Proc. Nati. Acad. Sci. USA Vol. 91, pp. 6245-6248, June 1994 Medical Sciences Decay-accelerating factor (CD55), a glycosylphosphatidylinositol- anchored complement regulatory , is a receptor for several echoviruses ( receptor/picornavirus/echovirus 7) JEFFREY M. BERGELSON*, MELVIN CHAN, KEITH R. SOLOMON, NICOLE F. ST. JOHN, HUAMAO LIN, AND ROBERT W. FINBERG Laboratory of Infectious Diseases, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA 02115 Communicated by Baruj Benacerraf, March 29, 1994

ABSTRACT Echoviruses are human pathogens belonging autologous complement (4). Deficient expression ofDAF and to the picornavirus family. Decay-accelerating factor (DAF) is other GPI-anchored complement regulatory results a glycosylphosphatidylhinostol (GPI)-anchored surface protein in the hematologic disorder paroxysmal nocturnal hemoglo- that protects cells from lysis by autologous complement. Anti- binuria. In this report, we show that DAF also mediates cell DAF monoclonal antibodies prevented echovirus 7 attachment surface attachment and infection by several echovirus sero- to susceptible cells and protected cells from infection. HeLa types. cells specifically lost the capacity to bind echovirus 7 when treated with phosphatidylinositol-specific phospholipase C, an MATERIALS AND METHODS enzyme that releases GPI-anchored proteins from the cell surface, indicating that the virus receptor, like DAF, is a and Virus Assays. Echoviruses 1 (Farouk), 6 GPI-anchored protein. Although Chinese hamster ovary cells (D'Amori), 7 (Wallace), 8 (Bryson), 11 (Gregory), 12 (Travis do not bind echovirus 7, transfectants expressing human DAF 2-85), 20 (JV-1), and 21 (Farina) were obtained from the bound virus efficiently, and binding was prevented by pre- American Type Culture Collection and grown in HeLa cells. treatment with an anti-DAF monoclonal antibody. Anti-DAF Antibody inhibition ofviral cytopathic effect and inhibition of antibodies prevented infection by at least six echovirus sero- plaque formation were measured as described (2). ypes. These results indicate that DAF is the receptor mediating Monoclonal Antibodies (mAbs). To obtain mAb IF7, attachment and infection by several echoviruses. BALB/c mice were immunized with live HeLa cells, their splenocytes were fused to P3x63.Ag8U.1 myeloma cells, and Despite their importance in the pathogenesis of viral illness, hybridomas were selected in hypoxanthine/aminopterin/ specific receptors have been identified for only a few of the thymidine medium. Culture supernatants were assessed for viruses that cause human disease. Among the echoviruses, their ability to protect HeLa cell monolayers from infection nonenveloped RNA viruses responsible for febrile illnesses by echovirus 7 using a colorimetric assay as described for including aseptic meningitis, echoviruses 1 and 8 initiate echovirus 1 (1). Anti-DAF mAbs 1H4, 8D11, and 11D7 (5) infection by attaching to the integrin VLA-2 (1, 2). However, were obtained from Wendell Rosse (Duke University), and these two closely related serotypes account for only a small IA10 (6) was from Victor Nussenzweig (New York Univer- percentage of echovirus infections, and anti-VLA-2 antibod- sity). The anti-DAF antibody MEM 118 and the anti-CD-59 ies have no effect on infection by many of the 30 echovirus mAb MEM 125 were from the 5th International Conference serotypes. Similarly, although echoviruses 1 and 8 bind to on Human Leukocyte Differentiation Antigens (November and infect Chinese hamster ovary (CHO) cells transfected 3-7, 1993, Boston). The anti-5'-nucleotidase (CD73) mAb with human VLA-2, otherechoviruses do not, suggesting that 1E9 was provided by Linda Thompson (Oklahoma Medical other echoviruses may bind other receptors (1, 2). Foundation). mAb D2 recognizes CDw109 (7), and DE9 Most cell surface proteins are anchored in the cell mem- recognizes the integrin f31 subunit (CD29) (1). brane by a hydrophobic transmembrane peptide domain. Immunoprecipitation and Immunodepletion. lodination of Some, however, lack a classic transmembrane domain and HeLa cells by the lactoperoxidase/glucose oxidase method, are instead anchored to the membrane by a fatty acid tail, extraction in buffer containing 1% Nonidet P-40, and precip- attached to the protein C terminus by a glycosylphosphati- itation with mAbs and rabbit-anti-mouse immunoglobulin- dylinositol (GPI) linkage (reviewed in ref. 3). Studies ofthese coated protein A-Sepharose beads were performed essen- proteins have been facilitated by identification ofthe enzyme tially as described (1). In some experiments, cell extracts phosphatidylinositol-specific phospholipase C (PI-PLC), were immunodepleted three times with mAb IA10 or a which, by cleaving the anchor, releases GPI-linked proteins control mAb before final immunoprecipitation. Because we from the cell surface. Since the first report of the GPI-linked found that mAb IF7 failed to bind cells in the absence of trypanosomal variant surface glycoprotein, an increasing divalent cations, all procedures were performed in buffers number of GPI-anchored mammalian proteins have been containing 2 mM CaCl2 and 2 mM MgCl2, and EDTA and shown to function in cell adhesion, enzymatic processes, EGTA were omitted. small molecule transport, and transmembrane signaling. No Radiolabeled Virus Binding Assays. Viruses were radiola- identified virus receptor has been shown to possess a GPI beled by growth in medium containing [35S]methionine and anchor. purified by velocity sedimentation in sucrose gradients as Decay-accelerating factor (DAF; CD55) is a GPI-anchored described (2). HeLa cell monolayers in 24-well plates or glycoprotein that functions in protecting cells from lysis by Abbreviations: DAF, decay-accelerating factor; GPI, glycosylphos- The publication costs of this article were defrayed in part by page charge phatidylinositol; PI-PLC, phosphatidylinositol-specific phospholi- payment. This article must therefore be hereby marked "advertisement" pase C; mAb, monoclonal antibody. in accordance with 18 U.S.C. §1734 solely to indicate this fact. *To whom reprint requests should be addressed.

6245 Downloaded by guest on September 26, 2021 6246 Medical Sciences: Bergelson et al. Proc. Natl. Acad. Sci. USA 91 (1994) monolayers of CHO cells transfected with human DAF (8) A B C (provided by Douglas M. Lublin, Washington University, St. Louis) or CHO cells transfected with the a2 subunit of VLA-2 (2) were preincubated with mAbs and then washed 100- 4W- and exposed to radiolabeled echoviruses (10,000-30,000 cpm) for 1 hr at room temperature. Monolayers were then 70- - washed and dissolved for determination of cell-bound radio- activity as described (9). Purified antibodies and hybridoma 49- - supernatants were used at 10-20 gg/ml and ascites fluids were diluted 1:100. 1 2 9 10 11 12 For measurements of virus binding after PI-PLC treat- 3 4 5 6 7 8 13 ment, HeLa cells were dispersed with EDTA, suspended in FIG. 1. Immunoprecipitation and immunodepletion with IF7 and PI-PLC buffer (RPMI 1640 medium/0.2% bovine serum anti-DAF mAb IA10. Surface-iodinated HeLa cells were extracted albumin/50 tkM 2-mercaptoethanol/l0 mM Hepes/0.1% with 1% Nonidet P-40 and precipitation was performed as described. azide), and divided in two portions, one of which received (A) Immunoprecipitation. Lanes: 1, control mAb MOPC 195; 2, IF7; PI-PLC (0.5 unit per 106 cells; Oxford Glycosystems, 3, IA10. (B) Immunoprecipitation after immunodepletion with mAb Rosedale, NY). Both treated and mock-treated cells were D2 (anti-CDwlO9), which recognizes 180- and 150-kDa GPI- anchored proteins on HeLa cells. Lanes: 4, MOPC 195; 5, IA10; 6, incubated for 1 hr at 370C and then washed and dispensed into IF7; 7, D2; 8, DE9, which recognizes the 130-kDa integrin f31 Eppendorf tubes before addition of radiolabeled virus. subunit. (C) Immunoprecipitation after immunodepletion with anti- DAF mAb IA10. Lanes: 9, MOPC 195; 10, IF7; 11, IA10; 12, D2; 13, RESULTS DE9. Numbers on left are kDa. A mAb That Prevents Cell Attachment by Echovirus 7. To HeLa cells (Fig. 1, lane 2). When HeLa cells were incubated identify potential receptors for echovirus 7, we took an with PI-PLC, IF7 expression detected by indirect immuno- approach that was successful for identification of VLA-2 as fluorescence decreased dramatically, indicating that IF7 rec- the receptor for echovirus 1 (1). Mice were immunized with ognized a GPI-anchored protein (data not shown). We there- HeLa cells and their splenocytes were fused to myeloma cells fore suspected that IF7 might recognize DAF, a 70-kDa to generate hybridomas. Hybridoma supernatants were then GPI-anchored protein highly expressed on HeLa cells. IF7 screened for the capacity to protect susceptible cells from and the well-characterized anti-DAF antibody IAl0 (6) rec- infection by echovirus 7. Three fusions were screened, and ognized proteins of identical mobility (Fig. lA). Immunode- one protective antibody, mAb IF7, was obtained. pletion with a control antibody recognizing CDw1O9, another mAb IF7 Recognizes DAF. IF7 immunoprecipitated a 70- GPI-anchored protein expressed on HeLa cells, did not affect kDa protein from detergent extracts of surface-iodinated immunoprecipitation of the 70-kDa proteins by IF7 or IAl0 A B Preincubated with: 8- * ControlAb -o E2 IF7 4000 c 2000- .8

- a 1000 - 2000 -o-i CZ z o _0 __Echo Eco- a: 0- ---I 0 - Echo 7 Echo 1 Echo 6 Echo 1

C anti- VLA-2 anti- 5'-NT anti- CD59 IAl 0 MEM 118 -T 8D11 11D7 anti-DAF MAbs 1 H4 IF7 No MAb 0 4000 8000 Radiolabeled virus bound (cprm)

FIG. 2. Inhibition of radiolabeled virus binding by mAb IF7 and other anti-DAF antibodies. (A) IF7 inhibits attachment of echovirus 7. (B) IF7 inhibits attachment of echovirus 6. (C) Other anti-DAF antibodies inhibit attachment of echovirus 7. HeLa cell monolayers in 24-well tissue culture plates were incubated with mAbs for 1 hr at room temperature and then antibody was removed and radiolabeled viruses were added (10,000-20,000 cpm). After 1 hr, monolayers were washed and dissolved for determination of cell-bound virus by scintillation counting. Detailed conditions for assays have been described (9). Results are shown as mean radiolabeled virus bound + SD for triplicate monolayers. Downloaded by guest on September 26, 2021 Medical Sciences: Bergelson et al. Proc. Natl. Acad. Sci. USA 91 (1994) 6247

(Fig. 1B). In contrast, immunodepletion with IA10 removed I-I all protein precipitable by IF7 (Fig. 1C). These results E 6000- C.) indicate that IF7 recognizes DAF. CL * Control MAb mAb IF7 prevented attachment of echoviruses 6 and 7 to 0 0 MAb IF7 HeLa cells (Fig. 2 A and B) with no effect on attachment of -o echovirus 1, which is known to bind the integrin VLA-2 (1). Other anti-DAF mAbs also inhibited virus binding (Fig. 2C). Ul) mAbs 8D11 and 11D7 partially inhibited attachment of echo- en virus 7, while 1H4 blocked binding as effectively as IF7. One U1) anti-DAF mAb, IA10, had no effect on virus binding, sug- .5 gesting that this antibody may recognize an epitope distant from the virus binding site. Virus binding was not inhibited by CU the anti-VLA-2 mAb AA10 or by antibodies to other GPI- cc un- anchored proteins expressed on HeLa cells [anti-CD59 (Fig. CHO-DAF CHO-al 2 2C; five other mAbs were tested and are not shown), anti- 5'-nucleotidase (Fig. 2C), and anti-CDw1O9 (data not FIG. 4. Virus binding to CHO cell transfectants expressing shown)]. Like IF7, mAb 1H4 protected cells from virus human DAF. CHO cells transfected with human DAF or with the a2 subunit of VLA-2 were pretreated with anti-DAF mAb IF7 or with infection. Preincubation of HeLa monolayers with 1H4 in- a control mAb and then exposed to radiolabeled echovirus 7. Results hibited plaque formation by echoviruses 6 and 7 by >99%, are shown as mean radioactive virus bound ± SD for triplicate with no effect on plaque formation by echovirus 1 (data not monolayers. shown). DAF Is a Receptor for Echovirus 7. The observation that scribed (2) to measure IF7 effects on viruses that did not form anti-DAF antibodies prevented virus attachment and infec- countable plaques in HeLa cells. IF7 significantly inhibited tion suggested that DAF was a cellular receptor for echovirus cytopathic effects caused by echoviruses 11 and 20, indicat- 7. Since PI-PLC releases DAF from the HeLa cell surface, ing that these viruses also interact with the DAF receptor we predicted that HeLa cells treated with this enzyme would (data not shown). Pretreatment ofHeLa cell monolayers with lose the capacity to bind virus. Whereas little or no echovirus IF7 reduced subsequent plaque formation by echoviruses 6, 7 bound to PI-PLC-treated cells, virus bound efficiently to 7, 12, and 21 with no inhibition of plaque formation by HeLa cells incubated in buffer without PI-PLC (Fig. 3). echoviruses 1 and 8 (Fig. 5). PI-PLC treatment did not prevent attachment of echovirus 1. This demonstrated that the receptor for echovirus 7, like DAF, was a GPI-anchored protein. DISCUSSION Because CHO cells do not bind echovirus 7 (2), we used The data presented here demonstrate that DAF (CD55) is a CHO cells transfected with human DAF to demonstrate that receptor for at least six echovirus serotypes. Anti-DAF mAbs DAF is directly responsible for virus binding. Radiolabeled block virus attachment and protect cells from infection, and echovirus 7 bound to CHO cells transfected with human DAF rodent cells transfected with human DAF gain the capacity to cDNA but did not bind to control transfectants (]Fig. 4). bind virus. Echovirus 1 did not bind to DAF transfectants (data not We previously found that an integrin molecule, VLA-2, shown). Anti-DAF mAb IF7 prevented echovirus 7 attach- serves as the receptor for echoviruses 1 and 8, two closely ment to DAF transfectants (Fig. 4), confirming that DAF related echovirus serotypes (2). Some echovirus serotypes mediated virus interaction with these cells. are not inhibited by either anti-DAF or anti-VLA-2 antibodies Anti-DAF Antibodies Prevent Infection by Other Echovirus (unpublished observation), suggesting that at least one other Serotypes. mAb IF7 protected cells from infection not only by echovirus 7 but also by other echoviruses. We used a quantitative assay of virus-induced cytopathic effects de- Echo 1 E 7000- E Echo 8 0. C.) Echo 6 -D * Mock-treated C 0 U PIPLC-treated .0 Echo 7 Echo 12

a) ._5U)o0 Echo 21 (n 7r- i 0 -40 -20 0 20 40 60 80 100 -0 CZ Inhibition of plaque formation (%)

Echo 7 Echo 1 FIG. 5. Inhibition of plaque formation by mAb IF7. HeLa cell monolayers were incubated with IF7 supernatant or with a control FIG. 3. Radiolabeled virus binding to HeLa cells after PI-PLC antibody for 0.5 hr at room temperature and then antibody was treatment. HeLa cells were incubated with PI-PLC or in buffer alone. removed and each virus was added in minimal essential medium/5% Cells were then washed in virus binding buffer, distributed into calf serum. After 1 hr at room temperature, wells were overlaid with Eppendorf tubes (5 x 105 cells per tube), pelleted gently, and 0.8% agar plaquing medium and plaques were allowed to develop at resuspended in 0.1 ml of radiolabeled echovirus 7 or echovirus 1 37°C for 40-44 hr as described (2). Results are shown as percentage (-20,000 cpm) in virus binding buffer plus 0.1% azide. After 1 hr at inhibition ± 95% confidence intervals, calculated for results from room temperature, cells were washed and dissolved for scintillation triplicate monolayers. Percentage inhibition = [(plaques per control counting. Results are shown as mean radioactive virus bound per 5 monolayer - plaques per IF7-treated monolayer)/plaques per con- x 105 cells ± SD for triplicate samples. trol monolayer] x 100. Downloaded by guest on September 26, 2021 6248 Medical Sciences: Bergelson et al. Proc. Natl. Acad Sci. USA 91 (1994) echovirus receptor molecule remains to be identified. ing DAF) participate in transmembrane signaling and are Whether the receptors that mediate virus attachment to physically associated with src family tyrosine kinases (26- cultured cells are determinants of tissue tropism in vivo 28). Caveolae also contain an inositol trisphosphate-sensitive remains an open question. The prototype echovirus strains calcium channel (29). Virus interaction with DAF may result used in our experiments were isolated in the 1950s and are in transmission of signals that play a role in the virus life cycle adapted to growth in HeLa cells. It will be of great interest or in virus-induced cell damage. to study the effects ofanti-DAF antibodies on freshly isolated viruses. We thank Dr. Douglas Lublin for generously providing us with Based on its nucleotide sequence, DAF is predicted to CHO cells transfected with human DAF and DAF deletion mutants, have five extracellular domains (10, 11). Close to the mem- and we are grateful to Drs. Victor Nussenzweig, Wendell Rosse, and brane is a serine/tyrosine-rich domain, which is the site of Linda Thompson for mAbs. We thank Ms. Claudette Thompson for help with statistical calculations. This work was supported by a grant extensive O-linked glycosylation. There are also four short from the National Institutes of Health consensus repeat (SCR) domains typical of complement (AI31628). regulatory proteins. Other investigators used deletion mu- 1. Bergelson, J. M., Shepley, M. P., Chan, B. M. C., Hemler, tants to show that mAb 1H4 recognizes an epitope within M. E. & Finberg, R. W. (1992) Science 255, 1718-1720. SCR3 (5), and our unpublished data indicate that mAb IF7 2. Bergelson, J., St. John, N., Kawaguchi, S., Chan, M., Stubdal, recognizes SCR2. Because these two antibodies were most H., Modlin, J. & Finberg, R. W. (1993) J. Virol. 67, 6847-6852. effective in blocking virus attachment and infection, they 3. Cross, G. (1990) Annu. Rev. Cell Biol. 6, 1-39. may identify regions ofthe molecule close to the virus binding 4. Lublin, D. & Atkinson, J. (1989) Annu. Rev. Immunol. 7, site. 35-57. Identification of DAF as an echovirus 5. Coyne, K., Hall, S., Thompson, E., Arce, M., Kinoshita, T., receptor may clarify Fujita, T., Anstee, D., Rosse, W. & Lublin, D. (1992) J. two earlier observations regarding echovirus-cell interac- Immunol. 149, 2906-2913. tions. Initial characterization of the echoviruses showed that 6. Kinoshita, T., Medof, M., Silber, R. & Nussenzweig, V. (1985) certain serotypes-including all those reported here to be J. Exp. Med. 162, 75-92. inhibited by anti-DAF antibodies, but not serotypes 1 and 7. Haregewoin, A., Solomon, K., Hom, R., Soman, G., Bergel- 8-are capable of agglutinating human erythrocytes (12, 13). son, J., Bhan, A. & Finberg, R. (1994) Cell. Immunol., in press. DAF was originally isolated from erythrocytes (14). The 8. Lublin, D. & Coyne, K. (1991) J. Exp. Med. 174, 35-44. relation between the putative erythrocyte virus receptor (15) 9. Bergelson, J. M., Chan, B. M. C., Finberg, R. W. & Hemler, and the receptor present on nucleated cells is uncertain, and M. E. (1993) J. Clin. Invest. 92, 232-239. it will be to learn whether anti-DAF antibodies 10. Caras, I., Davitz, M., Rhee, L., Weddell, G., Martin D. J. & interesting Nussenzweig, V. (1987) Nature (London) 325, 545-549. prevent virus-induced hemagglutination. 11. Medof, M., Lublin, D., Holers, V., Ayers, D., Getty, R., In other experiments 20 years ago, a protein factor present Leykam, J., Atkinson, J. & Tykocinski, M. (1987) Proc. Natl. in the supernatants ofhighly permissive human cell lines was Acad. Sci. USA 84, 2007-2011. shown to enhance infection of less susceptible cells by 12. Hsiung, G.-D. (1962) Ann. N.Y. Acad. Sci. 101, 413-422. echovirus 6 (16) and bind directly to virus (17). In this regard, 13. Goldfield, M., Srihongse, S. & Fox, J. (1957) Proc. Soc. Exp. it is interesting that DAF is present in HeLa cell supernatants Biol. Med. 96, 788-791. (18) and that purified DAF can be reincorporated in cell 14. Nicholson-Weller, A., Burge, J., Fearon, D., Weller, P. & membranes to correct the erythrocyte defect in paroxysmal Austen, K. (1982) J. lImmunol. 129, 184-189. nocturnal hemoglobinuria (19). data that 15. Philipson, L., Bengtsson, S., Brishammar, S., Svennerholm, L. Preliminary suggest & Zetterqvist, 0. (1964) Virology 22, 580-590. soluble DAF binds radiolabeled echovirus 7. It should be 16. Righthand, F. & Karzon, D. (1967) Proc. Soc. Exp. Biol. Med. possible to determine whether the previously reported en- 124, 1248-1254. hancing factor is a soluble form of DAF. 17. Righthand, F. (1972) Infect. Immun. 6, 805-809. Recent work suggests that many GPI-anchored proteins 18. Medof, M., Walter, E., Rutgers, J., Knowles, D. & Nussen- within the cell are densely clustered in caveolae, small zweig, V. (1987) J. Exp. Med. 165, 848-864. invaginations of the cell membrane visible by electron mi- 19. Medof, M., Kinoshita, T., Silber, R. & Nussenzweig, V. (1985) croscopy (20). Caveolae are not clathrin coated and are Proc. Natl. Acad. Sci. USA 82, 2980-2984. marked by the presence of a 22-kDa integral membrane 20. Anderson, R. G. W. (1993) Curr. Opin. Cell Biol. 5, 647-652. protein called caveolin (21). Based on studies of the GPI- 21. Rothberg, K., Heuser, J., Donzell, W., Ying, Y.-S., Glenney, J. & Anderson, R. (1992) Cell 68, 673-682. anchored receptor for 5-methyltetrahydrofolate, caveolae 22. Anderson, R., Kamen, B., Rothberg, K. & Lacey, S. (1992) have been proposed to be the sites for cellular uptake ofsmall Science 255, 410-411. molecules by a process distinct from classic receptor- 23. Rothberg, K., Ying, Y., Kolhouse, J., Kamen, B. & Anderson, mediated endocytosis (22). Caveolae have not been seen to R. (1990) J. Cell Biol. 110, 637-649. pinch off and dissociate from the cell surface, and the folate 24. Keller, G.-A., Siegel, M. & Caras, I. W. (1992) EMBO J. 11, receptor does not enter an endosomal compartment (23). 863-874. Although some GPI-anchored proteins, including DAF, may 25. Tausk, F., Fey M. & Gigli, I. (1989) J. Immunol. 143, 3295- be internalized, they appear to be excluded from clathrin- 3302. coated pits, and internalization occurs only in 26. Sargiacomo, M., Sudol, M. & Lisanti, M. (1993) J. Cell Biol. noncoated 122, 789-807. vesicles (24, 25). The identification of a GPI-anchored virus 27. Stefanova, I., Horejsi, V., Ansotegui, I., Knapp, W. & Stock- receptor suggests that some viruses may enter cells by an inger, H. (1991) Science 254, 1016-1019. unusual pathway. 28. Shenoy-Scaria, A., Kwong, J., Fujita, T., Oloszowy, M., Caveolae have been proposed as sites for transmembrane Shaw, A. & Lublin, D. (1992) J. Immunol. 149, 3535-3541. (20, 26). GPI-anchored proteins (includ- 29. Fujimoto, T. (1993) J. Cell Biol. 120, 1147-1157. Downloaded by guest on September 26, 2021