|Home | About | Journals | Submit | Contact Us | Français|
Gangliosides have been shown to be plasma membrane receptors for both murine polyomavirus and SV40, while JC virus uses serotonin receptors. In contrast, little is known of the membrane receptor and entry pathway for BK virus (BKV), which can cause severe disease in immunosuppressed bone marrow and renal transplant patients. Using sucrose flotation assays, we investigated BKV binding to and interaction with human erythrocyte membranes and determined that this interaction was dependent on a neuraminidase-sensitive, proteinase K-resistant molecule. BKV was found to interact with the gangliosides GT1b and GD1b. The terminal α2-8-linked disialic acid motif, present in both of these gangliosides, is likely to be important for this interaction. We also determined that the addition of GD1b and GT1b to LNCaP cells, which are normally resistant to BKV infection, made them susceptible to the virus. In addition, BKV interacted with membranes extracted from the endoplasmic reticulum (ER) and infection was blocked by the addition of brefeldin A, which interferes with transport from the ER to the Golgi apparatus. These data demonstrate that BKV uses the gangliosides GT1b and GD1b as receptors and passes through the ER on the way to the nucleus.
BK virus (BKV), a member of the human polyomavirus (Py) family, was first isolated in 1971 from the urine of a renal transplant recipient (17). BKV is ubiquitous in the human population and infects a large proportion of individuals around the world (5). Primary infection with BKV most often occurs as an asymptomatic infection during childhood, although there are reports describing transient cystitis in some patients (8, 16). Antibodies against BKV are detected in 50% of children at 3 years of age and in nearly 100% of children by age 10 (37, 43). Low levels of BKV replication in lymphoid tissues, coupled with the detection of BKV DNA in the tonsils, suggest that the initial infection may occur in the tonsils (23), with infected monocytes spreading the virus to other tissues and organs, particularly the kidney (38). The subsequent urinary tract infection is usually subclinical but can be reactivated upon immunosuppression, often leading to hemorrhagic cystitis or nephritis (2, 13, 26, 27). Clinical BKV infection has become a greater problem in recent years due to larger numbers of bone marrow and kidney transplants, which require recipients to undergo immunosuppressive therapy. In addition, newer generations of drugs used in transplantation are more effective at suppressing the immune system. Evidence has suggested that BKV reactivation may be responsible for up to 5% of all renal graft dysfunction (1, 13, 30, 33). Along with individuals undergoing immunosuppressive therapy, individuals with compromised immune systems due to human immunodeficiency virus infection are at risk for clinically active BKV (20, 51). BKV is also commonly detected in the urine of pregnant women due to either immunologic or hormonal changes (9, 10).
The BKV virion is a 45-nm icosahedral particle enclosing a 5.2-kilobase circular, double-stranded DNA molecule. This genome codes for six known proteins: two early nonstructural polypeptides, the large tumor (TAg) and small tumor (tAg) antigens; and four late polypeptides, an agnoprotein thought to be important for assembly of the virion for SV40 and an inhibitor of DNA break repair for JCV and three capsid proteins, VP1, VP2, and VP3 (7, 11, 12, 40). Transcription of the BKV genome occurs divergently from a single regulatory region: the early proteins, TAg and tAg, are encoded by one strand, while the late proteins, VP1, VP2, VP3, and the agnoprotein, are encoded by the opposite strand (42).
How BKV binds to the host cell and traffics to the nucleus remains unclear. Although it is known that BKV enters the cell using a caveola-dependent mechanism and is dependent on an intact microtubule network but not an intact cytoskeleton for transport to the nucleus, many additional interactions, including the identity of its receptor, remain to be determined (14, 15). The capsid protein VP1, which comprises the majority of the virus's 72 capsomeres, is likely responsible for the interaction between the virus and the cell surface (3). It also has been shown for Py that small changes in VP1 which alter binding lead to large changes in infectivity (6, 46, 47). It has been reported that BKV hemagglutinates human erythrocytes and that this activity is sensitive to neuraminidase digestion (32, 41). Other studies have also shown that the addition of soluble gangliosides attenuates BKV-mediated hemagglutination (45). Treatment of Vero cells with neuraminidase, which cleaves sialic acid residues in the sugar moiety contained in many gangliosides and glycoproteins, inhibits BKV infection, and the addition of a mixture of Vero cell gangliosides to these treated cells restores infection (44). Gangliosides have been identified as receptors for the polyomaviruses SV40 and Py, both of which also enter the cell via a caveola-dependent mechanism (19, 49).
These studies indicating that gangliosides can affect BKV interactions with the cell combined with the finding that it uses an entry pathway similar to those of SV40 and Py, both of which also can use gangliosides as plasma membrane receptors, led us to hypothesize that BKV might interact with a specific set of gangliosides on human cells. We report the identification of the gangliosides GD1b and GT1b as receptors for BKV. We demonstrate that BKV binds to a molecule on human erythrocyte membranes that is resistant to proteinase digestion while sensitive to neuraminidase treatment. Using a sucrose flotation assay, we find that BKV binds to GD1b and GT1b, and the addition of these gangliosides allows for the infection of cells that are otherwise resistant to BKV. We further elucidate part of the pathway BKV takes from the plasma membrane to the nucleus. SV40 and Py have been shown to traffic to the endoplasmic reticulum (ER) before moving to the nucleus (18, 34), and we hypothesized that BKV follows a similar route during its initial stage of infection since it uses the same class of receptor molecule. Brefeldin A (BFA), a drug that blocks ER to Golgi trafficking, inhibits productive BKV infection, demonstrating that BKV must pass through the ER on its way to the nucleus.
Monolayer cultures of human prostate carcinoma (LNCaP) cells were maintained in Dulbecco's modified Eagle's medium (Gibco/BRL) containing 10% fetal bovine serum and supplemented with 100 units/ml penicillin and 100 μg/ml streptomycin. Human kidney proximal tubule epithelial (HPTE) cells, isolated from cadaveric tissue, were maintained in UltraMDCK media (BioWhittaker) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin, 5 mg/liter insulin, 5 mg/liter transferrin, 61 mg/liter ethanolamine, and 10 mg/liter selenium (BioWhittaker); 60 μg/liter epidermal growth factor (Sigma); 0.0651 ng/liter triiodothyronine (BioWhittaker); and 33 μg/liter retinoic acid (Sigma) (29). Cells were passaged at a 1:3 ratio upon nearing confluence. All cultures were maintained at 37°C with 5% CO2.
For infections, LNCaP cells were grown to 70% confluence and incubated with medium containing 3.2 μM concentrations of the ganglioside GM1, GM2, GM3, GD1a, GD1b, or GT1b (Matreya) for 18 h or with medium alone. Medium containing the gangliosides was removed, and the cells were washed with additional medium to remove any unincorporated gangliosides. The cells were then infected with either BKV or SV40 at a multiplicity of infection (MOI) of 5 fluorescence-forming units (ffu)/cell. Total cellular protein lysates were prepared at 5 days postinfection. HPTE cells were grown to 70% confluence and, in some experiments, treated with 5 μg/ml brefeldin A (Sigma) for 1 h prior to infection. Cells were then washed with medium to remove excess brefeldin A and infected with BKV at an MOI of 5 ffu/cell. Total cellular protein lysates were prepared at 48 h postinfection.
Protein lysates were prepared using E1A lysis buffer (25) containing 0.05 M NaF and complete EDTA-free proteinase inhibitors (Roche). Protein concentrations were determined using the Bio-Rad protein assay, and equal amounts of protein were electrophoresed. For TAg and VP1 detection, proteins were separated using 8 and 12% sodium dodecyl sulfate (SDS)-polyacrylamide gels, respectively. Proteins were transferred to nitrocellulose membrane at 25 V for 45 min with a Bio-Rad Trans-Blot SD transfer apparatus using a solution of 39 mM glycine, 48 mM Tris base, 0.037% SDS, and 20% methanol (pH 8.3). PAb416 (24) was used to detect TAg expression, P5G6 (gift of D. Galloway) was used to detect BKV VP1, and 6C5 was used to detect GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (Abcam). Primary antibodies were detected using a sheep anti-mouse immunoglobulin G/horseradish peroxidase-linked secondary antibody (Amersham) and observed using the ECL Plus Western blotting detection system (Amersham).
BKV was propagated in HPTE cells as previously described (31). Viral lysates, including all cell debris, were then adjusted to pH 7.4 with 0.5 M HEPES (pH 8.0) and centrifuged at 8,000 × g for 30 min at 4°C. The supernatant was removed, and the pellet was resuspended in 5 ml of buffer A (10 mM HEPES [pH 7.9], 1 mM CaCl2, 1 mM MgCl2, and 5 mM KCl). The pH of the lysate was then adjusted to 6.0 with 0.5 M HEPES (pH 5.4), and 5 units of type V neuraminidase (Sigma) was added and incubated for 1 h at 25°C. The solution was returned to pH 7.4 through the addition of 0.5 M HEPES (pH 8.0) and then heated to 40°C and centrifuged at 16,000 × g for 5 min. The supernatant was removed and saved, while the pellet was further incubated with 5 ml of buffer A containing 0.1% deoxycholate at 25°C for 15 min and then centrifuged at 16,000 × g for 5 min. The supernatant was removed, combined with the previous supernatant, and centrifuged through a 15% sucrose cushion into 35 g/100 ml CsCl at 141,000 × g in a Beckman Optima LE-80K ultracentrifuge using a Beckman tube (28 by 89 mm) in a Beckman SW28 rotor for 3.5 h at 20°C. Virus was dialyzed against buffer A, and titers were determined using the fluorescent focus assay as previously described (31). SV40 was grown and titers were determined as previously described (4).
Liposomes were formed with phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol, and phosphatidylserine (Avanti Polar Lipids), dissolved in chloroform, and used at a ratio of 28:8:4:1, respectively (49). Gangliosides were added to the lipid mixture to a final concentration of 12 μM. Chloroform was removed from the lipids by evaporation during low-speed centrifugation, and liposomes were resuspended in binding buffer containing 150 mM potassium acetate, 250 mM sucrose, 50 mM HEPES (pH 7.6), and 2 mM magnesium acetate, sonicated for 30 min in a sonicating water bath at room temperature, and shaken overnight at 4°C.
Human erythrocyte membranes were prepared from 7 ml of freshly harvested type O human blood. Blood was centrifuged at 500 × g for 10 min to separate cells from serum. The cells were washed three times with 10 ml of buffer containing 150 mM NaCl and 50 mM HEPES (pH 7.8) and lysed using 10 ml of water, and membranes were collected by centrifugation at 12,000 × g for 15 min. Membranes were washed three times with 1.3 ml of 20 mM Tris-HCl and 20 mM NaCl (pH 7.5) and resuspended in the 200 μl of wash buffer. Rough ER membranes were isolated from canine pancreas as previously described (21).
To remove proteins, 100 μl of human erythrocyte membranes was incubated with 20 μl of proteinase K-conjugated agarose beads (Sigma) for 30 min at 25°C. Proteinase K-beads were removed by centrifugation. Proteinase K digestion was confirmed by electrophoresing dilutions of treated and untreated membranes on a 4 to 20% SDS-polyacrylamide gel and staining using a Colloidal Blue staining kit (Invitrogen). To remove sialic acid residues, 5 μl of human erythrocyte membranes was incubated two times with 50 mU of neuraminidase from Arthrobacter ureafaciens (EMD Biosciences) for 8 h. A total of 10 μl 150 mM potassium acetate, 250 mM sucrose, 50 mM HEPES (pH 7.6), and 2 mM magnesium acetate was then added to 5 μl of proteinase K- or neuraminidase-treated membranes for each flotation assay.
A total of 5 μl of purified BKV, containing 100,000 ffu of virus, was mixed with 15 μl of either liposomes, 1 mg/ml bovine serum albumin (BSA), erythrocyte membranes, or pancreatic membranes and incubated at 37°C for 30 min at 700 rpm in an Eppendorf thermomixer. A 20-μl aliquot of the lipid/virus mix was then mixed thoroughly with 80 μl of a 73% sucrose solution. This mixture was placed at the bottom of a Beckman centrifuge tube (7 by 20 mm), and 50 μl of a 40% sucrose solution was added on top of the sample, followed by 50 μl of a 25% sucrose solution. The sample was then centrifuged for 1 h at 390,000 × g and 4°C using a Beckman Optima MAX-E ultracentrifuge with a Beckman TLA-100 rotor. Twenty-one-microliter fractions were collected and analyzed by immunoblotting.
To begin to determine the BKV receptor on the plasma membrane, we first investigated whether BKV binding to human erythrocyte membranes could be detected using a sucrose flotation assay. Erythrocytes were used for this initial assay because BKV can hemagglutinate these cells, and membranes isolated from erythrocytes consist of mainly plasma membrane due to the lack of any internal organelles. Purified BKV was incubated with either BSA or erythrocyte membranes, and a sucrose flotation assay was performed. An immunoblot for the major BKV capsid protein, VP1, was then used to detect the presence of the virus in each fraction (Fig. (Fig.1A).1A). When BKV was incubated with BSA, VP1 was observed in only the bottom fractions of the gradient, illustrating the natural property of the virus to remain at the bottom. In contrast, when BKV was incubated with erythrocyte membranes, VP1 was detected in all fractions, demonstrating that BKV was bound to the membranes. The presence of virus in all fractions is likely due to a heterogeneous population of membranes binding differing amounts of virus and floating to different densities in the gradient.
Once we determined that BKV bound to the erythrocyte membranes, we further characterized this interaction through digestion of the membranes with either proteinase K or neuraminidase, which removes either protein components or sialic acid residues, respectively. Purified BKV was incubated with proteinase K-digested erythrocyte membranes and assayed using sucrose flotation (Fig. (Fig.1A).1A). VP1 was detected in all fractions after proteinase K digestion at intensities equal to those observed in fractions from gradients containing undigested erythrocyte membranes. Digestion of the proteins present on the membranes was confirmed by Colloidal Blue staining of a gel in which aliquots of digested and undigested samples were separated (Fig. (Fig.1B).1B). These results suggest that although BKV is capable of binding to human erythrocyte membranes, the molecule important for this interaction is not a protein. Next we determined whether the molecule that binds BKV contains sialic acid residues. To this end, erythrocyte membranes were treated with neuraminidase, which cleaves sialic acid attached to proteins and lipids. Purified BKV was incubated with these membranes and assayed as described above (Fig. (Fig.1C).1C). While VP1 was detected in all fractions of the sample containing undigested membranes, VP1 was detected in only the bottom fractions after incubation of the membranes treated with neuraminidase. These results illustrate that sialic acid residues play an important role in BKV binding to erythrocytes.
As these results indicated that the molecule which binds BKV contains sialic acid residues and is unlikely to be a protein, we asked whether gangliosides bind to BKV. It has been shown previously that gangliosides act as receptors for a wide variety of molecules and viruses, most relevantly two additional polyomaviruses, Py and SV40 (18, 49). We therefore prepared liposomes formed with or without the ganglioside GM1, GT1b, or GD1b (Fig. (Fig.2A),2A), incubated cesium gradient-purified BKV with these liposomes, and assayed binding using the sucrose flotation procedure (Fig. (Fig.2B).2B). When BKV was incubated with buffer, liposomes, or liposomes containing GM1, we detected VP1 in only the bottom fractions. However, upon incubation of purified BKV with liposomes containing ganglioside GT1b or GD1b, interactions between the virus and the ganglioside were strong enough to allow flotation of the virus. Analysis of the structures of these three gangliosides reveals that GD1b and GT1b contain a common disialic acid motif (Fig. (Fig.2A).2A). We also observe that the gangliosides GM1 and GD1b are identical with the exception of the terminal α2-8-linked sialic acid. These results suggest that the terminal α2-8-linked sialic acid, which is also present in GT1b, is important for maximum interaction of the virus with the ganglioside.
Although these results indicate that BKV interacts with gangliosides as measured by a sucrose flotation assay, further studies were required to determine whether gangliosides allow BKV entry into cells to cause infection. We observed that the human prostate carcinoma cell line, LNCaP, is naturally resistant to infection with BKV but susceptible to infection with SV40. We had previously observed a similar, but reversed, situation in HPTE cells, which can be infected by BKV but not by SV40 (31); the addition of the ganglioside GM1, shown to be a receptor for SV40 (49), conferred SV40 susceptibility to the HPTE cells. We therefore asked whether the addition of gangliosides GD1b and GT1b to LNCaP cells would allow infection with BKV. LNCaP cells were mock infected, infected with SV40, or infected with BKV after preincubation with various gangliosides (Fig. (Fig.3).3). Protein lysates were extracted from these cells at 5 days postinfection and probed for TAg expression by immunoblotting. In SV40-infected cells, large amounts of SV40 TAg expression were observed, but no detectable BKV TAg expression was noted in untreated BKV-infected cells. The addition of the gangliosides GD1b and GT1b to the cells before infection with BKV resulted in expression of BKV TAg in LNCaP cells. These results confirm that BKV interacts with the gangliosides and demonstrate that this interaction allows BKV to infect cells which are normally resistant to the virus.
We next investigated whether BKV traffics through the ER prior to reaching the nucleus. HPTE cells were infected with BKV at an MOI of 5 ffu/cell after incubation with or without BFA, an inhibitor of ER to Golgi trafficking. Protein extracts from lysates prepared 48 h postinfection were probed for the expression of TAg (Fig. (Fig.4A).4A). We found that while TAg expression was readily detected in the cells not exposed to BFA, no TAg expression was detected in the BFA-treated samples. This result suggested that the path taken by BKV to the nucleus passes through the ER. As it was shown previously that GM1 and GD1a act as carriers to bring SV40 and Py, respectively, from the plasma membrane to the ER (18, 34), we hypothesized that the ganglioside to which BKV bound would be transported in a similar manner and also found on ER membranes. To test this possibility, we examined the interaction between the virus and ER membranes. Purified BKV was incubated with ER membranes or BSA and centrifuged on a sucrose gradient, and fractions were probed for the presence of VP1 (Fig. (Fig.4B).4B). We observed virus in all fractions when it was incubated with ER membrane but in only the bottom fractions when it was incubated with BSA. These results demonstrate that the ER membrane contains receptors for BKV, suggesting that BKV travels through the ER before it reaches the nucleus.
The recognition of the host cell by the virus is a key determinant of the efficiency of a viral infection. This initial interaction at the plasma membrane also can determine which route the virus will take to the nucleus. For polyomaviruses, this interaction leads to caveola-mediated entry in the cases of SV40, BKV, and Py or clathrin-mediated entry in the case of JCV (19, 36, 49). The integrity of this initial stage of infection may be critical for correct conformational changes, unfolding, and eventual disassembly of the capsid. In this study, we report that BKV exhibits interactions at the plasma membrane similar to those reported for SV40 and Py, which use gangliosides as receptors, but different from that of JCV, which has been reported to use protein molecules as receptors.
Our experiments define GD1b and GT1b as molecules that bind BKV in vitro and as receptors for the virus on cells. The lack of binding of BKV to other gangliosides in our studies suggests that it is the carbohydrate moiety to which the virus binds rather than the identical ceramide lipid moiety present in all the gangliosides. While others have described sialic acid or gangliosides in general as potential receptors for BKV (44, 45), this is this first study in which specific molecules and moieties have been identified as receptors. We demonstrated, using a sucrose flotation assay, that BKV interacts with erythrocyte membranes. Through treatment with proteinase K and neuraminidase we determined that the interaction between BKV and the membranes was not dependent on proteins but instead was dependent on sialic acid residues, which are commonly found in gangliosides. We then demonstrated that BKV interacted specifically with the gangliosides GD1b and GT1b, but not GM1, when these gangliosides were incorporated into liposomes. Additionally, BKV was unable to infect LNCaP cells without the addition of GD1b or GT1b to the cells. After we obtained these results, it was also reported that while normal prostate epithelial cells express the gangliosides GD1b and GT1b, these gangliosides are absent on LNCaP cells (39).
The interaction between BKV and the gangliosides GD1b and GT1b, but not GM1, leads us to conclude that a strong interaction between the virus and the ganglioside is dependent upon a disialic acid motif. This result suggests that the binding pocket on the BKV capsid is deeper than that of SV40, but possibly similar to that of Py, which also uses GT1b as a plasma membrane receptor (47, 49). Also interesting is that the two gangliosides which confer susceptibility to infection both contain a terminal α2-8-linked sialic acid residue. It is possible that this specific linkage is required for a productive interaction between the virus and the receptor.
We demonstrated that BKV interacts with ER membranes and that trafficking of BKV through the ER is critical for delivery of the viral genome to the nucleus, as evidenced by TAg expression. Intracellular pH changes are also necessary for efficient BKV infection (15), but exactly where these changes take place remains to be determined. The standard route of transport during ganglioside synthesis is production of the ceramide lipid in the ER, addition of the specific sugar moieties in the Golgi apparatus, and transport of the completed ganglioside to the plasma membrane. These findings, when taken together with our results, suggest that a fraction of the gangliosides at the plasma membrane traffic back to the ER, and the BKV bound to these gangliosides follows a similar pathway. Previous studies demonstrating interactions of SV40 and Py with gangliosides support this model (18, 35), but it is also possible that any or all of these viruses interact with additional molecules present in the ER. While all three viruses enter the cell by a caveola-dependent mechanism, and exiting from the ER is a critical step, neither SV40 nor Py has been shown to colocalize with the Golgi apparatus (15, 18, 35, 49). It is possible that these viruses, as well as BKV, use a BFA-sensitive pathway that does not involve the Golgi apparatus, or that they pass too rapidly through the Golgi apparatus to be detected.
While BKV can infect a wide variety of cells in culture, in the human host there are three specific cell types in which the virus is commonly found: tonsillar lymphocytes, peripheral blood mononuclear cells, and kidney epithelial cells (22, 38). Gangliosides likely play a role in the infection of each of these cell types. The initial BKV infection of a child likely occurs in the tonsils, where GD1b and GT1b have both been observed (48). BKV infection then progresses to peripheral blood mononuclear cells, from which these two gangliosides have also been extracted (50). Finally, BKV remains subclinical for the lifetime of the host in the kidney, where it has been shown that GD1b and GT1b are present in both cortical tubular and medullary tissue (28). The mechanism of transmission of BKV remains unknown, but it is not thought to be transmitted from mother to fetus. While pregnant women commonly secrete BKV in their urine, the activity of the enzyme responsible for the creation of the α2-8-linked sialic acid residue in both GD1b and GT1b is reduced nearly 80% in the tissue surrounding the fetus (52). It is possible that the downregulation of GD1b and GT1b in these tissues provides a level of protection to the fetus from the virus reactivation that occurs during pregnancy.
Given the importance of BKV infection and reactivation in transplant patients, it is critical to understand the life cycle of the virus. The identification and characterization of BKV receptors and trafficking offer potential targets for pharmaceutical agents. Thus, further studies to elucidate interactions between BKV and its target cell types are required for a complete understanding of the mechanisms used by the virus to infect the host and cause disease.
Subsequent to the submission of this manuscript, Dugan et al. reported that BKV can use an N-linked glycoprotein as a receptor (A. S. Dugan, S. Eash, and W. J. Atwood, J. Virol. 79:14442-14445, 2005).
We thank the members of the Imperiale and Tsai laboratories for their help with this work, Denise Galloway for anti-BKV VP1 antibodies, Bob Garcea for virus purification protocols, and Kathy Spindler for useful discussions and suggestions.
This work was supported by NIH grant AI060584.