Genetic host specificity of hepatitis E virus

Abstract

Hepatitis E virus (HEV) causes epidemic and sporadic cases of hepatitis worldwide. HEV genotypes 3 (HEV3) and 4 (HEV4) infect humans and animals, with swine being the primary reservoir. The relevance of HEV genetic diversity to host adaptation is poorly understood. We employed a Bayesian network (BN) analysis of HEV3 and HEV4 to detect epistatic connectivity among protein sites and its association with the host specificity in each genotype. The data imply coevolution among ∼70% of polymorphic sites from all HEV proteins and association of numerous coevolving sites with adaptation to swine or humans. BN models for individual proteins and domains of the nonstructural polyprotein detected the host origin of HEV strains with accuracy of 74-93% and 63-87%, respectively. These findings, taken together with lack of phylogenetic association to host, suggest that the HEV host specificity is a heritable and convergent phenotypic trait achievable through variety of genetic pathways (abundance), and explain a broad host range for HEV3 and HEV4.

Keywords: Adaptation; Bayesian network; Coevolution; HEV ORFs; Hepatitis E virus; Prediction.

Fig. 1

Coordination among HEV protein sites modeled with BNs. The BN models show genome-wide epistatic connectivity among aa sites (a structural coefficient threshold = 2.0) and association to host variable (a structural coefficient threshold = 0.95). Nodes represent polymorphic aa sites and arcs between them represent dependency. Nodes are color-coded according to the ORF and ORF1-domains and numbered according to the aa positions in the respective ORFs. Orf1(x) encompasses UNK and denotes aa sites that fall outside known ORF1 domains (n = 16 in BNHEV3 and n = 18 in BNHEV4). (A) A learned BN of HEV3 sequences (n = 65) and (B) A learned BN of HEV4 sequences (n = 55).

Fig. 2

Contribution of proteins and ORF1-domains to BN_HEV3 and BN_HEV4. (A) Bar charts show number of aa sites involved in BNs for each protein and ORF1-domain. (B) Number of links among aa sites from all proteins and ORF1-domains observed in BNs. Numbers outside and inside of parenthesis are for BN_HEV3 and BN_HEV4, respectively.

Fig. 3

Strengths of epistatic influences. Strength of linkages (primary y-axis) and number of links (secondary y-axis; crosses joined with dashed lines) among HEV proteins and ORF1-domains (x-axis) was computed from learned BN_HEV3 and BN_HEV4. (A) Overall strengths measured for each protein or ORF1-domain in entire BNs. Directionality of influences is color coded. (B–J) Strength and number of links to all proteins and ORF1-domains in BN_HEV3 and BN_HEV4 observed for each protein and ORF1-domain.

Fig. 4

The most interconnected (n ⩾ 5) and influential (global KL-divergence ⩾ 2.0) nodes in BN_HEV3 and BN_HEV4. Bars show global strength of influence (primary y-axis) and number of links (secondary y-axis) for aa sites identified in BNs (Fig. 1). The numbers of total and intra-protein (or intra-domain) links are shown with crosses and rhombi, respectively.

Fig. 5

HEV3 and HEV4 protein sites most associated with host specificity. Bars show relative MI values (primary y-axis) for aa sites identified in BNs (Fig. 1). MI for the Plp site 557 (MI = 0.16) in BN_HEV3 and Pol site 1692 (MI = 0.19) in BN_HEV4 were assigned a relative value of 1. P values for each site are identified with crosses and shown as percentage (secondary y-axis).

Fig. 6

HEV3 and HEV4 host-specific epistatic motifs. (A) HEV3-BN_Swine contains 40 arcs connecting 42 aa sites; (B) HEV3-BN_Human contains 44 arcs connecting 48 aa sites; (C) HEV4-BN_Swine contains 34 arcs connecting 40 aa sites; and (D) HEV4-BN_Human contains 60 arcs connecting 66 aa sites. All links are statistically significant (p ⩽ 10⁻⁵) and highly correlated (avg. r = 0.9326 and r = 0.8791 – A and B, respectively; r = 0.8075 and r = 0.7934 – C and D, respectively). Color coding and numbering are as in Fig. 1.

Fig. 7

Host-specific separation of HEV3 and HEV4 strains in LP-modeled physicochemical space. Shown are LP plots of physicochemical properties for aa sites from Pol and Pp (Table 4). Probability mapping of human and swine strains is color-coded, with human space shown in blue and swine in red. Color density is proportional to probability values. (A) LP map of HEV3 variants (n = 65) using Pol aa physicochemical properties or markers (n = 16); (B) LP map of HEV4 variants (n = 55) using Pol markers (n = 16) and (C) LP map of HEV4 variants (n = 55) using the Pp markers (n = 10). Below the mappings are line charts showing the prediction results (probability scores) on validation datasets from the above corresponding LP maps; y-axis represents probability [0–1]; p(H) and p(S) are probabilities of the human (blue line) and swine (red line) origin of a strain, respectively. GenBank accession numbers (x-axis) are shown for each test sequence; black triangles and circles identify HEV strains obtained from humans and swine, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)