The trans-splicing technique also relies on a co-transduction approach in which the transgene is split such that one vector (vector A) contains a promoter, a 5′ portion of the gene, followed by a splice donor sequence (Fig. 2). The downstream vector (vector B) contains a splice acceptor sequence the remaining 3′ cDNA (or gene), and a poly A tail [4]. A concatemerization event in the correct orientation (theoretically about a 16 % chance) generates a single DNA molecule containing the intended large transgene expression cassette (Fig. 2) [4, 24]. Upon production of the pre-mRNA the intron is spliced out along with the ITR recombination junction, establishing an intact open reading frame large than can be packaged in a single vector [23, 25]. The trans-splicing method has been used in several Duchenne muscular dystrophy mouse models and has shown detectable restoration levels of dystrophin post-transduction [26–28]. There are also studies showing efficient whole-body and retina transduction that may also be promising for efficiently restoring large transgene expression [7, 29]. Even though efficient transduction is noticed in these experiments the transduction efficiency of these trans-splicing vectors is decreased compared to that of a single intact vector in skeletal muscle, eye, and liver [9]. The reason for this decreased transduction efficiency is due to several factors including (1) the efficiency of multiple vector transduction of a single cell, (2) the efficiency of the cellular machinery to reconstruct the large transgene in the desired orientation, (3) the decreased accumulation of mRNA due to the instability of the pre-mRNA, and (4) the gene splitting site [30]. The use of synthetic introns, including exonic splicing enhancers, can be used to at least partially counteract these rate-limiting steps [26, 31, 32].

The hybrid trans-splicing technique is essentially a combination of the overlapping and trans-splicing large gene delivery approaches described above. These hybrid vectors are designed in the same manner as trans-splicing vectors with one key difference: the inclusion of an overlapping sequence within the intron of both the 5′ and 3′ vectors (Fig. 2) [6, 13]. To increase the likelihood of the desired large gene reconstruction, particular sequences that are considered “recombinogenic” can be used as the overlapping sequence [6, 12, 13, 33]. Following co-transduction, the full-length transgene can either be formed by HR at the overlapping sequence or the ITR homology. Alternatively, the ITRs may join via NHEJ [9, 20]. Then, transcription, splicing and translation generate the desired protein as described above for trans-splicing vectors. Current reports comparing different overlapping “recombinogenic” sequences within the hybrid trans-splicing strategy suggest that a 77 nt sequence derived from the F1 phage genome is the most efficient in promoting large transgene reconstruction, at least in the tested cells [12]. This hybrid AK vector context, as well as the normal trans-splicing approach, has demonstrated relevance for the treatment of retinal diseases including Stargardt’s disease and Usher syndrome type 1B [12].

Regarding rAAV large gene delivery , reports have demonstrated the ability of the AAV capsid to package genomes over 5 kb [34–37]. For instance, Allocca et al. surveyed the packaging capacity of different capsid serotypes [34]. In the eye, it was demonstrated that AAV serotype 5 capsid, in particular, could package an expression cassette containing the adenosine triphosphate-binding cassette (ABCA4) at an unprecedented size of 8.9 kb, which was verified by alkaline gel electrophoresis [34]. Importantly, this AAV5-ABCA4 vector mediated effective transduction and resulted in the correction of disease phenotypes observed in the retina of a Stargardt’s disease mouse model [34]. The excitement for this groundbreaking report was later tempered by the reports of several groups that concluded that rAAV genomes over 5 kb were not packaged in their entirety, but instead DNA “fragments” of the intended oversized cassette were encapsidated (Fig. 3). Characterization of this process, demonstrated that oversized genome packaging initiates from the 3′ end and the external 5′ end is truncated when the capacity of the capsid is reached [8, 10, 11]. However, this process remains largely uncharacterized as heterogeneous DNA species (all <5 kb) are packaged in the apparently intact AAV capsids [8–11, 38]. Interestingly, these fAAV preparations retained their ability to mediate large gene transduction, albeit at a reduced efficiency compared to intact AAV in a tissue- and administration-dependent manner. A mechanistic evaluation of fAAV transduction in cell culture suggested that a large gene reconstruction event relies on HR, as the repair proteins Rad51C and XRCC3 were necessary while the NHEJ mediator DNA-PKcs was dispensable [9]. Despite the lack of an explanation for the earlier report demonstrating AAV5 large gene packaging [34], what is clearly evident is that a new rAAV large gene strategy emerged (fAAV) which, as in the other strategies, relies on the intracellular reconstruction of the oversized genome fragments, perhaps by canonical homology directed repair, at least in vitro [1, 9].