Role of glycosaminoglycans for binding and infection of hepatitis B virus

Authors


*E-mail dieter.glebe@viro.med.uni-giessen.de; Tel. (+49) 641 99 41203; Fax (+49) 641 99 41209.

Summary

Many parts of the life cycle of hepatitis B virus (HBV) infection of hepatocytes have been unravelled, but the attachment and entry process leading to infection is largely unknown. Using primary Tupaia hepatocyte cultures as an in vitro infection system, we determined that HBV uses cell-surface heparan sulfate proteoglycans as low-affinity receptor, because HBV infection was inhibited by heparin (IC50: 5 μg ml−1) or other higher-sulfated polymers, but not by lower-sulfated glycosaminoglycans, such as chondroitin sulfate. Pretreatment of primary hepatocytes with heparinase decreased viral binding and inhibited HBV infection completely. Interestingly, after preS1-dependent viral binding at 16°C to the cell surface, subsequent infection could still be inhibited by HBV preS1-lipopeptides, but not by heparin any more, suggesting a shift of the virus to a high-affinity receptor. In summary, we suggest following multistep attachment process: in vivo, HBV is initially trapped within the liver in the space of Dissé by heparan sulfate proteoglycans. Thereafter, HBV binds via its preS1 attachment site and the N-terminal myristic acid to a yet unknown, high-affinity receptor that confers uptake in a yet unknown compartment.

Introduction

Hepatitis B virus (HBV) belongs to the family of Orthohepadnaviruses that causes acute and chronic infections of the liver. With over 370 million people chronically infected, chronic HBV infections are a major cause of liver cirrhosis and hepatocellular carcinoma in many regions of the world (Baumert et al., 2007). Although our knowledge of the molecular biology of this virus has increased over the past years, resulting in an effective vaccine and new therapeutics for chronic HBV carriers (Tillmann, 2007), the viral and cellular determinants of viral binding and entry of this virus are still enigmatic (Glebe and Urban, 2007). This was mainly due to the lack of a suitable and practicable infection system for HBV in the past. In vivo, HBV infects only humans and some primates (e.g. chimpanzees) and till now, no easy small animal system is available for the study of HBV infection. For in vitro studies, primary human hepatocyte cultures (PHHs), obtained after perfusion of liver pieces after surgical resection, were for many years the only system to study the early steps of this highly liver-specific infection (Glebe, 2006). However, these cultures are, for obvious reasons, very limited in availability. Furthermore, PHHs are not easy to handle and were reported to differ in susceptibility due to the very heterogeneous quality of different liver cell preparations (Gripon et al., 1988). These limitations were partly overcome by the observation that primary hepatocyte cultures from Asia tree shrews, Tupaia belangeri (primary Tupaia hepatocytes; PTH), are susceptible to HBV (Walter et al., 1996; Köck et al., 2001). Interestingly, this small mammal does not belong to the order primates (Nishihara et al., 2002) but forms an own order called Scandentia. Nevertheless, we (Glebe et al., 2003; 2005) and others (Köck et al., 2001) were able to show that HBV infection of PTH is homologous to infection of PHH and a newly established HBV-susceptible cell line (Gripon et al., 2002). HBV virions appear after negative staining in electron microscopy as spheres of 45 nm in diameter. Besides the virions, HBV-infected hepatocytes constitutively secrete also nucleocapsid-free subviral particles (SVPs), composed of the hepadnaviral surface proteins. HBV-SVPs exist in a spherical (22 nm in diameter) or filamentous (same diameter, but variable length) form. The three co-carboxyterminal HBV surface (HBs) proteins (L-, M-, S-HBs) are distinguished by three domains: preS1 only in LHBs, preS2 in LHBs and MHBs, and S in all three HBs proteins. The SHBs is the major component of the virion envelope and the SVPs; however, virions and filamentous SVPs contain more LHBs than spheres (Heermann et al., 1984). Till now, a still-growing list of potential binding partners for all three HBV surface proteins in human serum and on cellular membranes has been proposed; however, none have even been proven to act as a functional HBV receptor facilitating HBV infection (Glebe and Urban, 2007). Recently, one group described binding and hence purification of HBV and hepatitis C virus (HCV) from human plasma samples by heparin columns (Zahn and Allain, 2005). Therefore, the question arises, whether HBV uses sulfated glycosaminoglycans (GAGs) for attachment to hepatocytes. GAGs are long, unbranched polysaccharides, consisting of repeating disaccharides, usually an amino sugar and an uronic acid. While N-acetylglucosamine and glucuronic acid form heparan sulfate (HS), the backbone of chondroitin sulfate (CS) consists of N-acetylgalactosamine and glucuronic acid (Esko and Selleck, 2002). GAG chain synthesis starts with the addition of an amino sugar to a tetrasaccharide that is attached via O-glycosylation to a serine residue, followed by a glycine. Mature GAG chains undergo distinct modifications, like N- and O-sulfation and epimerizations, leading to the great variety of GAGs. Especially HS chains are found mainly on membrane proteoglycans, like glypicans and syndecans. Glypicans are heparan sulfate proteoglycans (HSPGs) and linked via a glycosylphosphatidylinositol (GPI) anchor to the cell membrane. Syndecans contain also mainly HS chains, but are type I transmembrane proteins (Hacker et al., 2005). Heparin is known as a highly sulfated soluble subtype of HS. In vivo, HS is an extracellular GAG and attached to a core protein, while heparin is an intracellular GAG, synthesized in granulated cells and is cleaved from its serglycin core protein (Horner, 1986; Kresse et al., 1993). Although HS are found on virtually every cell, highly sulfated liver-specific HS provide binding sites for diverse ligands and receptors involved in regulating growth control, lipid metabolism, haemostasis, signal transduction and cell adhesion (Esko and Lindahl, 2001; Kreuger et al., 2006). Besides its physiological role, e.g. mediating binding of apolipoprotein E to hepatocytes (Dong et al., 2001), liver-HS provides binding sites for various liver-targeting pathogens, like malaria circumsporozoite (Rathore et al., 2001), dengue virus (Chen et al., 1997) and HCV (Barth et al., 2006; Koutsoudakis et al., 2007). In this study, we investigated the role of GAGs during the attachment and infection of HBV to susceptible primary hepatocytes.

Results

Binding of HBV to heparin

Recently, one group reported an interaction of HBV with heparin columns (Zahn and Allain, 2005). Heparin is a complex polysaccharide consisting of repeated disaccharides of uronic acid and glucosamine. The disaccharide chains are modified by N- and O-sulfation that contribute to a specific pattern of negative charge. To determine specificity of binding of HBV surface proteins to heparin, we coated 96-well plates with heparin and determined binding of purified HBV particles (Fig. 1). Using increasing concentrations of different sulfated polymers as competitor, we determined that the highly sulfated heparin could inhibit binding to heparin-coated plates completely at 10 μg ml−1. CS that has a lower sulfation grade, resulted in a dose-dependent reduction of binding, but no complete inhibition, even at high concentrations (1000 μg ml−1). To investigate the role of sulfation of heparin for the binding, we used also specifically de-N- and de-O-sulfated heparins for competition of binding to heparin plates. In de-N-sulfated heparin, N-sulfated glucosamine residues of heparin are removed, while in de-O-sulfated heparin, O-sulfate esters of heparin are removed. Interestingly, complete inhibition of HBV binding to heparin could neither be achieved with de-O-sulfated heparin, nor de-N-sulfated heparin. However, use of de-O-sulfated heparin resulted in 50% reduction of HBV binding, if used at very high concentrations (1000 μg ml−1). This suggests that the overall negative charge, given by the degree of sulfation of the polyanions, is important for viral binding and not primarily a specific pattern of sulfation.

Figure 1.

Binding of HBV to heparin-coated wells. Highly purified HBV SVPs (100 ng ml−1) from chronic HBV carriers were incubated with indicated concentrations of polyanions for 1 h at 16°C and added to heparin-coated (25 μg ml−1) 96-well plates. Plates were incubated for 2 h at 37°C, and heparin-bound SVPs were detected using an conformational-dependent mAb to SHBs in a colorimetric reaction as described in Experimental procedures. Ø, no polyanions. The dotted line indicates the cut-off.

Binding and infectivity of HBV to primary hepatocytes and different cell lines

In order to measure binding and infection of HBV to cells from different origins, we incubated purified HBV or SVPs with freshly isolated PTH, primary rat hepatocytes (PRH) and established human cell lines (HepG2 and Hela). To differentiate between binding and uptake, we incubated the virus with the cells for 1 h at 16°C (Fig. 2). Incubation at this temperature inhibited cellular endosomal uptake as detected by the absence of bulk-flow endocytosis below 18°C by a soluble peroxidase assay (Shurety et al., 1998) (Fig. 2A, picture h). Nevertheless, this temperature allowed binding of HBV, detected as a ring-like staining of HBV particles in immunocytochemistry (Fig. 2A, picture c) in contrast to the staining at 37°C, suggesting uptake of HBsAg (Fig. 2A, picture a). We observed a specific staining for HBsAg not only to susceptible PTH (Fig. 2A, picture a) but also to HepG2 and Hela cells that are non-susceptible for HBV infection (Fig. 2A, pictures d and f). To evaluate the amount of binding of the purified virus to the different cells, we measured viral binding by real-time polymerase chain reaction (PCR) (Fig. 2B). Interestingly, only about one-tenth of the viral inoculum was bound by susceptible PTH irrespective of the concentration ranging from 107 to 103 genome-equivalents ml−1. However, non-susceptible HepG2 and Hela cells bound purified HBV with the same magnitude like susceptible PTH. Next, we performed binding and infection experiments using freshly isolated PRH and PTH (Fig. 3). Surprisingly, both rat and Tupaia hepatocytes bound approximately 10% of the viral inoculum, ranging from 105 to 107 genomes ml−1 (Fig. 3A), while only PTH got infected in a dose-dependent manner as detected by secretion of HBeAg 9–12 days after infection (Fig. 3B). Interestingly, when we added highly sulfated polyanions during infection, e.g. dextran sulfate (100 μg ml−1), infection wascompletely blocked (Fig. 3B), while binding was only decreased by a factor of 3. Pretreatment of hepatocytes with dextran sulfate and subsequent removal before addition of virus had no effect on binding or infection of HBV (data not shown). As a control for specificity of HBV infection of PTH, we added myristoylated HBV preS1 peptide of amino acids 2–48 that has been shown to inhibit HBV infection of PHHs and PTH in nanomolar concentration (Glebe et al., 2005; Gripon et al., 2005). However, when applied at 100 nm, the peptide was unable to inhibit binding of virus to PTH and PRH, although infection of PTH could be completely inhibited (Fig. 3B). These data indicate that at least in vitro, HBV binds to susceptible and non-susceptible cells within the same magnitude (one-tenth of a given HBV inoculum). The observation that highly sulfated polyanions are able to inhibit HBV infection completely, but could not inhibit HBV binding to PTH and PRH, suggests that interaction of HBV with negatively charged cell-surface structures is not species-specific. Moreover, the inability of infection-interfering HBV myr-preS1 lipopeptides to inhibit HBV binding at nanomolar concentrations suggests a post-entry role of this kind of inhibitor.

Figure 2.

Binding of HBV to PTH and different cell lines.
A. Immune staining of Tupaia hepatocytes after incubation with purified HBV SVPs from human plasma (2 μg ml−1 each) for 1 h at 37°C with a monoclonal antibody (mAb) against the S-domain (red) and counterstained with Meyer's hemalaun (blue). Binding of plasma-derived particles is detected as seen for PTH (a), HepG2 cells (d) and Hela cells (f) after incubation for 1 h at 37°C, while (c) indicates incubation of Tupaia hepatocytes with purified HBV SVPs for 1 h at 16°C. Control slides without HBsAg incubation: PTH (b), HepG2 (e) and Hela cells (g). All panels have same magnification (bar = 50 μm). The inhibition of cellular uptake below 18°C was measured by the absence of bulk-flow endocytosis of soluble peroxidase (h) as described in Experimental procedures. The dotted line indicates the cut-off.
B. Cellular binding assay for HBV. Highly purified HBV from chronic carriers in the indicated concentrations was incubated with different confluent cell cultures in 12-well plates for 4 h at 16°C. Bound HBV particles were determined by real-time PCR as described in Experimental procedures (log copies ml−1).

Figure 3.

Binding and infection of HBV to primary hepatocytes from Tupaia and rat.
A. Highly purified HBV from chronic carriers in the indicated concentrations was incubated with different cell cultures for 4 h at 16°C without competitor or after pre-incubation of virus with dextran sulfate (100 μg ml−1) or myristoylated HBV preS1 peptides 2–48 (100 nM) for 1 h at 16°C. Bound particles were determined by real-time PCR as described in Experimental procedures.
B. Highly purified HBV from chronic carriers in the indicated concentrations was incubated with different cell cultures for 4 h at 16°C without competitor or after pre-incubation of virus with dextran sulfate (100 μg ml−1) or myristoylated HBV preS1 peptides 2–48 (100 nM) for 1 h at 16°C. After extensive washing at 4°C, cells were shifted to 37°C for 12 h to allow HBV uptake. Medium was changed every 3 days. HBeAg production of infected cultures was determined from day 9 to day 12 post infection. The dotted line indicates the cut-off (multiples of cut-off signal).

Infection of PTH in the presence of polyanions

To determine the role of various polyanions for HBV infection, we pre-incubated different concentrations of polyanions, heparin, CS, dextran sulfate and the uncharged dextran with purified virus from HBV carriers before addition to susceptible PTH (Fig. 4A). After binding at 16°C, cells were washed extensively, and internalization of bound viral particles was achieved by shifting the temperature to 37°C for 12 h. Outcome of infection was determined by measuring secretion of newly synthesized HBsAg of infected PTH 9–12 days after infection as described. It turned out that heparin was a much more efficient inhibitor of HBV infection (IC50: approximately 5 μg ml−1) than CS (IC50: approximately 800 μg ml−1) (Fig. 4A). Dextran sulfate could inhibit HBV infection in nearly the same magnitude as heparin (IC50: < 25 μg ml−1), while pre-incubation of purified virus with the uncharged dextran was unable to inhibit HBV infection. As a control, myristoylated preS1 peptides aa 2–48 containing the essential binding domain (aa 9–18, genotype D) inhibited HBV infection, while a myristoylated preS1 peptide lacking this domain (myr-preS1 aa 19–48) failed to inhibit infection as described (Glebe et al., 2005).

Figure 4.

Influence of different polyanions on HBV infection of PTH during or after viral binding.
A. Highly purified HBV (100 genomes per hepatocyte) from chronic carriers were incubated with PTH for 4 h at 16°C after pre-incubation of virus with different concentrations of indicated polyanions for 1 h at 16°C. Incubation with 100 nM infection-interfering myristoylated HBV preS1 peptides 2–48, or inactive peptide (preS1 domains 19–48) for 1 h at 16°C served as a control. After extensive washing at 4°C, cells were shifted to 37°C for 12 h to allow HBV uptake. Medium was changed every 3 days. HBsAg production of infected cultures was determined from day 9 to day 12 post infection and is displayed in percentage of the uncompeted control infection (Ø). The dotted line indicates the cut-off.
B. Highly purified HBV from chronic carriers were incubated with PTH (100 genomes per hepatocyte) for 4 h at 16°C. After extensive washing at 4°C, cells were incubated with different concentrations of indicated polyanions and HBV peptides (100 nM) for 1 h at 16°C. After further washing steps at 4°C, cells were shifted to 37°C for 12 h to allow HBV uptake. Medium was changed every 3 days. HBsAg production of infected cultures was determined from day 9 to day 12 post infection and is displayed in percentage of the uncompeted control infection (Ø). The dotted line indicates the cut-off.

Treatment of cells with different enzymes that cleave cell-surface bound glycans

These observations suggested that charged HSPGs might be a primary binding factor for HBV on PTH. To prove this hypothesis, we pretreated PTH with heparinase, known to efficiently cleave sulfated HS chains from the cell surface. The enzyme was removed by extensive washing, and purified HBV was added at 16°C for 4 h. During these conditions, appearance of newly synthesized, or recycled re-sulfated cellular HS was unlikely, because the estimated half-life of HS proteoglycanes at the plasma membrane of cultured primary hepatocytes was reported to be 3–6 h at 37°C (Oldberg et al., 1977; Egeberg et al., 2001). Afterwards, the cultures were washed again at 4°C and shifted to 37°C to allow uptake of bound virus. As shown in Fig. 5A, treatment with heparinase reduced HBV infection in a dose-dependent manner and resulted in complete loss of susceptibility at 6 U ml−1. This result confirms the importance of HSPGs for the infection process of HBV. To rule out the possibility that other glycans than HS might contribute to infection, we also pretreated the cells either with peptido N-glycanase F (PNGase F), resulting in cleavage of N-linked sugars, or with sialidase, known to cleave negatively charged sialic acid from N- and O-glycans. However, none of these treatments had any effect on HBV infection (Fig. 5A). The removal of the distinct sugars from cell-surface proteins was tested by shifting of the heavily N-glycosylated and sialylated asialoglycoprotein receptor (ASGPR) in SDS-PAGE immunoblots from plasma-membrane preparations of PTH (data not shown). Because the ASGPR is a liver-specific lectin, it had been speculated that it might serve as a specific receptor for HBV, especially during treatment of hepatocytes with sialidase (Owada et al., 2006). However, neither pre-incubation with sialidase, nor incubation in the presence of sialidase, resulted in significant increase or decrease of infection (Fig. 5A). Furthermore, no effect of sialidase treatment on HBV infection was detected when non-susceptible rat hepatocytes or HepG2 cells were used (data not shown). To determine the effect of heparinase treatment of PTH on HBV binding, we treated PTH at 16°C with heparinase and determined binding of purified HBV at 16°C for 4 h. As shown before (Fig. 3A), approximately 90% of the viral input did not bind to HBV and remained within the supernatant of the cells (Fig. 5B). However, even heparinase treatment of PTH that resulted in complete inhibition of infection, caused only a threefold decreased viral binding.

Figure 5.

Influence of different cellular glycans on HBV binding and infection of PTH.
A. Primary hepatocytes were incubated with indicated concentrations of heparinase III, PNGase F, or sialidase or myristoylated preS1 peptides (100 nM) for 4 h at 16°C. After washing at 4°C, highly purified HBV from chronic carriers were incubated for 4 h at 16°C in the absence of enzymes, while in the case of sialidase, enzyme was also present in one panel during the whole infection period. After extensive washing at 4°C, cells were shifted to 37°C for 12 h to allow HBV uptake. Medium was changed every 3 days. HBsAg production of infected cultures was determined from day 9 to day 12 post infection and is displayed in percentage of the uncompeted control infection (Ø). The dotted line indicates the cut-off.
B. Primary hepatocytes were incubated with indicated concentrations of heparinase, for 4 h at 16°C. After washing at 4°C, highly purified HBV from chronic carriers were incubated (100 genomes per hepatocyte) for 4 h at 16°C in the absence of enzymes. After extensive washing at 4°C, bound particles were determined by real-time PCR as described in Experimental procedures. Ø; no enzyme.

Role of sulfation of hepatocellular proteoglycans for HBV infection

Heparan sulfate proteoglycans are heavily charged due to the addition of sulfate groups to their sugar backbones. Cultivation of cells in sulfate-free medium and addition of sodium chlorate inhibits effective sulfation of CSPGs and HSPGs in a dose-dependent manner. It has been reported that low concentrations of chlorate (3–5 mM) reduced mainly sulfation of CSPG, while higher chlorate concentrations (10 mM) are needed for inhibition of HSPG sulfation (Fjeldstad et al., 2002). Furthermore, inhibition of sulfation of HSPGs was shown to significantly decrease infection with vaccinia virus (Chung et al., 1998) and inhibits adeno-associated virus 2 infection (Qiu et al., 2000) in vitro. Therefore, we pre-incubated PTH 48 h before infection with increasing concentrations of sodium chlorate in sulfate-free medium, and added purified virus in either the presence or absence of the drug (Fig. 6). Pretreatment of PTH with sodium chlorate resulted in a decrease of susceptibility towards HBV. Infection in the presence of sodium chlorate showed slightly stronger inhibitory potential (IC50: approximately 10 mM) than in the absence of the drug (IC50: 25 mM) during infection. Unfortunately, higher concentrations of sodium chlorate were toxic to PTH when applied for 48 h. To rule out any drug-related effects on the virus itself, we infected PTH in the presence of sodium chlorate with indicated concentrations, but omitted pre-incubation of the cell cultures with sodium chlorate. However, this had no significant effect on HBV infection (Fig. 6). These observations indicate that although cleavage of HS on hepatocytes can decrease HBV binding only by a factor of 3, HBV infection is absolutely dependent on the presence of highly sulfated HSPGs on the surface of susceptible primary hepatocytes.

Figure 6.

Influence of sulfation of GAGs on HBV infection of PTH. Primary hepatocytes were incubated with indicated concentrations of sodium chlorate in sulfate-free medium for 48 h at 37°C before (pre-incubation) or for 12 h at 37°C during infection with HBV. Incubation with 100 nM infection-interfering myristoylated HBV preS1 peptides 2–48, or inactive peptide (preS1 domains 19–48) during infection served as a control. Medium was changed every 3 days. HBsAg production of infected cultures was determined from day 9 to day 12 post infection and is displayed in percentage of the uncompeted control infection (Ø). The dotted line indicates the cut-off.

Effect of polyanions after viral binding to PTH

To further elucidate the effect of polyanions after interaction of HBV with hepatocyte membranes, we first let the virus bind to the cells at 16°C for 4 h, washed away the unbound virus at 4°C and thereafter added the same concentration of polyanions as used for pre-incubation (Fig. 4B). After an additional hour at 16°C, the cells were shifted to 37°C for 12 h to allow infection. Interestingly, after viral binding, addition of polyanions had only minor effect on HBV infection (Fig. 4B). This suggests that HBV has switched from low-affinity binding to high-affinity binding on the surface of hepatocytes at 16°C. However, this process might not be completed after 4 h because, with high concentrations of heparin (1000 μg ml−1) and dextran sulfate (100 μg ml−1), a decrease in infectivity by 50% and 20% respectively was observed. Nevertheless, HBV infection could be efficiently inhibited by adding myristoylated HBV preS1 peptides 2–48 in nanomolar concentration, even after HBV binding. This is in line with previous results demonstrating that HBV preS1-lipopeptides could inhibit HBV infection when applied 2, 4 or 12 h before addition of virus (Glebe et al., 2005).

Discussion

Previously, we showed that HBV surface proteins bind to HBV-susceptible PTH (Glebe et al., 2003; 2005). In this study, we analysed HBV binding in more detail. Using highly purified HBV from the plasma of chronically infected patients, only 10% of the viral input bound to susceptible PTH, and this proportion could not be increased by raising the concentration. Binding of duck hepatitis B virus (DHBV) to primary duck hepatocytes is also inefficient and does also not reach saturation (Klingmuller and Schaller, 1993). Interestingly, the same proportion of viral binding was observed with non-susceptible cells, such as PRH cultures, a human hepatoma cell line (HepG2), and even with Hela cells. These data confirm previous observations on non-productive binding of HBV from various sources to different cell lines, but this binding was usually not related to infection (for review, see Glebe and Urban, 2007). Given the finding that HBV binds to heparin columns (Zahn and Allain, 2005), we analysed the interaction of HBV with heparin in more detail. We detected that purified HBV SVPs bound very well to heparin-coated wells, and this specific binding could only be inhibited completely by heparin itself. Use of lower-sulfated forms of heparin or CS resulted in either no inhibition (de-N-sulfated heparin) or dose-dependent reduction of HBV binding (CS and de-O-sulfated heparin). This is in line with previous publications using heparin-affinity chromatography for HBsAg purification (Einarsson et al., 1978; Tajima et al., 1992). Interestingly, binding of HBV to susceptible PTH and non-susceptible PRH was only moderately inhibited by similar concentrations of highly sulfated polymers, such as heparin or dextran sulfate. This is in contrast to the data obtained by Ying et al. (2002), who found nearly complete reduction (97%) of HBV binding to PHH and non-susceptible cell lines in the presence of 100 μg ml−1 heparin. However, in contrast to the present work, those studies presented by Ying et al. were conducted with crude supernatants of stably HBV-producing cell lines (e.g. HepAd38). It is well known that these cell lines secrete, besides HBV virions, a large amount of partially and non-enveloped core particles (Sun and Nassal, 2006). The HBV core protein, however, contains an arginine-rich carboxyterminal domain that has been shown to mediate attachment of nucleocapsids to cell-surface HS (Vanlandschoot et al., 2005; Cooper and Shaul, 2006). On the other side, the viral particles used in our study, derived from human plasma and purified through a sucrose-gradient, contain fully enveloped viral particles, which were not permeable for nucleotide triphosphates (Mest, 1997). Therefore, the effects detected by Ying et al. may be due to contaminations of the inoculum with partially or non-enveloped viral core particles. Furthermore, Ying et al. did not study the effect of polyanions on HBV infection. In our study, a complete inhibition of HBV binding by highly sulfated polyanions could not be achieved by the use of purified plasma-derived HBV and primary hepatocytes; however, these polymers (heparin and dextran sulfate) could inhibit HBV infection of PTH completely when given together with the viral inoculum. To determine the role of sulfation of cell-surface GAGs for HBV infection, we inhibited sulfation of cellular GAGs by sodium chlorate that is known to inhibit heparin-dependent infection of different viruses (Chung et al., 1998; Qiu et al., 2000). Indeed, sulfation of cell-surface GAGs seems to be important for HBV infection, because inhibition of sulfation decreases HBV infection in a dose-dependent manner. In a similar approach, this has been demonstrated for a HBV-susceptible human hepatoma cell line with HBV derived from a recombinant source (Schulze et al., 2007). HCV, a highly liver-specific RNA virus, was reported to use distinct N-sulfated cell-surface HS for primary attachment (Barth et al., 2006), but in our study, neither HBV binding to heparin-coated wells, nor infection of PTH (data not shown), could be inhibited by de-N- or de-O-sulfated heparin. This argues against the requirement of a distinct sulfation pattern of cell-surface GAGs in the case of HBV. The ubiquity of HS on mammalian cells may be one reason why purified HBV binds with the same degree to susceptible and non-susceptible primary hepatocytes, and even non-liver cells. The finding that pretreatment of cellular surface proteins with heparinase inhibits viral infection while PNGase and sialidase had no effect, further supports the involvement of highly sulfated cell-surface HS in the early steps of HBV infection. Interestingly, highly sulfated polymers like heparin or dextran sulfate have no effect on DHBV infection in vivo and in vitro (Offensperger et al., 1991). Although DHBV shares many molecular similarities with HBV, DHBV probably uses other receptors for primary attachment than GAGs (for a recent review, see Glebe and Urban, 2007). In a recent publication, Owada et al. (2006) reported that treatment of HBV with sialidase during incubation with human hepatoma cell line HepG2 would result in infection via the ASGPR, a liver-specific lectin that is responsible for uptake of desialylated glycoproteins from blood into hepatocytes (Spiess, 1990) and has been proposed as a receptor for HBV (Treichel et al., 1994; 1997). In previous studies, we could show that N- and O-glycans of HBV surface glycoproteins could be desialylated by treatment with sialidase (Schmitt et al., 1999; 2004), resulting in increased binding to HepG2 cells (Glebe and Gerlich, 2004). This binding, however, did not lead to infection of HepG2 cells or increased susceptibility of PTH (Glebe and Gerlich, 2004).

Recently, we demonstrated that HBV SVPs consisting only of SHBs were unable to bind primary hepatocytes, while SVPs formed by a fusion protein of preS1 and the S-domain bound to PTH like serum-derived SVPs, composed of all three HBV surface proteins (Glebe et al., 2005). Because it is known that the first 77 amino acids of the preS1 domain are important for infection of PHH with HBV (Le Seyec et al., 1999), this domain might serve as a binding partner for negatively charged HS on the cell-surface of hepatocytes.

Given the ubiquitous abundance of GAG on cellular and extracellular matrices, the questions arises, why experimental inoculation of very low concentrations of HBV (10–100 particles) could induce acute HBV infection in chimpanzees (Ulrich et al., 1989; Hsia et al., 2006; Yugi et al., 2006). One possibility might be that the preS domains are covered in the blood by a serum protein partially protecting HBV from non-specific attachment to blood vessels. A number of serum proteins have been proposed for this function, including albumin (Krone et al., 1990) and apolipoprotein H (Mehdi et al., 1996; Stefas et al., 2001). In a recent publication, Deng et al. (2007) propose lipoproteinlipase (LPL) as a binding factor for the preS1 domain. Because LPL itself binds to HS proteoglycanes (Spillmann et al., 2006), this might facilitate binding of HBV to hepatocytes in vivo.

Heparin and dextran sulfate have also been shown to inhibit HIV infection in vitro (Mitsuya et al., 1988); therefore, attempts have been made to use dextran sulfate and similar substances as a potential antiviral for HIV-infected patients (McClure et al., 1992). However, direct parental administration of those substances interferes with blood clotting, while their molecular weight is usually too high to be enterally absorbed. However, they provide a potential for the use as a topically applied microbicide for prevention of sexual transmission of HIV. Different candidate agents of highly sulfated polymers have been proven to provide up to 90% protection of SIV transmission in the macaque model (Weber et al., 2001) and are currently in human phase II and III trials (Mayer et al., 2003). Because HBV is also a sexually transmitted disease, use of polyanions to block sexual transmission of HIV might also work on HBV; however, this has to be validated.

Experimental procedures

Reagents

Heparin sodium salt, N-acetylated-de-O-sulfated heparin, de-N-sulfated heparin, CS, dextran sulfate, dextran, heparinase III, soluble peroxidase, sialidase and sodium chlorate were purchased from Sigma, Taufkirchen, Germany. PNGase F was from New England Biolabs.

Isolation and purification of HBV virions and SVPs from plasma of HBV-infected patients

Hepatitis B virus and natural HBsAg (genotype D, HBsAg subtype ayw2) was isolated from HBeAg-positive plasma of two asymptomatic chronic HBV carriers with normal transaminase levels in serum. One carrier (ID326) had 8 × 109 HBV genomes ml−1 and 120 μg ml−1 HBsAg, and the second (ID304) had 1.6 × 109 HBV genomes ml−1 and 55 μg ml−1 HBsAg. The purification was performed as described (Glebe and Gerlich, 2004). In brief, HBV and SVPs from 350 ml human plasma were pelleted through 10% and 15% sucrose for 15 h at 25 000 r.p.m. in a SW28 rotor (Beckman, Munich, Germany). The resuspended pellets were pooled and ultracentrifuged into a discontinuous sucrose density gradient (15%, 25%, 35%, 45% and 60%) as described above. Virus-containing fractions at 40–45% sucrose were identified by quantitative real-time PCR (LightCycler system, Roche, Germany) using primers and hybridization probes against the HBV X-region as described (Jursch et al., 2002). The assay was calibrated using the Eurohep reference plasma, which has been converted to a World Health Organization international standard sample (Saldanha et al., 2001).

Peptide synthesis and purification

Peptides were synthesized and purified in the Department of Biomolecular Chemistry at the Zentrum für Molekulare Biologie, Heidelberg (ZMBH), Germany, and by Peptide Specialty Laboratories GmbH, Heidelberg, Germany.

Isolation and culture of PTH

Primary hepatocytes of T. belangeri (Asian tree shrews) were isolated as described (Glebe et al., 2003). In brief, the livers were perfused via the portal vein with HANKS solution (Invitrogen, Karlsruhe, Germany) containing 5 mM EGTA, followed by perfusion with DMEM (Invitrogen) containing 0.05% collagenase (Sigma). Hepatocytes were selectively pelleted three times at 40 g for 6 min at 4°C. Hepatocytes were resuspended in Tupaia hepatocyte medium (THM) and plated on collagen-coated 12-well plates (105 hepatocytes per well) as described. Plating efficiency was measured prior to infection and at the end of the experiment by a modified MTT assay as described (Glebe et al., 2005). The MTT assay involves the metabolic conversion of the water-soluble compound MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, obtained from Roche] to an soluble formazan only at the membranes of intact living cells. Variability as determined by MTT assay for each preparation was found to be 10% or lower (data not shown). The organ harvest from Tupaias has been approved by the local animal protection committee.

Cell lines

Human cervix carcinoma cell line Hela and human hepatoma cell line HepG2 were obtained from the American Type Culture Collection. Cells were cultivated in DMEM (Invitrogen) with addition of 10% fetal calf serum at 37°C in a humidified incubator. For binding experiments, cells were trypsinized and cultivated on collagen-coated cell culture wells or coverslips until use.

Binding of HBV to heparin

Ninety-six-well ELISA plates pretreated by plasma polymerization (EpranEx, Plasso Technology, Portobello, UK) were coated with 25 μg ml−1 of heparin in PBS at 4°C for 12 h. After extensive washing (three times with PBS/0.2% Tween 20, two times with PBS) and subsequent blocking with 1% BSA (crystallized, Sigma) in PBS for 2 h at 37°C, highly purified HBV SVPs from chronic HBV carriers (100 ng ml−1) were incubated with indicated concentrations of polyanions for 1 h at 16°C and added to heparin-coated plates. Plates were incubated for 2 h at 37°C, and after removal of the supernatant, the plate was again washed with PBS and PBS/0.2% Tween 20, and peroxidase-conjugated anti-HBs from the Enzygnost HBsAg ELISA kit (Dade Behring, Germany) was added to each well and incubated for 1 h at 37°C. After washing as described above, an o-phenylenediamine/H2O2 substrate (tablets from DAKO) was added for 15 min at room temperature, and the amount of coloured product was measured by OD492.

Cellular binding assay for HBV

Primary Tupaia hepatocytes or indicated cell lines were plated on collagen-coated 12-well plates in THM (1 × 105 cells per well) as described above. Purified virions from plasma were diluted in THM to yield the final concentrations given in the individual experiments. Cells were incubated for the times and temperatures indicated, and unbound virus was removed by extensive washing with ice-cold THM. HBV DNA was purified from cellular lysates using a DNA extraction kit (Highpure, Roche, Germany). HBV DNA was quantified by real-time PCR (LightCycler system, Roche, Germany) using primers and hybridization probes against the HBV X-region as described (Jursch et al., 2002).

Binding and uptake of plasma-derived HBV SVPs

Primary Tupaia hepatocytes, PRH or indicated cell lines were plated on collagen-coated coverslips in THM (1 × 103 cells per coverslip) as described above. Purified SVPs from plasma containing all three HBV surface proteins were diluted in THM to yield a final concentration of 2 μg ml−1. Cells were incubated for the times and temperatures indicated and washed several times with ice-cold THM. Cells were fixed with 3% paraformaldehyde for 0.5 h at 4°C and permeabilized with 0.1% Triton X-100 in PBS for 0.5 h at room temperature. HBsAg staining was performed with the APAAP staining kit (DAKO, Hamburg, Germany) as described (Glebe et al., 2003) using monoclonal anti-HBs (Novocastra, Newcastle, UK).

Soluble horseradish peroxidase (HRP) uptake assay

Uptake of HRP was determined as described (Shurety et al., 1998) with minor modifications. In brief, primary hepatocytes were plated on collagen-coated 12-well cell culture dishes as described above. For HRP uptake, cells were incubated with 0.1 mg ml−1 HRP in DMEM for 4 h either at 37°C or 18°C. Afterwards, excess HRP was removed by several washes with ice-cold DMEM. Cells were scraped off the dishes into PBS containing 0.1% Triton X-100. After 10 min at 4°C, unsoluble cellular debris was removed by centrifugation at 5000 g for 10 min at 4°C. Soluble cellular extracts were assayed for HRP activity by adding an o-phenylenediamine/H2O2 substrate (tablets from DAKO) for 15 min at room temperature, and the amount of coloured product was measured by OD492. Resultant OD492 values were normalized for protein concentration of the cellular extracts.

Infection-inhibition of PTH cultures using polyanions or HBV preS1-lipopeptides

Highly purified HBV (100 genomes per hepatocyte) from chronic carriers were incubated with PTH for 4 h at 16°C after pre-incubation of virus with different concentrations of indicated polyanions for 1 h at 16°C. Incubation with 100 nM infection-interfering myristoylated HBV preS1 peptides 2–48, or inactive peptide (preS1 domains 19–48) for 1 h at 16°C served as a control. After extensive washing at 4°C, cells were shifted to 37°C for 12 h to allow HBV uptake. For post-incubation of polyanions, highly purified HBV from chronic carriers were incubated with PTH (100 genomes per hepatocyte) for 4 h at 16°C. After extensive washing at 4°C, cells were incubated with different concentrations of indicated polyanions and HBV peptides (100 nM) for 1 h at 16°C. After further washing steps at 4°C, cells were shifted to 37°C for 12 h to allow HBV uptake. Medium was changed every 3 days, and supernatant from day 9 to day 12 was measured for the appearance of secreted HBV e antigen (HBeAg) and HBsAg. All experiments were conducted at least in three independent series, and the results of one representative experiment are shown in each case.

Enzymatic removal of cell-surface associated GAGs

Heparinase III, sialidase or PNGase F was incubated with PTH in the appropriate digestion buffer according to the protocol given by the manufacturer. After several washings with THM, viral binding was performed for 4 h at 16°C with indicated concentrations of purified virus. Subsequently, cells were washed at 4°C, and were either analysed for viral binding as described above, or shifted to 37°C to allow infection.

Inhibition of cellular GAG sulfation

For inhibition of sulfation of cell-associated proteoglycans, PTH were cultivated for the times indicated in sulfate-free medium (Joklik modified Earls medium, Sigma) in the presence or absence of different concentrations of sodium chlorate, an known inhibitor of cellular ATP-sulfurylase. HBV infection was performed afterwards as described either in the presence or absence of sodium chlorate.

Assay for HBV-specific proteins

HBeAg was determined quantitatively by a commercially available ELISA (AxSym, Abbott Laboratories, Delkenheim, Germany) that reacts specifically with HBeAg but not with HBV core particles as described (Glebe et al., 2005). HBsAg was measured by an in-house sandwich ELISA as described (Glebe et al., 2003). Results were obtained as ratios of signal to cut-off and were converted to percentage of the non-inhibited control.

Acknowledgements

We thank S. Broehl for excellent technical assistance, C. Zartner and K.P. Valerius for maintaining the Tupaia colony, and T. Discher, G. Bein and J. Misterek for supply of HBV-positive plasma. Supported by Grant SFB A2 (to D.G. and Wolfram H. Gerlich) from the Deutsche Forschungsgemeinschaft (DFG). The authors thank Wolfram H. Gerlich for continuous support of this work.

Ancillary