Abstract
Listeria monocytogenes is responsible for severe food-borne infections, but the mechanisms by which bacteria cross the intestinal barrier are unknown. Listeria monocytogenes expresses a surface protein, internalin, that interacts with a host receptor, E-cadherin, to promote entry into human epithelial cells. Murine E-cadherin, in contrast to guinea pig E-cadherin, does not interact with internalin, excluding the mouse as a model for addressing internalin function in vivo. In guinea pigs and transgenic mice expressing human E-cadherin, internalin was found to mediate invasion of enterocytes and crossing of the intestinal barrier. These results illustrate how relevant animal models for human infections can be generated.
Understanding how bacteria cross the intestinal barrier is a key issue in the study of foodborne diseases. Listeria monocytogenes causes listeriosis, an infection characterized by bacterial dissemination from the intestinal lumen to the central nervous system and the fetoplacental unit (1). As recently shown, L. monocytogenes is also responsible for gastroenteritis (2, 3). How this bacterium crosses the intestinal barrier is unknown. In vitro, theL. monocytogenes surface protein internalin promotes bacterial internalization into human epithelial cells (4). Its receptor is E-cadherin (5), a protein that mediates the formation of adherens junctions between epithelial cells and is also expressed on the basolateral face of polarized epithelial cells (6). In contrast to human E-cadherin (hEcad), mouse and rat E-cadherins are not receptors for internalin, and internalin plays no role in entry into mouse and rat epithelial cells (7). We have shown that this specificity relies on the nature of the sixteenth amino acid, a proline in hEcad, and a glutamic acid in mouse and rat E-cadherins (7). We concluded that although rat and mouse can be successfully used to study the T cell response to intravenous (IV) infection of L. monocytogenes, they are inappropriate models for studying internalin function in vivo. In mice, oral infections are not reproducibly lethal, and bacterial translocation across the intestinal barrier is low. Moreover, in these animals, specific targeting to the brainstem and the fetoplacental unit is not seen, even after IV inoculation.
Guided by the pioneering work of Racz et al.(8), we observed that a guinea pig epithelial cell line allows entry of L. monocytogenes strain EGD at an efficiency 200 times that of isogenic internalin deletion mutant EGDΔinlA (7). These results correlated with the finding that guinea pig E-cadherin harbors a proline at position 16 and led us to propose that the role of internalin in vivo could be addressed with guinea pigs or transgenic mice expressing hEcad. We report the results of oral infections of both guinea pigs and transgenic mice expressing hEcad on their enterocytes and show that internalin is a virulence factor mediating crossing of the intestinal barrier.
IV infections of mice with L. monocytogenes are lethal. The median lethal dose (LD50) of EGD or EGDΔinlA is similar (∼105 bacteria) (9). In contrast, oral infections with even 5 × 1010 EGD or EGDΔinlA are not lethal, and few bacteria cross the intestinal barrier (7, 9). We infected guinea pigs to assess L. monocytogenes virulence via the oral route in a model in which the role of internalin could be tested. Oral infection with 5 × 1011 EGD resulted in 100% mortality, and infection with 1010 EGD resulted in 100% survival; the LD50 is 1011 bacteria (Web fig. 1) (9). In contrast, infection with 5 × 1011 EGDΔinlA resulted in 100% survival. Thus, internalin acts as a virulence factor in guinea pigs. Because E-cadherin is expressed in various epithelial cells, internalin may be critical after crossing of the intestinal barrier at a later stage of infection. We thus tested the role of internalin after an IV infection (9). The LD50 values of EGD and EGDΔinlA after IV infection of guinea pigs were identical (∼5 × 107 bacteria), and the number of bacteria present in the liver and spleen 72 hours after IV inoculation with 106 EGD or EGDΔinlA is similar (9,10). The absence of a detectable role for internalin in infections initiated via the IV route, together with internalin-dependent mortality after oral infection, suggested a critical role for internalin at an early stage of infection, between bacterial ingestion and access to the bloodstream. We followed bacterial invasion in the small intestine of guinea pigs and translocation into the mesenteric lymph nodes, liver, and spleen using different inocula (5 × 1011, 1011, and 1010 bacteria). Wild-type bacteria invade the intestinal tissue and reach deeper organs, whereas EGDΔinlA bacteria do not (Fig. 1). This process is dose dependent (10) and internalin dependent (Fig. 1). Intestinal invasion is not promoted by the presence of Peyer's patches, as the ratio of bacterial counts of EGD versus EGDΔinlA is more than one order of magnitude higher in intestinal portions without Peyer's patches than in isolated Peyer's patches (Web fig. 2) (9). Intestinal sections immunolabeled with antibody to L. monocytogenes and electron microscopy revealed the presence of intraenterocytic L. monocytogenes 12 hours after bacterial ingestion (Fig. 2). Despite extensive searching, EGDΔinlA was not detected within enterocytes. To verify these findings, we used L. innocua, a nonpathogenic species that becomes invasive when transformed with inlA. Ex vivo infection of intestinal samples with L. innocua (inlA), in contrast to wild-type L. innocua, resulted in the invasion of enterocytes, confirming that internalin was able to mediate invasion of those cells (Web fig. 3) (9).
Role of internalin in the bacterial invasion of guinea pig (A) small intestine, (B) liver, (C) mesenteric lymph nodes, and (D) spleen. Bacterial counts in organs of guinea pigs infected orally with 1010 L. monocytogenes wild-type strain EGD (purple plots) and the internalin deletion mutant EGDΔinlA (white plots) are shown (9). Mean colony-forming units (CFU) from four different animals are given at each time point.
Internalin-mediated invasion of enterocytes and crossing of the intestinal barrier in guinea pigs. (A) Histological sections of intestinal samples after oral infection of guinea pigs with wild-type (WT) L. monocytogenes (left) and the internalin mutant ΔinlA(right). Arrows show bacteria, which appear red; scale bars, 10 μm (9). (B) Thin sections of an intestinal sample from guinea pig infected with wild-type L. monocytogenes. Intestinal portions were dissected and fixed 12 hours after infection (5, 9). The bottom panels show enlargements of three intraenterocytic L. monocytogenes in the top panel: one is still in a phagocytic vacuole (left); one is partially surrounded with actin in a two-membrane vacuole, suggesting that it had spread from one cell to the next (middle); and one is dividing in the cytosol (right).
In animals infected orally with wild-type L. monocytogenes, intravillous abscesses, i.e., foci of bacterial multiplication and recruitment of polymorphonuclear cells and monocytes, were detected as early as 12 hours after bacterial ingestion. Their size and number correlated with bacterial counts and reached a maximum 72 to 120 hours after infection (Figs. 1 and 2A). In animals infected with high doses of EGDΔinlA (5 × 1011 bacteria), bacteria were occasionally seen in the intestine, either in Peyer's patches or in regions where the epithelial monolayer was damaged, but intravillous abscesses similar to those observed with EGD were not seen with EGDΔinlA.
To demonstrate and elucidate the role of the internalin E-cadherin interaction in vivo, we generated transgenic mice that express hEcad solely in enterocytes. This was achieved by placing the hEcad gene under the control of the iFABP promoter (9,11). iFABP encodes the intestinal isoform of the fatty acid binding protein whose expression is restricted to epithelial cells of the small intestine (12). Several transgenic founders were obtained, and the corresponding lines were established and tested (9, 11). One of them, exhibiting the expected hEcad expression pattern, i.e., restricted to enterocytes, was further characterized (Web fig. 4) (9). Parental and transgenic intestinal morphologies were indistinguishable (10). Oral infections of nontransgenic mice with 5 × 1010 wild-type bacteria or the internalin mutant resulted in complete survival of the animals (Fig. 3A). In contrast, infection of transgenic mice resulted in ∼85% mortality with this infective dose of wild-type bacteria and zero mortality with the same dose of the internalin mutant (Fig. 3A). Moreover, a kinetic analysis of bacterial counts in intestine and mesenteric lymph nodes (Fig. 3B) and immunohistochemical studies (Fig. 3C) showed that the internalin hEcad interaction mediated invasion of enterocytes and subsequent intravillous bacterial multiplication, abscess formation, and access to mesenteric lymph nodes, liver, and spleen (10).
Infection of transgenic mice expressing hEcad on their enterocytes by L. monocytogenesand the internalin deletion mutant. (A) Survival of nontransgenic mice and iFABP-hEcad transgenic mice after oral infection with 5 × 1010 wild-type (WT) L. monocytogenes (purple plots) and the internalin mutant ΔinlA (black plots) (9). (B) Bacterial counts in organs of nontransgenic mice (left) or iFABP-hEcadtransgenic mice (right) infected with 1010 wild-typeL. monocytogenes (purple plots) and the internalin mutant ΔinlA (white plots) (mean CFU from three animals for each time point) (9). (C) Histological sections of intestinal samples after oral infection ofiFABP-hEcad transgenic mice with wild-typeL. monocytogenes (left) and the internalin mutant ΔinlA (right) (scale bars, 10 μm) (9).
In conclusion, these data establish that interaction of internalin with E-cadherin on enterocytes is an early critical step for the onset of listeriosis in vivo. From data obtained in the mouse and rat species, the intestinal phase of human listeriosis was previously assumed to be almost silent and intestinal translocation of bacteria was assumed to be a nonspecific event (13). Listeriosis was thought to be mainly due to the ability of L. monocytogenes to survive inside phagocytic cells of deeper tissues once across the epithelial barrier. But as reported recently (2, 3), ingestion of large numbers of L. monocytogenes (up to 3 × 1011 bacteria per individual) is associated with gastroenteritis. Our results confirmed this enteropathogenicity and offer a molecular basis for this observation. In addition, by promoting invasion of enterocytes and translocation across the intestinal barrier, internalin also mediates access to deeper tissues and should now be considered a virulence factor as important as other well-characterized virulence factors, such as listeriolysin O and ActA (14).
E-cadherin is mostly expressed at adherens junctions of enterocytes (6) and does not appear to be accessible from the apical side of enterocytes in the intestinal brush border. How does internalin reach E-cadherin? The intestinal mucosa is a fast renewing tissue (12). Enterocytes, at the tip of villi, are replaced by ascending enterocytes produced in intestinal crypts, and the transient opening of cell-cell junctions between enterocytes may allow E-cadherin to interact with bacteria (12, 15). Consistent with this hypothesis, L. monocytogenes was more frequently found at the tip of intestinal villi in guinea pigs and transgenic mice. Yet, invasion of enterocytes in vivo remains a rare event, as abscesses were not detected in all villi, although each intestinal villus possesses thousands of cells.
InlB, another invasion surface protein of L. monocytogenesthat mediates entry into various cell types in vitro (16), might promote internalin access to E-cadherin in vivo. Indeed, InlB has been shown to interact with Met, the hepatocyte growth factor receptor, which is also expressed on enterocytes and whose activation can result in disruption of cell-cell junctions (17, 18). Listeria monocytogenes may also enter enterocytes at the apical side, as L. innocuaexpressing internalin can infect enterocytes and is detected just beneath microvilli (9).
The tropism of Listeria internalin for enterocytes differs from that of Yersinia invasin, which targets apically expressed β1-integrins in M cells and permits translocation ofYersinia through these cells (19, 20). The internalin E-cadherin interaction is thus a potential tool for specific drug targeting to enterocytes and delivery across the intestinal barrier. In humans, L. monocytogenes can also cross the fetoplacental and the blood-brain barriers, which contain E-cadherin–expressing cells (21–23). Whether internalin is involved in these other steps of the infection is unknown. Interestingly, cytomegalovirus and Toxoplasma gondii have the same double tropism for the maternofetal unit and the brainstem.
Transgenic mice have been instrumental for studying the molecular mechanisms of viral infections, including poliomyelitis [mice expressing human poliovirus receptor (24)] and measles [mice expressing human CD46 (25)]. As shown here, transgenic mice can also provide models for human bacterial diseases.
↵* To whom correspondence should be addressed. E-mail: pcossart{at}pasteur.fr
