CRISPR-Cas9 gene editing of hepatitis B virus in chronically infected humanized mice

Abstract

Chronic hepatitis B virus (HBV) infection is a major public health problem. New treatment approaches are needed because current treatments do not target covalently closed circular DNA (cccDNA), the template for HBV replication, and rarely clear the virus. We harnessed adeno-associated virus (AAV) vectors and CRISPR-Staphylococcus aureus (Sa)Cas9 to edit the HBV genome in liver-humanized FRG mice chronically infected with HBV and receiving entecavir. Gene editing was detected in livers of five of eight HBV-specific AAV-SaCas9-treated mice, but not control mice, and mice with detectable HBV gene editing showed higher levels of SaCas9 delivery to HBV⁺ human hepatocytes than those without gene editing. HBV-specific AAV-SaCas9 therapy significantly improved survival of human hepatocytes, showed a trend toward decreasing total liver HBV DNA and cccDNA, and was well tolerated. This work provides evidence for the feasibility and safety of in vivo gene editing for chronic HBV infections, and it suggests that with further optimization, this approach may offer a plausible way to treat or even cure chronic HBV infections.

Keywords: AAV; CRISPR/Cas9; Gene editing; HBV; Humanized mouse.

Graphical abstract

Figure 1

Gene delivery to humanized chimeric mouse livers (A) Livers from liver-humanized FRG mice were stained to show human (red, anti-human albumin [hAlb]) and mouse (green, anti-mouse albumin [mAlb]) hepatocytes. (B) GFP expression in humanized livers of FRG mice 14 days after retro-orbital delivery of scAAVLK03-smCBA-GFP vector at a dose of 5 × 10¹⁰ vector genomes/mouse. Livers were stained using an anti-GFP antibody. (C) In situ GFP expression in whole livers at necropsy detected using an AMG EVOS fluorescence microscope 14 days after retro-orbital administration of a low (5 × 10¹⁰ vector genomes) or high (2 × 10¹¹ vector genomes) dose of scAAVLK03-smCBA-GFP. (D) Co-labeling with anti-GFP (green) and anti-human albumin (red) antibodies in humanized livers from a FRG mouse treated with 2 × 10¹¹ vector genomes of scAAVLK03-smCBA-GFP. (E) Liver gene delivery in liver-humanized Alb-uPA/scid mice. Liver tissue was harvested from Alb-uPA/scid mice 14 days after intravenous retro-orbital delivery of scAAV-LK03-smCBA-GFP vector at a dose of 5 × 10¹⁰ vector genomes/mouse. Images show human albumin (red, hAlb) and GFP (green, anti-GFP).

Figure 2

Experimental setup for AAV-SaCas9 treatment of HBV-infected humanized FRG mice AAV vectors were generated with capsid LK03 and express SaCas9 in combination with either two control GFP-specific sgRNAs (GFP early/late) or two HBV genotype C-specific sgRNAs (HBV early/late). Chronically HBV-infected FRG mice were given entecavir for 17 days to suppress viremia and then injected intravenously with either anti-GFP or anti-HBV AAV SaCas9 vectors, and followed for levels of viremia for 4 (early) or 9 (late) weeks, before necropsy. Mice were infected with a genotype C HBV inoculum 144 days prior to AAV administration and entecavir administration was halted at 4 weeks after AAV administration. vg, vector genomes; ITR, inverted terminal repeat; EFM, EF1α short/minute virus of mice intron hybrid promoter; pA, SV40 poly(A); sgRNA, single guide RNA; hH1, human H1 promoter; hU6, human U6 promoter.

Figure 3

HBV-specific and off-target gene editing PCR amplicons spanning the C7 and C14 HBV target sites, or 20 closely related off-target sites for C7 or C14 within the human or mouse genomes (five of each) were amplified from liver genomic DNA and subjected to deep sequencing for analysis of mutations within the indicated 26-bp SaCas9 sgRNA target sequences. Mutations detected within each target site were identified using a custom script, and the mutation rate is based on mutations with multiple reads (non-singletons). (A and B) Mutation rates for the C7 and C14 target sites were determined for all mice (A), and the relative mutation rate for each off-target site was compared to the mutation rate of mice from the GFP-specific control sgRNA treatment group (B). The chromosomal location and starting nucleotide location for each off-target site are indicated along with the number of mismatches versus C17 or C14. Error bars indicate SD.

Figure 4

Levels of human serum albumin and human hepatocyte chimerism The percentage of human hepatocytes at the time of death was determined by quantifying levels of human and mouse RPP30 by qPCR on total DNA extracted from livers using primer/probe sets against human and mouse RPP30. Values were corrected for the presence of non-hepatocyte mouse cells as described in Materials and Methods. (A and B) Levels of human hepatocyte chimerism are shown for each treatment group (A) and for individual mice (B). (C and D) Levels of human serum albumin were quantified by ELISA and are shown for all animals prior to initial HBV challenge (C) or for surviving animals at day 56 after AAV administration (200 days after HBV challenge) (D). (E) The percentage of human hepatocytes at death for animals surviving 56 days or longer after AAV is also shown. Asterisks indicate mice with detectable HBV gene editing. Error bars indicate SD; p values were generated using one-sided t tests. Values for the six mice that survived for 56 days or longer are color coordinated (colored dots).

Figure 5

Liver-associated viral DNA levels at the time of death (A–C) Levels of total HBV DNA (A), cccDNA (B), and total HBV DNA per cccDNA molecule (C) found in human hepatocytes of each liver were assayed by ddPCR using primers that detect total HBV DNA, cccDNA, and hRPP30. Grouped (left panel) and individual mouse (right panel) values are shown. Mouse livers were snap-frozen at necropsy, or as soon as possible following death. Total HBV DNA levels were analyzed using total DNA extracted from livers by a QIAGEN DNeasy blood and tissue kit. cccDNA levels were analyzed using DNA that was extracted from livers using a modified Hirt procedure, and then digested with T5 exonuclease. Error bars indicate SD; p values were generated using one-sided t tests.

Figure 6

Longitudinal viral loads Chronically HBV-infected FRG mice were given entecavir to suppress viremia, then injected intravenously with either anti-GFP or anti-HBV AAV-SaCas9 vectors as shown and followed for levels of viremia for 4 (anti-GFP early, anti-HBV early) or 9 (anti-GFP late, anti-HBV late) weeks. Mice were infected with a genotype C HBV inoculum 144 days prior to AAV administration, and entecavir administration was halted at 4 weeks after AAV administration to monitor levels of viremia rebound in control or treated mice. (A and B) All mice were monitored weekly during entecavir (days −17 to 28) and AAV (days 0–62) treatment, and viral loads for grouped (A) and individual (B) mice are shown. Error bars indicate SD.

Figure 7

Characterization of HBV-infected and AAV-SaCas9-treated humanized FRG mouse livers Serial liver sections from humanized liver FRG mice were stained with hematoxylin and eosin (H&E), subjected to RNAscope for the presence of HBV and SaCas9 RNA, or co-labeled by immunohistochemistry for human cytokeratin 18 (hCK18) in combination with HBV surface antigen (HBsAg), activated caspase-3, or Ki67. Representative serial sections are shown for animal GFP-L3, which received control GFP6 and GFP7 sgRNAs. Scale bars, 1 mm.

Figure 8

Quantification of HBV and SaCas9 RNA levels (A) Liver sections from three AAV-SaCas9 control and three AAV-SaCas9-treated humanized liver FRG mice were subjected to RNAscope using custom HBV and SaCas9-specific probes. Cells in each image were assigned based on DAPI staining, and those containing foci indicative of HBV and/or SaCas9 probe hybridization were quantified using a RRScell image analysis algorithm, which was used to assign individual cell types. (B) Examples of assigned HBV⁺/SaCas9⁺ cells (red circles), HBV⁺/SaCas9⁻ cells (green circles), HBV⁻/SaCas9⁺ cells (blue circles), and HBV-/SaCas9- cells (purple circles) present within the boxed region of mouse HBV-L1 shown in (A). (C) where appropriate, semi-automated image masking by RRScell was used to analyze non-tumor regions only (yellow areas). (D–F) Quantification of the percentage of HBV⁺ cells per DAPI⁺ cell (D), SaCas9⁺ cells per HBV⁺ cell (E), and HBV⁺ cells per SaCas9⁺ cell (F) is shown for individual humanized mouse livers. Tumors (T) and approximate tumor margins (dotted lines) are indicated where present.