INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Parvovirus B19, a small, autonomous, single-stranded DNA virus, was discovered in the sera of asymptomatic blood donors (7). It is known to be the etiologic agent of a variety of clinical disorders in humans (1, 4, 13, 14, 16, 17, 20, 24, 36, 38). Parvovirus B19 shows a remarkable tropism for human erythroid progenitor cells and is restricted in its in vitro replication to primary cells from human bone marrow (26, 27, 40, 41, 45), fetal liver (47), umbilical blood (42), and peripheral blood (37) and to two erythropoietin-differentiated megakaryocytic cell lines (23, 39). Factors responsible for the highly restricted tropism of productive B19 infection are (i) the blood group P antigen (synonyms: globoside, globotetraosylceramide, Gb4), the cellular receptor for parvovirus B19 (4); (ii) putative intracellular factors, largely restricted to human erythroid cells, which are required for optimal transcriptional activation of the B19 p6 promoter and viral replication (11, 19, 25, 45); and (iii) reduced B19 capsid protein expression in nonpermissive cells due to a block in full-length transcription of the viral genome, atypical mRNA splicing, and impaired ribosome loading of structural gene transcripts (5, 21, 28). Results from clinical pathology studies, however, have suggested that B19 viral entry also occurs into several human nonerythroid cell types (13, 14, 16, 17, 24, 36, 38). In order to investigate the mechanism of parvovirus B19 binding and entry into different hematopoietic and nonhematopoietic human cells and primary cells and to reevaluate the role of P antigen in these processes, we used a recombinant parvovirus B19 vector that would allow us to specifically study the mechanism of interaction of parvovirus B19 capsids with human cells in the absence of any parvovirus B19 genomic sequences. We have previously reported the construction of such a recombinant vector in which the parvovirus B19 capsids contain a recombinant adeno-associated virus type 2 (AAV) genome consisting of the bacterial -galactosidase (lacZ) reporter gene driven by the cytomegalovirus immediate early gene promoter (29). In those studies, the recombinant parvovirus B19 vector was shown to successfully transduce the erythroid population in normal human low-density bone marrow mononuclear (LDBM) and CD34⁺ progenitor cells as determined by 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) staining. In addition, a low-level transduction of nonerythroid LDBM cells could also be detected (29). In subsequent studies reported here, using more sensitive reporter genes (firefly luciferase [Luc] and enhanced green fluorescent protein [EGFP]) and detection conditions (flow cytometry), we have reevaluated the role of erythrocyte P antigen in parvovirus B19 binding and entry in several hematopoietic and nonhematopoietic human cell lines and in two primary nonhematopoietic human cell types. We provide evidence that P antigen is necessary and sufficient for parvovirus B19 binding to human cells; however, it is not sufficient for viral entry. Our data also suggest that a putative cell surface coreceptor is involved in the entry of parvovirus B19 into human cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells, plasmids, and viruses. Human 293, HeLa, HL-60, and M07e cells were maintained in Iscove's modified Dulbecco's medium (IMDM), and K562 and HEL cells were maintained in RPMI 1640 medium containing 10% newborn calf serum and antibiotics. Human umbilical vein endothelial cells (HUVEC) and normal human lung fibroblasts (NHLF) (Clonetics) were maintained in growth media supplied by the manufacturer. Human bone marrow cells were obtained after informed consent from healthy volunteer donors as approved by the Institutional Review Board for studies involving human subjects and were maintained in IMDM containing 20% fetal bovine serum, growth factors, and antibiotics. The generation of recombinant B19 hybrid viruses has been described previously (29). Recombinant plasmids containing the EGFP or the Luc genes under the control of the cytomegalovirus immediate early gene promoter cloned within AAV inverted terminal repeats were used to generate recombinant B19 vectors, vCMVp-EGFP/B19, and vCMVp-Luc/B19, respectively. In some experiments, Trans³⁵S label (specific activity, 1,206 Ci/mmol; ICN Pharmaceuticals Inc., Irvine, Calif.) was used to produce radiolabeled recombinant B19 as well as AAV particles, as previously described (18). All virus preparations were purified by CsCl equilibrium density gradient centrifugation and were subjected to DNase I (Boehringer Mannheim, Indianapolis, Ind.) treatment at 37°C for 30 min, followed by heat inactivation of human adenovirus type 2 (Ad2) at 56°C for 30 min. Quantitative slot blot analysis and transduction of 293 cells were performed to determine the physical and infectious titers, respectively (40, 41).

Isolation and infection of human erythroid (glycophorin A⁺ [GlyA⁺]) and nonerythroid (glycophorin A [GlyA]) LDBM cells. Glycophorin A, a glycoprotein specifically expressed on bone marrow and peripheral blood cells of the erythroid lineage, was used to separate erythroid and nonerythroid LDBM cells. To this end, bone marrow samples were immediately diluted with an equal volume of IMDM containing 20 U of heparin/ml. LDBM cells were obtained by Ficoll-Hypaque (Pharmacia, Piscataway, N.J.) density centrifugation (48), labeled with anti-glycophorin A-conjugated magnetic beads, and passed through a magnetic separation column (Miltenyi Biotech, Sunnyvale, Calif). The Gly A cells were allowed to flow through the column, and the magnetic bead-bound Gly A⁺ cells were eluted with MACS buffer (0.5% bovine serum albumin [BSA] and 5 mM EDTA in 1× phosphate-buffered saline [PBS]). The purity of Gly A⁺ LDBM cells was determined by flow cytometry and ranged between 95 and 97%. To determine the extent of contamination of the GlyA population with the GlyA⁺ cells, this population was incubated with anti-transferrin receptor (CD71) antibody since GlyA antibody could not be used. Approximately 3% of cells in the GlyA population were CD71 positive, as determined by flow cytometry.

Analysis of expression of P antigen. To determine the presence of P antigen (globoside) on the cell surface, unpermeabilized cells were washed with cold PBS-1% BSA twice, incubated with rabbit anti-human globoside antibody (Matreya) and mouse anti-rabbit phycoerythrin-conjugated secondary antibody (Boehringer-Mannheim), and analyzed by flow cytometry. Cells incubated with only the secondary phycoerythrin-conjugated antibody were used as controls.

Virus binding assays. Binding experiments were carried out as described by Mizukami et al. (22) with the following modifications: adherent cells were detached by brief trypsinization, and 3 × 10⁵ adherent and suspension cells were incubated with ~2 × 10⁵ total cpm of [³⁵S]methionine-labeled viral particles (3 × 10⁶ particles of recombinant AAV or 1 × 10⁶ particles of recombinant parvovirus B19) for 60 min on ice. Cells without addition of radiolabeled virions and tubes without cells incubated with radiolabeled virions alone were included as controls. All cells were washed extensively with PBS-1% BSA to remove any unbound virions, and cell-associated radioactivity was determined in a liquid scintillation counter. Nonspecific binding was subtracted from each value to calculate the specific binding. In competition experiments, 5 × 10⁹ [³⁵S]methionine-labeled recombinant AAV or parvovirus B19 particles were preincubated with 10 µg of globoside (globotetraosylceramide; Sigma, St. Louis, Mo.) for 90 min in a total volume of 500 µl on ice before performing the binding assays. All experiments were performed in duplicate, and the data shown represent mean values from three independent experiments.

Transduction assays. Equal numbers of cells (2 × 10⁵ for LDBM cells or 1 × 10⁶ for cell lines and primary cells) were either mock infected or infected with either 2,000 physical particles per cell (ppc) of recombinant vCMVp-EGFP/B19 (LDBM cells) or 8,000 ppc of recombinant vCMVp-Luc/B19 vectors for 2 h at 37°C. Half of the cells were cultured in complete medium containing an Ad2 multiplicity of infection of 10 for 48 h. EGFP expression was detected by flow cytometry, and Luc activity was determined in cell lysates using a Luc reporter system (Promega). Values of mock-infected cells were subtracted from those obtained for infected cells. Fewer than 10 relative light units/µg of protein were considered negative. The data shown represent the mean of three independent experiments performed in duplicates.

Kinetics of viral entry. To study the cellular uptake of parvoviral particles, radiolabeled recombinant AAV and B19 virions were bound to the cells for 60 min on ice as described for the binding assay. Unbound particles were removed by washing with cold PBS-1% BSA, and the cells were transferred for 2 to 30 min to 37°C to allow viral entry to occur. Uninternalized virions were removed by trypsinization and washing with PBS-1% BSA. Cell-associated radioactivity was determined in cell lysates as described for the virus binding assays.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Primary human nonerythroid cells can be transduced by the recombinant parvovirus B19 vector. In our previous studies, we reported transduction of the erythroid population of normal human LDBM cells and erythroid-differentiated CD34⁺ human progenitor cells by a recombinant parvovirus B19-lacZ vector, but we also consistently observed a low-level transduction of nonerythroid LDBM and CD34⁺ cells, as determined by X-Gal staining (29). Using higher ppc ratios (2,000 ppc versus 200 ppc) of a recombinant parvovirus B19-EGFP vector, we reevaluated the transduction efficiency in LDBM cells that had been separated into erythroid and nonerythroid populations by using an antibody against the erythroid cell surface glycoprotein, glycophorin A. Twenty-four and 48 h after infection, EGFP was expressed in ~70 and 79%, respectively, of GlyA⁺ (erythroid) LDBM cells, as determined by flow cytometry (Fig. 1A). In accordance with our previous results, transgene expression could also be observed in GlyA (nonerythroid) LDBM cells (~20 and 15% transduction efficiency; Fig. 1A). Thus, the transduction efficiency determined in both populations by using more sensitive conditions was substantially higher than in our previous study where we also used lower ppc ratios (29). In order to determine whether the cellular receptor for parvovirus B19, the blood group P antigen (globoside), was expressed on nonerythroid LDBM cells, we performed indirect immunofluorescence analyses of unpermeabilized cells using flow cytometry. Approximately 75% of GlyA⁺ LDBM cells were positive for P antigen compared with ~30% of GlyA LDBM cells (Fig. 1B). The mean fluorescence intensity, a measure of the expression level of P antigen per cell, was significantly higher in the GlyA⁺ population than in the GlyA population of LDBM cells (30.78 ± 0.67 versus 4.51 ± 0.13 arbitrary fluorescence units; P < 0.05). A dual-color analysis of LDBM cells infected with the recombinant parvovirus B19-EGFP vector and stained for P antigen 24 h postinfection revealed that EGFP expression was almost exclusively confined to the P antigen-positive population in both GlyA⁺ and GlyA LDBM cells (data not shown). These results demonstrate that P antigen is expressed in both erythroid and nonerythroid populations in normal human LDBM cells and that recombinant parvovirus B19 vectors can successfully transduce P antigen-positive cells in both populations.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Flow cytometry analyses of parvovirus B19 vector-mediated transgene expression (A) and P antigen expression (B) in erythroid and nonerythroid cell populations obtained from primary LDBM cells. These assays were performed as described in Materials and Methods.

P antigen is expressed on established human hematopoietic cell lines, but viral binding does not correlate with the level of expression. Since GlyA⁺, and especially GlyA, LDBM cells represent highly heterogeneous populations, we wished to use established hematopoietic human cell lines to further investigate the role of P antigen in recombinant parvovirus B19 vector binding and entry. We chose two erythroleukemic cell lines, HEL and K562; a promyelocytic leukemia cell line, HL-60; and a megakaryocytic leukemia cell line, M07e; and determined their P antigen expression levels as described above. As can be seen in Fig. 2A, P antigen is expressed in HEL, K562, and HL-60 cells but not in M07e cells. In order to test the functionality of P antigen expressed on HEL, K562, and HL-60 cells, we performed virus binding assays. To this end, [³⁵S]methionine-labeled recombinant AAV and parvovirus B19 vectors were incubated with each cell type, followed by extensive washing, and cell-associated radioactivity was determined. As expected, M07e cells, which lack the cellular receptor and coreceptor for AAV, heparan sulfate proteoglycan, and fibroblast growth factor receptor 1 (33, 44), did not bind recombinant AAV vectors. No binding with recombinant B19 vectors was detected since these cells also lack P antigen. Binding of recombinant AAV virions to K562 cells was most efficient compared with that to HEL and HL-60 cells. All three cell types also bound recombinant parvovirus B19 virions (Fig. 2B). Although the same numbers of cells were used, the binding efficiency of AAV and parvovirus B19 could not be directly compared in these experiments since viral stocks with the equivalent radioactivity contained higher numbers of AAV particles than of parvovirus B19 particles. These results suggest that P antigen is required for parvovirus B19 binding but that levels of P antigen expression do not directly correlate with the extent of viral binding.

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Flow cytometry analyses of P antigen expression (A) in and 35S-labeled viral binding (B) to established human hematopoietic cell lines K562, HEL, HL-60, and M07e. These assays were carried out as described in Materials and Methods. cpm, counts per minute.

P antigen expression also occurs on established and primary nonhematopoietic human cells, and parvovirus B19 can bind to these cells. In order to investigate the mechanisms of parvovirus B19 binding and entry into nonhematopoietic human cells, we chose two established cell lines of epitheloid origin: an embryonic kidney cell line, 293, and a cervical carcinoma cell line, HeLa. In addition, we included two primary human cell types which might be of interest for future gene therapy approaches, HUVEC and NHLF. Flow cytometry analysis of cell surface-expressed P antigen revealed the presence of P antigen on all four cell types, with the lowest expression levels observed in HeLa cells (Fig. 3A). Binding assays using [³⁵S]methionine-labeled recombinant AAV and parvovirus B19 particles were performed as described above, except that binding was carried out in suspension cultures in order to maintain comparable experimental conditions for suspension and adherent cells. These results are shown in Fig. 3B. As can be seen, AAV binding occurred with all four cell types tested. Interestingly, binding of parvovirus B19 was also observed.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Flow cytometry analyses of P antigen expression (A) in and 35S-labeled viral binding (B) to established human cell lines 293 and HeLa and primary cell lines HUVEC and NHLF. These assays were performed as described in Materials and Methods. cpm, counts per minute.

View larger version (16K): [in this window] [in a new window]   FIG. 4.   Effect of globoside treatment on the binding of radiolabeled AAV or parvovirus B19 to 293 cells. These assays were performed as described in Materials and Methods. ()Gb4, absence of Gb4; (+)Gb4, presence of Gb4.

Recombinant parvovirus B19 vectors can transduce both hematopoietic and nonhematopoietic human cells. Having documented that parvovirus B19 could indeed bind to both hematopoietic and nonhematopoietic cell lines and to primary human cells, it was next of interest to examine whether these cells were also permissive for the B19 capsid-mediated steps of infection, such as vector internalization, transfer to the nucleus, uncoating, and release of the recombinant genome into the nucleus. Since the parvovirus B19 vector contained a recombinant AAV genome and since AAV second-strand synthesis is a rate-limiting step for transduction (34), all cell types were coinfected with Ad2, known to promote the viral second-strand DNA synthesis and transgene expression. Transduction assays were carried out in the absence of coinfection with Ad2 as well. Initially, each cell type was transfected with a recombinant plasmid containing the cytomegalovirus promoter-driven Luc reporter gene to ensure that the cytomegalovirus promoter was active in these cells. Abundant transgene expression was detected in all cell types (data not shown). Subsequently, 10⁶ cells of both the hematopoietic and nonhematopoietic lines were transduced with 8,000 ppc of the recombinant B19-Luc vector for 2 h, following which 50% of each cell type were infected with an Ad2 multiplicity of infection of 10. Forty-eight hours after infection, transgene expression was determined as described in Materials and Methods. As is evident in Fig. 5A, recombinant B19 vectors did not infect M07e cells as expected, since these cells do not express P antigen and are unable to bind the virus. No transgene expression could be detected in HEL, K562, and HL-60 cells, even with coinfection with Ad2, although these cells express P antigen and efficiently bind the virus. The transduction data for nonhematopoietic cells are shown in Fig. 5B. Note that the y axis is in a logarithmic scale. The efficiency of transgene expression in the absence of coinfection with Ad2 was low. Following Ad2 coinfection, transgene expression increased substantially in all four cell types. The highest levels of transgene expression were seen in NHLF, suggesting that recombinant B19 vectors entered these cells and delivered the recombinant genomes into the nucleus. Transgene expression in these cells, however, was largely dependent on unimpaired second-strand DNA synthesis. These results indicate that P antigen is necessary and sufficient for binding of recombinant B19 virions; however, it is not sufficient for viral entry. In addition, these data demonstrate that nonhematopoietic established cell lines (293) as well as primary human cells (HUVEC and NHLF) can be transduced by recombinant B19 vectors.

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Analyses of parvovirus B19 vector-mediated transgene expression in established human hematopoietic cell lines (A) and in established and primary nonhematopoietic human cells (B). These assays were carried out as described in Materials and Methods. RLU, relative light units. Ad, Ad2.

Kinetics of parvovirus B19 entry into cells parallels that of AAV. The lack of transduction of cells by recombinant B19 vectors can be due to one or more of the following: lack of viral binding, impaired viral entry, trafficking, uncoating, and delivery of the viral genome to the nucleus. We ruled out the lack of vector binding as a possible cause of lack of transduction of K562, HEL, and HL-60 cells. In order to investigate the early, post-receptor binding steps in parvovirus B19 infection and to evaluate the role of P antigen in these processes, we compared the kinetics of parvovirus B19 entry into different human cells with that of AAV, which has been described recently (2). All entry assays were performed with cells in suspension. These results are shown in Fig. 6A. Consistent with data reported by Bartlett et al. (2) for adherent HeLa cells, AAV particles were rapidly taken up by HeLa and 293 cells, with approximately 50 to 67% of surface-bound virions internalized within 30 min, confirming our previous observation that brief trypsinization did not significantly impair the AAV receptor and coreceptor function in these assays. Preincubation of AAV particles with globoside did not have an effect on AAV uptake. Interestingly, parvovirus B19 virions entered 293 and HeLa cells with a similar kinetic, leading to an uptake of approximately 60% of virions within the first 30 min. Preincubation of recombinant B19 virions with globoside, however, abolished viral uptake (Fig. 6A). Recombinant AAV virions also entered NHLF efficiently (~70%) (Fig. 6B). Although K562 cells bound AAV particles very efficiently (Fig. 2A), transduction of these cells with recombinant AAV vectors has been reported to be low (33; data not shown). Impaired entry of AAV into K562 cells (~27%) (Fig. 6B) may contribute to this phenomenon. Recombinant B19 virions were taken up efficiently by NHLF, but the extent of viral entry into K562 and HL-60 cells was substantially lower and absent, respectively (Fig. 6B). Taken together, these data suggest that despite the use of different receptors and perhaps coreceptors, AAV and parvovirus B19 appear to use a very similar mechanism(s) to enter human cells.

View larger version (18K): [in this window] [in a new window]   FIG. 6.   Kinetics of entry of AAV or parvovirus B19 into established human cell lines 293 and HeLa (A) and into K562, HL-60, and primary NHLF (B). These assays were performed as described in Materials and Methods.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Since the discovery of erythrocyte P antigen as a cellular receptor for parvovirus B19 in 1993, little progress has been made in understanding the underlying mechanism of viral entry into human cells. Two factors have contributed to this. The first factor is the remarkably narrow tissue-tropism of parvovirus B19. For example, a productive infection and efficient replication of parvovirus B19 are limited to cells in the erythroid lineage in primary human hematopoietic cells which are heterogeneous. Although we and others have identified two human megakaryocytic leukemia cell lines that are permissive for parvovirus B19 infection, the efficiency of replication has been shown to be very poor. The second factor is the lack of availability until recently (29) of a recombinant parvovirus B19 vector with which early steps in the viral infection can be studied. However, the detection of parvovirus B19 capsid proteins and viral DNA in several human nonerythroid cell types in clinical pathology studies (13, 14, 17, 24, 36, 38) suggests that the viral entry might not be limited to cells in the erythroid lineage. With the use of a recombinant parvovirus B19 vector, we were able to confirm the role of P antigen as the primary attachment receptor for parvovirus B19. However, we could not observe a direct correlation between cell surface expression levels of P antigen and the efficiency of recombinant parvovirus B19 vector binding. At least two factors could account for this observation. First, the binding of parvovirus B19 to its primary receptor, P antigen, might be more efficient and/or stable in some cell types if interaction with an as-yet-unidentified coreceptor is required. The expression level of this putative coreceptor might be different among different cell types. Second, it is also conceivable that the accessibility of cell surface P antigen for parvovirus B19 might vary in different cell types due to differences in the composition of membrane-associated polysaccharides and/or polypeptides. Despite efficient binding of parvovirus B19 to the hematopoietic cell lines K562 and HL-60, these cells could not be transduced by the recombinant vector, because the virus failed to efficiently enter these cells. These data corroborate our contention that P antigen is necessary but not sufficient for a successful infection by parvovirus B19 and strongly suggest the existence of a putative cellular coreceptor that is required for efficient entry of parvovirus B19 into human cells. The use of a cellular receptor for virus attachment and the requirement of a coreceptor for viral entry have been established for a number of viruses. It has recently been shown that the cellular receptor for AAV attachment is heparan sulfate proteoglycan (44). Subsequently, we identified a cellular coreceptor for AAV, fibroblast growth factor receptor 1 (33), and Summerford et al. documented the involvement of V5 integrin in AAV entry (43). Adenoviruses are bound to human cells via three different cell surface molecules: the coxsackievirus and adenovirus receptor, major histocompatibility complex class I molecules, and integrin _M2 (3, 15, 35); for entry into the cell, however, the interaction with coreceptors, identified as the _V integrins, V₃ and V₅, is required (46). Studies on the cellular receptor for human immunodeficiency virus type 1 (HIV-1), human CD4, revealed that mouse cells which were transfected with human CD4 allowed HIV-1 attachment to the cells but that the virus was unable to enter. Different chemokine receptors (mainly CXCR4, CCR5, CCR3, and CCR2b) have subsequently been identified as coreceptors for different HIV-1 strains (6, 8-10). Interestingly, several recent studies suggest that in addition to the CD4 antigen and different chemokine receptors, neutral glycosphingolipids (GSLs) might play a role in HIV-1 infection. Incorporation of human erythrocyte GSLs into nonhuman CD4⁺ or GSL-depleted human CD4⁺ cells rendered these cells susceptible to HIV-1 envelope glycoprotein-mediated fusion, suggesting a potential role for GSLs in the proper arrangement of plasma membrane receptor molecules for optimal interaction with viral components (12, 30-32).

Parvovirus B19-mediated transduction of primary HUVEC and NHLF, which we show express P antigen, suggests that this vector might be an attractive alternative to the more commonly used recombinant AAV vectors for gene transfer purposes. In addition, since these cells allow viral entry, their homogeneous nature might be exploitable to reveal the identity of the putative coreceptor(s) for parvovirus B19. The identification of the putative cellular coreceptor(s) for parvovirus B19 will shed light on the underlying mechanisms of viral pathogenesis as well as have important implications in the use of recombinant B19 vectors as a tool in gene therapy approaches.

Finally, although these studies reinforce the role of cell surface P antigen as a primary receptor for binding of parvovirus B19, it is difficult to envisage that the virus would evolve a strategy to use a cellular receptor that is expressed most abundantly on mature erythrocytes which lack nuclei. Because parvovirus replication in general occurs in the cell nucleus, it would be suicidal for parvovirus B19 to infect and enter mature erythrocytes. We predict that mature erythrocytes lack the putative coreceptor(s). We also hypothesize that P antigen binds and concentrates parvovirus B19 in membrane microdomains and allows interaction with the appropriate additional cellular coreceptor(s) expressed only on permissive cells, which, in turn, leads to viral entry. In the absence of this putative coreceptor(s) on mature erythrocytes, the virus might exploit the high levels of P antigen present on these cells for a highly efficient systemic dissemination.