Adeno-Associated Virus Genome Interactions Important for Vector Production and Transduction

Abstract

Recombinant adeno-associated virus has emerged as one of the most promising gene therapy delivery vectors. Development of these vectors took advantage of key features of the wild-type adeno-associated virus (AAV), enabled by basic studies of the underlying biology and requirements for transcription, replication, and packaging of the viral genome. Each step in generating and utilizing viral vectors involves numerous molecular interactions that together determine the efficiency of vector production and gene delivery. Once delivered into the cell, interactions with host proteins will determine the fate of the viral genome, and these will impact the intended goal of gene delivery. Here, we provide an overview of known interactions of the AAV genome with viral and cellular proteins involved in its amplification, packaging, and expression. Further appreciation of how the AAV genome interacts with host factors will enhance how this simple virus can be harnessed for an array of vector purposes that benefit human health.

Introduction

The study of virus–host interactions has contributed significantly to our understanding of cell biology and physiology. Since the adeno-associated virus (AAV) is not known to be associated with any human pathology, there has been less incentive to identify drug targets to combat infections compared with other pathogenic viruses. Conversely, the minimal pathogenesis and toxicity of wild-type (wt)AAV has made it an attractive gene therapy vector. Basic knowledge of wtAAV and its interactions with the cell helped drive the development of recombinant adeno-associated virus (rAAV) vectors, and guide its production and applications.^1–4

AAV is defined as a dependo-parvovirus, as it requires the functions of a helper virus to complete the replication cycle. The first rAAV vectors were generated in the 1980s by replacing the two AAV genes with a transgene. This vector plasmid was co-transfected with a complementing plasmid, which provided the genes for AAV structural and nonstructural proteins, into cells infected with the adenovirus (AdV) helper.^5–7 Identification of the four gene products of the AdV helper virus required for AAV replication facilitated plasmid-based protocols to generate rAAV vectors. Development of these protocols, together with purification schemes, enabled the production of high-titer rAAV particles that are free of contaminating AdV and capable of highly efficient gene transfer.⁸

rAAV is powerful in its simplicity: Vector genomes contain only a short viral sequence element, and can achieve transgene expression for the life of a postmitotic target cell. rAAV vectors have demonstrated efficient gene transduction in preclinical gene therapy studies and are now approved for clinical use in the United States.³ Transduction efficiency varies depending on the target cell type and serotype capsid being utilized. rAAV vectors have been used for a plethora of therapeutic approaches, including gene addition to add new functions, gene replacement to provide a working version of a gene with disease mutations, modification of protein expression by targeting RNA, and gene editing for targeted genome modifications.⁹ A deeper understanding of the factors that interact with both viral and vector forms of the AAV genome will ultimately lead to increased efficacy of rAAV gene delivery for an array of disease indications. Here, we review AAV genome interactions that are relevant to the production of rAAV vectors and their applications.

The AAV Genome and its Products

The AAV genome is ∼4.7 kb of single-stranded DNA (ssDNA) flanked at both ends by structural elements known as the inverted terminal repeats (ITRs) (Fig. 1A). The ITRs are 145 nucleotides in length, and they can form hairpin structures by self-annealing (Fig. 1B).¹⁰ These secondary structures include the Rep binding element (RBE) and a terminal resolution site (TRS), which together constitute the AAV origin of replication. The ITRs are also required as packaging signals for genome encapsidation after replication.

Figure 1.

The AAV genome and its products. (A) Diagram of the wtAAV genome, illustrating the layout of the rep (orange) and cap (green) genes, viral promoters (p5, p19, p40), polyadenylation signal (pA, gray), and ITR sequences. All elements are drawn to scale. (B) Magnified view of the ITR (boxed region in A), detailing the complementary A/A′ (blue), B/B′ (green), and C/C′ (red) sequences. The D-sequence (brown) is contiguous with the ssDNA genome body. The terminal resolution site (TRS) consists of the cleavage site (black arrow) and the flanking bold bases. (C) The nine known protein products of the AAV genome. Transcripts (gray lines) are aligned with their respective promoters in (A). Boxes labeled with protein names constitute ORFs translated from the rep (orange) and cap (green) genes. +1 indicates these protein products are translated from a +1 nucleotide shifted ORF within the cap gene. The Rep enzymatic domains relevant to interactions with the AAV genome are indicated above. (D) Schematic of an rAAV vector genome, illustrating the maximum cassette size that can be packaged of 5 kb (drawn to scale with A). The CMV promoter (pink) is shown as an example, with the blue bar representing available space for the gene of interest. CMV, cytomegalovirus; ITR, inverted terminal repeat; ORF, open reading frames; pA, polyadenylation signal; rAAV, recombinant adeno-associated virus; RBE, Rep-binding element; ssDNA, single-stranded DNA; TRS, terminal resolution site; wt, wild-type.

The AAV genome is highly compact, with complex overlapping coding regions, alternate splicing schemes, and multiple translation initiation sites from canonical and noncanonical start codons. Similar to other parvoviral genomes, the AAV genome harbors two main genes, rep and cap, which encode nine distinct protein products (Fig. 1C). Transcription of the rep gene is initiated from the p5 or p19 promoters to, respectively, produce the large (Rep78 and Rep68) and small (Rep52 and Rep40) nonstructural Rep proteins. The different C-terminal regions of the Rep proteins are generated by alternate splicing. The N-terminal domain unique to Rep78/68 contains the DNA-binding domain that recognizes specific sequences within the ITR.^11–13 All four Rep proteins have common helicase and ATPase domains that function in genome replication and/or encapsidation.^14,15

Transcription of the cap gene is initiated from a single promoter termed p40. The cap minor splice product expresses the viral protein (VP)1 structural protein, whereas the major splice product expresses the VP2 and VP3 structural proteins. This region also encodes the nonstructural Assembly-Activating Protein (AAP), which performs multiple functions related to capsid assembly.^16–19 AAP is translated from a +1 nucleotide shifted reading frame from that of the VP proteins. Recently, an additional protein product has been identified in this +1 ORF (open reading frame), the Membrane-Associated Accessory Protein (MAAP).²⁰ The precise function of MAAP remains to be elucidated, but it is presumably associated with the production phases of the replication cycle.

Construction of an rAAV vector genome requires only that the transgene sequence is flanked by the ITRs,²¹ and that the entire length of the vector genome does not exceed the upper packaging limit of 5 kb.^22,23 A typical transgene cassette includes a relatively small promoter, for example, the cytomegalovirus immediate-early enhancer and promoter, and a cDNA version of the gene of interest to minimize the transgene size (Fig. 1D). Since the viral genes have been removed from AAV vector constructs, the ITRs are the only viral genetic element that will persist within the cell in long-term transduction with rAAV vectors. Occupying less than 6% of the vector genome length, the short ITRs are powerful for their size, performing an impressively large number of functions,¹⁰ many of which are discussed herein.

Genome Interactions Important for AAV Transcription and Replication

Replication for any DNA virus involves transcription of viral genes to generate proteins that then aid in amplification of the viral DNA genome. When cells are infected with the wtAAV virus, the first step in productive replication is conversion of the released ssDNA genome into a double-stranded DNA (dsDNA) template for transcription (Fig. 2A, step 4). The host factors that govern this step for wtAAV are similar to those that facilitate this process during transduction by rAAV vectors (Fig. 2B). The viral ITR structure provides a primer for DNA synthesis by host polymerases that extend that strand to the end of the viral genome. When plasmids are transfected for rAAV production, transcription can be initiated directly from this dsDNA. In both cases, AdV helper proteins are required to activate the AAV p5 and p19 promoters, which drive expression of the Rep78/68 and Rep52/40 proteins, respectively. The Rep proteins themselves are involved in both negative and positive regulation of viral transcription.²⁴ Rep interactions with the Sp1 transcription factor are known to mediate gene activation.²⁵ There is also evidence to suggest that the ITR can act as both an enhancer and an initiator for transcription,^26,27 implying that it is recognized by cellular proteins that regulate transcription.

Figure 2.

The cellular path of AAV genome fates and interactions. Schematic representation of (A) the wtAAV infection pathway and (B) the decoupling of this pathway in rAAV vector production and transduction. The AAV gene products are denoted in colors corresponding to Fig. 1C. Major steps of the pathway are enumerated in (A) as follows: (1) Entry. (2) Trafficking. (3) Uncoating. (4) Second strand synthesis. (5) Transcriptional activation. (6) Genome replication. (7) Packaging. Diagrams are not to scale and are not exhaustive of the processes involved. For example, transgene expression occurs from monomeric vector genomes in addition to concatemers. (C) Summary of host and helper viral proteins that interact with the AAV genome, arranged by the processes and phases with which they have been identified. Green text indicates factors involved in positive regulation, and red text denotes negative regulators of viral replication or vector transduction. The production phase of this panel employs similar factors as in replication, and thus they were omitted from the production diagram for simplicity; there are no known factors that interact with the AAV genome exclusively in the vector production context. *Factors that interact with the AAV genome in an AdV-dependent manner; ^†factors that bind the AAV genome as part of a complex with the Rep proteins. AdV, adenovirus.

The model of AAV DNA replication consists of unidirectional strand-displacement replication in which the Rep proteins play a central role. The ITRs contain motifs that serve as the viral origin of replication for this process (Fig. 1B). Through its self-complementary sequence and resulting secondary structure, the ITR provides a base-paired 3′ hydroxyl group for unidirectional DNA synthesis mediated by host replication machinery. The larger two Rep proteins (Rep78/68) bind specifically to dsDNA containing a repeating motif of GCTC within the ITR referred to as the Rep binding element (RBE) or site (RBS).²⁸ The Rep protein functions on the viral genome as an oligomer, with multiple regions required for multimerization and enzymatic activities.^29–31 The ITRs are faithfully replicated through a process termed terminal resolution,³² whereby the endonuclease activity in the N-terminus of the Rep78/68 protein³³ generates a nick at the TRS. This nick provides the necessary 3′ hydroxyl group for replication through the ITR. The TRS sequence is embedded within a dsDNA region of the ITR, and the Rep helicase activity is therefore necessary to generate an accessible ssDNA substrate for Rep endonuclease activity.³⁴ Beyond the tight hairpin of the ITR is the D-sequence, which may have a role in the replication and packaging of vector genomes.³⁵

When producing rAAV, the factors involved are similar to those active in wtAAV replication. The key difference is that the AAV and helper viral genes are introduced ectopically into the cell, for example by triple-plasmid transient transfection (Fig. 2B). In this scenario, the viral elements are transcribed from two plasmids, and replication of the ITR-flanked genome occurs from an ITR-containing third plasmid. Once expressed, the viral elements interact with host factors to proceed through the remaining steps of the replication cycle, producing rAAV particles containing the replicated ssDNA vector genome. The relative abundance of each component provided in the vector scenario can affect production outcomes. For example, excess levels of Rep78/68 resulted in significantly lower rAAV yields,³⁶ which, in hindsight, can be explained by later studies elucidating Rep's roles in transcriptional regulation at the AAV p40 promoter. Studies on basic AAV biology will, undoubtedly, continue to inform advances for the field of gene therapy.

Contributions of cellular protein interactions

In addition to the viral Rep proteins, amplification of the viral genome involves assembly of host replication factors at the origin of replication. Several in vitro replication assays demonstrated replication of AAV templates using cell extracts and purified proteins.^37,38 In cell-free assays, bacterially expressed Rep68 is capable of mediating AAV DNA replication in an AdV-independent manner,^38,39 suggesting that once Rep has been produced, the cell contains all factors required. Although the natural host species of AAVs are limited to vertebrates, it has been demonstrated that production of rAAV can occur in both mammalian and insect cells,⁴⁰ and possibly (although inefficiently) in yeast.^41,42 This implies that the host cellular factors required for AAV replication are conserved among eukaryotes.

Many cellular factors interact with the AAV genome either directly or indirectly via viral Rep proteins. Genetic screens have provided a powerful approach to identify relevant cellular factors. A one-hybrid screen designed to identify cellular proteins that recognize the RBE within the AAV ITR identified cellular zinc-finger 5 protein (ZF5) as a negative regulator of AAV replication.⁴³ Ectopic expression of ZF5 led to ITR-dependent repression of the p5 promoter, as well as reduction in both replication of wtAAV and production of rAAV vectors.⁴³ Rep interactions also impact the cellular proteins associated with the viral genome. Genetic two-hybrid screens in yeast identified the positive coactivator 4 (PC4) protein as a Rep-interacting protein.^44,45 Over-expression of PC4 downregulated AAV promoters in the absence of helper virus but enhanced AAV replication in the presence of AdV.^44,45 A proteomic screen of all four Rep proteins found that Rep binds to the protein phosphatase PP1 to enable phosphorylation of KAP1 and release of transcriptional repression.⁴⁶ Additional cellular proteins can bind to Rep and enhance its activities. For example, the high mobility group protein 1 stimulates Rep binding, TRS nicking, ATPase activity, and repression of the p5 promoter.⁴⁷ Affinity purification of proteins associated with Rep has identified cellular proteins with roles in replication, repair, and RNA processing,^48,49 although it is unclear how many of these interactions occur with Rep bound to the viral genome.

The AAV DNA replication has been reconstituted in vitro by using purified replication factor C, replication protein A (RPA), and proliferating cell nuclear antigen (PCNA), as well as a fraction of cell extract containing polymerase activity and other unidentified factors.^37,50 Sensitivity to polymerase inhibitors, as well as the requirement for PCNA, suggests that either polymerase delta or eta are being used by AAV in this system. Involvement of these proteins in AAV replication at the cellular level has not been established, although RPA is also found at viral replication centers.⁵¹ One common feature of the different conditions under which AAV replicates is the enhancing role of ssDNA-binding proteins: AdV-DBP from AdV, ICP8 from Herpes Simplex Virus (HSV), and RPA from the cell.^51,52 These three proteins are also found at viral replication centers, interact with the larger Rep proteins, and enhance their activities.⁵¹

Contributions of helper viruses

Helper viruses impact defined steps that enhance AAV replication.⁵³ The AdV E1a protein activates AAV promoters by relieving repression of the p5 promoter by the cellular repressor YY1, and by recruiting the p300 histone acetyltransferase.⁵⁴ A complex of E1B55K and E4orf6 proteins aids AAV replication and may indirectly facilitate cytoplasmic accumulation of AAV mRNAs.⁵⁵ The AdV virus associated RNA VAI stimulates translation of AdV RNAs and appears to do the same for AAV.⁵⁶ The AdV-DBP is directly required for increasing polymerase processivity during AAV replication.^51,57 The set of helper functions supplied by HSV is a very different group of proteins from that used by AdV. The minimal HSV proteins that support AAV replication in cell culture are the helicase–primase complex of UL5, UL8, and UL52, together with the UL29 gene product ICP8.⁵⁸ This cohort induces AAV replication in the absence of a viral polymerase, indicating that these helper proteins facilitate recruitment of cellular replication machinery. Expression of these four HSV proteins alone results in formation of a complex at discrete subnuclear sites, and AAV Rep protein co-localizes at these centers.⁵¹ The HSV ICP8 protein alone can also enhance AAV DNA replication in an in vitro assay.⁵⁹ The enzymatic functions of the helicase–primase complex do not appear to be required; their role may be merely to ensure that ICP8 protein (and possibly other cellular factors) is in the correct subnuclear location for AAV DNA. The HSV polymerase UL30 may also contribute to AAV replication as shown in cells⁶⁰ and in an in vitro reconstituted system of purified HSV replication proteins.⁵⁹

Although the sets of functions provided by helper viruses clearly differ, one of the shared features of the AAV helper viruses may be the establishment of nuclear viral replication centers and recruitment of cellular factors to aid AAV replication.^49,51,61 Viral infections are subject to spatial regulation, with the formation of specific nuclear structures where viral transcription and replication occur.⁶² Productive replication of AAV relies on helper virus functions to establish these replication compartments.

Genome Interactions Involved in Packaging

The term “packaging” colloquially refers to the production of virus or vector in a generic sense, but in the case of AAV it more specifically refers to the act of genome encapsidation. This step in virion assembly entails insertion of the ssDNA genome into a preassembled capsid,⁶³ in contrast to some other viruses in which capsomers assemble around the viral genome as a scaffold. Packaging is an imperfect process and can lead to appreciable proportions of empty capsids in vector preparations.⁶⁴ Improving genome packaging efficiency is of high interest for vector production efforts. Known interactions with the AAV genome that mediate the packaging reaction are limited, but this may reflect a minimal number of interactions that are actually involved.

Interactions of the AAV Rep proteins have a leading role in packaging.³⁵ During genome replication, specifically after terminal resolution, Rep78/68 remains covalently attached to the nascent ssDNA genome⁶⁵ and docks on the five-fold (5F) axis of the assembled capsid (Fig. 2A, step 7).⁶⁶ The Rep52/40 proteins are essential for packaging and accumulation of ssDNA progeny genomes.⁶⁷ They also bind the capsid with reasonably high affinity,⁶⁸ and their DNA helicase activity (but not that of Rep78/68) is required to “pump” the genome into the capsid through a channel formed at the 5F axis of symmetry.¹⁵ The interaction of Rep78/68 with the genome is RBE-specific, whereas Rep40 can bind and unwind nonspecific dsDNA substrates.^69,70 These properties allow virtually any vector sequence flanked by ITRs to be packaged by the Rep proteins. Intriguingly, although the Rep helicase activity can unwind the AAV ITR secondary structure during genome replication and packaging, it appears to be less efficient with non-ITR secondary structures located within the genome body. This is reflected by high proportions of truncated genomes packaged when the vector genome contains repetitive or self-complementary regions, such as shRNA expression cassettes.⁷¹ In addition to the ITR's role as the origin of replication, the D-sequence of the ITR, which is not involved in the hairpin secondary structure, is required for packaging.⁷² Mechanisms underlying this role remain unknown. There is some evidence for a cellular protein that binds the D-sequence,^72–74 but it appears to function at the level of transduction, and has not been implicated in packaging.

Naturally, the AAV genome must interact with the capsid into which it is packaged. Consistent with this, mutations at the rim of the 5F pore disrupt the ability for Rep to dock and mitigate genome insertion. Mutations lining the channel itself produce capsids that bind Rep but have a significant reduction in encapsidated genomes.^66,75 The nature of the packaging defect for these mutants has not been elucidated, but most of these mutations constitute a polar to nonpolar change in amino acid identity. These polarity mutations likely disrupt the shape or charge of the 5F pore such that the ssDNA genome is sterically or electrostatically inhibited from entering. Specific residues on the lumenal surface of the capsid are known to make contact with the bases and backbone of the ssDNA genome in a sequence-independent manner.⁷⁶ Once the genome is packaged, it likely exerts a reasonable amount of force on the lumenal capsid surface.⁷⁷ Capsid mutations at the VP–VP interface that permit assembly but cannot be packaged⁶⁶ are believed to be defective due to the structural integrity required to withstand internal pressure from the genome.

Spatiotemporal control of packaging

Spatiotemporal coordination of packaging has only been studied for the prototypical serotype AAV2.⁷⁸ The large Rep proteins and accumulating AAV genomes colocalize in the nucleoplasmic replication centers preceding cap expression. Once capsids assemble in nucleoli, they are redistributed to the nucleoplasm and eventually the cytoplasm. This redistribution occurs independently of replicated AAV genomes, yet depends on Rep expression, at least for AAV2. The Rep proteins can be detected in nucleoli but only after assembled capsids are detected. At this time, Rep52/40 proteins take on a punctate distribution enriched in and around nucleoli, whereas Rep78/68 appear more uniform throughout the nucleoplasm. One way to interpret these observations collectively is that Rep78/68 shuttles a covalently linked nascent genome to the nucleolar site of capsid assembly, where capsids are accumulating and effectively shielding each other's 5F pores. Capsids on the nucleolar “surface” thus have only one or two 5F pores sterically available, where small and large Rep proteins can dock and begin genome insertion. This interaction is of reasonably high affinity, and genome/Rep/capsid complexes mobilize to the nucleoplasm as packaging progresses. This hypothetical model would, however, only apply to serotypes that assemble in nucleoli.

Considerations of host factors for packaging

There are no specific host factors identified to date to be involved in packaging. Since reconstitution of genome packaging in cell-free assays requires addition of cell lysate,⁷⁹ there must be a role for some host proteins. Successful genome encapsidation seems to require active replication of the ssDNA genome,⁷⁹ and therefore host factors may be more indirectly involved through active replication rather than directly in packaging. Also unclear is whether strand displacement is achieved solely by host polymerase or whether the small Rep proteins aid in this concomitantly with packaging. Perhaps the largest curiosity regarding packaging mechanisms is that in a typical infection or vector production scenario, the majority of packaged genomes are full-length from ITR to ITR with few truncated products. Since only one genome with a size <5 kb can fit inside a capsid, only a single packaging event can thus occur per capsid; initiation of multiple packaging reactions per capsid would result in multiple truncated and partially encapsidated genomes per virion. The icosahedral symmetry of the capsid forms twelve 5F pores that are presumably identical, yet Rep docks and packages into only one pore. This suggests a model where competitive exclusion of Rep/genome complexes is involved, perhaps mediated by host factors or via steric hindrance as suggested in the previous section.

Genome Interactions that Impact Vector Transduction

The rAAV particle is relatively simple yet is capable of cell entry and nuclear delivery of foreign genetic material that can be stably expressed for the lifetime of that cell. This is achieved by a nonenveloped capsid of only 25 nm diameter with a single known enzymatic activity,⁸⁰ whose sole cargo is the ssDNA genome. This simplicity and the absence of AAV or helper virus during vector transduction implies that the host cell performs myriad functions in the intricate process of transduction. We will limit this discussion to direct interactions with the vector genome itself that occur during transduction. Other host interactions mediating transduction have been extensively reviewed elsewhere.^81–85

The vector genome is released after nuclear entry, although mechanisms involved in uncoating inside cells are entirely unknown (Fig. 2A, B). Biophysical features of AAV capsid disruption and genome release have been studied in vitro,⁷⁷ but interactions that are important for AAV genome uncoating in the host cell have not yet been defined. The kinetics of uncoating vary by capsid serotype,⁸⁶ and the uncoating process may be rate-limiting for transduction in some settings.⁸⁷ Once the ssDNA genome is exposed, the ITR secondary structure allows for self-priming and second-strand synthesis by host polymerases, in the same manner that occurs during wtAAV replication. To bypass this step, self-complementary vectors (scAAV) were developed that self-anneal throughout the full transgene length once uncoated, forming a dsDNA template that is transcription-ready.^88,89

In contrast to wtAAV, which relies on a helper virus and Rep for expression from viral promoters, the expression cassette in vector genomes is active in the native host environment. In transgene design, the choice of promoter can include those from other viruses when strong and ubiquitous transgene expression is desired, or promoters that are mammalian in origin and cell type specific. In addition, the ITRs of many AAV serotypes are known to have low-level promoter activity that allows for maximal capacity for the transgene.^26,27,90

Vector promoters presumably recruit the canonical mammalian transcription machinery (transcription factors, the Preinitiation Complex, etc.), although this has not yet been specifically investigated. Recently, the E3 ubiquitin ligase RNF121 was demonstrated to be essential for transcription from rAAV genomes.⁹¹ Other than a decrease in active RNAPII association with vector genomes in the absence of RNF121, the underlying mechanisms of the RNF121 requirement for vector expression remain to be elucidated. Intriguingly, these studies demonstrated that RNF121 is only required for expression from vector-delivered AAV genomes: Pure rAAV genomes extracted from capsids and transfected into RNF121-deficient cells showed no defect in transgene expression.⁹¹ In addition, there are studies that implicate the capsid proteins as an in cis requirement for transcription from rAAV genomes. Capsid mutants near the two-fold axis of symmetry exhibit a strange phenotype, whereby they are capable of assembly, packaging, cell entry, nuclear entry, and uncoating but show a marked decrease in transcription from the delivered vector genomes.^92,93 A functional, direct interaction between the VP proteins and the genome that they uncoat has not yet been demonstrated, but this transcriptional phenotype suggests interesting mechanisms that are unique to AAV biology, whereby capsid proteins act in cis to regulate expression of their genetic cargo.

As transduction progresses, the majority of monomeric vector genomes, now dsDNA, recombine into dimers and larger concatemers through recombination of their ITRs.⁹⁴ These concatemers increase in molecular weight over time as more vector genomes recombine.⁹⁵ Presumably, host enzymes catalyze the intermolecular ligation of vector genomes, but little is known about the mechanisms and factors involved in concatemerization and persistence. Concatemers can persist for the life of the cell as transcriptionally active episomes⁹⁶ or can be silenced over time,⁹⁷ which implies that the host cell epigenetically regulates vector genomes. Concatemeric episomes with nucleosomal structure have been observed in primate skeletal muscle up to several years after rAAV treatment.⁹⁸

The association of wtAAV or rAAV vector genomes with histones allows subterfuge from host defenses that recognize naked DNA, yet it conversely provides a substrate for negative or positive epigenetic regulation. For example, KAP1 binds and silences wtAAV2 viral genomes by recruiting histone modifying enzymes such as the SETDB1 methyltransferase and the NuRD deacetylase complex.⁴⁶ Accordingly, AAV2 viral genomes are associated with trimethylated H3K9, a mark of transcriptional silencing.⁴⁶ In one study, ChIP by an anti-acetyl histone H3 antibody precipitated vector genomes when cells were treated with HDAC inhibitors,⁹⁹ suggesting that active chromatin remodeling occurs on vector genomes. A mark of active genes, H3K27Ac has been shown to be associated with vector genomes,⁹¹ but a few other studies have examined rAAV vector chromatin signatures. It is also not clear how many episomes might be retained in each cell, and within the concatemer how many of the transgenes are expressed at any time. Understanding how the cell processes vector genomes can inform gene therapy strategies that promote ideal chromatin signature for fine-tuned expression in target tissues. In addition, an understanding of rAAV genome processing could greatly enhance the trans-splicing approach,^100,101 which relies on unidirectional and stoichiometric concatemerization to overcome AAV's small genome capacity.

At present, interactions with vector genomes postdelivery is an understudied area of AAV vector biology. Since the only feature common to all vectors is the ITR, cellular proteins that recognize these unusual DNA structures are likely to impact genome fate and transduction efficiency. The first cellular protein reported to bind the ITR was a ssDNA-binding protein that recognizes the D-sequence (ssD-BP) and was subsequently identified as the chaperone-associated protein FKBP52.⁷³ It is unclear which functions of this protein are relevant to its role in the AAV lifecycle, but it was suggested that phosphorylated FKBP52 binds to the D-sequence and blocks second-strand synthesis.^73,74 The phosphorylation status of the protein correlates with rAAV transduction efficiency and treatments that enhance transduction.^73,74 Transcription factors belonging to the RFX protein family were also identified as host factors that bind the D-sequence and regulate rAAV transgene expression.¹⁰² It is likely that other proteins binding to elements of the ITR could also impact transgene expression from rAAV vectors.

Considerations of Cellular Innate and Intrinsic Defenses

Immune responses to AAV vectors can significantly affect the safety and efficacy of rAAV-mediated gene therapy. Most studies on immune responses to AAV focus on the capsid, which is the primary interface between the therapeutic transgene and the patient and can illicit acute immune responses. As the persistent therapeutic entity, cellular responses against the genome itself are also undesirable. These are less concerning for safety and toxicity, but they can impact the establishment and longevity of transgene expression. Innate immune responses directed toward wtAAV or rAAV genomes have been observed. TLR9-mediated recognition of the vector genome triggers type I interferon production irrespective of transgene sequence.¹⁰³ This response is reported to be substantially greater for scAAV genomes,¹⁰⁴ although the canonical ligand for TLR9 activation is ssDNA.¹⁰⁵

Similar to innate immune sensing of potentially pathogenic DNA, eukaryotic cells possess machinery that monitors for DNA damage and breaks, resulting in activation of signaling checkpoints that enable repair. Components of the cellular DNA repair network have been implicated in the response to rAAV vectors and processing of genome outcomes. The AAV ITRs have been suggested to elicit a cellular DNA damage response involving signaling by cellular kinases.^106,107 Since rAAV vector transduction is higher in cells deficient for the ATM kinase,¹⁰⁸ the cellular DNA damage response may limit AAV infection. The Mre11/Rad50/Nbs1 (MRN) complex that recognizes dsDNA breaks and free ends during DNA repair also senses AAV ITRs.¹⁰⁹ This was originally revealed by discoveries of the substrates for the AdV E1B55K/E4orf6 complex, which promotes wtAAV replication and rAAV vector transduction.^110,111 The AdV E1B55K/E4orf6 proteins were found to induce degradation of the MRN complex, which impacts DNA damage signaling.¹¹² Components of the MRN complex colocalize with AAV genome foci, but these host proteins impede concatemer growth and transgene expression¹¹³ as well as wtAAV viral replication.¹⁰⁹ Although originally suggested to impact second-strand synthesis for rAAV, the direct binding of MRN to the ITR structure may be more relevant for inhibiting wtAAV replication or rAAV transduction in a way that is independent of its signaling or enzymatic activities.¹¹⁴ Deleting parts of the ITR may reduce the inhibition of transgene expression by MRN and ATM, although vector productivity is also impacted.¹¹⁵ The MRN complex may also play a role in AAV integration into the host genome.¹¹⁶ Other cellular repair factors and pathways have been implicated in the response to AAV ITRs.^117–120 The cellular proteins Ku86 and Rad52 are proteins involved in repair of DNA breaks, and they have been suggested to bind directly to AAV ITRs.^118,121 rAAV transduction is increased in Ku86-defective cells, whereas it is inhibited in Rad52 knockout cells.¹¹⁸ The DNA-PKcs and Ku proteins are required for AAV replication, and their role in nonhomologous end-joining repair may lead to rAAV genome concatemerization.^120,121 Rad52 is involved in cellular break-induced replication and has been proposed to fulfill an analogous role in AAV replication.¹²² The relative binding of these factors to ITRs may determine the pathway used for processing of AAV genomes.

Interactions for AAV Genome Persistence, Integration, and Gene Targeting

One intriguing feature of the wtAAV virus is the observation that infection can result in latent infections.¹²³ Both wtAAV and rAAV can persist as extrachromosomal episomes detected as circular tandem head-to-tail concatemers. High levels of Rep expression are toxic to the host cell,^124,125 which necessitates that gene expression is minimized from persistent wtAAV genomes. In contrast, robust long-term transgene expression from rAAV vectors in vivo suggested that at least some of the episomal forms remain transcriptionally active.^126,127 In the absence of productive replication, the viral DNA can also become integrated into the cellular genome. At least in dividing cultured cells, wtAAV can be found integrated in tandem arrays of several genome equivalents in a head-to-tail configuration.¹²⁸ The wtAAV remains stably integrated in these cells with minimal expression of Rep proteins, but it can be activated by helper virus infection. The precise mechanism by which the integrated genome is rescued to allow productive replication is not completely understood. Although integration was initially believed to occur randomly, it was demonstrated that wtAAV is capable of site-specific integration into a region of the host genome on human chromosome 19.¹²⁹ Integration is targeted within a region of several hundred nucleotides of this locus, which was called AAVS1.¹³⁰ Vectors lacking the rep gene no longer integrate in a site-specific manner, suggesting a role for Rep proteins in targeting integration. The large Rep proteins (Rep68/78) are sufficient to target integration of a DNA sequence flanked by AAV viral ITRs.^131,132 The basis for Rep-mediated targeting of integration was revealed by discovery of Rep binding and nicking sites within the AAVS1 integration locus.¹³³ The DNA requirement for these elements was further confirmed by transfection and episomal recombination,^133,134 and the only viral genetic element required in cis is the ITR. A complex is formed between the viral ITR and the human sequence by Rep proteins that tether the DNA molecules for recombination.¹³⁵

Standard rAAV vectors do not retain the rep gene, and therefore they lack the targeted integration capacity of wtAAV.⁹⁶ Their integration frequency is considered to be very low, especially in tissue comprised of nondividing cells. The rare event of vector integration is usually not associated with toxicity, although there have been some suggestions of associations with oncogenic events in animal models.¹³⁶

The ssDNA molecule delivered into the nucleus by standard rAAV vectors can serve as a template for homologous recombination and gene targeting.^137–139 These gene targeting recombination approaches have also been demonstrated to work in vivo in some organs and can be used for gene addition.^140,141 The frequency of rAAV-induced recombination can be further augmented by introducing a dsDNA break at the targeting site.^142,143 Site-specific endonucleases have been combined with rAAV donor templates to achieve precise engineering of the human genome.¹⁴⁴ Features that could contribute to the efficient recombination rates with rAAV vectors include the ssDNA nature of the genome, which is directly available for homologous recombination, and potentially the recruitment of cellular repair factors by the ITRs.¹⁴⁵ As more is learned about the cellular proteins involved in mediating rAAV-induced gene targeting, it is possible that specific genetic elements could be incorporated into rAAV vector genomes to further enhance the efficiency. The tethering concept of the Rep protein could also be harnessed as a way of promoting rAAV genome interactions with the host genome.

Concluding Thoughts

The simplicity of the AAV genome requires strategies to exploit the cellular machinery to ensure productive infection and co-existence with the host. Every step of the virus cycle is governed by molecular interactions between the virus and the host. Determining the molecular interactions that take place during infection is important for understanding both the infectious cycle and the behavior of vectors. A graphical summary of these interactions as discussed in this review is presented in Fig. 2C.

Although our understanding of the natural biology of AAVs is somewhat limited, vector development has spurred studies into interactions with the host. Viral gene therapy vectors take advantage of the innate ability of viruses to transfer genetic material to the nucleus of the host cell for endogenous gene expression. A thorough understanding of the molecular interactions that take place during transduction will aid the development and optimization of these promising gene transfer vectors. Recombinant viral vectors have, in turn, provided powerful tools for studying the molecular interactions that characterize wtAAV infection. There have been a number of high-throughput screens designed to identify factors that bind AAV capsids or impact rAAV transduction.^146–150 These have uncovered host factors that limit entry and transduction, but they have mainly not identified factors that interact directly with the genome. Proteomic approaches with labeled genomes have identified cellular proteins on the genomes of the AdV and HSV helper viruses,^151,152 and it would be interesting to see how these are impacted by AAV co-infection and whether similar approaches could identify cellular proteins on AAV and rAAV vector genomes.

Addressing key questions about recognition and processing of the AAV genome will improve their potential as gene therapy vectors. It is unclear where in the nucleus uncoating takes place and whether this impacts genome fate. It is also unclear to what extent different vector tropism and intracellular trafficking pathways affect genome interactions and how these impact transduction efficiencies. Further dissection of factors that control infectivity will allow lower vector doses to be used, increasing patient safety, whereas an investigation of replication and packaging mechanisms will support increased vector production efficiencies and reduce drug costs. Understanding interactions with the genome and identification of endogenous cellular regulators of the AAV lifecycle is also important for ensuring efficient long-term transduction. It will be crucial to determine the consequences of introducing foreign genetic material into the host cell, whether it be damage responses to ITRs or genetic rearrangements resulting from integration into the host genome. Understanding basic biology will continue to propel advances in AAV gene therapy applications.

Author Disclosure

No competing financial interests exist.

Funding Information

Work on AAV and helper proteins in the Weitzman laboratory has been supported by National Institutes of Health grants (AI067952, CA097093, AI051686).