A versatile method to profile hepatitis B virus DNA integration

Abstract

Background:

HBV DNA integration into the host genome is frequently found in HBV-associated HCC tissues and is associated with hepatocarcinogenesis. Multiple detection methods, including hybrid capture-sequencing, have identified integration sites and provided clinical implications; however, each has advantages and disadvantages concerning sensitivity, cost, and throughput. Therefore, methods that can comprehensively and cost-effectively detect integration sites with high sensitivity are required. Here, we investigated the efficiency of RAISING (Rapid Amplification of Integration Site without Interference by Genomic DNA contamination) as a simple and inexpensive method to detect viral integration by amplifying HBV-integrated fragments using virus-specific primers covering the entire HBV genome.

Methods and Results:

Illumina sequencing of RAISING products from HCC-derived cell lines (PLC/PRF/5 and Hep3B cells) identified HBV-human junction sequences as well as their frequencies. The HBV-human junction profiles identified using RAISING were consistent with those determined using hybrid capture-sequencing, and the representative junctions could be validated by junction-specific nested PCR. The comparison of these detection methods revealed that RAISING-sequencing outperforms hybrid capture-sequencing in concentrating junction sequences. RAISING-sequencing was also demonstrated to determine the sites of de novo integration in HBV-infected HepG2-NTCP cells, primary human hepatocytes, liver-humanized mice, and clinical specimens. Furthermore, we made use of xenograft mice subcutaneously engrafted with PLC/PRF/5 or Hep3B cells, and HBV-human junctions determined by RAISING-sequencing were detectable in the plasma cell-free DNA using droplet digital PCR.

Conclusions:

RAISING successfully profiles HBV-human junction sequences with smaller amounts of sequencing data and at a lower cost than hybrid capture-sequencing. This method is expected to aid basic HBV integration and clinical diagnosis research.

INTRODUCTION

Over 290 million individuals are carriers of the HBV worldwide, and an estimated 820,000 annual deaths were caused by HBV-related diseases such as cirrhosis and HCC in 2019.¹ HBV is a small 3.2 kb DNA virus that specifically infects hepatocytes and can integrate into the human genome. Several studies have revealed that HBV DNA integration to the host genome occurs in 75%–90% of HBV-related HCC tissues and is closely related to the development of HCC.^2–4 However, viral integration is not essential for producing new virions and is thought to be a byproduct of viral replication. Numerous studies have demonstrated that viral integration induces chromosomal instability, insertional mutagenesis, cis-mediated activation of downstream genes, and persistent expression of viral proteins such as hepatitis B x protein and hepatitis B surface proteins.^3,5–11 Moreover, HBV DNA integration explains why HBsAg loss is difficult to achieve by nucleo(t)ide analog therapy.¹² Despite many studies analyzing clinical specimens, the molecular mechanisms involved in HBV DNA integration and its exact pathological significance remain unclear due to the lack of convenient methods to analyze HBV DNA integration.

Various methods have been adopted to detect viral genome integration; however, each has advantages and disadvantages.^12,13 Classical Southern blotting is an inexpensive method to visualize chimeric DNA resulting from the integration sites, albeit with poor sensitivity and labor intensiveness. PCR-based methods such as Alu-PCR and inverse nested PCR are highly sensitive but not comprehensive, depending on the Alu-sequence and restriction enzyme sites. Next-generation sequencing (NGS) technologies have enabled the analysis of the viral genome integration process and its implications. Hybrid capture-sequencing, the most commonly used method, enables a highly sensitive and comprehensive analysis by enriching integrants with capture probes tiling the whole HBV genome.^12,14 However, its high cost limits its versatility and hampers the progress of research in this field. A recent report that the remnant of integration-derived chimeric DNA in plasma cell-free DNA (cfDNA) correlates with the recurrence of HCC¹⁵ has raised the possibility that detecting integration possesses a diagnostic utility to monitor residual tumor and recurrence after tumor resection. Therefore, methods that can comprehensively and cost-effectively detect integration sites with high sensitivity are required as a useful tool for basic research and a promising diagnostic tool.

Recently, we developed a highly sensitive, simple, and inexpensive PCR-based method—Rapid Amplification of Integration Site without Interference by Genomic DNA contamination (RAISING)—to detect the integration of retroviruses [human T-cell leukemia virus type-1 (HTLV-1) and bovine leukemia virus] using viral genome-specific primers.^16–18 In the present study, we aimed to detect HBV DNA integration using RAISING by amplifying HBV-integrated fragments using virus-specific primers covering the entire HBV genome. We comprehensively determined the viral integration sites in HBV-associated HCC-derived cell lines, as well as de novo integration sites in HBV-infected cells, liver-humanized mice, and clinical specimens. Engrafting the immunodeficient mice with the cell lines, HBV-human junction DNA, identified by the RAISING-based sequencing, was detectable from their plasma cfDNA, suggesting its potential usefulness in the clinical diagnosis of HCC. Our study provides a novel and powerful tool for detecting HBV DNA integration.

METHODS

Rapid Amplification of Integration Site without Interference by Genomic DNA contamination

RAISING to detect HBV DNA integration (Figures 1A, B) was performed according to the protocol for HTLV-1¹⁷ with slight modifications. Briefly, 200 or 500 ng of total DNA was treated with RNase A, followed by single-stranded DNA synthesis using KOD-Plus-Neo polymerase (TOYOBO) and virus-specific F1 primers. After column purification using the Monarch PCR & DNA Cleanup Kit (New England Biolabs), the product was polyAG-tailed using Terminal Transferase (New England Biolabs), followed by second-stranded DNA synthesis using Q5 Hot Start High-Fidelity DNA Polymerase (New England Biolabs) and oligo (dT) primer with an anchor sequence. After the first PCR by adding viral F2 primers, nested PCR was performed using KOD-Plus-Neo polymerase, viral F3 primers, and Illumina adaptor-tagged anchor sequence primer. HBV-specific primers (listed in Supplemental Tables S1–S3, http://links.lww.com/HC9/A682) were designed so that they cover the whole HBV genome. The products were visualized by electrophoresis on a 2% agarose gel, or purified using the Wizard SV Gel and PCR Clean-Up System (Promega) for subsequent sequencing.

FIGURE 1.

Workflow of RAISING targeting HBV DNA integration. (A) Schematic representation of the workflow for identifying HBV-human chimeric genome sequences. HBV-integrated fragments are amplified by RAISING, and then the products are sequenced with NGS. (B) Workflow of RAISING. First, the elongation of chimeric single-stranded DNA was achieved using HBV-specific F1 primers, followed by poly A tailing (with poly AG extension) and second-stranded DNA synthesis by oligo (dT) primer with an anchor sequence. The products were further enriched by nested PCR using HBV-specific F2/F3 and anchor primers with or without an NGS adaptor at 5′. N and V are mixed bases (N: A+C+G+T, V: A+C+G). (C) Experimental design of RAISING samples. Total genomic DNA was extracted from cells, chimeric mouse liver tissues, and liver biopsies of human specimens and analyzed for HBV DNA integration. The profiles of HBV DNA integration were analyzed by RAISING using human HCC-derived cell lines [PLC/PRF/5 (Table 1, upper) and Hep3B cells (Table 1, lower)], cell culture and animal model systems for reproducing de novo HBV infection, and clinical specimens of patients with HBV-related HCC. PLC/PRF/5 and Hep3B cells and clinical specimens were also analyzed by the hybrid capture-sequencing (Tables 1, 2, and Supplemental Table S9, http://links.lww.com/HC9/A682). HepG2-hNTCP-C4 cells and primary human hepatocytes were inoculated with HBV for 16 hours and washed to remove the free virus. The cells were cultured for 12 or 23 days in the presence of 1 µM entecavir and then harvested (Table 3). Human-hepatocyte chimeric mice were intravenously injected with HBV. At 14 weeks after HBV inoculation, antiviral therapy was performed for 8 or 10 weeks for other studies (Supplemental Table S7, http://links.lww.com/HC9/A682). Adjacent nontumor liver biopsies were harvested from patients with HBV-related HCC (Table 4 and Supplemental Table S8, http://links.lww.com/HC9/A682). Abbreviations: ddPCR, droplet digital polymerase chain reaction; NGS, next-generation sequencing; RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

Cell culture

PLC/PRF/5 [obtained from the Japanese Collection of Research Bioresources (JCRB) Cell Bank] and Hep3B cells were cultured at 37°C, 5% CO₂, in growth medium comprising DMEM (high glucose) (Wako) supplemented with 10% fetal bovine serum, 100 units/mL penicillin, 100 µg/mL streptomycin, 10 mM HEPES (pH 7.4), 100 µM nonessential amino acids, and 1 mM sodium pyruvate.¹⁹ HepG2, HepG2-hNTCP-C4, and Hep38.7-Tet cells were cultured at 37°C, 5% CO₂, in a growth medium consisting of DMEM/F-12 + GlutaMax (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum, 100 units/mL penicillin, 100 µg/mL streptomycin, 10 mM HEPES (pH 7.4), and 5 µg/mL insulin.^20,21 Primary human hepatocytes (PXB cells) (PhoenixBio) were cultured at 37°C, 5% CO₂, in dHCGM.²² Entecavir was purchased from Santa Cruz.

Preparation of genomic DNA and cfDNA

Genomic DNA was extracted from cells, mouse liver tissues, and liver biopsies of human specimens using a QIAamp DNA Mini kit (Qiagen) or SepaGene kit (Sekisui Medical) according to the manufacturer’s protocol or with phenol/chloroform followed by isopropyl alcohol precipitation. cfDNA was extracted from mouse plasma using a QIAamp Circulating Nucleic Acid Kit (Qiagen) according to the manufacturer’s protocol.

Illumina sequencing

The libraries for Illumina sequencing were size-selected using SPRIselect beads (Beckman Coulter), targeting 500 bp for MiSeq. Selected libraries were sequenced using MiSeq (v3, 300 bp × 2) according to the manufacturer’s instructions.

NGS data analysis

Sequence reads were trimmed using Sickle (https://github.com/najoshi/sickle) to exclude the sequence reads shorter than 50 nucleotides and/or those with quality scores lower than 20. After data trimming, homology searches of sequence reads were performed using Magic-BLAST (https://ncbi.github.io/magicblast/) to human (GRCh38.p13) and HBV reference sequences (GenBank accession number: EU054331, X02763, GQ477489, AB219428, D00331, GQ924620, GQ358158, AB246345, U95551).²³ We extracted only sequence reads that had an appropriate viral sequence length (30 nucleotides) and high homology with the human genome sequence (≥95% match score). Chimeric sequences supported by 5 or more reads with an overhang size >10 were filtered and indicated as integration sites.

Junction-specific nested PCR

Junction-specific primer sets were designed to span the junction sequences of the host and virus (Supplemental Figure S2, http://links.lww.com/HC9/A682; Supplemental Figure S3, http://links.lww.com/HC9/A682; Supplemental Figure S5, http://links.lww.com/HC9/A682; and Supplemental Table S4, http://links.lww.com/HC9/A682). Nested PCR was performed using the KOD-Plus-Neo polymerase. The PCR products were visualized by electrophoresis on 2% agarose gels and subsequently subjected to Sanger sequencing.

Hybrid capture-sequencing

Library preparation and target capture for HBV sequences were performed using SureSelect XT reagents (Agilent Technologies) according to the manufacturer’s protocol. The libraries were pooled and enriched using the SureSelect DNA capture custom kit (Agilent Technologies). Capture probes were designed to target 9 HBV strains (GenBank accession numbers: EU054331, X02763, GQ477489, AB219428, D00331, GQ924620, GQ358158, AB246345, and U95551). The captured libraries were sequenced using the NovaSeq6000 platform (Illumina) with paired-end reads of 150 bp according to the manufacturer’s instructions.

HBV preparation and infection

HBV (genotype D) derived from the culture supernatant of Hep38.7-Tet cells cultured in the absence of tetracycline was prepared as described.²⁴ The HBV was inoculated into HepG2-hNTCP-C4 cells at 12,000 [Figure 1C and Table 3 (upper)] or primary human hepatocytes at 8000 [Figure 1C and Table 3 (lower)] genome equivalents per cell in the presence of 4% polyethylene glycol 8000 (PEG 8000) (Sigma-Aldrich) at 37°C for 16 hours, as described.²⁵

TABLE 3.

HBV-human junctions in HBV-infected HepG2-hNTCP-C4 cells and primary human hepatocytes determined by RAISING-sequencing

Sample	No. total reads	HBV junction position	HBV genome	Human chromosome	Human junction position	Human gene
HepG2-hNTCP-C4 cells	13	3166	P, S	Chr12	120995394	HNF1A
	11	3165	P, S	Chr11	5946294	OR56A3
	11	3180	P, S	Chr12	120995632	HNF1A
	10	3166	P, S	Chr12	120995816	HNF1A
	10	3180	P, S	Chr12	120994967	HNF1A
	9	3168	P, S	Chr12	120995765	HNF1A
	9	3180	P, S	Chr7	7705852	RPA3
	8	503	P, S	Chr1	630538	AL669831.3
	8	1242	P	Chr6	12017287	HIVEP1
	7	496	P, S	Chr15	50125750	ATP8B4
	7	1760	X	Chr2	16922581	(intergenic)
	5	341	P, S	Chr1	629360	AL669831.3
	5	505	P, S	Chr7	158071793	PTPRN2
	5	1718	X	Chr4	177918044	LINC01098
	5	3180	P, S	Chr5	134925420	PCBD2
	5	3180	P, S	ChrX	50742917	SHROOM4
Primary human hepatocytes	61	1496	P, X	Chr2	27496085	(intergenic)
	15	1829	C, X	Chr14	103430901	MARK3
	11	1812	X	Chr2	234648039	(intergenic)
	5	985	P	Chr8	21705351	GFRA2

Abbreviation: RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

Human-hepatocyte chimeric mice

Sixteen-week-old human-hepatocyte chimeric mice (PXB mice) (PhoenixBio), generated by transplantation of human hepatocytes into uPA/SCID mice,²⁶ were intravenously injected with 10⁶ copies/body of HBV genotype C (Figure 1C and Supplemental Table S7, http://links.lww.com/HC9/A682). They were ethically sacrificed 22 or 24 weeks after HBV infection. This animal experiment was performed in strict accordance with both the Guide for the Care and Use of Laboratory Animals and the guidelines stipulated by the local Committee for Animal Experiments at PhoenixBio and Hamri. HBV (genotype C) was purchased from PhoenixBio.

Clinical specimens

Adjacent nontumor liver biopsies were obtained from patients diagnosed with HBV-related HCC at the Division of Gastroenterology and Hepatology, Department of Internal Medicine, the Jikei University School of Medicine, and the Division of Hepatobiliary and Pancreatic Disease, Department of Internal Medicine, Hyogo Medical University. All samples were collected after obtaining written informed consent and analyzed with the approval of the ethics committees of the Jikei University School of Medicine, Hyogo Medical University, and the National Institute of Infectious Diseases. This study was conducted in accordance with both the Declarations of Helsinki and Istanbul.

Xenograft mouse models

Four-week-old female NOD/SCID mice (Jackson Laboratory Japan) were housed under pathogen-free conditions. To prepare xenograft mouse models, PLC/PRF/5 or Hep3B cells (5×10⁶ cells/head) suspended in a 1:1 mixture of cell culture medium and Cultrex Basement Membrane Extract (R&D Systems) were subcutaneously injected into the right flank of each mouse. After 6–11 weeks, when the tumor volume reached approximately 1000 mm³, whole blood was collected and randomly pooled from 2 to 3 heads. Tumor volumes were calculated as length × width × width × 0.5.²⁷ All procedures were performed in accordance with the guidelines and with the approval of the Animal Care and Use Committee of the National Institute of Infectious Diseases (approval no. 122083).

Droplet digital PCR

Primers and TaqMan probes were designed based on the HBV-human junction sequences in PLC/PRF/5 and Hep3B cells, as determined by RAISING-sequencing (Supplemental Table S5, http://links.lww.com/HC9/A682). cfDNA was subjected to QX200 droplet digital PCR (ddPCR) (Bio-Rad) as described.¹⁵

RESULTS

Detection of HBV DNA integration in tumor-derived cell lines by RAISING

We applied the RAISING to detect HBV DNA integration in HBV-associated human HCC-derived cell lines, PLC/PRF/5 and Hep3B cells [Figure 1C (top)], containing HBV integrants without producing virions. To detect viral integration comprehensively, the primer set (nos. 1–14) used at first step of RAISING was designed to cover the entire HBV genome (Supplemental Figure S1A, http://links.lww.com/HC9/A682). The RAISING products amplified from the PLC/PRF/5 or Hep3B cells are shown in Supplemental Figure S1B-a, http://links.lww.com/HC9/A682 and Supplemental Figure S1C-a, http://links.lww.com/HC9/A682. The RAISING products appeared as smeared bands by agarose gel electrophoresis because the length of the final products varies according to the termination of single-stranded DNA synthesis as RAISING adapts to single-stranded DNA synthesis using virus-specific primers. The products were further Sanger sequenced to confirm the presence of HBV-human junction sequence. In the sequence amplified by primer no.13 from PLC/PRF/5 cells, we found a chimeric sequence comprising HBV and human genome chromosome 17 (Supplemental Figure S1B-b, http://links.lww.com/HC9/A682). The determined junction was consistent with those previously reported using NGS technology.^28–31 Similarly, the product amplified by primer no. 8 from Hep3B cells was a chimera of HBV and human genome chromosome 13 (Supplemental Figure S1C-b, http://links.lww.com/HC9/A682), which coincides with the results of a previous report.³¹ We further analyzed these RAISING products using NGS technology (MiSeq platform). As shown in Table 1, NGS identified various HBV-human junctions in many human chromosomes and HBV genes in PLC/PRF/5 and Hep3B cells. These results indicate that RAISING could detect HBV DNA integration in HCC-derived cell lines.

TABLE 1.

HBV-human junctions in PLC/PRF/5 and Hep3B cells determined by RAISING-sequencing and hybrid capture-sequencing

Sample	Detection method	No. total reads	HBV junction position	HBV genome	Human chromosome	Human junction position	Human gene	% of total chimeric reads		(Figure 2B)
PLC/PRF/5 cells	RAISING-sequencing	3804	1391	P, X	Chr5	1297478	(intergenic)	0.92	*	→ #1
		3577	491	P, S	Chr17	82105786	CCDC57	0.87	**	→ #2
		1602	1408	P, X	Chr3	131451702	(intergenic)	0.39	***	→ #3
		1465	817	P, S	Chr12	109573899	MVK	0.36	****	→ #4
		935	62	P, S	Chr4	180586417	(intergenic)
		475	2138	C	Chr13	33088561	(intergenic)	0.12	*****
		58	257	P, S	Chr16	70170602	AC009060.1
		31	257	P, S	Chr16	69947689	(intergenic)
		9	919	P	Chr5	8736559	(intergenic)
		9	1719	X	Chr19	16239292	AP1M1
		7	449	P, S	Chr5	78401537	SCAMP1
		6	271	P, S	Chr11	13870025	AC027779.1
		6	926	P	Chr18	3719830	DLGAP1
		6	937	P	Chr16	69957337	CLEC18A
	Hybrid capture-sequencing	57,356	491	P, S	Chr17	82105786	CCDC57	0.08	**
		35,788	637	P, S	Chr4	180587608	(intergenic)
		24,825	2138	C	Chr13	33088561	(intergenic)	0.04	*****
		23,527	817	P, S	Chr12	109573899	MVK	0.03	****
		20,135	1408	P, X	Chr3	131451702	(intergenic)	0.03	***
		18,677	1391	P, X	Chr5	1297478	(intergenic)	0.03	*
		12,291	3081	P, S	Chr1	143222538	(intergenic)
		3128	3081	P, S	Chr1	143240684	(intergenic)
		1614	3081	P, S	Chr10	41845764	(intergenic)
		513	3081	P, S	Chr1	143265918	(intergenic)			(Figure 2C)
Hep3B cells	RAISING-sequencing	6248	2061	C	Chr13	91169307	(intergenic)	0.74	†	→ #1
		1431	1796	X	Chr3	75917792	ROBO2	0.17	††	→ #2
		344	1796	X	Chr4	72824684	(intergenic)	0.04	†††
		58	1796	X	Chr11	37205841	(intergenic)	0.01	††††
		42	2061	C	Chr13	91169319	(intergenic)
		8	1796	X	Chr4	124126569	(intergenic)
		8	861	P	Chr9	26959590	IFT74
		7	1796	X	Chr5	145896541	GRXCR2
		5	1796	X	Chr10	129015069	(intergenic)
		5	1796	X	Chr14	42166780	(intergenic)
		5	1796	X	Chr3	98232650	(intergenic)
	Hybrid capture-sequencing	72,240	1796	X	Chr3	75917792	ROBO2	0.09	††
		64,122	2061	C	Chr13	91169307	(intergenic)	0.08	†
		40,875	1796	X	Chr4	72824684	(intergenic)	0.05	†††
		16,083	1796	X	Chr11	37205841	(intergenic)	0.02	††††
		13496	1829	C, X	Chr21	7261295	(intergenic)
		4200	1829	C, X	Chr4	49118564	(intergenic)
		2641	1829	C, X	ChrY	56696309	(intergenic)

Note: The junctions commonly detected in RAISING-sequencing and hybrid capture-sequencing are indicated by asterisks (*) or daggers (†), and their detection rates of integration events are calculated.

Abbreviation: RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

RAISING-sequencing has the capability for detection of HBV DNA integration comparable to hybrid capture-sequencing

We further validated the HBV-human junction sequences in PLC/PRF/5 or Hep3B cells, as determined by RAISING and NGS analysis (RAISING-sequencing), using junction-specific nested PCR (Figure 2A). Focusing on the 4 most frequently found junctions in PLC/PRF/5 cells [junctions #1–4, Table 1 (upper)], inner and outer primer pairs were designed so that the amplicons spanned the junctions (Supplemental Figure S2, http://links.lww.com/HC9/A682). Indeed, the expected products for junctions #1–4 were detected in the PLC/PRF/5 genomic DNA (Figure 2B-a), and subsequent Sanger sequencing confirmed that their sequences were identical to those determined by RAISING-sequencing (Figure 2B-b). Similarly, the top 2 junctions from the Hep3B cells [junctions #1–2, Table 1 (lower)] were validated by nested PCR and sequencing (Figure 2C and Supplemental Figure S3, http://links.lww.com/HC9/A682). These results confirmed the presence of chimeric sequences identified by RAISING-sequencing in these cell lines.

FIGURE 2.

Validation of HBV-human junction sequences in PLC/PRF/5 and Hep3B cells by junction-specific nested PCR. (A) Primers for junction-specific nested PCR were designed so that they span the junction sequences determined by the RAISING-sequencing in Table 1. (B, C) The PCR products by the primers targeting the junctions identified from PLC/PRF/5 [B, Table 1 (upper)] and Hep3B cells [C, Table 1 (lower)]. GAPDH gene was used as an internal control; DW and HepG2 cells as negative controls. HBV-human junction sequences are indicated using orange and blue boxes, respectively. (D) The sensitivity of RAISING in Supplemental Figure S1B-(b), http://links.lww.com/HC9/A682 was assessed by serially diluting PLC/PRF/5 cells genomic DNA with HepG2 cells genomic DNA. PLC/PRF/5 cells genomic DNA input percentage is indicated at the top. Abbreviations: DW, distilled water; GAPDH, Glyceraldehyde-3-phosphate dehydrogenase; RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

Our previous study demonstrated that RAISING could detect HTLV-1-human junction sequences at a frequency as low as 0.032%.¹⁷ In this study, we assessed the detection sensitivity of the HBV-human junction sequence using RAISING. PLC/PRF/5 genomic DNA was serially diluted with that of HepG2 cells without HBV DNA integration into their genomes. The results revealed that RAISING could amplify HBV-integrated fragments at a frequency of 0.1% or more (Figure 2D), with a similar sensitivity like that of HTLV-1-human junctions.

Furthermore, we compared the performance of RAISING-sequencing with that of hybrid capture-sequencing, a method frequently used for profiling HBV-human junction sequences (Table 1).(12–14) HBV-human junction sequences with a relatively high frequency in each method shared between both methods were completely identical, demonstrating very low levels of Taq-derived mutations during DNA-amplification steps in both RAISING-sequencing and hybrid capture-sequencing. For PLC/PRF/5 cells, the reads mapped to HBV were 99.5% (409,424/411,353) and 45.2% (31,761,938/70,275,382), of total reads generated by RAISING-sequencing and hybrid capture-combined NGS (NovaSeq platform), respectively [Table 2 (left)]. In addition, the RAISING-sequencing generated 11,990 chimeric reads (2.91% of total reads), 99.2% (11,889 reads) of which covered the junctions supported by both the RAISING-sequencing and hybrid capture-sequencing. Meanwhile, 478,291 chimeric reads were obtained by hybrid capture-sequencing, 198,188 reads of which were junctions located 3′ end of the viral reference sequences. Among the chimeric reads, 180,342 reads covered the consensus junctions supported by both methods, equivalent to 37.7% of the total chimeric reads and 91.0% of the 5′-HBV-human-3′ chimeric reads.

TABLE 2.

Comparison between RAISING-sequencing and hybrid capture-sequencing in tumor-derived cell lines (PLC/PRF/5 and Hep3B cells)

Note: Numbers in brackets indicate reads or rate focusing one-sided junction to compare with RAISING-sequencing.

Abbreviation: RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

Similarly, analyzing total DNA from Hep3B cells revealed that 820,426 reads out of 840,127 (97.7%) obtained by RAISING-sequencing, and 18,088,353 out of 83,945,238 (21.5%) by hybrid capture-sequencing, were mapped to HBV [Table 2 (right)]. The RAISING-sequencing generated 8161 chimeric reads (0.97% of the total reads), and 8143 (99.8% of the total chimeric reads) supported the consensus junctions. The hybrid capture-sequencing generated 272,585 chimeric reads (0.32% of the total reads), among which 216,572 reads adopted 5′-HBV-human-3′ structure, and 193,360 (70.9% of the total chimeric reads and 89.3% of the 5′-HBV-human-3′ type) supported the consensus junctions. These results indicate that RAISING-sequencing outperforms hybrid capture-sequencing in concentrating junction sequences and enables the detection of HBV DNA integration with smaller sequencing data, maintaining consistency with hybrid capture-sequencing.

RAISING-sequencing detected HBV DNA integration induced by de novo HBV infection

Recent studies have demonstrated that HBV DNA integration occurs in the early stages of infection and is not affected by the inhibition of reverse transcription by nucleos(t)ide analogs treatment.^32–34 Therefore, we examined whether the RAISING-sequencing could detect HBV DNA integration caused by de novo infection [Figure 1C (center)]. First, we used HepG2-hNTCP-C4 cells, an HBV-susceptible HepG2-derived cell line overexpressing human sodium taurocholate cotransporting polypeptide (NTCP).²⁰ The cells were infected with HBV for 16 hours and cultivated for 12 days with entecavir, a nucleos(t)ide analog, which was added to reduce the HBV rcDNA burden in the samples. Indeed, we observed an increase in the detection of junction sequences with entecavir treatment (Supplemental Table S6, http://links.lww.com/HC9/A682). Total genomic DNA was extracted from cells, amplified by RAISING, and sequenced using NGS (MiSeq platform). The assay identified HBV-human junction sequences from diverse human chromosomes [Table 3 (upper)], which is consistent with previous reports.^32–34 In addition, HBV-human junction sequences were obtained from more physiologically relevant samples, both HBV-infected primary human hepatocytes [Figure 1C (center) and Table 3 (lower)] and HBV-infected liver humanized uPA/SCID mice [Figure 1C (center) and Supplemental Table S7, http://links.lww.com/HC9/A682].

We further performed RAISING-sequencing of the adjacent nontumor liver biopsy specimens obtained from patients with HBV-related HCC [Figure 1C (bottom) and Supplemental Table S8, http://links.lww.com/HC9/A682]. The HBV-human junction profiles obtained are listed in Table 4. The number of chimeric reads substantially differed among the 3 samples, likely reflecting the amount of intrahepatic viral DNA in each sample. Most of the chimeric reads were concentrated on the C terminus half of the hepatitis B x protein gene (Supplemental Figure S4, http://links.lww.com/HC9/A682), which is consistent with previous reports.^3,4,35–37 The identified junctions were further validated by junction-specific nested PCR (Figure 2A and Supplemental Figure S5, http://links.lww.com/HC9/A682), and the same sequences were retrieved as those obtained by RAISING-sequencing (Figure 3A–C). These data indicate that RAISING-sequencing can robustly detect de novo HBV DNA integration occurring randomly with low frequency and reproduce previous reports of integration profiling of HBV-HCC samples.

TABLE 4.

HBV-human junctions in HCC clinical samples determined by RAISING-sequencing

Patient ID	No. total reads	HBV junction position	HBV genome	Human chromosome	Human junction position	Human gene	(Figure 3)
HCC-6	13	1800	X	Chr15	20503423	HERC2P3	→ #1
(nontumor tissue)	5	1804	X	Chr7	1081173	C7orf50	→ #2
HCC-33	1009	1753	X	Chr18	67700005	AC114689.3	→ #1
(nontumor tissue)	77	1713	X	Chr2	200147042	(intergenic)
	40	1779	X	Chr3	139268965	MRPS22
	38	1634	X	Chr4	185836717	SORBS2
	28	1733	X	Chr2	226775817	IRS1
	22	1819	C, X	Chr14	71388644	SIPA1L1
	20	1654	X	Chr4	103537128	(intergenic)
	16	1753	X	Chr4	73897495	(intergenic)
	16	1828	C, X	Chr9	95148096	FANCC
	10	1820	C, X	Chr1	173944363	RC3H1
HCC-46	5965	1676	X	ChrX	27176279	AC107419.1	→ #1
(nontumor tissue)	1051	1646	X	Chr12	77598319	NAV3	→ #2
	790	1785	X	Chr4	102824385	UBE2D3	→ #3
	587	1751	X	Chr1	198862191	MIR181A1HG
	218	1818	C, X	Chr10	124502003	LHPP
	181	1829	C, X	Chr2	215420541	FN1
	84	1821	C, X	Chr1	235499700	B3GALNT2
	78	1928	C	Chr2	177509515	AGPS
	46	1829	C, X	Chr11	11206373	(intergenic)
	22	809	P, S	Chr8	88734382	AC090578.1
	22	1827	C, X	Chr2	195217190	(intergenic)
	18	1680	X	Chr2	235683893	AGAP1
	10	1775	X	Chr18	64561691	(intergenic)
	9	1826	C, X	Chr8	135688215	(intergenic)
	8	1687	X	Chr12	34822229	(intergenic)
	8	1824	C, X	Chr11	46364499	DGKZ

Abbreviation: RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

FIGURE 3.

Validation of HBV-human junction sequences in clinical specimens by junction-specific nested PCR. HBV-human junction sequences in the genomic DNA of clinical specimens were verified by junction-specific nested PCR using primer sets designed to span the junction. (A–C) The PCR products were generated using primers designed to span the junctions found in HCC-6 (A), HCC-33 (B), and HCC-46 (C) by RAISING-sequencing (Table 4). GAPDH gene was used as an internal control; DW and HepG2 cells as negative controls. HBV-human junction sequences are indicated using orange and blue boxes, respectively. Abbreviations: DW, distilled water; GAPDH, Glyceraldehyde-3-phosphate dehydrogenase; RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

HBV-human junctions, identified by RAISING-sequencing, were detected in plasma cfDNA using ddPCR

Recently, HBV-human junction DNA, identified by hybrid capture-sequencing, was reportedly detected as cfDNA in the peripheral blood in patients with HBV-HCC, implying its usefulness as a tumor marker, especially for recurrence.¹⁵ Therefore, we examined whether HBV-human junction DNA identified by RAISING-sequencing could be also detected in plasma cfDNA. Xenograft mouse models were subcutaneously engrafted with PLC/PRF/5 or Hep3B cells, and plasma cfDNA was harvested when the tumor volume reached approximately 1000 mm³ (6–11 weeks after injection) (Figure 1C). As shown in Table 5, junctions #1 and #3 of PLC/PRF/5 cells, identified by RAISING-sequencing [Table 1 (upper)], were detected in the plasma cfDNA of the engrafted mice using ddPCR. Similarly, junctions #1–2 of Hep3B cells [Table 1 (lower)] were detected in the plasma cfDNA of Hep3B xenografts. The results demonstrate that the chimeric DNA determined by RAISING-sequencing was detectable in the plasma cfDNA of the xenograft mouse model.

TABLE 5.

HBV-human chimeric genomes in plasma cfDNA of tumor xenograft mouse models determined by ddPCR

		HBV-human chimeric genomes (copies/mL plasma)
PLC/PRF/5 xenograft mouse model	Tumor volume (mm3)	Junction #1	Junction #3
Pool (control)	–	0.0	0.0
Pool 1	698±53	18.6	19.2
Pool 2	1694±851	9.4	27.4
Pool 3	1604±605	27.7	27.7
		HBV-human chimeric genomes (copies/mL plasma)
Hep3B xenograft mouse model	Tumor volume (mm3)	Junction #1	Junction #2
Pool (control)	–	0.0	0.0
Pool 1	2479±1385	884.9	445.3
Pool 2	2462±1270	36.4	39.0
Pool 3	883±621	65.9	0.0

Abbreviations: cfDNA, cell-free DNA; ddPCR, droplet digital polymerase chain reaction; RAISING, Rapid Amplification of Integration Site without Interference by Genomic DNA contamination.

DISCUSSION

The integration of the HBV genome is frequently detected in hepatic tumor tissues. In this study, we demonstrated that HBV genome integration could be detected by RAISING, a novel method for detecting viral integration. RAISING-sequencing of the HBV-HCC-derived cell lines, PLC/PRF/5 and Hep3B cells, revealed viral integration profiles consistent with previous reports.^28–30 In addition, the assay could determine de novo HBV DNA integration in the early stages of infection in HepG2-hNTCP-C4 cells and primary human hepatocytes, in line with recent studies.^32–34 Furthermore, RAISING-sequencing successfully identified the chimeric sequences in HBV-infected tissues, including humanized livers and clinical specimens. We identified a junction in the fibronectin 1 (FN1) gene (Table 4), which is reportedly a frequent target in adjacent nontumor liver tissues.^3,6,38–41 Consistent with previous reports,^32,42,43 the HBV-human junctions identified in this study included microhomologous sequences between HBV and human genomes (Supplemental Figure S6, http://links.lww.com/HC9/A682). These results suggest that RAISING-sequencing can be used to determine HBV DNA integration sites, similar to conventional methods.

Many studies elucidated the close association between HCC development and HBV DNA integration, which occurs in more than 75%–90% of HBV-related HCC.^2–4 However, the lack of methods for assessing this integration hinders in-depth investigation of the molecular mechanisms and pathophysiological significance of HBV DNA integration. The simple and inexpensive PCR-based method is comparable to hybrid capture-sequencing with smaller sequencing data (Table 2 and Supplemental Table S9, http://links.lww.com/HC9/A682) and is also applicable to analysis using cell-based assays. Compared to inverse nested PCR, RAISING-sequencing is more expensive and more complicated to operate because it requires NGS analysis, but it has a notable advantage in the comprehensive detection of integration sites. Therefore, we propose that RAISING-sequencing is expected to facilitate the assessment of HBV DNA integration and contribute to elucidating the molecular mechanism of integration and its significance in HBV carcinogenesis.

Recent studies determining the HBV-human junction sequences in patients with HBV-HCC by hybrid capture-sequencing reported that the junctions are detectable in plasma or urine cfDNA.^15,44 Especially, Li et al(15) reported that the remnant chimeric DNA in the plasma after surgery is correlated with early relapse. These studies raise the possibility that chimeric DNA is a promising tumor marker for predicting recurrence. Similarly, we showed that the junction sequences determined by RAISING-sequencing were detected in the blood samples of the xenograft mouse models (Table 5). Thus, it might be intriguing to use blood chimeric DNA as a relapse marker in HBV-HCC cases where liver specimens are available and junction sequences can be determined by RAISING-sequencing (eg, after surgery). Nevertheless, further studies are necessary to apply this strategy to the clinical diagnosis, especially to determine the criteria (eg, frequency and sequence in HBV-human junctions) for choosing integration sites where ddPCR primers should be designed.

The reported RAISING-sequencing method for HBV has a limitation in that it depends on short-read sequencing (Illumina, 300 bp) and cannot determine the genomic structure of HBV-human chimeric reads. Adopting third-generation sequencing platforms, such as PacBio and Nanopore, may make it possible as other studies did.^{30,31,42,45–49} Additionally, these techniques may increase the applicability of RAISING-sequencing by covering the entire HBV genome with fewer primers. Moreover, the current strategy is not fit to detect HBV-human junctions from cfDNA directly, due to its trace amounts in blood samples and fragmentation that hampers triple primers annealing. Applying whole genomic amplification might allow the detection of junctions from cfDNA.⁵⁰

In conclusion, we established RAISING-sequencing, a simple and inexpensive method, to determine HBV-human junction sequences using smaller sequencing data. This method provides a robust tool for clarifying the molecular mechanisms of HBV DNA integration and can serve as a potential diagnostic tool for HBV-HCC.