Liver Injury Increases the Incidence of HCC following AAV Gene Therapy in Mice

Abstract

Adeno-associated virus (AAV) integrates into host genomes at low frequency, but when integration occurs in oncogenic hotspots it can cause hepatocellular carcinoma (HCC). Given the possibility of recombinant AAV (rAAV) integration leading to HCC, common causes of liver inflammation like non-alcoholic fatty liver disease (NAFLD) may increase the risk of rAAV-induced HCC. A rAAV targeting the oncogenic mouse Rian locus was used, and as expected led to HCC in all mice infected as neonates, likely due to growth-related hepatocyte proliferation in young mice. Mice infected with rAAV as adults did not develop HCC unless they were fed a diet leading to NAFLD, with increased inflammation and hepatocyte proliferation. Female mice were less susceptible to rAAV-induced HCC, and male mice with NAFLD treated with estrogen exhibited less inflammation and immune exhaustion associated with oncogenesis compared to those without estrogen. Adult NAFLD mice infected with a non-targeted control rAAV also developed HCC, though only half as frequently as those exposed to the Rian targeted rAAV. This study shows that adult mice exposed to rAAV gene therapy in the context of chronic liver disease developed HCC at high frequency, and thus warrants further study in humans given the high prevalence of NAFLD in the population.

Graphical abstract

Figure 1

Schematic of Rian Targeting Vector and a Control tdTomato Vector The AAV-Rian vector consists of a CAG promoter flanked by arms of homology to Rian. The green arrows indicate the positions of primers, one inside the CAG promoter and one outside of the arm of homology, used to detect CAG integration by homologous recombination. AAV-tdTomato is a non-editing vector that expresses the tdTomato transgene driven by the CAG promoter.

Figure 2

HFD-Induced Liver Injury Does Not Exacerbate AAV-Induced HCC in the Neonatal Mouse Model (A) Outline of experimental design (n = 3–14 per group). (B) Tumor incidence in neonatal mice infected with AAV-Rian or AAV-tdTomato in presence or absence of diet induced liver injury. Tumor incidence was not statistically significant between diet groups by chi-square analysis (p > 0.05). Numbers above the bars represent the sample size. (C) Tumor burden at 6 months. No statistical significance was achieved between diet groups by one-way ANOVA followed by Bonferroni post hoc (p > 0.05). Numbers above the bars represent the sample size. (D) Images of gross liver specimens from 6-month-old mice. (E) Representative H&E staining of livers from mice on RD and HFD. (F) Representative gel of CAG insertion in Rian tumors by integration PCR, where one primer targets the CAG region and the other primer targets a region outside of the homology arm. Four tumors, two from female and two from male tumors, were selected. No integration was detected in mice on RD infected with AAV-tdTomato.

Figure 3

HFD-Induced Liver Injury and Partial Hepatectomy Exacerbates AAV-Induced HCC in the Adult Mouse Model (A) Outline of experimental design (n = 5–20 per group). (B) Tumor incidence in adult mice infected with AAV. HFD increased Rian induced tumor incidence (chi-square test, p < 0.05). Numbers above the bars represent the sample size. (C) Tumor burden in adult mice infected with AAV. Both HFD and partial hepatectomy increased tumor burden in AAV-Rian-CMV infected mice (one-way ANOVA, Bonferroni post hoc, p < 0.01). Numbers above the bars represent the sample size. (D) Images of gross liver specimens from 9-month-old mice. (E) Representative H&E staining of livers from mice on RD and HFD. (F) Representative gel of CAG insertion in Rian tumors by integration PCR. Target PCR product was not observed in mice on RD infected with either Rian or tdTomato virus but was observed in mouse on HFD infected with Rian and in mice that had received a partial hepatectomy. The bands seen in the RD/tdTomato and RD/Rian lanes were sequenced and did not show CAG or Rian sequences, suggesting that these bands are artifacts.

Figure 4

Tumors Exhibit a Prototypical Rian Signature and Have a Similar Transcriptomic Profile to Human HCC Subclassifications (A) Heatmap showing the AAV-Rian signature and the “AAV-Rian UP” and “AAV-Rian Down” gene sets generated from the Wang et al. study, and analysis of the expression of the genes in the Rian Locus (GRCm38/mm10, chr12:108860000–110418000). (B) HCC classification, survival signatures, oncogenic pathways, and lipid metabolism activation in the tumor samples of the study. p Values were calculated comparing diets within male Rian tumor samples.

Figure 5

Estrogen Partially Ameliorates HFD-Induced Liver Injury (A) Outline of experimental design. (B) Representative H&E liver images of mice on RD or HFD for 1 month or 6 months. Females have slightly lower HFD-induced injury at both time points, and male mice on HFD+E2 also had less injury than males on HFD that didn’t receive E2. (C) Heatmap showing the expression of immune-related pathways, oncogenic pathways, lipid metabolism, and liver function in the background liver samples.

Figure 6

Estrogen Treatment Partially Suppressed Hepatocyte Proliferation (A) BrdU staining (brown dots) identifies dividing hepatocytes. (B) Morphometric analysis of BrdU hepatocytes. Estrogen partially reversed HFD-induced injury (Student’s t test, p < 0.05).