RNA Trans-Splicing Modulation via Antisense Molecule Interference

In recent years, RNA trans-splicing has emerged as a suitable RNA editing tool for the specific replacement of mutated gene regions at the pre-mRNA level. Although the technology has been successfully applied for the restoration of protein function in various genetic diseases, a higher trans-splicing efficiency is still desired to facilitate its clinical application. Here, we describe a modified, easily applicable, fluorescence-based screening system for the generation and analysis of antisense molecules specifically capable of improving the RNA reprogramming efficiency of a selected KRT14-specific RNA trans-splicing molecule. Using this screening procedure, we identified several antisense RNAs and short rationally designed oligonucleotides, which are able to increase the trans-splicing efficiency. Thus, we assume that besides the RNA trans-splicing molecule, short antisense molecules can act as splicing modulators, thereby increasing the trans-splicing efficiency to a level that may be sufficient to overcome the effects of certain genetic predispositions, particularly those associated with dominantly inherited diseases.


Introduction
RNA-based therapies constitute elegant options to restore gene function in various genetic diseases. Therapeutic intervention at the posttranscriptional level during the course of naturally occurring mRNA maturation in the cell represents a viable opportunity to manipulate the mRNA sequence prior to protein translation. Examples of such strategies include splice-modulating therapies, such as antisense oligonucleotide (ASO)-mediated exon skipping and spliceosome-mediated RNA trans-splicing (SMaRT), both of which are currently under development for the therapy of several monogenetic diseases [1][2][3].
Others [4,5] and we ourselves [6,7] have applied SMaRT as an emerging RNA editing tool for the specific correction of recessively and dominantly inherited mutations. So far, SMaRT has successfully been implemented in RNA repair studies for genetic disorders such as epidermolysis bullosa (EB) [8,9], muscular dystrophy [4,10], Alzheimer's disease [11], and Huntington's disease [12]. SMaRT uses the plasmid and rationally designed short ASOs, applied as oligonucleotides, that bind potential splicing elements within the KRT14 target region. Dominant mutations in KRT14 are responsible for the generalized severe form of epidermolysis bullosa simplex (EBS-gen sev). Incorporation of the mutant K14 into the intermediate filament (IF) network compromises its integrity, leading to its collapse into protein aggregates under conditions of stress [6,33,34]. Using our established fluorescence-based screening system [21,22] we investigated the impact of KRT14-specific antisense molecules on the trans-splicing efficiency of our KRT14-RTM and on the general cis-splicing pattern within the KRT14 target region. We achieved increased RNA repair levels after the addition of specific antisense molecules of varying lengths in our screening system. Thus, we were able to identify potential splicing modulators that may further improve the RNA trans-splicing technology to a level sufficient to support its implementation in a clinical setting.

Generation of Antisense Molecules for Trans-Splicing Enhancement
Aiming to increase the KRT14 repair efficiency via RNA trans-splicing in vitro, we adapted our well established fluorescence-based screening system [21] for the selection of randomly and rationally generated antisense molecules capable of directing the splicing reaction from cis to trans in the presence of an RTM ( Figure 1A). We compared the functionality of asRNAs, some covering the majority of the respective KRT14 targeting region, and rationally designed ASOs that bind specifically to selected exon-intron boundaries in close proximity to the RTM binding site located within intron 7 of KRT14 ( Figure 1A). Both antisense molecule variants are expected to increase the trans-splicing efficiency of the previously described and generated KRT14-specific RTM [6] via blocking or interfering with competitive, naturally occurring cis-splicing events within the KRT14 target region. The RTM was designed to replace the first seven exons of KRT14, covering all known EBS-associated mutations in this gene. The RTM was previously selected via our fluorescence-based screening system and confirmed to be functional in an EBS patient-derived keratinocyte line [6], although its efficiency warranted improvement to achieve a full phenotypic reversion in these cells.  [6,33,34]. Using our established fluorescence-based screening system [21,22] we investigated the impact of KRT14-specific antisense molecules on the trans-splicing efficiency of our KRT14-RTM and on the general cis-splicing pattern within the KRT14 target region. We achieved increased RNA repair levels after the addition of specific antisense molecules of varying lengths in our screening system. Thus, we were able to identify potential splicing modulators that may further improve the RNA trans-splicing technology to a level sufficient to support its implementation in a clinical setting.

Generation of Antisense Molecules for Trans-Splicing Enhancement
Aiming to increase the KRT14 repair efficiency via RNA trans-splicing in vitro, we adapted our well established fluorescence-based screening system [21] for the selection of randomly and rationally generated antisense molecules capable of directing the splicing reaction from cis to trans in the presence of an RTM ( Figure 1A). We compared the functionality of asRNAs, some covering the majority of the respective KRT14 targeting region, and rationally designed ASOs that bind specifically to selected exon-intron boundaries in close proximity to the RTM binding site located within intron 7 of KRT14 ( Figure 1A). Both antisense molecule variants are expected to increase the trans-splicing efficiency of the previously described and generated KRT14-specific RTM [6] via blocking or interfering with competitive, naturally occurring cis-splicing events within the KRT14 target region. The RTM was designed to replace the first seven exons of KRT14, covering all known EBS-associated mutations in this gene. The RTM was previously selected via our fluorescence-based screening system and confirmed to be functional in an EBS patient-derived keratinocyte line [6], although its efficiency warranted improvement to achieve a full phenotypic reversion in these cells. The RTM, the main component of the SMaRT system, carries the wild-type coding sequence of KRT14 spanning from exon 1 to exon 7, splicing elements for efficient splicing and a binding domain specific for intron 7 of KRT14. Antisense molecules were randomly (red) and rationally (green) generated in order to block competitive cissplicing elements within the KRT14 target region, thereby facilitating the trans-splicing-induced RNA repair of our pre-selected RTM; (B) for asRNA library generation the PCR-amplified KRT14 target region (1142 bp), spanning from exon 5 to intron 7, was fragmented by sonication and the resulting fragments were randomly cloned into a pcDNA 4.0 expression plasmid. Colony PCR and sequence analysis of individual bacterial clones revealed the presence of various target sequences in sense and antisense orientation. Vector: pcDNA 4.0 expression vector without KRT14 sequence; RTM, RNA trans-splicing molecule; 5′ SS, 5′ splice site.
To generate the asRNA library, the KRT14 target region encompassing 1144 nt spanning exon 5 to intron 7 immediately 5′ proximal to the BD target site, was PCR-amplified and fragmented via sonication. The resulting fragments were randomly cloned into the pcDNA 4.0 vector (Invitrogen, Figure 1. The combined application of antisense strategies. (A) The RTM, the main component of the SMaRT system, carries the wild-type coding sequence of KRT14 spanning from exon 1 to exon 7, splicing elements for efficient splicing and a binding domain specific for intron 7 of KRT14. Antisense molecules were randomly (red) and rationally (green) generated in order to block competitive cis-splicing elements within the KRT14 target region, thereby facilitating the trans-splicing-induced RNA repair of our pre-selected RTM; (B) for asRNA library generation the PCR-amplified KRT14 target region (1142 bp), spanning from exon 5 to intron 7, was fragmented by sonication and the resulting fragments were randomly cloned into a pcDNA 4.0 expression plasmid. Colony PCR and sequence analysis of individual bacterial clones revealed the presence of various target sequences in sense and antisense orientation. Vector: pcDNA 4.0 expression vector without KRT14 sequence; RTM, RNA trans-splicing molecule; 5 SS, 5 splice site.
To generate the asRNA library, the KRT14 target region encompassing 1144 nt spanning exon 5 to intron 7 immediately 5 proximal to the BD target site, was PCR-amplified and fragmented via sonication. The resulting fragments were randomly cloned into the pcDNA 4.0 vector (Invitrogen, Carlsbad, CA, USA) downstream to the human cytomegalovirus (CMV) promotor and maintained as individual plasmids expressing KRT14 sequences in either the sense or antisense orientation with respect to the target region ( Figure 1B).
We obtained a library of 136 asRNAs of varying lengths from 19 nt to 620 nt. Using this library, we initially performed triple transfection experiments, introducing an artificial KRT14 screening minigene (KRT14-scMG; 0.5 µg) and the RTM screening vector (KRT14-scRTM; 0.5 µg) together with the individual asRNA plasmids (3 µg). The engineered KRT14-scMG consists of the KRT14 target region from exon 5 to the end of intron 7 (KRT14 nt: 3320-4462, NCBI Gene ID: 3861) fused to the 3 half (nt: 337-720) of the GFP coding sequence. The screening RTM is comprised of the dsRed fluorescent reporter molecule sequence, a short linker sequence, the 5 (nt: 1-337) half of the GFP-coding sequence, a functional 5 splice site (5 SS) for efficient splicing, and the BD specific for intron 7 of KRT14 ( Figure 2) [6]. Successful and accurate trans-splicing between the pre-mRNA transcribed from the KRT14-scMG and the KRT14-scRTM resulted in the reconstitution of full-length GFP and translation of a functional protein in treated HEK293 cells detectable by flow cytometry and fluorescence microscopy ( Figure 2). The intensity of the GFP signal correlates with the efficiency of the trans-splicing reaction. As such, changes in the GFP signal intensity in the presence of the different asRNAs would reflect the positive or negative impact of asRNAs on the trans-splicing efficiency [21,22]. Carlsbad, CA, USA) downstream to the human cytomegalovirus (CMV) promotor and maintained as individual plasmids expressing KRT14 sequences in either the sense or antisense orientation with respect to the target region ( Figure 1B). We obtained a library of 136 asRNAs of varying lengths from 19 nt to 620 nt. Using this library, we initially performed triple transfection experiments, introducing an artificial KRT14 screening minigene (KRT14-scMG; 0.5 µg) and the RTM screening vector (KRT14-scRTM; 0.5 µg) together with the individual asRNA plasmids (3 µg). The engineered KRT14-scMG consists of the KRT14 target region from exon 5 to the end of intron 7 (KRT14 nt: 3320-4462, NCBI Gene ID: 3861) fused to the 3′ half (nt: 337-720) of the GFP coding sequence. The screening RTM is comprised of the dsRed fluorescent reporter molecule sequence, a short linker sequence, the 5′ (nt: 1-337) half of the GFPcoding sequence, a functional 5′ splice site (5′ SS) for efficient splicing, and the BD specific for intron 7 of KRT14 ( Figure 2) [6]. Successful and accurate trans-splicing between the pre-mRNA transcribed from the KRT14-scMG and the KRT14-scRTM resulted in the reconstitution of full-length GFP and translation of a functional protein in treated HEK293 cells detectable by flow cytometry and fluorescence microscopy ( Figure 2). The intensity of the GFP signal correlates with the efficiency of the trans-splicing reaction. As such, changes in the GFP signal intensity in the presence of the different asRNAs would reflect the positive or negative impact of asRNAs on the trans-splicing efficiency [21,22]. Screening constructs for antisense molecule selection. The reporter-based screening system consists of an artificial KRT14 screening minigene (KRT14-scMG), containing the KRT14 target region spanning from exon 5 to intron 7 and the 3′ part of GFP, and the RTM screening vector (KRT14-scRTM) carrying a dsRed reporter molecule and the missing 5′ GFP part, splicing elements for efficient splicing and a binding domain specific for intron 7 of KRT14. Accurate trans-splicing between the KRT14-scMG and the KRT14-scRTM (red) restores the GFP expression (green) in the cell manifested in a yellow signal. Cell nuclei were 4′,6-diamidin-2-phenylindol (DAPI) stained (blue). BD, binding domain; 5′ SS, 5′ splice site.

asRNA34 Facilitates RTM-Mediated RNA Editing
We performed further triple transfection experiments in the HEK293 system using asRNA34 in order to confirm the results from our library screen ( Figure 4A). In all described experiments, the total amount of DNA transfected was the same in each setting to exclude variabilities that could arise from transfection efficiency and cell toxicity when using different amounts of DNA. We used pcDNA 4.0 empty vector where needed to ensure equal levels of transfected plasmids. Combining the results of all experiments, we were able to achieve full-length GFP expression in~42% of HEK293 cells treated with the KRT14-scMG (0.5 µg), KRT14-scRTM (0.5 µg), and asRNA34 (3 µg). In contrast, in cells receiving pcDNA4.0 (3 µg) instead of asRNA34, only~9% of treated cells were GFP-positive by flow cytometric analysis ( Figure 4B). The geometric mean of GFP-positive cells was 3.7 in the absence of asRNA34, and rose to 13.2 in the presence of asRNA34, representing a 3.6-fold increase in GFP intensity.
Additionally, asRNA34 had a dose-dependent effect on GFP expression on the protein level, as assayed by flow cytometry and on RNA expression, evaluated by semi-quantitative real time PCR (sqRT-PCR) ( Figure 5A,B). Together, these results revealed an increase in trans-splicing efficiency of up to 60-fold in the presence of 3 µg asRNA34 instead of 1 µg ( Figure 5B). asRNA34 has a length of 324 nt and is predicted to be complementary to a region of KRT14 spanning the exon 5-intron 5 and intron 5-exon 6 junctions ( Figure 4A). As such, asRNA34 is expected to impact cis-splicing of intron 5 and not intron 7. To understand how interfering with cis-splicing of intron 5 could positively impact trans-splicing, we first investigated the impact of increasing asRNA34 concentrations on cis-splicing levels at all exon-exon junctions within the KRT14-scMG pre-mRNA in the experiments described above. Towards this end, primers spanning the exon-exon junctions were designed to assess the number of mRNA molecules harbouring correctly spliced exons by sqRT-PCR.
The analysis of competitive cis-splicing events within the KRT14-scMG upon co-delivery of different levels of asRNA34-expressing plasmids indicated a potential stabilizing effect of the asRNA on the target pre-mRNA as increased cis-splicing levels were detectable via sqRT-PCR analysis ( Figure 5C). This finding was contrary to our initial assumption that blocking potential splicing elements within the KRT14-scMG resulted in reduced cis-splicing rates. In contrast, the amount of transfected RTM primarily impacted only the trans-splicing level, and not the cis-splicing level as shown via flow cytometric analysis ( Figure 5D) and sqRT-PCR ( Figure 5E,F

Splice Site-Specific Antisense Oligonucleotides Increase RNA Trans-Splicing Levels
Although asRNA34 could confer a significant improvement on the KRT14-scRTM's RNA transsplicing efficiency, we also observed increased cis-splicing of intron 7 with increasing concentrations of the asRNA, which would be in direct competition with our desired trans-splicing reaction. To avoid the unwanted effect on cis-splicing, we generated shorter site-specific ASOs (Table 4) designed to directly block competitive 5′ splice sites (ASO1, 2, 3, 6, 7, 8), potential splice site enhancer sequences (ASO4, 9) ( Figure S1), or both, within the KRT14-scMG sequence. In total, nine ASOs 20-24 nts in length were rationally designed to cover exon 6 to intron 7 including the exon-intron junctions upstream of the RTM binding site within the KRT14 target region ( Figure 6A). Coady et al. (2008) recently showed that the inhibition of cis-splicing of adjacent introns can have a positive impact on the trans-splicing efficiency of a given RTM [32]. Therefore, in addition to ASOs specific for the exonintron 7 junction, we also tested ASOs 1, 2, and 3, which target the exon-intron 6 junction further upstream to the RTM binding site. A scrambled ASO (scrASO), which was designed not to target any region within the KRT14-scMG, was included as a control in initial triple transfection experiments. Flow cytometric analysis of transfected HEK293 cells revealed a significant increase of GFP- and ASO9, binding the exon 7-intron 7 junction and the 5′ end of exon 7, respectively, represented the most promising ASOs, resulting in 33% and 30% GFP-positive cells in the screen. We confirmed the results with these two ASOs in additional experiments, obtaining a similar increase in GFPpositive cell numbers from 15% (scrASO) to 30% (ASO7) and 29% (ASO9) ( Figure 6C). Furthermore, Western blot analysis of whole cell lysates from triple-transfected HEK293 cells validated the elevation of trans-spliced chimeric products as indicated by an increase in GFP expression ( Figure 6D). . The asRNA34-mediated trans-splicing increment was accompanied by increased cis-splicing levels, detected via sqRT-PCR and normalized to GAPDH mRNA levels (C) (the mean value + SD of one representative experiment is shown), indicating a stabilizing effect on the target pre-mRNA by antisense molecule interference. Empty pcDNA 4.0 expression plasmids were transfected into HEK293 cells to adjust total plasmid levels delivered into the target cells. Increasing the amount of transfected RTMs results in an increased GFP expression as detected via flow cytometry (D) and sqRT-PCR (E), although the cis-splicing rate within the KRT14-scMG remains largely unaffected (F). An unpaired Student's t-test (two-tailed) was performed with the GraphPad Prism software (GraphPad Software, San Diego, CA, USA) to prove statistical significance (Tables 2 and 3). Significance values are given in asterisks (ns = not significant; * p-value ≤ 0.05; ** p-value ≤ 0.01). The asRNA34-mediated trans-splicing increment was accompanied by increased cis-splicing levels, detected via sqRT-PCR and normalized to GAPDH mRNA levels (C) (the mean value + SD of one representative experiment is shown), indicating a stabilizing effect on the target pre-mRNA by antisense molecule interference. Empty pcDNA 4.0 expression plasmids were transfected into HEK293 cells to adjust total plasmid levels delivered into the target cells. Increasing the amount of transfected RTMs results in an increased GFP expression as detected via flow cytometry (D) and sqRT-PCR (E), although the cis-splicing rate within the KRT14-scMG remains largely unaffected (F). An unpaired Student's t-test (two-tailed) was performed with the GraphPad Prism software (GraphPad Software, San Diego, CA, USA) to prove statistical significance (Tables 2 and 3). Significance values are given in asterisks (ns = not significant; * p-value ≤ 0.05; ** p-value ≤ 0.01).

Splice Site-Specific Antisense Oligonucleotides Increase RNA Trans-Splicing Levels
Although asRNA34 could confer a significant improvement on the KRT14-scRTM's RNA trans-splicing efficiency, we also observed increased cis-splicing of intron 7 with increasing concentrations of the asRNA, which would be in direct competition with our desired trans-splicing reaction. To avoid the unwanted effect on cis-splicing, we generated shorter site-specific ASOs (Table 4) designed to directly block competitive 5 splice sites (ASO1, 2, 3, 6, 7, 8), potential splice site enhancer sequences (ASO4, 9) ( Figure S1), or both, within the KRT14-scMG sequence. In total, nine ASOs 20-24 nts in length were rationally designed to cover exon 6 to intron 7 including the exon-intron junctions upstream of the RTM binding site within the KRT14 target region ( Figure 6A). Coady et al. (2008) recently showed that the inhibition of cis-splicing of adjacent introns can have a positive impact on the trans-splicing efficiency of a given RTM [32]. Therefore, in addition to ASOs specific for the exon-intron 7 junction, we also tested ASOs 1, 2, and 3, which target the exon-intron 6 junction further upstream to the RTM binding site. A scrambled ASO (scrASO), which was designed not to target any region within the KRT14-scMG, was included as a control in initial triple transfection experiments. Flow cytometric analysis of transfected HEK293 cells revealed a significant increase of GFP-expressing cells from 11% (scrASO) up to 33% depending on the ASO introduced ( Figure 6B). ASO7 and ASO9, binding the exon 7-intron 7 junction and the 5 end of exon 7, respectively, represented the most promising ASOs, resulting in 33% and 30% GFP-positive cells in the screen. We confirmed the results with these two ASOs in additional experiments, obtaining a similar increase in GFP-positive cell numbers from 15% (scrASO) to 30% (ASO7) and 29% (ASO9) ( Figure 6C). Furthermore, Western blot analysis of whole cell lysates from triple-transfected HEK293 cells validated the elevation of trans-spliced chimeric products as indicated by an increase in GFP expression ( Figure 6D).  We additionally verified this increase in trans-splicing efficiency at the mRNA level via sqRT-PCR analysis ( Figure S2), revealing an up to 2-4-fold increase in the levels of reconstituted GFP mRNA in cells harboring the selected ASOs as compared to scrASO-treated controls ( Figure 7A). In contrast to the results obtained with asRNA34, the shorter ASO7 and ASO9 had no significant effect on the general cis-splicing events occurring within the KRT14-scMG. Rather the levels of cis-splicing within the target pre-mRNA were slightly reduced (~2-fold) in the presence of ASO7, whereas ASO9 exhibited no significant impact on these events ( Figure 7A). Increasing the concentrations of ASO7 added to the trans-splicing reaction, revealing the highest trans-splicing efficiency at a concentration of 60 nmol ( Figure 7B). We additionally verified this increase in trans-splicing efficiency at the mRNA level via sqRT-PCR analysis ( Figure S2), revealing an up to 2-4-fold increase in the levels of reconstituted GFP mRNA in cells harboring the selected ASOs as compared to scrASO-treated controls ( Figure 7A). In contrast to the results obtained with asRNA34, the shorter ASO7 and ASO9 had no significant effect on the general cis-splicing events occurring within the KRT14-scMG. Rather the levels of cis-splicing within the target pre-mRNA were slightly reduced (~2-fold) in the presence of ASO7, whereas ASO9 exhibited no significant impact on these events ( Figure 7A). Increasing the concentrations of ASO7 added to the trans-splicing reaction, revealing the highest trans-splicing efficiency at a concentration of 60 nmol ( Figure 7B).  Table 5). The cis-splicing activity within the Ex5/6-Ex6/7 KRT14-scMG region was significantly decreased for ASO7 and ASO9. Within the Ex6/7-GFP region the cis-splicing activity was decreased for ASO7 but unaffected for ASO9; (B) the co-delivery of the KRT14-scMG, the KRT14-scRTM, and different ASO7 concentrations into HEK232 cells resulted in the most promising trans-splicing efficiency at an ASO concentration of 60 nmol. HEK293 cells treated with the screening constructs and respective scrambled ASO concentrations were set to 1. Significance values are given in asterisks (** p-value ≤ 0.01; *** p-value ≤ 0.001).

Discussion
In recent years, RNA trans-splicing has been predominantly applied in RNA repair studies for various genetic diseases or for the RTM-mediated delivery of "suicide genes" into tumour cells [35,36]. Reprogramming at the pre-mRNA level represents an elegant approach for the editing of aberrant RNA molecules that maintains endogenous regulation of the target transcript, making the technology potentially safe and well suited for the therapy of certain genetic conditions. These include disorders caused by mutations affecting large genes or dominantly inherited mutations, difficult to correct via conventional full-length cDNA replacement strategies. In this respect, we have already demonstrated promising results with a COL7A1-targeting RTM capable of correcting ~40% of all RDEB-associated mutations encoded within the ~3300-nucleotide long target pre-mRNA [7]. In the case of dominant negative mutations in EBS-gen sev, Cao et al. showed with an inducible EBS mtK14 mouse model, that dominance of the mutation is dependent on the ratio of wild-type to mutated alleles. A reduction in expression of the mutated allele by ~50% appeared sufficient to 4-fold increase in GFP expression in ASO7 and a~2-fold increase in ASO9-treated cells in comparison to cells treated with the scrambled ASO. An unpaired Student's t-test (two-tailed) was performed with GraphPad Prism software (GraphPad Software, San Diego, CA, USA) to prove the statistical significance between scrASO and ASO7, as well as scrASO and ASO9 ( Table 5). The cis-splicing activity within the Ex5/6-Ex6/7 KRT14-scMG region was significantly decreased for ASO7 and ASO9. Within the Ex6/7-GFP region the cis-splicing activity was decreased for ASO7 but unaffected for ASO9; (B) the co-delivery of the KRT14-scMG, the KRT14-scRTM, and different ASO7 concentrations into HEK232 cells resulted in the most promising trans-splicing efficiency at an ASO concentration of 60 nmol. HEK293 cells treated with the screening constructs and respective scrambled ASO concentrations were set to 1. Significance values are given in asterisks (** p-value ≤ 0.01; *** p-value ≤ 0.001). Table 5. Statistical significance cis-splicing Figure 7. Significance values are given in asterisks (ns = not significant; ** p-value ≤ 0.01; *** p-value ≤ 0.001).

Discussion
In recent years, RNA trans-splicing has been predominantly applied in RNA repair studies for various genetic diseases or for the RTM-mediated delivery of "suicide genes" into tumour cells [35,36]. Reprogramming at the pre-mRNA level represents an elegant approach for the editing of aberrant RNA molecules that maintains endogenous regulation of the target transcript, making the technology potentially safe and well suited for the therapy of certain genetic conditions. These include disorders caused by mutations affecting large genes or dominantly inherited mutations, difficult to correct via conventional full-length cDNA replacement strategies. In this respect, we have already demonstrated promising results with a COL7A1-targeting RTM capable of correcting~40% of all RDEB-associated mutations encoded within the~3300-nucleotide long target pre-mRNA [7]. In the case of dominant negative mutations in EBS-gen sev, Cao et al. showed with an inducible EBS mtK14 mouse model, that dominance of the mutation is dependent on the ratio of wild-type to mutated alleles. A reduction in expression of the mutated allele by~50% appeared sufficient to overcome the EBS phenotype [37]. In this respect, RNA trans-splicing-mediated correction of the mutant RNA is expected to not only increase the amount of wild-type KRT14 transcripts, but also simultaneously decrease the level of mutant transcripts by competing with the normal cis-splicing events. Indeed, we could achieve partial reversion of the phenotype of EBS keratinocytes associated with a dominant-negative mutation in KRT14 with an appropriate KRT14-targeting RTM [6].
With the recent advances in the identification and selection of the BD for the RTM [18,21,22], the main obstacle that remains is the low efficiency of the technique, which limits its potential for future in vivo or clinical application. Therefore, we have developed a fluorescence-based screening system aimed at identifying factors with the potential to increase the trans-splicing efficiency, such as antisense RNA molecules [28]. In the course of this study, we could identify antisense RNA molecules capable of enhancing the trans-splicing efficiency induced by the KRT14-targeting RTM by up to 20-fold. The system proved practical for the screening of an unbiased library of randomly generated sequences of varying lengths and binding positions along the target pre-mRNA, as well as short, site-specific ASOs rationally designed to block splice sites and splicing enhancer elements. Our results revealed both antisense strategies to be attractive for future RNA repair studies in patient cells, but also highlight several points for consideration.
In the described screening system, the KRT14-scRTM induced a trans-splicing reaction between the 5 GFP coding sequence on the RTM and the 3 GFP sequence located immediately after intron 7 on the target pre-mRNA molecule. Notably, some of the best-acting antisense molecules identified (asRNA34 and ASO9) were not expected to directly inhibit the competing cis-splicing of intron 7. ASO9 was expected to bind to the 5 region of exon 7, whereas asRNA34 bound a 324-nt region spanning all of exon 5, intron 5, and half of exon 6. As such, asRNA34 was rather expected to interfere with cis-splicing of intron 5, whereas ASO9 may or may not interfere with cis-splicing of intron 6. The mechanisms by which these molecules enhanced trans-splicing efficiency are not yet completely understood. However, blocking potential splicing enhancer and silencer sequences located within the KRT14-scMG is likely to impact both types of splicing reactions (cis and trans). asRNA34, in particular, would block many predicted enhancer and silencer sequences, and thereby affect the general spliceosome-mediated splicing efficiency ( Figure S3) in an unpredictable manner. In this respect, increasing amounts of asRNA34 added to the cells led to a significant enhancement of both cisand trans-splicing reactions. Besides blocking potential splicing motifs within the MG sequence, asRNA34 may also exert a general stabilizing effect on the target pre-mRNA that promotes both cisand trans-splicing reactions. Thus, while we achieved increased trans-splicing rates by the addition of asRNA34, the concurrent upregulation of cis-splicing in the target pre-mRNA would be expected to negate any of its associated benefits, as no alteration in the ratios of wild-type (corrected) transcripts to mutant transcripts can be expected.
In contrast to long asRNA sequences, the mechanisms of action by which short ASOs exert their effects are more clearly defined, as they are designed to inhibit splicing either by directly blocking the splice donor/acceptor sites located within the exon-intron junctions (as in the case of ASO7), or by blocking potential splicing enhancer motifs adjacent to the intron-exon junction (ASO9). Both ASO7 and ASO9 resulted in enhanced trans-splicing of the RTM without any detectable increase in cis-splicing events within the target-pre-mRNA, suggesting that these shorter ASOs are better suited for our purposes. The most auspicious ASO7, targeting the exon 7-intron 7 boundary, additionally exhibited an inhibitory effect on the competitive cis-splicing of intron 7 as expected. The observed 2-fold inhibition in cis-splicing of intron 7, combined with the 4-fold increase in trans-splicing will not only reduce the level of mutated transcripts in the patient cells but also shift the ratio of wild-type to mutant transcripts further towards wild-type. This is expected to decrease the dominant negative effect of the disease-causing mutation to a level that may be sufficient to fully restore the wild-type phenotype in patient cells.
This study describes a straightforward and practical protocol that can be adapted to the screening of any type of splicing modulator, including pharmacological compounds. Our results also demonstrate that there are different mechanisms by which antisense technologies can impact splicing events and further underscore the need to comprehensively evaluate the effects of such molecules on different processes and outcomes including, for example, RNA stability. To this end, future studies will not only focus on evaluation of this therapeutic strategy in EBS patient-derived keratinocytes, but also on the investigation of mechanisms of action and precise quantification of corrected and mutant transcripts using next-generation sequencing platforms.

Construction of asRNA Library and Antisense Oligonucleotides
An asRNA library was constructed by the fragmentation of the PCR-amplified KRT14 target region spanning from exon 5 to intron 7 by sonication according to an already established protocol [21,28]. PCR amplification was performed using genomic DNA from a healthy donor and the GoTaq DNA polymerase (Promega, Madison, MI, USA). The resulting fragments were treated with a DNA Terminator End Repair Kit (Lucigen, Middleton, WI, USA), cloned into a pcDNA 4.0 plasmid (Invitrogen, Carlsbad, CA, USA) using the restriction site for EcoRV, and analyzed for their correct orientation (complementary to the target region) by sequence analysis. PCR products were digested with the corresponding restriction enzymes for 1 h at 37 • C, purified with a Illustra GFX PCR DNA and Gel Band Purification Kit (GE Healthcare, Chalfont St. Giles, UK) and ligated with T4 DNA ligase (Thermo Fisher Scientific, Waltham, MA, USA), all according to the manufacturer's protocols. The ligation was then transformed into chemically competent SoloPack ® Gold Competent Cells (Agilent Technologies, Santa Clara, CA, USA) and plasmid preparations were carried out using a Plasmid Mini Prep Kit (Sigma-Aldrich, Taufkirchen, Germany), according to the manufacturer's protocol. Sequence analysis of all plasmids and PCR products was performed using a 3500 ABI automated sequence analyzer and ABI PRISM dye terminator cycle sequencing kit (Applied Biosystems, Foster City, CA, USA).

Screening Constructs
Screening constructs used for detection of cis-splicing and correct trans-splicing were designed and cloned according to the protocol of Bauer et al. [21]. Concisely said, the KRT14-scRTM carries a 5 split-GFP part consisting of the first nucleotides (nt: 1-336) from a full length acGFP (pacGFP vector, Clontech, Mountain View, CA, USA) and a BD sequence that specifically targets Intron 7 of KRT14 cloned into the pcDNA3.1D/V5-HIS (Invitrogen, Carlsbad, CA, USA) backbone. The KRT14-scMG spanning exon 5 to intron 7 was generated using a KRT14 exon 5 forward primer (5 -GATCAAGCTTCACCACAGAGGAGCTGAACCGCGAGGTGGC-3 ) and a KRT14 intron 7 reverse primer (5 -GATCGGATCCGGGGAAGAGGTGGGAAGAGGACGTTACC-3 ) for amplification via GoTaq DNA polymerase (Promega, Madison, MI, USA). Afterwards, the amplified PCR product was cloned into the screening vector backbone downstream to the CMV-promoter (pcDNA3.1, Invitrogen, Carlsbad, CA, USA) using HindIII and BamHI restriction sites [21,28]. The KRT14-scMG also carries the 3 split-GFP portion consisting of the rearward part of acGFP (nt: 337-720) to enable a reconstitution of the full length acGFP upon accurate trans-splicing.

Flow Cytometric Analysis
The amount of trans-splicing events, manifested in the expression of GFP, was measured 48-72 h after transfection using a Beckman Coulter FC-500 FACS analyzer (Beckman Coulter, Vienna, Austria). The transfected HEK293 cells were washed once with phosphate buffer saline (Dulbecco's PBS) and detached from the cell culture plates using Trypsin-EDTA. Trypsin-EDTA was inactivated with Solution A (7.25 g HEPES; 1.8 g glucose; 0.22 g NaCl; 0.27 g NaHPO 4 ; 1 mL phenol red; 900 mL H 2 O; pH 7.4) supplemented with 10% FBS. After centrifugation for 5 min at 350× g, 10,000 cells were analyzed according to their GFP expression and geometric mean with the Kaluza 1.3 software (Beckman Coulter, Brea, CA, USA).

RNA Isolation and cDNA Synthesis
HEK293 cells were harvested 48-72 h post transfection and RNA was isolated using the Innuprep RNA Mini Kit (Analytik Jena, Jena, Germany) according to the manufacturer's protocol. Afterwards cDNA synthesis was performed with 3 µg RNA using the iScript™ cDNA Synthesis Kit (BioRad, Hercules, CA, USA) according to the manufacturer's protocol.

sqRT-PCR Analysis of Cis-and Trans-Splicing Levels
sqRT-PCR was performed to detect fused GFP transcripts in treated HEK293 cells. For sqRT-PCR analysis, a 5 GFP-specific forward primer (5 -GCTGACCCTGAAGTTCATCTG-3 ), a 3 GFP-specific reverse primer (5 -CGCCGATGGGGGTATTCTGCTGG-3 ), cDNA of treated cells and GoTaq ® qPCR Master Mix (Promega, Madison, WI, USA) was used. The PCR was performed using a Bio-Rad CFX™ system (BioRad, Hercules, CA, USA) with the following conditions: 95 • C for 2 min, and 50 cycles of 20 s at 95 • C, 20 s at 62 • C, and 25 s at 72 • C. Experiments were carried out in duplicates and repeated two times. Correct PCR products were verified by direct sequence analysis.

Western Blot Analysis
Enhancement of asRNA/ASO-induced trans-splicing manifests in an increase in full-length GFP expression, which was quantified at the protein level via Western blot analysis 48-96 h post transfection. HEK293 cells were washed with PBS, harvested, and resuspended in radioimmunoprecipitation assay buffer RIPA (Santa Cruz Biotechnology, Heidelberg, Germany). Samples were denatured for 5 min at 95 • C in 4× SDS loading buffer (0.25 M Tris-HCl; 8% SDS; 30% glycerol; 0.02% bromphenol blue; 0.3 M β-mercaptoethanol; pH 6.8). SDS-gel-electrophoresis was performed at 140 V for a maximum of 2 h using a NuPAGE 12% Bis-Tris gel (Invitrogen, Carlsbad, CA, USA) and NuPAGE MOPS running buffer. Proteins were electro-blotted onto a nitrocellulose membrane (Amersham Hybond-ECL, Amersham, Buckinghamshire, UK) at 0.25 A for 75 min. The membrane was blocked with Western Blocking Reagent (Roche, Basel, Switzerland) in Tris buffered saline (TBS) + 0.2% Tween (TBS-T) for 1 h at RT and incubated with a primary anti-GFP rabbit IgG antibody (MBL-598, MBL International, Woburn, MA, USA) in a dilution of 1:750 in TBS-T at 6 • C overnight. After three washing steps (3 × 10 min) with TBS-T the membrane was incubated with a secondary HRP Envision+ labelled anti rabbit antibody (1:100 in TBS-T, Agilent Technologies, Santa Clara, CA, USA) and incubated for 90 min at room temperature. Finally, GFP expression was visualized using the Luminata™ Forte Western HRP Substrate (Millipore, Burlington, MA, USA). α-actinin (H-2) mouse monoclonal IgG antibody (1:1000 in TBS-T; Santa Cruz. Dallas, TX, USA) and a secondary HRP Envision+ labelled anti mouse antibody (1:100 in TBS-T Tween, Agilent Technologies, Santa Clara, CA, USA) were used to detect α-actinin which served as protein loading control. The relative quantification of GFP expression was measured and calculated using the Image Lab 5.2.1 (Bio-Rad) software.

Statistical Analysis of SqRT-PCR Data
For statistical analysis, an unpaired Student's t-test (two-tailed) was performed with the GraphPad Prism 5.03 software (GraphPad Software, San Diego, CA, USA) to prove statistical significance between the values of controls (pcDNA 4.0 & scrASO) and samples (asRNA34, ASO7, ASO9). A sample size of at least n = 4 was analyzed for mean ± SEM to provide a p-value.

Analysis of Splicing Enhancer and Silencer
DNA sequence motifs, that probably regulate/influence the general splicing mechanism within the KRT14 targeting region, were predicted using the prediction websites http://regrna.mbc.nctu.edu. tw/html/prediction.html and http://www.umd.be/HSF3/HSF.shtml [38]. Conflicts of Interest: Johann W. Bauer is an inventor on US (US8735366) and European (EP2320952) patent "Improved pre-mRNA trans-splicing molecules (RTM) and their uses".