Can Adeno-Associated Viral Vectors Deliver Effectively Large Genes?

Abstract

Gene therapy with adeno-associated viral (AAV) vectors has reached the clinical stage for many inherited and acquired diseases. However, due to a cargo capacity limited to <5 kb, AAV-mediated treatment of diseases that require transfer of larger genes still appears elusive. This is a major drawback of a platform that has otherwise been repeatedly found to be safe and effective. Thus, great efforts have been directed toward the identification of strategies to overcome this limitation. Among the most studied approaches is the use of dual vectors, in which a transgene is split across two separate AAV vectors. Mechanisms acting at either the DNA, pre-mRNA, or protein levels have been explored to restore full-length transgene expression in infected cells. Here, we will review them as well as additional strategies developed to deliver large genes with AAV. We discuss the pros and cons of these strategies and the aspects that still need to be addressed.

Introduction

Recombinant adeno-associated viral (AAV) vectors are currently considered the leading platform for in vivo gene therapy. Since the initial isolation of an infectious AAV clone,^1,2 many efforts have been devoted to refining this virus as a gene therapy vehicle. Various reasons are behind AAV vectors gaining popularity over the years, including the lack of pathogenicity of wild-type AAV, their low immunogenicity, and the availability of different serotypes with broad but distinct tropism and ability to transduce a variety of target tissues, including liver, retina, cardiac muscle, and central nervous system.^3
–5 These aspects, as well as the substantial innovations that have been established over the years, have boosted the investigation of AAV as a human gene therapy platform.

Although initially explored for the delivery of genes mutated in monogenic inherited diseases, AAV vectors were later adapted for targeting more complex conditions or delivering innovative tools, as those for genome editing. More than three decades of pre-clinical research have thus culminated in three AAV-based gene therapy products for inherited diseases affecting different tissues receiving market approval from the European Medicines Agency and the American Food and Drug Administration between 2012 and 2019.^6
–8 Despite the encouraging clinical success of these AAV-based gene therapy products, there are still a number of challenges of this gene delivery platform that limit its broad use. Addressing these challenges is one of the main current focus of the gene therapy field.

One of the drawbacks of AAV is that, being the smallest virus described to date, its optimum genome-packaging capacity is <5 kb. Considering that recombinant AAV vectors are generated by replacing the whole AAV genome, except the two 145 bp-long inverted terminal repeats (ITRs), still this allows the incorporation of less than 4.5 kb of foreign DNA, which should include at a minimum the coding sequence (CDS) of the therapeutic gene as well as the promoter and the polyadenylation (polyA) sequence. Therefore, this cargo capacity is clearly insufficient to deliver genes with a CDS that exceeds 4 kb, including those causative of common inherited diseases (Table 1).

Table 1.

Large therapeutic genes and multiple adeno-associated viral vector-based strategies for their delivery

Disease	Gene	Accession No.	CDS (kb)	Reconstitution Strategy	Target Tissue	References
Cystic fibrosis	CFTR	NM_000492	4.4	Segmental trans-splicing	Lung	⁸⁴
Glycogen storage disease III	GDE	NM_000028	4.6	Dual AAV	Muscle, liver	⁵⁹
Nonsyndromic deafness	OTOF	NM_001287489	6	Dual AAV	Ear	⁵⁶
Dysferlinopathies	DYSF	NM_003494	6.2	Dual AAV	Muscle	^41,57
Usher 1B	MYO7A	NM_000260	6.7	Dual AAV	Retina	^{52,55,71,73,74}
Stargardt disease	ABCA4	NM_000350	6.8	Dual AAV/protein trans-splicing	Retina	^{52,53,55,74,75/99}
Hemophilia A	F8	NM_000132	7.1	Dual AAV	Liver	⁶²
Leber congenital amaurosis 10	CEP290	NM_025114	7.5	Protein trans-splicing	Retina	⁹⁹
Usher 1D	CDH23	NM_022124	10.1	Triple AAV	Retina	⁷⁸
Duchenne muscular dystrophy	DMD	NM_000109	11	Dual-triple AAV/protein trans-splicing	Muscle	^{29,51,58,61,66,67,69,70,76,77/92}
Alstrom syndrome	ALMS1	NM_015120	12.5	Triple AAV	Retina	⁷⁸

AAV, adeno-associated viral vectors; CDS, coding sequence.

In this review, we will discuss the various strategies that have been explored to adapt AAV for delivering large genes.

Fitting Large Genes in a Single AAV Vector

A seemingly straightforward way to overcome the limited cargo capacity of AAV is to shrink the transgene expression cassette.

Efforts have been directed toward generation of either shortened versions of the regulatory elements (i.e., promoters and polyA),^9

–12 retaining their efficiency and specificity, or development of shortened versions of therapeutic genes,^9,11,13 to encode a truncated but functional protein. Both strategies have been explored with some degrees of success but have also shown some limitations. Indeed, generation of shorter versions of the regulatory elements often led to reduced levels of transgene expression. A combination of short synthetic enhancers and promoters was explored to overcome this issue.¹⁴ Development of truncated versions of a therapeutic gene requires a deep understanding of the structure–function relationship of the protein and is, therefore, gene specific, requiring extensive optimization and testing for each gene.

Historically, the first attempt at optimizing the delivery of a large gene via AAV vectors is represented by the AAV2-mediated delivery of the cystic fibrosis transmembrane conductance regulator (CFTR) gene.

The large size of the CFTR CDS (∼4.4 kb), indeed, leaves little space for the regulatory elements in the vector; therefore, a number of efforts have been directed toward shrinking the CFTR expression cassette by either generating truncated forms of CFTR that retain their functionality or identifying short promoter/polyA signals.¹⁵ Flotte et al. first attempted the delivery of an AAV vector in which the expression of a truncated version of CFTR was driven by the ITRs of AAV. These, indeed, were found to have a low promotorial activity, resulting in detectable levels of CFTR both in vitro and in vivo.^16

–19 However, the levels of expression in patients with cystic fibrosis (CF) enrolled in the first clinical trial were found to be generally low.^18,20,21

Further studies have shown that targeted deletions in the CFTR CDS, to allow inclusion in the AAV genome of longer and more efficient promoters and/or enhancers, result in higher efficiency of CFTR reconstitution in cells,^{9,14,15,22
–24} confirming that reducing the length of the expression cassette by minimizing the size of the regulatory elements, at the cost of their strength, might limit the efficacy of the gene therapy approach.

Micro-gene therapy is under extensive clinical evaluation also for gene therapy of Duchenne muscular dystrophy (DMD), with three ongoing clinical trials in the United States and one being planned in Europe.²⁵ The large size of the dystrophin gene (CDS: ∼11 kb) has been a major obstacle in developing methods for DMD gene therapy for a long time. However, the initial discovery of a highly functional mini-version of the dystrophin protein in mildly affected Becker muscular dystrophy patients,²⁶ as well as the numerous studies that have generated considerable knowledge regarding the structural domains of dystrophin, prompted rational design of smaller but highly functional mini- and micro-dystrophin versions suitable for AAV-mediated gene therapy applications. These have been shown to provide some degrees of therapeutic benefit.^27

–30

The size of the CDS of additional large proteins has been optimized to fit into AAV vectors.^31

–34 Some of these mini-protein versions have already reached the clinical stage,³⁵ whereas other forms of therapeutic applicability are still under preclinical investigation, to deeply evaluate retention of the proper functionality.³⁶ For many diseases, however, the generation of truncated versions that retain protein function has so far not been achieved. Consequently, the development of additional methods to express large therapeutic proteins by AAV gene transfer has remained an important area of research.

Another interesting strategy that has been used by several research groups to deliver large genes through AAV vectors is to encapsidate the full-length expression cassette (>5 kb) in a single AAV, without any size optimization.^37

–41 These so called “oversize” AAV vectors have been found to successfully express full-length large proteins both in vitro and in vivo.^37

–41 However, it was found that the genome contained in oversize AAV vectors was not a pure population of intact large-size genomes but rather a mixture of genomes highly heterogeneous in size.^39,42

–45 Reassembly of these truncated genomes in the target cell nucleus after infection has been hypothesized to be the mechanism responsible for the reconstitution of the full-length expression cassette in infected cells and, thus, for the full-length protein expression seen in different studies.^{39,40,42

–45}

Despite the well-documented ability of this straightforward approach to mediate large gene transduction in vivo, the genome heterogeneity of oversize AAV vectors opens fundamental questions regarding the safety of the approach, which limit its further clinical translation without additional optimizations.

An alternative strategy that is being investigated to deliver large genes is to leverage the capsid of the human bocavirus 1 (HBoV1),^46,47 an autonomous parvovirus relative of AAV, with a 5.5 kb genome. Indeed, it has been shown that an oversize AAV2 vector genome, including the CFTR expression cassette, can be effectively packaged in the HBoV1 capsid, resulting in efficient transduction of the human-polarized airway epithelia.⁴⁷ Further investigation of the tropism of this vector for additional tissues could expand the portfolio of strategies available for delivery of large genes.

Exploring the Use of Multiple AAV Vectors to Deliver Large Genes

Although the ability to deliver large transgenes in a single AAV vector has been a longstanding goal of the gene therapy field, the strategy that has been most extensively studied and tested over the past years to overcome the limited AAV cargo capacity is to co-administer two or more AAV vectors that carry separate parts of a gene. Joining of the two halves at either the DNA, pre-mRNA, or protein level is then explored to reconstitute the expression of the large therapeutic protein (Fig. 1), as described later.

Figure 1.

Overview of multiple AAV-based approaches for expression of large therapeutic proteins. The CDS of a large gene is split in two halves (5′ and 3′), flanked by the ITRs, which are separately packaged into two AAV capsids. On co-transduction of the same cell, different mechanisms are explored to reconstitute full-length protein expression through joining of the two halves at either the DNA (A), pre-mRNA (B), or protein (C) levels. (A) Schematic representation of dual AAV-mediated reconstitution of a large gene. In this approach, the 5′-vector includes the promoter and the 5′CDS, whereas the 3′-vector includes the 3′CDS and the polyA signal. In the trans-splicing strategy, a SD signal is included after the 5′CDS in the 5′-vector, whereas an SA signal is included before the 3′CDS in the 3′-vector. The full-length expression cassette is reconstituted through ITR-mediated concatemerization of the two vectors and splicing, which removes the ITR structure at the junction point. In the overlapping strategy, a region of the transgene is repeated in both vectors so that the full-length expression cassette is reconstituted through homologous recombination. In the hybrid strategy, the two vectors are designed as in the trans-splicing strategy, with the addition of a highly recombinogenic exogenous sequence (HR) downstream of the SD in the 5′-vector and upstream of the SA in the 3′-vector. Thus, the full-length expression cassette can be reconstituted by either ITR-mediated concatemerization (as in the trans-splicing strategy) or through homologous recombination mediated by the recombinogenic sequence, followed by splicing to remove the exogenous sequences at the junction point. (B) Schematic representation of pre-mRNA trans-splicing-mediated reconstitution of a large mRNA. In this approach, the 5′-vector includes the 5′CDS and the SD, whereas the 3′-vector includes the SA, the 3′CDS, and the polyA; both vectors include the promoter and an HD. Pairing of the two mRNAs through the complementary HDs is followed by splicing, which leads to the removal of the HD in the junction point and full-length mRNA reconstitution. (C) Schematic representation of protein trans-splicing-mediated reconstitution of a large protein. The 5′-vector includes the 5′CDS and 5′intein (n-intein), whereas the 3′-vector includes the 3′CDS and 3′intein (c-intein); both vectors include the promoter and the polyA. Pairing of the two half polypeptides is mediated via inteins self-recognition; subsequent inteins self-excision from the host protein results in full-length protein reconstitution. AAV, adeno-associated viral vectors; CDS, coding sequence; HD, hybridization domain; HR, highly recombinogenic exogenous sequence; ITRs, inverted terminal repeats; polyA, polyadenylation; SA, splicing acceptor; SD, splicing donor.

Dual AAV-mediated reconstitution of a large gene

Various dual AAV strategies to reconstitute a large gene in target cells have been described, which rely on different mechanisms: trans-splicing, overlapping, and hybrid AAV vectors (Fig. 1A).

The trans-splicing approach takes advantage of the inherent ability of AAV's ITRs to concatemerize to reconstitute full-length genomes.^48,49 In this approach, the two vectors carry two separate halves of the transgene, and they are designed as follows: the 5′-vector has the promoter, the 5′ half of the CDS, and a splicing donor (SD) signal; the 3′-vector has a splicing acceptor (SA) signal, the 3′ half of the CDS, and the polyA signal. On reconstitution of the full-length genome through tail-to-head ITR-mediated concatemerization of the two AAV genomes, the SD and SA signals will allow splicing of the concatemerized ITR structure that forms at the junction point to obtain an intact large mRNA molecule.

The overlapping approach relies on homologous recombination between an overlapping sequence of the CDS included in both AAV vectors to reconstitute a full-length genome.⁵⁰ Thus, the 5′-vector carries the promoter and the 5′ half of the CDS, whereas the 3′-vector carries the 3′ half of the CDS and the polyA signal.

The hybrid approach is a combination of the trans-splicing and the overlapping approaches, as it is based on the addition of a highly recombinogenic exogenous sequence to the trans-splicing vectors to increase recombination efficiency.⁵¹ This recombinogenic sequence is placed downstream of the SD signal in the 5′-vector and upstream of the SA signal in the 3′-vector. On either ITR-mediated concatemerization of the two AAV vectors or homologous recombination mediated by the exogenous recombinogenic sequence, the splicing signals will allow splicing of the recombinogenic sequence and/or recombinant ITR structure to restore the large gene mRNA.

Each of these strategies has some advantages as well as drawbacks for successful application.

Although the overlapping approach is the simplest in design since it does not require foreign or artificial DNA elements compared with the other approaches, its efficiency is strictly both transgene- and target cell-dependent. Indeed, both the extent of activity of the repair mechanism on which overlapping vectors rely for large gene reconstitution in the target cells and the intrinsic ability of the overlapping region to mediate efficient homologous recombination will significantly impact the efficiency of the platform.^52,53

Vice versa, ITR concatemerization, on which trans-splicing vectors rely, is a spontaneous process that has been described to occur in target cells on infection.⁴⁹ However, for efficient trans-splicing to occur, there are a number of rate-limiting steps to be overcome (for a detailed discussion see Ghosh and Duan⁵⁴). These include formation of the concatemer in the productive orientation (i.e., tail-to-head) and effective splicing, the last being influenced by both the sequence of the gene that surrounds the splitting point and the efficiency of the splicing signals.

The hybrid approach was, thus, developed with the aim of overcoming the limitations of the two previous approaches.⁵¹ The use of a highly recombinogenic exogenous sequence is expected to increase the chances of directing dual AAV vectors joining in the proper orientation compared with the trans-splicing approach. Concomitantly, it should assure efficient homologous recombination in a transgene-independent manner, differently from the overlapping approach. Yet, similarly to the trans-splicing approach, the sequences surrounding the splicing signals have an impact on splicing efficiency, so careful selection of the splitting point is requested to achieve maximal efficacy of large gene reconstitution.

Assessment of the potential of dual AAV vectors in reconstituting both reporter and therapeutic genes in vitro and in vivo has been performed, having as a target various tissues, including sensory organs,^55,56 muscle,^41,57
–59 lung,⁶⁰ heart,⁶¹ and liver.^59,62 The different studies have led to variable results. This is somewhat expected as there are several factors to consider: the design of the platform, target cell types, AAV vector serotype and dose used, as well as transduction levels achieved. These factors all impact the success of transduction from dual AAV vectors. Thus, even slight differences between the studies might account for variable outcomes.

Historically, the trans-splicing approach has been the first to be developed^48,63
–65 and initially tested for erythropoietin⁴⁸ and factor VIII⁶² gene delivery. Some years later, successful reconstitution of the mini-dystrophin gene was also achieved by intramuscular injection of trans-splicing vectors.⁵⁸ Despite the mini-dystrophin studies that clearly highlighted the importance for this platform of the selection of the optimal splitting point to achieve efficient transgene expression,^58,66 the success in reconstituting the gene and improving the phenotype of the DMD mouse model prompted further investigations of the ability of trans-splicing vectors to reconstitute large transgenes.^57,67,68

Overlapping vectors have been more consistently used to reconstitute genes in the muscle, compared with other tissues, as the retina, confirming that some tissues are more amenable than others to homologous recombination-based approaches.

Intravenous dual AAV overlapping-mediated delivery of the mini-dystrophin gene has been found to result in levels of expression that, although lower than in wild-type muscles, were able to improve muscle performance.^29,69 Interestingly, the efficiency of reconstitution from overlapping vectors was also found to be superior to three different sets of hybrid vectors, suggesting that specific regions of the DMD gene might be particularly prone to homologous recombination.⁷⁰ Similarly, a study comparing different strategies for dysferlin delivery to the muscle identified the overlapping as the most efficient approach, although the hybrid vectors relied on a region of homology derived from an intron of the dysferlin gene, which might have been not sufficiently recombinogenic.⁴¹

Differently, in the retina, low levels of overlapping vectors-mediated transduction have been generally found in the terminally differentiated photoreceptors across different studies,^52,71,72 whereas more efficient reconstitution was found in the retinal pigmented epithelium.⁵² Optimization of the overlapping region and use of serotypes with high retinal transduction ability were found to be prerequisites to achieve sustained levels of transgene expression.^53,73

In the absence of identification of optimal overlapping regions in the transgene or in the presence of poor splitting sites, the hybrid approach has been found to significantly increase the efficiency of transgene reconstitution compared with overlapping or trans-splicing vectors.^51,52

In the retina, various studies have shown the ability of hybrid vectors to reconstitute large transgenes,^52,72
–74 at levels that were higher compared with the other dual AAV strategies side-by-side tested^52,72,73 and that resulted in improvement of the retinal phenotype of animal models of inherited retinal diseases.^52,75 As a confirmation, prompted by the promising results obtained in the delivery of the large MYO7A gene to the retina of the Usher 1B mouse model,⁵² a phase I/II clinical trial has been planned to investigate the efficacy of hybrid vectors in Usher 1B patients (https://cordis.europa.eu/project/rcn/212674_it.html).

Recently, the possibility of expanding the dual-vector system to a triple-vector system, increasing the genome capacity of AAV vectors even up to 14 kb, has also been investigated by different groups.^76
–78 Although full-length gene reconstitution has been shown to occur with triple AAV vectors in different studies,^76
–78 resulting in transient improvement of the phenotype of a mouse model of an inherited retinal disease,⁷⁸ the expression levels overall were found to be weak. Nonetheless, future optimizations that improve single or multiple AAV transduction might allow the use of triple AAV vectors for delivery of proteins that are too large for even dual-vector systems.

Two major drawbacks are associated with dual and triple AAV vector platforms. First, production of short unwanted protein products derived from the free single halves, which have not been engaged in full-length gene reconstitution, have been observed in various studies. Optimization of dual AAV vectors design as well as inclusion of in-frame degradation sequences, which impact only production of the unwanted protein products without altering full-length protein production, has been used to overcome this issue.^53,75

Second, with the exception of a few reports,^60,79 all the strategies have shown lower efficiency compared with a single normal-size AAV vector. Whether these levels are sufficient to achieve therapeutically meaningful improvements in some diseases is still an open question that needs to be investigated in future clinical trials.

Pre-mRNA trans-splicing-mediated reconstitution of a large mRNA

The discovery that primary transcripts of eukaryotic genes can undergo the process of RNA trans-splicing (i.e., spliceosome-mediated splicing between two different pre-mRNA molecules)^80,81 opened new possibilities for gene therapy, based on correction of defective genes at the RNA rather than at the DNA level.

Although initially trans-splicing has been explored to correct mutations in endogenously produced mutated pre-mRNAs by delivery of an exogenous corrected fragment of the endogenous gene,⁸² adaptations of this process have been later developed. In the approach defined as “segmental trans-splicing” (STS), two engineered individual DNA fragments encode for the 5′ and 3′ fragments of the pre-mRNA of a large gene and share an intronic hybridization domain that can favor trans-splicing, leading to joining of the two half-transcripts into an intact full-length mRNA⁸³ (Fig. 1B). Interestingly, AAV-mediated delivery of the 5′- and 3′-halves of the CFTR cDNA in human CF airway epithelial cells followed by STS was found to result in full-length CFTR reconstitution and restoration of CF cells' functionality.⁸⁴

Despite this pivotal proof-of-concept study, however, the inability to achieve significant levels of pre-mRNA trans-splicing, as well as the several scientific and technical challenges related to the use of RNA-based repair strategies^82,85 have limited further investigation of STS in combination with AAV vectors as a strategy for corrections of mutations in large genes.

Protein trans-splicing-mediated reconstitution of a large protein

Another system that has been tested to reconstitute large proteins via AAV vectors relies on the use of the protein trans-splicing (PTS) mechanism.

PTS is a post-translational process catalyzed by intervening proteins called split-inteins. Split-inteins are expressed as two independent polypeptides (N-intein and C-intein) at the extremities of two host proteins (N-polypeptide and C-polypeptide) and remain inactive until encountering their complementary partner.⁸⁶ On association with the counterpart, the reconstituted intein precisely excises itself from the host proteins while mediating ligation of the N- and C-polypeptides via a peptide bond, in a traceless manner (Fig. 1C). This PTS process occurs without any known cofactor or source of energy. Proper folding of the split-inteins as well as the presence of a nucleophilic residue as first amino acid of the C-polypeptide, to assist catalysis, are the only requirements.⁸⁷

The first natural split-intein was identified in 1998.⁸⁶ Since then, a plethora of natural and artificial split-inteins have been developed and exploited in different biotechnological applications,^88,89 including reconstitution of large genes. The first proof of concept that split-inteins could effectively reconstitute large genes was obtained by using the functional B-domain deleted FVIII (BDD-FVIII). BDD-FVIII light and heavy chains were found to be efficiently ligated from split-inteins, resulting in increased circulating FVIII and coagulation activity in the plasma.^90,91

Given this promising result, split-intein use has been further adapted to AAV-based applications, to deliver therapeutic genes that are mutated in muscle, liver, and retinal diseases. The first evidence that split-intein-based gene therapy approaches could be used to deliver large genes via AAV vectors was obtained by using the mini-dystrophin gene.⁹² On co-delivery in mdx mice of the two halves of the mini-dystrophin cDNA fused to N- and C-inteins, efficient production of the two polypeptides was shown. However, no detailed investigation on the functionality of the reconstituted protein was performed.⁹²

The most extensive use of the AAV-split-inteins platform has been for the reconstitution of the large and widely used clustered regularly interspaced short palindromic repeats (CRISPR)–Cas9 genome editing nuclease, both in vitro and in vivo.^93
–95 Indeed, many of the developed Cas9 proteins have a CDS that exceeds the canonical AAV cargo capacity⁹⁶; thus, identification of split-intein-Cas9 sets that retain Cas9 endonuclease efficiency on reconstitution has been an exciting field of research over the past few years.

The possibility to split Cas9 proteins in halves has, in addition, opened the unprecedent opportunity to deliver engineered Cas9-fusion proteins,⁹⁴ which have even larger sizes than wild-type Cas9, via AAV vectors. Editing of a disease-causing mutation using CRISPR-associated base editors in a mouse model of phenylketonuria was found to restore physiological blood phenylalanine levels.⁹⁷ Similarly, targeted repression of a master regulator gene of photoreceptor determination, mediated by split-inteins-Cas9 fused to a repressor domain, prevented vision loss in a mouse model of retinitis pigmentosa.⁹⁸

We have recently demonstrated that AAV-intein-based platforms can be explored to reconstitute large therapeutic proteins in the retina of animal models and in human photoreceptors from retinal organoids. Delivery of AAV-intein vectors encoding for the ABCA4 protein in a mouse model of Stargardt disease resulted in significant amelioration of the phenotype.⁹⁹ Importantly, in a side-by-side comparison between the AAV-intein and dual AAV hybrid platforms, we found that the levels of AAV intein-mediated large protein reconstitution largely exceeded those achieved via AAV genome recombination by dual AAV vectors.⁹⁹ In addition, in the large pig retina we have found levels of AAV-intein-mediated protein reconstitution comparable to those achieved with a single AAV vector.

One of the drawbacks of the AAV-intein platform is that part of the cloning capacity will be taken up by regulatory elements that need to be replicated within each vector. Due to this design, the packaging capacity of the platform might still be insufficient for some large genes. Thus, use of short, and often weaker, regulatory elements would be required.

Thanks to the availability of different inteins, which do not cross-react,^91,99,100 we have shown that it is possible to expand the AAV intein system to a triple vector-based system in which the three polypeptides are fused with two different sets of inteins, thus mediating directional full-length protein reconstitution.⁹⁹ Along this line, triple AAV-intein-mediated delivery of the large CEP290 protein resulted in more efficient full-length protein expression than that obtained with two AAV-intein vectors, which were lacking additional potent regulatory elements, and in therapeutic efficacy in the mouse model of Leber congenital amaurosis type 10.⁹⁹

Additional drawbacks associated with the AAV-intein platform need to be addressed in the near future to expand the use of this promising platform. First, construct design appears to be particularly critical compared with other platforms and will need careful evaluation for each large protein. Second, although the AAV-intein platform appears not to be toxic,⁹⁹ the presence of both intein excised from the mature protein and non–trans-spliced single polypeptides opens questions regarding the safety of the platform that will require formal long-term toxicity studies to be finally addressed.

Conclusions and Future Directions

Expanding the cargo capacity of AAV vectors beyond 5 kb has been a longstanding goal of the gene therapy field. Both adaptations of large expression cassettes to fit into a single AAV vector and adaptations of the AAV platform to deliver large genes have been explored. Some strategies, having shown more success, have gained more attention and, thus, their efficiency has been investigated in multiple tissues. The results are often conflicting and have highlighted that there is not one strategy that can fit every need.

Overall, dual AAV strategies have been more reproducibly found to reconstitute large genes, although clearly there are optimizations required for each platform to achieve maximal success, as described earlier. Limitations related to the efficiency of gene reconstitution, which still remains lower than that of a single vector, as well as to the safety of the platform, because of the potential production of shorter protein products, have not been definitively addressed. Due to these limitations, dual AAV applications to target diseases that affect small and confined tissues, as the eye, could have higher chances of success, because of the high rates of co-transduction achievable⁷⁴ and limited safety concerns raised from the production of the shorter protein products.

More recently, strategies that rely on intein-mediated PTS for large protein reconstitution are being increasingly explored. Interestingly, the efficiency of reconstitution appears to be higher compared with other platforms.

However, the design of AAV-intein constructs is particularly critical since it needs to take into account both preservation at the junction points of amino acid residues needed for efficient PTS and splitting of the therapeutic proteins outside of structural domains to avoid incorrect polypeptide folding. Thus, screening of various splitting points might be required to identify a suitable set of AAV-intein vectors for each therapeutic gene. Despite the optimizations, some half polypeptides might show limited stability or incorrect targeting inside the cell, resulting in inefficient PTS. In addition, safety issues related to the production of additional protein products remain.

In conclusion, although still at the pre-clinical stage, the research around large gene delivery using AAV vectors is showing encouraging signs of progression. The choice of the best strategy for AAV-mediated large gene delivery will most probably depend on the protein to be reconstituted, the target cells, and the nature of the disease. New emerging strategies, as either genome editing^101,102 or targeted exon skipping using anti-sense oligonucleotides,^103
–105 could further enrich the portfolio of available strategies for correction of diseases due to mutations in large genes. With all these advancements, more and more diseases will become amenable to AAV-mediated gene delivery.

Footnotes

Author Disclosure

No competing financial interests exist.

Funding Information

This work was supported by the University of Naples Federico II under STAR Program (to I.T.).

References

Atchison

, Casto

, Hammon

. Adenovirus-associated defective virus particles. Science, 1965; 149:754–756.

Hoggan

, Blacklow

, Rowe

. Studies of small DNA viruses found in various adenovirus preparations: physical, biological, and immunological characteristics. Proc Natl Acad Sci U S A, 1966; 55:1467–1474.

Wang

, Tai

PWL

, Gao

. Adeno-associated virus vector as a platform for gene therapy delivery. Nat Rev Drug Discov, 2019; 18:358–378.

Keeler

, Flotte

. Recombinant adeno-associated virus gene therapy in light of luxturna (and Zolgensma and Glybera): where are we, and how did we get here?. Annu Rev Virol, 2019; 6:601–602.

Trapani

, Auricchio

. Has retinal gene therapy come of age? From bench to bedside and back to bench. Hum Mol Genet, 2019; 28(R1):R108–R118.

Yla-Herttuala

. Endgame: glybera finally recommended for approval as the first gene therapy drug in the European Union. Mol Ther, 2012; 20:1831–1832.

FDA approves hereditary blindness gene therapy. Nat Biotechnol 2018;36:6.

Hoy

. Onasemnogene abeparvovec: first global approval. Drugs, 2019; 79:1255–1262.

Ostedgaard

, Rokhlina

, Karp

, et al. A shortened adeno-associated virus expression cassette for CFTR gene transfer to cystic fibrosis airway epithelia. Proc Natl Acad Sci U S A, 2005; 102:2952–2957.

10.

McFarland

, Zhang

, Atchaneeyaskul

, et al. Evaluation of a novel short polyadenylation signal as an alternative to the SV40 polyadenylation signal. Plasmid, 2006; 56:62–67.

11.

Scallan

, Lillicrap

, Jiang

, et al. Sustained phenotypic correction of canine hemophilia A using an adeno-associated viral vector. Blood, 2003; 102:2031–2037.

12.

Powell

, Rivera-Soto

, Gray

. Viral expression cassette elements to enhance transgene target specificity and expression in gene therapy. Discov Med, 2015; 19:49–57.

13.

Toole

, Pittman

, Orr

, et al. A large region (approximately equal to 95 kDa) of human factor VIII is dispensable for in vitro procoagulant activity. Proc Natl Acad Sci U S A, 1986; 83:5939–5942.

14.

Yan

, Sun

, Feng

, et al. Optimization of recombinant adeno-associated virus-mediated expression for large transgenes, using a synthetic promoter and tandem array enhancers. Hum Gene Ther, 2015; 26:334–346.

15.

Guggino

, Cebotaru

. Adeno-associated virus (AAV) gene therapy for cystic fibrosis: current barriers and recent developments. Expert Opin Biol Ther, 2017; 17:1265–1273.

16.

Flotte

, Afione

, Conrad

, et al. Stable in vivo expression of the cystic fibrosis transmembrane conductance regulator with an adeno-associated virus vector. Proc Natl Acad Sci U S A, 1993; 90:10613–10617.

17.

Flotte

, Afione

, Solow

, et al. Expression of the cystic fibrosis transmembrane conductance regulator from a novel adeno-associated virus promoter. J Biol Chem, 1993; 268:3781–3790.

18.

Wagner

, Reynolds

, Moran

, et al. Efficient and persistent gene transfer of AAV-CFTR in maxillary sinus. Lancet, 1998; 351:1702–1703.

19.

Conrad

, Allen

, Afione

, et al. Safety of single-dose administration of an adeno-associated virus (AAV)-CFTR vector in the primate lung. Gene Ther, 1996; 3:658–668.

20.

Flotte

, Schwiebert

, Zeitlin

, et al. Correlation between DNA transfer and cystic fibrosis airway epithelial cell correction after recombinant adeno-associated virus serotype 2 gene therapy. Hum Gene Ther, 2005; 16:921–928.

21.

Aitken

, Moss

, Waltz

, et al. A phase I study of aerosolized administration of tgAAVCF to cystic fibrosis subjects with mild lung disease. Hum Gene Ther, 2001; 12:1907–1916.

22.

Zhang

, Wang

, Fischer

, et al. Efficient expression of CFTR function with adeno-associated virus vectors that carry shortened CFTR genes. Proc Natl Acad Sci U S A, 1998; 95:10158–10163.

23.

Wang

, Fischer

, Zhang

, et al. Efficient CFTR expression from AAV vectors packaged with promoters—the second generation. Gene Ther, 1999; 6:667–675.

24.

Sirninger

, Muller

, Braag

, et al. Functional characterization of a recombinant adeno-associated virus 5-pseudotyped cystic fibrosis transmembrane conductance regulator vector. Hum Gene Ther, 2004; 15:832–841.

25.

Duan

. Systemic AAV micro-dystrophin gene therapy for Duchenne muscular dystrophy. Mol Ther, 2018; 26:2337–2356.

26.

England

, Nicholson

, Johnson

, et al. Very mild muscular dystrophy associated with the deletion of 46% of dystrophin. Nature, 1990; 343:180–182.

27.

Bostick

, Shin

, Yue

, et al. AAV-microdystrophin therapy improves cardiac performance in aged female mdx mice. Mol Ther, 2011; 19:1826–1832.

28.

Gregorevic

, Blankinship

, Allen

, et al. Systemic microdystrophin gene delivery improves skeletal muscle structure and function in old dystrophic mdx mice. Mol Ther, 2008; 16:657–664.

29.

Odom

, Gregorevic

, Allen

, et al. Gene therapy of mdx mice with large truncated dystrophins generated by recombination using rAAV6. Mol Ther, 2011; 19:36–45.

30.

Harper

, Hauser

, DelloRusso

, et al. Modular flexibility of dystrophin: implications for gene therapy of Duchenne muscular dystrophy. Nat Med, 2002; 8:253–261.

31.

Zhang

, Li

, Su

, et al. Gene therapy using a miniCEP290 fragment delays photoreceptor degeneration in a mouse model of Leber congenital amaurosis. Hum Gene Ther, 2018; 29:42–50.

32.

Burton

, Nakai

, Colosi

, et al. Coexpression of factor VIII heavy and light chain adeno-associated viral vectors produces biologically active protein. Proc Natl Acad Sci U S A, 1999; 96:12725–12730.

33.

Krahn

, Wein

, Bartoli

, et al. A naturally occurring human minidysferlin protein repairs sarcolemmal lesions in a mouse model of dysferlinopathy. Sci Transl Med, 2010; 2:50ra69.

34.

Chao

, Mao

, Bruce

, et al. Sustained expression of human factor VIII in mice using a parvovirus-based vector. Blood, 2000; 95:1594–1599.

35.

Peyvandi

, Garagiola

. Clinical advances in gene therapy updates on clinical trials of gene therapy in haemophilia. Haemophilia, 2019; 25:738–746.

36.

Lostal

, Bartoli

, Roudaut

, et al. Lack of correlation between outcomes of membrane repair assay and correction of dystrophic changes in experimental therapeutic strategy in dysferlinopathy. PLoS One, 2012; 7:e38036.

37.

Allocca

, Doria

, Petrillo

, et al. Serotype-dependent packaging of large genes in adeno-associated viral vectors results in effective gene delivery in mice. J Clin Invest, 2008; 118:1955–1964.

38.

Grieger

, Samulski

. Packaging capacity of adeno-associated virus serotypes: impact of larger genomes on infectivity and postentry steps. J Virol, 2005; 79:9933–9944.

39.

Hirsch

, Li

, Bellon

, et al. Oversized AAV transductifon is mediated via a DNA-PKcs-independent, Rad51C-dependent repair pathway. Mol Ther, 2013; 21:2205–2216.

40.

Grose

, Clark

, Griffin

, et al. Homologous recombination mediates functional recovery of dysferlin deficiency following AAV5 gene transfer. PLoS One, 2012; 7:e39233.

41.

Pryadkina

, Lostal

, Bourg

, et al. A comparison of AAV strategies distinguishes overlapping vectors for efficient systemic delivery of the 6.2 kb Dysferlin coding sequence. Mol Ther Methods Clin Dev, 2015; 2:15009.

42.

Dong

, Nakai

, Xiao

. Characterization of genome integrity for oversized recombinant AAV vector. Mol Ther, 2010; 18:87–92.

43.

Hirsch

, Agbandje-McKenna

, Samulski

. Little vector, big gene transduction: fragmented genome reassembly of adeno-associated virus. Mol Ther, 2010; 18:6–8.

44.

Lai

, Yue

, Duan

. Evidence for the failure of adeno-associated virus serotype 5 to package a viral genome > or = 8.2 kb. Mol Ther, 2010; 18:75–79.

45.

, Yang

, Colosi

. Effect of genome size on AAV vector packaging. Mol Ther, 2010; 18:80–86.

46.

Fakhiri

, Schneider

, Puschhof

, et al. Novel chimeric gene therapy vectors based on adeno-associated virus and four different mammalian bocaviruses. Mol Ther Methods Clin Dev, 2019; 12:202–222.

47.

Yan

, Keiser

, Song

, et al. A novel chimeric adenoassociated virus 2/human bocavirus 1 parvovirus vector efficiently transduces human airway epithelia. Mol Ther, 2013; 21:2181–2194.

48.

Yan

, Zhang

, Duan

, et al. Trans-splicing vectors expand the utility of adeno-associated virus for gene therapy. Proc Natl Acad Sci U S A, 2000; 97:6716–6721.

49.

Yang

, Zhou

, Zhang

, et al. Concatamerization of adeno-associated virus circular genomes occurs through intermolecular recombination. J Virol, 1999; 73:9468–9477.

50.

Duan

, Yue

, Engelhardt

. Expanding AAV packaging capacity with trans-splicing or overlapping vectors: a quantitative comparison. Mol Ther, 2001; 4:383–391.

51.

Ghosh

, Yue

, Lai

, et al. A hybrid vector system expands adeno-associated viral vector packaging capacity in a transgene-independent manner. Mol Ther, 2008; 16:124–130.

52.

Trapani

, Colella

, Sommella

, et al. Effective delivery of large genes to the retina by dual AAV vectors. EMBO Mol Med, 2014; 6:194–211.

53.

McClements

, Barnard

, Singh

, et al. An AAV dual vector strategy ameliorates the Stargardt phenotype in adult Abca4(−/−) mice. Hum Gene Ther, 2019; 30:590–600.

54.

Ghosh

, Duan

. Expanding adeno-associated viral vector capacity: a tale of two vectors. Biotechnol Genet Eng Rev, 2007; 24:165–177.

55.

Trapani

. Adeno-associated viral vectors as a tool for large gene delivery to the retina. Genes (Basel), 2019; 10:pii: E287.

56.

Akil

, Dyka

, Calvet

, et al. Dual AAV-mediated gene therapy restores hearing in a DFNB9 mouse model. Proc Natl Acad Sci U S A, 2019; 116:4496–4501.

57.

Lostal

, Bartoli

, Bourg

, et al. Efficient recovery of dysferlin deficiency by dual adeno-associated vector-mediated gene transfer. Hum Mol Genet, 2010; 19:1897–1907.

58.

Lai

, Yue

, Liu

, et al. Efficient in vivo gene expression by trans-splicing adeno-associated viral vectors. Nat Biotechnol, 2005; 23:1435–1439.

59.

Vidal

, Pagliarani

, Colella

, et al. Rescue of GSDIII phenotype with gene transfer requires liver- and muscle-targeted GDE expression. Mol Ther, 2018; 26:890–901.

60.

Halbert

, Allen

, Miller

. Efficient mouse airway transduction following recombination between AAV vectors carrying parts of a larger gene. Nat Biotechnol, 2002; 20:697–701.

61.

Ghosh

, Yue

, Shin

, et al. Systemic trans-splicing adeno-associated viral delivery efficiently transduces the heart of adult mdx mouse, a model for Duchenne muscular dystrophy. Hum Gene Ther, 2009; 20:1319–1328.

62.

Chao

, Sun

, Bruce

, et al. Expression of human factor VIII by splicing between dimerized AAV vectors. Mol Ther, 2002; 5:716–722.

63.

Sun

, Li

, Xiao

. Overcoming adeno-associated virus vector size limitation through viral DNA heterodimerization. Nat Med, 2000; 6:599–602.

64.

Duan

, Yue

, Yan

, et al. A new dual-vector approach to enhance recombinant adeno-associated virus-mediated gene expression through intermolecular cis activation. Nat Med, 2000; 6:595–598.

65.

Nakai

, Storm

, Kay

. Increasing the size of rAAV-mediated expression cassettes in vivo by intermolecular joining of two complementary vectors. Nat Biotechnol, 2000; 18:527–532.

66.

Lai

, Yue

, Liu

, et al. Synthetic intron improves transduction efficiency of trans-splicing adeno-associated viral vectors. Hum Gene Ther, 2006; 17:1036–1042.

67.

Ghosh

, Yue

, Long

, et al. Efficient whole-body transduction with trans-splicing adeno-associated viral vectors. Mol Ther, 2007; 15:750–755.

68.

Reich

, Auricchio

, Hildinger

, et al. Efficient trans-splicing in the retina expands the utility of adeno-associated virus as a vector for gene therapy. Hum Gene Ther, 2003; 14:37–44.

69.

Zhang

, Yue

, Li

, et al. Dual AAV therapy ameliorates exercise-induced muscle injury and functional ischemia in murine models of Duchenne muscular dystrophy. Hum Mol Genet, 2013; 22:3720–3729.

70.

Zhang

, Duan

. Novel mini-dystrophin gene dual adeno-associated virus vectors restore neuronal nitric oxide synthase expression at the sarcolemma. Hum Gene Ther, 2012; 23:98–103.

71.

Lopes

, Boye

, Louie

, et al. Retinal gene therapy with a large MYO7A cDNA using adeno-associated virus. Gene Ther, 2013; 20:824–833.

72.

Carvalho

, Turunen

, Wassmer

, et al. Evaluating efficiencies of dual AAV approaches for retinal targeting. Front Neurosci, 2017; 11:503.

73.

Dyka

, Boye

, Chiodo

, et al. Dual adeno-associated virus vectors result in efficient in vitro and in vivo expression of an oversized gene, MYO7A. Hum Gene Ther Methods, 2014; 25:166–177.

74.

Colella

, Trapani

, Cesi

, et al. Efficient gene delivery to the cone-enriched pig retina by dual AAV vectors. Gene Ther, 2014; 21:450–456.

75.

Trapani

, Toriello

, de Simone

, et al. Improved dual AAV vectors with reduced expression of truncated proteins are safe and effective in the retina of a mouse model of Stargardt disease. Hum Mol Genet, 2015; 24:6811–6825.

76.

Koo

, Popplewell

, Athanasopoulos

, et al. Triple trans-splicing adeno-associated virus vectors capable of transferring the coding sequence for full-length dystrophin protein into dystrophic mice. Hum Gene Ther, 2014; 25:98–108.

77.

Lostal

, Kodippili

, Yue

, et al. Full-length dystrophin reconstitution with adeno-associated viral vectors. Hum Gene Ther, 2014; 25:552–562.

78.

Maddalena

, Tornabene

, Tiberi

, et al. Triple vectors expand AAV transfer capacity in the retina. Mol Ther, 2018; 26:524–541.

79.

Ghosh

, Yue

, Duan

. Viral serotype and the transgene sequence influence overlapping adeno-associated viral (AAV) vector-mediated gene transfer in skeletal muscle. J Gene Med, 2006; 8:298–305.

80.

Konarska

, Padgett

, Sharp

. Trans splicing of mRNA precursors in vitro. Cell, 1985; 42:165–171.

81.

Solnick

. Trans splicing of mRNA precursors. Cell, 1985; 42:157–164.

82.

Berger

, Maire

, Gaillard

, et al. mRNA trans-splicing in gene therapy for genetic diseases. Wiley Interdiscip Rev RNA, 2016; 7:487–498.

83.

Pergolizzi

, Ropper

, Dragos

, et al. In vivo trans-splicing of 5' and 3' segments of pre-mRNA directed by corresponding DNA sequences delivered by gene transfer. Mol Ther, 2003; 8:999–1008.

84.

Song

, Lou

, Boyer

, et al. Functional cystic fibrosis transmembrane conductance regulator expression in cystic fibrosis airway epithelial cells by AAV6.2-mediated segmental trans-splicing. Hum Gene Ther, 2009; 20:267–281.

85.

Yang

, Walsh

. Spliceosome-mediated RNA trans-splicing. Mol Ther, 2005; 12:1006–1012.

86.

, Hu

, Liu

. Protein trans-splicing by a split intein encoded in a split DnaE gene of Synechocystis sp. PCC6803. Proc Natl Acad Sci U S A, 1998; 95:9226–9231.

87.

Aranko

, Volkmann

. Protein trans-splicing as a protein ligation tool to study protein structure and function. Biomol Concepts, 2011; 2:183–198.

88.

Aranko

, Wlodawer

, Iwai

. Nature's recipe for splitting inteins. Protein Eng Des Sel, 2014; 27:263–271.

89.

Elleuche

, Poggeler

. Inteins, valuable genetic elements in molecular biology and biotechnology. Appl Microbiol Biotechnol, 2010; 87:479–489.

90.

Zhu

, Liu

, Chi

, et al. Protein trans-splicing based dual-vector delivery of the coagulation factor VIII gene. Sci China Life Sci, 2010; 53:683–689.

91.

Zhu

, Liu

, Wang

, et al. Inter-chain disulfide bond improved protein trans-splicing increases plasma coagulation activity in C57BL/6 mice following portal vein FVIII gene delivery by dual vectors. Sci China Life Sci, 2013; 56:262–267.

92.

, Sun

, Wang

, et al. Protein trans-splicing as a means for viral vector-mediated in vivo gene therapy. Hum Gene Ther, 2008; 19:958–964.

93.

Truong

, Kuhner

, Kuhn

, et al. Development of an intein-mediated split-Cas9 system for gene therapy. Nucleic Acids Res, 2015; 43:6450–6458.

94.

Chew

, Tabebordbar

, Cheng

, et al. A multifunctional AAV-CRISPR-Cas9 and its host response. Nat Methods, 2016; 13:868–874.

95.

Schmelas

, Grimm

. Split Cas9, not hairs—advancing the therapeutic index of CRISPR technology. Biotechnol J, 2018; 13:e1700432.

96.

Adli

. The CRISPR tool kit for genome editing and beyond. Nat Commun, 2018; 9:1911.

97.

Villiger

, Grisch-Chan

, Lindsay

, et al. Treatment of a metabolic liver disease by in vivo genome base editing in adult mice. Nat Med, 2018; 24:1519–1525.

98.

Moreno

, Fu

, Zhu

, et al. In situ gene therapy via AAV-CRISPR-Cas9-mediated targeted gene regulation. Mol Ther, 2018; 26:1818–1827.

99.

Tornabene

, Trapani

, Minopoli

, et al. Intein-mediated protein trans-splicing expands adeno-associated virus transfer capacity in the retina. Sci Transl Med, 2019; 11:pii: eaav4523.

100.

Lockless

, Muir

. Traceless protein splicing utilizing evolved split inteins. Proc Natl Acad Sci U S A, 2009; 106:10999–11004.

101.

Ruan

, Barry

, Yu

, et al. CRISPR/Cas9-mediated genome editing as a therapeutic approach for Leber congenital amaurosis 10. Mol Ther, 2017; 25:331–341.

102.

Aguti

, Malerba

, Zhou

. The progress of AAV-mediated gene therapy in neuromuscular disorders. Expert Opin Biol Ther, 2018; 18:681–693.

103.

Arnett

, Chamberlain

. Therapy for neuromuscular disorders. Curr Opin Genet Dev, 2009; 19:290–297.

104.

Collin

, den Hollander

, van der Velde-Visser

, et al. Antisense Oligonucleotide (AON)-based Therapy for Leber Congenital Amaurosis Caused by a Frequent Mutation in CEP290. Mol Ther Nucleic Acids, 2012; 1:e14.

105.

Slijkerman

, Vache

, Dona

, et al. Antisense oligonucleotide-based splice correction for USH2A-associated retinal degeneration caused by a frequent deep-intronic mutation. Mol Ther Nucleic Acids, 2016; 5:e381.