Human Mannose-binding Protein Functions as an Opsonin for Influenza A Viruses Kevan L. Hartshorn, * Kedamath Sastry, * Mitchell R. White, * E. Margot Anders, Michael Super,t R. Alan Ezekowitz,t and Alfred 1. Tauber* *Departments ofMedicine and Pathology, Boston University School ofMedicine and the Department ofMedicine, Boston City Hospital, Boston, Massachusetts 02118; Divisions oftHematology/Oncology and Infectious Diseases, Department ofPediatrics, Children's Hospital and Dana Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts 02115; and 1Department ofMicrobiology, University ofMelbourne, Victoria, Australia Abstract sponses (1). However, frequent mutations occur in HA genes such that antigenic determinants of the virus are altered, ren- Influenza A viruses (IAVs) cause substantial morbidity and dering previously immune individuals susceptible to repeat in- mortality in yearly epidemics, which result from the ability of fection (2). These changes result in frequent epidemics and, the virus to alter the antigenicity of its envelope proteins. De- occasionally, pandemics of influenza. Although considerable spite the rapid replication of this virus and its ability to infect a morbidity and mortality may be associated with these out- wide variety of cell types, viremia is rare and the infection is breaks, there is heterogeneity in the clinical outcome for differ- generally limited to the upper respiratory tract. The preim- ent individuals infected with the same viral strain (3). For half mune host defense response against IAV is generally, there- a century it has been recognized that serum factors can inhibit fore, successful. We have previously provided (and summa- influenza hemagglutination (4). Recently, it has been estab- rized) evidence that neutrophils contribute to defense against lished that one class ofserum HA inhibitors, the "beta" inhibi- IAV, although neutrophil dysfunction and local tissue damage tors, are mannose-binding lectins (4, 5). Several strains ofIAV may be less salutory byproducts of this response. Here we pro- that have been selected on the basis of resistance to growth vide evidence that the serum lectin mannose-binding protein inhibition by beta inhibitors have been characterized (4, 5). In directly inhibits hemagglutinin activity and infectivity of sev- each case a specific mutation has been found in the resistant eral strains of IAV. In addition mannose-binding protein acts as strain (as compared with the parent, sensitive, strain), that an opsonin, enhancing neutrophil reactivity against IAV. Op- leads to loss of a specific glycosylation site. For two of the sonization of IAV by mannose-binding protein also protects the strains, the parent virus is known to have at this site a high neutrophil from IAV-induced dysfunction. These effects are mannose glycan attachment that lies in close proximity to the observed with physiologically relevant concentrations of man- cell attachment site ofthe HA. These findings suggest that beta nose-binding protein. Two different allelic forms of recombi- inhibitors of IAV act by binding to these high mannose struc- nant mannose-binding protein are found to have similar effects. tures, thus masking the cell attachment domain of the HA. We believe, on the basis of these data, that mannose-binding Two candidate serum lectins that may constitute beta inhib- protein alone and in conjunction with phagocytic cells is an itors are mannose-binding protein (MBP) and conglutinin (5- important constituent of natural immunity (i.e., preimmune de- 7). MBP and conglutinin are members ofa family ofmamma- fense) against IAV. (J. Clin. Invest. 1993.91:1414-1420.) Key lian serum lectins that have in common a Ca2+-dependent lec- words: neutrophils * hemagglutination * recombinant proteins* tin-binding domain at the COOH terminus and a collagenous mannose-binding protein , influenza A virus domain followed by a short area ofdisulfide bonds at the NH2 terminus (6). Both are present as multimeric structures in Introduction serum. Structurally both resemble C I q and are able to bind to Clq receptors through their collagen domain (6, 7). The product ofinfluenza A virus (IAV)' hemagglutinin (HA) This study has focused on the interaction of recombinant genes is expressed on the surface of infected cells and evokes MBPs with several strains of IAV. MBPs have been character- specific and long-lived humoral and cell-mediated immune re- ized in the serum ofmice, rats, rabbits, and humans (6, 7). The physiological role for these proteins appears to be in first line A preliminary report of some of the data presented in this paper has host defense and they may be considered as pattern-recogni- been published in abstract form ( 1991. J. Cell Biol. 115:361a.) tion molecules that function as ante-antibodies (8). MBPs are Address correspondence to Kevan L. Hartshorn, Department of acute-phase reactants (9, 10) that recognize a wide variety of Medicine, Boston University School ofMedicine, 80 East Concord St., disparate oligosaccharides that are present on the cell walls of Boston, MA 02118-2394. certain Gram-negative bacteria ( 1), fungal pathogens ( 12), Received for publication 13 August 1992 and in revisedform 12 and the envelope glycoprotein of human immunodeficiency November 1992. virus ( 13). There are at least two allelic variants ofMBP. In the vast majority ofindividuals there is one aberration in the colla- 1. Abbreviations used in this paper: HA, hemagglutinin; HAU, hemag- glutinin units; IAV, influenza A virus; MBP, mannose-binding protein; gen region at the eighth repeat as Gly-Gln-Gly is present rather PFU, plaque-forming units; rMBP, recombinant MBP. than Gly-X-Y-Gly. In 5-7% ofthe population Asp replaces Gly at the fifth collagen repeat, in addition to the irregularity ob- J. Clin. Invest. served more commonly ( 14). Homozygosity for this latter phe- © The American Society for Clinical Investigation, Inc. notype appears to segregate with low serum MBP levels, which 0021-9738/93/04/1414/07 $2.00 in turn have been associated with the phenotype of recurrent Volume 91, April 1993, 1414-1420 infections (14, 15). However, in recent studies both allelic 1414 Hartshorn, Sastry, White, Anders, Super, Ezekowitz, and Tauber forms of MBP have been shown to have similar functional Canberra, Australia) and the beta-inhibitor resistant mutant properties in vitro. One important difference is that the recom- (Mem7IH-BelN/BS) was developed by Dr. E. Margot Anders (Univer- binant MBP containing the Gly to Asp mutation at the fifth sity ofMelbourne, Parkville, Victoria, Australia). The H3N2/A/Bang- collagen repeat was not able to fix complement via the classical kok/ 1/79 (Bankok 79) strain was a gift of Dr. Robert Webster (St. Jude's Hospital, Memphis, TN). HA titers were determined by titra- pathway ( 14). In this paper we assess the interaction of both tion ofvirus samples in PBS followed by addition ofthoroughly washed forms of MBP with various influenza viral strains selected for human type 0 red blood cells. known differences in HA glycosylation. Incubation of MBPs with concentrated IAV stocks was carried out MBP-ligand complexes would be expected to encounter for 0.5 h at 370C in PBS with 1 mM Ca2" and Mg2", except in certain circulating cells, in particular neutrophils, which have been experiments where 2 mM EGTA or 10 mg/ml mannan (Sigma Chemi- shown to ingest bacteria opsonized with MBP ( 11). Neutro- cal Co.) were added, as indicated. For experiments with neutrophils, phils alone may play a role in the preimmune host defense aliquots of virus or MBP-treated virus were then added to cell suspen- against influenza ( 16, 17). We and others have characterized sion at a 1:40 vol/vol ratio. In describing these experiments, the final the interaction of human neutrophils with unopsonized IAV concentration of MBP in the neutrophil suspension is given. and found evidence that these cells are activated by IAV. In the Virus plaque formation was assessed on Madin-Darby canine kid- ney cells: duplicate confluent monolayers in six-well plates were ex- absence of antibody, neutrophils are stimulated by cell-free posed to virus preparations with orwithout MBP in 10-fold dilutions in virus to generate a pertussis toxin-insensitive respiratory burst serum-free CRCM-30 medium (Sigma Chemical Co.) for 1 h at 370C, response, characterized by hydrogen peroxide (H202), but no followed by washing twice in PBS. The monolayers were then overlaid superoxide (O°-) production ( 18, 19). Neutrophils are also re- with CRCM-30 media with 0.1% BSA, 0.4% agarose, 10 ,ug/ml trypsin, ported to specifically adhere to IAV-infected epithelial mono- and 10 ;g/ml DEAE-Dextran. The agarose solution was warmed until layers without mediation of antibody (20). After exposure to melting and allowed to cool to the touch, and then mixed with the other influenza viruses in vitro and in vivo, however, neutrophils and ingredients and layered over the cells before solidification. The mono- monocytes exhibit depressed responsiveness to various stimuli layers were then kept at 370C in a 5% CO2 incubator for 72 h, followed and depressed microbicidal activity against bacteria (21, 22). by removal ofthe agarose overlay. After staining with crystal violet, the In this report we demonstrate that both forms of recombinant plaques were counted by direct inspection and titers ofplaque-forming units (PFU)/ml calculated. Values given are mean PFU/ml for two MBP inhibit IAV HA activity and act as opsonins, enhancing wells. MDCK cells were obtained from the American Type Culture neutrophil responsiveness to the virus. Collection (Rockville, MD). Neutrophils from healthy volunteer donors were isolated to > 95% purity as previously described using dextran precipitation, followed by Methods a Ficoll-Hypaque gradient separation for removal ofmononuclear cells and hypotonic lysis to eliminate contaminating erythrocytes (21 ). Cell Reagents. FMLP, cytochalasin B, horseradish peroxidase type II, sco- viability was > 98% as determined by trypan blue staining, and cells poletin, SOD, cytochrome c, neuraminidase type X (protease activity were used within 5 h ofisolation.