In May 2009, hepatitis–splenomegaly syndrome affected a flock of 37-week-old broiler breeder hens in Shandong, China. This flock had a history of decreased egg production. Affected chickens had regressive ovaries, extensive necrosis and hemorrhage of the liver, and enlarged liver and spleen. Antibodies against avian HEV ORF2 were detected in 80 of 94 serum samples from the same chicken flock, according to ELISA (5,10) with the truncated ORF2 protein used by Guo et al (10) and chicken serum diluted 1:100 in 0.5% Tween-20 phosphate-buffered saline containing 2.5% nonfat dry milk and 10% Escherichia coli lysate. On the basis of previous results, we used a cutoff optical density of 0.43 (11). Using a published method (12), we detected an avian HEV ORF2 RNA gene with 242 bp in 7 of 10 fecal and 5 of 8 bile samples.

From the bile samples that were positive for the avian HEV ORF2 gene, we used nested RT-PCR with 5 overlapping fragments to amplify the near-complete genomic sequence of avian HEV. Primers were designed on the basis of the other 4 avian HEV near-complete sequences in GenBank (Table 1). The RT-PCR conditions and reaction mixture were designed according to the SuperScript II One-Step RT-PCR System instructions (Invitrogen, Carlsbad, CA, USA). To identify the extreme 3′ genomic sequence, we used a modified RACE (3′ rapid amplification of cDNA ends) technique. The sense primer F5 (Table 1) was chosen from the ORF2 region, and the antisense primers included a commercially available anchored adaptor primer and an amplification primer (Invitrogen). Using inner PCR primers, we sequenced the PCR products of 5 fragments in both directions (Table 1); the sequence data were collected by an ABI3730 Genetic Analyzer (JinSiTe Biotech Co., Nanjing, China).

We assembled the near-complete genome of avian HEV, which was 6,660 nt long including the 3′ poly A tail, by using 5 overlapping fragments sequences and Lasergene 7.0 EditSeq computer programs (DNAStar, Madison, WI, USA) and designated it China avian HEV (CaHEV). CaHEV contained a complete ORF1 gene encoding a nonstructural protein of 1,522 aa, an ORF2 gene encoding a capsid protein of 606 aa, an ORF3 gene encoding a cytoskeleton-associated phosphoprotein of 87 aa, and a 3′ noncoding region of 121 nt. The sequences of CaHEV were deposited into GenBank under accession no. GU954430.

The near-complete genomic and different region sequence analyses performed by using ClustalW (www.clustal.org) and Lasergene 7.0 MegAlign software indicated that CaHEV shared the highest identity (98.3%) with European avian HEV isolate (EaHEV) and 82.0%–82.6% with 3 other avian HEV isolates (Table 2). Moreover, compared with the different regions of 4 other avian HEV strains, the ORF1 gene of CaHEV shared 80.7%–98.3% nt and 92.7%–98.8% aa sequence identities, the ORF2 gene shared 84.1%–98.5% nt and 98.3%–99.7% aa sequence identities, the ORF3 gene shared 93.9%–98.9% nt and 88.6%– 97.7% aa identities, and the 3′ noncoding region shared 78.9%–97.6% nt identities (Table 2).

ORF1 of CaHEV contained most mutations compared with prototype avian HEV (prototype aHEV); 5, 16, and 29 nonsilent mutations occurred in the methyltransferase, helicase, and RNA-dependent RNA polymerase (RdRp) functional domains, respectively (data not shown). However, only 2 mutations occurred in motif VII of RdRp domain (Figure 1, panel A), which contains 8 motifs responsible for virus replication (13). The 2 mutations in motif VII of the CaHEV RdRp domain are L(1432)M and I(1434)V. Australian avian HEV isolate (AaHEV) also has the mutation in the latter position and was a transition from I(1433) to T (Figure 1, panel A). This position is well conserved among mammalian HEV isolates by the presence of V, which is the same as CaHEV (Figure 1, panel A).

In the ORF2 region, 6 nonsilent mutations (C4R, R5G, G27S, T42A, T303V, and Q473M) were determined for CaHEV and compared with prototype aHEV. One mutation of Q(473)M, in the antigenic domain II, was seen in EaHEV and in CaHEV (Figure 1, panel B). Because this domain is unique to avian HEV, as predicted by Haqshenas et al. (14) and Guo et al. (10), this point mutation may change the antigenicity of the epitopes in domain II of the capsid protein. In antigenic domain IV, a mutation of R(600)K occurred in the “avirulent aHEV” compared with other 4 avian HEV strains, including CaHEV, from the sick chickens (Figure 1, panel C). This mutation may affect the virulence of avian HEV as speculated by Billam et al. (8) and Billic et al. (9). The 3 putative N-linked glycosylation sites (²⁵⁵NLS [1], ⁵¹⁰NST [2], and ⁵²²NGS [3]) are shared between prototype aHEV and CaHEV (data not shown). However, the second site is ⁵¹⁰NNT in “avirulent aHEV” and AaHEV strains and is eliminated in the EaHEV strain (9). In human and swine HEV strains, these sites are ¹³⁷NLS (1), ³¹⁰NLT (2), and ⁵⁶²NLS (3). Recently, the potential N-linked glycosylation in ORF2 was shown to prevent formation of infectious particles, but its role in other functions of HEV, e.g., virus virulence and cell tropism, remain to be elucidated (15). In the ORF3 gene, including only 83 aa, 10 nonsilent mutations were found compared with the prototype aHEV, and 9 mutations were the same as EaHEV (data not shown).

Phylogenetic trees of the near full-length sequence of avian and mammalian HEV strains were constructed by using the neighbor-joining distance method and Lasergene 7.0 software. A bootstrap test of 1,000 replicates was used to evaluate the reliability of the groups. Avian HEV was segregated into a distinct branch separate from mammalian HEV; according to the genotype separation corresponding to their geographic origin suggested by Bilic et al. (9), CaHEV belongs to avian HEV genotype 3 (Figure 2).

Avian HEV infection of a chicken flock in Shandong, China, was identified by detection of avian HEV ORF2 antibodies and viral RNA. A near-complete avian HEV genome from the flock was determined, and sequence analysis indicated that this avian HEV strain displayed the highest identity (98.3%) with EaHEV and belonged to avian HEV genotype 3.