by RT-PCR analysis.Additionally, we cannot rule out that difference between theMstn and Dmd target genes, such as transcript levels and ratesof turnover could attribute to the change in pattern of efficacyin vivo. Efficient delivery of charged nucleic acids to muscleis an unsolved challenge in the AON therapeutic field and it islikely that an effective delivery vehicles or high dosing wouldbe required to perform a robust assessment of bispecificAONs in vivo. Nevertheless, for targeting Mstn, these resultswould suggest that a cleavable bispecific SSO has similaractivity to separate SSOs following local delivery.

To address the question of how AON length may affectcellular uptake, we used Cy5-labeled SSOs to assess uptakeinto H2k mdx myotubes. Qualitative assessment by fluores-cent microscopy suggested that bispecific and monospecificAONs were taken up by cells at similar levels (data notshown). However, when FACS analysis was performed, Cy5signal was higher for cells transfected with monospecificSSOs (Fig. 5a). The reduction in Cy5 signal observed for theBI-S can be partly attributable to the reduced efficiency of theCy5 coupling reaction to the longer AON, which we wouldestimate is about 70% as efficient (data not shown). A limi-

tation of this approach is that we cannot distinguish betweeninternalized and surface bound Cy5 label, however followingwashing steps we would expect the majority of the signal tobe intracellular Cy5 signal. Importantly though, these resultsare highly suggestive that enhanced uptake of cleavablebispecifics is not the underlying reason of equivalent efficacyto monospecific SSOs. Given the novelty of using BI-S andBI-XS SSOs for a splice-switching approach, we assessedwhether cell toxicity could be an issue. However, no notablecellular toxicity was present in any of the AONs used in thisstudy for the concentration interval used for the in vitroexperiments.

One obstacle we encountered over the course of experi-ments with the BI-S SSO was a partial loss in activity overtime (data not shown). Therefore, we exploited an alternativelinker dubbed XS (DTPA), originally developed for an al-ternative purpose (i.e., attaching oligonucleotide derivativesto gold particles). To our knowledge, the XS (DTPA) linkerhas not been used for this purpose before. The advantagecompared to standard disulfide linkers is that it should bemore stable outside the cellular environment and under re-ductive conditions be cleaved so as to not leave free thiolgroups and therefore be of a more similar chemical nature asthe monospecific SSOs (Fig. 6c). The BI-XS SSO dose-dependently induced exon skipping exceeding the activity ofthe non-cleavable bispecific (BI-C18). However, BI-XSdisplayed somewhat lower activity than monospecific ver-sions or the parent BI-S SSO. This is most likely a conse-quence of the inability to fully cleave this SSO underreductive conditions. Indeed, following overnight exposureto DTT, the BI-XS AON appears more resistant to cleavagethan the BI-S AON. Future studies with the BI-XS AON willfocus on assessing its activity in vivo to determine if theapparent enhanced stability of the compound outweighs theobserved reduction in exon skipping efficacy.

In conclusion, we have developed a robust system for solidphase synthesis of bispecific SSOs harboring internal disulfideinterlinkages that are cleaved in reducing environment. Weshow that these are as least as potent as their monospecificcounterparts. For more quantitative assessment of bispecificAON activity in vivo it is likely that a significantly higherdosing amounts over a longer time period would be necessaryto induce a measurable functional improvement and theseexperiments along with in vivo testing of the BI-XS AON willbe the subject of future studies. Additionally, it may also beprudent to measure the activity of bispecific AONs where theefficacy of the monospecific elements are more closelymatched, or where the target transcripts have similar rates ofturnover, as this would allow direct comparison of activityover a wider range of concentrations. Nevertheless, weclearly show that these bispecific oligonucleotides have ro-bust in vitro activity and could prove very useful for multi-exon skipping approaches, or multitargeting in general,where equal delivery of each antisense sequence is prefera-ble. Also, from a regulatory therapy perspective, it is ad-vantageous to have single therapeutic compounds overcocktails. Furthermore, we have recently been able to includefurther target sequences through generation of trispecificAONs, additionally comprising a miRNA-31 target site, sinceit has been implicated in regulation of dystrophin expression(Cacchiarelli et al., 2011), which will be further evaluated.Whilst the prospect for these bispecific or trispecific AONs is

22 BESTAS ET AL.

exciting, application in a preclinical or clinical setting willrequire the development of delivery vectors that allow forsystemic delivery of these relatively large compounds andremains the greatest challenge for this approach.

Acknowledgments

S.ELA is supported by the Swedish Research Council(Vetenskapsradet-med Unga Forskare) and the Swedish So-ciety of Medical Research. CIES is supported by the SwedishResearch Council. Waters (Sweden and United States) isgreatly acknowledged for support and collaboration in the LC-MS area. We thank Tolga Sutlu at the Center of Hematologyand Regenerative Medicine (HERM), Karolinska Institutet, forhis kind help with the FACS analysis.

Author Disclosure Statement

The authors declare that there is no competing financialinterest.

References

AARTSMA-RUS, A. (2012). Overview on DMD exon skip-ping. Methods Mol. Biol. 867, 97–116.

AARTSMA-RUS, A., JANSON, A.A., KAMAN, W.E., BREM-MER-BOUT, M., DEN DUNNEN, J.T., BAAS, F., VAN OM-MEN, G.J., and VAN DEUTEKOM, J.C. (2003). Therapeuticantisense-induced exon skipping in cultured muscle cells from sixdifferent DMD patients. Hum. Mol. Genet. 12, 907–914.

AARTSMA-RUS, A., JANSON, A.A., KAMAN, W.E.,BREMMER-BOUT, M., VAN OMMEN, G.J., DEN DUNNEN,J.T., and VAN DEUTEKOM, J.C. (2004). Antisense-inducedmultiexon skipping for Duchenne muscular dystrophy makesmore sense. Am. J. Hum. Genet. 74, 83–92.

AARTSMA-RUS, A., KAMAN, W.E., WEIJ, R., DEN DUN-NEN, J.T., VAN OMMEN, G.J., and VAN DEUTEKOM,J.C. (2006). Exploring the frontiers of therapeutic exonskipping for Duchenne muscular dystrophy by double tar-geting within one or multiple exons. Mol. Ther. 14, 401–407.

ADAMS, A.M., HARDING, P.L., IVERSEN, P.L., COLEMAN,C., FLETCHER, S., and WILTON, S.D. (2007). Antisenseoligonucleotide induced exon skipping and the dystrophin genetranscript: cocktails and chemistries. BMC Mol. Biol. 8, 57.

ALTER, J., LOU, F., RABINOWITZ, A., YIN, H., RO-SENFELD, J., WILTON, S.D., PARTRIDGE, T.A., and LU,Q.L. (2006). Systemic delivery of morpholino oligonucleo-tide restores dystrophin expression bodywide and improvesdystrophic pathology. Nature Med. 12, 175–177.

AOKI, Y., YOKOTA, T., NAGATA, T., NAKAMURA, A.,TANIHATA, J., SAITO, T., DUGUEZ, S.M., NAGARAJU,K., HOFFMAN, E.P., PARTRIDGE, T., and TAKEDA, S.(2012). Bodywide skipping of exons 45–55 in dystrophicmdx52 mice by systemic antisense delivery. Proc. Natl. Acad.Sci. U. S. A. 109, 13763–13768.

BAUMAN, J., JEARAWIRIYAPAISARN, N., and KOLE, R.(2009). Therapeutic potential of splice-switching oligonu-cleotides. Oligonucleotides 19, 1–13.

BOGDANOVICH, S., PERKINS, K.J., KRAG, T.O., WHIT-TEMORE, L.A., and KHURANA, T.S. (2005). Myostatinpropeptide-mediated amelioration of dystrophic pathophysi-ology. FASEB J. 19, 543–549.

BOGDANOVICH, S., and POPOVIC, D. (2002). Onset ofglassy dynamics in a two-dimensional electron system insilicon. Phys. Rev. Lett. 88, 236401.

CACCHIARELLI, D., INCITTI, T., MARTONE, J., CESANA,M., CAZZELLA, V., SANTINI, T., STHANDIER, O., andBOZZONI, I. (2011). miR-31 modulates dystrophin expres-sion: new implications for Duchenne muscular dystrophytherapy. EMBO Rep. 12, 136–141.

CHUNG, H.J., HONG, C.A., LEE, S.H., JO, S.D., and PARK,T.G. (2011). Reducible siRNA dimeric conjugates for effi-cient cellular uptake and gene silencing. Bioconjug. Chem.22, 299–306.

CIRAK, S., ARECHAVALA-GOMEZA, V., GUGLIERI, M.,FENG, L., TORELLI, S., ANTHONY, K., ABBS, S., GAR-RALDA, M.E., BOURKE, J., WELLS, D.J., et al. (2011).Exon skipping and dystrophin restoration in patients withDuchenne muscular dystrophy after systemic phosphoro-diamidate morpholino oligomer treatment: an open-label,phase 2, dose-escalation study. Lancet 378, 595–605.

DUMONCEAUX, J., MARIE, S., BELEY, C., TROLLET, C.,VIGNAUD, A., FERRY, A., BUTLER-BROWNE, G., andGARCIA, L. (2010). Combination of myostatin pathway in-terference and dystrophin rescue enhances tetanic and specificforce in dystrophic mdx mice. Mol. Ther. 18, 881–887.

EL ANDALOUSSI, S.A., HAMMOND, S.M., MAGER, I., andWOOD, M.J. (2012). Use of cell-penetrating-peptides in ol-igonucleotide splice switching therapy. Curr. Gene Ther. 12,161–178.

FAIRBROTHER, W.G., YEH, R.F., SHARP, P.A., andBURGE, C.B. (2002). Predictive identification of exonicsplicing enhancers in human genes. Science 297, 1007–1013.

GOEMANS, N.M., TULINIUS, M., VAN DEN AKKER, J.T.,BURM, B.E., EKHART, P.F., HEUVELMANS, N., HOLLING,T., JANSON, A.A., PLATENBURG, G.J., SIPKENS, J.A., et al.(2011). Systemic administration of PRO051 in Duchenne’smuscular dystrophy. New Engl. J. Med. 364, 1513–1522.

HEEMSKERK, H.A., DE WINTER, C.L., DE KIMPE, S.J.,VAN KUIK-ROMEIJN, P., HEUVELMANS, N., PLATEN-BURG, G.J., VAN OMMEN, G.J., VAN DEUTEKOM, J.C.,and AARTSMA-RUS, A. (2009). In vivo comparison of 2’-O-methyl phosphorothioate and morpholino antisense oligo-nucleotides for Duchenne muscular dystrophy exon skipping.J. Gene Med. 11, 257–266.

HOFFMAN, E.P., BROWN, R.H., JR., and KUNKEL, L.M.(1987). Dystrophin: the protein product of the Duchennemuscular dystrophy locus. Cell 51, 919–928.

KANG, J.K., MALERBA, A., POPPLEWELL, L., FOSTER,K., and DICKSON, G. (2011a). Antisense-induced myostatinexon skipping leads to muscle hypertrophy in mice followingocta-guanidine morpholino oligomer treatment. Mol. Ther.19, 159–164.

KANG, J.K., MALERBA, A., POPPLEWELL, L., FOSTER,K., and DICKSON, G. (2011b). Antisense-induced myostatinexon skipping leads to muscle hypertrophy in mice followingocta-guanidine morpholino oligomer treatment. Mol. Ther.19, 159–164.

KEMALADEWI, D.U., HOOGAARS, W.M., VAN HEININ-GEN, S.H., TERLOUW, S., DE GORTER, D.J., DENDUNNEN, J.T., VAN OMMEN, G.J., AARTSMA-RUS, A.,TEN DIJKE, P., and ‘T HOEN, P.A. (2011). Dual exonskipping in myostatin and dystrophin for Duchenne musculardystrophy. BMC Med. Genomics 4, 36.

KOLE, R., KRAINER, A.R., and ALTMAN, S. (2012). RNAtherapeutics: beyond RNA interference and antisense oligo-nucleotides. Nat. Rev. Drug Discov. 11, 125–140.

LANGLEY, B., THOMAS, M., BISHOP, A., SHARMA, M.,GILMOUR, S., and KAMBADUR, R. (2002). Myostatin

BISPECIFIC SPLICE SWITCHING OLIGONUCLEOTIDES 23

inhibits myoblast differentiation by down-regulating MyoDexpression. J. Biol. Chem. 277, 49831–49840.

LEE, S.J., and MCPHERRON, A.C. (2001). Regulation ofmyostatin activity and muscle growth. Proc. Natl. Acad. Sci.U. S. A. 98, 9306–9311.

LU, Q.L., MANN, C.J., LOU, F., BOU-GHARIOS, G., MOR-RIS, G.E., XUE, S.A., FLETCHER, S., PARTRIDGE, T.A.,and WILTON, S.D. (2003). Functional amounts of dystrophinproduced by skipping the mutated exon in the mdx dystrophicmouse. Nature Med. 9, 1009–1014.

LU, Q.L., RABINOWITZ, A., CHEN, Y.C., YOKOTA, T.,YIN, H., ALTER, J., JADOON, A., BOU-GHARIOS, G., andPARTRIDGE, T. (2005). Systemic delivery of antisense oli-goribonucleotide restores dystrophin expression in body-wideskeletal muscles. Proc. Natl. Acad. Sci. U. S. A. 102, 198–203.

MAGEE, T.R., ARTAZA, J.N., FERRINI, M.G., VERNET, D.,ZUNIGA, F.I., CANTINI, L., REISZ-PORSZASZ, S., RAJ-FER, J., and GONZALEZ-CADAVID, N.F. (2006). Myos-tatin short interfering hairpin RNA gene transfer increasesskeletal muscle mass. J. Gene Med. 8, 1171–1181.

MALERBA, A., KANG, J.K., MCCLOREY, G., SALEH, A.F.,POPPLEWELL, L., GAIT, M.J., WOOD, M.J., and DICK-SON, G. (2012). Dual Myostatin and dystrophin exon skip-ping by morpholino nucleic acid oligomers conjugated to acell-penetrating peptide is a promising therapeutic strategyfor the treatment of duchenne muscular dystrophy. Mol. Ther.Nucleic Acids 1,e62.

MANN, C.J., HONEYMAN, K., MCCLOREY, G., FLETCH-ER, S., and WILTON, S.D. (2002). Improved antisense oli-gonucleotide induced exon skipping in the mdx mouse modelof muscular dystrophy. J. Gene Med. 4, 644–654.

MCCLOREY, G., MOULTON, H.M., IVERSEN, P.L.,FLETCHER, S., and WILTON, S.D. (2006). Antisenseoligonucleotide-induced exon skipping restores dystrophinexpression in vitro in a canine model of DMD. Gene Ther. 13,1373–1381.

MCCROSKERY, S., THOMAS, M., MAXWELL, L., SHAR-MA, M., and KAMBADUR, R. (2003). Myostatin negativelyregulates satellite cell activation and self-renewal. J. CellBiol. 162, 1135–1147.

MCPHERRON, A.C., LAWLER, A.M., and LEE, S.J. (1997).Regulation of skeletal muscle mass in mice by a new TGF-beta superfamily member. Nature 387, 83–90.

MOK, H., LEE, S.H., PARK, J.W., and PARK, T.G. (2010).Multimeric small interfering ribonucleic acid for highly effi-cient sequence-specific gene silencing. Nat. Mater. 9, 272–278.

MORGAN, J.E., BEAUCHAMP, J.R., PAGEL, C.N., PECK-HAM, M., ATALIOTIS, P., JAT, P.S., NOBLE, M.D.,FARMER, K., and PARTRIDGE, T.A. (1994). Myogenic celllines derived from transgenic mice carrying a thermolabileT antigen: a model system for the derivation of tissue-specificand mutation-specific cell lines. Dev. Biol. 162, 486–498.

MUNTONI, F., and WOOD, M.J. (2011). Targeting RNA to treatneuromuscular disease. Nat. Rev. Drug Discov. 10, 621–637. ‘1

NAKATANI, M., TAKEHARA, Y., SUGINO, H., MATSU-MOTO, M., HASHIMOTO, O., HASEGAWA, Y., MUR-AKAMI, T., UEZUMI, A., TAKEDA, S., NOJI, S., et al.(2008). Transgenic expression of a myostatin inhibitor derivedfrom follistatin increases skeletal muscle mass and amelioratesdystrophic pathology in mdx mice. FASEB J. 22, 477–487.

ROBERTS, T.C., ANDALOUSSI, S.E., MORRIS, K.V.,MCCLOREY, G., and WOOD, M.J. (2012a). Small RNA-Mediated Epigenetic Myostatin Silencing. Molecular therapy.Nucleic acids 1,e23.

ROBERTS, T.C., BLOMBERG, K.E.M., MCCLOREY, G.,ANDALOUSSI, S.E.L., GODFREY, C., BETTS, C.,COURSINDEL, T., GAIT, M.J., SMITH, C.I.E, and WOOD,M.J.A. (2012b). Expression Analysis in Multiple MuscleGroups and Serum Reveals Complexity in the MicroRNATranscriptome of the mdx Mouse with Implications forTherapy. Mol. Ther. Nucleic Acids 1,e39.

RUBENSTEIN, M., HOLLOWELL, C.M., and GUINAN, P.(2012). Differentiated prostatic antigen expression in LNCaP cellsfollowing treatment with bispecific antisense oligonucleotidesdirected against BCL-2 and EGFR. Med. Oncol. 29, 835–841.

SMITH, P.J., ZHANG, C., WANG, J., CHEW, S.L., ZHANG,M.Q., and KRAINER, A.R. (2006). An increased specificityscore matrix for the prediction of SF2/ASF-specific exonicsplicing enhancers. Hum. Mol. Genet. 15, 2490–2508.

THOMAS, M., LANGLEY, B., BERRY, C., SHARMA, M.,KIRK, S., BASS, J., and KAMBADUR, R. (2000). Myostatin,a negative regulator of muscle growth, functions by inhibitingmyoblast proliferation. J. Biol. Chem. 275, 40235–40243.

VAN VLIET, L., DE WINTER, C.L., VAN DEUTEKOM, J.C.,VAN OMMEN, G.J., and AARTSMA-RUS, A. (2008). As-sessment of the feasibility of exon 45–55 multiexon skippingfor Duchenne muscular dystrophy. BMC Med. Genet. 9, 105.

WAGNER, K.R., FLECKENSTEIN, J.L., AMATO, A.A.,BAROHN, R.J., BUSHBY, K., ESCOLAR, D.M., FLANI-GAN, K.M., PESTRONK, A., TAWIL, R., WOLFE, G.I.,et al. (2008). A phase I/IItrial of MYO-029 in adult subjectswith muscular dystrophy. Ann. Neurol. 63, 561–571.

WOOD, M.J. (2010). Toward an oligonucleotide therapy forDuchenne muscular dystrophy: a complex developmentchallenge. Sci. Transl. Med. 2, 25ps15.

WU, B., LU, P., BENRASHID, E., MALIK, S., ASHAR, J.,DORAN, T.J., and LU, Q.L. (2010). Dose-dependent restorationof dystrophin expression in cardiac muscle of dystrophic miceby systemically delivered morpholino. Gene Ther. 17, 132–140.

WU, B., LU, P., CLOER, C., SHABAN, M., GREWAL, S.,MILAZI, S., SHAH, S.N., MOULTON, H.M., and LU, Q.L.(2012). Long-term rescue of dystrophin expression and improve-ment in muscle pathology and function in dystrophic mdx mice bypeptide-conjugated morpholino. Am. J. Pathol. 181, 392–400.

WU, B., MOULTON, H.M., IVERSEN, P.L., JIANG, J., LI, J.,LI, J., SPURNEY, C.F., SALI, A., GUERRON, A.D., NA-GARAJU, K., et al. (2008). Effective rescue of dystrophinimproves cardiac function in dystrophin-deficient mice by amodified morpholino oligomer. Proc. Natl. Acad. Sci. U. S. A.105, 14814–14819.

YIN, H., MOULTON, H.M., BETTS, C., MERRITT, T., SEOW,Y., ASHRAF, S., WANG, Q., BOUTILIER, J., and WOOD,M.J. (2010). Functional rescue of dystrophin-deficient mdxmice by a chimeric peptide-PMO. Mol. Ther. 18, 1822–1829.

Address correspondence to:Samir EL Andaloussi, PhD

Clinical Research CenterDepartment of Laboratory Medicine

Karolinska Institutet at NovumHalsovagen 7

SE-141 57 HuddingeSweden

E-mail: samir.el-andaloussi@ki.se

Received for publication September 23, 2013; accepted afterrevision November 26, 2013.

24 BESTAS ET AL.