Short Consensus Repeat Domain 1 of Decay-Accelerating Factor Is Required for Enterovirus 70 Binding

Abstract

Enterovirus 70 (EV70), like several other human enteroviruses, can utilize decay-accelerating factor (DAF [CD55]) as an attachment protein. Using chimeric molecules composed of different combinations of the short consensus repeat domains (SCRs) of DAF and membrane cofactor protein (CD46), we show that sequences in SCR1 of DAF are essential for EV70 binding. Of the human enteroviruses that can bind to DAF, only EV70 and coxsackievirus A21 require sequences in SCR1 for this interaction.

The complement regulator decay-accelerating factor (DAF [CD55]) can serve as an attachment protein for a number of different human enteroviruses (10), including several echoviruses (1, 8, 33), coxsackie B viruses (5, 27), and coxsackievirus A21 (29). DAF binding does not appear to be a necessary step for several of these enteroviruses to infect cells in culture (2, 3, 28–31), but it contributes significantly to virus adsorption and may function as a facilitator of virus entry. We showed previously that enterovirus 70 (EV70), the causative agent of acute hemorrhagic conjunctivitis (35), also uses DAF as an attachment protein (13). For a human enterovirus, EV70 has an unusual tropism for the conjunctiva and a broad in vitro host range (35, 36), which could be explained by the specific host cell surface molecules that EV70 utilizes for adsorption and entry. Because interaction with DAF may be an important step in the life cycle of certain enteroviruses, including EV70, and because of the potential significance of enterovirus-DAF interactions in viral pathogenesis, we were interested in localizing the EV70 binding site on DAF.

MAbs to SCRs 1 to 3 inhibit EV70 binding to HeLa cells.

As a regulator of complement activation, DAF contains four characteristic extracellular short consensus repeat domains (SCRs) of approximately 60 amino acids each (14, 20, 22, 23). Previously we reported that monoclonal antibodies (MAbs) specific for SCRs 1 and 3, but not SCR2, inhibited EV70 attachment to and infection of HeLa cells (13). This pattern of antibody blockade differed from the patterns seen for echoviruses (1) and coxsackie B viruses (5, 27), suggesting that EV70 binds to sequences in DAF that differ from those recognized by other enteroviruses. Because binding to SCRs 1 and 3, but not SCR2, was surprising, we tested a second sample of the SCR2-specific MAb (IF7) and discovered that it efficiently blocked binding of radiolabelled EV70 to HeLa cells (Fig. 1) and infection (data not shown) in a concentration-dependent manner. Therefore, antibodies recognizing DAF SCRs 1, 2, and 3, but not SCR4, inhibit EV70 binding to HeLa cells, and this correlated with their ability to protect HeLa cells against EV70 infection. The reason for the failure of antibody in our first sample of IF7 to block EV70 binding and infection of HeLa cells is not known.

FIG. 1.

Inhibition of EV70 binding to HeLa cells by DAF-specific MAbs. Confluent monolayers of HeLa cells grown in 24-well cell culture dishes were incubated with various concentrations of the following MAbs for 1 h at 37°C: 11D7 (SCR1 specific), IF7 (SCR2 specific), 1H4 (SCR3 specific), and 8D11 (SCR4 specific). The MAbs have all been described previously (1, 9) and were quantified by antibody sandwich enzyme-linked immunosorbent assay (12) using a standard curve constructed with known amounts of mouse immunoglobulin G1 as the primary antibody. For assays performed with MAb IF7, all solutions contained 2 mM CaCl₂ and 2 mM MgCl₂. Virus purification and binding assays were carried out as described previously (13). Cells were washed once and then incubated for 45 min at 33°C with 5,000 cpm of ³⁵S-labelled EV70 per well. The amount of virus bound (mean ± standard deviation for duplicate samples) is shown as a percentage of virus bound to cells that received no MAb. At a final concentration of 18 μg/ml, MAb 8D11 also showed no inhibition of virus binding.

DAF SCR1 is necessary for EV70 binding.

Antibody blockade excluded only sequences within DAF SCR4 as potential sites for interaction with EV70. To further localize the EV70 binding region, we tested chimeric receptors (Fig. 2), in which DAF domains were replaced by the corresponding domains of membrane cofactor protein (MCP [CD46]), a closely related member of the regulator of complement activation family (14), for their ability to bind radiolabelled EV70. Following transfection of NIH 3T3 cells with plasmid DNA and infection with recombinant vaccinia virus vTF7-3, as described previously (13), receptor expression was monitored by flow cytometry. MAbs specific for all four DAF SCRs and for MCP were used to confirm equivalent levels of expression of the different receptors. The binding of SCR1-specific (11D7) and SCR4-specific (8D11) MAbs was the least affected by SCR substitution and demonstrated that the different DAF-MCP chimeras were expressed at approximately equal levels (Table 1). In virus binding assays done in parallel, DAF-TM and all of the chimeric receptors with single SCR substitutions, with the exception of DM1, bound substantial amounts of radiolabelled EV70 (Fig. 3). Binding of EV70 to cells expressing DM3 was equivalent to that observed for cells expressing DAF. The amount of virus that bound to cells expressing DM2 or DM4 was approximately 40 and 70%, respectively, of that binding to cells expressing DAF. Cells expressing MCP-PI did not bind EV70; therefore, MCP sequences cannot contribute to virus binding. The virus-binding ability of some of the chimeras may be underestimated, since cells expressing DAF bound approximately two to three times more MAb than cells expressing chimeric receptors (Table 1), consistent with findings reported previously for CHO cells transfected with cDNA encoding DAF and SCR deletion mutants (5). This could reflect greater amounts of DAF on the cell surface; however, the ability of DAF-specific MAbs, particularly IF7 (SCR2 specific) and 1H4 (SCR3 specific), to recognize the cognate SCR in different chimeras varied considerably (data not shown), presumably because domain replacement perturbs conformation sufficiently to alter MAb binding. Nevertheless, these results clearly demonstrated that sequences in DAF SCR1 are necessary for EV70 binding.

FIG. 2.

Schematic representation of chimeric DAF-MCP receptors. DAF SCRs are represented as open ovals, and MCP SCRs are numbered and shaded. The sequences of the different chimeric cDNAs have all been described previously (8, 15, 16) with the exception of DM34, which was constructed by replacing the XbaI/BsrGI fragment of DM234 with the corresponding fragment from DM3. DMn indicates a molecule in which SCRn of DAF has been replaced by the corresponding SCR of MCP. Short horizontal lines represent the approximate locations of N-linked glycosylation sites. The serine-threonine-proline (STP) domain and glycosyl phosphatidylinositol anchor of DAF are represented by an open box and zigzag lines, respectively. The STP, transmembrane, and cytoplasmic domains of MCP are represented by a shaded box, a chain, and a thin vertical line, respectively.

TABLE 1.

Expression of DAF and DAF-MCP chimeras in NIH 3T3 cells

Construct	Relative level of expressiona
DAF	1.0
DAF-TM	0.55
DM1	0.39b
DM2	0.39
DM3	0.44
DM4	0.58c
DM34	0.91c
DM234	0.47c
MCP-PI	0.58d

^aBecause MCP is sensitive to trypsin (21), NIH 3T3 cells expressing the indicated protein were scraped from tissue culture plates and dispersed by pipetting. Cells from replicate wells were pooled, and the level of expression of each chimera, relative to that of DAF, was assessed by flow cytometry using an EPICS XL-MCL cytometer (Coulter). Except for the indicated chimeras, separate analyses were performed using MAb 11D7 and 8D11 (0.2 μg/ml) as the primary antibody. The secondary antibody was fluorescein isothiocyanate-conjugated sheep anti-mouse immunoglobulin G (1:20; Amersham). A minimum of 5,000 events was counted in the established gate, and mean fluorescence intensities were determined from single-parameter histograms generated with EPICS XL 1.5 software. Controls for autofluorescence and nonspecific secondary antibody binding were included in each analysis. Each value represents an average of between two and four independent determinations with each antibody. 

^bMAb 8D11 was used as the primary antibody. 

^cMAb 11D7 was used as the primary antibody. 

^dMAb E4.3 (Serotec) was used as the primary antibody. 

FIG. 3.

EV70 binding to NIH 3T3 cells expressing DAF-MCP chimeras. NIH 3T3 cells were transfected with vector alone (pCRα or pcDNA3) or with pcDNA3-DAF, pCRα-DAF, or plasmids encoding DAF-MCP chimeras, as described previously (13). DAF-TM cDNA was inserted in a pcDNA3 background. The other chimeras were in a pCRα background, with the exception of DM2 and MCP-PI, which were in pSRαEN. Cells were infected 24 h later with vaccinia virus vTF7-3 (11) at a multiplicity of infection of 20, and incubation was continued for another 20 h. Cells were washed once and then incubated for 45 min at 33°C with 5,000 cpm of ³⁵S-labelled EV70 per well. The amount of virus bound (mean ± standard deviation) is shown as a percentage of virus bound relative to transfected cells expressing DAF, once background binding was subtracted. The data in this figure are from two experiments done in triplicate. Binding to cells transfected with pCRα or pcDNA3 ranged from 2 to 4% of input virus. Binding to cells transfected with pCRα-DAF or pcDNA3-DAF ranged from 15 to 25% of input virus.

DAF SCR1 is not sufficient for EV70 binding.

To determine if DAF SCR1 sequences were sufficient for EV70 binding, two other chimeric receptors, DM34 and DM234, were tested for their ability to bind EV70. Cells expressing the DM34 receptor (which contains DAF SCRs 1 and 2) bound radiolabelled EV70 as well as did cells expressing DAF, but cells expressing DM234 (which contains SCR1 of DAF) bound only very small amounts of EV70 (Fig. 4). While the amount of SCR1-specific MAb bound to cells expressing DM34 was roughly twice the amount bound to cells expressing DM234 (Table 1), these results confirm that the binding site for EV70 includes sequences found within SCR1 of DAF. In addition, they suggest that optimal binding involves interaction between EV70 capsid sequences and sequences in both SCR1 and SCR2 of DAF. Alternatively, DAF SCR2 may have a scaffolding role (6) in presenting a virus-binding site on SCR1 in the correct conformation. It should also be pointed out that the glycosylation site located between SCR1 and SCR2 of DAF is not present in DM1, DM2, or DM234 (Fig. 2). Early work with EV70 showed that neuraminidase treatment of erythrocytes inhibited EV70-mediated hemagglutination (32). This is consistent with a role for N-linked carbohydrate in EV70 binding and implicates sialic acid in DAF-EV70 interactions. A fourth possibility is that SCR2 of MCP interferes with EV70 binding because of conformational effects on SCR1 or because the N-linked glycan in MCP SCR2 of DM2 and DM234 (15) (Fig. 2) hinders access of virus to the binding site.

FIG. 4.

EV70 binding to NIH 3T3 cells expressing chimeras with multiple SCR substitutions. NIH 3T3 cells were transfected with pCRα or with plasmids encoding DAF, DM34, or DM234. Cells were infected 24 h later with vaccinia virus vTF7-3 at a multiplicity of infection of 20, and incubation was continued for another 20 h. Cells were washed once and then incubated for 45 min at 33°C with 4,000 to 7,000 cpm of ³⁵S-labelled EV70 per well. The amount of virus bound (mean ± standard deviation) is shown as a percentage of virus bound relative to transfected cells expressing DAF, once background binding was subtracted. The data in this figure are from three experiments done in triplicate. Binding to cells transfected with pCRα ranged from 3 to 4% of input virus. Binding to cells transfected with pCRα-DAF ranged from 14 to 20% of input virus.

Sequences distal to SCR1 are not essential for EV70 binding.

The virus binding data also indicate that sequences in SCR3, SCR4, and the membrane-anchoring domain of DAF are not required for EV70 binding. However, cells expressing DM4 and DAF-TM displayed reduced virus adsorption compared to cells expressing DAF (Fig. 3), and cells expressing DM2 were able to bind greater amounts of EV70 than cells expressing DM234 (Fig. 3 and 4). These sequences may contribute to the overall conformation of DAF and, as a result, to EV70 binding.

In summary, we have demonstrated that the ability of DAF to bind EV70 resides predominantly in the SCR1 domain. Other enteroviruses also use DAF as an attachment protein (1, 5, 8, 10, 27, 29, 33), but to date, only coxsackievirus A21 has been shown to require SCR1 sequences for binding to DAF (29). SCRs 2 and 3 are involved in coxsackievirus B3 binding (5), a site in or near SCR3 appears to be crucial for interaction of DAF with coxsackievirus B5 (27), and SCRs 2, 3, and 4 (1, 8, 27) are required for echovirus 7 binding. The ability of different enteroviruses to recognize different SCR domains may reflect spatial requirements for subsequent interactions with coreceptors. Virions may have to be positioned appropriately to efficiently engage surface molecules and trigger events required for virus entry, as has been demonstrated for measles virus by using long, hybrid MCP molecules (7). While the evidence for this among the picornaviruses is scant, an extended chimeric poliovirus receptor bound poliovirus efficiently and mediated infection but produced a prolonged eclipse phase compared with that produced by poliovirus receptor itself (6). It remains to be determined if the positioning of virus binding domains in DAF will have major effects on enterovirus infection.

Is binding to DAF a necessary step in EV70 infection? It does not appear to be a necessary step for infection of cells in culture by other enteroviruses (2, 29, 30). Expression of human or murine coxsackievirus-adenovirus receptors, but not DAF, rendered rodent and rhabdomyosarcoma cells susceptible to coxsackie B virus infection (2, 3, 30, 31). Interaction with intercellular adhesion molecule 1 (CD54) induces a conformational change in coxsackievirus A21 that is required for virus entry into cells (28, 29). Similarly, exposure of echovirus 7 to DAF is insufficient for conversion of virions to A particles (24), and it has been suggested that some echoviruses can utilize more than one receptor (25). Additional cell surface molecules such as β2-microglobulin (10, 34) and 44-kDa (18, 19) and 100-kDa (26) proteins have been implicated in either enterovirus attachment or entry into permissive cells. Still, by facilitating virus adsorption, DAF could be an important determinant of enterovirus tissue tropism and pathogenesis. The ability to bind to DAF may influence the efficiency with which a particular enterovirus or enterovirus variant infects specific cell types in vivo. It has been shown, for example, that different clinical isolates of coxsackie B viruses (4), different cardiovirulent strains of coxsackievirus B3 (17), and different echoviruses (25) vary considerably in their abilities to use DAF as a receptor. Experiments to more fully characterize the interactions between EV70 and DAF and their importance in determining its unique tissue tropism and host range are ongoing.