Novel mechanism of antibodies to hepatitis B virus in blocking viral particle release from cells


  • Dr. Neumann is on the speakers' bureau of and received grants from Roche, Merck, and Human Genome Sciences. Dr. Ijaz received royalties from Abbott Murex.


Antibodies are thought to exert antiviral activities by blocking viral entry into cells and/or accelerating viral clearance from circulation. In particular, antibodies to hepatitis B virus (HBV) surface antigen (HBsAg) confer protection, by binding circulating virus. Here, we used mathematical modeling to gain information about viral dynamics during and after single or multiple infusions of a combination of two human monoclonal anti-HBs (HepeX-B) antibodies in patients with chronic hepatitis B. The antibody HBV-17 recognizes a conformational epitope, whereas antibody HBV-19 recognizes a linear epitope on the HBsAg. The kinetic profiles of the decline of serum HBV DNA and HBsAg revealed partial blocking of virion release from infected cells as a new antiviral mechanism, in addition to acceleration of HBV clearance from the circulation. We then replicated this approach in vitro, using cells secreting HBsAg, and compared the prediction of the mathematical modeling obtained from the in vivo kinetics. In vitro, HepeX-B treatment of HBsAg-producing cells showed cellular uptake of antibodies, resulting in intracellular accumulation of viral particles. Blocking of HBsAg secretion also continued after HepeX-B was removed from the cell culture supernatants. Conclusion: These results identify a novel antiviral mechanism of antibodies to HBsAg (anti-HBs) involving prolonged blocking of the HBV and HBsAg subviral particles release from infected cells. This may have implications in designing new therapies for patients with chronic HBV infection and may also be relevant in other viral infections. (HEPATOLOGY 2010;)

Viruses elicit a range of antiviral antibodies, but only some of these have direct antiviral activity and are referred as neutralizing antibodies, because they render virions noninfectious by blocking viral entry into cells.1 Such antibodies bind to epitopes that interfere with the interaction of the viral surface protein and its receptor by steric hindrance,2 by directly targeting the receptor-binding site on viruses,3 or by inducing conformational changes that abrogate the functionality of the viral surface protein.4 In addition, the antiviral activities of antibodies against virus particles in the circulation can include clearance via fragment crystallizable (Fc)-mediated effector systems, such as complement-dependent virolysis or phagocytosis.5

In hepatitis B virus (HBV) infection, antibodies directed to a conserved region (a-determinant) of the HBV surface antigen (HBsAg) are known to confer protection by high-affinity binding of HBsAg, the main component of the virus envelope, as well as the 22-nm subviral particles.6 The efficacy of antibodies to HBsAg (anti-HBs) in preventing HBV infection has been established both when given as passive immunoprophylaxis, for example, to prevent mother-to-child HBV transmission or to prevent HBV reinfection of the liver graft following liver transplantation,7 as well as by the success of universal active immunization using recombinant HBsAg, resulting in high anti-HBs titers.8 The mechanisms of anti-HBs protection are not understood, although the common belief is that these are based on binding HBV particles in circulation, thus preventing the infection of liver cells. According to this paradigm, cells that have already been infected will not be affected by anti-HBs. Importantly, the protection achieved by anti-HBs, at least in the liver transplant setting, is not sterile because HBV DNA is detectable in the new liver even in cases with effective prophylaxis.9

Recently, in vitro experiments have shown that HBs-specific immunoglobulin G (IgG) is internalized into hepatocyte-derived cell lines and inhibits the secretion of HBsAg and virions from these cells.10 The HBsAg and anti-HBs were colocalized within the cells, and the specificity of intracellular HBsAg–anti-HBs interaction was further demonstrated by abrogating the anti-HBs inhibitory effect in cells transfected with HBV genomes expressing antibody-escape mutant HBsAg.10 To investigate further the phenomenon of intracellular blocking of HBV release by antibodies and its potential for therapeutic application, we analyzed both in vivo and in vitro the effect of two human monoclonal antibodies to HBsAg, HBV-Ab17 and HBV-Ab19, which have been shown to have high neutralizing activity against HBV.11, 12 We used mathematical modeling of serum HBV DNA and HBsAg levels to gain information about viral dynamics during a single or multiple infusions of a combination of the two monoclonal anti-HBs (HepeX-B) in patients with chronic hepatitis B. We then replicated this approach in vitro, using cells secreting HBsAg, and compared the prediction of the mathematical modeling obtained from the in vivo kinetics.


DMEM, Dulbecco's modified Eagle medium; ELISA, enzyme-linked immunosorbent assay; Fc, fragment crystallizable; HBV, hepatitis B virus; HBsAg, hepatitis B surface antigen; HCV, hepatitis C virus; IgG, immunoglobulin G.

Materials and Methods


Human monoclonal antibodies to HBsAg (HBV-Ab17 and HBV-Ab19) were generated as described.11 The antibodies bind different epitopes on HBsAg; HBV-Ab17 recognizes a conformational epitope, whereas HBV-Ab19 recognizes a linear epitope between amino acids 140-149. The specific activities of HBV-Ab17 and HBV-Ab19 are 554 IU/mg and 2090 IU/mg, respectively, and their affinity constants (Kd) are 7.6 × 10−10 M and 5 × 10−10 M, respectively.12 HepeX-B is a 3:1 (mg:mg) mixture of HBV-Ab17 and HBV-Ab19. The serum half-lives of HepeX-B following a single 10 mg or 40 mg infusion in healthy volunteers were 22.3 ± 5.5 and 24.2 ± 4.4 days, respectively (Rachel Eren and Shlomo Dagan, unpublished data). For the in vitro experiments, a human monoclonal antibody (IgG1) against the envelope protein (E2) of hepatitis C virus (HCV-Ab68) was used as an isotype control.

Clinical Design

Serum HBV DNA and HBsAg levels were determined in patients with chronic hepatitis B, who participated in Phase 1A and 1B clinical trials for evaluation of HepeX-B.13 Phase 1A was an open-label, single-dose study with a total of 15 patients, each receiving a single dose of HepeX-B (range = 0.26-40 mg) by an intravenous infusion over 2-8 hours. Serum samples were taken at 0, 0.5, 1, 2, 4, 8, 12, 24, 48, and 96 hours after infusion. Phase 1B was an open-label study with ascending multiple doses of HepeX-B.13 Four sequential cohorts of three patients each were given 10, 20, 40, or 80 mg as four identical doses of HepeX-B at weekly intervals. Serum samples were taken at 0, 4, 12, and 24 hours after each infusion. Serum HBV DNA levels were quantitated by Amplicor HBV Monitor assay with a limit of detection 200 copies/mL (Roche Diagnostics, Branchburg, NJ). Serum HBsAg levels were determined by an automated immunoassay (IMX system; Abbott GmbH Diagnostika, Wiesbaden-Delkenhaim, Germany), using a purified HBsAg preparation as standard. The limit of detection of this assay is 0.125 ng/mL.

Design of In Vitro Experiments

The PLC/PRF/5 cell line was established from hepatocellular carcinoma.14 These cells contain integrated HBV DNA fragments and produce 22-nm noninfectious HBsAg particles.15-17 The HBsAg production was shown to be constant on a per-cell basis during culture.18, 19 In the present study, PCL/PRF/5 cells were cultured in Dulbecco's modified Eagle medium (DMEM; Invitrogen, Paisley, UK), supplemented with 10% fetal bovine serum (Invitrogen), 500 U/mL penicillin, 500 μg/mL streptomycin, and 2 mM L-glutamine. The cells were seeded in 24-well plates at 50,000 per well. After 48 hours, the cells were confluent, which was the starting time point (T0) of the experimental conditions outlined below.

(1) Internalization of Anti-HBs and Effect on Intracellular HBsAg.

At T0, the supernatants were removed and replaced with medium only (DMEM/5% fetal bovine serum, control); medium plus HBV-Ab17 at two concentrations (0.2 and 0.5 mg/mL); medium plus HBV-Ab19 (0.2 and 0.5 mg/mL); HepeX-B (0.5 mg/mL); or medium plus isotype IgG (0.2 and 0.5 mg/mL) as a further control. After 48 hours (T48), the supernatants and the cells were collected separately. The cell lysates were tested for cellular IgG and HBsAg by western blot. The HBsAg secreted in the supernatants was quantitated by enzyme-linked immunosorbent assay (ELISA).

(2) Kinetics of HBsAg Secretion in the Presence of Anti-HBs.

In this set of experiments, at T0 the supernatants of PLC/PRF/5 cells were replaced with: (1) medium only; (2) isotype IgG control (0.5 mg/mL); (3) human AB serum, as another source of nonimmune IgG (0.5 mg/mL); (4) HBV-Ab17 (0.5 mg/mL); (5) HBV-Ab19 (0.5 mg/mL); or (6) HepeX-B (0.5 mg/mL). The cells were cultured continuously for 48 hours, during which period an aliquot of the supernatants was taken (without adding new medium) at 3, 6, 12, 24, 36, and 48 hours, and the HBsAg levels were determined by ELISA.

(3) Kinetics of HBsAg Secretion from Cells Pretreated with Anti-HBs.

The supernatants at T0 were replaced with the same controls or antibodies, as outlined above. During the first 24 hours, an aliquot of the supernatants was collected at 3, 6, 12, and 24 hours. After 24 hours (T24), the supernatants in all wells were replaced with fresh medium, without nonimmune IgG or anti-HBs, and the cells were kept in culture for a further 24 hours. During the second 24-hour period, aliquots were collected at the same time points as during the first period, i.e., at 27, 30, 36, and 48 hours, and the HBsAg levels were quantitated by ELISA.

Immunoblot Analysis

After 48 hours, the cells were trypsinized and washed two times in phosphate-buffered saline and resuspended in lysis buffer. The presence of human IgG and HBsAg in PLC/PRF/5 cells was analyzed by western blot in cells cultured in the presence or absence of monoclonal anti-HBs IgG, as described.10

ELISA for Quantitation of Secreted HBsAg

Aliquots of cell culture supernatants were added to wells coated with P2D3 monoclonal anti-HBs,20 and the detection was with peroxidase-conjugated monoclonal anti-HBs (D2H5).21 The HBsAg in supernatants was quantitated in nanograms per milliliter, based on a parallel testing of eight standard amounts of HBsAg between 5 and 2000 ng, which were included in the same run of the assay. Details of the western blot and ELISA methodology are given in the Supporting Material.

Mathematical Modeling

In Vivo and In Vitro Models.

Mathematical modeling of the antiviral effect of the antibodies on viral kinetics in patients was performed by extension of the standard model by Neumann et al.22 to examine the dynamics of HBsAg particles and the hepatitis B virions in the serum, as measured by HBV DNA, and a number of possible antiviral effects. In addition, we developed a model in order to simulate the in vitro kinetics of supernatant HBsAg produced by PLC/PRF/5 cells in culture and the possible effects of antibodies on that process. Both models, their parameters' estimates and fitting procedures are explained in detail in the Supporting Material.


Viral Dynamics After a Single Anti-HBs Infusion

Analysis of viral kinetics after a single infusion of 40 mg HepeX-B showed a rapid HBV DNA decline starting 0.5 hours after initiation of infusion and continuing throughout the 8-hour infusion period, reaching 2.5-3.3 log10 copies/mL reduction from baseline with a half-life of 0.33-0.53 hours (Fig. 1 and Table 1). A parallel HBsAg decline to undetectable levels (<0.125 ng/mL) was observed in all three patients, with a half-life of 0.09-0.19 hours and maximal decline of 4.3-4.6 log10 copies/mL relative to baseline.

Figure 1.

HBV DNA and HBsAg kinetics in vivo during and following HepeX-B infusion. Plasma HBV DNA and HBsAg were measured frequently during a single infusion of 40 mg HepeX-B (A,B; three patients, respectively), and following multiple infusions of 40 mg (C,D) or 80 mg (E,F). An immediate rapid decline after the beginning of infusion is observed for HBV DNA and is even more pronounced for HBsAg, with the exception of one nonresponding patient, shown as triangles in (C) and (D). After each infusion, a rebound in viremia is observed, which was delayed in some of the patients. Dashed lines represent the end of infusion period. UD, undetectable, i.e, below assays limit.

Nonlinear fitting of the HBV DNA and of HBsAg kinetics to a viral dynamics model (Supporting Material, Equations 1-4) allowed us to test various hypotheses for the antiviral mechanism of HepeX-B (Fig. 2). First, blocking de novo infection (1 ≥ η > 0) cannot be the major antiviral mechanism because it can only result in a viral decline slope of the order of the loss rate of infected cells, half-life larger than 1 day, which is not in agreement with the rapid viral decline observed here. Second, accelerated loss of infected cells (k > 1) or blocking of virion production and/or release from infected cells (0 < εV ≤ 1) by themselves are also not sufficient explanations, because the expected decline would follow the clearance rate of serum HBV virions. The half-life of HBV virions (ln(2)/cv) ranges between 3-24 hours, as found in previous studies of HBV kinetics,15, 23-26 which is too slow compared to the very rapid decline observed during HepeX-B infusion (Fig. 1). Third, assuming an accelerated clearance of HBV virions from circulation (aV > 1) cannot by itself explain the rapid decline in serum HBV DNA (Fig. 2). The observed half-life of the order of 0.33-0.53 hours gives a minimal (maximal) estimate of accelerated clearance of HBV particles of aV = 5.7 (72.7), by assuming the most rapid (slow) observed virion half-life of 3 (24) hours. However, such acceleration of clearance can give rise to a decline of only log (1/aV) = 0.75-0.86 log10 copies/mL (Fig. 2), because viral load during accelerated clearance will reach a new steady state at V0/a (Supporting Material, Equation 6). This is considerably less than the viral decline observed here, and thus accelerated clearance is a necessary but not sufficient explanation of the antiviral mechanism.

Figure 2.

Simulation of various models for a potential antiviral mechanism of HepeX-B in comparison to HBV DNA decline. When the antiviral mechanism is assumed to be blocking infection (η > 0, as expected from neutralizing antibodies), the resulting decline slope (upper most dotted line) is maximally the loss rate of infected cells (half-life longer than 1 day) and thus cannot explain the fast decline observed in vivo (red squares). Also when assuming that the mechanism is blocking virion release from infected cells (εv > 0) or increased loss of infected cells (κ > 1), the resulting slope (dash-dotted line), which corresponds to the intrinsic half-life of serum HBV virions (longer than 3 hours), is still too slow. Only when assuming that HepeX-B accelerates (here, av = 44) the clearance of virions from circulation, can the simulated decline slope match that of the in vivo data (dashed and solid lines). However, if acceleration of the clearance is the only mechanism assumed (or if it is combined with blocking de novo infection, not shown) then the viral load at end of infusion is expected to be only 44-fold lower than baseline (dashed line), instead of 751-fold. On the other hand, if an acceleration of av = 751 is assumed, the expected slope of decline (bottom dotted line) would have been much faster than that observed. The in vivo decline can be simulated (solid line) if one assumes that HepeX-B provides both acceleration of virion clearance from serum (av = 44) as well as partial blocking εv = 94%) of release and/or virion production from infected cells.

The observed HBV DNA kinetics during HepeX-B infusion can be explained by assuming a combination of two mechanisms of action: an accelerated HBV clearance from the circulation (aV > 1), mediated by the infused antibodies, together with partial blocking of virion release from infected cells (1 > εV > 0), as shown in Fig. 2. Using this assumption, we performed nonlinear fitting for all patients (Fig. 3 and Table 1A), assuming either a slow or a rapid intrinsic half-life of HBV virions, thus giving, accordingly, low and high estimates of the effectiveness in blocking virion release of εV = 76.2%-96.1% (97%-99.5%).

Table 1A. Results of Non-Linear Fitting of the In Vivo Frequent Sampling Data
HepeX-B Dose (mg)Patient NumberBaseline ALT (U/L)HBV-DNA DynamicsHBsAg Dynamics
Baseline HBV-DNA (log cp/mL)Half-Life Decline (hours)Total Decline (log cp/mL)Clearance AccelerationBlocking EffectivenessHBeAg StatusBaseline HBsAg (log ng/mL)Half-Life Decline (hours)Total Decline (log ng/mL)Clearance AccelerationBlocking Effectiveness
Low (c = 5.5)High (c = 0.69)low (c = 0.69)high (c = 5.5)low (c = 0.69)high (c = 0.02)Low (c = 0.02)High (c = 0.69)
Figure 3.

Nonlinear fitting of HBV DNA and HBsAg kinetics with a model assuming a combined effect of blocking virion release together with acceleration of virus clearance from circulation. This model yields good nonlinear fitting simultaneously for HBV DNA (squares, left y-axis) and HBsAg (triangles, right y-axis) in the three patients with frequent samples (A-C) and allows us to estimate the viral dynamics parameters (Table 1A). Because HBsAg became undetectable in all patients, the equivalent parameters are only minimal estimates corresponding to the solid blue curves, but in fact could be faster (e.g., the dashed curves). In one patient (C), we observed a slower relapse, thus indicating a gradual decline in the effectiveness of blocking viral release (dotted line).

The analysis of HBsAg kinetics following HepeX-B infusion yields similar conclusions with higher acceleration of clearance and effectiveness in blocking release (Fig. 3). Analysis of HBsAg decline during treatment with lamivudine shows decline with a half-life of 38 days, which we use as a maximal estimate of HBsAg half-life, with a minimal estimate of 1 day (data not shown). Nonlinear fitting of HBsAg decline during HepeX-B gives a half-life of 0.09-0.19 hours, thus corresponding to a minimal (maximal) estimate of the acceleration of HBsAg clearance from circulation of aA = 126-282 (4,800-10,729) times an intrinsic half-life of 1 (38) days. As for the HBV DNA kinetics, accelerated clearance of HBsAg cannot by itself explain the total decline, and thus a maximal (minimal) estimate of effectiveness in blocking release of HBsAg particles from infected cells is εA = 98.6-99.5% (46.2%-82.4%). Note, that because all patients reached undetectable levels of HBsAg, the decline is probably larger and thus the estimates of effectiveness of blocking HBsAg release are most probably underestimated here.

Viral Dynamics After Multiple Anti-HBs Infusions

In addition to the three patients who received a single infusion of 40 mg HepeX-B, six patients received multiple HepeX-B doses of 40 and 80 mg (Fig. 1C-F), but those patients had less frequent sampling and the duration of infusion was only 4 hours. We have therefore tested in the first three patients with frequent samples whether the estimation of decline half-life and total decline from 0.5 to 4 hours is accurate enough. We found that although both values were slightly underestimated with less frequent samples, the estimate of effectiveness in blocking release with the 4-hour sample was still accurate within 90%-110% of the nonlinear fitting (Table 1B). In five of six patients who received multiple doses of HepeX-B, a rapid decline with a half-life 0.31-0.66 hours and magnitude of 1.6-3.3 log10 copies/mL was observed. The mathematical analysis, with low and high estimates of the accelerated clearance and effectiveness, indicates also in these patients that the antiviral activities of HepeX-B antibodies include both antibody-mediated accelerated clearance and partial blocking of viral particles release from infected cells (Table 1B). One patient (patient 202) with relatively high baseline levels of HBV DNA and HBsAg did not show significant declines during HepeX-B infusions.

Table 1B. Analysis of In Vivo Kinetics Based on Nonfrequent Sampling
  1. *Patient participated in both phase 1A and phase 1B studies. NEG, negative; Non-Resp, nonresponder; POS, positive.

4 × 40203386.70.353.08.668.992.8%99.1%POS3.20.294.183314277.7%99.41%
4 × 40105474.50.661.64.636.43.1%87.9%NEG3.20.333.672273633.3%98.2%
4 × 40202198.813.90.1Non-RespNon-RespPOS4.65.10.2Non-RespNon-Resp
4 × 803011376.50.313.39.676.596.2%99.5%NEG3.70.294.182310875.5%99.36%
4 × 80106877.70.343.18.971.394.1%99.3%NEG2.90.333.672274635.0%98.3%
4 × 80311*1205.50.382.77.862.787.8%98.5%NEG3.40.284.284320681.2%99.51%
Mean  6.40.442.57.358.669.6%96.2%

Both HBV DNA and HBsAg levels returned to baseline levels ±0.5 log10 within 24-48 hours after the infusion in the three patients with frequent samples, and within 1-7 days in the six patients with less frequent samples. There was no cumulative effect of HepeX-B on the decline of HBV DNA or HBsAg in the patients who received 4 weekly infusions. However, HBV DNA and HBsAg levels at 24 hours after infusion were, in general, lower than expected from the rebound kinetics predicted by the model, if the antiviral effect of HepeX-B disappears immediately after the end of the infusion (Supporting Material, Equation 9). Notably, for the 80 mg dose (Fig. 1E,F), at 24 hours after 5 of 12 infusions, HBsAg was still undetectable and HBV DNA was at least 3 log10 lower than baseline. Simulation of the slow rebound kinetics indicates a delay (10-16 hours) in release of viral particles after infusion and a prolonged effect of the antibodies after the end of infusion with a half-life of the order of 1-10 days (Fig. 3C). Because of the infrequent sampling after the infusion it is not possible to quantify these effects precisely.

In Vitro Effects of Anti-HBs

The assumption that HepeX-B can block the release of viral particles from cells was tested in a series of in vitro experiments using PLC/PRF/5 cells, which are known to have stable production of HBsAg.16-19, 27 The western blot analysis of cell lysates after 48-hour culture showed dose-dependent internalization of both control IgG (nonspecific for HBV), as well as of IgG with anti-HBs specificity (Fig. 4A), which is in line with our previous findings.10 The cellular uptake of HBV-Ab19 appears to be higher than HBV-Ab17, as indicated by the different density of the western blot bands (Fig. 4A). In the same cytoplasmic extracts, the western blot revealed a marked intracellular accumulation of HBsAg, which was observed only in cells cultured in the presence of anti-HBs, but not in control cells (Fig. 4B). The combination of HBV-Ab17 and HBV-Ab19 (HepeX-B) had a greater effect for HBsAg retention within the cells, than did each of these two antibodies alone.

Figure 4.

Immunoblot analysis of cytoplasmic extracts from PLC/PRF/5 cells cultured in the presence of human monoclonal anti-HBs antibodies. HBV-Ab17 and HBV-Ab19 were used individually at two concentrations (0.5 and 0.2 mg/mL), or as mixture (HepeX-B). Human IgG isotype and medium only served as controls. (A) Detection of cell-associated IgG-γ chain (50 kDa). Both IgG isotype control and IgG anti-HBs show dose-dependent uptake in the cells. (B) Detection of HBsAg (24 and 27 kDa). The cellular uptake of specific anti-HBs antibodies leads to dose-dependent increase of HBsAg within the cells in contrast to cell-associated isotype IgG or medium only control.

We also determined the effect of anti-HBs (HBV-Ab17 or HBV-Ab19 alone, or in combination as HepeX-B) on the kinetics of HBsAg secretion (Fig. 5A). In the control supernatants, the HBsAg levels rose rapidly in the first hours and then continued with a slower increase to an average level of 3054 ± 342 ng. However, in the presence of HBV-Ab17 (or HBV-Ab19), the HBsAg levels were markedly reduced to only 0.4 (50.5) ng, which is 3.9 (1.8) log10 ng lower than the controls. Moreover, in the presence of HepeX-B, HBsAg remained undetectable (< 0.1 ng) throughout the 48 hours of the experiment. Fitting of the full kinetics data with the in vitro model (Supporting Material, Equations 10 and 11) indicated that HBV-Ab17 or HBV-Ab19 give rise to a partial blocking of HBsAg release with effectiveness of 99.98% or 98.8%, respectively. A minimal estimate of the effectiveness in blocking release in the presence of HepeX-B is 99.998%. Similar results were obtained by estimation of HBsAg at 48 hours using the analytical solution (Supporting Material, Equations 12 and 13).

Figure 5.

Supernatant HBsAg kinetics in vitro shows blocking of particle release by PLC/PRF/5 cells incubated with HepeX-B. (A) Significantly lower HBsAg levels were observed when cells were incubated for 48 hours with HBV-Ab17 (red squares), HBV-Ab19 (green triangles) or HepeX-B (blue diamonds) as compared to medium (empty diamonds), isotype-control (stars), or human AB serum (empty circles). Fitting with the mathematical model (solid lines) indicates blocking of particle release. (B) Cells were incubated with monoclonal antibodies for 24 hours, after which the supernatants were replaced with medium only. Blocking release continued in the absence of anti-HBs in the supernatant, with a gradual decline of its effectiveness. UD, undetectable, i.e, below assay limit.

In order to verify whether the antibodies to HBsAg have a prolonged effect on HBsAg release from PLC/PRF/5 cells, we repeated the experiment, but after the first 24 hours, we replaced the supernatants with fresh medium and continued measuring the HBsAg levels in supernatants for an additional 24 hours (Fig. 5B). During the first 24 hours, when either control medium or antibodies to HBsAg were present, the kinetics were similar to the first experiment. After we replaced the supernatants with fresh medium at 24 hours, the HBsAg levels in the new supernatants started from 0 and the kinetics of HBsAg secretion from control cells (treated with medium only) was slightly faster, probably due to a larger number of cells per well relative to the first 24 hours. In contrast, cells that were treated initially with HBV-Ab17 (undetectable HBsAg during first 24 hours) showed slower kinetics of HBsAg secretion (i.e., de novo increase in supernatants) also after the antibodies were removed from the medium. Fitting of the kinetics with the model indicated a delay of 2.6 hours in HBsAg release from the cells and also a slow decline, with a half-life of 6.9 days, of the effectiveness in blocking release from an initial 99.99% at 24 hours. Similarly, a delay of 10.7 hours and a slow decline of the blocking release effectiveness with a half-life of 11.5 days were observed for the cells initially treated with HepeX-B.


By modeling viral dynamics during in vivo and in vitro treatment with antibodies to HBsAg, this study reveals a novel antiviral mechanism of antibodies to HBV. Apart from the “conventional” antiviral activities against viral particles in the circulation, anti-HBs antibodies are internalized in liver cells and exert intracellular antiviral activities with prolonged blocking of viral particles release from infected cells. Both experimental approaches—in vivo infusions of HepeX-B and in vitro treatment of cells with the same antibodies—produced concordant results with a reduction of viremia and/or HBsAg titers, as well as a prolonged antiviral effect after anti-HBs administration. The combination of two human antibodies—HBV-Ab17, recognizing a conformational epitope, and HBV-Ab19, recognizing a linear epitope on HBsAg—had an additive effect. Viral dynamics revealed that when used alone, HBV-Ab17 had a stronger effect, but with a shorter duration when the antibody was not present (Fig. 5B). In contrast, HBV-Ab19 had a weaker effect in blocking the release of viral particles from the cells, but its effect was more prolonged which may be due to a greater uptake within cells, as indicated by the western blot (Fig. 4).

Experimental data indicate that in some viral infections, antibody binding to viral antigens expressed on the cell surface can modulate viral replication within cells. For example, treatment of alphavirus-infected rat neurons with monoclonal antibodies to E2 envelope protein was found to mediate viral clearance from the neurons28; antibodies to measles virus added to virus-infected cells were shown to interfere with viral protein expression inside the cells29; the addition of pseudorabies-specific immunoglobulins to pseudorabies-infected monocytes induced internalization of plasma membrane–bound viral protein via endocytosis.30 A different type of antibody-mediated interaction with infected cells was observed by IgA anti-hemagglutinin antibodies.31 Polymeric IgA anti-hemagglutinin was found to be actively transported into epithelial cells by polymeric Ig receptor and to mediate intracellular neutralization of influenza virus by binding to viral proteins within the cell and preventing viral assembly.31

This study reveals that the antiviral effect of anti-HBs against HBV involves not only binding of viral particles in the circulation, but it also involves intracellular antiviral activity by blocking viral particle release from the cells. We previously demonstrated that anti-HBs is endocytosed into hepatocyte-derived cell lines regardless of the presence or absence of HBsAg.10 This is likely to occur as a receptor-mediated endocytosis of IgG via the major histocompatibility complex class I–like Fc-receptor, FcRn. We have shown that FcRn is expressed on several liver cell lines and Fc elimination abrogated the IgG biding to the cells, as well as its effect on HBsAg secretion.10 FcRn is the transport receptor for IgG and protects IgG from catabolism after entry into cells.32, 33 This process is likely to operate during chronic HBV infection, because anti-envelope antibodies have been detected in the serum of virtually all patients with chronic hepatitis B, when sensitive assays are used.34 Intracellular binding and blocking the secretion of HBV particles may have a role for containment of HBV when it is present at low level within cells, for example, in subjects with spontaneously resolved HBV infection35 or in liver transplant recipients having effective long-term hepatitis B Ig prophylaxis without clinical HBV recurrence.9, 36

The antiviral activity of HBV-neutralizing antibodies may have clinical implications for treatment of chronic hepatitis B. Posttreatment rebound of HBV replication occurs frequently after stopping direct antivirals even after prolonged treatment for many years. Application of anti-HBs, in combination with a potent antiviral agent directly blocking HBV replication, will target different sites of the HBV replicative cycle within cells and may reduce relapse rates, analogous to the approach that was applied successfully after highly active antiretroviral therapy withdrawal in patients with human immunodeficiency virus infection.37 Furthermore, after prolonged and effective control of HBV replication in patients treated with potent antivirals, an add-on application of anti-HBs may facilitate HBsAg clearance in appropriately selected patients.


The authors thank Eithan Galun, principal investigator of the clinical investigation,13 which provided data for mathematical modeling, and Dov Terkieltaub for technical assistance.