Nature Medicine | Contaminating Plasmid Sequences and Disrupted Vector Genomes in the Liver Following Adeno-Associated Virus Gene Therapy

This study, using metagenomic sequencing and in situ hybridization techniques,首次 revealed the presence of manufacturing plasmid contamination and complex vector genome structures in the livers of patients after adeno-associated virus gene therapy, uncovering a novel molecular mechanism potentially linked to hepatotoxicity.

 

Literature Overview

The article titled "Contaminating Plasmid Sequences and Disrupted Vector Genomes in the Liver Following Adeno-Associated Virus Gene Therapy," published in the journal Nature Medicine, reviews and summarizes the clinical and molecular features of a child with spinal muscular atrophy (SMA) who developed severe hepatitis after receiving onasemnogene abeparvovec gene therapy. The research team conducted long-read and short-read metagenomic sequencing and in situ hybridization analysis on liver biopsy tissue, systematically revealing residual manufacturing plasmid sequences, highly rearranged vector genome structures, and the presence of multiple viral sequences in the liver post-treatment. The study further explores the potential association between these molecular events and AAV gene therapy-related hepatotoxicity. The paragraph is coherent and logically structured, ending with a Chinese period.

Background Knowledge

Adeno-associated viruses (AAVs) are widely used gene therapy vectors, with several products already approved by the FDA for treating monogenic disorders such as spinal muscular atrophy (SMA) and hemophilia. However, intravenous infusion of AAV vectors is often accompanied by hepatotoxicity, with some patients experiencing severe liver injury or even acute liver failure, particularly at high doses or in older, heavier patients. The mechanisms underlying this toxicity remain incompletely understood and may involve immune responses to the viral capsid, vector genome, or transgene product, impurities from vector production, or direct cellular toxicity. AAV vector production relies on a triple-plasmid transfection system: a vector plasmid containing the therapeutic gene (e.g., pSMN), a packaging plasmid encoding AAV replication and capsid genes (e.g., pAAV2/9), and a helper plasmid providing auxiliary viral functions (e.g., pHelper). Despite adherence to cGMP standards, residual plasmid fragments, empty capsids, or recombination products may persist during production. Additionally, prior studies have indicated that complex tandem repeat structures can form in non-human primates. However, whether these complex structures form in human patients and whether they are driven by residual plasmids or helper viruses has lacked direct evidence. This study focuses on an SMA child who developed hepatitis after high-dose onasemnogene abeparvovec treatment, aiming to elucidate the underlying molecular mechanisms of liver pathology and fill a critical knowledge gap regarding the molecular basis of AAV-related hepatotoxicity.

 

 

Research Methods and Experiments

The research team performed multi-omics analysis on liver biopsy tissue obtained seven weeks after infusion of onasemnogene abeparvovec in a 7-year-old girl with SMA type 1. First, histopathology and immunohistochemistry were used to evaluate liver inflammation and injury. Subsequently, host-depleted DNA and RNA metagenomic sequencing was performed using the Illumina platform to unbiasedly identify non-human nucleic acid sequences. Initial classification used the Kraken2/Bracken pipeline, followed by alignment of sequencing reads to reference sequences of AAV, adenovirus, and manufacturing plasmids. To verify plasmid contamination, specific PCR assays were designed to detect adenoviral sequences outside the pHelper region. Oxford Nanopore long-read sequencing was further employed to resolve complex genome structures, with rearrangement analysis performed by alignment to complete plasmid references. Finally, RNAscope in situ hybridization was used to localize the therapeutic gene SMN1 and plasmid-specific sequences (e.g., bacterial origin of replication and AAV9 capsid gene) on tissue sections, with comparisons made to healthy and disease control tissues.

Key Conclusions and Perspectives

• Liver biopsy histopathology showed moderate portal and lobular inflammation with CD4+ and CD8+ T cell infiltration, consistent with AAV-associated hepatitis, but adenovirus immunostaining was negative
• Short-read metagenomic sequencing detected AAV, adenovirus, and human herpesvirus 6B (HHV-6B) sequences; however, HHV-6B had full-genome coverage and PCR confirmation, whereas adenovirus signals were limited to partial gene regions and specific PCR was negative, indicating HHV-6B was a true infection while adenovirus signals originated from manufacturing plasmid contamination rather than wild-type infection
• Sequencing reads widely mapped to all three manufacturing plasmids (pSMN, pAAV2/9, pHelper), especially the backbone regions of pAAV2/9 and pSMN, confirming the presence of plasmid DNA contamination in the liver post-treatment
• Long-read sequencing revealed extensive vector genome breakage, concatemerization, and complex rearrangements, including head-to-head, head-to-tail, and alternating repeat patterns, resembling products of rolling-circle replication
• In situ hybridization confirmed that 28.5% of hepatocytes expressed SMN1, while 5.1%–5.8% of cells carried plasmid-specific sequences, validating both vector transduction and plasmid contamination
• Multiple vector-human genome fusion sites were identified, suggesting potential random integration, though no clear hotspots were observed and fusion transcripts were rare, indicating mostly non-functional integration
• The study hypothesizes that residual plasmid sequences (e.g., AAV rep genes or adenoviral helper genes) or co-infecting HHV-6B may act as auxiliary factors driving in vivo vector genome amplification and rearrangement, thereby promoting complex structure formation

Research Significance and Prospects

This study provides the first direct molecular evidence in human patients of manufacturing plasmid contamination and complex vector genome structures in the liver following AAV gene therapy. These findings challenge the traditional view that AAV vectors primarily exist as stable circular episomes in vivo, suggesting instead that their genomes can undergo significant rearrangements. Plasmid contamination may act as an immunostimulant, activating innate immune pathways (e.g., TLR9), or promote vector amplification through rep protein expression, exacerbating hepatocyte damage. Complex tandem structures may affect transgene expression efficiency or induce DNA damage responses, contributing to inflammation.

Future studies need to validate whether these findings are common across patients receiving AAV therapy, especially comparing those with and without hepatotoxicity. More efficient plasmid removal processes and optimized vector designs to reduce rearrangement propensity are needed. The prevalence of helper viruses (e.g., HHV-6) in other patients and their impact on treatment safety should also be evaluated. This work provides a new perspective on understanding AAV hepatotoxicity mechanisms and may guide the development of safer vector manufacturing processes and clinical monitoring strategies.

 

 

Conclusion

This study conducted deep molecular analysis of a child with SMA who developed hepatitis after AAV gene therapy, revealing the presence of manufacturing plasmid DNA contamination, complex rearranged vector genome structures, and HHV-6B infection in the liver. Metagenomic sequencing and in situ hybridization together confirmed that not only does the therapeutic vector genome undergo extensive concatemerization and breakage, but production plasmid sequences also persist in hepatocytes. These aberrant structures may be driven by replication elements from residual plasmids or helper viruses (e.g., HHV-6B), triggering immune responses or cellular stress that lead to liver injury. This work is the first to demonstrate in humans that AAV vectors can form highly complex DNA structures in vivo, suggesting current manufacturing processes may introduce potentially pathogenic impurities. The study emphasizes the importance of optimizing AAV purification protocols and monitoring patients for helper virus infections, providing key molecular targets for developing safer gene therapy strategies. These findings have significant implications for the safety assessment, quality control standards, and clinical management of AAV gene therapy.

 

Sarah Buddle, Li-An K Brown, Sofia Morfopoulou, Francesco Muntoni, and Judith Breuer. Contaminating plasmid sequences and disrupted vector genomes in the liver following adeno-associated virus gene therapy. Nature Medicine.