A single-cell digital PCR method tailored for quantification of HBV DNA positive cells in liver biopsy tissues.

Introduction
Hepatitis B virus (HBV) infection remains a global public health challenge. According to the WHO, 254 million people were living with chronic hepatitis B (CHB) in 2022, with an estimated 1.2 million new infections annually and approximately 1.1 million related deaths [1]. China accounts for a significant portion of this burden, with a Hepatitis B surface antigen (HBsAg) prevalence of 5–6% [2, 3]. HBV persists in infected hepatocytes as covalently closed circular DNA (cccDNA) and integrated HBV [4]. This persistent HBV DNA poses the greatest challenge for achieving a complete cure [5]. Thus, accurate quantification of intrahepatic HBV DNA is crucial for assessing treatment efficacy and predicting prognosis [6].
In accordance with conventional practice, intrahepatic HBV DNA levels (mainly cccDNA) are typically evaluated indirectly through serological assays [7, 8]. Serum HBV DNA is commonly used to measure viral replication and guide antiviral treatment. However, HBV cccDNA can still be detected within the liver even when serum HBV DNA levels are below the detection limit [9]. While serum HBsAg reflects HBV cccDNA activity and its clearance is a key clinical endpoint, the correlation between serum HBsAg and intrahepatic cccDNA diminishes with long-term antiviral therapy [10, 11]. Recent studies suggest that serum HBV RNA and hepatitis B core-related antigen (HBcrAg) levels could reflect intrahepatic HBV cccDNA replication and transcription status when HBV DNA levels are undetectable [12]. These markers may serve as indicators of cccDNA activity following HBsAg clearance; however, their clinical applicability remains to be validated [13, 14].
Current methods for quantifying HBV cccDNA in liver biopsy tissues involve extracting bulk DNA and using detection techniques such as Southern Blot [15], quantitative PCR (qPCR), or digital PCR (dPCR) [16]. Although these techniques provide averaged signals from cell populations, they fail to measure cellular heterogeneity in infection status within the tissue. Additionally, variations in tissue quality and experimental procedures may compromise the sensitivity and accuracy of current bulk tissue-based methods, particularly for CHB patients with low intrahepatic HBV cccDNA levels (i.e., < 1 copy/cell) [17]. Therefore, quantification of intrahepatic HBV cccDNA at the single-cell level is needed to gain insight into the heterogeneity of infected cells.
Digital PCR (dPCR) employs the compartmentalization of samples to achieve absolute nucleic acid quantification, enabling accurate detection of low or even single-copy targets [18]. Widely used platforms include droplet-based (ddPCR) and chamber-based (cdPCR) systems. ddPCR involves partitioning samples into nanoliter-sized droplets through a two-phase water–oil split reaction system. In comparison to cdPCR, ddPCR has a larger channel size and reaction volume, rendering it particularly well-suited for the processing of single-cell suspensions and the detection of single-cell-specific genes or proteins [19–21]. The potential for this technology to be applied in the investigation of pathogen genes at the single-cell level remains to be seen.
In this study, we developed a single-cell infection detection PCR (scID-PCR) method specifically designed for liver biopsy samples to analyze the heterogeneity of intrahepatic HBV DNA, including both cccDNA and integrated forms. We validated the method’s stability and accuracy using HepG2.2.15 and Hep3B2.1–7 cell models. To demonstrate its clinical relevance, scID-PCR was applied to liver biopsy samples from patients with CHB and hepatocellular carcinoma (HCC). The scID-PCR method enables single-cell resolution evaluation of infection status and provides valuable insights into the prognosis of CHB patients.

Results

Design of scID-PCR for quantifying intrahepatic HBV-positive cells
To quantify HBV DNA positive (HBV-positive) cells in human liver biopsy tissues, we implemented a single-cell droplet digital PCR (sc-ddPCR) workflow [20], adapting it for pathogen detection (Fig. 1A). This process involves encapsulating single cells into droplets, followed by in-droplet TaqMan PCR targeting β-actin and cccDNA. β-actin positivity (β-actin +) confirms successful single-cell encapsulation, while dual positivity for β-actin and cccDNA (β-actin + & cccDNA +) indicates an HBV-infected cell (Materials and Methods). When the full-length HBV genome (3.2 kb) integrates into the host genome, it retains the same sequence as cccDNA. Therefore, our method cannot differentiate between integrated HBV DNA and cccDNA, as both can result in cccDNA-positive droplets that represent HBV-positive cells [22].
A critical aspect of this workflow is maintaining the appropriate cell-to-droplet ratio to ensure single-cell encapsulation, which is essential for reliable results. Since encapsulation occurs randomly, it follows a Poisson distribution [23, 24]. Commercial ddPCR devices typically produce between 20,000 and 100,000 droplets per standard 20 μL ddPCR reaction mix [25]. To optimize single-cell encapsulation, we modeled the probability distributions for encapsulating 2,000, 4,000, and 8,000 cells into 20,000, 50,000, and 100,000 droplets, respectively (Fig. 1B, Table S1, Materials and Methods). Based on these analyses, an input of 2,000–4,000 cells per standard 20 μL ddPCR reaction mix was found to best balance between maximizing detection signals and minimizing multi-cell droplets (Fig. 1C). This optimized input cell number is compatible with widely used ddPCR systems, such as the QX2000 system (Bio-Rad) and the MicroDrop-100 digital PCR system (Guangdong Forevergen Medical Technology Co., Ltd).

Optimization of scID-PCR for the quantification of intrahepatic HBV-positive cells
To ensure the specificity and sensitivity of intrahepatic HBV DNA detection, we carefully optimized primers and probes. Primer–probe pairs for HBV DNA were specifically designed to target the gap coding region of rcDNA, enabling accurate amplification of both cccDNA and integrated HBV DNA [26] (Table S2, Materials and Methods). Amplification efficiency and specificity were evaluated using qPCR with HBV DNA as the template. The selected primers, β-actin-2 and cccDNA-3, demonstrated high specificity, as indicated by single, sharp peaks in melting curves (Figure S1), and high sensitivity, reflected by lower Ct values (Table S3). Further optimization of the annealing temperature was performed using scID-PCR with HepG2.2.15 cells, yielding optimal results (Figure S2). To assess the pan-genotypic capacity of the selected primers-probe pairs, we performed multiple sequence alignments of the scID-PCR primer–probe target regions across HBV genotypes A–H. The results showed high conservation for genotypes A–E, ensuring reliable detection for these subtypes. However, subtypes F, G, and H exhibited lower compatibility (Figure S3). Therefore, our scID-PCR is well-suited for detecting HBV genotypes A-E.
Under identical sc-ddPCR conditions, the signal-to-noise ratio for DNA input was significantly higher than that for cell input (Figure S4). Two primary factors likely contributed to this difference. First, components within the cell lysate may inhibit PCR reactions. To mitigate this, we introduced a PCR enhancer (KAPA Enhancer 1). Second, differences in template copy numbers between DNA input and cell input ddPCR could affect amplification. In DNA input ddPCR, each droplet contains a single-copy template, whereas in cell input ddPCR, β-actin is present in two copies, and cccDNA may exist in multiple copies. This could lead to limited PCR substrate availability in droplets, reducing amplification efficiency. To address this, we increased the quantities of dNTPs and primers, and also increased the number of PCR cycles. These combined strategies successfully enhanced the signal-to-noise ratio, ensuring reliable detection and quantification (Fig. 2A).
In our initial experiments, we observed a substantial number of β-actin- & cccDNA + droplets (Fig. 2B), indicating the presence of cell-free nucleic acids in the single-cell suspension supernatant. This issue could lead to false positives in β-actin + & cccDNA + droplet counts. Specifically, cell-free HBV DNA may be encapsulated into droplets containing cccDNA-negative cells, or both cell-free HBV DNA and β-actin may co-encapsulate into the same droplet. Similarly, the presence of cell-free β-actin could falsely indicate successful cell encapsulation in β-actin + droplets. The cell-free HBV DNA and β-actin DNA may have originated from cell rupture during handling. To resolve this, we treated the suspension with DNase I to digest these nucleic acids and removed the enzyme using a laminar flow washing method to minimize cell damage (Materials and Methods). After washing, cells were immediately encapsulated. The DNase I treatment effectively reduced the proportion of β-actin- & cccDNA + droplets (Fig. 2C).

Verification of scID-PCR using HBV cell models
To evaluate the stability and accuracy of scID-PCR in quantifying HBV-infected cells, we tested its performance using HepG2.2.15 and Hep3B2.1–7 cell lines. HepG2.2.15 cells, which contain integrated HBV sequences, served as a cccDNA surrogate, while Hep3B2.1–7 cells, harboring intact HBV, were used as a widely recognized infection model [27]. HBV-positive cells were first identified at the RNA and protein levels using RNAscope (catalog number 560071, ACD) and immunofluorescence (IF, HbcAg) (Materials and Methods). The RNAscope probe targets the 1244–1916 region of the HBV genome, which overlaps with the region amplified by our selected primer–probe pairs. RNAscope analysis revealed consistent proportions of HBV-positive cells in both cell lines (HepG2.2.15: 51.4 ± 2.7%; Hep3B2.1–7: 43.2 ± 2.5%) when compared to scID-PCR measurements (HepG2.2.15: 51.2 ± 0.9%; Hep3B2.1–7: 38.5 ± 1.3%) (Fig. 3, Materials and Methods). Similarly, IF results (HepG2.2.15: 56.0 ± 4.4%; Hep3B2.1–7: 42.3 ± 1.5%) closely aligned with scID-PCR findings (Fig. 3, Materials and Methods).

To further validate the accuracy of scID-PCR in detecting HBV cccDNA, we performed serial dilutions of HBV-positive cells mixed with HBV-negative peripheral blood mononuclear cells (PBMCs) to evaluate detection accuracy. The scID-PCR results showed a clear, proportional decline in HBV-positive cells percentage, consistent with the series dilution of both HepG2.2.15 and Hep3B2.1-7 cells (Fig. 4A). Minor discrepancies between observed proportions and expected dilution factors were likely due to variations in cell counting and pipetting accuracy. Nonetheless, scID-PCR consistently demonstrated stable and precise detection of cccDNA at the single-cell level.
To assess the impact of replication intermediates (RIs) on scID-PCR reliability, HepG2.2.15 and Hep3B2.1-7 cells were treated with the nucleoside analog entecavir (ETV), which inhibits viral polymerase activity and reduces the production of virions and replication intermediates [28]. Following ETV treatment, the proportion of HBV-positive cells detected by scID-PCR remained largely unchanged (Fig. 4B). This result indicates that the presence of RIs does not affect the reliability of scID-PCR in detecting infected cells.

Detection of HBV-positive cells in liver biopsy tissues
As a proof of concept, we evaluated HBV-positive cells in liver biopsy tissues from five CHB patients and one patient with non-HBV-related HCC (Table S4, Materials and Methods). The CHB patients were treated with a combination of nucleos(t)ide analogs (NAs) and pegylated interferon-α (PEG-IFN-α). Before treatment (0 weeks, 0w), all CHB patients tested serologically positive for hepatitis B surface antigen (HBsAg > 0.05 IU/mL), and three were positive for hepatitis B e antigen (HBeAg > 1.0 S/CO). After 48 weeks of therapy (48 weeks, 48w), all five patients were HBeAg-negative, four were HBsAg-negative, three were HBV DNA-negative, and two were negative for both HBV DNA and HBsAg (Table S5, Materials and Methods).
To investigate intrahepatic heterogeneity of HBV infection, liver biopsy samples were collected at 48 weeks post-treatment. Tumor and adjacent non-tumor tissues were also obtained from the HCC patient. Fresh liver tissues were enzymatically dissociated into single-cell suspensions, with all samples showing excellent viability and cell counts (Fig. 5A, Table S6, Materials and Methods).
Using the scID-PCR, we detected HBV-positive cells in liver tissues from all CHB patients but not in the non-HBV-related HCC patient (Table S5). Among the CHB patients, those who remained serologically HBV DNA-positive at 48 weeks exhibited higher proportions of HBV-positive cells in liver tissues (CHB-1: 35.81%; CHB-2: 38.83%) compared to those who were HBV DNA-negative (CHB-3: 19.29%; CHB-4: 22.64%; CHB-5: 12.88%). We observed a correlation tends between the proportion of HBV-positive cells in the liver and serum HBV DNA (R2 = 0.899) and HBsAg (R2 = 0.372) levels, which became less apparent when serum HBV DNA levels fell below the detection limit (Fig. 5B/C). These findings align with previous studies but uniquely provide intrahepatic cccDNA levels after serum HBV DNA clearance. Notably, no correlation was found between serum HBeAg levels (Figure S5) and the proportion of HBV-positive cells in liver tissues.
Based on changes in HBsAg, HBeAg, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and HBV-positive cell proportions (R), patients were classified into three groups (Fig. 5D). Group 1 (CHB-1 and CHB-2) exhibited high initial HBsAg levels (CHB-1: 791.12 IU/mL; CHB-2: 1184.78 IU/mL, 0w) and responded well to therapy, with HBsAg levels dropping below 0.05 IU/mL at 48 weeks. However, they retained high liver HBV-positive cell ratios (CHB-1: 35.81%; CHB-2: 38.83% at 48w). Group 2 (CHB-3 and CHB-4) exhibited low initial HBsAg levels (CHB-3: 0.40 IU/mL; CHB-4: 0.23 IU/mL, 0w) and demonstrated varied therapeutic responses; CHB-3 showed no significant response, with HBsAg levels increasing to 0.96 IU/mL at 48 weeks and a HBV-positive cell proportion of 19.29%. In contrast, CHB-4 responded well to therapy (HBsAg < 0.05 IU/mL at 48w), with a HBV-positive cell proportion of 22.64%. Group 3 (CHB-5) exhibited moderate initial HBsAg levels (700.75 IU/mL, 0w) and responded well to therapy, with HBsAg levels dropping below 0.05 IU/mL at 48 weeks. CHB-5 had the lowest proportion of HBV-positive cells (12.88% at 48w) but exhibited significant liver damage, as indicated by elevated ALT and AST levels (> 40 U/L). This suggests that clearance of infected cells may have contributed to the observed liver injury.

Discussion
The persistence of HBV DNA within infected hepatocytes remains a major challenge to achieving a complete cure for CHB. Current methods for measuring cccDNA levels lack the resolution needed to capture the heterogeneity of infected cells, limiting their ability to provide detailed insights into the dynamics of intrahepatic infection [29].
Several approaches have been developed to address this challenge. Zhang et al. introduced a technique using split amplification and nested PCR, in which single nuclei are isolated and lysed, and the lysate is divided into multiple wells for PCR targeting both cccDNA and IFN-α as an internal control. This method requires over 24 individual PCR reactions per nucleus, with results analyzed by agarose gel electrophoresis [30]. Huang et al. developed a droplet digital PCR (ddPCR) approach, where single cells are manually isolated, lysed, and digested for the non-cccDNA, then directly used for cccDNA quantification [31]. Tu et al. proposed cinqPCR, which uses restriction enzymes to invert specific HBV sequences, allowing cccDNA measurement normalized to cellular DNA. However, this technique is applied to bulk DNA extracts from many cells [32]. While these methods effectively detect cccDNA, they suffer from limitations such as lack of single-cell resolution, low throughput, time-intensive protocols, and high labor demands. These constraints can lead to biased results, particularly when analyzing small sample sizes.
To address these challenges, we developed the scID-PCR method, a droplet-based method that encapsulates individual cells and performs in-droplet TaqMan PCR targeting both β-actin and HBV cccDNA. This approach enables high-throughput, rapid, and efficient detection of HBV-infected cells at single-cell resolution. We validated the robustness of scID-PCR in both cell models (HepG2.2.15 and Hep3B2.1–7) and liver biopsy samples from CHB patients. After antiviral therapy, the method detected HBV-positive cells in all CHB patients, including those with undetectable serum HBV DNA. This underscores its utility in identifying residual infection, which is often missed by conventional methods. Moreover, the proportion of HBV-positive cells in liver tissue provided a more nuanced picture of infection status, allowing us to distinguish patients with better prognoses and more accurately evaluate treatment outcomes. These results highlight the limitations of relying solely on serological markers and emphasize the importance of intrahepatic assessment in guiding CHB management.
For proof of concept, we integrated clinical serum markers, such as HBsAg, HBeAg, ALT, and AST, with the proportion of HBV-positive cells in liver tissue. This enabled a more nuanced understanding of treatment response. Patients with reduced serum HBsAg and HBeAg levels during therapy, coupled with a low proportion of HBV-positive cells in the liver (< 30%), were more likely to achieve a functional cure (e.g., CHB-4). Conversely, patients who showed a serological response (e.g., CHB-1 and CHB-2) but retained a high proportion of HBV-positive cells (> 30%) may require extended treatment. In cases where patients exhibited signs of liver inflammation (e.g., elevated ALT and AST) alongside a low proportion of HBV-positive cells (e.g., CHB-5), a combination of antiviral and anti-inflammatory therapies could improve outcomes. For non-responders (e.g., CHB-3), alternative therapeutic strategies may need to be explored. These findings demonstrate the value of scID-PCR as a complementary tool to serological assays, offering a more comprehensive evaluation of infection status and aiding in personalized treatment decisions.
The proportion of HBV-positive cells in liver tissue was strongly correlated with serum HBV DNA levels, weakly with HBsAg, and showed no correlation with HBeAg. Serum HBV DNA reflects active replication from infected hepatocytes, explaining the strong correlation. HBsAg, which is derived from both cccDNA and integrated HBV DNA, may remain stable in long-term treated patients because of ongoing expression from integrated sequences, weakening its link to intrahepatic HBV-positive cells. HBeAg is a transcriptional product but is more affected by viral mutations (e.g., core promoter), which can suppress its expression even when the infection persists. Thus, the observed correlations are reasonable [9–11] and highlight the added value of tissue-based detection for understanding HBV infection dynamics.
Despite its strengths, scID-PCR has some limitations. It does not provide an exact count of HBV DNA copies within individual cells, as each droplet contains a whole cell that may carry varying—and often multiple—copies of HBV DNA. Therefore, precise quantification per cell is not possible. Conventional methods estimate copy numbers by comparing the target signal to an internal reference gene. Although scID-PCR includes a reference gene (β-actin), using differences in fluorescence intensity to infer relative copy number is not advised. Variations in amplification efficiency and differences in probe fluorescence between the target and reference can compromise accuracy.
One limitation of scID-PCR is its reliance on liver tissue samples. Although liver biopsy is minimally invasive, it is often not well accepted by patients. In addition, fresh liver tissue must be processed promptly for cell isolation and preservation, which presents logistical challenges. To support broader clinical applications, future research should focus on investigating correlations between blood-based biomarkers and intrahepatic residual HBV. Additionally, this study focused on validating the method and included a limited number of CHB samples, which restricts the generalizability of the findings. Future studies with larger patient cohorts are necessary to confirm these observations and refine the clinical implications of scID-PCR data.
By enabling single-cell resolution, scID-PCR represents a significant advancement in HBV DNA detection and improves our understanding of HBV biology, particularly in patients with low levels of viral replication, such as those who have seroconverted. Furthermore, this method has the potential for broader applications in detecting other latent or low-abundance infectious pathogens [33], such as HSV, VZV, EBV, CMV, or even latent SARS-CoV-2.

Materials and methods

Samples, participants, and ethics
The HepG2.2.15 cell line was obtained from the Central Lab and Liver Disease Research Center, the Affiliated Hospital of Yunnan University (Kunming, China). The Hep3B2.1–7 cell line was purchased from Pricella (CL-0102).
Five CHB patients were recruited from an anti-HBV pilot cohort. Patients with CHB, aged 18–65 years, who had received NAs treatment for 1–5 years and had serum HBV DNA ≤ 1,000 copies/ml and HBsAg ≤ 3,000 IU/ml were eligible for inclusion. Exclusion criteria included receiving IFN or systemic antiviral therapy within the previous 6 months; coinfection with HIV, HCV, or HDV; the presence of cirrhosis and decompensated liver disease; pregnancy or lactation; and any other contraindication to IFN therapy.
All enrolled patients received a 48-week combination treatment of NAs and PEG-IFN-α. Venous blood samples were collected at baseline (week 0) and at the end of treatment (week 48) for serological testing. Liver biopsy samples were obtained at week 48 via ultrasound-guided puncture.
The treatment response was defined as HBV-DNA loss. Patients who remained HBsAg- or HBV DNA-positive at the end of treatment were classified as non-responders. For this study, we included three responders and two non-responders. Detailed demographic and clinical data are provided in Table S4.
1 HCC patient without HBV infection was recruited in this study. Liver tumor tissues and the adjacent non-tumor tissues were collected from the tumor resection surgery.

Primers and probes
The HBV genome primarily exists as relaxed-circular DNA (rcDNA), a 3.2 kb molecule consisting of a complete minus strand and an incomplete plus strand. Within infected cells, rcDNA is converted into covalently closed circular DNA (cccDNA). To specifically distinguish cccDNA from rcDNA, primers were designed to target the gap region unique to rcDNA, enabling the precise amplification of the cccDNA sequence. Primers and probes were designed using Primer Premier 6.0 and the Real-time PCR (TaqMan) Primer Design tools from GenScript. The specificity of each primer pair was verified using NCBI Primer-BLAST against the “Human genomic + transcript” database, with “Homo sapiens” as the selected organism. Following this validation, two β-actin and three cccDNA primer pairs were chosen for further use (Table S2).

qPCR
The qPCR was used to assess the amplification efficiency and specificity of the primers. qPCR reactions were performed using LightCycler® 96 PCR Detection System (Roche Diagnostics, Mannheim, Germany). A 20 µL qPCR reaction mix comprised 10 µL of 2 × FastStart Essential DNA Green Master (Roche Diagnostics, Mannheim, Germany), 1 µM of forward primer, 1 µM of reverse primer, and 10 ng HepG2.2.15 DNA sample. The conditions of amplification were an initial pre-incubation cycle of 10 min at 95 °C, followed by 40 cycles of denaturation for 10 s at 95 °C, annealing for 15 s at 55 °C and extension for 20 s at 72 °C. Melt curve analysis from 65 °C to 95 °C with a ramp rate of 2.2 °C/second.

Serum HBsAg, HBeAg, AST, and ALT measurements
HBsAg and HBeAg levels were measured by chemiluminescent microparticle immunoassay (CMIA) using an Architect i2000SR analyzer (Abbott Diagnostics, North Chicago, IL, USA). Serum HBsAg was determined quantitatively, while serum HBeAg was determined qualitatively. The positive cut-off value for HBsAg is ≥ 0.05 IU/mL. HBeAg is interpreted using a ratio of the sample relative light unit (RLU) rate to the cut-off RLU (S/CO). AST and ALT levels were determined according to standard procedures in the clinical laboratories in the hospitals.

Serum HBV DNA quantification
HBV DNA was quantified by Roche Cobas TaqMan HBV test (Roche Diagnostics, Mannheim, Germany), according to the manufacturer’s instructions, with a linear range of 20 to 108 IU/ml. HBV DNA testing results below 20 IU/ml were considered negative.

Liver biopsy dissociation
Fresh liver biopsy tissues were immediately stored and transported at 4 °C in MACS® Tissue Storage Solution (Miltenyi Biotec). For enzymatic dissociation, tissues were washed twice with precooled dissociation medium (RPMI1640 + 10% FBS) and minced into 2–4 mm pieces using sterile scalpels. The tissue fragments were transferred to a T25 flask containing 4.5 mL of fresh dissociation medium, along with 250 μL Collagenase I (20 mg/mL, Sigma), 100 μL Dispase II (100 mg/mL, Sigma), and 50 μL DNase I (100 μg/mL, Roche). The mixture was incubated at 37 °C with 5% CO₂ and shaken at 160 RPM for 30 min to 2 h or until the tissue fragments were fully dissociated. The resulting cell suspension was passed through a 70 µm strainer to remove debris and centrifuged at 400 × g at 4 °C for 5 min. Red blood cells were lysed using eBioscience™ ACK Lysis Buffer (Invitrogen), followed by another centrifugation at 500 × g at 4 °C for 5 min. The cell pellet was resuspended in 1 mL PBS (Gibco), and the single-cell suspension was counted using Countstar with AO/PI staining. Cells could then be cryopreserved in CryoStor CS10 cell cryopreservation media (Sigma, USA) for future use.

Digest the cell-free DNA in single-cell supernatant
This protocol applies specifically to cryopreserved cells or cell lines and is not required for fresh liver biopsy dissociation cells. Cell-free DNA, including cccDNA and genomic DNA released from dead cells, can generate false-positive signals in ddPCR. To eliminate this, DNase I was used to digest cell-free DNA in the single-cell suspension. Specifically, 4 μl DNase I (Roche) was added to 50 μl of single-cell suspension (2 × 105 cells in PBS or dissociation medium) and incubated at 37 °C with 5% CO₂ for 15 min. Then, the cells were washed using the Laminar Wash System (Curiox Biosystems). After washing, cells were then gently collected, counted using Countstar with AO/PI staining, and immediately processed for sc-ddPCR to ensure accurate results.

sc-ddPCR
Single-cell suspension was used as template input for sc-ddPCR. The 20μL ddPCR reaction mix as: 10μL ddPCR supermix (Bio-Rad), 0.9μL cccDNA primer (F + R, 20 μM), 0.3μL cccDNA probe (20 μM), 0.9μL β-actin primer (F + R, 20 μM), 0.3μL β-actin probe (20 μM), 0.4μL dNTP (10 mM each, NEB), 2.6μL 5 × KAPA enhancer 1 (KAPA Biosystems), and 0.6μL membrane destructive agent (FOREVERGNE), and 4μL single-cell supernatant (2000–4000 cells). The PBS was used to replace the single-cell suspension for negative control (NTC). Gently pipetted mix the ddPCR reaction mix and immediately generated water-in-oil droplets using the ddPCR (FOREVERGNE or Bio-Rad) system following the manufacturer’s instructions. The droplets were transferred to a 96-well plate for the PCR protocol using a C1000 Thermal Cycler (Bio-Rad). The thermal cycling program included the cell lysis step at 85 °Cfor 60 min, pre-start DNA polymerase at 95 °C for 10 min, and 54 cycles of start DNA polymerase at 95 °C for 30 s, annealing and elongation at 55 °C for 60 s, stored at 16 °C. Then, the 96-well plate was transferred to the ddPCR system (FOREVERGNE or Bio-Rad) for FAM and VIC signal readouts.

Immunofluorescence assay
HBV core protein (HBcAg) was analyzed by immunofluorescence on HepG2.2.15 and Hep3B2.1–7 cells. Approximately 5 × 104 cells per well were seeded in 4-well Chambered Coverglass (Capitol Scientific, Austin, TX) overnight. Cells were fixed using 4% paraformaldehyde for 30 min, followed by incubation with 0.5% Triton-X 100 for 15 min at room temperature. After being blocked with 5% BSA at room temperature for 30 min, cells were incubated with anti-hepatitis B virus core antigen–antibody (1:200, Abcam, USA) at 4 °C overnight, incubated with an Alexa Fluor® 488 conjugated goat anti-mouse IgG (1: 200, Abcam, USA) secondary antibody for one hour at room temperature 5. Finally, the cells were stained with 4,6-diamidino-2-phenylindole (DAPI, Vector, CA) for nuclear indication. Images were captured using a confocal laser scanning microscope (TCS-NT, Leica Microsystems, Heidelberg, Germany). Fluorescent foci were quantified using ImageJ software (available from http://rsb.info.nih.gov/ij/). Threshold levels were set using a sample without primary antibodies in each staining group. The statistical results were presented as mean ± SME.

RNAscope assay
HepG2.2.15 and Hep3B2.1–7 cells were seeded at 5 × 104 cells/mL in Nunc Lab™-Tek™ II Chambered Coverglass and incubated overnight. The RNAscope assay was performed using the RNAscope™ Multiplex Fluorescent Reagent Kit v2 with the HybEZ™ II Hybridization System (Advanced Cell Diagnostics) following the manufacturer’s instructions. Slides were fixed with 4% paraformaldehyde (PFA) for 30 min at room temperature, dehydrated in a series of ethanol concentrations (50%, 75%, 100%), treated with hydrogen peroxide for 10 min, and incubated with Protease III solution for 10 min. Between steps, slides were rinsed with PBS.
HBV mRNA within the cells was hybridized using RNAscope Z probes targeting the 1244–1916 region of the HBV genome. Hybridization was performed for 2 h at 40 °C in a HybEZ oven. Negative control probes were used in companion cells to confirm signal specificity. Following hybridization, slides underwent sequential amplification with AMP1 (30 min at 40 °C), AMP2 (30 min at 40 °C), and AMP3 (15 min at 40 °C). Amplifiers were tagged with fluorescent dyes by applying channel-specific horseradish peroxidase (HRP) and the red dye TSA Vivid Fluorophore 570 for 30 min at 40 °C. HRP blockers were then applied to seal the amplified structures. Slides were counterstained with DAPI and imaged using a confocal microscope.
Images from multiple sections per group were analyzed using ImageJ software. Thresholds were set using negative control samples. Cells containing red signals (HBV-RNA) were categorized as HBV-positive, and those with both red (HBV-RNA) and blue (nucleus) signals were classified as HBV-positive cells. Morphologically intact cells were randomly selected, counted manually under 20 × magnification, and averaged per biological replicate. Results were presented as mean ± SEM.

Statistical analysis

Modeling the probability distribution for cell encapsulation in droplets
For each combination of cells and droplets, we calculated the mean number of cells per droplet (λ) as follows:
The probability of a droplet encapsulating exactly κ cells is given by the Poisson probability mass function:
Using the “dpois” function from the “stats” package, the probabilities were computed for κ = 0, 1, 2, … cells per droplet for each combination of cells and droplets.

Proportion of the cccDNA+ cells
The HBV-positive (cccDNA+) droplets are defined as the droplets that the FAM signals higher than NTC (negative control, the input is PBS); The β-actin-positive (β-actin+) droplets are defined as the droplets that the VIC signals higher than NTC. The β-actin+ indicated a cell was encapsulated in the droplet; The cccDNA and β-actin double positive (β-actin+ & cccDNA+) indicated an infected cell was encapsulated in the droplet.
The proportion of HBV-positive cells (R) was calculated as follows:
a: number of cccDNA+ & β-actin+ droplets.
b: number of cccDNA− and β-actin+ droplets.

Supplementary Information
Below is the link to the electronic supplementary material.