Recurrent integration of domestic cat hepatitis B virus DNA near feline CCNE1 supports an oncogenic role in hepatocellular carcinoma in cats

Abstract

Hepatitis B virus (HBV) is the major cause of hepatocellular carcinoma (HCC) in humans. Domestic cat hepatitis B virus (DCHBV) naturally infects cats worldwide, but the oncogenic potential of this hepadnavirus is unclear. We investigated whether DCHBV contributes to feline HCC. Feline liver biopsies diagnosed with HCC (cases) and lymphocytic cholangitis (controls) were tested for DCHBV DNA by PCR. DCHBV-positive HCCs were further characterised by in situ hybridisation (ISH), whole-genome sequencing (WGS) and phylogenetic analysis. Targeted capture sequencing was used to identify and map viral DNA integrations. DCHBV DNA was detected in 17/71 (23.9 %) HCCs versus 0/88 controls (P < 0.001). ISH confirmed hepatocyte-specific viral localization. Phylogenetic analysis placed six viruses in genotype A, and a seventh divergent virus in genotype B virus. Virus-host chimeric sequences, consistent with integration sites, were identified in 11/16 PCR-positive HCCs. Eight of the 11 integration sites were independently confirmed with WGS. Viral termini in integrated DCHBV sequences corresponded to double-stranded linear DNA, the substrate for HBV integration. Five unique DCHBV integrations fell within, or were adjacent to, the promoter of the feline homologue of proto-oncogene CCNE1, a recurrent target for HBV integration in human HCC. Our findings reveal a compelling association between DCHBV detection and HCC in cats. Critically, virus integration in DCHBV-associated HCC is described for the first time, supporting that, like HBV, DCHBV can promote hepatocarcinogenesis by insertional mutagenesis. Clarification of fundamental DCHBV virology in vitro, and the consequences of natural infection could advance disease-prevention strategies for feline and human patients.

Highlights

• •Domestic cat hepatitis B virus is associated with hepatocellular carcinoma in cats.
• •Targeted capture sequencing reveals DCHBV integration in the cancer cell genome.
• •The cyclin E1 protooncogene, CCNE1, is targeted in feline and human HBV-associated cancers.
• •Clarification of the clinical consequences of natural DCHBV infection is needed.
• •Future studies of DCHBV could inform feline and human hepatology.

1. Introduction

Hepatitis B virus (HBV) is the leading cause of human hepatocellular carcinoma (HCC) [1]. A group I carcinogen, HBV exerts its oncogenic effects by indirect mechanisms related to chronic hepatitis that usually precede cancer, and directly from the consequences of oncogenic viral proteins and virus integration into the host genome [[2], [3], [4]].

HBV is the type species of the Hepadnaviridae, a family of small, enveloped, hepatotropic, partially double-stranded DNA viruses [5]. Classified alongside HBV in the genus Orthohepadnavirus are hepatitis B-like viruses that infect diverse mammalian hosts including rodents, bats, equids, shrews and non-human primates [5]. However, systematic investigation of the pathogenesis of non-human hepadnaviruses is limited to rodent viruses developed as experimental models of HBV infection. For example, woodchuck hepatitis virus (WHV) has an oncogenic capacity that exceeds that of HBV, rapidly and almost inevitably causing HCC in chronically infected Eastern woodchucks (Marmota monax) [6]. While hepatitis B-like viruses infecting other rodents, bats and donkeys have been detected concurrently, or in association with host hepatopathology, our understanding of the clinical consequences of chronic infection for most mammalian hosts is limited [[7], [8], [9], [10], [11]].

The domestic cat hepatitis B virus (DCHBV, or domestic cat hepadnavirus [DCH]; species Orthohepadnavirus felisdomestici) was discovered in 2018 [12]. The cat presents unique features as a hepadnavirus host. The enormous popularity of cats as pets has seen feline medicine become a major field of veterinary science. This provides context in which to interpret hepadnavirus-induced pathologies, as well as access to case material from naturally-infected cats. This structure supports a comparative oncology approach to advance novel treatments for pets and people. A similar program is already well-established for spontaneous cancers affecting dogs [13].

DCHBV has a typical 3.2 kb, partially double-stranded (ds)DNA genome with four overlapping ORFs encoding the core, polymerase, surface and X proteins [5,12]. Two DCHBV genotypes are described, with all but 4 of 68 viruses sequenced clustering in so-called Genotype A [14]. DCHBV infection of cats is widespread; molecular surveys conducted in Asia, North and South America, Europe, and Oceania report DCHBV viraemia in <1 % to >17 % of cats tested [15]. Serological markers for DCHBV are not widely available, but anti-core antibody detection in a study of 256 cats suggests that more than twice as many cats are exposed to DCHBV than are detected by viraemia alone [16].

DCHBV has been associated with both chronic hepatitis and HCC in one small study of archived diagnostic liver biopsies [17]. DCHBV DNA was detected by PCR in 6/14 (43 %) chronic hepatitis cases and 8/29 (28 %) HCC, and in situ hybridization (ISH) localized DCHBV to hepatocytes in virus-positive cases. In contrast, all 43 control samples, comprising biliary carcinoma, biliary inflammation, non-neoplastic/non-inflammatory liver lesions and histologically normal liver, tested DCHBV-negative [17].

Herein, we investigated the relationship between DCHBV and HCC in naturally-infected cats. First, we aimed to confirm the proposed association between DCHBV and HCC in a statistically robust study. As a control group, we used biopsies with a diagnosis of lymphocytic cholangitis, a chronic, progressive, likely immune-mediated inflammatory disease centered on the biliary tree that is histologically distinct from chronic hepatitis [18]. Second, we aimed to determine whether DCHBV integrates into the host genome in feline HCC, as observed in human HBV-associated HCCs, and to characterize the integration sites [2].

2. Materials and methods

2.1. Ethics statement

Samples used in this study were archived biopsies that had been obtained with owner consent, either for diagnostic purposes or at necropsy, and had been deidentified. Hence, institutional ethical approval was not required.

2.2. Sample and data collection

Two groups of archived formalin-fixed paraffin-embedded (FFPE) feline liver samples were collected in a multicentre study. Biopsies with a confirmed diagnosis of HCC were designated "cases" and those with lymphocytic cholangitis as "controls". Samples were identified by a database search from veterinary diagnostic laboratories located in the USA, UK and Taiwan. The recorded diagnosis was confirmed by histopathological review performed by board-certified veterinary pathologists (J.P.C. and J.M.C). Criteria to confirm lymphocytic cholangitis were a minimum of 12 portal tracts available for examination, and the presence of a predominantly lymphocytic inflammatory infiltrate targeting bile ducts, and signs of bile duct injury (such as proliferation or damage) [19]. Biopsies with a recorded diagnosis of lymphocytic cholangitis were excluded if neutrophils were present in the walls or lumen of bile ducts, or if lymphoma was suspected based on the histological pattern (including lack of bile duct targeting, absence of bile duct injury, or primarily lobular involvement). Sample data were recorded, including submission date, laboratory location, and, where available, patient age, sex, and neuter status.

2.3. DNA extraction and PCR testing

DNA was extracted from 8 μm scrolls of FFPE liver with Promega Maxwell RSC DNA FFPE kit. PCR for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was performed as described previously to assess the integrity of the DNA [17]. Samples testing negative for the 80 bp GAPDH product were excluded. PCR for DCHBV DNA was performed as described previously [20]. The identity of amplicons migrating at the expected band size (258 bp) was confirmed by Sanger sequencing (Eton Bioscience, North Carolina, USA) and Basic Local Alignment Search Tool BLASTN analysis using the National Center for Biotechnology Information (NCBI) GenBank database [21,22].

2.4. DCHBV genome sequencing using overlapping primers

Amplification and sequencing of the entire DCHBV genome was carried out using overlapping primers on DNA extracted from liver tissue that tested positive for DCHBV by PCR [12]. Viral sequences were edited using Geneious Prime 2022.2.2 (https://www.geneious.com). High-quality sequences (>95 %) were subjected to BLASTN and FASTA nucleotide searches using the default values to find homologous hits in the NCBI database.

2.5. Phylogenetic analysis

The nucleotide sequences of the DCHBV variants obtained in this study, as well those obtained from NCBI/GenBank, were aligned using MAFFT v7.490 [23]. This resulted in a total of 72 sequences on which evolutionary analysis could be undertaken. Following sequence alignment, phylogenetic trees were estimated for the polymerase, core, surface and X ORFs separately, utilizing the maximum likelihood method in IQ-Tree v2.1.0, with 1000 SH-like approximate likelihood ratio test (SH-aLRT) and Ultrafast bootstrap (UFboot) replicates to test nodal support, and the nearest neighbour interchange to search for the optimal tree topology [24]. In each case, the ModelFinder program in IQ-Tree v2.1.0 was used to determine the best-fit model of nucleotide substitution [25].

2.6. In situ hybridization

Feline HCC DNA samples that tested positive for DCHBV by PCR were processed for ISH as described previously [17]. Briefly, ISH was performed on 5 μm sections using the Leica BOND RXm platform (Leica Biosystems, Deer Park, IL, USA), DCHBV probe V-FeHepadnavirus, targeting the viral polymerase and X DNA and transcripts, and RNAScope 2.5 LSx Red detection kit (Advanced Cell Diagnostics Inc., Hayward, CA, USA). Positive and negative control probes were feline peptidylprolyl isomerase B (FePPIB) (Cat. No. 455011, Advanced Cell Diagnostics), and dihydrodipicolinate reductase (DapB) of Bacillus subtilis (Cat. No. 312038, Advanced Cell Diagnostics). Slides were stained with the manufacturers 2.5 Red protocol.

2.7. Sequencing for DCHBV integration sites

Two approaches, specifically targeted sequencing and whole genome sequencing (WGS), were used to investigate DCHBV integration sites. For targeted sequencing, FFPE liver-derived DNA was available from 16 HCC biopsies that tested DCHBV-positive by PCR and 4 DCHBV-negative HCCs. A total of 5 ng of total nucleic acid extract was used to prepare libraries using the EF Library preparation kit (version 1, Twist Biosciences, CA, USA) and dual indexes (TruSeq compatible Twist UMI Adapter System, Twist Biosciences). Minor modifications were made to the manufacture's library preparation protocol. Briefly, fragmentation time was reduced to 2 min and 15 s, and a total of 16 amplification cycles was used during library amplification. DNA library concentration was assessed using the Qubit dsDNA High Sensitivity Quantitation Assay (Thermo Fisher Scientific) on the Qubit 4 Fluorometer (Thermo Fisher Scientific) and up to 96 libraries were pooled in a total of 2 ml for multiplex hybridisation. Library pools were dried down for 17 h prior to standard hybridisation with the Twist Comprehensive Viral Research Panel (Twist Biosciences). This panel contains probes that capture the complete genome of 3153 viruses, including the DCHBV genome. Following post-capture amplification and purification, captured libraries were quantified using the Qubit dsDNA High Sensitivity Quantitation Assay (Thermo Fisher Scientific) on Qubit 4 Fluorometer (Thermo Fisher Scientific) and library fragment size was assessed using the Bioanalyzer High Sensitivity D5000 ScreenTapes (Agilent). The molarity of each capture pool was determined and, if required, additional post capture pools were combined in equimolar concentrations prior to sequencing using 300 cycle Illumina sequencing chemistry. Sequencing was performed with an expected 1.5 million reads per library.

WGS was also performed on FFPE liver-derived DNA from 8 DCHBV-positive HCCs. DNA extracted from these tissues yielded a gDNA peak >600 bp on Genomic DNA ScreenTape Analysis version 5.1 (Agilent Technologies, Inc., Santa Clara, CA, USA). Illumina DNA tagmentation was used for DNA library construction (Illumina, San Diego, CA, USA), with the input DNA normalized to 500 ng per sample. WGS was performed on the Illumina NextSeq 2000 sequencer using P3 and P4 flow cells with XLEAP-SBS reagents (Illumina).

In both WGS and targeted sequencing libraries, Trim Galore version 0.6.7 was used to remove low-quality bases (<Q20) and to remove adapter sequences from the raw FASTQ files [26]. For WGS, MEGAHIT version 1.2.9 (https://hpc.ilri.cgiar.org/megahit-software) was used to assemble the trimmed, paired-end Illumina reads into contigs, under default settings. Trimmed sequences were mapped to the reference genome (NC_040719.1 DCHBV isolate Sydney2016, complete genome). Non-DCHBV reads were filtered by aligning to the DCHBV genome using Bowtie2, employing the following settings “--no-unal --local --very-sensitive-local”. Isolation of feline-viral chimeric reads was performed by aligning filtered reads to the feline genome (F.catus_Fca126_mat1.0, NCBI RefSeq assembly: GCF_018350175.1 [27]). A custom Python script was developed to analyse chimeric reads in order to identify DCHBV integration and to BLAST (NCBI) against both DCHBV and the feline genome. Integration junctions with more than 5 supporting reads were considered as true positives. Sequencing results were compiled and the genomic annotation of both viral and feline sequences were added using a second Python script (by calling HOMER's AnalyzePeak.pl script [28]). DCHBV integrations found with each sequencing method can be found in the supplementary data (Tables S1 and S2).

2.8. Statistical analysis

To assess the matching efficiency of HCC cases and lymphocytic cholangitis controls, potential associations between patient characteristics (sex, neuter status, breed) and the outcome variable (case or control) were evaluated using Chi-square tests (Table 1). The age distributions of cases and controls were assessed from violin plots. The average age of cats was compared between the cases and controls using a two-sample T-test. A Fisher's exact test was conducted to evaluate the association between DCHBV detection (PCR -positive or negative) and HCC (being case or control). Among HCC cases, the associations between the variables of interest (age, sex, breed, neuter status and location of sample submission) and DCHBV PCR result were evaluated using Chi-square or Fisher's exact tests, and the average age was compared between the DCHBV-positive and DCHBV-negative cases using a two-sample T-test.

Table 1.

Patient characteristics and comparison for hepatocellular carcinoma cases and lymphocytic cholangitis controls arising in 159 cats.

Variable	Categories	Hepatocellular carcinoma cases (n = 71)	Lymphocytic cholangitis controls (n = 88)	P-valuea
Sex	Female	31 (43.7 %)	43 (48.9 %)	0.513
	Male	40 (56.3 %)	45 (51.1 %)
Neuter status	Neutered	62 (87.3 %)	80 (90.1 %)	0.467
	Intact	9 (12.7 %)	8 (9 %)
Breed	Mixed breed	62 (91.1 %)	74 (85 %)	0.249
	Purebred	6 (8.8 %)	13 (14.9 %)

^aP-values are from Chi-square tests.

3. Results

3.1. Characteristics of the study groups

3.1.1. Hepatocellular carcinoma - cases

A total of 71 cases of HCC (cases) from the USA (n = 31 total, submitted in New York [n = 13], Texas [n = 12 ] and North Carolina [n = 6]), UK (n = 26) and Taiwan (n = 14), were confirmed on histological review. Age was available for 66 of these cases (Fig. 1). The mean age was 151 months, standard deviation (SD) 37 months, ranging from 48 to 228 (median = 156). The sex, neuter status and breed of HCC cases are summarized in Table 1.

Fig. 1.

The age distribution of cats diagnosed with hepatocellular carcinoma (HCC; n = 66) or lymphocytic cholangitis (LC; n = 88) are presented using Violin plots. The median ages are represented by hollow circles within the interquartile range (bold lines).

3.1.2. Lymphocytic cholangitis - controls

A total of 88 lymphocytic cholangitis biopsies (Controls) were confirmed on histological review (submitted in USA [n = 68), and UK [n = 20]). The mean age of these cats was 120 months (SD = 39), ranging between 15 and 191 months (median = 123) (Fig. 1). Variables; sex, neuter status and breed of lymphocytic cholangitis controls are summarized in Table 1.

3.2. Matching of case to control groups

No significant differences were identified between case and control groups with respect to sex, neuter-status, or breed (Table 1). The mean age of the HCC cases (151 months; SD = 37) was significantly greater than the mean age of the lymphocytic cholangitis control group (120 months; SD = 39) using the T-test (P < 0.001). Although age-matching was not achieved, the age range of DCHBV-positive HCC cases fell within that of controls, so any effect on interpretation is expected to be small.

3.3. Comparison of feline HCC cases testing positive or negative for DCHBV by PCR

Notably, DCHBV DNA was amplified from 17/71 (23.9 %) HCC cases and 0/88 (0 %) lymphocytic cholangitis controls, indicating a significant difference between the two groups (P < 0.001, Fisher's exact test). DCHBV DNA was detected significantly more often in HCC arising in females than males (P = 0.045; Table 2). None of breed, neuter status or sample submission location were significantly associated with the detection of DCHBV in HCC cases (Table 2). The mean age of 16 cats with DCHBV-positive tumours was 150 months (SD = 17), ranging between 120 and 180 months, and for 50 cats with DCHBV-negative tumours was 151 (SD = 41), ranging from 48 to 228 months (Fig. 2). There was no significant difference in the mean age of DCHBV-positive versus DCHBV-negative HCC (P = 0.917 using the T-test).

Table 2.

Patient characteristics for 71 feline hepatocellular carcinoma cases that tested positive or negative for DCHBV DNA on PCR.

Variable	Categories	DCHBV-positive	DCHBV-negative	P-value
Sex	Female	11 (35.5 %)	20 (64.5 %)	0.045a
	Male	6 (15 %)	34 (85 %)
Neutered status	Neutered	14 (22.6 %)	48 (77.4 %)	0.439b
	Intact	3 (33.3 %)	6 (66.7 %)
Breed	Mixed breed	16 (26.2 %)	46 (73.8 %)	0.999b
	Purebred	1 (14.3 %)	5 (85.7 %)
Location of submission	USA	4 (12.9 %)	27 (87.1 %)	0.084a
	UK	7 (26.9 %)	19 (73.1 %)
	Taiwan	6 (42.9 %)	8 (57.1 %)

^aP-value from Chi-square test.

^bP-value from Fisher's exact test.

Fig. 2.

The age distribution of cats diagnosed with hepatocellular carcinoma (n = 66), by DCHBV PCR status (positive = 16; negative = 50) are presented using violin plots. The median ages are represented by hollow circles within the interquartile range (bold lines).

3.4. DCHBV In situ hybridization

Of 17 DCHBV-positive HCCs, 11 yielded a positive hybridization signal (Fig. 3). Of these, signal was present within the tumor in nine cases, and in the adjacent liver parenchyma in two cases. Strong signal was often present in clusters and islands of hepatocytes. In the non-tumor parenchyma, these were often located in periportal areas and extending into the remainder of the parenchyma (Fig. 3). Appropriate signals with positive (feline PPIB) and negative (DapB) control probes were obtained for all samples. While ISH is a highly sensitive method for viral DNA detection, a smaller number of cells is sampled than by PCR, which may account for DCHBV PCR-positive, ISH-negative cases (6/17).

Fig. 3.

Domestic cat hepatitis B virus in situ hybridization. A and B: Case fel-HCC-TW4. Hepatocellular carcinoma with strong cytoplasmic and nuclear signal (red) throughout the neoplastic cells. C and D: Case fel-HCC-UK3. Non-tumor parenchyma adjacent to a hepatocellular carcinoma. Strong cytoplasmic and nuclear signal in clusters of hepatocytes, most prominent in periportal areas. A and C bar = 250 μm; B and D bar = 25 μm.

3.5. Phylogenetic analysis of DCHBV sequences

This study contributed seven DCHBV sequences, representing 83 %–100 %, of the genome obtained from feline HCCs submitted in USA (GenBank accession numbers PQ206354, PQ480066, PQ480067), Taiwan (PQ480065, PQ206355) and UK (PQ999195, PQ858708). Our phylogenetic analysis of the four ORFs from DCHBV revealed that the HCC cases here fell into two distinct lineages indicative of independent transmission events. Specifically, all but one of the HCC cases fell in the large cluster of DCHBV sequences, denoted genotype A, found in cats globally (i.e. the top clade in Fig. 4). Within this large cluster, three of the HCC sequences (PQ206355, PQ480066, PQ480065 from cases fel-HCC-TW5, fel-HCC-USA2 and fel-HCC-TW2, respectively) were very closely related and formed a monophyletic group in the polymerase and surface phylogenies, compatible with direct transmission among them, while the other sequences PQ858708 (case fel-HCC-UK6), PQ999195 (case fel-HCC-UK3) and PQ480067 (fel-HCC-USA3) fell on separate branches such that they have independent origins. In marked contrast, sequence PQ206354 (case fel-HCC-USA1) fell within a phylogenetically distinct lineage, genotype B, with four other viruses, three sampled in Brazil, and one in Japan, and was also the most divergent lineage within that group, representing the split between the A and B genotypes, with branch lengths scaled to the number of nucleotide substitutions per site.

Fig. 4.

Maximum likelihood phylogenetic trees of nucleotide sequences for each ORF (polymerase, surface, core and X protein) of DCHBV. Coloured circles at the branch represent the sequence country of origin and the black full and half circle at the node represent SH-aLRT and UFboot values. The sequences described in this study are highlighted with a red star. All trees are mid-point rooted, representing the split between the A and B genotypes, with branch lengths scaled to the number of nucleotide substitutions per site.

3.6. DCHBV integration site analysis

Utilizing magnetic capture-based enrichment enabled targeted sequencing at reduced depth. DCHBV-specific reads were detected in 13 out of 16 DCHBV-positive HCCs, with chimeric reads (containing both DCHBV and host sequences) detected in 11 of these (Table S1). As expected, no integration junctions were found in any of 4 DCHBV-negative HCCs, showing the specificity of this approach. The sequences can be accessed using SRA BioProject accession number PRJNA1293714.

DCHBV integration events identified using targeted sequencing were confirmed with WGS of the 8/8 samples investigated (Table S2). Despite greater reads per sample using WGS (mean of 478M compared with 3.3M paired-end reads using targeted sequencing, Fig. S1A), fewer DCHBV-specific reads were detected by WGS (mean 405 vs 5,088, Fig. S1B). Importantly, junctions detected by both sequencing approaches were essentially identical with respect to host coordinates and sites of the DCHBV junction.

Using a cut-off of five supporting reads, a total of 52 DCHBV integration junctions were identified by either method (Tables S1 and S2), with 36 of the 52 junctions represented by either an upstream or downstream junction alone. Putative upstream and downstream junctions of the same integration events were identified in eight instances (Fig. S2A). Two of these eight instances were found to have the junctions with discordant orientations (Fig. S2B), suggestive of viral inversion events.

Consistent with integration of other hepadnaviruses, both upstream and downstream junctions clustered within ∼50bp of 1710 nt the putative epsilon site of DCHBV (Fig. 5A and B) [[29], [30], [31], [32]]. Thus, for most integrations, the promoter and ORF of the DCHBV surface gene were present and intact, suggesting that integrated forms could contribute to surface antigen expression as observed in human HBV infections [33,34]. There was also additional enrichment of upstream junctions between 2038 nt and 2247 nt, suggesting additional forms or pathways of DCHBV DNA integration other than the canonical dslDNA forms.

Fig. 5.

Analysis of DCHBV DNA integration junctions with respect to viral and host genomes. Distribution of (A) downstream and (B) upstream integration junctions. Putative epsilon site highlighted with a line. The region in which deletions had been detected in previous studies (suggesting sites where the ends of dslDNA had re-ligated to form covalently-closed circular DNA through illegitimate recombination) are coloured in green [20,31,32]. (C) Genomic distribution of DCHBV DNA integration sites compared to feline genome. Integration events were enriched by a factor of 2.4 in either introns or exons. (D) DCHBV DNA integrations were mapped to each chromosome, normalized to the chromosome length and number of integrations found. Significant enrichment was observed in chromosome E2 (highlighted in red).

With respect to the host genome, DCHBV integrations were ∼2.4-fold more likely to occur into both introns and exons, after normalizing for the functional composition in feline genome (Fig. 3C). While most DCHBV integrations were distributed across the entire genome, enrichment was observed in chromosome E2 (Z-score = 3.15, Fig. 3D). This enrichment was driven by five unique integration events inferred by nine integration junctions in the vicinity of the Cyclin E1 (CCNE1) gene (Fig. 6), suggesting a mechanistic role of DCHBV integration in liver cancer. Two integration sites were present in the intergenic region between CCNE1 and CE2H19orf12 (located ∼60 kb downstream of CCNE1). Two integrations were located within the promoter/UTR region of CCNE1. Further in-depth analysis of the chimeric reads found an additional integration within the fourth intron of CCNE1, which was initially excluded due to having only two supporting reads. Thus, we find evidence of enriched, repeated, independent integrations in the CCNE1 locus in liver tissues of DCHBV-positive cats, suggesting an underlying mechanism for DCHBV infection causing liver disease.

Fig. 6.

DCHBV DNA integrations within CCNE1. The cat CCNE1 and CE2H19orf12 genes in chromosome E2 are shown in red. CCNE1 exon and introns are represented in dark and light green, respectively, and gene orientation represented with grey arrows. Nine junctions from 5 unique integrations near and within CCNE1 from different samples represented. Junctions 1–2, 3–4, 5–6, and 8–9 represent junctions of the same integration, respectively. The feline genome is shown in brown and the DCHBV DNA sequence is shown in blue.

4. Discussion

This study identifies DCHBV as a risk factor for HCC in cats through both epidemiological and molecular approaches. In addition, we show for the first time that DCHBV integrates into the host genome and shows enrichment in putative proto-oncogenes, a strategy used by human HBV to promote liver cancer in humans [35].

Building on a previous report that detected DCHBV in HCC biopsies, but not in various non-HCC lesions or histologically normal liver, the current study used a larger, independent sample set to increase statistical power, and a control group with a single diagnosis, lymphocytic cholangitis [17]. ISH localised DCHBV nuclei acids in the cytoplasm and nucleus of neoplastic hepatocytes, with the signal distribution, intensity, and pattern matching the findings of Pesavanto et al. (2019) [17]. Combining the results of this study with that of Pesavento et al. (2019), which was otherwise conducted similarly, suggests that DCHBV could contribute to 25 % of feline HCCs [17]. By comparison, HBV causes around 50 % of HCC in people, whereas the contribution of rodent hepadnaviruses to liver cancer in natural infection is not readily discernible [36].

The potential burden of disease from DCHBV in cats requires clarification. HCC is the most common non-haematopoietic hepatic malignancy in cats [37], but clinical cases are recognized infrequently, and primary liver cancer represents only 1–3 % of all feline malignancies in published studies [[37], [38], [39]]. However, HCC may escape detection in cats. For example, owners may be unwilling or unable to support diagnostic investigations. In addition, clinical signs may be absent in some cases, which is suggested by reports of HCC as an incidental finding at necropsy [38].

DCHBV-associated HCCs were identified in older cats (mean age ∼12.5 years), which is consistent with previous reports [38]. However, DCHBV infection is commonly detected in young cats (<2 years) [40]. Thus, if DCHBV is indeed a causative factor in HCC, then this carcinogenic process may occur over an extended time period, similar to human HBV-associated HCC, and contrasting WHV-associated HCC in woodchucks, which usually occurs within 2 years in chronic infection [2,6]. This suggests that additional factors are involved in inducing DCHBV-associated liver cancer, such as hepatic inflammation. On this point, early evidence suggests that chronic hepatitis, a common prequel to HBV-associated HCC in humans, may be associated with DCHBV infection in cats [17,41]. Moreover, histopathological changes seen in viral hepatitis in people but previously considered uncommon in cats, such as interface hepatitis, are reported in DCHBV-infected cats [17,20,41].

No sex predisposition for developing HCC has been reported in cats and none was apparent here [38]. However, the risk of DCHBV-positive HCC was higher in females than in males. Most cats in this study had undergone ovariohysterectomy or castration, surgical procedures that are recommended routinely for pets. In humans, oestrogen and testosterone have been implicated as negative and positive risk factors, respectively, for HCC [36,42,43]. If sex hormones influence HCC risk similarly in cats and humans, the absence of these hormones post-neuter surgery could increase the risk of virus associated-HCC in females compared with males, as observed in this study.

The proportion of DCHBV-associated HCC was higher among biopsies submitted in Taiwan than in the USA or UK. Interestingly, the reported prevalence of DCHBV viraemia in cats in Taiwan, around 11 %, is higher than that reported in the USA (<1 %) or the UK (2 %) [20,32,44,45]. HBV endemicity and disease burden vary geographically, and natural exposure of woodchucks to WHV varies from 0 to 60 % in different US states [6,36]. Further work is needed to clarify how DCHBV endemicity and disease burden might vary geographically.

Also of note was that our study identified a fifth genotype B virus, and the first from a cat sampled in the USA, although the provenance of this case could not be confirmed. More broadly, the phylogenetic pattern observed here, particularly the presence of long branch lengths, is strongly suggestive of currently hidden genetic diversity of DCHBVs. Future studies are needed to understand the extent of DCHBV diversity and whether genotype influences disease risk, as is the case with HBV infection [46].

Our discovery that DCHBV integrates into the feline genome suggests that DCHBV has the potential to drive hepatocyte transformation via pathways similar to those used by hepatitis B (HBV) and woodchuck hepatitis virus (WHV), including insertional mutagenesis and chromosomal instability [30,35]. The DCHBV integration sites identified matched the expected termini of the putative double-stranded linear DNA (dslDNA) intermediates, given the relative location of viral direct repeat sequences [31]. This also corresponds to a region in which deletions have been detected in previous studies of DCHBV, putatively representing the ends of dslDNA re-ligating to form covalently-closed circular DNA [20,32]. Together, these results support the existence of a dslDNA form of DCHBV with the capacity to integrate, consistent with other hepadnaviruses [[29], [30], [31]]. Putative viral inversion events in integrated DCHBV sequences resemble those reported in HBV integration studies using long-read sequencing technologies [47]. We speculate, therefore, that integrated DCHBV DNA could provide a persistent source of surface protein, as has been shown for human HBV [33,34]. Further research into the fundamental virology of DCHBV is needed to completely characterize the similarities and differences of viral replication compared to other hepadnaviruses.

Repeated DCHBV integration within or near the promoter of the feline CCNE1 homologue in several feline HCC cases raises the possibility that DCHBV may act by insertional mutagenesis, driving CCNE1 overexpression. CCNE1 encodes cyclin E1 which is required for cell cycle G1/S phase transition [48]. Overexpression of CCNE1 and subsequent dysregulation of cell cycle control occurs in ovarian cancers and other human malignancies [48]. HBV integrates throughout the genome but, importantly, CCNE1 has been identified as a recurrent integration target in HBV-associated HCC [49]. In contrast, WHV-associated HCCs almost always feature viral integrations in association with myc oncogenes that trigger rapid tumor development [50]. It is possible that the apparently common, but not universal, CCNE1 association reported here for DCHBV integration lies somewhere between the HBV-WHV spectrum in driving oncogenesis through insertional mutagenesis following viral DNA integration.

Our findings add integration and insertional mutagenesis to the long list of similarities between DCHBV and HBV, which are already known to share comparable genomic structures and size, putative entry receptor, and heterogeneous global distribution [12,15,32]. Given these similarities, a unique opportunity exists to study DCHBV infection in cats to inform both human and veterinary medicine. The value of cats as human companions drives a global market for species-specific diagnostics and therapeutics, while HBV medicine seeks new approaches to treat chronic infection and prevent hepadnavirus-associated diseases. Of relevance here is the natural susceptibility of cats to the immunosuppressive lentivirus, feline immunodeficiency virus (FIV), presenting additional opportunities to understand viral interactions pertinent to HBV-human immunodeficiency virus coinfection [51,52]. In addition, reverse transcriptase inhibitors, a mainstay of HBV treatment, have been used “off-label’ for specific indications in feline patients, reviewed in Ref. [15]. Novel DCHBV-specific tools, including diagnostic panels to correlate with clinical data in naturally-infected cats, as well custom cell lines and organoids, could to support cross-disciplinary collaborations aiming to unravel the translational and reverse-translational opportunities that might emerge from DCHBV infection of cats.

CRediT authorship contribution statement

João P. Cavasin: Writing – review & editing, Supervision, Resources, Methodology, Investigation. Min-Chun Chen: Writing – review & editing, Methodology, Investigation, Formal analysis, Data curation. Harout Ajoyan: Writing – review & editing, Validation, Methodology, Investigation, Formal analysis, Data curation. Melanie J. Dobromylskyj: Writing – review & editing, Resources. Wei-Hsiang Huang: Writing – review & editing, Resources. Mason Jager: Writing – review & editing, Resources. Kate Van Brussel: Writing – review & editing, Formal analysis. Rebecca Rockett: Writing – review & editing, Investigation. Omid Nekouei: Writing – review & editing, Formal analysis. Penny Watson: Writing – review & editing, Resources. Jason Bestwick: Writing – review & editing, Resources. Yan Ru Choi: Writing – review & editing, Investigation. Jonathan A. Lidbury: Writing – review & editing, Supervision, Methodology. John M. Cullen: Writing – review & editing, Supervision, Resources. Edward Holmes: Writing – review & editing, Supervision, Formal analysis. Jörg M. Steiner: Writing – review & editing, Supervision, Project administration, Conceptualization. Thomas Tu: Writing – review & editing, Validation, Supervision, Methodology, Formal analysis, Data curation. Julia A. Beatty: Writing – review & editing, Writing – original draft, Supervision, Project administration, Conceptualization.

Financial support

No external funding was received for this study. Partial support for the study was provided by grant from CityUHK to JB (No. 9380111).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data relevant to this study are presented herein or available on publicly available databases, as indicated