The role of transepithelial transport by M cells in microbial invasion and host defense

The lining of the intestine is a vast monolayer of highly polarized epithelial cells that provides an effective barrier to most of the macromolecules, microorganisms and toxins present in the intestinal lumen. One component of this barrier is the thick and complex coat of glycoproteins and glycolipids on the apical brush borders of intestinal absorptive cells. The epithelium is nevertheless vulnerable to enteric pathogens that express surface adhesins, enzymes, and other specialized mechanisms for colonization of epithelial surfaces and invasion of the mucosa. It is thus not surprising that the intestinal mucosa is heavily populated with cells of the immune system. Indeed, the intestinal lining contains more lymphoid cells and produces more antibodies than any other organ in the body (Mestecky and McGhee, 1987; Seilies et al., 1985). The vast majority of antibodies produced at this site are of the IgA isotype, and are exported into secretions.

Transepithelial transport plays two crucial roles in the mucosal immune response (Fig. 1). First, samples of antigens and microorganisms must be transported from the intestinal lumen across the epithelium in order to be processed in lymphoid tissues and elicit a mucosal immune response (Neutra and Kraehenbuhl, 1992; Kraehenbuhl and Neutra, 1992). This appears to be the special role of M cells, a unique epithelial cell type located exclusively in the lymphoid follicle-associated epithelia over sites containing organized mucosal lymphoid tissue (Bockman and Cooper, 1973; Neutra et al., 1987; Owen 1977). Antigen processing and presentation at these sites results in IgA-committed, antigen-specific B lymphocytes that proliferate locally, then leave the mucosa, migrate via the bloodstream, and finally ‘home’ to mucosal or glandular sites throughout the intestine as well as other mucosal and secretory tissues (Cebra et al., 1976; McDermott and Bienenstock, 1979). These disseminated cells terminally differentiate to become subepithelial plasma cells that produce polymeric IgA antibodies (Mestecky and McGhee, 1987; Kraehenbuhl and Neutra, 1992). Transepithelial transport in the basal to apical direction is then required for transport of IgA into glandular and mucosal secretions, and this is mediated by polymeric immunoglobulin (poly-Ig) receptors in a variety of epithelial and glandular cells (Mostov et al., 1980; Kuhn and Kraehenbuhl 1982; Apodaca et al., 1991).

Organized mucosal lymphoid tissues, recognized by the presense of lymphoid follicles, are present in the oral cavity, the bronchi, and throughout the small and large intestines (Owen and Ermak, 1990; Owen and Nemanic, 1978). Single follicles or small clusters are located along the length of the gastrointestinal (GI) tract, with increasing frequency in the colon and rectum (O’Leary and Sweeney, 1986; Fujimura et al., 1992) and aggregated follicles are found in the lingual and palatine tonsils, adenoids, appendix and Peyer’s patches in the small intestine. Whether single or aggregated, mucosal lymphoid follicles are separated from the lumen by a highly specialized epithelium containing antigen-transporting M cells (Fig. 2).

Since transepithelial transport by M cells seems to be required for initiation of a secretory immune response, increasing the efficiency of this transport is an important component of mucosal immunization strategies (Neutra and Kraehenbuhl, 1992). The M cell apical surface lacks the closely packed microvilli and thick enzyme-rich coat of absorptive enterocytes, and displays many broad microdomains from which endocytosis occurs (Neutra et al., 1987). M cell apical membranes nevertheless contain abundant glycoconjugates that can serve as binding sites for cationic macromolecules and possibly for lectin-like microbial surface molecules (Neutra et al., 1987; Bye et al., 1984; Owen and Bhalla, 1983). Microorganisms, particles and lectins that adhere either selectively or nonselectively to M cell apical membranes are endocytosed and transcytosed with high efficiency (Neutra et al., 1987; Pappo and Ermak 1989). This is consistent with the observation that materials that can adhere to mucosal surfaces tend to evoke strong secretory immune responses (DeAizpurua and Russell-Jones, 1988; Mayrhofer, 1984). Certain pathogenic viruses and bacteria adhere selectively to M cells and are efficiently transported, but niether the microbial surface molecules that mediate adherence nor the M cell surface molecules that serve as receptors have been identified. It is probable that microorganisms, which may be considered multivalent particulate ligands, could find their receptors particularly accessible on M cells. It was recently demonstrated that immunoglobulins in the lumen adhere selectively to the apical membranes of M cells, and IgA-antigen complexes were found to adhere to M cells much more efficiently than antigen alone (Weltzin et al., 1989). This raises the possibility that slgA could in some cases promote re-uptake of antigens and microorganisms by M cells, perhaps to boost the secretory immune response to pathogens that have not been effectively cleared from the lumen. Studies of this transport system are hampered by the fact that M cells are relatively rare occupants of the intestinal lining, however, and the M cell phenotype has not been replicated in culture.

M cells take up macromolecules, particles and microorganisms by adsorptive endocytosis via clathrin-coated pits and vesicles (Neutra et al., 1987), fluid-phase endocytosis in either coated (Neutra et al., 1987) or uncoated vesicles (Bockman and Cooper, 1973; Owen, 1977), and phagocytosis involving extension of cellular processes and reorganization of submembrane actin assemblies (Winner et al., 1991). Bacteria that adhere to M cells can form broad areas of close interaction: during uptake of Vibrio cholerae, for example, the bacterial outer membrane and the M cell apical membrane are separated by a uniform 10-20 nm space. Endocytosed material is delivered into apical endosomal tubules and vesicles (Neutra et al., 1987) some of which contain the late endosome or lysosome membrane marker lgpl20 and (in some species) MHC class II antigen (Allan et al., 1993). M cell endosomes are acidified (Allan et al., 1993), but it is not known whether they contain proteases and whether endocytosed materials are partially degraded during transepithelial transport.

The transcytotic pathway of M cells is dramatically shortened by invagination of the basolateral membrane to form a large intra-epithelial ‘pocket’. Apical endosomes of M cells deliver their content not to lysosomes but rather directly to the membrane domain lining the pocket, and endocytosed materials are released by exocytosis into the pocket as early as 10 minutes after apical endocytosis (Neutra et al., 1987). Since bacteria and particles are readily released from the membrane into the intra-epithelial space, it is likely that that initial binding involves multiple, low-affinity interaction sites and that the milieu of the transport vesicle or intra-epithelial pocket allows rapid dissociation. Whether apical membrane molecules are replaced by de novo synthesis or by recycling of membrane microdomains from the pocket is not known. Antigens that are transported by M cells may interact first with the antigen-presenting cells and lymphocytes present in the intraepithelial pocket formed by M cells (Ermak et al., 1990) but it is not known whether an immune response is generated in this sequestered site. In any case, the subepithelial tissue immediately below the follicle-associated epithelium contains IgM⁺ B cells, CD4⁺ T cells, dendritic cells and macrophages, and in this cellular network antigens and microorganisms are likely to be efficiently processed and presented (Ermak and Owen, 1986).

A common mechanism such as lectin-carbohydrate recognition may allow the M cell to ‘sample’ pathogenic luminal organisms, either by binding of lectin-like bacterial adhesins to M cell surface glycoconjugates or, conversely, binding of bacterial surface oligosaccharides to M cell surface lectins. M cell binding of noninvasive bacteria such as Vibrio cholerae (Winner et aL, 1991; Owen et al., 1986) results in efficient sampling by the mucosal immune system, and a strong secretory immune response. In the case of cholera, secretion of anti-microbial slgA appears to play a major role in limiting the duration of mucosal disease and preventing re-infection (Svennerholm et al., 1984; Jertbom et al., 1986). A variety of pathogenic bacteria and viruses that bind to M cells, however, exploit the transport mechanism that was intended for mucosal protection by using this transepithelial pathway as an invasion route. For example, transport of Salmonella typhi into Peyer’s patch mucosa (Kohbata et al., 1986) results in a vigorous anti-Salmonella mucosal immune response, but this occurs too late to prevent spread of bacteria to the liver and spleen and disseminated systemic disease (Chau et al., 1981; Hohmann et al., 1978). Similarly, M cell binding and transport of Shigella flexneri (Wassef et al., 1989) and Yersinia enterocoliitica (Grutzkau et al., 1990) allows these organisms to gain access to the lamina propria, where they cause mucosal disease by basolateral invasion of epithelial cells and infection of mucosal macrophages (Isberg, 1990; Sansonetti, 1991).

Viral pathogens also exploit the M cell transport system. This has been amply demonstrated in studies of reovirus pathogenesis in mice (Wolf et al., 1981). Poliovirus adhered to human M cells in organ culture (Sicinski et al., 1990), and the retrovirus HIV-1 adhered to M cells of rabbit and mouse follicle-associated epithelia in mucosal explants (Amerongen et al., 1991). Selective adherence to M cell apical membranes effectively targets these pathogens for efficient transport into the intestinal mucosa. In the case of reovirus type 3 and possibly poliovirus, entry into Peyer’s patch mucosa is followed by invasion of neuronal target cells and spread to the central nervous system (Nibert et al., 1991). If HIV is transported by human M cells, the virus would be delivered directly to target T cells within and under the follicle-associated epithelium. Although viral particles are sampled by the mucosal immune system and immune responses may be generated, entry of virus into target cells puts them beyond the reach of anti-viral antibodies. Identification of viral surface components that mediate M cell adherence, and the corresponding M cell receptors, remains an important research priority.

There is considerable indirect evidence that specific secretory IgA in the fluids bathing mucosal surfaces can prevent contact of antigens and pathogens with epithelial cells, a phenomenon called ‘immune exclusion’ (Tomasi, 1983; Killian et al., 1988). The molecular mechanisms that underlie this protection, however, are not completely understood and it is likely that this unique effector molecule can play multiple roles in the changing mucosal environment of the intestinal lumen. Secretory IgA is an effector molecule, which functions outside the body in environments usually devoid of complement and phagocytic cells. Most IgA antibodies studied to date are not ‘neutralizing’ in the classic sense as they generally do not opsonize in vitro, do not bind complement or cause release of complement fragment C5a, and do not directly cause bacterial lysis (Tomasi, 1983; Killian et al., 1988; Childers et al., 1989). It has been shown that specific IgA antibodies injected systemically may be ineffective against systemic microbial challenge even when the same antibodies can protect against mucosal challenge when they are present in mucosal secretions (Subbarao and Murphy, 1992; Michetti et al., 1992). IgA antibodies secreted in response to mucosal bacteria and viruses are directed primarily against surface antigens or secreted toxins. Since slgA is a dimer with 4 antigen binding sites, it can efficiently crosslink target macromolecules and micro-organisms in the intestinal lumen, thus inhibiting motility and facilitating entrapment in mucus and clearance by peristalsis. In the context of the mucosa, slgA can also collaborate with bacteriostatic proteins and lytic cells to enhance microbial killing in novel ways (Killian et al., 1988; Childers et al., 1989).

Oral or mucosal vaccination has been clearly shown to be the optimal method for induction of secretory IgA and effective immune protection against many enteric pathogens. Thus there is rapidly growing interest in testing and use of novel oral vaccines in experimental animals and humans (Subbarao and Murphy, 1992; Levine and Edelman, 1990; Mekalanos, 1992). The immune responses generated by enteric viral and bacterial infections and by mucosal vaccines usually include other components in addition to secretory IgA, however, such as production of systemic IgG antibodies and cell-mediated immunity (Cancellieri and Fara, 1985). Thus it has been difficult to judge the relative importance of slgA in protection, or to determine whether ‘immune exclusion’ by slgA alone can be sufficient to prevent mucosal infection. It is technically difficult to collect, purify and analyse polyclonal slgA from secretions and even more difficult to determine which antigenic determinants and antibody specificities contributed to protection. For these reasons, several laboratories including our own have produced monoclonal IgA antibodies and have used them to address these issues in vivo and in vitro.

Mucosal (but not systemic) immunization favors the formation of antigen-sensitized lymphoblasts of the IgA isotype in organized mucosal lymphoid tissues such as Peyer’s patches. In our laboratories, we have found that fusion of cells isolated directly from Peyer’s patches after a series of mucosal immunizations with viruses, bacteria and protein antigens is an effective method for obtaining IgA hybridomas (Weltzin et al., 1989; Winner et al., 1991; Michetti et al., 1992; Apter et al., 1991). The antigen specificities of the IgA antibodies thus obtained serve to identify the microbial surface molecules that are most immunogenic in the mucosal system. Most of the monoclonal IgAs that we have produced and analysed have been directed against microbial surface components, and this is consistent with the proposed function of slgA in immune exclusion. Mucosally derived IgA hybridomas produce IgA primarily in dimeric form and these antibodies are recognized by polymeric immunoglobulin receptors on epithelial cells (Weltzin et al., 1989). Thus they can be delivered into secretions of mice in vivo or across epithelial monolayers in vitro via the normal transepithelial transport system with addition of secretory component (Winner et al., 1991; Subbarao and Murphy, 1992) and can bind recombinant secretory component in solution (Michetti et al., 1991). These methods provide an unlimited source of protease-protected IgA antibodies of known epitope specificities that can be separately tested in protection assays.

To obtain continuous secretion of monoclonal IgA antibodies into intestinal secretions and bile of mice via the normal receptor-mediated epithelial transport system, hybridoma cells are injected subcutaneously on the upper backs of Balb/c mice, resulting in hybridoma ‘backpack’ tumors that release IgA into the circulation (Winner et al., 1991). In tissues such as liver and intestine where capillaries are permeable, dimeric IgA binds to basolateral epithelial poly-Ig receptors and is delivered into secretions as monoclonal slgA (Fig. 3). We have used the backpack tumor method to analyse the specificity and the mechanisms of IgA protection against micro-organisms in the intestine, using as models two enteric bacterial pathogens: Vibrio cholerae, a non-invasive organism that colonizes the mucosal surface and secretes cholera toxin, and Salmonella typhimurium, an invasive pathogen.

Vibrio cholerae causes severe diarrheal disease by colonizing the mucosal surface of the small intestine. When vibrios enter the luminal microenvironment the toxR regulatory gene is activated resulting in expression of several virulence factors including the pilus protein that promotes mucosal colonization, and cholera toxin (CT) (Miller et al., 1989). Binding of CT to its glycolipid receptor GM1 on epithelial cells initiates a series of intracellular events that result in massive chloride secretion (Holmgren, 1981; Lencer et al., 1992). V. cholerae evokes polyclonal secretory IgA antibodies (slgA) directed against both CT and bacterial surface components including the outer membrane lipopolysaccharide (LPS) (Svennerholm et al., 1984) and these are thought to be involved in limiting the primary infection and providing protection against subsequent reinfection (Svennerholm et al., 1984; Jertbom et al., 1986). Using the hybridoma backpack tumor method, we have shown that secretion of monoclonal slgA directed against a strain-specific carbohydrate epitope of the surface LPS is sufficient to prevent diarrhea and death in mice after a lethal oral dose of V. cholerae (Winner et al., 1991). Furthermore, a single oral dose of 5 mg of the same anti-LPS IgA protected suckling mice against subsequent oral challenge for up to 3 hours, although protection was lost as the IgA was cleared from the upper small intestine by peristalsis (Apter et al., 1991).

To test the ability of IgA to block the effects of CT and prevent diarrheal disease, hybridomas were generated that produce monoclonal IgA antibodies directed against CT; all of the IgAs recognized the B subunit and none was directed against the binding site for GMi, the intestinal cell membrane toxin receptor (Apter et al., 1991). These antibodies were applied to a well-defined intestinal epithelial cell monolayer culture system (Lencer et al., 1992; Dharm-sathaphom and Madara, 1990) to directly measure protection against toxin action in vitro, and were tested in vivo by passive oral administration or the backpack tumor method. When applied to human T84 colon carcinoma cells grown on permeable supports, the IgAs blocked CT-induced Cl^-secretion in a dose-dependent manner and completely inhibited binding of rhodamine-labelled CT to apical cell membranes (Apter et al., 1991). Oral feeding of the monoclonal IgAs protected against oral CT administration in vivo. Both in vitro and in vivo, however, high doses of IgA were required for CT neutralization. In contrast, anti-CTB IgA failed to protect suckling mice against a lethal oral dose of live V. cholerae organisms, whether the antibodies were delivered perorally or secreted intraintestinally in mice bearing backpack tumors (Apter et al., 1991). These results corroborate previous human studies in which volunteers orally immunized with multiple large doses of purified CT were not protected against oral challenge with V. cholerae, whereas volunteers immunized orally with live V. cholerae organisms were completely protected (Levine et al., 1979, 1983). Since CT is not a component of the bacterial surface, anti-CT IgA alone is not likely to prevent V. cholerae colonization, and secretion of toxin from adherent organisms on the mucosal surface could deliver high levels of CT directly into the enterocyte glycocalyx and onto apical membranes. Our results show that IgAs directed against the bacterial surface provide most efficient mucosal protection.

The disease caused by Salmonella typhimurium infection in mice is similar to typhoid fever caused by S. typhi in humans (Edelman and Levine, 1986). In both cases, organisms are ingested orally, rapidly invade the intestinal epithelium, proliferate in the mucosa, and then spread to the liver and spleen resulting in a potentially lethal systemic disease (Hohmann et al., 1978; Edelman and Levine, 1986). Entry into the mucosa occurs first via transepithelial transport by M cells (Kohbata et al., 1986) and then by adherence, invasion and damage of absorptive enterocytes (Takeuchi, 1975). Immunization strategies designed to protect against this disease are aimed at preventing the initial interaction of bacteria with epithelial cells. Indeed, the presense of antiSalmonella IgA in intestinal secretions is known to be closely correlated with protection in experimental animals and humans (Chau et al., 1981; Edelman and Levine, 1986). Using monoclonal IgA antibodies, we have recently demonstrated that secretory IgA antibodies alone can provide this protection.

A series of anti-Salmonella IgA hybridomas were generated from Peyer’s patch cells and a monoclonal polymeric IgA (Sal4) was characterized that recognizes a surface-exposed carbohydrate epitope on wild type S. typhimurium (Michetti et aL, 1992). Mice bearing subcutaneous Sal4 hybridoma tumors secreted monoclonal slgA into their gastrointestinal tracts and were protected against a lethal oral challenge with this bacterium. This protection was directly dependent on specific recognition by Sal4 IgA, since mice secreting Sal4 IgA from hybridoma tumors were not protected against an equally virulent S. typhimurium mutant that lacks the epitope recognized by Sal4. In these in vivo studies, secretion of monoclonal IgA was shown to prevent the initial event in pathogenesis, uptake of Salmonella by M cells into the Peyer’s patch mucosa. We also used monolayer cultures of epithelial cell lines (MDCK and HT29) to test IgA protection against invasion of epithelial cells by Salmonella. It had been previously shown that invasion of MDCK cells in vitro correlates closely with invasion of CaCo2 cells and with infection of the small intestinal mucosa in vivo (Finley and Falkow, 1990; Finlay et al., 1988). Using these in vitro models, we have demonstrated that specific IgA antibodies can protect absorptive epithelia against invasive Salmonella in the absence of mucus, peristalsis and other protection mechanisms (Michetti et al., unpublished data). Thus, IgA alone can prevent epithelial contact and mucosal invasion.

In summary, our studies using monoclonal IgA antibodies have confirmed that the principal role of slgA in the intestine is to prevent contact of antigens and pathogens with epithelial surfaces. We have established that slgA of appropriate specificity, if present in sufficiently large amounts, can provide protection of the intestinal mucosa in the absence of other immune protection mechanisms. Identification of protective secretory IgA antibodies provides a means to identify the microbial antigens that should be included in effective oral vaccines. Elucidation of the mechanisms that govern adherence and transepithelial transport by M cells will allow us to target such vaccines to the inductive sites of the mucosal immune system.

We thank the former members of our laboratories who conducted much of the work summarized in this review: Pierre Michetti, Richard Weltzin, Helen Amerongen, Julie Mack, Scott Winner and Felice Apter. We are indebted to our collaborators John Mekalanos, Michael Mahon and James Slauch from the Department of Microbiology and Molecular Genetics at Harvard Medical School. The authors are supported by NIH Research grants HD17557, DK21505 and AI29378 and NIH Center grant DK34854 to the Harvard Digestive Diseases Center (M.R.N.); Swiss National Science Foundation grant 31.246404.89 and Swiss League against Cancer grant 373.89.2 (J.P.K.).