Damien Marsic

Altering Tropism of rAAV by Directed Evolution.

Directed evolution represents an attractive approach to derive AAV capsid variants capable of selectively infect specific tissue or cell targets. It involves the generation of an initial library of high complexity followed by cycles of selection during which the library is progressively enriched for target-specific variants. Each selection cycle consists of the following: reconstitution of complete AAV genomes within plasmid molecules; production of virions for which each particular capsid variant is matched with the particular capsid gene encoding it; recovery of capsid gene sequences from target tissue after systemic administration. Prevalent variants are then analyzed and evaluated.

High-accuracy biodistribution analysis of adeno-associated virus variants by double barcode sequencing

Biodistribution analysis is a key step in the evaluation of adeno-associated virus (AAV) capsid variants, whether natural isolates or produced by rational design or directed evolution. Indeed, when screening candidate vectors, accurate knowledge about which tissues are infected and how efficiently is essential. We describe the design, validation, and application of a new vector, pTR-UF50-BC, encoding a bioluminescent protein, a fluorescent protein and a DNA barcode, which can be used to visualize localization of transduction at the organism, organ, tissue, or cellular levels. In addition, by linking capsid variants to different barcoded versions of the vector and amplifying the barcode region from various tissue samples using barcoded primers, biodistribution of viral genomes can be analyzed with high accuracy and efficiency.

Direct Head-to-Head Evaluation of Recombinant Adeno-associated Viral Vectors Manufactured in Human versus Insect Cells

The major drawback of the Baculovirus/Sf9 system for recombinant adeno-associated viral (rAAV) manufacturing is that most of the Bac-derived rAAV vector serotypes, with few exceptions, demonstrate altered capsid compositions and lower biological potencies. Here, we describe a new insect cell-based production platform utilizing attenuated Kozak sequence and a leaky ribosome scanning to achieve a serotype-specific modulation of AAV capsid proteins stoichiometry. By way of example, rAAV5 and rAAV9 were produced and comprehensively characterized side by side with HEK293-derived vectors. A mass spectrometry analysis documented a 3-fold increase in both viral protein (VP)1 and VP2 capsid protein content compared with human cell-derived vectors. Furthermore, we conducted an extensive analysis of encapsidated single-stranded viral DNA using next-generation sequencing and show a 6-fold reduction in collaterally packaged contaminating DNA for rAAV5 produced in insect cells. Consequently, the re-designed rAAVs demonstrated significantly higher biological potencies, even in a comparison with HEK293-manufactured rAAVs mediating, in the case of rAAV5, 4-fold higher transduction of brain tissues in mice. Thus, the described system yields rAAV vectors of superior infectivity and higher genetic identity providing a scalable platform for good manufacturing practice (GMP)-grade vector production.

Evaluation of engineered AAV capsids for hepatic factor IX gene transfer in murine and canine models

BackgroundAdeno-associated virus (AAV) gene therapy vectors have shown the best outcomes in human clinical studies for the treatment of genetic diseases such as hemophilia. However, these pivotal investigations have also identified several challenges. For example, high vector doses are often used for hepatic gene transfer, and cytotoxic T lymphocyte responses against viral capsid may occur. Therefore, achieving therapy at reduced vector doses and other strategies to reduce capsid antigen presentation are desirable.MethodsWe tested several engineered AAV capsids for factor IX (FIX) expression for the treatment of hemophilia B by hepatic gene transfer. These capsids lack potential phosphorylation or ubiquitination sites, or had been generated through molecular evolution.ResultsAAV2 capsids lacking either a single lysine residue or 3 tyrosine residues directed substantially higher coagulation FIX expression in mice compared to wild-type sequence or other mutations. In hemophilia B dogs, however, expression from the tyrosine-mutant vector was merely comparable to historical data on AAV2. Evolved AAV2-LiC capsid was highly efficient in hemophilia B mice but lacked efficacy in a hemophilia B dog.ConclusionsSeveral alternative strategies for capsid modification improve the in vivo performance of AAV vectors in hepatic gene transfer for correction of hemophilia. However, capsid optimization solely in mouse liver may not predict efficacy in other species and thus is of limited translational utility.

An Immune-Competent Murine Model to Study Elimination of AAV-Transduced Hepatocytes by Capsid-Specific CD8+ T Cells

Multiple independent adeno-associated virus (AAV) gene therapy clinical trials for hemophilia B, utilizing different AAV serotypes, have reported a vector dose-dependent loss of circulating factor IX (FIX) protein associated with capsid-specific CD8+ T cell (Cap-CD8) elimination of transduced hepatocytes. Hemophilia B patients who develop transient transaminitis and loss of FIX protein may be stabilized with the immune-suppressive (IS) drug prednisolone, but do not all recover lost FIX expression, whereas some patients fail to respond to IS. We developed the first animal model demonstrating Cap-CD8 infiltration and elimination of AAV-transduced hepatocytes of immune-deficient mice. Here, we extend this model to an immune-competent host where Cap-CD8 transfer to AAV2-F9-treated mice significantly reduced circulating and hepatocyte FIX expression. Further, we studied two high-expressing liver tropic AAV2 variants, AAV2-LiA and AAV2-LiC, obtained from a rationally designed capsid library. Unlike AAV2, Cap-CD8 did not initially reduce circulating FIX levels for either variant. However, FIX levels were significantly reduced in AAV2-LiC-F9-treated, but not AAV2-LiA-F9-treated, mice at the study endpoint. Going forward, the immune-competent model may provide an opportunity to induce immunological memory directed against a surrogate AAV capsid antigen and study recall responses following AAV gene transfer.

549. Fine Tuning of Transduction Efficiency of rAAV Vectors via Modulation of Capsid Composition

Recombinant Adeno-associated virus (rAAV) vectors have emerged as one of the most versatile and successful gene therapy delivery vehicles. Even though the industry is poised for the expansion into several application areas represented by orphan diseases, a simple and scalable rAAV production technology is still lacking. We have recently developed the OneBac system to allow scalable, high-titer production of the full range of rAAV serotypes by infection of stable insect Sf9 cell lines with a single baculovirus. Some of the serotypes produced in this system, however, were characterized by a low transduction efficiencies compared to HEK 293-derived vectors. Here we describe an approach for resolving this drawback by modulating the ratios of VP1:VP2:VP3 capsid composition to derive particles with a higher VP1 content. This is accomplished by modifying a canonical Kozak sequence preceding VP1 ATG start codon. A range of Kozak sequences has been tested and the most favorable have been identified for AAV5 and AAV9 serotypes. These newly designed capsid genes were incorporated into Sf9 stable cell lines mediating packaging of AAV5 and AAV9 vectors which exceeded transduction efficiencies of HEK 293-derived counterparts by 2-3 folds. Curiously, the optimal ratios of VP1:VP2:VP3 were serotype-specific requiring fine-tuning to achieve a compromise between higher VP1 content mediating higher transduction potencies vs. lower packaging efficiencies impeding production yield. In summary, we have developed a novel approach of significantly enhancing biological potencies of AAV vectors derived from baculovirus system thus facilitating translational applications.

723. AAV Mediated Cancer Targeting: Systemic Trafficking to Tumor Is More Important Than Vector Tumor Cell Interaction

Adding tumor specific ligands to enhance vector tumor cell interaction is the conventional concept to generate tumor targeting adeno-associated viral vector (AAV). However, it remains poorly proved whether high AAV tumor cell interaction contributes to high tumor localization in vivo following systemic delivery. Here, we conducted directed evolution selections on patient derived xenograft models using a complex AAV capsid library. Uniquely, we compared the pressure for AAV tumor cell interaction alone (intratumoral library injection) and multi-layer pressure including traveling to the tumor and infection (intravenous library injection). Distinct patterns of AAV capsid motifs were identified after intratumoral and intravenous screenings. Motifs isolated from intratumoral screenings were named tumor specific motifs and those isolated from intravenous screenings were named systematic trafficking motifs. AAVs with tumor specific motifs but not systemic trafficking motifs showed significantly increased tumor cell transduction in vitro, indicating enhanced vector target cell interaction after intratumoral-based selections. Interestingly, following systemic delivery, AAVs with systemic trafficking motifs mediated hundreds of folds higher transgene expression than those with tumor specific motifs and wild type AAV in vivo. The combination of both motifs further increases the tumor tropism but not the transduction efficiency in vivo. When analyzing the AAV genome biodistribution by quantifying the genome copy number, the systemic trafficking motifs greatly reduced the native AAV tropism, which potentiated higher accessibility of AAV to the tumor. In contrast, AAVs only bearing tumor specific motifs maintained native AAV tropism and failed to mediate increased genome localization in tumor. Furthermore, in two independent patient derived xenograft models and two different tumor types, our novel AAV vector armed with combined motifs all showed hundreds of folds increase in transduction efficiency with no detectable off-targeting expression. In conclusion, contradictory to the prevailing theory, our study demonstrated that the AAV tumor cell interaction did not contribute to increased tumor localization but just specificity in vivo. Therefore, targeting is not only depending on AAV tumor cell interaction but also, and more importantly, depending on the accessibility of AAV to the tumor cells following systemic delivery. Future cancer directed AAV vector design should take into account the complex processes during systemic delivery as well as the vector tumor cell interaction.

304. Deriving Useful Data from Next-Gen Sequencing of AAV Capsid Libraries

The relatively short read lengths produced by the major Next-Gen sequencing platforms (up to 300 nt for Illumina, up to 200 nt for Ion Proton) are poorly suited for analyzing large combinatorial libraries of longer nucleotide sequences, such as those involving AAV capsid genes. Despite its much lower throughput and accuracy, the PacBio single-molecule real-time technology is currently the only option for long templates, with average read lengths of 10 to 15 kb. Using the Circular Consensus Sequencing (CCS) mode, in which template DNA fragments are circularized, allows a significant increase in accuracy due to the fact that each template is being sequenced multiple times. To interpret PacBio CCS data, we have previously reported a first version of the CapLib code, which was developed to identify variable regions in AAV combinatorial capsid libraries.DNA fragments, derived from purified DNA-containing AAV particles, 869 bp in length and including 27 variable nucleotide positions, were sequenced in CCS mode using the P6-C4 chemistry. A total of 26,897 reads were obtained, with a mean read length of 814 nt, a mean read quality of 0.9956 and a mean number of passes of 21.34. Only 5,456 reads had the correct size of 869 nt, and of these, only 1,638 had a sequence that matched the reference sequence, indicating that only 6% of reads were potentially error-free and that the vast majority had multiple insertions and deletions.In order to extract more useful information from the sequencing data, a new version of the CapLib software was developed. It is designed to correct sequencing reads in silico by assuming that constant nucleotide positions are wild-type and focusing on the detection of the variable positions. The premise was validated by Sanger sequencing of multiple clones, confirming that mutations were present only in the intended positions. Depending on the parameter values used, up to 14,000 reads could be recovered by CapLib 2.In addition to recovering PacBio CCS reads, CapLib 2 can also assemble Sanger sequencing data, translate recovered reads into protein sequences and perform detailed analyses of the dataset. It can also analyze clones resulting from directed evolution experiments and compare them with the original library.

264. Absence of CD8 T Cell Responses Against AAV Capsid Using Reduced Vector Dose or Library Selected Liver Targeted AAV2 Based Capsids

Predicting immune responses directed against the AAV capsid have proven to be a challenge in the translation of liver directed AAV gene therapy into the clinic. Previously we have described the first mouse model that allows us to evaluate the potential for CD8 T cell responses directed against AAV transduced hepatocytes and to define protocols and conditions to minimize this response (Martino et al. Blood 2013). Empirical data from both AAV2 and AAV8 clinical trials for hemophilia B have suggested that a threshold vector dose is needed for activation of capsid specific CD8 T cells and subsequent elimination of factor IX expressing hepatocytes. We sought to experimentally confirm this observation in our murine model by evaluating the impact of reducing AAV2 and AAV8 vector doses by one log. No reduction in FIX expression was observed for either AAV2 or AAV8 transduced mice receiving 1×1010 vg and capsid specific CD8 T cells, suggesting that that the reduced vector dose dropped the number of viral particles below a threshold for efficient MHC I presentation.In another set of studies we evaluated two recently described AAV2 capsid variants (AAV2-LiA and AAV2-LiC), which were selected for murine hepatocyte tropism from a highly complex rationally designed capsid library (Marsic et al. Mol Ther 2014). Both AAV2-LiA and AAV2-LiC vectors were not targeted by capsid specific CD8 T cells and resulted in stable FIX expression for the duration of the experiment. Our previous studies conducted on the AAV2-M3 (Y-F 444,500,730) capsid showed that these three mutations provided sustained FIX expression in the presence of capsid CD8 T cells. By design, AAV2-LiA contains two of the previously described (Y-F) mutations at positions 444 and 500 and AAV2-LiC contains one at position 500. These data now extend on this study and suggest that either a single Y-F mutation at position 500 or double at positions 444 and 500 may be sufficient in reducing capsid antigen presentation on hepatocytes. Although it is unclear if these single and double Y-F mutants are alone sufficient or if there are synergistic mutations in AAV2-LiA and AAV2-LiC that contribute to reduced capsid antigen presentation. Nonetheless, we were unable to detect the presence of infiltrating CD8 T cells both at the one week and four week time points where previous studies conducted on a similar AAV2 vector demonstrated loss of FIX expression and the presence of infiltrating CD8 T cells at the one week time point. Interestingly although AAV2-LiA was less efficient than AAV2-LiC in transgene expression in immune competent mice, our studies conducted in immune deficient mice showed that both vectors provided similar levels of FIX expression. These data suggest that AAV liver transduction efficiency is altered in immune deficient mice. Therefore, we would offer caution in the analysis and interpretation of AAV transgene expression data generated solely in immune deficient animals.Together, these data support a vector dose dependency threshold for capsid antigen presentation on hepatocyte MHC I molecules and demonstrates that the novel AAV2-LiA and LiC capsids avoid detection by capsid specific CD8 T cells.