Understanding capsid assembly and genome packaging for adeno-associated viruses

Abstract

Adeno-associated viruses (AAVs) are promising therapeutic viral vectors. Their capsid is assembled from viral proteins VP1, VP2 and VP3, aided by an assembly-activating protein, followed by replication protein mediated packaging of their 4.7-kb genome with inverted terminal repeats as packaging signals. To aid improvement of AAV vectors, knowledge of viral determinants of successful capsid assembly and genome packaging is important. We review the current knowledge of these two processes and efforts to overcome limited DNA packaging capacity and limit the packaging of unwanted foreign DNA in vector development. Residues involved in essential capsid assembly and genome packaging interactions cannot be manipulated in vector engineering. This information thus aids strategies to improve vector production and to increase AAV packaging capacity toward improved efficacy of this vector system.

EXECUTIVE SUMMARY

Capsid assembly

• The nucletotide sequence of the cap gene contains all the information required to express three structural proteins which when assembled have the ability to perform multiple functions in the virus replicative life cycle.

• The viral protein (VP) 3 common region is sufficient to assemble the virus capsid in the presence of the assembly-activating protein.

• The VP twofold and fivefold interactions are important in the assembly pathway, and their disruption should be avoided when generating new vectors.

• Capsid assembly occurs in the nucleoli, assembly-activating protein is required for this process (except for AAV4, AAV5 and AAV11) and it can be supplied in trans during vector production.

Genome packaging

• AAV genomes flanked by inverted terminal repeats are packaged as ssDNA into empty preformed capsids.

• The packaging mechanism couples the replication and encapsidation process.

• The small Rep proteins play a central role in the translocation of the genome into the capsid.

• The limit of packageable recombinant AAV genomes size is approximately 5.2 kb.

• Single oversized AAV vectors and dual AAV vectors allow transfer of genes exceeding the AAV packaging capacity.

• Self-complementary AAV vectors allow a fast onset of transgene expression due to the packaging of both complementary strands in one capsid.

• Packaging signal homologs should be avoided in any construct except for the transgene cassette to prevent collateral packaging of nonvector DNA into AAV capsids.

The adeno-associated viruses (AAVs) belong to the Parvoviridae family. Currently, 13 human and nonhuman primate AAV serotypes and over a 150 genotypes have been described [1]. Their small, nonenveloped capsids, of approximately 260 Å in diameter, package a linear ssDNA genome of approximately 4.7 kb. Both ends of the AAV genome contain identical inverted terminal repeats (ITRs) of approximately 150 nucleotides (NTs) that form T-shaped hairpin secondary structures (Figure 1). The ITRs are sectioned into four regions that contain cis-elements, including the Rep-binding element (RBE) and the terminal resolution site (trs), which are indispensable for genome replication and packaging. The ITRs flank two large open reading frames (ORFs) which code for a series of replication (Rep) proteins and viral proteins (VPs), and the assembly-activating protein (AAP; Figure 1). The multifunctional Rep proteins play important roles in different aspects of the viral replication cycle [2]. All four of the AAV Rep proteins (Rep78, Rep68, Rep52 and Rep40; Figure 1) share a 305 amino acid (aa) stretch that contains a helicase/ATPase domain [3] and nuclear localization signals (NLS) [4]. The C-terminus of Rep68/40 differs from those of Rep78/52 due to differential splicing (Figure 1). The larger Rep78/68 proteins are extended at their N-terminus in comparison to Rep52/40 (Figure 1) [5]. Their N-terminal 224-aa contains a DNA-binding and endonuclease domain [6]. While the large Rep proteins, Rep78/68 are indispensable for genome replication, the small Rep proteins, Rep52/40 are essential for genome packaging. The structural/viral proteins VP1, VP2 and VP3 are encoded by the second ORF, they form the viral capsid, and all share a common C-terminal sequence (Figure 1). Similar to the Rep proteins the VPs are also generated by differential splicing and the utilization of alternate start codons. VP1 is encoded within the entire cap ORF, it is approximately 735 aa in length, 87 kDa in molecular weight (MW) and contains a phospholipase A2 (PLA2) site that is important for endo/lysosomal escape during cellular trafficking and viral infectivity in its unique N-terminus (VP1u) [7]. The VP2 sequence is encoded by alternative splicing, overlaps with the C-terminal region of VP1, it is approximately 598 aa in length and 73 kDa in MW. VP3 is the most abundant VP, it is approximately 533 aa in length, 61 kDa in MW and its sequence is contained within VP1 and VP2. The viral capsid is assembled from the overlapping VP3 region [8]. The virus capsid is a critical component of its infectious lifecycle: it protects the viral genome, determines the tissue type that is targeted (tissue tropism) or treated, and remains intact while it escapes the endo/lysosomal pathway and traffics to the nucleus where the genome is released [9] for another round of replication. For productive replication, the AAVs depend on the coinfection with helper viruses, such as, adenoviruses or herpesviruses [2]. The functions that these helper viruses provide include transcription factors for the AAV promoters, replication factors and factors that modify the host cell environment for efficient AAV transcription, translation and genome replication [2].

Figure 1. Genome organization of adeno-associated virus.

An overview of the AAV2 genome is shown with its two open reading frames flanked by ITRs. The AAV transcripts are depicted below with their translation products. The upper part illustrates the secondary structure of the left ITR with the Rep-binding site, the adjacent terminal resolution site and the D-sequence. AAP: Assembly-activating protein; AAV: Adeno-associated virus; ITR: Inverted terminal repeat; RBE: Rep-binding element; trs: Terminal resolution site; VP: Viral protein.

The AAVs have emerged as attractive gene delivery vectors. In recombinant AAV vectors (rAAVs), a desired transgene expression cassette is packaged into the capsids instead of the wild-type (WT) viral genome [10]. These vectors can be used to transduce a desired target tissue to achieve long-term expression of this transgene in postmitotic cells with no known pathogenicity. To construct rAAVs, only the flanking cis-active ITRs have to be retained for the packaging of the transgene cassette while the rep and cap genes can be supplied in trans [10]. For the majority of rAAV vectors, the ITRs and rep gene of serotype 2 (AAV2) are utilized resulting in pseudo-typed rAAVs [11]. In contrast, a wide variety of natural or engineered capsid genes can be used to package the transgene into the capsid with the desired characteristics and tissue tropism. In order to obtain the most efficient AAV vectors for a specific application, new capsid variants are constantly being isolated from natural sources [1] or developed either by rational design or by directed evolution approaches [12]. During the engineering of novel AAV capsid variants it is important to maintain essential functions, such as, the capsid-assembly and genome-packaging capabilities. Most of the research on these topics was conducted on AAV2, and it is therefore the focus of this review.

Capsid assembly

Composition & structure of assembled AAV capsids

The AAV capsid is composed of 60 subunits of the overlapping VP1:VP2:VP3, in an approximate ratio of 1:1:10, assembled with T = 1 icosahedral symmetry and a diameter of approximately 260 Å [13,14]. The 3D structures of numerous AAV serotypes and clonal isolates have been determined by x-ray crystallography and/or cryo-electron microscopy and image reconstruction [15–19,20]. The AAV VP monomer N-terminus (VP1u, VP1/2 common region, and ~15 VP3 N-terminal residues) is not seen in the crystal structures but has been suggested to be located on the inside of the capsid by cryo-reconstruction [21]. The remaining VP3 has a topology which contains a conserved alpha-helical region (αA) and a βA strand followed by an eight-stranded β barrel motif formed by the βBIDG and CHEF sheets (Figure 2A). Located between the secondary structure elements are interconnecting loops which are named based on their location between the strands, for example, the loop between the βD and βE strands is the DE loop, and that between strands βH and βI is the HI loop (Figure 2A).

Figure 2. The adeno-associated virus structure.

(A) Cartoon representation of AAV2 VP (PDB ID: 1LP3) monomer with the αA helix colored in red, β-strands colored in blue and the interconnecting loops colored in pink. The approximate icosahedral twofold, threefold and fivefold axes are represented by an oval, triangle and pentagon, respectively. (B) Surface representation (gray) of AAV2 capsid generated from 60 VP monomers with the position of one VP monomer highlighted in the same colors as in (A) and a second monomer in gray to enable easy orientation in the virus capsid.

AAV: Adeno-associated virus; VP: Viral protein.

This image was generated using the PyMol program [22].

The T = 1 icosahedral capsid of the AAVs (Figure 2B) is assembled via two-, three- and fivefold interactions which form a dimer, trimer and pentamer, respectively (Figure 3). The twofold interface is characterized by a depression which is formed by the loops after the βI strands of two VPs oriented in opposite direction and a wall formed by the αA helix (Figure 3A). This region has the least number of VP–VP interface interactions and represents the thinnest region of the virus capsid shell [23,24]. The threefold axis is surrounded by three protrusions formed by interactions of the BC, EF and interdigitating GH loops (Figure 3B). These protrusions are formed from interdigitating loops from two VPs and have the most extensive VP–VP interactions [23,24]. Residues forming these protrusions are radially the most distant from the virus capsid center. The fivefold axis contains a cylindrical channel formed by the interaction of five DE loops (β-ribbons; Figure 3C). The HI loops of each VP monomer lie on top of the βCHEF sheets of adjacent VP monomers and together form a depression around this channel. The channel connects the inside of the viral capsid to the outside and is reported to be the site of genome packaging and uncoating, and VP1u externalization for its PLA2 function required for endo/lysosomal capsid escape en route to the nucleus.

Figure 3. Adeno-associated virus symmetry-related interactions.

(A) VP dimer with the reference monomer colored in blue and the interacting monomer colored in cyan. Theicosahedral twofold axis (indicated by 2) is located adjacent to the conserved interacting α-A helix. (B) VP trimer formed from interdigitating variable loops. The icosahedral threefold axis is indicated by 3, the reference monomer colored in blue and the interacting monomers colored in cyan and orange. (C) VP pentamer with DE loops surrounding the icosahedral fivefold axis (indicated by 5). The reference monomer is colored in blue and the symmetry-related monomers are colored in cyan, orange, yellow and pink. The approximate icosahedral twofold, threefold and fivefold axes are represented by an oval, triangle and pentagon, respectively.

VP: Viral protein.

This image was generated using the PyMol program [22].

Mechanism of capsid assembly

The first glimpse into the mechanism of AAV capsid assembly was provided by the pulse-chase experiment by Myers and Carter which showed that the process involves the rapid formation of empty capsids from a pool of VPs, followed by the slow rate limiting step of packaging the ssDNA genome into the preformed empty capsid (see next section) [25]. Immunofluorescence and immunoprecipitation experiments have shown that the VP proteins are produced in the cytoplasm and are then transported to the nucleus, specifically to the nucleoli where capsid assembly occurs [26]. This transport is facilitated by nuclear localization signals present on the N-terminus of the VP within VP1u and VP1/VP2 common regions. Sedimentation sucrose gradient analysis of nuclear and cytoplasmic fractions of an AAV infection demonstrated that the cytoplasmic fraction contains a population of VP monomers and oligomers ranging in size from 10 to 15S consistent with the sedimentation coefficients for trimers and pentamers, respectively, of VP1, VP2 and VP3. The nuclear fraction contained VP oligomers with sedimentation coefficients between 60 and 110S consistent with empty and full capsids, respectively [27].

The AAV virus capsid is reported to be composed of VP1:VP2:VP3 in an approximate ratio of 1:1:10 based on densitometry analysis of purified AAV capsid VPs separated by SDS-PAGE [14,28–29]. More recently, mass spectrometry analysis of AAV1 by an orbitrap mass spectrometer showed that the incorporation of VP1, VP2 and VP3 into the virus capsid was stochastic and likely related to the rate of expression of each VP [30]. Mutational studies on the entire cap gene showed that the VP3 common sequence is enough to assemble AAV capsids by knocking out the start codons of VP1 and VP2 [31–34]. However, in addition to VP3, the AAP is also essential for capsid assembly [35]. This 23-kDa protein is also encoded within the ORF of VP2 and VP3, initiated from a non-canonical CTG start codon, and translated in a different reading frame compared with the VP sequence. The AAP is reported to function as a scaffold protein that facilitates the localization of VP3 to the nucleus [35].

Specific residues important for the formation of the AAV capsid have been identified via mutagenesis of residues distributed over the entire VP1 sequence and peptide insertions at specific VP positions followed by characterization of their ability to reduce or abolish capsid assembly. Residues located at the two- and fivefold interfaces of the VP, secondary structure elements and close to the fivefold channel play a role in capsid assembly (Table 1 & Figure 4) [36–37]. Alterations of residues in the symmetry-related interfaces are predicted to abrogate the formation of assembly intermediates or building blocks that are important components of the assembly pathway. Changes in the secondary structure elements of the VPs are likely to cause misfolding and ultimately degradation by the ubiquitin/proteasomal pathway. This residue-level information is critical toward maintaining the capsid structure when generating new and improved vectors for gene therapy applications.

Table 1.

Adeno-associated virus 2 capsid assembly mutants and their location on the virus capsid.

AAV2 residues	VP location / sidechain interaction	Ref.
H229, D231	Close to βA, twofold interaction	[38]
D237, R238	Close to αA, twofold axis (wall)	[38]
H255, K258	BC loop, fivefold interaction	[38]
R286, H288	βC strand	[38]
S292	αA, twofold interaction	[38]
R294, D295	αA, twofold interaction	[38,39]
R307, K309, R310	βD strand, interior surface	[40]
E322, V323, T324	DE loop, fivefold interaction	[36]
M604, V605	GH loop, threefold interaction	[40]
D608, R609, D610	GH loop, close to threefold axis	[38]
A667	HI loop, fivefold interaction	[39]
E681,1682, E683	βI strand, interior of the capsid	[38,39]
K688	βF strand, twofold interaction	[37,39]
E689, K692, R693	Post βI, twofold interaction	[38,39]
K706	Post βI, fivefold interaction	[39]

Figure 4. Residue determinants of adeno-associated virus 2 capsid assembly.

(A) Cartoon representation of the AAV2 VP monomer with amino acid positions affecting capsid assembly indicated as spheres. The location of the residues, listed in Table 1, are colored in blue for basic residues, red for acidic residues, green for polar residues and yellow for hydrophobic residues. (B) Surface representation of the AAV2 capsid with amino acids which affect assembly colored as in (A). The approximate icosahedral twofold, threefold and fivefold axes are represented by an oval, triangle and pentagon, respectively.

AAV: Adeno-associated virus; VP: Viral protein.

This image was generated using the PyMol program [22].

The role of scaffold proteins & DNA in AAV capsid assembly

Assembled capsids are first detected in the nucleus, specifically in the nucleoli during a productive infection [26]. This is followed by the observation of capsids colocalizing with both Rep proteins and the ssDNA [26, 41–42]. However, assembled capsids have been detected in the nucleus in the absence of the replication proteins suggesting that their presence is not required for assembly. Additionally, two nuclear proteins, nucleolin and nucleophosmin [43,44], have been shown to associate with both VPs and assembled capsids in the nucleoli. The role of these two proteins in the assembly process has not been determined.

Recently, AAV capsid assembly studies have focused on the characterization of AAP and its role in this process for the different AAV serotypes. AAP was reported to bind to the AAV VP multimers across the twofold axis of symmetry (thus, a dimer interface) at the β strands located at the inner surface of the capsid and target these VPs to the nucleus where it facilitates capsid assembly [35]. In the most recent studies, it has been shown that all serotypes tested, with the exception of AAV4, AAV5 and AAV11, require AAP in order to assemble their capsid [35,39, 45–47]. Thus, the VP3 of these exceptions (AAV4, AAV5 and AAV11) enter the nucleus via an AAP-independent mechanism that is yet unknown and challenge what is known about the role of AAP in AAV capsid assembly.

Genome packaging

The packaging of the viral genome into preformed capsids is an important step in the AAV viral cycle since it enables progeny virions of WT viruses to undergo another round of replication. For rAAV vectors, this enables the encapsulation of desired transgenes. The following sections review what is known with respect to the mechanisms involved in this process from studies on WT AAV virions and rAAV vectors. Generally, it is believed that the packaging mechanisms of both types of AAV capsids are identical.

AAV packaging mechanism

The AAV packaging process occurs in the nucleus of the cell producing the AAV capsid and relies on the presence of: assembled empty capsids, replicated AAV genomes and the viral Rep proteins. For the packaging of a certain DNA molecule into AAV capsids, the WT AAV genome or any transgene cassette, the DNA sequence must be flanked by the cis-acting ITRs. These not only represent the packaging signals but also aid the amplification of the DNA genome by a self-primed replication mechanism (Figure 5A). The products of the replication are DNA genomes of positive and negative polarity that are packaged into the AAV virions with equal frequency [48]. By replacing the D-sequence in one of the flanking ITRs with an unrelated sequence, the packaging of genomes with only positive or negative polarity can be achieved [49]. Mutation of this region in both ITRs leads to a defect in replication and packaging [50].

Figure 5. Adeno-associated virus genome replication.

(A) Simplified replication model of a single-stranded AAV genome by a self-priming mechanism leading to the generation of new genome copies. Newly generated AAV DNA segments are colored in green, orange and blue. Displaced AAV genomes are reported to be encapsidated into empty preformed capsids. (B) Model for the replication of an artificial self-complementary genome. Note that the terminal resolution site of the left ITR in a self-complementary AAV genome is deleted. The mutated ITR leads to an altered replication mechanism compared with (A) and an internal ITR that connects complementary DNA strands.

AAV: Adeno-associated virus; ITR: Inverted terminal repeat; scAAV: Self-complementary adeno-associated virus; trs: Terminal resolution site.

The current model for AAV packaging couples the replication and packaging process (Figure 5A). This is supported by the observation that the Rep proteins, capsids and the AAV DNA colocalize in the nucleoplasm in infected cells [26]. During AAV replication, the N-terminus of the large Rep proteins (Table 2 & Figure 6A) [51] bind specifically to the RBE within the ITRs and cleave the trs [52] which allows self-priming and induces further replication. Eventually the displacement of the ssDNA genome occurs due to DNA replication of the complementary strand (Figure 5A). In that process, the single-stranded binding proteins, E2A (adenovirus) or ICP8 (herpes virus) provided by the helper viruses are involved [2, 51]. Furthermore, herpes virus directly interacts with AAV replication with its viral helicase– primase complex [2]. After one single-stranded genome is fully displaced, the DNA is reported to be translocated into the capsid in a 3′ to 5′ direction by the helicase/ATPase domain of the small Rep proteins [53]. The translocation of the ssDNA into the capsid is believed to occur through the channel at the fivefold symmetry axis.

Table 2.

Adeno-associated virus 2 Rep protein mutants leading to a DNA packaging defect.

Mutation	Phenotype	Ref.
M225G	Rep52/40 not expressed (mutated start codon), genomes are replicated but not packaged	[53]
K340H	Mutated ATP binding site, affects ATPase and helicase activity, small Rep proteins with this mutation are defective for packaging	[53]
E379K E379Q K391I K391T K404I K404T	Mutated helicase domain, affects helicase activity, small Rep proteins with this mutation are defective for packaging	[53]
K404A K406A	Mutation reduces the ability of the small Rep proteins to bind ssDNA	[54]
Δ225–274-Rep52/40	Small Rep proteins with N-terminal deletion, reduced binding to capsid	[55]
R107A K136A R138A	Mutated DNA-binding domain of large Rep proteins, no replication of AAV genomes	[51]
Y156F	Mutation of endonuclease domain, replication defect	[51]

Figure 6. Adeno-associated virus 2 Rep protein structure.

(A) Surface representation (gray) of the N-terminus (aa 1–202) of a Rep68 monomer (PDB ID: 5DCX). Highlighted are the residue positions which affect DNA replication as listed in Table 2. (B) Surface representation (gray) of a Rep40 monomer (aa 225–490; PDB ID: 1U0J). Highlighted are the positions of the residues which affect DNA packaging as listed in Table 2. Please note that the coordinates for aa 402–407 have not been deposited.

aa: Amino acid. AAV: Adeno-associated virus.

This image was generated using the PyMol program [22].

While the large Rep proteins (Rep78/68) are reported to bind to the ITRs, the small Rep proteins (Rep52/40) are reported to interact with ssDNA in the nucleus [54], with the affinity of the former interaction being greater than the latter [55]. The absence of the small Rep proteins, a mutation in the helicase/ATPase domain or a capsid-binding deficient Rep52/40 variant leads to significantly reduced amounts of genome-containing AAV particles (Table 2 & Figure 6B) [53]. These observations support a central role for the small Rep proteins in AAV genome packaging. With respect to the role of the capsid in this process, mutations of residues assembling the channel at the fivefold, for example T329/T330A, in the AAV2 capsid show a severe reduction of Rep–capsid interaction and genome packaging [42]. Another mutation, R432A, which is located near the threefold symmetry axis also prevents packaging of genomes into the capsid [38]. This residue is however not exposed on the capsid surface (interior or exterior) and capsids with the R432A change are still capable of binding Rep proteins [42]. Instead, R432A affects the stability of the capsid, inducing structural changes far away from the actual site of the mutation, such as, an increased disorder of the N-terminal region of the capsid, in addition to VP structural rearrangements [56]. This observation suggests that packaging fidelity requires a certain level of stability that needs to be maintained. This may be due to a need to compensate for the increase in pressure that has been reported to be associated with genome packaging for parvoviruses [57, 58]. Interestingly, residues within the inside of AAV capsid have been shown to interact with the single NT ordered in several AAV capsid structures [23, 59–61]. The binding site consists of residues V420, P421, N609, I622, H630, P631, S632, G637, F638, G639 and L640 from one VP, and residues D626 and K641 from a threefold related VP [23, 59–61]. These residues participate in nonspecific (with deoxyribose sugar and phosphate) and specific (with base) hydrophobic and polar interactions with the NT. Significantly, regardless of whether the structures are determined for ‘empty’ capsids, virions packaging the green fluorescent protein gene, luciferase gene or a therapeutic gene, this NT is ordered inside this conserved DNA-binding pocket. However, although the conservation of this interaction and the interacting residues suggest a role in DNA packaging and possibly capsid stability, these phenotypes are yet to be tested.

Genome size limitation in AAV packaging

The WT AAVs package a viral genome of approximately 4.7 kb into their capsids. Packaging of vector genomes with variable sizes revealed that encapsidation of genomes is most efficient in a range of 4.1–4.9 kb [62]. Attempts to package larger transgenes have been made, however this only occurs with lowered efficiency and a clear upper packaging threshold at 5.2 kb [63,64]. Larger genomes are truncated at the 5′ end of the ssDNA (Figure 7A) and segments are packaged into different capsids [65]. Subtracting the size of the ITRs from this maximum packaging capacity leaves up to 4.9 kb for the desired transgene cassette which includes the gene of choice and other required transcriptional regulatory elements. One approach to package the largest-possible gene is to reduce the size of the promoters, polyA signals and other control elements [66–68]. Although some of these minimal elements might work in some target tissues, they still limit the gene size to approximately 4.6 kb. This size is sufficient for genes encoding a protein of approximately 1500 aa and thus, suitable for packaging the majority of human protein-coding genes when expressed as cDNA. However, a small percentage of human genes (<6%) exceeds this size limit [69].

Figure 7. Strategies for the expression of genes exceeding the packaging capacity of adeno-associated virus capsids.

(A) Single oversized vector genomes >5.2 kb are partially packaged. Truncated genomes of opposite polarity might anneal and reassemble to the full length transgene by DNA repair pathways. (B) Dual overlapping or (C) trans splicing vectors separate the transgene to normal-sized vector genomes with regions that overlap in the vector genomes. The full-length transgene is restored either by homologous recombination or splicing.

In order to express proteins from oversized genes using AAV vectors, single oversized [70] and ‘dual’ vectors [71] have been developed. Although the exact mechanism for single oversized vectors is not fully understood, these become truncated and packaged into different capsids, and it is believed that genomes of opposite polarity might anneal [65] and reassemble to the full-length transgene by DNA repair pathways (Figure 7A) [72]. With AAV dual vectors [73], which have been shown to result in higher transduction efficiencies compared with single oversized vectors, the transgene cassette is separated into two or more normal-sized vector genomes which are packaged into different capsids. One method to reassemble the fragments is homologous recombination by introducing a region that overlaps in both vector genomes (Figure 7B) [74]. Another method utilizes a ‘trans splicing’ mechanism to express the full transgene encoded on the different vector genomes (Figure 7C). For this approach, the 5′ end fragment vector genome possesses a splice donor site while a splice acceptor site is inserted at the beginning of the 3′ end fragment vector genome [75]. In order to make this reassembly as efficient as possible, hybrid dual vectors have been developed which combine both strategies [71]. No matter which dual vector method is used, they all require the coinfection of both vectors in each individual cell. With this strategy, very large genes can be expressed using AAV vectors. This has been shown to be an effective strategy for the largest human gene, dystrophin, whose cDNA was packaged into three trans-splicing vectors leading to the expression of the full-length protein in mice [76]. An additional strategy to express oversized genes involves the miniaturization of large genes. In this approach, large portions of the coding sequence are deleted leaving only the essential domains which are then packaged into a single capsid. This method has also been utilized for the expression of a minidystrophin gene [77].

Packaging of self-complementary vector genomes into AAV capsids

As stated above, AAV vectors package ssDNA. This genome requires the synthesis of the cDNA strand to generate a transcription-competent dsDNA template following nuclear entry. For this step, host’s cell replication proteins are needed. In postmitotic cells, such as, neuronal cells, these proteins are largely not expressed. This can lead to a delayed onset of transgene expression. In order to circumvent this problem, self-complementary AAV (scAAV) vectors have been developed [78] which allow immediate transcription of the transgene in both dividing and nondividing cells. For the generation of these vectors, the trs in one of the flanking ITRs is deleted. This prevents the Rep proteins from nicking this sequence which leads to the generation of an internal ITR that connects the cDNA strands of the transgene cassette during genome replication (Figure 5B). However, packaging of scDNA genomes is associated with a significant disadvantage because the genome size is reduced by approximately 50% compared with ssDNA genome-packaging vectors. With a coding capacity of approximately 2.5 kb, these vectors are only suitable for small proteins and regulatory RNA expression cassettes. Multiple studies have shown superior transgene expression efficiency of scAAV compared with ssAAV at early time points [79]. However, in long-term studies, transduction from ssAAV genes eventually reaches the scAAV levels [80], thus the pros and cons with respect to onset of expression and transgene size must be considered.

Collateral packaging of nonvector DNA sequences into AAV capsids

For the clinical applications of AAV vectors, the capsids should ideally only contain the desired vector genome cassettes with none to minimal contaminations with foreign DNA sequences. Toward achieving this goal, vector genomes to be packaged into AAV capsids are only flanked by the ITRs. However, different types of contaminating DNA sequences have been described as being copackaged with the transgenes, including fragments of the host cell genome from the cells that are used during vector production [81]. It has been reported that genome sequences which are close homologs to RBE, such as, the AAVS1 site, were detected in purified AAV vector preparations [82]. Another source of DNA impurities are the helper gene constructs that are transfected or infected into the vector producer cells. Rep expression cassettes, driven by the AAV p5 promoter, have been reported to result in significant rep gene copackaging possibly due to the presence of a cryptic RBE in the p5 promoter [83]. Third, a phenomenon termed ‘reverse packaging’ has been described that results in the packaging of DNA sequences outside of the ITR-flanked vector genome cassette [84]. Measures to reduce the packaging of nonvector DNA sequences include the removal of potential RBEs from helper constructs [85] or the increase of the backbone of the vector genome plasmid to sizes larger than the packaging capacity of AAV vectors to minimize reverse packaging [86]. Traditionally, the contamination with foreign DNA sequences is determined by quantitative PCR. However, this limits the quantification to only those sequences that are specifically detected by primers of known sequences. Recent next-generation sequencing based technologies allow an unbiased identification and quantification of DNA impurities [87].

Conclusion & implications for gene therapy applications

Efficient capsid assembly and genome packaging are essential steps in WT virus infection as well as the production of vectors for therapeutic application. For the AAVs, specific regions of the capsid have been identified as important determinants of assembly, including the two- and fivefold interfaces. The threefold interface, with its numerous interactions, appears to be redundant in this process. This suggests that dimer and pentamer formation are important steps in this process, although further studies are required to determine their temporal order. Nuclear proteins, particularly nucleolin and nucleophosmin, are involved in this assembly process, although their exact role requires further elucidation. With respect to genome packaging, the Rep proteins and VPs, particularly the VP3 overlapping sequence, play a role. Fidelity of the interactions between these viral components dictate the ITR-mediated packaging of WT genomes as well as transgenes into AAV capsids. Further knowledge of AAV determinants for capsid assembly and genome packaging provide important guides for vector engineering. This is particularly useful in efforts to generate vector variants with the desired phenotypes of improved transduction efficiency in addition to improved vector production and the packaging of transgenes with oversized genomes. These efforts can avoid the modification of essential residues while nonessential aa positions provide sites for engineering.

Future perspective

Understanding the steps involved in viral infection, particular for viral systems being exploited for clinical applications, such as, the AAVs, impacts efforts to improve therapeutic efficacy. In addition to information on determinants of capsid assembly and genome packaging, information will continue to be acquired on AAV capsid determinants of tissue recognition, tissue tropism, transduction efficiency, efficient endo/lysosomal pathway trafficking and host immune system interactions. Information will become available on Rep determinants of genome packaging, particularly the footprint for their interaction on the AAV capsid. Residues engaged in capsid–AAP interactions will be finer mapped and the reason for the serotype-specific requirement for AAP involvement in AAV capsid assembly will be better understood. Additional host factors involved in capsid assembly and genome packaging will be identified, and regions with which they engage on the AAV capsid will become known. Vectors will be designed which combine all the available information for vector optimization. This will include the design of personalized vectors guided by the patients’ immune profile with improved cellular trafficking properties to the target tissue to be treated.