Huntington's Disease (HD) is an autosomal dominant neurodegenerative disorder caused by an expanded polyglutamine repeat (polyQ) in one allele of the Huntingtin (HTT) gene (The Huntington's Disease Collaborative Research Group, 1993). More than 35 polyQs causes disease, and the number of repeats is inversely correlated to the onset age and severity (Andrew et al., 1993; Duyao et al., 1993; The Huntington's Disease Collaborative Research Group, 1993). The disease affects ~6 per 100,000 in Caucasian populations (Pringsheim et al., 2012), with devastating consequences including progressive motor dysfunction, chorea, cognitive impairment, psychiatric abnormalities (Vonsattel et al., 2008), sleep disturbances (Arnulf et al., 2008), eventually leading to death (DiFiglia et al., 1995). These symptoms are due to the dysfunction and eventual death of specific projection neurons within the brain (Ross and Tabrizi, 2011). In particular, medium spiny neurons of the striatum, followed by the pyramidal neurons of the cortex, are most affected by disease. The loss of these neurons can occur years before overt clinical symptoms (Rosas et al., 2008; Aylward et al., 2011).

The HTT gene encodes a protein that is evolutionarily conserved, with homologs found as distant as the multicellular amoeba Dictyostelium discoideum (Eichinger et al., 2005). This degree of conservation implicates HTT as a necessary gene for the survival of multicellular organisms. Consistent with this notion, a Htt-null in murine models results in early embryonic lethality (e7.5) (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995). HTT is ubiquitously expressed and has roles in a multitude of cellular functions including protein transport (DiFiglia et al., 1995; Gutekunst et al., 1995; Block-Galarza et al., 1997), protein-protein interactions (Takano and Gusella, 2002), transcriptional regulation (Kegel et al., 2002; Sugars and Rubinsztein, 2003), inhibition of apoptosis (Rigamonti et al., 2000), and embryonic development (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995). While the expanded allele is associated with disease development, from an evolutionary perspective, the increasing CAG trinucleotide repeat length in HTT homologs correlates with an increase in brain evolution (Zuccato and Cattaneo, 2016). Potentially, the expanded repeat assists in the development of more complex nervous systems and that further pathogenic expansion of HD is an unintended side effect of this evolutionary process (Zuccato and Cattaneo, 2016). Like many other neurodegenerative diseases, it is currently unknown why this ubiquitously expressed protein specifically causes neuronal degeneration. There are a multitude of theories of the contribution of mutant HTT (mtHTT) protein to cell death including: specific mtHTT cleavages (Lunkes et al., 2002; Graham et al., 2006; Ratovitski et al., 2009) and modifications (Steffan et al., 2004; Gu et al., 2009; O'Rourke et al., 2013; Vicente Miranda et al., 2016), buildup of toxic aggregates (DiFiglia et al., 1997; Scherzinger et al., 1997) or soluble mtHTT oligomers (Takahashi et al., 2008; Leitman et al., 2013; Ramdzan et al., 2017), and haploinsufficiency of wild-type HTT (Humbert, 2010).

RNA splicing is the editing of nascent pre-mRNA, thereby producing mRNA by removing the intronic regions. Part of this highly dynamic and regulated process involves recognition of intronic regions by the spliceosomal complex, including: a “GU” donor site at the 5′ end of the intron, an “A” branchpoint near the 3′ end of the intron, and a poly-pyrimidine tract with an “AG” splice acceptor site at the 3′ end of the intron (Supplementary Figure 1). cis-Splicing occurs when introns are spliced out of a pre-mRNA and the flanking exons of the gene are ligated together. However, splicing can also occur between exons of two different RNA molecules, termed trans-splicing. trans-Splicing generates a chimeric mRNA and has been demonstrated to occur naturally in trypanosomes (Murphy et al., 1986; Sutton and Boothroyd, 1986), rat hepatocytes (Caudevilla et al., 1998), and even human cells (Flouriot et al., 2002; Wu et al., 2014). A technology predicated upon these naturally occurring trans-splicing events was developed in 1999, where an endogenous transcript was modified via spliceosome-mediated pre-mRNA trans-splicing to include an exogenous exon (Puttaraju et al., 1999). Previously, we have presented proof-of-principle experiments demonstrating that exon one replacement of HTT by trans-splicing can be achieved in cultured cells, both into a HTT mini-gene and the endogenous HTT pre-mRNA, suggesting a novel mechanism for mtHTT repair (Rindt et al., 2012).

These exogenous exons were termed pre-mRNA trans-splicing molecules (PTMs) and have three main modalities: a region that specifically binds to the target endogenous RNA, an artificial intron, and the exogenous “replacement” exon. Each sequence can be modified accordingly to improve efficacy and specificity to the trans-splicing event into the endogenous RNA (Garcia-Blanco, 2003). To further refine and increase the efficacy of HTT trans-splicing, a novel series of PTMs were designed to specifically bind to either the 5′ or 3′ ends of HTT intron 1. The coding sequence of each PTM-replaced exon 1 with a non-disease exon 1 sequence including 21 CAG trinucleotide repeats. PTMs that targeted the 3′ end of HTT intron 1 or the intron 1 branch point resulted in enhanced levels of trans-spliced RNA compared to the original HD trans-splicing RNA, demonstrating a correction of the HTT allele at the molecular level. To determine whether delivery of the optimized PTMs increased functional HTT protein, PTMs were expressed in a neural culture derived from HD patient induced pluripotent stem cells (iPSCs). In each instance PTM expression correlated with improvements in two previously established HTT CAG-expanded dependent phenotypes: susceptibility to BDNF withdrawal and ATP deficiency. Together, these results demonstrate that trans-splicing can be optimized to increase functional HTT protein, suggesting that trans-splicing could potentially serve as a therapy for HD.

In the development of the original PTM that targeted replacement of HTT exon 1, few modifications were made to optimize the trans-splicing efficiency (Rindt et al., 2012). Therefore, the primary objective in this report was to enhance trans-splicing by targeting different regions within the HTT RNA. In this new series of PTMs, a non-disease encoding HTT exon 1 with 21 CAG repeats was fused to a synthetic stretch of RNA that functioned as the targeting sequence or the “tether” (Figure 1A). Additionally, to enhance splicing to the endogenous HTT pre-mRNA, three tandem repeats of a previously identified ISE element were incorporated into the PTM, as well as the HTT intron 1-derived branch point. PTMs were targeted to the 5′ or 3′ end of intron 1, with T9 overlapping the branch point upstream of exon 2 (Figure 1B). To initially examine the efficiency of trans-splicing, a previously developed mini-gene system was utilized that comprises a sub-genomic fragment of the HTT gene spanning exons 1, 2, and 3, separated by shortened introns 1 and 2, respectively. Fourty-two CAG tandem repeats were engineered into the exon 1 mini-gene sequence. Following co-transfection of the mini-gene with the PTMs into HEK293 cells, RT-PCR was used to specifically detect the trans-splicing product (Figure 1C). As expected, the original T7 PTM, which tethered to the 5′ end of intron 1, resulted in low but detectible levels of trans-splicing. Similarly, low levels of trans-splicing were also observed using the similarly targeted T8 PTM. In contrast, T4 and T9, the PTMs that were proximally positioned to exon 2 and disrupted the cis branch point within the mini-gene pre-mRNA, resulted in greater levels of trans-splicing (Figure 1C). Site-specific mutagenesis of the PTM branch point reduced trans-splicing efficiency in each instance, highlighting the importance of intact and functional splicing signals within the PTMs.

Having identified lead trans-splicing candidates using the mini-gene system, the branchpoint-containing PTMs were examined in a more challenging context: trans-splicing to endogenous HTT. HEK293 cells were transfected with PTM-expressing vectors and RNA was subsequently collected and analyzed via RT-PCR. The T4 and T9 PTMs were the most consistently active in this assay, resulting in high levels (10–15%) of trans-splicing product in multiple independent experiments (Figure 1D). Collectively, these results identify the T4 and T9 PTMs as trans-splicing RNAs that lead to enhanced replacement of the expanded HTT allele in a mini-gene system and with endogenous HTT in HEK293 cells. These therefore will serve as lead candidates in the following experiments that examine trans-splicing in disease-relevant contexts.

This study provides further validation for the use of trans-splicing as a powerful potential therapy tool for genetic disorders such as HD. In this report, the objective was to determine whether targeting the PTM tether to different regions within the HTT pre-mRNA would lead to increased trans-splicing activity. In a variety of disease-relevant experimental contexts, targeting the cis branch point within the HTT pre-mRNA improved efficiency of the PTM in a variety of disease contexts.

As HD is a monogenic disorder, gene-based correction of mtHTT is a compelling prospective therapeutic tool. There are several paradigms for “gene therapy,” each with important benefits as well as potential drawbacks (for a complete review please see Keiser et al., 2016; Yang et al., 2016). RNA interference (RNAi) and antisense oligonucleotides (ASOs) are single-stranded nucleotide sequences that work by binding specifically to the target mRNA, thereby knocking down gene expression. ASOs in particular do hold great promise as a treatment for HD. They have been demonstrated to efficiently knock-down mtHTT in an allele-specific manner (Østergaard et al., 2013; Skotte et al., 2014; Southwell et al., 2014) and are currently in HD patient clinical trials (Rollnik, 2017). One drawback however, includes the inability of ASOs to cross the blood-brain-barrier, therefore requiring direct administration to the brain, and the need for repeatedly administration over the course of a lifetime, as they are not made in the cells. Short-hairpin RNAs and microRNAs act in a similar fashion to RNAi and ASOs to knock-down gene expression, but can be virally delivered. This allows for continued gene expression after a single dose, however dosage control is more challenging. Another method to reduce/eliminate mtHTT expression at the protein level is the use of virally-delivered intracellular antibodies (intrabodies). However, proper intrabody folding and low solubility in the reducing cytoplasmic environment are issues of functionality and stability for this proposed therapy (Biocca et al., 1995). New technologies have emerged in the past decade to alter gene expression or even permanently edit the mutant genes themselves, including zinc finger nucleases, transcriptional activator-like effector nucleases (TALENs), or clustered regulatory interspaced short palindromic repeat/CRISPR associated protein 9 (CRISPR/Cas9) (for review see Merkert and Martin, 2016). While these technologies can lead to the genetic correction of mtHTT at the DNA level, the current in vivo technology is limited, primarily due to the relatively low efficiency of genetic correction and the need to correct the expanded allele in many tissues throughout the body (Kolli et al., 2017; Yang et al., 2017). Future studies utilizing these technologies will hopefully continue to improve upon what is already demonstrated to be promising techniques, the current limitations lead us to optimize the strategy of trans-splicing as a potential addition to the therapy toolkit for HD.

Trans-splicing also has its own set of strengths and weaknesses. One positive attribute is the inherent control of gene expression: trans-splicing cannot lead to gene overexpression, since the maximal amount of the trans-splicing event relies upon the endogenous HTT promoter. This is important in terms of appropriate gene “dosing.” A temporary knock-down of all HTT (both mutant and wild-type), termed a “HTT holiday” (Lu and Yang, 2012), has been shown to be tolerated in animals (McBride et al., 2011) and to reverse some symptoms of disease (Boudreau et al., 2009; Drouet et al., 2009; Kordasiewicz et al., 2012). However, other disease contexts demonstrate the complexity of HD, since a 50% decrease in murine Htt expression induces behavioral and cognitive abnormalities in ~1/3 of the mice, coupled with neurodegeneration in adulthood (Nasir et al., 1995; O'Kusky et al., 1999). Additional studies have demonstrated that targeted inactivation of Htt in the adult mouse brain results in neurodegeneration, further demonstrating that a delicate balance is required when envisioning HD therapeutics (Dragatsis et al., 2000). These studies would implicate the need for balance between reducing mtHTT and maintaining some functional level of wildtype HTT. However, studies using a cre-lox system to deplete neural (using the Nestin or CamKII promoters) Htt expression in adult mice (>4 months) demonstrated no changes in total brain volume or expression of neuronal proteins (Wang et al., 2016). With these conflicting studies there is obviously still work to be done examining the necessity of HTT expression in the adult brain. trans-Splicing avoids the haploinsufficiency issues as the molecular reaction converts mtHTT to wildtype HTT and any off-target trans-splicing to the non-pathogenic allele still results in wildtype HTT expression. While this report does not address the potential of allele-specific targeting, the net effect of non-allele specificity is a steady-state level of HTT expression, with at least some mutant to wild-type HTT conversion. Ideally, in the future, further modifications could be made to allow for allele-specific trans-splicing by targeting the tethering region to SNPs in the mtHTT, as has been done for HD ASO therapy (Skotte et al., 2014). In moving forward toward in vivo studies the CMV promoter would ideally be replaced with something that has shown sustained and robust in vivo expression within the CNS, such as the CBA promoter (Gray et al., 2011).

Perhaps one of the most significant hurdles to trans-splicing involves in vivo efficiency (Berger et al., 2016). While efficiencies have been reported ranging from <1 to ~40%, here we demonstrate that the efficacy of trans-splicing to the HTT mini-gene is ~10–15%. However, as wildtype HTT has been demonstrated to have an anti-apoptotic effect (Rigamonti et al., 2000, 2001; Gervais et al., 2002), the small reduction in mtHTT, paired with the increase in corrected HTT, may be sufficient to confer a therapeutic benefit. Consistent with this notion, trans-splicing mediated induction of HTT protein lead to the phenotypic reversal within neural cells derived from HD patient iPSCs. Another potential problem lies with recent reports of toxic RNA mtHTT species, in addition to the toxic protein (Bañez Coronel et al., 2012, 2015). It is currently unknown what effect trans-splicing would have on these transcripts. Ideally, the endogenous exon 1 pre-mRNA that was not spliced into the mature mRNA would be targeted for degradation, a likely path as this transcript does not possess a polyA signal and should be rapidly degraded through normal cellular processes. However, if there were additional toxic transcripts produced from the trinucleotide repeat (Bañez Coronel et al., 2012, 2015), these products would most likely not be targeted by this technology. Lastly, there are potential issues of non-disease gene disruption via off-target effects of the PTM or integration of the viral cassette into the genome. This is a consideration that will need to be examined for all gene therapy paradigms in the future. Therefore, in moving forward toward a translational product, the PTMs would most likely need to use a non-integrating viral vector, such as AAV. While direct administration of virus to the CNS is an option, using a viral vector able to cross the blood-brain-barrier after peripheral administration would be ideal (Deverman et al., 2016). To date, trans-splicing to spurious genomic sites has not been detected in our hands, however, a variety of “-omics” platforms, including RNA- and whole genome sequencing should be utilized before advancing further toward the clinic.

One interesting finding in this study was a ~47 kDa band, detected via both a GFP and a HTT antibody. This potential GFP-HTT fusion product was only detected in iPSC-derived neural progenitors that expressed PTMs. Theoretically, the reversal of the energy and cell death phenotypes seen in the HD iPSC-derived neural progenitors or striatal cultures, respectively, is due to the exon 1 replacement of mtHTT with the PTMs. However, we cannot exclude the possibility that the PTM expression of an exogenous HTT exon 1 fused with GFP is not contributing to the phenotype reversal. While it is known that HTT has pro-survival roles, the speculation presented here that a non-CAG expanded exon 1 alone could be of benefit in the presence of mtHTT in its self is extremely interesting and worthy of future pursuit as a potential avenue of therapy.

Trans-Splicing has been proposed for a potential therapeutic agent in other genetic disorders (Supplementary Table 2). These studies have all demonstrated efficacy of trans-splicing in vitro, and some in vivo (for a complete review please see Berger et al., 2016). Presently, this study builds upon our previous publication (Rindt et al., 2012) to demonstrate that modification of the binding modality of the PTM increases trans-splicing efficacy from 1–5% (Rindt et al., 2012) to 10–15% by altering the tethering region to specifically target the branch point of the endogenous intron 1. One interesting finding of this study is the increased aggregate retention seen via filter-trap assay, possibly as a result of the N-terminal exon 1 fragment seen via Western blotting analyses. This increased aggregate retention was seen in all samples expressing a PTM, to varying degrees which likely correlate with the original infection efficiency. This increase in aggregate retention did not result in increased aggregate formation or localization seen via immunocytochemistry. While historically HTT aggregate formation has been thought of as pathogenic (Khoshnan et al., 2002; Colby et al., 2004; Takeuchi et al., 2014), more recent papers have questioned this paradigm (Arrasate et al., 2004). Evidence for each side exists. For example, aggregate-suppressing therapeutic candidates can suppress toxicity (Warrick et al., 1999; Heiser et al., 2000; Lecerf et al., 2001; Kazantsev et al., 2002; Yoshida et al., 2002; Wang et al., 2005). Alternatively, many cell types that do not undergo degeneration (Kuemmerle et al., 1999) express even more aggregates than those that do (Vonsattel et al., 1985). Additionally, the soluble HTT population has instead been implicated in disease pathology (Arrasate et al., 2004). A recent study demonstrates that the aggregation of mtHTT may be protective against apoptosis, but may ultimately lead to a slow eventual cellular necrosis (Ramdzan et al., 2017). The findings presented here from the reversal of known patient iPSC-derived neural phenotypes potentially support the hypothesis that increased aggregation may be protective in HD, at least in the short-term. Longer-term studies of the effect of trans-splicing PTMs as a potential HD therapeutic, including those in mouse models, will therefore need to be performed.

The iPSC lines used in this study were approved by the IRB/SCRO at CSMC. All fibroblasts have been de-identified for sharing with the research community prior to generation of iPSCs, and no personal health information is available to any investigators beyond those who were involved in depositing them.

HR, CT, CL, and VM had substantial contributions to the conception or design of the work, the acquisition, analysis, or interpretation of data for the work. HR, CT, CL, and VM all drafted the manuscript and/or contributed critical revisions for important intellectual content and approved the final version. HR, CT, CL, and VM agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.