Short antisense-locked nucleic acids (all-LNAs) correct alternative splicing abnormalities in myotonic dystrophy

Myotonic dystrophy type 1 (DM1) is an autosomal dominant multisystemic disorder caused by expansion of CTG triplet repeats in 3′-untranslated region of DMPK gene. The pathomechanism of DM1 is driven by accumulation of toxic transcripts containing expanded CUG repeats (CUGexp) in nuclear foci which sequester several factors regulating RNA metabolism, such as Muscleblind-like proteins (MBNLs). In this work, we utilized very short chemically modified antisense oligonucleotides composed exclusively of locked nucleic acids (all-LNAs) complementary to CUG repeats, as potential therapeutic agents against DM1. Our in vitro data demonstrated that very short, 8- or 10-unit all-LNAs effectively bound the CUG repeat RNA and prevented the formation of CUGexp/MBNL complexes. In proliferating DM1 cells as well as in skeletal muscles of DM1 mouse model the all-LNAs induced the reduction of the number and size of CUGexp foci and corrected MBNL-sensitive alternative splicing defects with high efficacy and specificity. The all-LNAs had low impact on the cellular level of CUGexp-containing transcripts and did not affect the expression of other transcripts with short CUG repeats. Our data strongly indicate that short all-LNAs complementary to CUG repeats are a promising therapeutic tool against DM1.


INTRODUCTION
Myotonic dystrophy type 1 (DM1) is the most common muscular dystrophy in adults, affecting 1 in 6-10 000 live birth. Clinical features of DM1 mainly present myotonia, weakness and atrophy of skeletal muscles (1). This RNA dominant disorder is caused by expansion of CTG repeats (from ∼50 to ∼3000 repeats) in the 3 untranslated region (3 UTR) of the dystrophia myotonica protein kinase (DMPK) gene (2). The DM1 pathogenesis is based on the dominant gain-of-function of the DMPK transcript containing expanded number of CUG repeats that accumulate in cell nuclei as discreet foci (3). Expanded CUG repeats (CUG exp ) interact with RNA-binding proteins such as Muscleblind-like (MBNL) protein family, causing their nuclear sequestration and subsequently resulting in reduced protein activity in several cellular processes (4,5). Another pathogenic effect of DMPK mutation in DM1 is the induction of post-transcriptional upregulation of another RNA binding protein, CUGBP1 (CELF1). According to several reports, an unknown mechanism may lead to CUGBP1 phosphorylation resulting in its increased cellular stability (6). MBNLs and CUGBP1 are antagonistic regulators of alternative splicing, and both control developmentally regulated pre-mRNA maturation of many genes. Their functional imbalance leads to embryonic patterns of alternative splicing in adult DM1 tissues (7).
In the current study we aimed to check the potency of short synthetic oligomers composed exclusively of LNA subunits, to inhibit the toxic interaction of CUG exp RNA with nuclear proteins in vitro and in vivo. The LNA is a ribonucleotide homolog with a characteristic 2 -O,4 -Cmethylene bridge (19)(20)(21)(22)(23). LNA-containing oligomers are strict RNA mimics characterized by several attractive properties, such as high biological stability, low toxicity in vivo, potent biological activity and high melting temperature (T m ) when bound to either DNA or RNA (19,21,23). A single LNA subunit substitution into DNA or RNA oligomer typically increases T m by 2-8 • C (19,20). Due to these attributes, nucleic acids containing additional LNA subunits have been used as 'mixmers' (several LNA subunits in random positions of oligomer) or 'gapmers' (several LNA subunits on both sides of oligomer) in many applications such as RNase H-induced gene silencing (if the central part of gapmer contains DNA nucleotides) (19,20,22,23), RNAi (23)(24)(25) or steric blocking of micro-RNA activity (21,23,26,27). In contrast, all-LNA oligomers are able to stably bind target DNA or RNA molecules without causing their endonucleolytic cleavage by RNase H (19). Previously, oligomers composed exclusively of LNA subunits (all-LNA) with only eight LNA residues were successfully applied to bind the seed regions of micro-RNAs and block their activity in vitro and in vivo (28).
Here we report the use of short antisense all-LNA oligomers to bind efficiently to long CUG exp region of mutant DMPK transcripts in DM1 patients-derived cells and in skeletal muscle of mouse model of DM1. In cells or muscle tissue treated with all-LNAs, we observed significant reduction of nuclear CUG exp foci formation, prevention of MBNLs sequestration and specific correction of alternative splicing abnormalities for several MBNL-sensitive exons, without accelerating decay of transcripts with long or short CUG repeats.

Alternative splicing assays and semi-quantitative gene expression assays
The total RNA was isolated from cells or temporal arteritis (TA) muscles using TRI Reagent (Sigma) according to manufacturer's protocol. First strand of cDNA was synthesized using SuperScript II/III Reverse Transcriptase (Invitrogen) with Random Primers (Promega) according to manufacturer's instruction. The quality of the reverse transcription products was assessed by polymerase chain reaction (PCR) amplification of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA. RT-PCR was performed in using GoTaq Flexi DNA Polymerase (Promega). All PCR experiments involved at least three independent biological replicates and three technical repeats. Primer sequences and PCR conditions are specified in Supplementary Table S1. The RT-PCR products were stained with Ultra Safe Blue (Syngene) according to manufacturer's instructions and separated in 2% agarose gel. The DNA bands were visualized on G:BOX system (Syngene) and the intensity of each band was measured using GeneTools software (Syngene). EC 50 was determined in Graph Pad Prism using log(inhibitor) versus response model: Y = Bottom + (Top-Bottom)/(1+10 ∧ [=((X-LogEC 50 ))). Real-time quantitative RT-PCR analysis was performed using the 7900HT Fast Real-Time PCR System (Applied Biosystems). The Power SYBR Green PCR Master Mix (Applied Biosystems) and specific primer set (0.2 M of each) were used for amplification (Supplementary Table S1). The experiments were carried out in triplicate technical repeats for three biological replicates. The relative quantification in gene expression was determined using the Ct method. The GAPDH-specific signal was used as a reference.

Quantification of normal and mutant DMPK transcript level
The DM1-derived myoblast line (10009) is heterozygous for Bpm1 polymorphism in exon 10 of DMPK gene (29,30). This cell line was passaged and grown in EMEM medium to selected fibroblast cells for 2 weeks. In order to distinguish the two alleles of DMPK gene, PCR product (primer--Supplementary Table S1) was digested Bpm1 enzyme (ThermoScientific) (3 h/30 • C). The samples were run on a native 10% PA/TBE gel. The gel was stained with Syber Green (Invitrogen) and visualized on Gel imaging for fluorescence applications G:BOX (Syngene). The intensity of each band was measured using GeneTools software (Syngene).

Fluorescence in situ hybridization
The RNA fluorescence in situ hybridization (FISH) was performed with the use of DNA/LNA probe (CAG) 6 -CA, labeled at the 5 -end with Cy-3 as previously described (31). Microscopic slides were mounted using medium containing 2% propyl gallate (Sigma), 10% glycerol and 4 ,6diamidino-2-phenylindole (DAPI), and then sealed with fingernail polish. All FISH images were acquired on the Nikon A1Rsi microscope equipped with the Nikon Digi-talSight DS-Fi1c camera and processed with Nikon NIS Elements AR software. The quantification of RNA foci was performed using fluorescence part of microscope, filter DA/FI/TR-A-NTE (Semrock; range of excitation 387/478/555-25; emission range 433/517/613-25) and 60x oil immersion objective (Plan Apo VC 60x/1.4 Oil DIC N2). Quantification was performed by visual inspection of fluorescence spots present in cell nuclei. The number of foci was counted in at least 200 nuclei and the same experiment was repeated twice. Foci volume was estimated by Imaris software using images captured by confocal part of Nicon' microscope in the following excitation conditions: diode lasers 405 and 561 nm, dichroic mirror 405/488/561, emission filters 450/50 for DAPI and 595/50 for Cy-3. All images were acquired with sequential scanning to avoid spectral bleedthrough. Eighteen to twenty five optical sections were acquired.

Quantification of RNA/protein interaction in vitro
The MBNL1 isoform carrying sequence encoded by exons 1-4 of MBNL1 gene (NP 066368) containing glutathione-S-transferase, GST-tag, at N-terminus and His 6 -tag at Cterminus was prepared as previously described (32). Filter binding assay (FBA) was performed in a 30-l volume. To assess the MBNL1 affinity toward distinct labeled molecules, (CUG) 100 (0.1 nM), PO-LNA-CAG-8, PO-LNA-CAG-10 (0.5 nM) and PO-2 OMe-CAG-21 (0.2 nM) were incubated with the indicated concentration of the protein (ranging from 0 to 500 nM) in 1 x FBBa (FBB with 250-mM NaCl and 15-mM KCl) at 37 • C for 30 min. To analyze the inhibition of MBNL1binding to (CUG) 100 by PO-LNA-CAG and PS-LNA-CAG, 0.1 nM of labeled (CUG) 100 was first incubated with the indicated concentrations of oligomers (ranging from 0 to 10 M) at 37 • C for 30 min, and then 15 nM of MBNL1 was added to each sample and incubated as before. Twenty-five microliter of samples was loaded onto filter binding apparatus with nitrocellulose (Protran BA 85, Whatman R ) and nylon (Hybond TM N+, Amersham) membranes pre-wetted in 1x FBBa. The signal from membranes was visualized on IP through FLA-1500 and quantified using Multi Gauge software (FujiFilm). K i of the prevention of (CUG) 100 /MBNL1 complexes by LNA-CAG was measured by Graph Pad using two-phase decay curve.

Experimental mice and PO-LNA-CAG-10 treatment
Homozygous HSA LR transgenic mice (human skeletal actin long repeat; line 20b) were previously described (33). Wildtype mice on FVB background served as controls. All animal experiments were approved by the University of Rochester Institutional Animal Care and Use Committee. Injection and electroporation was performed under general anesthesia as described previously (34). Tibialis anterior (TA) muscles of adult (2-5-month-old) mice were pretreated by intramuscular injection of bovine hyaluronidase (15 l, 0.4 U/l) (Sigma-Aldrich, St. Louis, MO, USA) and after 2 h, injected with 10 g of PO-LNA-CAG-10 in 20 l of saline, or with saline alone. Immediately following muscle injection, electroporation was performed using electrodes placed parallel to the long axis of the muscle. Electroporation parameters were 100 V/cm, 10 pulses at 1 Hz and 20 ms duration per pulse. Northern blot analysis of CUG exp transcripts and RT-PCR assays monitoring alternative splicing of Atp2a1, Titin and Nfix were performed after 2 weeks from PO-LNA-CAG-10 injection as described previously (17). Electromyography (analysis of myotonia) was performed under general anesthesia as described (17). Myotonia was graded as follows: 0 indicates no myotonia; 1, occasional myotonic discharge in less than 50% of electrode insertions; 2, myotonic discharge in greater than 50% of insertions; 3, myotonic discharge with nearly every insertion.

Statistical analysis
Group data are expressed as mean ± standard deviation. Statistical significance of RT-PCR results was determined by a two-tailed Student's t-test using Microsoft Excel (*P < 0.05; **P < 0.01 and ***P < 0.001).

RESULTS
One of the pathogenic effect of a CTG repeat expansion is an undesirable interaction of toxic CUG exp RNA with several nuclear proteins (4). In this study, we tested whether very short antisense oligomers composed exclusively of LNA subunits can efficiently bind to long CUG exp hairpin structure in vitro and in vivo, in order to inhibit pathogenic RNA/protein interaction without causing activation of degradation pathways for long or short CUG repeats-containing transcripts. The stable occupation of CUG exp by LNA is meant to block the excessive binding of proteins such as MBNLs, hence restoring their normal availability in cellular environment and rescuing cellular processes in which they are involved, such as the alternative splicing regulation.

In vitro screening of antisense all-LNA oligomers complementary to CUG repeats
To determine the minimal length of oligomer composed exclusively of LNA residues sufficient to bind a CUG exp hairpin structure in vitro and to correct some of the DM1specific phenotypes in cellular models, we designed a series of all-LNA oligomers having phosphodiester linkages and differing in the length of the sequence complementary to CUG repeats (Table 1). These included: a 6-mer (PO-LNA-CAG-6), an 8-mer (PO-LNA-CAG-8), a 10-mer (PO-LNA-CAG-10) and a 12-mer (PO-LNA-CAG-12). The predicted melting temperatures (T m ) of designed oligomers differ significantly and amount to 19, 35, 49 and 56 o C, respectively (35). We expected that the thermodynamic parameters of PO-LNA-CAG-8, PO-LNA-CAG-10 and PO-LNA-CAG-12 could be sufficient for specific interaction with CUG repeats. The oligomers longer than 12 LNA subunits, with significantly higher T m , could bind non-specifically to RNA or DNA targets and could form stable intermolecular duplexes and hairpin structures, which could reduce their activity.
In the initial in vitro assays we tested whether any of the designed all-LNAs are able to bind the thermodynamically stable RNA hairpin structure formed by long CUG repeats and inhibit its interaction with recombinant MBNL1. Using EMSA, we monitored the affinity of tested oligomers to an in vitro transcript composed of ∼100 CUG repeats, referred to as (CUG) 100 (Figure 1a). The shortest oligomer did not invade the (CUG) 100 hairpin but unexpectedly, PO-LNA-CAG-8 showed slightly higher affinity toward the hairpin than longer PO-LNA-CAG-10, and much higher than PO-LNA-CAG-12. The dissociation constant values (K d ) for 8-mer and 10-mer oligomers were very low and accounted to ∼4 and ∼7 nM, respectively (Figure 1a and b). PO-LNA-CAG-12 appeared to have five to ten times higher K d (40 nM). As control, we used long 21-mer CAG repeat oligomer, with 2 -O-methyl modification in each position (PO-2 OMe-CAG-21) with predicted T m of 75 • C. The K d value for control oligomer was within the same range as for all-LNAs (5-10 nM).
Next, we asked whether tested oligomers have the potential to inhibit CUG repeat RNA/MBNL1 interaction in vitro. For this purpose we first used FBA and measured the affinity of MBNL1 to the transcript containing 35 CUG repeats (CUG) 35 , derived from normal DMPK allele, and a three times longer mutant transcript, (CUG) 100 . MBNL1 bound both types of transcripts with similar affinity (Supplementary Figure S1a and b). Then, using the same method, we tested the interaction between radiolabeled (CUG) 100 and MBNL1 in the absence or presence of each oligomer. PO-LNA-CAG-8 showed the lowest inhibition constant (K i ) which was about 100-fold lower than K i for a 10-mer and 300-fold lower than K i for a 12-mer. The inhibitory potential of the control 2 OMe-CAG-21 oligomers resembled the one observed for PO-LNA-CAG-12 (Figure 1c).
One of the possible explanations for such differences might be the competition of CUG hairpin with CAG repeat oligomers, since previous studies reported that short CUG or CAG repeat containing RNA oligomers can bind MBNL1 in vitro with relatively low K d (36). Therefore, we tested the ability of PO-LNA-CAGs and control oligomers to bind recombinant MBNL1 using FBA. We did not, however, observe any interactions between MBNL1 and tested oligomers in the 0-500-nM protein range (Supplementary Figure S1c and d).

Screening of antisense all-LNA oligomers in DM1 cells
To assess the potency of all-LNA oligomers in a physiologically relevant model of DM1, we employed three DM1 patients-derived fibroblast cell lines expressing mutant DMPK transcripts with ∼500, ∼1000 and ∼2000 CUG repeats, referred to as (CUG)500, (CUG)1000 and (CUG)2000, respectively. These cells display several DM1specific molecular phenotypes, such as the formation of CUG exp nuclear foci, sequestration of MBNL1 and aberrant alternative splicing of several MBNL-sensitive exons. In an initial attempt to find alternative exons that could serve as useful markers for splicing correction analyses in DM1 cells, we examined roughly 20 alternatively spliced transcripts known to be sensitive to MBNL level, and ∼20 transcripts regulated by other splicing factors either affected in DM1 (CUGBP1) or not affected in DM1 (PTBP1, PTBP2, NOVA1 and FOX2--used as the negative controls). The reverse transcription PCR (RT-PCR) assays showed that most of the tested alternative splicing events were not informative in our cellular models due to either low expression level or presence of only one alternative splicing isoform (100% of either exon inclusion or exon exclusion isoform) (Supplementary Figure S2a). For further studies we chose only six MBNL-sensitive splicing events, namely MBNL1 exon 7 (7), NFIX exon 7 (37), INSR exon 11 (38), MBNL2 exon 7, PHKA1 exon 19 and NCOR2 exon 45a (39) (Supplementary Figure S2b). Previously, it was reported that all of these were misregulated in skeletal muscles of DM1 patients and Mbnl1 knockout mice (39), but also were sensitive to silencing of MBNL1 in DM1 and non-DM1 fibroblasts (40). Simultaneously, we selected five MBNL-independent splicing events as controls. They include either exons affected in DM, namely SOS1 exon 26 and KIF13A exon 21 (39) or unaffected in DM, PPP3CB exon 13 (PTBP2-sensitive) (41), APLP2 exon 14 (NOVA1sensitive) (42), ECT2 exon 4 (FOX2-sensitive) (43) (Supplementary Figure S2c). In our cellular models we did not, a 'Upper case' denotes LNA modification; 'underline and bold', PS modification; ' m C; m A; m G', 2 OMe modification. b T m , prediction of melting temperature for oligomer/RNA duplex is based on www.exiqon.com/ls/Pages/ExiqonTMPredictionTool.aspx.  Figure S3). These data suggest that short all-LNAs could potentially be able to interact with CUG exp transcript located exclusively in the nucleus and may have limited effect on other cytoplasmic transcripts containing CUG repeats.
As a control we used PO-2 OMe-CAG-21 oligomer which showed high affinity toward CUG repeat RNA hairpin in our in vitro tests. Similar oligomer, but with additional phosphorothioate modification (PS-2 OMe), was previously shown in DM1 models as an effective tool inducing degradation of expanded CUG repeat RNA (15). We did not observe any influence of PO-2 OMe-CAG-21 on splicing changes of MBNL-sensitive exons in DM1 cells (Supplementary Figure S4), and we hypothesize that this could be caused by lower stability of phosphodiester containing oligomer compared to its PS-modified version. As the PO-LNA-CAG-10 showed the highest potency, we have decided to use it for all further analyses.

PO-LNA-CAG-10 specifically corrects MBNL-sensitive alternative splicing abnormalities
To be able to characterize the potency of PO-LNA-CAG-10 more precisely, we tested the broad range of oligomer concentrations (0.  Figure S5).
In order to determine specificity of PO-LNA-CAG-10, we designed three control oligomers containing 10 LNA residues: scrambled sequence with the same but mixed nucleotide composition (PO-LNA-CTR3) and two other oligomers (PO-LNA-CTR1 and PO-LNA-CTR2) with random sequences (Table 1). We observed that all control LNA oligomers transfected at 125 nM either did not have any effect on splicing of tested MBNL-sensitive transcripts, or their presence in cells shifted the splicing pattern toward the DM1-like (Supplementary Figure S6).
Next, we asked whether LNA oligomers could impact alternative splicing more globally. For this purpose, we analyzed five MBNL-independent alternative splicing events in the same DM1 cells. We did not observe any significant alternative splicing changes after treatment with PO-LNA-CAG-10 (Supplementary Figure S7). These results suggest that PO-LNA-CAG-10 very specifically corrects the alternative splicing of MBNL-sensitive exons in DM1 cells, most probably due to its high binding specificity toward CUG exp hairpin in mutant DMPK transcript and inhibition of toxic CUG exp /MBNLs interaction. Conversely, splicing pattern of alternative exons of three genes, PPP3CB, SOS1 and KIF13A, changed significantly after treatment with some of the control LNAs.
Finally, to determine the duration of PO-LNA-CAG-10 activity in dividing cells, we performed a time-course experiment ( Figure 3). The DM1 fibroblasts at ∼30% confluence were transfected with 125 nM of PO-LNA-CAG-10 and alternative splicing events of four MBNL-sensitive exons were analyzed from 3 h to 30 days post transfection. The maximal effect of PO-LNA-CAG-10, leading to the splicing pattern observed in non-DM1 cells, was acquired after 2 days from transfection. After 15 days the correction effect was ∼75% of maximal effect, despite cell divisions. Five days later the splicing pattern shifted to that of not treated DM1 cells (Figure 3). These results indicate that PO-LNA-CAG-10 is stable in cellular environment and active even after its dilution caused by cell proliferation.

PO-LNA-CAG-10 significantly reduces the number of CUG exp -containing foci without degradation of toxic CUG exp RNA
Previously published studies have shown that various antisense oligomers targeting CUG repeats, either in cellular or animal DM1 models, induce degradation of CUG expcontaining transcript (8)(9)(10)(11)(12)(13)(14)(15)(16). To determine whether PO-LNA-CAG-10 present in nuclei of DM1 cells causes degradation of mutant DMPK transcript we employed DM1 cells heterozygous for BpmI restriction site polymorphism present in exon 10 of DMPK (29,30). Both fibroblastic and myoblastic BpmI-polymorphic DM1 cell lines were transfected with 125 nM of PO-LNA-CAG-10. The semiquantitative RT-PCR-BpmI restriction fragments length polymorphism analyses showed only very mild (∼20%) reduction of mutant DMPK mRNA level in treated cells (Figure 4a). Moreover, there was no significant reduction of total level of DMPK transcript in these cells (data not shown). Additionally, we tested the effect of PO-LNA-CAG-10 on artificial transcript containing CUG exp in HEK293 and HeLa cells transiently transfected with DT960 vector expressing DMPK exons 11-15, with 960 interrupted CTG repeats in exon 15 (44). We observed that the delivery of PO-LNA-CAG-10 into these cells resulted in only ∼20% decrease in expression of artificial DMPK transcript (Figure 4b). Taken together, these data demonstrate that the tested LNA oligomer causes only slight but statistically significant reduction in the level of CUG exp containing transcripts.
We also analyzed the expression level of several other transcripts containing short CUG repeat tracts in (CUG)1000 cells and found that neither of the five tested transcripts, BPGM, CTSA, EIF2AK3, LRP8 and PAPSS2, showed statistically significant reduction of expression level after treatment with 125 nM of PO-LNA-CAG-10. The only exception was the CTSA transcript having significantly higher cellular level (Figure 4c and Supplementary Figure S8). These results strongly support the specificity of PO-LNA-CAG-10 toward mutant DMPK transcript and demonstrate that all-PO-LNA-CAGs have no impact on the level of transcripts containing complementary sequences in distinct parts of unrelated transcripts. Other details are the same as described in Figure 1d. Statistical significance was determined using two-tailed Student's t-test by comparison of average results from three independent experiments to results obtained for mock samples (*P < 0.05, **P < 0.01 and ***P < 0.001).
Next, we asked whether the interaction of PO-LNA-CAG-10 with expanded CUG repeats in DM1 cells could influence the number and size of nuclear CUG exp foci. Three DM1 cell lines differing in CUG-repeat expansion size, (CUG)500, (CUG)1000 and (CUG)2000, respectively, were treated for 48 h with PO-LNA-CAG-10, PO-LNA-CAG-8 or control oligomers. The results of RNA fluorescent in situ hybridization (FISH) revealed a significant reduction in the average number of nuclear CUG exp foci. In striking contrast to untreated or control LNA-treated cells which contained ∼60% of nuclei with three or more foci, all DM1 cells treated with PO-LNA-CAG-10 contained only ∼15% of such nuclei (Figure 5a and b and Supplementary Figure S9). Moreover, as much as ∼60% of PO-LNA-CAG-10-treated cells and only ∼15% of control LNA-treated cells contained no CUG exp foci or only a single focus. We also observed Nucleic Acids Research, 2015, Vol. 43, No. 6 3325 (Figure 5b and c). Our results indicate that the interaction of PO-LNA-CAGs with mutant DMPK transcript does not influence cellular degradation of targeted RNAs but instead, significantly, reduces both the number and the size of toxic CUG exp foci, most probably due to dispersion of the mutant DMPK transcripts in these cells.

Phosphorothioated LNA oligomers do not correct alternative splicing after unassisted delivery into DM1 cells
Recently, Stein and collaborators used an intracellular delivery method called gymnosis (45) that does not require the use of any transfection reagent or conjugates, but rather relies on unassisted free uptake of 'naked' oligomers added directly to cell culture medium. Although gymnosis was originally demonstrated using phosphorothioated LNA (PS-LNA) gapmers (45), efficient unassisted uptake was also described for unmodified all-PS-LNAs, termed 'tiny LNAs' (28,46,47). These reports encouraged us to test whether gymnosis could be utilized to deliver LNA-CAGs. Supply LNA-CAGs to cells without any carriers could increase the chances of systemically delivery oligomers in vivo. Our experiments demonstrated that neither PO-LNA-CAG-8 nor PO-LNA-CAG-10 was able to get into cells without transfection reagents (data not shown). Therefore, we wondered whether additional chemical modification of these oligomers could improve their bioavailability. For this purpose, two fully phosphorothioated all-LNAs consisting of To analyze the potency of PS-LNA oligomers in cells, we first delivered them by lipofection. PS-LNAs were able to induce alternative splicing correction to a similar extent as unmodified PO-LNA-CAGs (Supplementary Figure S11a). Next, fully phosphorothioated LNA oligomers were delivered gymnotically into (CUG)1000 fibroblasts by incubating the cells for 72 or 216 h in a medium containing from 0.5 to as much as 10 M of a specific oligomer. Neither experiment carried out for 72 h caused any significant alternative splicing changes of four tested MBNL-sensitive exons (data not shown). However, three times longer treatment with PS-LNA-CAG-8, at 10-M concentration but not at 0.5-M concentration, resulted with slight but statistically significant correction of three out of the five analyzed splicing events (exons 7, 45a and 11 of MBNL2, NCOR2 and INSR, respectively) (Supplementary Figure S11b). Moreover, FISH analyses of CUG exp foci in DM1-fibroblasts revealed that gymnotically delivered PS-LNAs reduced the average number of CUG exp foci in DM1 cells after both, 72-and 216-h-long treatment. The reduction was minor after 72 h, however, upon extended incubation time the effect was much more prominent and similar to the one observed upon canonical transfection of 25-nM PS-LNA with lipofectamine (Supplementary Figure S11c). Taken together, our data confirm that PS-LNAs can be effectively delivered into cells by gymnosis, but to reach comparable effectiveness as with lipofectamine-delivered PS-LNAs, the concentration of the oligonucleotides must be two orders of magnitude higher.

PO-LNA-CAG-10 corrects alternative splicing in DM1 muscle cells and in DM1 mouse model
To test all-LNA oligomers in more physiologically relevant DM1 models we used both the DM1 patient-derived muscle cells and a DM1 mouse model. The DM1 and non-DM1 myoblasts were transfected with 125 nM of PO-LNA-CAG-10 and three control all-LNA oligomers. After 48 h, we observed a significant correction of the alterna-tive splicing pattern of MBNL1 exon7, MBNL2 exon7 and NCOR2 exon45a only in PO-LNA-CAG-10-treated DM1 cells (Figure 6a). However, the percentage of exon inclusion indicative of the DM1-like splicing perturbations, although significantly reduced in PO-LNA-CAG-10-treated DM1 cells, did not decline to the level observed in non-DM1 myoblasts. Importantly, the correction effect was stronger when two consecutive transfection procedures were applied. Conversely, no significant differences in alternative splicing pattern of tested exons were observed in cells transfected with either of the control all-LNA oligomers (Figure 6a). These data strongly support the specificity and potency of PO-LNA-CAG-10. Finally, we tested the activity of PO-LNA-CAG-10 in human skeletal actin-long repeat (HSA LR ) mouse model in which skeletal muscles express the human alpha actin transcript containing ∼220 CUG repeats in the 3 UTR. This mouse model shows a number of molecular features observed in DM1 patients, including formation of nuclear CUG exp foci that sequester Mbnl proteins, alternative splicing abnormalities of Mbnl-sensitive exons and myotonic discharges in skeletal muscles (33). The TA muscles of two HSA LR mice were injected with 5 g of the PO-LNA-CAG-10 and the TA in the opposite hind limb was injected with saline only. Following the injection, each muscle was electroporated as previously described (34) to facilitate entry of oligomer into muscle fibers. Mice were sacrificed for analyses two weeks after a single oligomer injection. The RT-PCR assays were used to monitor an effect of PO-LNA-CAG-10 on correction of alternative splicing of MBNLsensitive exons. For this analysis we selected three alternative exons which showed strong splicing missregulation in HSA LR , namely Atp2a1 exon 22, Titin exon 5 and Nfix exon 7 (Figure 6b). All of them showed significant improvement of the splicing pattern in PO-LNA-CAG-10-treated muscles, and the splicing correction reached almost the level observed in wild-type mice.
Previous studies have shown that misregulated alternative splicing of Clcn1 results in repetitive action potentials (myotonic discharges) (48). Needle electrode recordings showed suppressive effect of myotonic discharges in PO-LNA-CAG-10-treated TA muscles (Figure 6c). More-over, similarly to DM1-patient-derived cells, the presence of PO-LNA-CAG-10 did not influence efficient downregulation of CUG exp transcript in vivo (Figure 6d). Altogether, our results strongly support the hypothesis that PO-LNA-CAG-10 can invade CUG exp hairpin in vivo without induction of strong targeted transcript degradation and efficiently remove MBNLs, and possibly other proteins, from pathogenic sequestration.

DISCUSSION
Chemical substitutions at the 2 -hydroxyl group with 2 -O-Me, 2 -F, 2 -MOE groups or LNA modification often improve the oligomers potency, stability and their overall pharmacokinetic properties (49). The LNA, containing a methylene linkage between the 2 -oxygen and 4 -carbon in the ribose ring, forms a class of modified nucleotides of immense interest. It has been demonstrated that duplex formation by LNA and its target strand is enthalpically favorable because of more efficient stacking of bases in duplexes, resulting in an A-type helical conformation. Biochemical and biophysical studies have shown that LNA has high affinity and specificity for the target strand. Because of these properties, LNA-containing oligomers have been used as probes in DNA sequencing and hybridization techniques (19) but also in RNAi (24) and antisense strategies (21,28,50). In this work, we applied all-LNA oligomers to inhibit RNA/protein interaction. Our results showed for the first time that very short synthetic oligomers com- posed of only eight or ten LNA subunits (PO-LNA-CAG-8, PO-LNA-CAG-10 and their phosphorothioated derivatives) can localize to cell nuclei, efficiently bind to thermodynamically stable CUG exp hairpin structures both in vitro and in vivo and inhibit toxic sequestration of nuclear proteins, including MBNLs.
Most of the previously published studies that involved targeting CUG repeats with ASOs have demonstrated that such an approach often triggers degradation of CUG expcontaining transcripts, either via nuclear RNase H path-way (12,14) or via an unknown mechanism (15,16), with the exception of morpholino CAG-25 (17). Neither the CAG-25 nor the short LNA-CAGs described in this study caused strong reduction in the level of CUG exp -containing transcripts below disease-inducing threshold in cellular and mouse models of DM1 (17). Our results also demonstrate that both types of antisense oligomers have similar potency to reduce the nuclear CUG exp foci and to correct the DM1specific alternative splicing abnormalities. The fact that all-LNAs are negatively charged and are about three times smaller than CAG-25 suggests that they could be efficiently delivered to affected tissues.
Previous studies showed that clinically promising antisense oligomers are expected to improve the target specificity, cellular stability and immunoresistance compared to RNA-based drugs (19). They mainly include additional modifications such as phosphorothioated linker. In general, the PS-modified oligomers display better affinity, higher nuclease resistance and better pharmacokinetic properties as well as lower toxicity compared to unmodified oligomers (19)(20)(21)(22)(23)28). It was reported that the renal clearance and excretion of PS-modified oligomers was significantly lower compared to non-PS oligomers, mainly due to binding plasma proteins by PS-oliogmers with high affinity (51). This binding has also consequences on higher chance of PS-LNA oligomers to enter cells without any carriers and nonspecific immune response (45,52). Previously, few strategies using phosphorothioated oligomers were tested in clinical trials (20,22,53). They include phosphorodiamidate-linked 2 OMe RNA in exon skipping therapy of Duchenne muscular dystrophy (54) and PS-LNA mixmer against miR-122 for treatment of chronic hepatitis C virus infection (26).
Therefore, to further improve bioavailability of short all-LNA oligomers, we tested the possibility of unassisted delivery of their phosphorothioated derivatives in cellbased models. After long treatment of DM1-cells with high amount, of PS-LNA-CAGs (10 M), the cellular concentration of the oligomer reached the level sufficient enough to reduce the efficacy of CUG exp foci formation, significantly inhibit MBNLs sequestration and induce alternative splicing correction. These results suggests potential application of these oligomers in systemic delivery in DM1 mouse models. Previously, naked all-PS-LNAs were shown to be efficiently delivered into many tissues by systemic delivery (28,46,47). Future in vivo strategies could also utilize oligomer-peptide or oligomer-lipid conjugates in the systemic delivery, as previously described for morpholino CAG-25 (55,56).
Our experiments demonstrated that all-LNAs are stable molecules in cellular environment. In proliferating cells the duration of beneficial effect of PO-LNA-CAG-10 on splicing correction was relatively long and reached ∼20 days in spite of significant dilution of the oligomer after several cell divisions. Moreover, the therapeutic effects in differentiated muscle fibers of HSA LR mice lasted minimum 14 days. This is at least twice as long as activity of siRNA directed against CUG exp in the same mouse model (9). Moreover, relatively high 10-g dose of PO-LNA-CAGs in intramuscular injection did not induce any significant immunestimulation or muscle degeneration effects, which for such a dose were commonly observed for RNA-based drugs (9) and LNA/DNA gapmers (14) targeting CUG repeats. Our observations indicate that short all-LNAs could be taken into consideration as a promising therapeutic tool for efficient inhibition of pathogenic interaction between RNA and other macromolecules.