2.1 Serum samples, genotyping, and check for mutations

Serum samples were collected from 54 HBeAg⁺ and treatment-naïve Chinese patients infected with subgenotype B2 or C2. Serum HBV DNA, HBsAg, and HBeAg were quantified by ADICON Clinical Laboratories using COBAS@ TaqMan@ platform and Abbott ARCHITECT i2000 platform, respectively. Liver biopsy or dynamic serum alanine transaminase (ALT) and HBV DNA levels within 6 months of follow up were available for 24 patients to classify them into the IT or IC phase of chronic infection. The 3.2-kb HBV genome was amplified from serum samples by polymerase chain reaction (PCR) and sequenced directly. Viral genotype was determined by phylogenetic analysis. BCP and PC mutants were defined by presence of A1762T/G1764A or G1896A mutation.

2.2 In vitro HBV infection using viremic serum samples

The HepG2 cell line stably transfected with an NTCP expression construct (HepG2/NTCP) and HepaRG cell line have been described.^{29, 30} Cells seeded in 96-well plates were used for infection, with confluent HepaRG cells differentiated for >2 weeks before infection. To compare in vitro transmission efficiency, 2 μL of serum sample was incubated with cells for 16 h in medium not supplemented with fetal bovine serum (FBS), with 4% polyethylene glycol (PEG) for HepG2/NTCP cells but without PEG for HepaRG cells. Cells were washed and further cultured in FBS-containing medium. HBsAg and HBeAg were measured from culture supernatant through enzyme-linked immunosorbent assay (ELISA) using commercial kits (Kehua), using a volume not causing signal saturation. Cells were harvested at Day 20 postinfection for Southern blot using a mixed probe of B2 and C2 subgenotypes.³⁰ To compare HBV infectivity using same genome copy number, serum samples were diluted with normal human serum.

2.3 1.1mer construct and production of cHBV particles

Generation of 1.1mer construct from serum samples has been described,³⁰ with a 14-nt distance between the end of CMV promoter and predicted transcription initiation site for pregenomic RNA. The 3.2-kb HBV genome was amplified by PF1'—PR1 primer pair, and an overlapping 0.1-kb fragment using PF2—PR2 primer pair. The two DNA fragments were cloned to pcDNA3.1zeo(-) vector using a DNA assembly kit. HepG2 cells in six-well plates were transfected with 2 μg of 1.1mer construct using TransIT-LT1 reagent (Mirus). Cell culture-derived HBV (cHBV) particles from combined Days 3 and 5 culture supernatant were PEG precipitated.²⁹ They were resuspended in 1/30th−1/70th original volume, and inoculated overnight with 4% PEG but no FBS, into HepG2/NTCP or HepaRG cells seeded in 96-well plates. To inoculate with same genome copy number of viral particles, virion DNA was quantified by PCR from a fraction of culture supernatant following immunoprecipitation with anti-preS1 and anti-S antibodies.³⁰

2.4 Analysis of HBV markers from transfected cells

HBV RNAs and replicative DNA from HepG2 cells transfected with the 1.1mer construct were detected by Northern and Southern blot with ³²P-labled full-length probe, while envelope and core proteins were detected by Western blot.¹⁶ Virions were immunoprecipitated from culture supernatant followed by PCR quantification of HBV DNA. Alternatively, virions and core particles were separated through native agarose gel electrophoresis followed by transfer to nitrocellulose filter and hybridization with HBV DNA probe.

2.5 Statistical analysis

Differences between the groups were examined using a Student t test or Wilcoxon Ranksum test by GraphPad Prism 6 (GraphPad Software, Inc.). A p value of <0.05 is considered as statistically significant. Data are presented as means ± standard deviations (SD).

3.1 Serum samples for functional characterization

Serum samples were collected from 54 HBeAg⁺ and treatment-naïve Chinese patients. Most data analysis was based on 48 samples with information on viral genotype, BCP and PC sequences: 30 of C2 and 18 of B2 subgenotypes (Table 1). They had HBV DNA titer of 1.07 × 10⁶−1.44 × 10⁸ IU/mL (Figure 1A). The G1896A PC mutation and A1762T/G1764A BCP mutations were found in 11 and 16 samples, respectively (Figure 1A; sample #35 had both PC and BCP mutations), as a pure population or mixture with WT sequence (Table 2). Consistent with the higher prevalence of PC mutation in genotype B but BCP mutations in genotype C,^{19, 31-33} 8 of 10 samples with just PC mutation belonged to B2 subgenotype, whereas 13 of the 15 samples with BCP mutations only were of C2 subgenotype (Table 2).

3.2 sHBV infected HepaRG cells better supported replication of BCP mutant in contrast to WT virus by HepG2/NTCP cells

To identify highly infectious HBV isolates and to compare transmission efficiency we first inoculated 2 μL of serum samples into HepG2/NTCP cells or differentiated HepaRG cells seeded in 96-well plates. Considering PEG-independent infection of HepaRG cells by serum-derived HBV (sHBV) particles,²⁹ inoculation was performed without PEG for HepaRG cells but with 4% PEG for HepG2/NTCP cells. Forty-four and 41 of the 48 samples were respectively infectious in HepaRG cells and HepG2/NTCP cells (Table 1), as defined by positive HBeAg and/or HBsAg from culture supernatant at Day 15 postinoculation. Thirty-nine samples were infectious in both cell types, five in HepaRG cells only, and two marginally infectious in HepG2/NTCP cells only. Consistent with our previous report,²⁹ HepaRG cells produced much higher HBsAg titer than HepG2/NTCP cells (Table 1; Figure 1B–E) and higher HBsAg/HBeAg ratio for most samples (Figure 1F vs. 1G; Supporting Information: Table S1). That was also true when comparison was made within WT isolates or among BCP mutants (Supporting Information: Table S2). In HepaRG cells sample #42 and #31 produced highest level of replicative DNA, followed sequentially by #6, #7, #14, #56, #38, #52, #18, #24, and #11 (Figure 1H). Most such samples harbored BCP mutations, with only #11 and #18 having pure WT sequence (Samples #14 and #56 had no clear sequencing data). In HepG2/NTCP cells level of replicative DNA was #42 > #55 > #50, with weak signals from #16, #53, #43, #44, and #48 (Figure 1I). They had pure WT sequence except #42, which had BCP mutations (Sample #48 had no clear sequencing data).

3.3 C2 samples had higher HBV DNA titers than B2 samples and higher transmission efficiency

The 30 C2 samples had higher median and mean titers of HBV DNA and HBeAg than the 18 B2 samples (Table 1; Supporting Information: Figure S1A,C). Slightly higher percentage of C2 samples was infectious than B2 samples in both cell types (Table 1). The C2 samples also produced higher titers of HBeAg and HBsAg than B2 samples, although for HepaRG cells the difference in HBsAg did not reach statistical significance (Table 1; Supporting Information: Figure S1D–G). In HepaRG cells 9 of the 11 isolates with high replication were of C2 subgenotype. Only #11 was a B2 isolate while the genotype for #14 was undetermined. Of the eight isolates with high DNA replication in HepG2/NTCP cells, all but #43 and #48 were C2 isolates. When only samples with WT BCP and PC sequences were compared, the 14 C2 serum samples had higher mean HBeAg and HBV DNA titers than the eight WT B2 samples (Figure 2A,C). They produced higher mean HBeAg and HBsAg titers than the B2 samples in HepG2/NTCP cells, with the difference in HBeAg not statistically significant (Figure 2I,J). In this cell line five WT C2 samples (#16, #44, #50, #53, #55) but only one WT B2 sample (#43) showed high replication (Figure 1I).

3.4 Reduced transmission efficiency of PC mutant than WT HBV

Serum samples of the eight PC mutants of B2 subgenotype had lower mean HBeAg, HBsAg, and HBV DNA titers than those of the eight WT B2 samples, although the difference in HBV DNA titer did not reach statistical significance (Figure 2A–C, left). Since in six of the eight samples (#1, #26, #33, #37, #39, #51) the PC mutant co-existed with WT virus, their median HBsAg/HBeAg ratio and HBV DNA/HBeAg ratio were not much elevated (Figure 2D,E). They produced lower HBsAg titer in both HepaRG and HepG2/NTCP cells (Figure 2H,J) suggesting reduced transmission efficiency. Surprisingly the two C2 samples harboring just PC mutation (#34, #36) had the highest HBV DNA titers among the 48 samples (Figure 1A), but they produced low HBeAg and HBsAg titers in HepG2/NTCP and especially HepaRG cells (Figure 1B–E).

Compared with the 14 WT C2 samples, the 13 C2 samples with BCP mutations but no CP mutation had lower mean HBeAg, HBsAg, and HBV DNA titers, with the difference in HBsAg titer being statistically significant (Figure 2A–C). These samples produced less HBeAg and HBsAg than the WT C2 samples in HepG2/NTCP cells, but no statistically significant difference was observed in HepaRG cells (Figure 2G–J). While all but #42 of the six high replicating C2 isolates in HepG2/NTCP cells had WT sequence, seven of the eight C2 isolates with high replication in HepaRG cells were BCP mutant (only #18 had WT sequence).

One concern with the above analysis was the small sample size, making it difficult to reach statistically significant findings. We, therefore, compared WT virus, BCP mutant, and PC mutant irrespective of viral genotype. Serum HBV DNA, HBsAg, and HBeAg titers were highest for the 22 WT isolates, intermediate for the 15 BCP mutants, and lowest for the 10 PC mutants, although the difference in HBV DNA titer did not reach statistical significance (Table 2). The median ALT level was highest for PC mutants and lowest for WT isolates. Median HBsAg and HBeAg titers in infected HepaRG and HepG2/NTCP cells were also highest for WT isolates and lowest for the PC mutants (Table 2). Consistent with reduced HBeAg expression by BCP mutations, such samples produced higher HBsAg/HBeAg ratio than WT isolates in HepaRG and especially HepG2/NTCP cells (Supporting Information: Table S2).

3.5 PEG could enhance sHBV infectivity, especially following prolonged sample storage

While data presented above revealed higher transmission efficiency of C2 than B2 isolates, and WT virus than PC mutant when same volume of serum sample was inoculated, it remained uncertain whether WT C2 isolates had higher infectivity than WT B2 isolates when same number of viral particles was inoculated. Also, to compare the two cell lines for susceptibility to HBV infection required identical inoculation condition (with or without PEG). We selected four B2 isolates and four C2 isolates for such comparison. All had high HBV DNA titers (1.48–4.18 × 10⁷ genomes/μL) with infectivity in both cell lines when 2 μL of serum sample was inoculated. All harbored WT BCP and PC sequences except for #42, which had BCP mutant as a pure population. This sample was included because it produced highest HBsAg and replicative DNA in both cell types (Figure 1D,E,H,I), with higher HBsAg/HBeAg ratio in HepG2/NTCP than HepaRG cells in contrast to most other samples (Figure 1F,G). Cells in 96-well plates were inoculated with serum samples diluted with normal human serum to reach 10 genomes/cell. When inoculated with 4% PEG, differentiated HepaRG cells produced higher titers of HBeAg and especially HBsAg than HepG2/NTCP cells, most striking for #20, #2, and #50 (Figure 3A–D). HepaRG cells also produced higher HBsAg/HBeAg ratio with the notable exception of #42 (Figure 3E,F), although the higher ratio in HepaRG cells was more striking in the initial screening (Table 3). Inoculation without PEG reduced HBV infectivity for both cell types whether according to HBeAg, HBsAg, or HBV DNA (Figure 3A–D,G,H). The PEG dependence was highest for #27, and least for #40 in HepaRG cells but #50 in HepG2/NTCP cells.

The PEG dependence of sHBV infectivity in HepaRG cells contradicted our previous finding,²⁹ for which viral particles concentrated from relatively fresh serum samples by PEG precipitation or ultracentrifugation through sucrose gradient were inoculated. Data presented in Figure 1 and Figure 3A–H were generated 3–11 months and 16–24 months after sample collection, respectively. Another infection experiment of some of these samples in HepG2/NTCP cells 12 months earlier than Figure 3B,D,F manifested much less PEG dependence, whether according to HBeAg or HBsAg (Figure 3I,J).

3.6 No evidence for higher infectivity of serum samples of three WT C2 isolates than four WT B2 isolates

Among the eight serum samples selected to compare infectivity, #42 had the highest HBV DNA titer. With same genome copy number for inoculation and in the presence of PEG, it no longer produced highest HBsAg titer but rather low HBeAg titer (Figure 3A–D). It produced higher HBsAg/HBeAg ratio than other samples, especially in HepG2/NTCP cells (Figure 3E,F). It also produced much higher replicative DNA than the other seven samples, especially in HepaRG cells (Figure 3G,H, top panels). In this regard, #42 had 1762/1764 BCP mutations as a pure population in contrast to WT sequence of the other seven samples, and such mutations augment HBV genome replication at the transcript level.^{13-16, 24} Comparison among the seven WT samples inoculated with PEG did not reveal higher infectivity of the C2 isolates. First, there was isolate-to-isolate variability within the same subgenotype. Second, findings from the two cell types were not always concordant. Third, ranking based on HBsAg did not always agree with that according to HBeAg. Indeed the seven samples differed in their HBsAg/HBeAg ratios (Figure 3E,F), and ranking from replicative DNA could also differ.

3.7 Three WT C2 clones as 1.1mer construct generated more replicative DNA and virions than three WT B2 clones in transfected HepG2 cells

To avoid complications from viral quasispecies (coexistence of BCP or PC mutant with WT virus), or possible host factors modulating HBV infectivity, we cloned HBV genome from viremic serum samples to generate cHBV particles. In this regard, a 1.1mer construct is often used to maximize pgRNA transcription, genome replication, and virion production.^34-36 For the eight serum samples compared for relative infectivity (Figure 3), three WT B2 isolates (#20, #27, #43) and two WT C2 isolates (#2, #28) were used to generate 1.1mer construct. Another WT C2 isolate for cloning was #55 with high HBV DNA titer and high transmission efficiency in HepG2/NTCP cells (Figure 1). For each isolate a 1.1mer clone with its HBV DNA sequence identical to the consensus sequence of the original serum sample was chosen. In transiently transfected HepG2 cells the 1.1mer construct of the three WT C2 clones in general produced higher level of core protein and replicative DNA than the three WT B2 clones (Figure 4B,C). They secreted more virions and naked core particles according to native agarose gel electrophoresis (Figure 4D). More virion secretion by the C2 clones was confirmed by immunoprecipitation followed by qPCR (Figure 4E).

3.8 The three WT C2 clones had slightly reduced infectivity than the three WT B2 clones

cHBV particles concentrated from culture supernatant of transfected HepG2 cells were inoculated to HepG2/NTCP cells or differentiated HepaRG cells in the presence of 4% PEG. When same genome copy number of viral particles was used, B2 clone 27.4 produced somewhat higher level of HBeAg, HBsAg, and replicative DNA than the two other B2 clones in HepG2/NTCP cells (Figure 5E,F,H). For the three C2 clones the three parameters did not agree with each other. Overall viral signals from C2 clones appeared reduced than B2 clones in this cell line, especially for replicative DNA (Figure 5E,F,H). Consistent with more efficient virion production by the C2 clones as 1.1mer construct in transfected HepG2 cells (Figure 4D,E), inoculation with viral particles concentrated from same volume of culture supernatant led to increased signals for the C2 than B2 clones (compare Figure 5A,B,D with Figure 5E,F,H). In HepaRG cells inoculated with same genome copy number of viral particles, the C2 clones produced lower HBsAg titer and HBsAg/HBeAg ratio, and less replicative DNA than the B2 clones (Figure 5I–L).

3.9 Impact of sHBV-to-cHBV conversion on HBsAg/HBeAg ratio

HepaRG cells produced higher HBsAg/HBeAg ratio than HepG2/NTCP cells, most striking for the genotype D clone serving as a positive control (compare Figure 5C,G,K). After molecular cloning to convert sHBV particles into cHBV particles, the HBsAg/HBeAg ratio in HepaRG cells inoculated with PEG remained similar for sample #20, #27, #43, and #55 but reduced for #2 and especially #28 when compared with data from Figure 1F (fresh serum samples inoculated without PEG) (Table 3). But if comparison was made with stored serum samples inoculated also with PEG (Figure 3E), the ratio was rather increased for #20, #27, #43. In this regard, the HBsAg/HBeAg ratio was reduced in Figure 3E than in Figure 1F for all the samples tested (Table 3). For HepG2/NTCP cells the ratio decreased for most samples. The much higher HBsAg/HBeAg ratio in HepaRG cells than HepG2/NTCP cells sustained or became even more striking (compare Figure 5G with Figure 5K).