Bile specimens (n = 200) were collected in September 2009 from rabbits raised on 20 farms in western France, in the departments of Maine et Loire (n = 6), Vendée (n = 6), Deux-Sèvres (n = 4), Calvados (n = 2), and Loire Atlantique (n = 2), the main geographic areas of rabbit farming in France. We sampled 10 rabbits from each farm when they were slaughtered at 70–90 days of age (Table 1). All rabbits were healthy and intended for human consumption. All samples were immediately stored at −80°C.

Liver specimens (n = 205) were collected during September 2007–November 2010 from 18 populations of wild rabbits, established in warrens; each population was considered epidemiologically independent. The populations were located in several departments of mainland France: Dordogne (n = 7), Finistère (n = 3), Deux-Sèvres (n = 2), Loire-Atlantique (n = 2), Haute-Garonne (n = 1), Charentes (n = 1), Morbihan (n = 1), and Pyrénées-Orientales (n = 1) (Table 1). The number of rabbits sampled in a given warren ranged from 1 to 44. They were >6 months of age, apparently healthy, and intended for human consumption. Each rabbit was eviscerated within a few hours of its death, and a sample of its liver was taken and immediately frozen at −80°C. Necropsies were performed on a group of 12 rabbits from the same warren in Haute-Garonne (W3), and samples of their intestine and cecum were taken, in addition to samples from the liver.

Serum specimens were collected from immunocompetent and immunocompromised patients who had received a diagnosis of hepatitis E from the department of virology at Toulouse University Hospital. All samples were stored at −80°C (23,24).

Samples (140 μL of rabbit bile and 50 mg of liver, intestine, and cecum) were disrupted with TRIzol (Invitrogen, Saint Aubin, France). RNA was extracted with QIAamp Viral RNA Mini Kits (QIAGEN, Courtaboeuf, France).

We used 1-step real-time reverse transcription PCR on the Light Cycler 480 instrument (Roche Diagnostics, Meylan, France) to amplify a 70-bp fragment. The primers and probes targeted the ORF3 region: forward primer HEVORF3-S: 5′-GGTGGTTTCTGGGGTGAC-3′, reverse primer HEVORF3-AS: 5′AGGGGTTGGTTGGATGAA-3′, and probe 5′-Fam-TGATTCTCAGCCCTTCGC-Tamra-3′ (25). Each 50-μL reaction mix contained 1 μL of SuperScript III Platinum One-Step Quantitative RT-PCR System (Invitrogen), 15 μL of RNA, primers (200 nmol/L) and probes (150 nmol/L), and 40 U of RNase Out (Invitrogen). Reverse transcription was carried out at 50°C for 15 min, followed by denaturation at 95°C for 1 min. DNA was amplified with 50 PCR cycles at 95°C (20 s) and 58°C (40 s). HEV RNA was quantified by using a transcribed RNA standard constructed from a genotype 3f HEV strain (GenBank accession no. EU495148). The limit of detection was 100 copies/mL.

Two fragments, one within ORF2 (189 bp) and the other within ORF1, encompassing the hypervariable region and X domain (≈1,400 bp), were amplified and sequenced in both directions by the dideoxy chain termination method (PRISM Ready Reaction Ampli Taq Fs and Dye Deoxy primers; Applied Biosystems, Paris, France) on an ABI 3130XL capillary DNA analyzer (Applied Biosystems, Foster City, CA, USA). The primers used for the ORF2 fragment were the following: forward primer HEVORF2-S: 5′-GACAGAATTRATTTCGTCGGCTGG-3′ and reverse primer HEVORF2-AS: 5′-TGYTGGTTRTCATAATCCTG-3′. The primers used for the ORF1 fragment were the following: forward primer HEVORF1-S: 5′-TGACGGCYACYGTKGARCTTG-3′ and reverse primer HEVORF1-AS: 5′-ACATCRACATCCCCCTGYTGTATRGA-3′.

The whole genomes of 2 rabbit strains (W1–11 and W7–57) and 1 human strain (TLS-18516-human) were amplified by overlapping RT-PCR. The primers are listed in Table 2.

The genotype was determined by using reference strains as previously described (26). Phylogenetic analyses were performed with genotype information on reference sequences based on the HEV classification proposed by Lu et al. (27). Sequences were aligned by using ClustalW (MEGA5, www.megasoftware.net; BioEdit version 7.0, www.mbio.ncsu.edu/bioedit/bioedit). Phylogenetic trees were created by the neighbor-joining (Kimura 2-parameter) method with a bootstrap of 1,000 replicates.

The 2 partial sequences of ORF1 and the 5 full-length sequences reported in this study have been deposited in GenBank. The accession numbers are JQ013789 and JQ013790 for ORF1, and JQ013791 to JQ013795 for the full-length sequences of W1–11, W7–57, TLS- 18516-human, TR19 (genotype 3c), and TR02 (genotype 3e), respectively.

All bile specimens from the 200 farmed rabbits and the liver specimens from the 205 wild rabbits were tested for HEV RNA (Table 1). Samples from 7 farms (35%) and 9 warrens (50%) tested positive for HEV RNA. HEV RNA was found in a 14 bile samples (7%) from farmed rabbits. The median HEV RNA concentration in the bile samples was 2.3 × 10⁷ copies/mL (range 100 copies/mL–10⁹ copies/mL). A total of 47 liver samples (23%) from wild rabbits were positive for HEV RNA; median HEV RNA concentration was 1.9 × 10⁶ copies/g (range 1,400 copies/g–5.8 × 10⁷ copies/g).

We tested the liver, intestine, and cecum samples from 12 wild rabbits from the same warren (W3) in triplicate to obtain a clear picture of the tissue distribution of HEV in infected rabbits. HEV RNA was detected in all the tissues from 4 rabbits (nos. 4, 7, 9, 12), in the liver and intestine of 1 rabbit (no. 5), and in the liver only of 1 rabbit (no. 6) (Table 3). The virus loads in the liver (mean 4.8 log copies/g), intestine (mean 4.0 log copies/g), and cecum (mean 3.6 log copies/g) were not significantly different.

Phylogenetic analyses, conducted on the basis of a 189-nt fragment within ORF2 of the 37 HEV strains from rabbits, HEV3 strains from humans circulating in France, and HEV reference sequences (HEV1, HEV2, HEV3, HEV4, rabbit HEV, rat HEV, wild-boar HEV) indicated that the 37 new ORF2 sequences from rabbit HEVs were clustered. One cluster contained 3 ORF2 sequences from previously characterized HEV from farmed rabbits from China, 2 ORF2 sequences from HEVs from farmed rabbits in France, and 13 ORF2 sequences from HEVs from wild rabbits in France (Figure 1). This cluster also contained an ORF2 sequence from a strain from a person in France (TLS-18516-human) (Figure 1). This strain was found in a serum sample from a 46-year-old man with an elevated alanine aminotransferase level (400 IU/L, reference <35 IU/L).

Phylogenetic analysis based on a 1,400-nt fragment within ORF1, indicated that the ORF1 sequences from HEV strains from rabbits in France (n = 4) or China (n = 3) and the ORF1 sequence from the human strain TLS-18516-human formed a distinct genetic group among sequences of HEV genotypes 1–4 (data not shown). The cluster of rabbit HEV sequences was also distinct from the HEV sequences from wild-boar and rat HEV genotypes that were characterized recently.

Comparison of the ORF1 sequences from rabbit HEV strains with reference ORF1 sequences from HEV genotypes 1–4 showed an insertion of 93 nt in the X domain of the ORF1 of all the rabbit HEV strains. This insertion was also found in the TLS-18516-human strain. The deduced amino acid sequences corresponding to this insertion, located between amino acids 938 and 939 (Burmese strain, M73218), were not very similar, except for 2 conserved amino acids at the C-terminal end.

We obtained the full-length genomic sequences of HEV strains from 2 wild rabbits in France and the TLS-18516-human strain. The phylogenetic tree, constructed by the neighbor-joining method using the full-length genomic sequences (including the sequences of genotypes 3f, 3e, and 3c, which were circulating in France), revealed that the HEV genomes from the rabbit strains and the TLS-18516-human strain belonged to the same clade. This clade was clearly separated from genotypes 1–4, found in other mammals and from the new HEV genotypes found in wild boars and rats (Figure 2).

The length of the rabbit strain W1–11 genome, excluding the poly(A) tract at the 3′ terminus, was 7,262 nt. The length of the rabbit strain W7–57 was 7,231 nt, and that of the TLS-18516-human strain was 7,259 nt. The nucleotide sequences of the rabbit strains and the TLS-18516-human strain were 80.3%–85% identical (Table 4). The nucleotide sequences of the rabbit or TLS-18516-human strains were 76.1% to 78.2% identical to those of genotype 3, 72.7% to 73.7% identical to those of genotype 1, 72.2% to 73.5% identical to those of genotype 2, and 72.9% to 74.9% identical to those of genotype 4. These comparisons therefore indicate that the sequences of the rabbit HEV strains and the TLS-18516-human strain are distinct from all known strains of HEV genotypes 1–4 and from the newly described HEV genotypes from wild boars and rats.