Antisense oligonucleotide induced exon skipping and the dystrophin gene transcript: cocktails and chemistries
Antisense oligonucleotides (AOs) can interfere with exon recognition and intron removal during pre-mRNA processing, and induce excision of a targeted exon from the mature gene transcript. AOs have been used in vitro and in vivo to redirect dystrophin pre-mRNA processing in human and animal cells. Targeted exon skipping of selected exons in the dystrophin gene transcript can remove nonsense or frame-shifting mutations that would otherwise have lead to Duchenne Muscular Dystrophy, the most common childhood form of muscle wasting.
Although many dystrophin exons can be excised using a single AO, several exons require two motifs to be masked for efficient or specific exon skipping. Some AOs were inactive when applied individually, yet pronounced exon excision was induced in transfected cells when the AOs were used in select combinations, clearly indicating synergistic rather than cumulative effects on splicing. The necessity for AO cocktails to induce efficient exon removal was observed with 2 different chemistries, 2'-O-methyl modified bases on a phosphorothioate backbone and phosphorodiamidate morpholino oligomers. Similarly, other trends in exon skipping, as a consequence of 2'-O-methyl AO action, such as removal of additional flanking exons or variations in exon skipping efficiency with overlapping AOs, were also seen when the corresponding sequences were prepared as phosphorodiamidate morpholino oligomers.
The combination of 2 AOs, directed at appropriate motifs in target exons was found to induce very efficient targeted exon skipping during processing of the dystrophin pre-mRNA. This combinatorial effect is clearly synergistic and is not influenced by the chemistry of the AOs used to induce exon excision. A hierarchy in exon skipping efficiency, observed with overlapping AOs composed of 2'-O-methyl modified bases, was also observed when these same sequences were evaluated as phosphorodiamidate morpholino oligomers, indicating design parameters established with one chemistry may be applied to the other.
KeywordsDuchenne Muscular Dystrophy Duchenne Muscular Dystrophy Mature mRNA Becker Muscular Dystrophy Target Exon
Antisense oligonucleotides (AOs) can be used to modify gene expression through the induction of a variety of mechanisms. Oligodeoxyribonucleotides can be used to target a gene transcript for RNaseH induced degradation, whereas oligomers composed of modified bases can redirect gene expression through RNA silencing , suppressing specific mRNA translation [2, 3, 4], enhancing mRNA stability  and redirecting pre-mRNA splicing patterns .
Protein-truncating mutations in the dystrophin gene typically lead to Duchenne muscular dystrophy (DMD), the most common severe childhood form of muscle wasting (review, ). Although the size of this gene, with 79 exons spanning some 2,400 kb, and distribution of expression have posed major challenges for gene repair or replacement strategies, these features have opened other avenues for intervention, such as AO induced exon skipping. Targeted removal of selected exons can excise or by-pass protein-truncating mutations from the dystrophin pre-mRNA during the splicing process. The application of AOs to induce targeted exon skipping in the dystrophin gene has been reported by several groups, examining different animal models [8, 9, 10, 11], regions of the human dystrophin gene transcript [12, 13, 14] and a variety of AO chemistries [12, 15, 16, 17, 18]. We have recently reported a comprehensive list of AOs that can induce skipping of all dystrophin exons, excluding the first and last exons . Many exons could be targeted for excision from the mature dystrophin mRNA with a high level of efficiency and in some cases, two exons were consistently removed using a single AO, suggesting tight coordination of recognition of these exons with intron removal. However, some exons were found to be extremely difficult to dislodge, despite the evaluation of many AOs directed to the target exon. These "recalcitrant" exons could be efficiently excised from the mature mRNA in response to some select combinations of apparently ineffective AOs.
AOs composed of 2'-O-methyl modified bases on a phosphorothioate backbone (2OMeAO) have been used extensively to induce targeted exon skipping in the dystrophin gene transcript and have some advantages over the phosphorodiamidate morpholino oligomers (PMO), including ease and cost of production, and efficient in vitro delivery when administered as cationic lipoplexes. However, PMOs appear to be better suited to in vivo application, where the increased stability and cellular uptake of uncomplexed compounds allows for higher levels of sustained dystrophin exon skipping, as well as an excellent safety profile [20, 21, 22]. In this report, we describe optimization and excision of recalcitrant dystrophin exons from the mature mRNA using AO cocktails for enhanced efficiency and/or specificity. The use of either 2OMeAOs or PMOs does not seem to influence exon skipping trends, indicating optimization of AO design with the 2OMe chemistry should be directly applicable to PMOs.
Results and discussion
We have designed AOs capable of individually excising 77 of the 79 exons from the dystrophin gene transcript , yet no universal motif has been identified as a reliable target for the consistent redirection of dystrophin pre-mRNA splicing. The rationale in our approach to AO design was to first direct AOs at motifs obviously implicated in exon processing and recognition, such as the acceptor and donor splice sites, as well as exonic splicing enhancers as predicted by ESEfinder . Once some dystrophin exon skipping was observed in AO-transfected human myogenic cells, a series of overlapping AOs were then designed to target that area, in an attempt to develop more effective AOs. In many cases, a single AO was eventually developed that would induce substantial levels of targeted exon skipping, and the study would then move to another dystrophin exon. Since this is regarded as a work in progress, the dystrophin exons were regarded as a reference point and classified into four types based upon the ease of excision from the mature mRNA . Type 1 dystrophin exons are removed most efficiently (greater than 30% after in vitro transfection at 100 nM), Type 2 are less easily dislodged, while Type 3 exons are poorly excised. Type 4 dystrophin exons are "special cases", where either a single AO removed multiple exons, or multiple AOs are required to excise a single targeted exon.
We observed that approximately two out of three AOs applied to the dystrophin pre-mRNA were able to induce some exon skipping, but there was considerable variation in levels of induced exon removal, relative to the intact dystrophin transcript. In some cases, the exon excision only occurred at low levels, or was sporadic. The frequency of these sporadic exon skipping events was greater than that observed in untreated cells, indicating some interference with the splicing process. However, the lack of reproducibility, or any dose-dependant responses indicated further refinement was essential.
Dystrophin exon 20 skipping
The same trend in inducing exon 20 skipping was observed in the mdx mouse dystrophin gene transcript. Individual AOs were essentially ineffective and one AO combination was most efficient at exon excision . However, it was of interest to note that the AO annealing coordinates of the AOs in the "mouse cocktail" were different from those directed at the human dystrophin gene transcript, indicating that it may not be possible to extrapolate AO design from one species to another. Further comparisons of induced exon skipping between the mouse and human dystrophin gene transcripts are currently underway.
Dystrophin exon 65 skipping
Dystrophin exon 10 skipping
The use of AO cocktails is not limited to enhancing exon removal from the dystrophin mRNA. In the case of human dystrophin exon 10, a combination of 2 AOs was required for specific exon excision. As with other recalcitrant exons, several AOs described in Additional file 3, were unable to induce any detectable removal of exon 10 when applied individually, except for H10A(-05+16), which only excised exon 10 together with blocks of flanking exons (data not shown). Several dystrophin transcripts missing exons 10–12, 9–12, 9–14 and 9–15 were sporadically detected, with 9–12 and 9–14 being most commonly observed. These dystrophin transcripts are in-frame and, extrapolating from a mildly affected Becker muscular dystrophy patient missing exon 9–22 , would be expected to produce a shorter dystrophin that could be of near-normal function. Two AOs, H10A(-09+16) and H10A(-05+24) that overlapped H10A(-05+16) had no effect on the processing of the dystrophin transcript (data not shown).
Some AOs directed at other parts of the dystrophin gene transcript had been shown to remove one or two exons in addition to the target, and this was assumed to reflect highly coordinated processing of both exons. Targeting human or canine dystrophin exon 8 always leads to transcripts missing exons 8 and 9 [11, 19], whereas directing AOs to human exons 17, 34 or 54 induces transcripts missing the targeted exon as well as 17+18, 34+35 and 54+55 respectively .
The induced skipping of exon 10 was distinct from these cases in that larger blocks of exons were involved, and the resultant patterns were somewhat variable. However, upon combining H10A(-06+15) with H10A(+98+119) or H10A(+130+149), specific exon 10 excision could be achieved, although there was still some evidence of additional shortened transcripts induced by the individual AOs.
Trends in AO cocktail design
Additional file 4 provides an overview of the predicted ESE splice motifs masked by the AOs reported in this study, with an indication of the maximum score and number of motifs shown in brackets. One feature that was common to all effective AO cocktails directed to exon 10, 20 and 65 was the targeting of predicted SC35 motifs by both AOs in the mixture. We do not propose that predicted SC35 motifs are the most important targets for induced exon skipping, as 8 out of 41 AOs targeting Type 1 exons, that are removed at high efficiency, do not appear to be directed at any predicted SC35 motifs . The relevance of the SC35 motifs to induced exon skipping is not known and requires further investigation.
Dystrophin exon 67 had previously been classified as a Type 3 exon, that is, only low levels of exon skipping were induced by the single AO, H67A(+22+47) . This AO was predicted to anneal to 3 SC35 motifs and, while substantial exon skipping was induced after transfection at 600 nM, weaker skipping at 300 nM and there was no detectable skipping at lower concentrations. However, upon combination of H67A(+22+47) with either of two other AOs directed at the acceptor, H67A(-10+17), or donor site H67D(+11-14), 50% exon 67 skipping was generated after transfection at 50 nM, with 42% and 23% exon 67 skipping induced after transfection at 10 and 2 nM, respectively (data not shown). The AOs directed at the exon 67 acceptor and donor sites were shown to be inactive when used individually at concentrations of 600 nM, and neither was directed to predicted SC35 motifs. Although the exon 67 cocktails do not conform to the observation that AOs targeting SC35 motifs are more effective in cocktails, this may reflect on the AO common to both cocktails, H67A(+22+47), which targeted three SC35 motifs and exhibited substantial exon skipping potential when applied at high concentrations.
AO cocktails that induced the most pronounced exon skipping did not necessarily block donor and acceptor sites, and other splicing motifs, nor did the most effective AO combination correlate with the total number of ESE sites targeted. For example, the most effective cocktail for exon 65, H65A(-11+14) and H65A(+26+50) masked three SC55 motifs, a single SF2/ASF and a SRp40 site. In contrast, a less effective AO combination directed at the same exon was directed at 4 SC35 sites, 2 SF2/ASF, 2 SRp40 and a single SRp55 motif. Exons 10 and 65 are removed by AOs directed at the acceptor site and internal ESE's, while exon 20 AOs anneal to internal ESE's. H10A(+98+119) and H65A(+63+87), annealed to all four predicted SR binding sites (SF2/ASF, SC35, SRp40 and SRp55), and these AOs were inactive when used individually.
AO chemistry comparisons in cocktails
The 2OMeAOs have some advantages over other AO chemistries, including PMOs, in that they can be readily synthesized in-house and may be efficiently transfected into cultured myogenic cells as cationic lipoplexes. PMOs are not readily taken up by cultured cells, unless high transfection concentrations are applied, scrape loading is employed [28, 29], or the PMOs are coupled to cell penetrating peptides to enhance delivery [30, 31, 32]. We have undertaken other comparisons between the 2OMeAOs and the PMOs directed at the mdx mouse nonsense mutation in exon 23 and observed that the PMOs offer much greater potential in vivo [16, 17, 33]. We now extend these studies to other targets, including dystrophin exons that had been difficult to displace and have found the same trends in exon skipping are observed with both chemistries.
Chemistry comparisons in AO design
All sequences developed as 2OMeAOs and shown to induce skipping of Type 1 exons  have now been prepared as PMOs (n = 41). These PMOs were shown to induce the same dystrophin exon removal patterns as those generated by the corresponding 2OMeAOs after transfection in either primary human myogenic cultures or muscle explants (data not shown). Detailed comparisons of sub-optimal PMOs to other dystrophin targets has not been undertaken, as the perfect concordance observed to date between the exon skipping trends with 2OMeAOs and PMOs would suggest that this would be unnecessary and a waste of resources. Our efforts are being directed to fine-tuning the next series of PMOs likely to enter clinical trials, as well as improving upon the efficiency of excision of Type 2 and 3 dystrophin exons, through improved AO design, and/or the use of AO cocktails. It will also be of interest to revisit some of the Type 1 dystrophin exons that are efficiently removed with a single AO, to ascertain if the application of AO cocktails can further enhance exon removal at very low AO concentrations. Although the use of AO cocktails will require the synthesis of two different compounds, these AOs would be used as a single preparation and safety evaluation would be undertaken on the combination. Should an AO cocktail be ten-fold more effective that an optimal single AO to induce targeted exon excision, there would be clear production and cost benefits. Perhaps more importantly, the use of AO cocktails may address safety and efficacy issues in that lower amounts of an AO preparation will need to be administered.
AO induced exclusion of dystrophin exons during pre-mRNA processing offers a potential treatment for removing or by-passing protein-truncating mutations that lead to DMD. For induced exon skipping to be a viable therapy, the most effective AO preparations must be developed so that minimal amounts can be administered. Many dystrophin exons could be efficiently removed from the mature mRNA by the intervention of a single AO during dystrophin pre-mRNA processing. Some exons require the action of two AOs, often ineffective when used individually, which somehow act in a synergistic fashion, presumably through prevention of spliceosome assembly by altering pre-mRNA folding or masking crucial protein binding sites. The SC35 motif appears to play some role as an amenable target for AO cocktails, but this association is not absolute. Clearly, there are many parameters involved in interfering with the pre-mRNA splicing process remaining to be elucidated. The trends in splice intervention may be seen with AOs composed of 2 different chemistries, 2OMeAO and PMO.
AO design and synthesis
2OMeAOs were prepared on an Expedite 8909 Nucleic acid synthesiser using the 1 μMol thioate synthesis protocol. AOs were designed to anneal to splicing motifs at the intron: exon boundaries, as well as ESE motifs predicted by the web based application, ESEfinder.
PMOs, and PMOs conjugated to the cell penetrating peptide, were synthesized by AVI Biopharma (Corvallis, Or). AO nomenclature is based upon that described by Mann et al, 2002 . The first letter designates the species, the number indicates the exon, the second letter specifies Acceptor or Donor splice sites, with the -/+ and numbers representing the annealing coordinates in the intronic and exonic domains respectively. For example, H65A(-11+14) would anneal across the acceptor site of human dystrophin exon 65, specifically to the last 11 bases of intron 64 and the first 14 nucleotides of exon 65.
Culture and transfection – Primary human myoblasts
The preparation of primary human myoblasts is described by Rando and colleagues, 1994 . Primary human myotubes were transfected in Opti-MEM (Invitrogen), 48 hrs after seeding, with Lipofectamine 2000 (L2K): AO at 1:1 w:w ratio according to manufacturer's instructions (Invitrogen). For each experiment, transfections were repeated three times to confirm reproducibility.
Culture and transfection – H-2Kb-tsA58 (H2K) mdx myoblasts
H2K-Mdx myoblasts  were cultured as described by Mann and colleagues 2001 . AOs were transfected with Lipofectin:AO at 2:1 w:w ratio, 24 hrs after seeding. Lipofectin was used according to manufacturer's instructions (Invitrogen, Melbourne). All transfections occurred in duplicate wells and were repeated three times to ensure consistency.
RNA extraction and RT-PCR have been described previously [8, 9]. Briefly, RNA was purified from duplicate cultures using an acid phenol extraction, before a one step RT-PCR was undertaken using specific primers, template and the Invitrogen Superscript III. After 30–35 cycles of amplification, an aliquot was removed and subjected to nested PCR using inner primer sets. Details of all primers used in these experiments are available upon request. The identity of the RT-PCR products was confirmed by direct DNA sequencing . Estimates of relative exon skipping efficiency were performed using the Vilber Lourmat Chemi-Smart 3000 system with Chemi-Capt software for image acquisition and Bio-1D software for image analysis.
The authors received funding from the National Institutes of Health (RO1NSO44146-02), the Muscular Dystrophy Association USA (MDA3718), the National Health and Medical Research Council of Australia (303216), Parent Project (UK) and the Medical and Health Research Infrastructure Fund (Western Australia).
- 4.Deas TS, Binduga-Gajewska I, Tilgner M, Ren P, Stein DA, Moulton HM, Iversen PL, Kauffman EB, Kramer LD, Shi PY: Inhibition of flavivirus infections by antisense oligomers specifically suppressing viral translation and RNA replication. J Virol 2005, 79: 4599-4609. 10.1128/JVI.79.8.4599-4609.2005PubMedCentralCrossRefPubMedGoogle Scholar
- 10.Bremmer-Bout M, Aartsma-Rus A, de Meijer EJ, Kaman WE, Janson AA, Vossen RH, van Ommen GJ, den Dunnen JT, van Deutekom JC: Targeted exon skipping in transgenic hDMD mice: A model for direct preclinical screening of human-specific antisense oligonucleotides. Mol Ther 2004, 10: 232-240. 10.1016/j.ymthe.2004.05.031CrossRefPubMedGoogle Scholar
- 14.Aartsma-Rus A, De Winter CL, Janson AA, Kaman WE, Van Ommen GJ, Den Dunnen JT, Van Deutekom JC: Functional analysis of 114 exon-internal AONs for targeted DMD exon skipping: indication for steric hindrance of SR protein binding sites. Oligonucleotides 2005, 15: 284-297. 10.1089/oli.2005.15.284CrossRefPubMedGoogle Scholar
- 33.Fletcher S, Honeyman K, Fall AM, Harding PL, Johnsen R, Steinhaus JP, Moulton HM, Iversen PL, Wilton SD: Morpholino oligomer mediated exon skipping averts the onset of dystrophic pathology in the mdx mouse. Molecular Therapy 2007. accepted, subject to minor revision.Google Scholar
- 38.Morgan JE, Beauchamp JR, Pagel CN, Peckham M, Ataliotis P, Jat PS, Noble MD, Farmer K, Partridge TA: Myogenic Cell Lines Derived from Transgenic Mice Carrying a Thermolabile T Antigen: A Model System for the Derivation of Tissue-Specific and Mutation-Specific Cell Lines. Developmental Biology 1994, 162: 486-498. 10.1006/dbio.1994.1103CrossRefPubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.