Hepatitis B virus infection initiates with a large surface protein–dependent binding to heparan sulfate proteoglycans


  • Potential conflict of interest: Nothing to report.


Contrary to many other viruses, the initial steps of the hepatitis B virus (HBV) infection, including attachment to hepatocytes, specific receptor interactions, and membrane fusion, are unsolved. Using HepaRG cells as an in vitro cell culture system, we here report that HBV entry into hepatocytes depends on the interaction with the glycosaminoglycan (GAG) side chains of cell-surface–associated heparan sulfate proteoglycans. Binding to GAGs requires the integrity of the pre-S domain as a part of the large (L-) viral envelope protein. HBV infection was abrogated by incubation of virions with heparin, but not the structurally related GAGs chondroitin sulfate A, B, and C. Infection was also abolished by suramin, a known inhibitor of duck hepatitis B virus infection or highly sulfated dextran sulfate. Polycationic substances such as poly-L-lysine, polybrene, and protamine also prevented infection, however, by addressing cellular components. Enzymatic removal of defined acidic carbohydrate structures from the cell surface using heparinase I/III or the obstruction of GAG synthesis by sodium chlorate inhibited HBV infection of HepaRG cells and, moreover, led to a reduction of HBV cell surface binding sites. The biochemical analysis showed selective binding of L-protein–enriched viral particles (virions or filaments) to heparin. GAG-dependent binding of HBV was improved by polyethylene glycol, a substance that specifically enhances HBV infection. Conclusion: HBV infection requires the initial attachment to the carbohydrate side chains of hepatocyte-associated heparan sulfate proteoglycans as attachment receptors. This interaction initializes the multistep entry process of HBV and cannot be bypassed by alternative routes. (HEPATOLOGY 2007;46:1759–1768.)

Hepatitis B virus (HBV) represents the medically important prototype of a family of small, enveloped DNA viruses (hepadnaviridae), which are widespread in mammals and birds. All members of this family possess high species and, at least with respect to the preferential site of replication, also tissue specificity.1 Hepadnaviral infections cause transient or persistent liver inflammation, with approximately 360 million people worldwide being chronic HBV carriers and approximately 650,000 deaths each year attributable to HBV-related progressive liver failure (cirrhosis or hepatocellular carcinoma).2

For a long time, in vitro studies of the HBV life cycle, particularly the early infection events (attachment, receptor binding, and fusion), were only feasible using primary human hepatocytes (PHH). The application of PHH was, however, limited through restrictions in accessibility and high variations in susceptibility to HBV infection.3 The establishment of the highly differentiable human hepatoma cell line HepaRG resolved this issue and facilitated the systematic analysis of the early HBV infection events.4

The HBV envelope consists of the large (L), middle (M), and small (S) surface proteins. They are encoded in a single open reading frame containing 3 in-phase start codons.1 L- and M-proteins share the 226 amino acids of the S-domain. The M-protein is characterized by an N-terminal extension of S by 55 amino acids (called preS2), whereas further extension with 108 (genotype D) or 119 (genotypes A and C) amino acids (designated preS1) define the L-protein.

Several observations indicate an essential role of the preS1 domain within the viral L-protein for early infection events: (1) DNA-containing virions, in contrast to the huge excess of empty spherical subviral particles, are enriched in L-protein.5 (2) In vitro infection of PHH with HBV requires myristoylation of glycine-2 in the preS1 domain.6, 7 (3) The integrity of the N-terminal 77 amino acids is essential for infectivity, whereas the preS2 domain is dispensable.8, 9 (4) Acylated peptides comprising the N-terminal 47 amino acids of the preS1 domain abolish HBV infection in vitro10–13 and in vivo (Petersen et al., unpublished observations, 2007).

For HBV, a variety of cellular binding factors have been described in the past3; however, none of them has been experimentally proven to be involved in HBV entry in infection studies.

Glycoproteins, glycolipids, and proteoglycans are constitutive components of the extracellular matrix and the plasma membrane. They have been selected by different pathogens, including protozoae (for example, leishmania14), bacteria (for example, Bordetella pertussis15), and viruses (such as herpes simplex viruses16 or human immunodeficiency virus17) as attachment sites. However, productive entry, resulting in the nucleocapsid release into the cytosol, requires additional steps, such as high-affinity receptor interactions and, for enveloped viruses, membrane fusion.

Glycosaminoglycans (GAGs) are unbranched polysaccharides composed of hexosamine/hexuronic acid repeats. They acquire negative charges through N- and O-sulfation of the carbohydrate moieties in the Golgi apparatus.18 GAGs are bound to core proteins or lipids to form glycoconjugates. GAG–ligand interactions are complex and characterized by either merely electrostatic forces or specific interactions involving hydrogen bonding or hydrophobic contacts.19 The strength and specificity of these contacts are determined by the cell type–specific structural heterogeneity of GAGs, reflected in variations in length and composition of the carbohydrate chains. Interestingly, hepatocytes constitute a unique GAG composition.20, 21 This, and the distinct architectural arrangement of hepatocytes in the liver,22 might contribute to the hepatotropism of pathogens, although their primary interaction is not exclusively restricted to liver cells.

Here we investigated the role of GAGs in HBV entry into hepatocytes. We provide evidence that HBV infection depends on these cell surface structures as attachment receptors. We further demonstrate that binding to heparin, a soluble GAG requires the integrity of the preferentially virion-associated L-protein, suggesting selectivity in the binding of virions to hepatocytes.


CHO, Chinese hamster ovary; DHBV, duck hepatitis B virus; ELISA, enzyme-linked immunosorbent assay; GAG, glycosaminoglycan; ge, genome equivalent; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HSPG, heparan sulfate proteoglycan; IU, International Units; L-protein, hepatitis B virus middle surface protein; LDH, lactate dehydrogenase; M-protein, hepatitis B virus medium surface protein; PCR, polymerase chain reaction; PEG, polyethylene glycol; PHH, primary human hepatocytes; S-protein, hepatitis B virus small surface protein.

Materials and Methods


Heparin, chondroitin sulfate A, B, and C, heparan sulfate, hyaluronic acid, suramin, dextran, dextran sulfate, sodium chlorate, protamine, poly-L-lysine, hexadimethrine bromide (polybrene), de-N-sulfated heparin, de-N-sulfated acetylated heparin, heparinase I, and heparinase III were purchased from Sigma-Aldrich.

Cell Lines.

HepaRG cells were cultivated as described.4 Chinese hamster ovary (CHO) cells K1 and pgsA74523 were grown in Dulbecco's modified Eagle's medium with the addition of 10% fetal bovine serum, 100 U/mL penicillin,and 100 μg/mL streptomycin.

Infection of HepaRG Cells With HBV.

As an infectious inoculum, we used an approximately 100-fold concentrated culture supernatant of HepG2.2.15 cells.12 If not stated otherwise, differentiated HepaRG cells were incubated with a 1:20 dilution of this virus stock in medium supplemented with 4% PEG 8000 (Sigma-Aldrich) for 16 hours at 37°C. At the end of the incubation, the cells were washed and further cultivated. Medium was changed every 3 days. To quantify the infection, hepatitis B surface antigen (HBsAg) and hepatitis B e antigen secreted into the culture supernatant from day 7 to 11 postinfection was determined by enzyme-linked immunosorbent assay (ELISA).

Infection competition experiments were performed in presence of the corresponding drug during virus inoculation. For pre-incubation studies, HepaRG cells or virus were mixed for 0.5 or 1 hour at 37°C with the substance followed by its removal before infection. All experiments were performed in duplicate and repeated at least 2 times independently. To analyze possible cytotoxic effects of the reagents, we performed lactate dehydrogenase (LDH) release assays (Cytotoxicity Detection KitPLUS, Roche Applied Science).

Inhibition of Cellular GAG Sulfation by Sodium Chlorate.

To prevent the sulfation of cell-associated proteoglycans, HepaRG cells were cultivated for 48 hours at 37°C in the presence of sodium chlorate. Subsequent virus inoculation was performed in the presence of the inhibitor.

Enzymatic Removal of GAGs From the Surface of HepaRG Cells.

The lyases were solved in their digestion buffers according to the manufacturers' protocol, added to prewashed cells, and incubated for 1 hour at 37°C. Subsequently, the cells were washed, and binding (4 hours, 37°C) or infection was performed in the absence of the enzymes. Enzyme concentrations are expressed in international units (IU) per milliliter (1 IU = 600 Sigma units).

Quantification of Cellular HBV Binding.

Attachment of HBV to cells was determined by quantitative polymerase chain reaction (PCR) detecting the viral DNA. Binding/uptake was performed for 4 hours at 37°C. Unbound virus was removed by washing. Viral DNA was prepared from the cellular lysates with the NucleoSpin Blood Kit (Macherey-Nagel) according to the manufacturers' protocol. For the SybrGreen-based real-time PCR (Invitrogen), the primer set Taq-HBV-F (5′-TCCCAGAGTGAGAGGCCTGTA-3′) and Taq-HBV-R (5′-ATCCTCGAGAAGATTGACGATAAGG-3′) was used. The reaction process started at 50°C for 2 minutes and 95°C for 15 minutes followed by 45 cycles at 95°C for 15 seconds and 60°C for 1 minute. All experiments were performed at least twice.

Binding of HBV Particles to Immobilized Heparin.

Concentrated culture supernatants of HepG2.2.15 cells were applied to a 1 mL Hi-Trap Heparin HP column (GE Healthcare) equilibrated with 0.02 M Tris/Cl (pH 7.4), 0.14 M NaCl. After loading, the column was washed with 10 to 20 column volumes of equilibration buffer. Stepwise elution was performed with 0.24 M, 0.34 M, 0.54 M, 1 M, and 2 M NaCl in 0.02 M Tris/Cl (pH 7.4). Fractions of 1 mL were collected and analyzed by western blot, immune- and DNA-dot blot using antibodies against the HBV L- (Ma18/75) and S- (goat-anti-HBsAg, Fitzgerald) protein or an HBV-specific [32P]-labeled DNA probe.

Trypsin Digestion of Viral Particles.

Elution fractions from a heparin column were adjusted to 140 mM NaCl in 0.02 M Tris/Cl (pH 7.4). These fractions were treated with 12.5 μg/μL trypsin for 5 hours at 37°C. After the incubation, the protease was inactivated by the addition of aprotinin (Fluka).


HBV Infection and Binding Is Specifically Inhibited by Heparin.

To test whether soluble GAGs inhibit HBV infection, we performed infection competition experiments in the presence of increasing concentrations of heparin, heparan sulfate, chondroitin sulfate A, chondroitin sulfate C, or dermatan sulfate (chondroitin sulfate B). The HBsAg secreted into the medium from day 7 to 11 postinfection was determined. To affirm that the measured HBsAg represents newly synthesized S-protein, we included a control inhibition experiment using the previously described entry inhibitor HBVpreS/2-48myr,10 which led to a reduction of HBsAg secretion below the detection limit (data not shown). As shown in Fig. 1A, heparin (approximately 2.4 sulfate groups/disaccharide) abolished HBV infection with a concentration that inhibits 50% of approximately 9.4 μg/mL. Heparan sulfate, a lower sulfated heparin derivative (0.8-1.4 sulfate groups/disaccharide) showed a reduction to 60% at 300 μg/mL. In contrast, GAGs with sulfate contents below 1.1 sulfate groups/disaccharide, such as chondroitin sulfate A and C but also dermatan sulfate with approximately 1.1 sulfate group/disaccharide, exhibited no effect on infection. To test whether heparin addresses viral structures, we preincubated HBV with heparin (630 μg/mL) and removed unbound GAGs before infection of HepaRG. The infection was reduced to 8% in comparison with the control (Fig. 1B, left). At no concentration could significant cytotoxic effects be observed (Supplementary Fig. 1). To examine whether this inhibition is attributable to an interference with cellular attachment, we analyzed the binding of heparin pretreated viruses to HepaRG cells by a quantitative real-time PCR. The quantification showed a decreased viral binding of 54% of the control (Fig. 1B, right). To exclude an effect on replication steps after virus binding, we added heparin to HepaRG cells at different times during and after inoculation with the virus. Heparin showed a time-dependent reduction in its ability to interfere with HBV infection when applied during inoculation (Supplementary Fig. 2A). At later times (48 and 72 hours) after initiation of infection, heparin had no effect. This emphasizes the exclusive action of heparin on an early step of HBV infection. The same held true for the inhibitors described later (Supplementary Fig. 2B).

Figure 1.

Heparin specifically inhibits HBV binding to and infection of HepaRG cells. (A) HepaRG cells were infected with HBV in the presence of heparin, heparan sulfate, chondroitin sulfate A, chondroitin sulfate C, and dermatan sulfate at increasing concentrations (0-300 μg/mL). HBsAg secreted from day 7 to 11 postinfection was determined by an ELISA. The values are given as percentage of the uncompeted control infection. The sensitivity of the ELISA (cutoff) is indicated by the dotted line. (B) HBV was preincubated with heparin (630 μg/mL) for 30 minutes at room temperature. Unbound GAGs were removed by ultrafiltration. Infection (left) was performed with the pretreated HBV particles under standard conditions. The amount of HBsAg in the supernatant of day 7 to 11 postinfection was determined. In parallel, binding (4 hours at 4°C) (right) of the heparin-treated HBV particles to HepaRG cells was quantified by real-time PCR and presented as cell-associated genome equivalents per well (106 cells).

Figure 2.

The degree of sulfation correlates with the ability to interfere with infection. (A) HepaRG cells were infected in the presence of unmodified, partially de–N-sulfated or completely de–N-sulfated heparin. The inoculum with the heparins was incubated overnight on the cells. At 11 days postinfection, the culture supernatants from day 7 to 11 postinfection were analyzed for HBsAg. (B) HepaRG cells were infected with HBV in the presence of suramin, dextran sulfate, dextran, and hyaluronic acid. HBsAg secreted between day 7 to 11 postinfection was determined. (C) Quantification of HBV binding (4 hours at 37°C) to HepaRG cells in presence of heparin (63 μg/mL and 630 μg/mL) or suramin (100 μg/mL). Cell-associated genome equivalents were quantified by real-time PCR.

Sulfation of Cellular Glycosaminoglycans Is a Prerequisite for HBV Infection.

To address the role of sulfation for infection inhibition, differentially sulfated heparin variants were analyzed. Completely de-N-sulfated heparin did not interfere with infection (Fig. 2A). In contrast, partially de-N-sulfated heparin exhibited an inhibition potential of 70% at the highest concentration tested (200 μg/mL), indicating that the degree of sulfation within the carbohydrate structure is important.

To investigate the contribution of the carbohydrate backbone on the infection inhibition potential, we performed neutralization experiments using the synthetic polyanions dextran sulfate (composed of glucose units with 2.3 sulfate groups/glucosyl residue) and suramin, a derivative of urea (6.0 sulfate groups/molecule), which has been reported to block duck hepatitis B virus (DHBV) and hepatitis delta virus infection.24 (Fig. 2B).

Although suramin (concentration that inhibits 50% of approximately 33 μg/mL) and dextran sulfate (concentration that inhibits 50% of approximately 9 μg/mL) neutralized HBV infection, dextran but also hyaluronic acid, an acidic, non-sulfated GAG, were inactive, even at 300 μg/mL.

To determine whether the inhibitory effect of suramin is attributable to an interference with HBV attachment to cells, we quantified cell association of virions in the presence of suramin in a concentration known to block infection (100 μg/mL). As a comparison, we analyzed the cellular binding of HBV in presence of 2 different heparin concentrations. For suramin, a 2.2-fold reduction in binding could be observed. At the highest concentration (630 μg/mL), heparin showed a 7.7-fold decreased binding in contrast to the control (Fig. 2C).

To provide additional evidence for the requirement of sulfated cell surface molecules for HBV infection, we inhibited the cellular adenosine triphosphate–sulfurylase by addition of sodium chlorate to the culture medium.25 As depicted in Fig. 3A chlorate concentrations greater than 50 mM reduced HBV infection of HepaRG cells to background level, indicating the requirement of sulfated cell surface structures. To test whether chlorate-mediated inhibition of HBV infection correlates with the reduction in virus binding, we quantified the ability of HepaRG cells to bind HBV after preincubation with increasing concentrations of sodium chlorate. As depicted in Fig. 3B, HepaRG cells devoid of sulfated GAGs were significantly impaired to bind HBV in comparison with the control (36% at 80 mM chlorate). To rule out side effects, HepaRG cells were cultured for 2 days in the presence of sodium chlorate (80 mM) and subsequently infected in absence or presence of the reagent. In both cases, HBsAg secretion was reduced to background levels. However, addition of the inhibitor 24-48 hours after inoculation with the virus had no effect (Supplementary Fig. 3).

Figure 3.

Sulfation of cellular GAGs is required for the interaction of HBV with hepatocytes. The sulfation of cellular proteoglycans was inhibited by the pretreatment of HepaRG cells with increasing concentrations of sodium chlorate for 48 hours before infection (A) or binding (4 hours at 4°C). (B) Infection efficacy was quantified by measurement of HBsAg secreted between days 7 and 11 postinfection. HBV binding to HepaRG cells was determined by real-time PCR and presented as cell-associated genome equivalents per 106 cells.

Neutralization of Negative Charges on HepaRG Cells by Polycations Abrogates HBV Infection.

To substantiate the importance of cellular polyanionic binding sites for HBV, we preincubated HepaRG cells with the heparin antagonists poly-L-lysine and protamine as well as polybrene and infected the cells in presence of the substances. As depicted in Fig. 4A, poly-L-lysine and polybrene strongly decreased HBV infection when applied at 2 or 1.5 μg/mL, respectively; protamine displayed a reduction to 19% at 200 μg/mL.

Figure 4.

The obstruction of negatively charged cellular interaction sites inhibits HBV infection. (A) HepaRG cells were incubated for 1 hour at 37°C with increasing concentrations of poly-L-lysine, polybrene, or protamine. Subsequently infection was performed in the presence of the substances. (B) HepaRG cells were preincubated with either polyanionic (heparin, suramin, and dextran sulfate) or polycationic (poly-L-lysine, polybrene, and protamine) compounds for 1 hour at 37°C before infection. Cells were washed and infected in absence of the substances. Secreted HBsAg of days 7 through 11 postinfection is presented as percentage of the control.

Preincubation of HepaRG cells with heparin, suramin, and dextran sulfate at concentrations that inhibit HBV entry when present during virus inoculation did not impair infection, whereas preincubation with poly-L-lysine, protamine, and polybrene did (Fig. 4B). Unexpectedly, the sulfated polymers heparin and dextran sulfate led to an approximately 1.6-fold and 2.3-fold increased infection. This might be explained by the creation of artificial binding sites through accumulation of soluble carbohydrates on the cell surface.

Enzymatic Removal of GAGs by Heparinases Reduces HBV Binding to and Infection of HepaRG Cells.

To specify the particular type of GAG responsible for HBV binding to hepatocytes, we took advantage of the different specificities of heparinase I and III. Although heparinase I cleaves heparin and heparan sulfate, heparinase III exclusively addresses the latter. As shown in Fig. 5A, heparinase I–treated HepaRG cells showed reduced HBV binding when compared with the control (24.5% at 71 mIU/mL and 12.6% at 143 mIU/mL). Similarly, heparinase III digestion decreased viral attachment by 68% at 38 mIU/mL, indicating that hepatocyte cell surface–exposed heparan sulfate is the key binding element for HBV.

Figure 5.

Enzymatic removal of cell-surface GAGs reduces HBV binding and infection. HepaRG cells were incubated for 1 hour at 37°C with the indicated concentrations of the GAG lyases heparinase I and III. The cells were washed, and binding (A) or infection (B) was performed in absence of the enzymes. The amount of cell-associated HBV was quantified via real-time PCR; infection was analyzed by secreted HBsAg from day 7 to 11 postinfection.

To prove the relevance of the heparinase I/III-mediated reduction of HBV binding for infection, HepaRG cells were preincubated with 4, 42, and 167 mIU/mL (heparinase I) or 21 mIU/mL (heparinase III), followed by inoculation with HBV for 16 hours in absence of the enzymes (Fig. 5B). Heparinase I decreased infection in a dose-dependent manner, with maximal reduction to 40% of the control. Heparinase III reduced infection approximately 2-fold. A more pronounced effect of the enzymes could be observed when the heparinases were present during virus inoculation (Supplementary Fig. 4). The slow kinetics of cellular HBV binding and uptake requires a prolonged incubation time for efficient infections. The inhibitory effect of heparinase I/III may therefore be partially counteracted by the re-emergence of newly synthesized heparan sulfate proteoglycans (HSPGs) on the surface of HepaRG cells.

Polyethylene Glycol (PEG) Enhances the Binding of HBV to Cell-Surface GAGs.

PEG has been shown to enhance HBV infection in PHH26 and HepaRG cells.4 To investigate whether the effect of PEG on HBV binding is mediated by cell surface-exposed GAGs, we used the CHO-K1 cell line and its GAG-deficient derivative CHO-pgsA745. CHO-pgsA745 cells lack xylosyltransferase and do not express heparan and chondroitin sulfates.23 In comparison, we quantified binding of HBV to differentiated and nondifferentiated HepaRG cells in the presence or absence of 4% PEG.

Quantification of HepaRG cell-associated HB virions in the presence of PEG showed that binding is independent of the cell differentiation state. In fact, nondifferentiated HepaRG cells bound HBV with slightly higher (10.4 ge/cell in comparison with 5.1 ge/cell) efficacy (Fig. 6A, right). Taken into account that 116 HBV ge/cell were applied but only 4.4% (differentiated cells) or 9.0% (nondifferentiated cells) were cell associated after 4 hours, we concluded that only a limited number of viral binding sites are available on the cell surface. Consistent with earlier studies, binding was reduced (13.7-fold in differentiated versus 16.2-fold in nondifferentiated cells) in the absence of PEG (Fig. 6A, left). A similar degree of HBV cell association including a PEG-mediated enhancement of binding (14.7-fold) was also observed for the nonsusceptible CHO-K1 cell line (Fig. 6B). Surprisingly, HBV binding to CHO-pgsA745 cells was not enhanced by PEG. The observation that HBV binding can only be reduced by heparin in the presence of PEG (Fig. 6B right) indicates that PEG promotes GAG-mediated virus–cell contacts.

Figure 6.

GAG-mediated cell-association of HBV is enhanced by PEG. (A) Binding of HBV to nondifferentiated and differentiated HepaRG cells was performed either in absence (left) or presence (right) of 4% PEG. Binding was quantified by a real-time PCR and is presented as cell-associated genome equivalents per well. (B) Binding of HBV (4 hours at 37°C) to CHO-K1 (black bars) and the GAG-deficient CHO-pgsA745 (gray bars) cells in the absence (left) and presence of PEG (middle and right) and heparin (right). HBV genome equivalents were quantified by real-time PCR and are given as absolute values and as ratios in the table.

L-Protein Rich HBV Particles Selectively Bind to Heparin.

To biochemically characterize the interaction of HBV particles to GAGs, we applied HepG2.2.15 cell-derived viral and subviral HBV particles to a heparin affinity chromatography column and determined their elution behavior. Most DNA-containing particles bound to heparin-sepharose under physiological salt conditions (140 mM NaCl) and eluted at NaCl concentrations of at least 240 mM (Fig. 7A left). To prove that these fractions contain virions, we (1) performed a western blot detecting L-protein (Fig. 7A, middle) which is mostly absent in 22 nm spherical subviral particles but enriched in virions and filaments5 and (2) simultaneously used this fraction in comparison with the input to infect HepaRG cells (Fig. 7A, right). The presence of L-protein and HBV-DNA in the 340 mM NaCl fraction and the unabated infectivity indicates binding of infectious virions to heparin. To investigate the contribution of the preS-part of the L-protein to heparin binding, we took advantage of previous observations demonstrating the feasibility to enzymatically remove preS1 and preS2 from L-protein by trypsin27, 28 without destroying the S-protein. To that aim, the 340 mM elution fraction of a heparin-affinity chromatography [Fig. 7B, top (underlined)] was trypsinized (5 hours at 37°C). As a control, the same fraction was incubated with phosphate-buffered saline. Although re-application of the untreated particles did not alter the ability to again bind to heparin and elute at 340 mM NaCl, HBV-DNA (Fig. 7B, bottom) and S-protein (Fig 7B, middle) were exclusively found in the flow-through fraction after trypsin treatment. The unchanged immunological traceability of S-protein in the flow-through fraction indicates that trypsin did not destroy the protein. However, and in accordance with the described sensitivity of the preS-domain against trypsin, only residual amounts of L-protein were detectable after protease treatment when using a preS-specific polyclonal antiserum (Fig. 7B, middle). This suggests a contributory role of the preS-domain of the HBV L-protein for GAG binding.

Figure 7.

Selective interaction of infectious, L-protein–enriched HBV particles with heparin-sepharose. (A) A 100-fold concentrate of a HepG2.2.15 cell supernatant, containing virions and subviral particles, but no naked cores, was applied to a heparin-sepharose affinity column. The input, flow through, and elution fractions (240, 340, 540, 1,000, and 2,000 mM NaCl) were analyzed for their HBV DNA content by dot-blot (left). The 340 mM elution fraction was analyzed for its L-protein content by a preS1-specific western blot using the monoclonal antibody Ma18/7 (middle). The infectivity of the 340 mM NaCl elution fraction on HepaRG cells in comparison with the respective input fraction was measured (right). The ratio of virions evident from the input and the 340 mM NaCl elution fraction of the heparin-affinity chromatography (left) is the same as the ratio between the control and the 340 mM NaCl elution fraction used to inoculate the cells for the infection of HepaRG cells (right). The infection was analyzed by quantification of HBsAg and hepatitis B e antigen secretion between days 7 and 11 postinfection. (B) A heparin-affinity chromatography was performed as described in (A). The fractions were analyzed for their L-protein content by an immune dot blot using the preS-specific polyclonal antiserum H863 (top). Trypsin digestion of the 340 mM elution fraction (underlined) was performed for 5 hours at 37°C, and the products were reapplied to the heparin column. The fractions were analyzed by immune dot blots detecting either the L- (preS) or the S-protein (middle). As a control, an undigested sample of the 340 mM fraction was reapplied to the column. Note that trypsin digestion results in an almost complete disappearance of preS-specific signals but does not diminish the S-signal, indicating that the integrity of S-protein remains unchanged. The fractions of the trypsin-treated sample were analyzed for their DNA content by a quantitative real-time PCR and are presented as genome equivalents/milliliter. The amount of the DNA in the fractions is additionally given as percentage of the input.


Glycoconjugates are cell-surface structures with important physiologic functions, such as cell–cell and cell–matrix interactions, adhesion, signaling, differentiation, and development.29 It is therefore not surprising that pathogens, including viruses selected these molecules as attachment receptors. We provide evidence that HBV infection requires the initial interaction with hepatocyte surface HSPGs and thus, for the first time, identify cellular molecules that serve as a primary attachment receptor for HBV.

Evidence for this conclusion is based on our observations that HBV binding to and infection of HepaRG cells is sensitive to (1) inhibition with distinct soluble GAGs, (2) the prevention of GAG sulfation by chlorate, and (3) the enzymatic removal of GAGs from the cell surface. Additionally, although not susceptible, cells with a defect in the biosynthesis of heparan and chondroitin sulfates exhibited a reduced ability to bind HBV. The fact that heparin, as a soluble GAG variant, and heparinase III, as a heparan sulfate–specific lyase, interfere with both HBV attachment and productive infection suggests that HSPGs act as initial HBV binding sites on the hepatocyte surface. This conclusion is also supported by a recent study by Leistner et al., using primary Tupaia hepatocytes.30 Consistent with our results, Ying et al.31 previously demonstrated an almost complete abrogation of HBV-binding to PHH and the nonsusceptible HepG2 cell line in the presence of heparin and dextran sulfate. However, they did not find a change in binding of HBV to HepG2 cells after heparinase treatment. This difference might be explained by the absence of PEG as a promoter of GAG-binding.

Our data indicate that heparin and other highly sulfated polyanions bind to positively charged structures on the surface of virions, resulting in inhibition of cellular binding and abrogation of infection. Although infection is reduced more than 100-fold, the decrease in virus binding is less pronounced (up to 8-fold). However, considering the low number of available cellular binding sites (5-30 ge/cell) this might be attributable to superimposed unspecific binding events that limit the detection range of our binding assay.

HSPGs are expressed by many cells. However, they differ in the types and the extent of proteoglycans they express and show variations in the heparan sulfate fine structure, such as modifications in chain length, the degree of sulfation, and the positions of the sulfate groups. The composition of cell-surface proteoglycans can change during differentiation or on activation of cells. Remarkably, liver heparan sulfates display some unique features, such as a low sulfation density in close proximity to the core protein and enriched N-sulfated glucosamine residues and 2-O-sulphated uronic acids at their distal ends.20, 21 This is consistent with our findings that highly sulfated GAGs (heparin, heparan sulfate) and GAG variants (dextran sulfate) were the most effective substances to inhibit binding and infection whereas decreased sulfation led to a progressive loss of inhibition. Thus, the degree of sulfation plays the key role in that interaction process whereas the nature of the polysaccharide backbone is less important.

In accordance with the presence of proteoglycans in many cell types, we and others3 found that binding of HBV is not restricted to susceptible HepaRG cells or PHH. Thus, at the level of attachment, the observed host range and the species specificity of HBV can only be explained within the limits of the foresaid preference of HBV for highly sulfated liver HSPGs. Productive infection, however, must be triggered by additional, more specific processes. One of these steps might be related to the inhibitory activity of acylated peptides comprising the N-terminal 47 amino acids of the preS1 domain.10, 12 Interestingly, these lipopeptides neither bind heparin nor interfere with binding of HBV to HepaRG cells at concentrations known to block infection (unpublished data). This indicates that they function in a postattachment step and emphasizes the assumption that HBV enters hepatocytes by a multistep process.

Previous reports4, 26 indicated that addition of PEG to PHH or HepaRG cells enhances HBV infection without inducing artificial fusion of the viral and the cellular membranes. This enhancement was related to an increased cell association of HBV DNA. However, the nature of the molecules involved in the interaction was unclear. We provide evidence that PEG promotes virus binding by increasing the GAG-dependent attachment of HBV to HepaRG cells by approximately the same factor as it increases infection. This conclusion is based on our observation that enhancement of binding in the presence of PEG was detected in all cells, with the exception of the GAG-deficient CHO-pgsA745. This result also excludes a possible unspecific PEG-mediated precipitation of virions on the cell surface and opens the question of whether PEG can also enhance infection of other pathogens that use GAGs as attachment receptors.

Initial reports on the role of suramin suggested that it inhibits the DHBV DNA polymerase activity.32 However, subsequent reports showed that DHBV replication was only inhibited when suramin was present during virus inoculation and not at later times.33 Suramin also interferes with hepatitis delta virus infection, an RNA virusoid that replicates independently of the reverse transcriptase.24 Using a semiquantitative PCR-based binding assay, Funk et al.34 demonstrated that suramin reduced the binding of DNA-containing DHBV particles to embryonic duck hepatocytes. We here extend this finding to HBV and propose a mechanism of action of suramin, namely, mimicking highly sulfated cellular structures that are required for attachment.

Although we and others35 present biochemical evidence that heparin-sepharose binds HB virions, we currently do not know the exact binding site. Our data strongly suggest that it is located in the preS domain of the L-protein because (1) L-protein–enriched virions display selectivity toward heparin. (2) Removal of the preS domain abrogates the virus–heparin interaction. Because of the functional mapping of sequence requirements for HBV and hepatitis delta virus infectivity,8, 36–38 the first 75 amino acids of the preS1 domain likely play a crucial role in this process.

Taken together, we propose a multistep entry pathway of HBV. It initializes through an L-protein–dependent binding to cell-associated HSPGs and is followed by a highly specific step involving a myristoylated N-terminal domain of preS1.


We thank Stephanie Held for technical assistance, Isabel Vogler, Silke Schmidt, and Kerry Mills for help with the establishment of the HBV binding assay, Caroline Gähler for the preparation of virus stocks, and Stefan Seitz for many helpful discussions. We also thank Jeffrey D. Esko for providing the CHO-pgsA745 cell line. We are indebted to Ralf Bartenschlager, who continuously supports our work.