RNase H-mediated degradation of toxic RNAin myotonic dystrophy type 1Johanna E. Leea,b, C. Frank Bennettc, and Thomas A. Coopera,b,d,1

Departments of aPathology and Immunology and dMolecular and Cellular Biology, and bInterdepartmental Program of Cell and Molecular Biology, BaylorCollege of Medicine, Houston, TX 77030; and cIsis Pharmaceuticals, Inc., Carlsbad, CA 92008

Edited by Eric N. Olson, University of Texas Southwestern, Dallas, TX, and approved February 1, 2012 (received for review October 18, 2011)

Myotonic dystrophy type 1 (DM1) is an RNA-dominant diseasecaused by abnormal transcripts containing expanded CUG repeats.The CUG transcripts aggregate in the nucleus to form RNA fociand lead to nuclear depletion of Muscleblind-like 1 (MBNL1) andstabilized expression of CUGBP Elav like family 1 (CELF1), both ofwhich are splicing regulatory proteins. The imbalance of theseproteins results in misregulation of alternative splicing and neuro-muscular abnormalities. Here, we report the use of antisenseoligonucleotides (ASOs) as a therapeutic approach to target thepathogenic RNA in DM1. We designed chimeric ASOs, termedgapmers, containing modified nucleic acid residues to induce RNaseH-mediated degradation of CUG-repeat transcripts. The gapmersselectively knockdown expanded CUG transcripts and are sufficientto disrupt RNA foci both in cell culture and mouse models for DM1.Furthermore, combination of gapmers with morpholino ASOs thathelp release binding of MBNL1 to the toxic RNA can potentiallyenhance the knockdown effect. Additional optimization will berequired for systemic delivery; however, our study provides analternative strategy for the use of ASOs in DM1 therapy.

microsatellite expansion | muscular dystrophy | phosphorothioate | gapmer

Myotonic dystrophy (DM) is the most common musculardystrophy in adults, affecting 1 in 8,000 individuals. It is

a multisystemic disease that affects mainly the skeletal muscle,heart, and central nervous system (1). DM1 patients have CTGtrinucleotide repeat expansions (>50–3,000 repeats) in the 3′untranslated region (3′ UTR) of the DMPK gene (2). The pre-dominant cause of DM1 pathogenesis is the gain-of-function ofmutant DMPK mRNA, which contains long CUG repeats thataccumulate in the nuclei as RNA foci (3). Two known pathwayscontribute to DM1 pathogenesis. First, the CUG repeats se-quester an RNA-binding protein, Muscleblind-like 1 (MBNL1),resulting in its depletion and loss of function (4). Second, therepeat RNA induces PKC-mediated phosphorylation of CUGBPElav like family 1 (CELF1), resulting in increased stability andgain-of-function (5). MBNL1 and CELF1 are antagonistic reg-ulators of alternative splicing and the imbalance of their activi-ties results in abnormal expression of embryonic splice variantsin adult tissue, some of which contribute to the pathogenesis ofthe disease (6).Increased understanding of DM1 pathogenesis has led to

therapeutic approaches including utilization of antisense oligo-nucleotides (ASOs), which can be used to block gene expression bysteric hindrance or to elicit RNase H-mediated cleavage of thetarget RNA (7). RNase H is a non-sequence-specific enzyme thatrecognizesRNA–DNAheteroduplexes and specifically cleaves theRNA strand (8). By introducingASOs complementary to a specificRNA sequence, RNase H can mediate cleavage and decay of thetarget RNA (9). The stability and efficiency of the ASO can beenhanced by substituting DNA with modified nucleotides withincreased affinity for RNA and resistance to nucleases, includinglocked nucleic acids (LNA) or 2′-O-Methoxyethyl (MOE) nucleicacids. These modified nucleic acids are not recognized by RNaseH; therefore, a center “gap” region with 7–10 nucleotides con-taining RNase H-compatible phosphorothioate (PS) DNA is re-quired (10, 11).

There have been several reports applying ASOs for potentialDM1 therapy. Wheeler et al. used morpholino ASOs that bind tothe toxic CUG repeats, blocking sequestration of Mbnl1 and res-cuing its loss-of-function (12). Another group used 2′-O-methyl(2’-OMe) phosphorothioate modified ASOs that reduced levels ofthe toxic CUG mRNA through unknown mechanisms that do notinvolve RNase H (13). Here we report a study to target degrada-tion of toxic RNA in DM1 specifically through an RNase H-me-diated mechanism.We generated gapmer ASOs with CAG repeatsequences that are sufficient to reduce expanded CUG transcriptsand RNA foci in both cell culture and mouse models of DM1.Importantly, the gapmers preferentially target expanded CUGrepeats compared with normal-size repeats. We also found thatcombined administration of gapmers with ASOs that displaceproteins from the toxic RNA can enhance the knockdown effect.Our study provides an additional approach for DM1 therapy andmay be applied to other RNA diseases.

ResultsRNase H-Mediated Degradation of Expanded CUG Repeats in CellCulture. To determine whether ASOs can be used to induce RNaseH-mediated degradation of CUGrepeatRNA,we designed gapmerASOs containing CAG sequences with 3–4 LNA or MOE nucleo-tides on the flanking ends and 8–10 PS nucleotides in the centerregion (Table 1). We first tested the gapmers in COSM6 cellstransiently transfected with a plasmid (DT960) containing DMPKexons 11–15, with 960 interrupted CTG repeats in exon 15 (Fig. 1 Aand B) (14). We found that both CAG gapmers with 14 nucleotides(LNA-CAG14) and 16 nucleotides (LNA-CAG16) resulted ina 70% decrease in DT960 mRNA by RT-PCR analysis (Fig. 1C).We designed the gapmers for RNase H accessibility; however,

previous studies have shown that ASOs can also initiate degra-dation through RNase H-independent pathways (12, 13). Todetermine whether the effect of the CAG gapmers is dependenton accessibility to RNase H, we tested a CAG “mixmer”(CAGmix) containing MOE and PS modifications that do notform a gap of more than 3 nucleotides, preventing binding ofRNase H (15). We found that at the same concentration (50nM), MOE-CAGmix decreased the DT960 transcript by 25%,but was significantly less efficient than the MOE-CAG14 gapmerthat resulted in 80% reduction (Fig. 1D). Thus, part of the re-duction may result from RNase H-independent mechanisms, butthe majority of the knockdown effect involves RNase H.

CAG Gapmers Are Sufficient to Disrupt RNA Foci in Cell Culture.Mutant DMPK transcripts containing long CUG repeats bindspecific proteins, including MBNL1, and accumulate in nuclearRNA foci. This is a main characteristic in DM1 patient cells andis also recapitulated in cell culture systems (3). To determine

Author contributions: J.E.L., C.F.B., and T.A.C. designed research; J.E.L. performed re-search; J.E.L. and T.A.C. analyzed data; and J.E.L., C.F.B., and T.A.C. wrote the paper.

Conflict of interest statement: C.F.B. is an employee of Isis Pharmaceuticals, Inc., and maymaterially benefit either directly or indirectly through stock options.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: tcooper@bcm.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1117019109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1117019109 PNAS Early Edition | 1 of 6

GEN

ETICS

whether CAG gapmers are able to disrupt RNA foci in cells, weperformed fluorescent in situ hybridization (FISH) using Cy3-la-beled probes that recognize CUG-containing RNA foci.We founda significant reduction of RNA foci in cells treated with LNA-CAG14 or LNA-CAG16 gapmers, whereas a control LNA gapmeragainst luciferase (α-luc) had no effect (Fig. 2A). To confirm thisresult, we designed aDT960-GFP plasmid that uses a tet-induciblebidirectional promoter to express both the expanded DMPKtranscript and GFP (Fig. 2B). By expressing both GFP and DT960RNA from the same plasmid, we were able to examine whetherGFP positive cells contain RNA foci with or without administra-tion of CAG gapmers. In the absence of gapmers, 95% of GFP-positive cells were found to have RNA foci. In cells treated withthe CAG gapmers, there was a significant decrease in the numberof GFP positive cells containing RNA foci, which was verified byquantification (Fig. 2 C and D). RT-PCR confirmed that the ex-panded CUG transcripts were indeed degraded (Fig. 3E). Takentogether, these results indicate that CAG gapmers are sufficient todisrupt CUG repeat-containing RNA foci.

CAG Gapmers Specifically Target Expanded CUG Repeats. DM1 is anautosomal dominant disease in which only one allele containsthe expanded CTG repeats, whereas the other allele containsnonpathogenic repeats of 5–38 CTGs. An ideal therapy willdiscriminate between RNA from mutant and wild-type alleles,and preferentially target the former. Targeting RNA from

mutant alleles will prevent loss of DMPK protein, which couldresult in detrimental consequences (16). To determine whetherCAG gapmers discriminate between expanded and nonexpandedRNA transcripts, we transiently transfected cells expressingDMPK mRNAs containing from 12 to 960 CUG repeats andadministered increasing concentrations of the LNA-CAG14gapmer. We found that up to the highest concentration of LNA-CAG14 gapmer tested (10 nM), there was no effect on levels ofRNA containing 12 CUG repeats, which is the average numberof repeats among non-DM1 individuals (Fig. 3) (17). For RNAcontaining 40 CUG repeats, the gapmer starts to have a signifi-cant knockdown at 3 nM (P < 0.001); whereas for 240 and 480repeats, the knockdown is significant at 1 nM (P < 0.05). ForRNA containing 960 repeats, significant knockdown is achievedat as low as 0.3 nM (P < 0.01) (Fig. 3). Thus, DMPK transcriptscontaining longer repeats are affected at lower concentrations.The data suggest that the CAG gapmers can potentially targetexpanded CUG repeats compared with normal length repeats.

CAG Gapmers Induce Degradation of Expanded DMPK Transcripts inSkeletal Muscle From a DM1 Mouse Model. To assess the efficacy ofCAG gapmers in vivo, we tested them in our EpA960/HSA-CreDM1 mouse model (18). The EpA960 transgene contains the hu-man DMPK exon 15 including 960 interrupted CTG repeats. Toinducibly express the transgene specifically in skeletal muscle,EpA960 mice are crossed with HSA-CreERT2 mice (19) andbitransgenic animals were mated to generate double homozygousEpA960/HSA-Cre mice. Mice 2–3 mo of age were injected withtamoxifen to induce expression of the mRNA containing 960 CUGrepeats. At least 2 wk postinduction,MOE-CAG14 gapmers (2 μg)were injected into the tibialis anterior (TA) muscle of one hindlimb, followed by in vivo electroporation. The TA muscle of thecontralateral hind limb was injected with MOE-CTG14 gapmer ascontrol and electroporated following the same procedure. Thelevel of EpA960 mRNA was determined by real-time RT-PCR 2wk after gapmer administration. We observed a 50% decrease inthe level of EpA960 transcript in theTAmuscle treatedwithMOE-CAG14 compared with the CTG control (Fig. 4A). This result wasconfirmed by standard RT-PCR assays (Fig. 4B). Consistent withthe EpA960 transcript level, we observed a 40% decrease in theaverage number of RNA foci per nuclei by FISH (Fig. 4 C and D).This result shows that CAG gapmers can induce degradation ofexpanded DMPK transcripts and disrupt RNA foci in DM1 mice.To assess the downstream effect of expanded CUG transcript

degradation on aberrant splicing, we examined the alternativesplicing of three DM1-associated splicing events: Cypher, Serca1,

Table 1. Antisense oligonucleotide sequences andmodifications

ASO name Sequence Modifications Length (nt)

LNA-CAG14 AGC AGCAGCAG CAG LNA/PS 14LNA-CAG16 CAG CAGCAGCAGC AGC LNA/PS 16GAC14 ACG ACGACGAC GAC LNA/PS 14α-luc TTC CCGTCATCGT CTTT LNA/PS 17MOE-CAG14 AGC AGCAGCAG CAG MOE/PS 14MOE-CTG14 TGC TGCTGCTG CTG MOE/PS 16CAGmix AG CAG CA G CA G CAG MOE/PS 14morCAG13 AGCAGCAGCAGCA morpholino 13morCAG25 AGCAGCAGCAGCAGCA

GCAGCAGCAmorpholino 25

Phosphorothioate (PS) nucleotides are in bold, locked nucleic acid (LNA)or 2′-O-Methoxyelthyl (MOE) nucleotides are underlined.

A B

C D

Fig. 1. RNase H-mediated degradation of expanded DMPKtranscript in cell culture. (A) Diagram of the DT960 minigeneconstruct. The DT960 minigene contains the human DMPKgenomic segmentwith exons 11–15 and 960 interrupted CTGrepeats expressed by a CMV promoter/enhancer. Primerpairs for RT-PCR [E15upF (forward) and E15upR (reverse)] arelocated in exon 15 upstream of the repeats (indicated byarrows). (B) Schematic of the experimental strategy. DT960plasmid was transiently transfected into COSM6 cells, fol-lowed by transfection of gapmers 24 h later. RNA isolation orfluorescence in situ hybridization (FISH) was performed thenext day. (C) (Upper) Standard RT-PCR showed 70% re-duction of DT960 RNA in cells treated with 50 nM of LNA-CAG14 or LNA-CAG16 gapmers. A gapmer complementaryto luciferase (α-luc) served as control. Poly-A binding protein(PABP) was used as internal control. The results of four in-dependent transfection experiments were averaged (***P <0.001). (Lower) Representative gel image used for quantifi-cation. (D) (Upper) Standard RT-PCR results. MOE-CAGmixtreatment resulted in 25% decrease of the DT960 RNA,whereas the MOE-CAG14 gapmer induced 80% reductioncompared with mock. The data represent the average ofthree independent transfection experiments (***P < 0.001).(Lower) Representative gel image used for quantification.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1117019109 Lee et al.

and Clcn1 (20–22). At time points more than 2 wk after tamoxifeninjection, all three events reverted toward the embryonic splicingpattern as expected (Fig. 4E, +Tam). MOE-CAG14 administra-tion (2 μg) partially reversed the embryonic splicing isoforms to-ward adult splicing isoforms compared with the MOE-CTG14control, consistent with the reduction of EpA960 transcripts. In-terestingly, however, treatment with a lower dose (0.5 μg) ofMOEgapmers had a slightly stronger effect on splicing (Fig. 4E) withoutreduction of EpA960 RNA (Fig. S1), implicating other mecha-nisms besides reduction of repeat RNA that could contribute tothe splicing change. Muscle injected with PBS and electroporatedwith the same procedures (mock) did not show splicing changescompared with untreated posttamoxifen mice (+Tam). Surpris-ingly, the splicing of Cypher and Serca1 shifted more toward theembryonic pattern in the muscle injected with the MOE-CTG14control. This result suggested that electroporation of the MOE

gapmers induced muscle damage and regeneration (see below),which causes a splicing switch toward embryonic patterns (23).

CAG Gapmers Induce Histological Abnormalities in Mouse SkeletalMuscle. We have shown that CAG gapmers reduce the level ofpathogenic CUG RNA in cell culture and mouse skeletal muscle.We next tested whether there are off-target effects on endoge-nous transcripts containing CUG repeats. There are at leasteleven genes in mice that express mRNAs containing ≥ 6 CUGrepeats (12, 13). To determine the nontarget effect of thegapmers, we examined the mRNA levels of three genes: Map-kap1, Mllt3 and Pcolce, which contain 26, 8 and 12 CUG repeatsrespectively. Real-time and standard RT-PCR showed no sig-nificant difference in the level of these transcripts in muscletreated with MOE-CTG14 control or MOE-CAG14 (Fig. 5 Aand B). This result is consistent with our cell culture experiments,where the CAG gapmers preferentially target the longer repeats.To further investigate secondary effects of the CAG gapmers

in mouse skeletal muscle, we examined the histology of the TAmuscle 2 wk postgapmer administration. PBS-treated muscle(mock) appeared essentially normal indicating that neither ta-moxifen nor the injection/electroporation procedure inducedpersistent histopathology. However, muscles treated with eitherthe MOE-CTG14 control gapmer or MOE-CAG14 showed focalregions with central nuclei and small myofibers containing mul-tiple nuclei (Fig. 5C). The histological abnormalities appear toreflect muscle damage and regeneration induced by gapmeradministration following electroporation, which explains theenhanced embryonic splicing patterns induced by the CTGcontrol gapmer (Fig. 4E). The fact that both the MOE-CTG andMOE-CAG14 gapmers induce regeneration suggests that it isthe gapmer chemistry that is toxic rather than the specific se-quence. In addition, the MOE-CAG14 gapmer also induces re-generation, strongly suggesting that their effects to reverse theembryonic pattern by degradation of expanded CUG RNA isstronger than reflected in Fig. 4 because a component of theembryonic pattern is due to regeneration in addition to expres-sion of expanded CUG RNA. We conclude that CAG gapmersdo not predominantly target endogenous CUG transcripts; how-ever, both CTG and CAG gapmers induce some level of muscledamage when electroporated into muscle tissue.

Combined Administration of CAG Morpholinos and MorpholinosEnhance CUG RNA Knockdown. A previous study demonstratedthat CAG-containing morpholinos (CAG25) bind to expandedCUG RNA, block sequestration of MBNL1, and reverse mo-lecular features caused by MBNL1 loss-of-function. Morpholinosare not recognized by RNase H; however, CAG25 caused a 50%reduction of CUG repeat RNA presumably by releasing RNAfrom foci and enhancing its degradation (12).

A

B

C

D E

Fig. 2. Disruption of RNA foci by CAG gapmers in cell culture. (A) RNA focicontaining DT960 RNA were detected by FISH using Cy3-labeled probes.Nuclei were counterstained with DAPI (Lower). All images were taken at thesame exposure time. (Scale bars: 20 μm.) (B) Diagram of the DT960-GFPconstruct. The expression of DMPK RNA and GFP are controlled by a bi-directional tetracycline responsive element (TRE), which is activated by thetransactivator in the presence of doxycycline (TetON). Primers (E15upF andE15upR) for RT-PCR are indicated with arrows. (C) FISH was performed oncells transfected with transfection reagent only (mock), GAC14 controlgapmer, LNA-CAG14 and LNA-CAG16. All images were taken at the sameexposure time. (Scale bars: 20 μm.) (D) Bar graph represents average percentof GFP+ cells containing RNA foci. For the results, ≥7 microscopic fields anda total of >90 cells were counted from three independent transfectionexperiments. (***P < 0.001). (E) Reduction of DT960 transcript was con-firmed by standard RT-PCR. Poly-A binding protein (PABP) was used as in-ternal control.

Fig. 3. CAG gapmers preferentially degrade RNAs containing expandedCUG repeats. Cells expressing identical DMPK transcripts except containing12, 40, 240, 480, or 960 CUG repeats were treated with increased dosage (0,0.1, 0.3, 1, 3, 10 nM) of LNA-CAG14 gapmer. LNA-CAG14 had no effect onRNA containing 12 CUG repeats. DMPK transcripts containing longer repeatsare affected at lower concentrations. Data represent average of three in-dependent experiments.

Lee et al. PNAS Early Edition | 3 of 6

GEN

ETICS

We wanted to determine whether morpholinos could enhancethe effect of the CAG gapmers through release of the expandedCUG RNA from nuclear foci. DT960-transfected cells were ex-posed to mixtures of LNA-CAG14 gapmer combined with in-creasing concentrations of a CAG morpholino containing 13nucleotides (morCAG13). morCAG13 alone was able to induce50% decrease in the level of DT960 RNA at 0.3 μM, consistentwith what has been reported (12). An enhanced knockdown effectwas achieved when 0.1 and 0.3 μMofmorpholino was added to 0.1nM of gapmer compared with cells treated with gapmer only.However, this enhancement was compromised when the mor-pholino concentration increased to 1 μM (Fig. 6A). One explana-tion is that the CAGmorpholino competes with the CAG gapmersfor the same CUG binding sites, therefore reducing the RNase H-mediated knockdown by gapmers at higher concentrations. Whenthe gapmer concentration increases to 1 nM, addition of mor-pholino no longer enhances degradation. Note that the concentra-

tion of morpholino tested is roughly 1,000 times that of thegapmer, indicating that much lower amount of gapmer is requiredfor significant knockdown.We then treated EpA960/HSA-Cre mice with a combination

of MOE-CAG14 (2 μg) and morCAG25 (20 μg) to determinewhether a similar synergistic effect occurs in vivo. AlthoughMOE-CAG14 by itself induced 50% decrease in the level of EpA960transcript, addition of morCAG25 further enhanced the reductionto 75% (Fig. 6B). These results indicate that combining ASOs thathelp release CUG transcripts from nuclear foci, with gapmersthat target the CUGs for degradation, can have a synergistic effectat certain concentrations.

DiscussionThe main goal of this study was to target degradation of ex-panded CUG repeats in DM1 by using gapmer ASOs. We

A

C

E

B

D

Fig. 4. CAG gapmer administration reduces expanded CUG RNA levels and aberrant splicing in a DM1 mouse model. (A) Real-time RT-PCR indicates a 50%decrease of the relative EpA960 transcript level in the TA muscle 2 wk after administration of MOE-CAG14 compared with the MOE-CTG14 control (n = 6 mice,triplicate assays per sample, **P < 0.01). β-actin was used as the internal control. (B) Standard RT-PCR shows a decrease in EpA960 transcript in muscle treatedwith MOE-CAG14 compared with the MOE-CTG14 control. Representative results from four treated mice (numbered) are shown. (C) Fewer RNA foci weredetected in muscle treated with MOE-CAG14 gapmer compared MOE-CTG14 control. Nuclei were stained using DAPI (Right). All images were taken at thesame exposure time. (Inset) Higher magnification of foci. (Scale bars: 20 μm.) (D) Quantification reveals 40% reduction of the number of foci per nucleus inMOE-CAG14 muscle compared with control. Five microscopic fields were counted per muscle. Bar graph represents the average of [number of foci/number ofnuclei] each field (n = 5 mice). (E) RT-PCR quantification of inclusion of alternative exons from three misregulated splicing events in DM1. Lanes are as follows:EpA960/HSA-Cre mice not injected with tamoxifen, n = 3 (-Tam); untreated mice 2 wk post tam, n = 4 (+Tam); mock-treated mice 2 wk post tam, n = 4 (mock);tamoxifen-induced EpA960/HSA-Cre mouse muscle injected with MOE-CTG14 control (2 μg), n = 6 ; MOE-CAG14 (2 μg), n = 6; MOE-CTG14 control (0.5 μg), n =7; MOE-CAG14 (0.5 μg), n = 7 (*P < 0.05, **P < 0.01, ***P < 0.001).

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1117019109 Lee et al.

designed gapmers with CAG repeat sequences containing LNAand MOE modifications, both of which are resistant to nucleasesand have increased affinity to RNA (24, 25). The gapmers weresufficient to knockdown the mutant CUG transcripts both in cellculture and DM1 mice, providing proof of principle that theRNase H pathway may be exploited for use in DM1 therapy.We showed that the CAG gapmers are able to preferentially

target long CUG repeats, which is likely to be due to the in-creased availability of hybridization sites with longer expansionsas well as preferential localization of expanded transcripts in thenucleus where RNase H is localized. 2′-OMe ASOs have alsobeen shown to selectively target expanded CUG repeats in DM1myoblasts (13). Another study showed that ASOs used to inhibittranslation of CAG expansions in Huntington disease only af-fected the expanded huntingtin allele but not the wild-type allele(26). This provides evidence that the ability to preferentiallytarget expanded repeat sequence may be common to other ASOtherapies.We found that CAG gapmers reduce expanded DMPK tran-

scripts by 80% and almost eliminate RNA foci in cell culture.However, a significant but smaller reduction (50%) was obtained inmouse skeletal muscle. Possible reasons for the decreased perfor-mance in mice include lower efficiency of delivery and variable lo-calization of gapmers in muscle tissue. We found that fluorescentlytagged ASOs transfected into cultured cells localize predominantlyin the nucleus, consistent with previous results (27). Moreover, 2′-OMe ASOs injected in mouse muscle are concentrated in nuclei

(13). It seems likely that a large fraction of the gapmers retained inmuscle enter the nucleus given the 50% reduction of CUGRNA. Inaddition, CAG oligos can form self-structures and it is likely ASOsdesigned to have decreased propensity to form structure will in-crease availability to bind to target RNA.Although the CAG gapmers reduced the level of toxic CUG

RNA in skeletal muscle tissue by half, we observed only a slightreversal of Cypher, Serca1, and Clcn1 splicing when comparingCAG and CTG control gapmers. It is possible that 50% reductiondoes not surpass a threshold needed to fully rescue these splicingevents. However, it is also likely that the regenerative responseinduced by the gapmers obfuscated the reversal of the embryonicsplicing pattern.We have recently shown that muscle regenerationproduces an adult-to-embryonic switch in splicing (23). Therefore,the rescue of splicing might have been mitigated by production ofthe embryonic splicing pattern in regenerating fibers. On the otherhand, we found that a lower dose of CAG gapmer treatment canresult in reversal of splicing even without reduction of toxic RNA.This finding suggests that the gapmers may work through otherRNase-H independent mechanisms to affect downstream splicingevents, perhaps through displacement of MBNL1 (15).There are at least two possible explanations of how ASO

treatment may result in muscle damage: (i) Sequence-specific off-target effects, which seem unlikely because both CTG and CAGgapmers resulted in similar histological changes. (ii) Sequence-dependent aptameric properties: ASOs with CAG and CTGrepeats are predicted to form higher order structures whichmay bemore immunostimulatory or induce aptameric effects (28, 29).Our CAG gapmers contain MOE modifications which are knownto decrease the immunostimulatory effects (30), but the centralregion is not MOE-modified. On the other hand, MOE modifiednucleotides prevent the ASO from degradation, which may en-hance the toxic properties when electroporated into cells. There islimited knowledge of adverse effects of chemically modified ASOsin muscle tissue and side effects can be minimized with futuredevelopment of antisense technology.We also tested an approach of combined administration of

RNase H-active gapmers with RNase H-inactive morpholinos incell culture and mouse models for DM1. We found that CAGmorpholinos could enhance the knockdown by CAG gapmers atspecific concentration ratios. However, once the morpholinoconcentration exceeds a threshold, degradation is inhibited. Thiseffect is expected because the morpholinos and gapmers com-pete for the CUG repeats and the high concentrations of mor-pholino are likely to displace the gapmers, inhibiting RNase Haccessibility. A synergistic effect of morpholinos and gapmerswas also observed in mice. A potential alternative approach is touse combined administration of CAG morpholinos with gapmerstargeting different sequences within the DMPK mRNA to avoidbinding competition. In addition, several reports have identifiedsmall molecules and peptides with the ability to disrupt RNA fociformation (31, 32). This opens up the possibility of combiningASOs with other nonantisense strategies for DM1 therapy.We did not test systemic delivery in this study due to toxicity

concerns; however, it is possible that the toxicity we observed was

A

C

B

Fig. 5. Secondary effects of CAG gapmers. (A) Real-time RT-PCR of threegene transcripts containing ≥8 CUG repeats. RNA levels are expressed asarbitrary units relative to β-actin, then normalized to the mean expression ofcontrol muscle. (n = 4 mice, triplicate assays per sample, P > 0.5 for each). (B)No consistent difference in expression level was seen by standard RT-PCR.Data from four mice is shown. (C) Hematoxylin and eosin staining of TAmuscle cross-sections. Central nuclei (arrowheads) and regions with multiplenuclei (asterisks) are present in muscle treated with MOE-CTG14 control andMOE-CAG14 gapmers. (Scale bar = 100 μm.)

A BFig. 6. Combined effect of CAG gapmers and mor-pholinos. (A) Following expression of DT960 RNA inCOSM6 cells, LNA-CAG14 gapmer (0, 0.1, 1 nM) wastransfected into COSM6 cells with increasing doses(0, 0.1, 0.3, 1 μM) of CAG morpholino containing 13nucleotides (morCAG13). β-actin was used as internalcontrol. (B) Real-time RT-PCR revealed enhanceddecrease of EpA960 transcript level when MOE-CAG14 (2 μg) was combined with morCAG25 (20 μg).EpA960 transcript levels from PBS-treated muscleserved as control (mock). n = 4–6 mice per group,triplicates were done per sample. β-actin was used asinternal control. (**P < 0.01, ***P < 0.001).

Lee et al. PNAS Early Edition | 5 of 6

GEN

ETICS

secondary to an inflammatory response stimulated by the ASOcombined with tissue damage due to the method of delivery.Different routes of administration as well as different chemistrieswill need to be tested to lower toxicity and improve efficacy. Thereare several systemically administered RNase H-based antisensedrugs in clinical trials, suggesting bright prospects in the future(10). The growing interest on therapeutic ASOs will acceleratedevelopment of novel nucleotide modifications and deliverymethods addressing issues of toxicity and bio-distribution.

Materials and MethodsOligonucleotides. LNA gapmers were purchased from Exiqon. MOE gapmerswere obtained from ISIS Pharmaceuticals. CAG morpholinos were purchasedfrom Gene Tools.

Cell Culture and Transfection. COSM6 cells were plated in six-well platescontaining DMEM supplemented with 10% certified FBS and 1% L-glutamine(all from Gibco). Twenty-four h after plating, 1 μg of CTG repeat plasmid wastransfected per well using Fugene6 (Roche). The next day, indicated con-centrations of gapmers were transfected using Lipofectamine 2000 (Invi-trogen). Morpholinos were transfected using the Neon transfection system(Invitrogen).

Transgenic Mice and ASO Injection. Animal experiments were approved by theInstitutional Animal Care and Use Committee of Baylor College of Medicine.EpA960 and HSA-Cre-ERT2 mice were used to generate double homozygousEpA960/HSA-Cre mice (18). Mice were injected daily for 5 consecutive dayswith 1 mg of tamoxifen. At least 2–4 wk after tamoxifen injection, mice wereanesthetized by i.p. injection of Avertin (0.5 mg/g weight). The tibialis an-terior (TA) muscle was pretreated with bovine hyaluronidase (Sigma-Aldrich)for 1–2 h and the indicated amount of ASO was injected intramuscularlyfollowed by in vivo electoporation (100 V/cm, 10 pulses at 1 Hz, 20-ms du-ration per pulse).

Fluorescence in Situ Hybridization (FISH). Cells were fixed in 4% para-formaldehyde and permeabilized with 0.02% Triton X-100 in PBS. Mousemuscle tissues were fixed overnight in 10% formalin. Tissues were thenparaffin-embedded and cut in cross-section at 10 μm. CUG transcripts weredetected using (CAG)5-Cy3-labeled LNA probes (Exiqon) as described (18).The nuclei were stained with DAPI using Vectashield (Vector).

RT-PCR. Total RNAwas isolated from skeletal muscle using TRIzol (Invitrogen).cDNA was generated from 4 μg of RNA using oligo dT and AMV reversetranscriptase (Life Science). To assay alternative splicing events, flankingprimer pairs were designed for Clcn1, Serca1, and Cypher. See Table S1 foroligo sequences. PCR products were separated on 5% nondenaturing poly-acrylamide gels and quantified using the Kodak Gel Logic 2200 and Mo-lecular Imaging software.

Real-Time RT-PCR. Taqman primers were used to quantify EpA960 repeatrecombined allele mRNA products as described (18). Primer pairs for Map-kap1, Mllt3 and Pcolce were purchased from Applied Biosystems. Real-timeRT-PCR was performed on the ABI Prism 7000 sequence detection system. Allsamples were normalized to β-actin (Applied Biosystems part no. 4352341E).

Statistics. All data are expressed as mean ± SD. Statistical significance wasdetermined using two-tailed Student t test. A P value of less than 0.05 wasconsidered significant.

ACKNOWLEDGMENTS. We thank Dr. C. Thornton and Dr. T. Wheeler(University of Rochester) for their expertise with the in vivo electroporationexperiments, M. Koshelev for initial help with the project, andD. Bundman fortechnical assistance. All histology was performed by the Center for Compar-ative Medicine pathology core facility at Baylor College of Medicine, witha special thanks to B. Bhatti. This workwas supported by theNational Institutesof Health Grant R01AR45653 (to T.A.C.), the Muscular Dystrophy Association(T.A.C.), and the Shanna and Andrew Linbeck Family Charitable Fund.

1. Harper P.S. (2001) Myotonic Dystrophy (W.B. Saunders, London).2. Brook JD, et al. (1992) Molecular basis of myotonic dystrophy: expansion of a tri-

nucleotide (CTG) repeat at the 3′ end of a transcript encoding a protein kinase familymember. Cell 68:799–808.

3. Davis BM, McCurrach ME, Taneja KL, Singer RH, Housman DE (1997) Expansion ofa CUG trinucleotide repeat in the 3′ untranslated region of myotonic dystrophyprotein kinase transcripts results in nuclear retention of transcripts. Proc Natl Acad SciUSA 94:7388–7393.

4. Miller JW, et al. (2000) Recruitment of human muscleblind proteins to (CUG)(n) ex-pansions associated with myotonic dystrophy. EMBO J 19:4439–4448.

5. Kuyumcu-Martinez NM, Wang GS, Cooper TA (2007) Increased steady-state levels ofCUGBP1 in myotonic dystrophy 1 are due to PKC-mediated hyperphosphorylation.Mol Cell 28:68–78.

6. Lee JE, Cooper TA (2009) Pathogenic mechanisms of myotonic dystrophy. Biochem SocTrans 37:1281–1286.

7. Bennett CF, Swayze EE (2010) RNA targeting therapeutics: molecular mechanisms ofantisense oligonucleotides as a therapeutic platform. Annu Rev Pharmacol Toxicol 50:259–293.

8. Stein H, Hausen P (1969) Enzyme from calf thymus degrading the RNAmoiety of DNA-RNA Hybrids: effect on DNA-dependent RNA polymerase. Science 166:393–395.

9. Walder RY, Walder JA (1988) Role of RNase H in hybrid-arrested translation by an-tisense oligonucleotides. Proc Natl Acad Sci USA 85:5011–5015.

10. Wahlestedt C, et al. (2000) Potent and nontoxic antisense oligonucleotides containinglocked nucleic acids. Proc Natl Acad Sci USA 97:5633–5638.

11. LimaWF, et al. (2004) Structural requirements at the catalytic site of the heteroduplexsubstrate for human RNase H1 catalysis. J Biol Chem 279:36317–36326.

12. Wheeler TM, et al. (2009) Reversal of RNA dominance by displacement of proteinsequestered on triplet repeat RNA. Science 325:336–339.

13. Mulders SA, et al. (2009) Triplet-repeat oligonucleotide-mediated reversal of RNAtoxicity in myotonic dystrophy. Proc Natl Acad Sci USA 106:13915–13920.

14. Philips AV, Timchenko LT, Cooper TA (1998) Disruption of splicing regulated bya CUG-binding protein in myotonic dystrophy. Science 280:737–741.

15. Monia BP, et al. (1993) Evaluation of 2′-modified oligonucleotides containing 2′-de-oxy gaps as antisense inhibitors of gene expression. J Biol Chem 268:14514–14522.

16. Jansen G, et al. (1996) Abnormal myotonic dystrophy protein kinase levels produceonly mild myopathy in mice. Nat Genet 13:316–324.

17. Ansved T, Edström L, Grandell U, Hedberg B, Anvret M (1997) Variation of CTG-repeatnumber of the DMPK gene in muscle tissue. Neuromuscul Disord 7:152–155.

18. Orengo JP, et al. (2008) Expanded CTG repeats within the DMPK 3′ UTR causes severeskeletal muscle wasting in an inducible mouse model for myotonic dystrophy. ProcNatl Acad Sci USA 105:2646–2651.

19. Schuler M, Ali F, Metzger E, Chambon P, Metzger D (2005) Temporally controlledtargeted somatic mutagenesis in skeletal muscles of the mouse. Genesis 41:165–170.

20. Kanadia RN, et al. (2006) Reversal of RNA missplicing and myotonia after muscleblindoverexpression in a mouse poly(CUG) model for myotonic dystrophy. Proc Natl AcadSci USA 103:11748–11753.

21. Kimura T, et al. (2005) Altered mRNA splicing of the skeletal muscle ryanodine re-ceptor and sarcoplasmic/endoplasmic reticulum Ca2+-ATPase in myotonic dystrophytype 1. Hum Mol Genet 14:2189–2200.

22. Mankodi A, et al. (2002) Expanded CUG repeats trigger aberrant splicing of ClC-1chloride channel pre-mRNA and hyperexcitability of skeletal muscle in myotonicdystrophy. Mol Cell 10:35–44.

23. Orengo JP, Ward AJ, Cooper TA (2011) Alternative splicing dysregulation secondaryto skeletal muscle regeneration. Ann Neurol 69:681–690.

24. Kurreck J, Wyszko E, Gillen C, Erdmann VA (2002) Design of antisense oligonucleo-tides stabilized by locked nucleic acids. Nucleic Acids Res 30:1911–1918.

25. Monia BP, Johnston JF, Sasmor H, Cummins LL (1996) Nuclease resistance and anti-sense activity of modified oligonucleotides targeted to Ha-ras. J Biol Chem 271:14533–14540.

26. Hu J, et al. (2009) Allele-specific silencing of mutant huntingtin and ataxin-3 genes bytargeting expanded CAG repeats in mRNAs. Nat Biotechnol 27:478–484.

27. Leonetti JP, Mechti N, Degols G, Gagnor C, Lebleu B (1991) Intracellular distribution ofmicroinjected antisense oligonucleotides. Proc Natl Acad Sci USA 88:2702–2706.

28. Gürsel M, Verthelyi D, Gürsel I, Ishii KJ, Klinman DM (2002) Differential and com-petitive activation of human immune cells by distinct classes of CpG oligodeox-ynucleotide. J Leukoc Biol 71:813–820.

29. Vollmer J, et al. (2004) Characterization of three CpG oligodeoxynucleotide classeswith distinct immunostimulatory activities. Eur J Immunol 34:251–262.

30. Henry S, et al. (2000) Chemically modified oligonucleotides exhibit decreased immunestimulation in mice. J Pharmacol Exp Ther 292:468–479.

31. Warf MB, Nakamori M, Matthys CM, Thornton CA, Berglund JA (2009) Pentamidinereverses the splicing defects associated with myotonic dystrophy. Proc Natl Acad SciUSA 106:18551–18556.

32. García-López A, Llamusí B, Orzáez M, Pérez-Payá E, Artero RD (2011) In vivo discoveryof a peptide that prevents CUG-RNA hairpin formation and reverses RNA toxicity inmyotonic dystrophy models. Proc Natl Acad Sci USA 108:11866–11871.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1117019109 Lee et al.