required for C3 convertase activity (Coyne et al., 1992), and evidence suggests that SCR1 interacts with CD97, a member of the seven transmembrane receptor ...
Journal of General Virology (1999), 80, 3145–3152. Printed in Great Britain ...................................................................................................................................................................................................................................................................................
Mapping the binding domains on decay accelerating factor (DAF) for haemagglutinating enteroviruses : implications for the evolution of a DAF-binding phenotype Robert M. Powell,2 Trevor Ward,1, 2 Ian Goodfellow,1 Jeffrey W. Almond2† and David J. Evans1 1 2
Division of Virology, Institute of Biomedical and Life Sciences, University of Glasgow, Church Street, Glasgow G11 5JR, UK School of Animal and Microbial Sciences, The University of Reading, Whiteknights, PO Box 228, Reading RG6 5AJ, UK
Decay accelerating factor (DAF) functions as a cell attachment receptor for a wide range of human enteroviruses, the interaction accounting for the haemagglutination phenotype exhibited by many members of this family. Haemagglutination inhibition assays using purified truncated soluble DAF (sDAF) receptors and short consensus repeat (SCR) domain-specific antibodies have been used to determine the domain(s) of DAF to which the viruses bind. Further sDAF-mediated virus neutralization and biosensor analysis have been used to confirm the virus-binding domains of DAF. Of the four distinct clusters of human enteroviruses, three contain representatives that bind DAF. The majority of DAF-binding enteroviruses occupy the ‘ CBV-like ’ cluster, and require SCR domains 2–4 for DAF binding. In contrast, the DAF-binding representatives of the ‘ ENV70-like ’ and ‘ PVlike ’ clusters require SCR1 for DAF interaction. These studies confirm that DAF binding is a widespread characteristic amongst phylogenetically divergent clusters within the enteroviruses and suggest that the ability to bind DAF may have evolved more than once within this group of viruses.
Introduction Picornaviruses are small non-enveloped icosahedral, positive-sense, single-stranded RNA viruses that cause a range of diseases in humans and other animals, including paralytic poliomyelitis, encephalitis and myocarditis (Melnick, 1996). Although exhibiting a high level of primary sequence and structural conservation, the enterovirus family of the Picornaviridae, which includes poliovirus (PV), echovirus (EV), the enteroviruses (ENV) and coxsackie A and B viruses (CAV and CBV), cause disease by replication in a divergent range of organ and tissue types. Tissue tropism is at least partly determined by the cellular receptor (Evans & Almond, 1998). Therefore, the identification of receptors and the subsequent characterization of virus binding is a prerequisite to understanding virus pathogenesis. The human enteroviruses are divided into sub-groups largely on the basis of pathogenicity in experimental animals Author for correspondence : David Evans. Fax j44 141 330 6249. e-mail David.Evans!vir.gla.ac.uk † Present address : Pasteur Me! rieux Connaught, 1541 Avenue Marcel Me! rieux, 69280 Marcy-L’Etoile, France.
0001-6472 # 1999 SGM
or, more recently, are numbered in order of identification. Since the virus receptor used influences tropism, and hence pathogenesis, a correlation between the enterovirus sub-grouping and the cellular receptor used for infection would be expected. For example, all serotypes of PV use a member of the immunoglobulin super-family of unknown cellular function [the PV receptor (PVR)] (Mendelsohn et al., 1989). However, the relatively general discrimination made on the basis of pathogenesis conceals significantly wider variation in the range of virus receptors used for cell attachment and infection. Furthermore, molecular techniques enable a distinction to be made between attachment and infection, which can be mediated by the same or different receptors. Since the identification of PVR, a number of additional enterovirus receptors have been identified including : the coxsackie and adenovirus receptor (CAR) ; the receptor for the coxsackie B viruses (Bergelson et al., 1997 a) ; intracellular adhesion molecule type 1 (ICAM-1), which is bound by several CAVs (Colonno, 1986 ; Pulli et al., 1995 ; Shafren et al., 1997 a) ; the α β integrin (VLA-2), the receptor for EV1 (Bergelson et al., # " 1992) ; and decay accelerating factor (DAF ; CD55), which was initially identified as the receptor for a subset of the haemagglutinating EV (Bergelson et al., 1994 ; Ward et al., DBEF
R. M. Powell and others
Fig. 1. Structure of DAF and domain recognition by SCR-specific MAbs. Schematic diagram of the structure of DAF indicating the four SCR domains, the heavily O-glycosylated serine/threonine-rich domain and the GPI anchor. The single N-linked glycosylation site indicated between SCR1 and SCR2 was removed during the construction of sDAF123 and sDAF12. The SCR-specificity of the domain-specific monoclonal antibodies (mAb) is also shown.
1994). Subsequent studies have shown that the haemagglutination (HA) phenotype of enteroviruses, including CAV21 (Shafren et al., 1997 a), CBV serotypes 1, 3 and 5 (Shafren et al., 1995) and EV serotypes 3, 6, 6h, 7, 11–13, 19, 21, 24, 25, 29, 30 and 33, is due to interaction with DAF on the erythrocyte (Powell et al., 1998). Although the HA phenotype of CBV3 and CAV21 is a consequence of DAF binding, cell binding and subsequent infection require CAR and ICAM-1, respectively (Shafren et al., 1997 a, b). The distinction between the simultaneous ability to bind two attachment molecules and the adaptation of a single isolate to alternative receptors is not always clear (Bergelson et al., 1995 ; Reagan et al., 1984 ; Shafren et al., 1995), but it is notable that in the known cases of multireceptor tropism amongst the enteroviruses, DAF is always implicated. DAF is a member of the regulator of complement activity protein family and prevents the deposition of C3 on the cell surface by enhancing decay of the C3 convertases, so protecting the cell from complement-mediated lysis (Lublin & Atkinson, 1989). The four short consensus repeat (SCR) domains of DAF are attached to the cell surface by a glycosylphosphatidyl inositol (GPI) anchor, and separated from the cell membrane by a heavily glycosylated serine- and threonine-rich domain (see Fig. 1). Only SCR domains 2–4 are required for C3 convertase activity (Coyne et al., 1992), and evidence suggests that SCR1 interacts with CD97, a member of the seven transmembrane receptor family (Hamann et al., 1996). We have previously demonstrated that DAF binding by EV7, in solution or to the cell surface, requires SCR domains DBEG
2–4 (Clarkson et al., 1995 ; Powell et al., 1997), which contrasts with the majority of known enterovirus–receptor interactions, in which the N-terminal domain of the receptor is generally implicated in virus binding (Evans & Almond, 1998). In the case of EV11, which has similar binding requirements to EV7, we have demonstrated, using surface plasmon resonance analysis, that almost 80 % of the binding affinity is located within SCR3 (Lea et al., 1998). A range of techniques has been used to map the SCRspecificity of DAF-binding viruses. These include the use of SCR-specific MAbs to block binding or infection, virus binding\infection studies using chimeric DAF receptors, and soluble DAF (sDAF) inhibition of virus binding, cell infection or HA. These studies suggest that SCR1 is the principal binding determinant of CAV21 and ENV70 (Karnauchow et al., 1998 ; Shafren et al., 1997 a), SCR2 is the domain bound by rhabdomyosarcoma (RD)-adapted CBV3 (Bergelson et al., 1995), and that, where known, SCR3 is implicated in binding the CBV and haemagglutinating EVs (Powell et al., 1998 ; Shafren et al., 1995). We have extended these studies using truncated sDAF derivatives, expressed in the yeast Pichia pastoris, to determine the SCR domains bound by a range of HA enteroviruses. Our results, together with published data from other laboratories, suggest that, although DAF binding is widespread amongst enteroviruses, there is a clear pattern to the SCR domain(s) required for binding by defined groups of enteroviruses. This suggests that DAF binding may have evolved several times within this group of viruses and that, by inference, there is a selective pressure to bind DAF, indicating that this phenotype is of considerable benefit to the virus.
Methods
Viruses and antibodies. EV and coxsackieviruses were a kind gift from Brian Megson, Public Heath Laboratory Service, Colindale, London, UK. The isolates had been obtained from the ATCC in 1988 and have been passaged in RD cells. The antibodies used in these studies (BRIC110, BRIC220 and 2D2-1) were a kind gift from G. Daniels at the Bristol blood transfusion service (UK), and have the domain binding specificity (Coyne et al., 1992) shown in Fig. 1. MAb 854 has been described previously (Daniels et al., 1998 ; Ward et al., 1994).
Expression of sDAF deletion mutants in Pichia pastoris. The expression of sDAF deletion mutants in the Pichia pastoris yeast expression system has previously been described in detail (Powell et al., 1997). sDAF containing all the SCRs, but excluding the serine\threonine domain was expressed, as were all contiguous 2 and 3 SCR constructs (sDAF , sDAF , sDAF , sDAF and sDAF ) and characterized for "#$ #$% "# #$ $% correct SCR folding by demonstrating interaction of sDAF fragments with a range of anti-DAF antibodies in an enzyme-linked immunoassay (data not shown). The single N-linked glycosylation site between SCRs 1 and 2 was deleted during the construction of sDAF and sDAF . "#$ "#
HA inhibition (HAI) assay using sDAF. One HA unit is the amount of infected tissue culture (TC) supernatant required to haemagglutinate 50 µl 0n5 % human erythrocytes. sDAF deletion mutants were made up to volume in PBS and serially diluted twofold in 50 µl PBS in a 96-well HA plate from a starting concentration of 8 µM. Four HA units
DAF-binding enteroviruses
Table 1. SCR domain mapping of DAF-binding enteroviruses Summary of the SCR domains involved in DAF binding by 12 serotypes of EV, CBV types 1 and 3, and CAV type 21, as determined by three separate methods : HAI, neutralization mediated by soluble derivatives of DAF containing the SCR domains indicated (neutralization), and surface plasmon resonance analysis of virus binding to the sDAF sub-fragment indicated (biosensor). Analysis method Interaction with sDAF domain :* Virus and serotype EV3 EV6 EV6h EV6d EV7 EV11 EV12 EV13 EV19 EV24 EV29 EV33 CBV1 CBV3 CAV21
1234 234 123 12 j j j j j j j j j j j j j j j
j j j j j j j j j j j j j j p
k k k k k k k k k k k k k k p
k k k k k k k k k k k k k k k
23
34
k k k k k k k k k k k k k k k
k k k k k k j k k k p k k k k
HAI HAI (MAb) (sDAF) †
†
†
† † † † † † †
† † † † †
† †
† †
† †
Neutralization
Biosensor
† † † †
† †
† †
† †
† † † †
* SCR domains of DAF that retain the virus binding determinants (j), or those that have no virus binding activity (k).p, SCR domains with demonstrable binding activity that is less than 50 % of that observable with sDAF . "#$% † Procedures applied to analysis of any particular virus type. of TC-infected virus culture were added in 10 µl PBS and the virus–receptor mixture was incubated for 1 h at 4 mC. Fifty microlitres of 0n5 % erythrocytes was added in PBS and the virus\cell suspension was mixed by gentle pipetting. Plates were incubated for 2 h at 4 mC and photographed.
HAI using a panel of anti-DAF monoclonal antibodies. A panel of anti-DAF MAbs directed against various domains were made up to volume in PBS and serially diluted twofold in PBS in a HA plate. The MAbs used, and starting concentrations, were : BRIC220 (SCR1), 10 µg\ml ; BRIC110 (SCR2), 10 µg\ml ; 854 (SCR3), 1 : 1000 of ascites ; and 2D2-1 (SCR4), 0n25 µg\ml. Fifty microlitres of 0n5 % erythrocytes was added to each well and the plates were incubated for 1 h to permit antibody binding. Four HA units of virus were added to each well and the cells were resuspended by gentle pipetting. Plates were incubated at 4 mC for 2 h and photographed.
Virus neutralization by sDAF. EV serotypes [10$ 50 % tissue culture infectious doses (TCID )] were incubated with twofold serial &! dilutions of sDAF deletion mutants starting from 8 µM. Following incubation at 37 mC for 1 h, the virus\DAF mixtures were applied to monolayers of RD cells in a 96-well plate and incubated for 24 h at 37 mC prior to examination for cytopathic effects by light microscopy and staining with crystal violet.
Mapping SCR domain interactions using a biosensor. EV6, 6h, 7 and 12 were grown in RD cells and purified by PEG precipitation followed by centrifugation through a 30 % sucrose cushion. The virus preparations (approximately 300 µg) were resuspended in PBS and analysed by PAGE and silver staining. Only viral proteins were visible
(data not shown). All analyses were performed on a BIAcore 2000 (BIAcore) biosensor. Viruses were covalently immobilized to a dextran matrix of a CM5 sensor chip using the amine coupling kit as described (BIAcore). Approximately 10K resonance units (RU) of each serotype were immobilized (data not shown). sDAF deletion mutants were assayed for binding by injecting 1 µM samples made up in commercially obtained HEPES-buffered saline (BIAcore) at a rate of 20 µl\min for 100 s. For dissociation of virus–receptor complexes between injections, 1200 seconds were permitted. Two negative controls were performed. Each sDAF fragment was also assayed for the ability to interact nonspecifically with a blank sensor chip and gave a negligible shift of 20p10 RU. Subsequently, an irrelevant protein (1 %, w\v BSA) was injected over the immobilized viruses and gave a shift of 150p30 RU (data not shown). The background binding due to non-specific interactions is estimated as the sum of these values, i.e. approximately 170 RU. The actual background values measured were 120 RU for EV6, 130 RU for EV6h, 150 RU for EV7 and 200 RU for EV12.
Results HAI assays
The HA phenotype of EV is explained by the binding of the virus to DAF expressed on the surface of erythrocytes (Powell et al., 1998). We have extended our earlier studies of the widespread nature of DAF binding amongst human EV to investigate the receptor domains involved in the virus– receptor interaction. Two types of HAI assays were performed DBEH
R. M. Powell and others
Fig. 3. HAI assays using soluble DAF. Six representative enteroviruses are shown in HAI assays using serial doubling dilutions of sDAF or truncated derivatives of sDAF. The numbers 1–4 indicate the SCR domains present in the competing sDAF protein. These results are tabulated in Table 1.
potential involvement of SCR2, all predominantly interact with SCR domains 3 and 4. In contrast, and in line with published results, CAV21 interacted in a distinctly different manner to the other viruses tested, and was only blocked by antibodies to SCR1.
Fig. 2. HAI assays using SCR-specific MAbs. Five representative enteroviruses are shown in HAI assays using serial doubling dilutions of the SCR-specific MAbs indicated. These results are tabulated in Table 1.
on a range of DAF-binding viruses, using antibodies to defined SCR domains of DAF or purified sDAF proteins. Representative results are discussed below and tabulated in Table 1. HAI assays with anti-DAF antibodies. A representative sample of DAF-binding viruses were screened in a HAI assay (Fig. 2, Table 1) following treatment of erythrocytes with each of a panel of anti-DAF MAbs that have previously been mapped to the four SCR domains of the receptor (see Fig. 1). A standardized number of HA units of each virus was tested against doubling dilutions of the antibodies. MAb BRIC220, which is specific for SCR1, the most membrane-distal domain of DAF, blocked HA by CAV21, but not by any of the other viruses tested. The SCR2-specific MAb BRIC110 inhibited HA by CBV3 and EV13. These two viruses, together with EV7 and EV12, were also blocked from binding DAF on erythrocytes by MAbs 854 and 2D2-1, which are specific for SCR3 and SCR4, respectively. These results suggest that, although there are qualitative differences between EV7, EV12, EV13 and CBV3 receptor binding, in particular the DBEI
HAI with sDAF. The HA reaction mediated by virus binding to DAF can be inhibited by the presence of sDAF or sub"#$% fragments of DAF that authentically mimic the virus binding site (Powell et al., 1997). We have extended our original studies of EV7 to map the binding sites of a range of HA enteroviruses using this approach and show the results obtained from a representative sample of genetically divergent serotypes in Fig. 3 and Table 1. The results obtained generally correlate well with those determined by HAI with anti-DAF MAbs. As previously reported, HA by EV7 is blocked by sDAF and "#$% sDAF , a pattern similar to that with EV6h and CBV3. In #$% addition to sDAF and sDAF , EV12 and EV29 are also "#$% #$% blocked by sDAF , implying that this two domain construct $% retains the ability to bind the virus, although the end-point titre is reduced in comparison with the larger receptor derivatives. The HAI pattern obtained with CAV21 is significantly different, and correlates less well with the data obtained with BRIC110 (Fig. 2), in that inhibition is only observed in the presence of sDAF ; the truncated sDAF "#$% derivatives that retain SCR1 do not block HA by this isolate. We have previously commented on the transitory nature of the HA phenotype displayed by certain EV serotypes which have a weak affinity for DAF (Powell et al., 1998), and consequently display weak HA reactions. This can be seen in Fig. 3 for EV29 (and to a lesser extent EV6h) in the presence of
DAF-binding enteroviruses
Fig. 4. Qualitative biosensor analysis of enterovirus–DAF binding. Approximately 10 000 RU of each of the four indicated serotypes was independently immobilized on the biosensor chip ; the interaction with truncated sDAF protein was monitored for 1200 s.
sDAF which display pinprick ‘ buttons ’ that are unrelated to "#$ sDAF concentration, and which can clearly be distinguished from HAI on the same plate. Virus neutralization by sDAF deletion mutants
We have previously demonstrated that sDAF , or "#$% derivatives that retain the ability to bind virus, are able to reversibly neutralize virus infection of susceptible cell lines by sterically preventing the virus from interacting with cellular DAF (Powell et al., 1997). We have investigated the ability of a range of enteroviruses to be neutralized by sDAF , or the "#$% truncated forms of the soluble receptor (see Table 1). In the case of EV6h, 6h’, 7, 12, 13, 29 and 33, the pattern of neutralization observed correlated precisely with the HAI analysis using sDAF or anti-DAF MAbs. EV6, EV24 and CBV1, which were not analysed by HAI, were all blocked by sDAF and sDAF , but not by the other truncated soluble "#$% #$% receptors. Direct measurement of SCR binding using a biosensor
Real time biological interactions can readily be detected by measuring changes in surface plasmon resonance ; the use of this technique for studying picornavirus–receptor interactions has recently been demonstrated for human rhinovirus and
ICAM-1 (Casasnovas & Springer, 1995) and EV11 binding to sDAF (Lea et al., 1998). We have used this approach at a qualitative level to investigate simple positive binding interactions for sDAF truncated receptors and EV6, 6h, 7 and 12. The approach used was essentially as described previously (Lea et al., 1998), with similar levels of virus immobilized on the sensor chip and the interaction with sDAF, or a non-specific protein (BSA), present in the soluble phase, determined. In each case, sample injection resulted in an increase in the signal (measured in RU) over the background level, reflecting bulk flow effects. Signals higher than those observed with the control BSA were considered specific. For example, EV7 generated a signal of 150 RU with 1 % BSA (data not shown), similar to that seen with sDAF or any of the two domain "#$ sDAF products (Fig. 4). In contrast, sDAF and sDAF "#$% #$% both generated significantly higher signals, indicative of a specific association between the virus and the receptor derivative, in line with the studies we have previously published (Lea et al., 1998). Using this approach with EV6, 6h, 7 and 12, results were obtained that are consistent with those demonstrated by HAI and\or soluble receptor-mediated virus neutralization. sDAF , sDAF and sDAF displayed no reactivity with any "#$ "# #$ of the viruses tested. As expected, all reacted with sDAF "#$% and, in line with other assays, sDAF . Of the viruses tested, #$% only EV12 reacted with sDAF .
%$Discussion Having initially been identified as an EV receptor (Bergelson et al., 1994 ; Ward et al., 1994), further studies have shown that DAF binding is widespread amongst the human enteroviruses, including at least 11 serotypes of EV, CBV1, 3 and 5, CAV21 and ENV70 (Karnauchow et al., 1996 ; Powell et al., 1998 ; Shafren et al., 1995, 1997 a). Proteins other than DAF may also be involved in virus binding or infection ; at least in the case of CBV3 and CAV21, there is compelling evidence that cell surface proteins including CAR or ICAM-1, respectively, are involved in cell infection (Shafren et al., 1997 a, b). Furthermore, additional cell surface proteins such as β-2 microglobulin (Ward et al., 1998) or the complement regulatory protein CD59 (I. Goodfellow & D. J. Evans, unpublished results), although probably not directly binding to the virus, are part of the virus–receptor complex that forms during cell infection. This situation is further complicated by the natural variation seen in the receptor specificity of primary isolates (Bergelson et al., 1997 b) and the apparent ease with which variant viruses can be selected that gain or lose the ability to interact with certain receptors. The HeLa cell receptor for ENV70 has been reported as DAF (Karnauchow et al., 1996), but the isolate available in our laboratory had been repeatedly passaged in RD cells and had lost the ability to haemagglutinate erythrocytes (data not shown), presumably due to the loss of the ability to bind DAF. Notwithstanding DBEJ
R. M. Powell and others
Table 2. Genetic clusters of human enteroviruses The human enteroviruses can be grouped into four clusters, of which the ‘ CAV16-like ’ A cluster contains no known DAF-binding viruses. The remaining three clusters are indicated, listing the viruses for which sequence data are available on which the groupings were made [based on the data of Po$ yry et al. (1996) and Pulli et al., (1995)]. Viruses known to bind DAF are underlined. Other receptor usage identified within each cluster is indicated. Viruses for which a receptor has yet to be identified are given in italics.
CBV-like (B-cluster) CAV type
9
CBV type ENV type EV type PV type Additional receptor usage
1, 3, 4, 5 69 1, 4, 6, 7, 11, 12, 27, 30 k CAR (CBV1, 3, 5) VLA-2 (EV1)
PV-like (C-cluster) 1, 11, 13, 15, 17, 18, 19, 20, 20a, 20b, 21, 22, 24 k k k 1–3 PVR (PV1–3) ICAM-1 (CAV13, 15, 18, 20, 21)
these points, the incidence of DAF binding amongst the human enteroviruses is striking. We are interested in the origin and evolution of DAF binding by the human enteroviruses. To this end, we have investigated the domains of DAF with which the DAF-binding haemagglutinating enteroviruses interact by HAI assay, virus neutralization and, for a limited number of viruses, by measuring the direct interaction of virus and receptor using a biosensor. We have previously demonstrated, using EV7, that HAI and virus neutralization using sDAF enables the identification of the receptor domains with which the virus interacts (Powell et al., 1997), and that these results correlate well with the observed binding to receptor variants expressed at the surface of transfected cells (Clarkson et al., 1995). The majority of viruses used in this study have been tested using at least two assays and the results obtained, summarized in Table 1, are generally in good agreement with alternative assays reported here and previous published studies. As expected, and irrespective of the type of assay used, all viruses interacted with sDAF , and the EV and CBV isolates also "#$% bound sDAF . These results are in agreement with our #$% previous studies of EV7, using sDAF and CHO cells expressing chimeric DAF receptors (Clarkson et al., 1995 ; Powell et al., 1997), in which we suggested that the integrity of the virus binding site on the receptor required the presence of the three SCR domains, even though blocking studies with MAb 854 (anti-SCR3), and subsequent biosensor analysis of EV11–DAF binding (Lea et al., 1998), suggested that SCR3 was the principal domain involved in the interaction. The HAI results using CBV3 are also in agreement with antibody inhibition of DAF binding in which SCR2 and SCR3 were implicated (Martino et al., 1998 ; Shafren et al., 1997 b). In contrast, DBFA
ENV70-like (D-cluster) k k k 68, 70 k
deletion of SCR2 alone abrogated binding to cell-surface expressed DAF by an RD-adapted CBV3 (CBV3-RD) (Bergelson et al., 1995), suggesting that there are differences in the specificity of binding within this group of viruses. This conclusion could also be drawn from the differential reactivity of sDAF to the viruses tested. EV12 and 29 alone exhibited $% binding to the two membrane-proximal SCR domains of sDAF in HAI assays (Fig. 3), a result in agreement with the antibodymediated HAI and biosensor studies on EV12 (Figs 2 and 4). CAV21 was efficiently blocked in an HAI assay by BRIC220 (Fig. 2), an SCR1-specific MAb, in agreement with previous domain mapping studies for this virus (Shafren et al., 1997 a). HAI assays using sDAF exhibited no inhibition of "# CAV21, sDAF showed only very weak inhibition and "#$ sDAF blocked HA significantly less well than for any of the "#$% other viruses tested (Fig. 3). The P. pastoris-expressed sDAF proteins used in these studies fold in an authentic manner, as confirmed by MAb binding analysis (Daniels et al., 1997), inhibition of complement fixation assays (I. Goodfellow & B. Sim, unpublished data) and preliminary NMR structural analysis (data not shown). However, the yeast-expressed sDAF differs from the cellular protein in that it is only "#$% partially glycosylated (approx. 70 % by SDS–PAGE ; data not shown) at the single N-linked site between SCR1 and SCR2 (Asn ), which was also removed during the construction of '" plasmids encoding sDAF and sDAF . We therefore in"# "#$ terpret the CAV21 HAI results as indicating that the nature or completeness of the glycosylation of sDAF , or the lack of "#$% carbohydrate on sDAF or sDAF may affect CAV21–DAF "#$ "# interactions. Although there is a precedent for this interpretation – ENV70, a virus that binds SCR1 (Karnauchow et al., 1998), exhibits a neuraminidase-sensitive HA phenotype
DAF-binding enteroviruses
(Utagawa et al., 1982) suggesting that sialic acid and N-linked carbohydrate may be involved in the binding – further studies are required to determine whether the differences noted in CAV21 binding reflect a direct interaction with the carbohydrate moiety, or if glycosylation may be required for the correct presentation of a binding site occupying the Nterminus of DAF. How has DAF binding evolved to be such a widespread characteristic of the human enteroviruses ? One possibility, which these studies do not rule out, is that the ancestral enterovirus possessed this characteristic which has subsequently been retained by many representatives due to selective pressure. Alternatively, the ability to bind DAF may have evolved multiple times within the human enteroviruses. The latter conclusion is supported by distribution of DAF-binding isolates amongst three of the four clusters of human enteroviruses identified in phylogenetic studies (see Table 2 ; Po$ yry et al., 1996 ; Pulli et al., 1995) : CAV21 in the ‘ PV-like ’ C-cluster, ENV70 in the ‘ ENV70-like ’ D-cluster, and the remainder (CBV1, 3 and 5, EV6, 7, 11, 12 and 30) in the ‘ CBV-like ’ Bcluster. In the absence of sequence data from the haemagglutinating EV omitted from Table 2, for which the DAFbinding specificity is known (EV6h, 6h’, 13, 19, 24, 29 and 33 ; see Table 1), it seems highly likely that these serotypes would also occupy the ‘ CBV-like ’ B-cluster. It is striking therefore, that the two broad classes of DAF interaction, with SCR1 (CAV21, ENV70) or SCR2–4 (the rest), are located in distinct clusters. Furthermore, the separate clustering of CAV21 and ENV70 (Table 2), and the presumed rhinovirus\poliovirus recombinant origin of CAV21 (Hughes et al., 1989), suggest that the ability to bind SCR1 may have evolved independently in these two viruses. Further studies are under way to determine whether the differences in binding catalogued between representatives of the ‘ CBV-like ’ cluster, for example the ability to bind sDAF by EV12 and 29, represent subtle $% variation within the group, or the independent evolution of a distinct binding footprint. These interpretations make the assumption that DAF binding is a physiologically relevant phenotype that has evolved because it offers a selective advantage to the virus within the host, which seems reasonable in light of the DAF binding observed in cardiovirulent coxsackieviruses and clinical isolates (Bergelson et al., 1997 b ; Martino et al., 1998). The nature of this advantage is the subject of further study. We acknowledge Brian Megson and Geoff Daniels for reagents, Mike Flint for critical comments on this manuscript and Brad Spiller for useful discussion. This study was supported by MRC Programme Grant No. G9006199.
References Bergelson, J. M., Shepley, M. P., Chan, B. M., Hemler, M. E. & Finberg, R. W. (1992). Identification of the integrin VLA-2 as a receptor for
echovirus 1. Science 255, 1718–1720.
Bergelson, J. M., Chan, M., Solomon, K. R., St. John, N. F., Lin, H. & Finberg, R. W. (1994). Decay-accelerating factor (CD55), a glycosyl-
phosphatidylinositol-anchored complement regulatory protein, is a receptor for several echoviruses. Proceedings of the National Academy of Sciences, USA 91, 6245–6249. Bergelson, J. M., Mohanty, J. G., Crowell, R. L., St. John, N. F., Lublin, D. M. & Finberg, R. W. (1995). Coxsackievirus B3 adapted to growth in
RD cells binds to decay-accelerating factor (CD55). Journal of Virology 69, 1903–1906. Bergelson, J. M., Cunningham, J. A., Droguett, G., Kurt-Jones, E. A., Krithivas, A., Hong, J. S., Horwitz, M. S., Crowell, R. L. & Finberg, R. W. (1997a). Isolation of a common receptor for coxsackie B viruses
and adenoviruses 2 and 5. Science 275, 1320–1323. Bergelson, J. M., Modlin, J. F., Wieland-Alter, W., Cunningham, J. A., Crowell, R. L. & Finberg, R. W. (1997b). Clinical coxsackievirus B
isolates differ from laboratory strains in their interaction with two cell surface receptors. Journal of Infectious Diseases 175, 697–700. Casasnovas, J. M. & Springer, T. A. (1995). Kinetics and thermodynamics of virus binding to receptor – studies with rhinovirus, intercellular-adhesion molecule-1 (Icam-1), and surface-plasmon resonance. Journal of Biological Chemistry 270, 13216–13224. Clarkson, N. A., Kaufman, R., Lublin, D. M., Ward, T., Pipkin, P. A., Minor, P. D., Evans, D. J. & Almond, J. W. (1995). Characterization of
the echovirus-7 receptor – domains of CD55 critical for virus binding. Journal of Virology 69, 5497–5501. Colonno, R. J. (1986). Cell surface receptors for picornaviruses. Bioessays 5, 270–274. Coyne, K. E., Hall, S. E., Thompson, E. S., Acre, M. A., Kinoshita, T., Fujita, T., Anstee, D. J., Rosse, W. & Lublin, D. M. (1992). Mapping of
epitopes, glycosylation sites, and complement regulatory domains in human decay accelerating factor. Journal of Immunology 149, 2906–2913. Daniels, G. L., Green, C. A. & Powell, M. R. (1997). Cromer antibodies : analysis with soluble recombinant decay-accelerating factor (srDAF) and the monoclonal antibody immobilization of erythrocyte antigens assay (MAIEA). Transfusion 37, S351. Daniels, G. L., Green, C. A., Powell, R. M. & Ward, T. (1998).
Hemagglutination inhibition of Cromer blood group antibodies with soluble recombinant decay-accelerating factor. Transfusion 38, 332–336. Evans, D. & Almond, J. (1998). Cell receptors for picornaviruses as determinants of cell tropism and pathogenesis. Trends in Microbiology 6, 198–202. Hamann, J., Vogel, B., van Schijndel, G. M. & van Lier, R. A. (1996).
The seven-span transmembrane receptor CD97 has a cellular ligand (CD55, DAF). Journal of Experimental Medicine 184, 1185–1189. Hughes, P. J., North, C., Minor, P. D. & Stanway, G. (1989). The complete nucleotide sequence of coxsackievirus A21. Journal of General Virology 70, 2943–2952. Karnauchow, T. M., Tolson, D. L., Harrison, B. A., Altman, E., Lublin, D. M. & Dimock, K. (1996). The HeLa cell receptor for enterovirus 70
is decay accelerating factor (CD55). Journal of Virology 70, 5143–5152. Karnauchow, T. M., Dawe, S., Lublin, D. M. & Dimock, K. (1998). Short consensus repeat domain 1 of decay-accelerating factor is required for enterovirus 70 binding. Journal of Virology 72, 9380–9383. Lea, S., Powell, R., McKee, T., Evans, D., Brown, D., Stuart, D. & van der Merwe, P. (1998). Determination of the affinity and kinetic
constants for the interaction between the human echovirus 11 and its cellular receptor CD55. Journal of Biological Chemistry 273, 30443–30447. Lublin, D. M. & Atkinson, J. P. (1989). Decay-accelerating factor : biochemistry, molecular biology, and function. Annual Review of Immunology 7, 35–58. DBFB
R. M. Powell and others Martino, T. A., Petric, M., Brown, M., Aitken, K., Gauntt, C. J., Richardson, C. D., Chow, L. H. & Liu, P. P. (1998). Cardiovirulent
coxsackieviruses and the decay-accelerating factor (CD55) receptor. Virology 244, 302–314. Melnick, J. L. (1996). Enteroviruses : polioviruses, coxsackieviruses, echoviruses, and newer enteroviruses. In Fields Virology, pp. 655–712. Edited by B. N. Fields, D. M. Knipe & P. M. Howley. Philadelphia : Lippincott–Raven. Mendelsohn, C. L., Wimmer, E. & Racaniello, V. R. (1989). Cellular receptor for poliovirus : molecular cloning, nucleotide sequence, and expression of a new member of the immunoglobulin superfamily. Cell 56, 855–865. Powell, R. M., Ward, T., Evans, D. J. & Almond, J. W. (1997). Interaction between echovirus 7 and its receptor, decay-accelerating factor (CD55) : evidence for a secondary cellular factor in A-particle formation. Journal of Virology 71, 9306–9312. Powell, R. M., Schmitt, V., Ward, T., Goodfellow, I., Evans, D. J. & Almond, J. W. (1998). Characterization of echoviruses that bind decay
accelerating factor (CD55) : evidence that some haemagglutinating strains use more than one cellular receptor. Journal of General Virology 79, 1707–1713. Po$ yry, T., Kinnunen, L., Hyypia$ , T., Brown, B., Horsnell, C., Hovi, T. & Stanway, G. (1996). Genetic and phylogenetic clustering of entero-
viruses. Journal of General Virology 77, 1699–1717. Pulli, T., Koskimies, P. & Hyypia$ , T. (1995). Molecular comparison of coxsackie A virus serotypes. Virology 212, 30–38. Reagan, K. J., Goldberg, B. & Crowell, R. L. (1984). Altered receptor
DBFC
specificity of coxsackievirus B3 after growth in rhabdomyosarcoma cells. Journal of Virology 49, 635–640. Shafren, D. R., Bates, R. C., Agrez, M. V., Herd, R. L., Burns, G. F. & Barry, R. D. (1995). Coxsackieviruses B1, B3, and B5 use decay
accelerating factor as a receptor for cell attachment. Journal of Virology 69, 3873–3877. Shafren, D. R., Dorahy, D. J., Ingham, R. A., Burns, G. F. & Barry, R. D. (1997a). Coxsackievirus A21 binds to decay-accelerating factor but
requires intercellular adhesion molecule 1 for cell entry. Journal of Virology 71, 4736–4743. Shafren, D. R., Williams, D. T. & Barry, R. D. (1997b). A decayaccelerating factor-binding strain of coxsackievirus B3 requires the coxsackievirus–adenovirus receptor protein to mediate lytic infection of rhabdomyosarcoma cells. Journal of Virology 71, 9844–9848. Utagawa, E. T., Miyamura, K., Mukoyama, A. & Kono, R. (1982).
Neuraminidase-sensitive erythrocyte receptor for enterovirus type 70. Journal of General Virology 63, 141–148. Ward, T., Pipkin, P. A., Clarkson, N. A., Stone, D. M., Minor, P. D. & Almond, J. W. (1994). Decay accelerating factor (CD55) identified as the
receptor for echovirus 7 using CELICS, a rapid immuno-focal cloning method. EMBO Journal 13, 5070–5074. Ward, T., Powell, R. M., Pipkin, P. A., Evans, D. J., Minor, P. D. & Almond, J. W. (1998). Role for β2-microglobulin in echovirus infection
of rhabdomyosarcoma cells. Journal of Virology 72, 5360–5365.
Received 26 May 1999 ; Accepted 7 September 1999
