Standardized Nomenclature and Reporting for PacBio HiFi Sequencing and Analysis of rAAV Gene Therapy Vectors

Eric Talevich, Elizabeth Tseng, Alpha Amadou Diallo, Nadia Sellami, Amicia Elliott, Brandi L. Cantarel, Nam K. Tonthat, Pranam Chatterjee, Phillip W.l. Tai, Claire AldridgebioRxiv 2024 · DOI

TL;DR

Researchers have developed a standardized nomenclature and analysis pipeline for PacBio HiFi sequencing data of AAV gene therapy vectors. This addresses a critical need for consistent reporting, enabling better quality control, comparability across studies, and ultimately, improved AAV vector design and manufacturing.

What Problem Does This Paper Address?

Recombinant adeno-associated viruses (rAAVs) are a cornerstone of gene therapy, with over 200 vector designs in human trials. Ensuring the safety and efficacy of these therapies hinges on a thorough understanding of the vector's genetic payload – its purity, integrity, and the absence of contaminants. While PacBio's HiFi long-read sequencing offers unprecedented resolution for characterizing AAV genomes, the lack of standardized computational analysis methods and reporting metrics has been a significant hurdle. This inconsistency makes it difficult to compare results from different production runs, hinders efforts to optimize vector design, and complicates quality control. This paper tackles this problem by proposing a unified approach to analyzing and reporting AAV sequencing data.

Methods

The core of this work is the development of a standardized bioinformatics analysis workflow, named Long-read AAV Analysis (LAAVA). The process begins with HiFi reads generated from PacBio sequencing. These reads are then mapped to a comprehensive set of reference sequences, including the AAV vector itself, packaging plasmids (RepCap and helper), and the host cell genome.

A key innovation is the classification of each read into specific "types" and "subtypes" based on its alignment characteristics. This allows for the identification of:

  • Vector Genome Purity: Distinguishing between full-length, partial, and fragmented AAV genomes.
  • Contaminants: Identifying and quantifying non-vector sequences such as host DNA, plasmid DNA (RepCap, helper), and chimeric recombinations.
  • ITR Configuration: Analyzing the "flip/flop" orientation of the inverted terminal repeats (ITRs), which is a critical quality attribute for AAV.

The workflow is implemented as a Nextflow pipeline, containerized with Docker for reproducibility and ease of deployment. The final output is a comprehensive report in HTML and PDF formats, summarizing key quality metrics and alignment statistics.

Datasets

The paper presents analyses of two example AAV sequencing runs: one single-stranded AAV (ssAAV) construct ("pAV-CMV-GFP") and one self-complementary AAV (scAAV) construct ("CBA-eGFP"). These datasets were generated using PacBio Sequel II systems. The raw sequencing data is made publicly available for community use and validation.

Key Insights and Ideas

  • The Power of Long Reads for AAV: HiFi sequencing's ability to capture full-length AAV genomes in a single read is transformative for detailed characterization, revealing nuances previously missed by short-read technologies.
  • Standardization is Crucial: The lack of standardized nomenclature and reporting has been a bottleneck. This work provides a much-needed framework for consistent analysis and comparison.
  • ITR Flip/Flop as a QC Metric: Analyzing the proportions of flip/flop configurations of ITRs offers a valuable quality control criterion for AAV production.
  • Identifying Subtle Issues: The detailed classification of reads can reveal subtle issues in vector design or manufacturing, such as unexpected snapback genome formation in ssAAV or oversized genomes in scAAV.
  • Open Source for Community Adoption: Providing the LAAVA software as open-source is key to promoting widespread adoption, transparency, and collaborative improvement of AAV analysis.

Key Results

  • ssAAV Analysis: The analysis of the ssAAV construct revealed notable issues, including a significant proportion of reads classified as scAAV (35.83%), suggesting template switching during replication. Approximately half of the AAV reads could not be classified into an orientation, potentially due to truncated ITRs or sequence divergence.
  • scAAV Analysis: The scAAV construct showed exceptionally high quality, with 97.69% of reads classified as scAAV and a high percentage of full-length reads (93.84%). A small number of "vector+backbone" reads were observed, which, if present at higher levels, could indicate issues beyond vector design.
  • Read Length Distributions: The analysis provided detailed read length distributions for full-length and partial reads of both ssAAV and scAAV, offering insights into genome integrity.

Limitations and Future Work

  • Artifacts and Interpretation: The authors acknowledge that artifacts can be introduced at various stages (production, library prep, sequencing). The report aims to quantify these, and the read definitions accommodate their possibility. Further work may be needed to refine the interpretation of certain read types, particularly the "other" category.
  • Expanding the Scope: While the current workflow focuses on AAV vector genomes, future work could expand to analyze other aspects of AAV products, such as capsid protein composition or integration events.
  • Community Feedback: The open-source nature of the software invites community participation, which will be crucial for its ongoing development and refinement.

Summary updated

Abstract

Despite recombinant adeno-associated viruses (rAAVs) being the leading platform for gene therapy, there is a lack of standardized computational analysis methods and reporting to assess the contents of each capsid through long-read sequencing. PacBio's highly accurate long-read HiFi sequencing enables comprehensive characterization of AAV genomes but requires bioinformatics expertise for analyzing, interpreting and comparing the results. To address this need and improve the understanding of functional viral payloads, our working group established standardized nomenclature and reporting for long-read sequencing data of rAAV vectors. The working group recommendations cover critical quality attributes (CQAs) related to vector purity (full-length vs. fragmented genomes) and identification of contaminants (host DNA, plasmid DNA). Our data analyses of de novo manufacturing runs by the recommended protocol revealed specificity of full and partially filled capsids and high-resolution characterization of partial/truncated vector species. Finally, we provide an open-source software implementing this standardized AAV analysis and reporting to promote transparency, facilitate data comparability, and improve rAAV vector design and quality control.
BibTeX
@misc{talevich2024standardized,
  author = {Eric Talevich and Elizabeth Tseng and Alpha Amadou Diallo and Nadia Sellami and Amicia Elliott and Brandi L. Cantarel and Nam K. Tonthat and Pranam Chatterjee and Phillip W.L. Tai and Claire Aldridge},
  title = {{Standardized Nomenclature and Reporting for PacBio HiFi Sequencing and Analysis of rAAV Gene Therapy Vectors}},
  year = {2024},
  doi = {10.1101/2024.05.07.592296},
  howpublished = {bioRxiv},
}
Figure