ABSTRACT
The intestinal stage of listeriosis was studied in a rat ligated ileal loop system. Listeria monocytogenes translocated to deep organs with similar efficiencies after inoculation of loops with or without Peyer’s patches. Bacterial seeding of deep organs was demonstrated as early as 15 min after inoculation. It was dose dependent and nonspecific, as the ΔinlAB, theΔhly, and the ΔactA L. monocytogenesmutants and the nonpathogenic species, Listeria innocua, translocated similarly to wild-type L. monocytogenesstrains. The levels of uptake of listeriae by Peyer’s patches and villous intestine were similar and low, 50 to 250 CFU per cm2 of tissue. No listeria cells crossing the epithelial sheet of Peyer’s patches and villous intestine were observed by transmission electron microscopy. The lack of significant interaction of listeriae and the follicle-associated epithelium of Peyer’s patches was confirmed by scanning electron microscopy. The follicular tissue of Peyer’s patches was a preferential site of Listeriareplication. With all doses tested, the rate of bacterial growth was 10 to 20 times higher in Peyer’s patches than in villous intestine. At early stages of Peyer’s patch infection, listeriae were observed inside mononuclear cells of the dome area. Listeriae then disseminated throughout the follicular tissue except for the germinal center. The virulence determinants hly and, to a lesser extent,actA, but not inlAB, were required for the completion of this process. This study suggests that Peyer’s patches are preferential sites for replication rather than for entry ofL. monocytogenes, due to the presence of highly permissive mononuclear cells whose nature remains to be defined.
Listeria monocytogenes is a ubiquitous gram-positive bacillus that causes serious infections in humans and animals (16). Epidemiological investigations of both epidemic and sporadic cases have shown that human listeriosis is a food-borne illness (10). Infection of pregnant women may result in abortion, stillbirth, and neonatal meningitis or sepsis (15). Meningitis, meningoencephalitis, and bacteremia are the most common presentations in nonpregnant adults. Immunocompromised and elderly individuals are primarily affected. However, apparently healthy individuals also contract listeriosis.
The establishment of a systemic infection after ingestion of the organism is an essential step in the pathogenesis of listeriosis (13). This systemic infection appears to be initiated in the small intestines of laboratory animals (20, 21). However, the mechanisms involved in the translocation of L. monocytogenes across the intestinal mucosa are largely unknown. The ileal Peyer’s patches rather than the intestinal villi are the initial sites of Listeria invasion in mice (20,21). This finding suggests that L. monocytogenespreferentially enters the host by crossing the follicle-associated epithelium (FAE) of the Peyer’s patches. The M cell could be the site of entry, as reported for other bacterial pathogens (33).
Several lines of evidence are consistent with an epithelial phase of invasion in listeriosis. Racz et al. found dividing listeriae in absorptive intestinal cells within 3 h of infection in an extensive electron microscopy study of guinea pigs (29). It has also been shown that L. monocytogenes enters cultured enterocyte-like cells (12, 22, 26) and initiates its cycle of intracellular infection inside these cells (32). Both apical and basolateral routes of entry into cultured enterocytes have been demonstrated (14, 18).
The intestinal step of Listeria infection has been studied experimentally by challenging rodents orally or intragastrically. Oral challenge is an approach that mimics the conditions of natural infection. It allows one to study important parameters interfering with the colonization of the gastrointestinal tract by L. monocytogenes, such as cellular immunity or indigenous bacterial flora (24, 38). However, the oral route has a number of limitations. Oral models are not very reproducible. Some authors have failed to achieve lethal infection by the oral route despite the use of high bacterial doses (ca. 109 bacteria per animal) (20, 31, 38). Others have reported similar 50% lethal doses by the intragastric and intraperitoneal routes (27). The number of bacteria actually delivered to the intestinal tract is low and highly variable among individuals, due particularly to the bactericidal activity of the gastric filter (20). It is not possible to assign a precise role to a given intestinal segment, especially in the translocation process, as the inoculated bacteria are distributed over the entire gastrointestinal tract. The ligated intestinal loop system overcomes these problems. It also makes quantification possible by adapting the standard gentamicin killing assay to the processing of tissue samples (25).
In this study, we used a ligated ileal loop system to examine the roles of intestinal villi and Peyer’s patches in the establishment ofListeria infection in the rat. We were particularly interested in determining whether the FAE was a preferential site forListeria invasion and whether the infection of Peyer’s patches was an obligate step in the translocation process.
MATERIALS AND METHODS
Bacterial strains and growth media. Listeria strains used in this study are listed in Table 1. Most experiments were performed with the wild-type L. monocytogenes strain EGD, obtained from T. Chakraborty (Institut für Medizinische Mikrobiologie, Giessen, Germany). The EGD strain provided by P. Cossart (Institut Pasteur, Paris, France), BUG600, was used as a control in experiments using its inlAB derivative BUG949. Vectors pAT18 and pAT28 (36) were introduced intoL. monocytogenes BUG600 and BUG949 by using published procedures (11). Transconjugants were selected on brain heart infusion (BHI; Difco) agar containing nalidixic acid (50 mg/liter), colistin (10 mg/liter), and either erythromycin (8 mg/liter) (pAT18) or spectinomycin (60 mg/liter) (pAT28). For infection of ligated rat intestinal loops, Listeria strains were grown in BHI broth containing antibiotics as appropriate at 37°C, with aeration, for 18 h. Bacteria were pelleted by centrifugation, washed once, and diluted appropriately in phosphate-buffered saline (PBS). The actual number of bacteria used to inoculate loops was controlled for each experiment by determining titers of viable bacteria on BHI agar plates. For coinfection experiments using strains carrying vectors pAT18 and pAT28, bacterial suspensions were mixed 1:1 and vortexed for 30 s before injection. Bacterial titers were determined by using selective media containing appropriate antibiotics.
Listeria strains used in this study
Rat intestinal loop assay.Wistar male rats (Janvier, Le-Genest-Saint-Isle, France) were used when they were 6 to 7 weeks old. Animals were anesthetized intraperitoneally with Nesdonal (50 mg/kg of body weight; Specia; Rhône Poulenc Rorer, Paris, France). When the surgical stage of anesthesia was reached, a small incision was made through the abdominal wall and the small intestine was exposed. A 5-cm ileal loop with or without a Peyer’s patch was formed by ligating the intestine with Vicryl 2 thread (Ethicon, Neuilly-sur-Seine, France). Care was taken to preserve intact blood supply to the loop. A 250-μl volume of bacterial suspension was injected into the closed loop through a 0.45-mm needle. The bowel was returned to the abdominal cavity, and the incision was closed. In most experiments, the rats were kept alive, under anesthesia, for 1 h. The abdomen was then reopened, the loop was deligated, and the intestinal content was flushed by injecting 5 ml of PBS containing 1 mg of gentamicin (Dakota Pharm, Créteil, France) per ml through a 0.45-mm needle. The bowel was returned to the abdominal cavity, and the incision was closed again. The rats regained consciousness and were allowed to eat and drink normally except that the drinking water contained 1 mg of gentamicin per ml to kill intraluminal listeriae. The rats were kept alive for 3 to 48 h and then were killed by injection of Dolethal (120 mg/kg) via heart puncture. In some experiments, the bacterial inoculum was kept in the loop for up to 3 h before the rats were killed.
Bacterial counts in tissue samples.Intestinal specimens consisted of Peyer’s patches and of pieces of villous intestine that were cropped to the same size as Peyer’s patches. Specimens were treated to kill extracellular bacteria, first by thoroughly washing with sterile PBS to eliminate mucus and debris and then by soaking for 3 h at 37°C in a gentamicin solution (100 mg/liter). The tissue specimens were washed twice with ice-cold PBS to eliminate residual gentamicin and then ground, and bacterial titers were determined by plating 0.1 ml of homogenates on BHI agar. Selective media containing appropriate antibiotics were used to count bacteria carrying pAT18 and pAT28 in coinfection experiments. Results were expressed as log10 CFU per tissue sample. The mesenteric lymph nodes (MLN), liver, and spleen were aseptically removed and ground. Bacterial counts were determined as described above. Results were expressed as log10 CFU per organ.
Histology and immunohistology.Tissue samples were fixed in 10% formalin and embedded in paraffin for histology. Sequential 2- to 3-μm sections were cut and stained with hematoxylin-eosin or by the Gram-Weigert procedure. Tissue samples to be processed for immunohistology were embedded in OCT compound (Miles Scientific, Naperville, Ill.) and frozen in liquid nitrogen. Cryostat sections 5 to 7 μm thick were cut, mounted on glass slides, and fixed in acetone. For immunolabeling, sections were first incubated with 20% goat serum in PBS containing 1 mM CaCl2 and 0.5 mM MgCl2for 30 min at room temperature. For detection of listeriae, sections were incubated sequentially with a rabbit antiserum to listerial O antigen 1/2 (J. Rocourt, Institut Pasteur, Paris, France), diluted 1/1,000, and a CY3-labeled goat F(ab′)2 to rabbit immunoglobulin G (Jackson ImmunoResearch Laboratories Inc., West Grove, Pa.), diluted 1/1,000. The sections were incubated with antibody for 30 min at room temperature and then washed three times in PBS. Immunolabeling of T, B, and monocyte-macrophage cell populations was performed to discriminate between the main regions of Peyer’s patches (dome, interfollicular region, and germinal center). The following mouse monoclonal antibodies from Serotec were used as primary antibodies: anti-rat pan-T-cell marker (clone MRC OX-52), anti-rat pan-B-cell marker CD45R, and anti-rat macrophage-related antigen ED1. Sections were incubated with appropriate dilutions of primary antibody (anti-pan-T, 1:100; anti-CD45R, 1:5; anti-ED1, 1:100) for 2 h at room temperature. Labeling was detected by using biotin-streptavidin-conjugated goat anti-mouse immunoglobulin and fluorescein isothiocyanate-conjugated streptavidin (Jackson ImmunoResearch Laboratories Inc.). All incubations were followed by three washes in PBS. Fluorescein isothiocyanate-phalloidin (Molecular Probes, Inc.) was used diluted 1/100 in PBS for F-actin staining. Slides were examined by fluorescence microscopy with a Leica DMRB microscope.
Electron microscopy.Tissue samples to be processed for transmission electron microscopy (TEM) were fixed in 2.5% glutaraldehyde and 2% paraformaldehyde (Sigma Chemical Co., St. Louis, Mo.) in 0.1 M cacodylate buffer (pH 7.3) containing 0.1 M sucrose, 5 mM CaCl2, and 5 mM MgCl2. Samples were postfixed in 2% aqueous osmium tetroxide (Merck, Darmstadt, Germany) for 1 h, dehydrated in a series of graded ethanol solutions, and embedded in Epon 812 (Janning, Vanves, France). Semithin sections were stained with toluidine blue for light microscopy. Ultrathin sections were stained with uranyl acetate and lead citrate and examined with an electron microscope (model CX200; JEOL, Croissy-sur-Seine, France) at 80 kV. Samples to be processed for scanning electron microscopy (SEM) were fixed in 3% glutaraldehyde in 0.2 M cacodylate buffer (pH 7.4) at room temperature for 30 min. Dehydration was carried out by using a series of alcohol dilutions (35 to 100%) and, finally, in a critical point apparatus. The samples were gold coated and examined with a JEOL 840A scanning electron microscope.
Statistical analysis.Student’s t test was used, and P values of <0.05 were considered statistically significant.
RESULTS
Quantitative study. (i) Listeria translocation to deep organs.Ileal loops with or without Peyer’s patches were inoculated with strain EGD (dose of ca. 109 CFU per loop) to determine whether Peyer’s patches were preferential sites forL. monocytogenes translocation. Spreading of the bacteria to MLN, liver, and spleen was studied over a 48-h period (Fig.1). The inoculation of loops with and without Peyer’s patches gave similar bacterial counts at 3 h, indicating that the presence of Peyer’s patches was not essential to the invasion of the host through the gut. The growth curves between 3 and 48 h were also similar in each of the organs studied, suggesting that the patterns of infection resulting from either mode of inoculation were similar.
Infection of MLN, liver, and spleen after inoculation ofL. monocytogenes EGD into loops with or without Peyer’s patches (PP). Loops were inoculated with ca. 109 CFU/loop. After 1 h of contact, they were deligated (time zero) and treated intraluminally with gentamicin (1 mg/ml). The number of viable bacteria was determined by killing groups of rats at intervals. Data points and error bars represent the mean and standard deviation of log10 CFU per organ (mean of four rats for each point).
We inoculated ileal loops with and without Peyer’s patches with doses of ca. 107, 108, or 109 CFU per loop to examine whether the translocation of L. monocytogenes EGD was a dose-dependent process. The spread of bacteria to MLN, liver, and spleen was recorded at 3 and 24 h (Fig. 2). The degree of bacterial translocation was related to the amount of inoculum for both kinds of loops. The dose dependence was most apparent at 24 h. At this time point, however, the differences between doses of 107 and 108 bacteria were statistically significant, but those between doses of 108 and 109 bacteria were not.
Effect of amount of inoculum on L. monocytogenes translocation from loops with or without Peyer’s patches (PP). Loops were inoculated with EGD at doses of ca. 109, 108, and 107 CFU/loop. After 1 h of contact, they were deligated (time zero) and treated intraluminally with gentamicin (1 mg/ml). The number of viable bacteria was determined by killing groups of rats at intervals. Data points and error bars represent the mean and standard deviation of log10 CFU per organ (mean of four rats for each point). ∗, bacteria recovered from three of four animals only.
Translocation was rapid irrespective of the presence or absence of Peyer’s patches. Bacteria were recovered from MLN, liver, and spleen in all challenged animals as early as 15 min after the bacteria were injected into the intestinal loops (dose of ca. 109 CFU per loop). The bacterial counts in each of these organs were similar for loops with and without Peyer’s patches (Table2).
Bacterial counts in deep organs 15 min after inoculation of L. monocytogenes EGD into intestinal loops
The nonpathogenic species Listeria innocua also crossed the intestinal barrier. The bacterial counts of L. innocuaCLIP1162 in MLN, liver, and spleen at 3 h (dose of ca. 109 CFU per loop) were not significantly different from those of L. monocytogenes EGD (Table3). However, CLIP1162 did not grow in these organs. There was a 300-fold difference in the bacterial counts of this strain and of EGD at 24 h (Table 3). This difference was statistically significant.
Infection with L. innocua CLIP1162
(ii) Listeria replication in Peyer’s patches and villous intestine.A procedure adapted from the gentamicin survival assay was used to study the growth of L. monocytogenes EGD and L. innocua CLIP1162 in Peyer’s patches and villous intestine over a 24-h period. There was no significant difference in the initial uptake of EGD by Peyer’s patches and villous intestine at any dose (Fig.3). Similar, low numbers of gentamicin-protected bacteria were recovered from both intestinal tissues at 3 h. There was a dose-dependent effect. Bacterial counts at 3 h were not affected by lowering doses from 109 to 108 bacteria but decreased dramatically when the dose was lowered from 108 to 107bacteria, which suggests that there are similar mechanisms of uptake by Peyer’s patches and villous intestine. Listeriae multiplied more readily in Peyer’s patches irrespective of the amount of inoculum used. At all doses, the number of bacteria at 24 h was 50 to 100 times greater in these organs than in villous intestine. The differences were statistically significant. Thus, L. monocytogenes invades villous intestine and Peyer’s patches with similar efficiencies but grows much more rapidly in Peyer’s patches. Bacterial counts 3 h after inoculation of ca. 109 CFU per loop were similar for CLIP1162 (Table 3), suggesting similar mechanisms of invasion for L. monocytogenes and L. innocua. However, CLIP1162 did not grow in Peyer’s patches or villous intestine (Table 3).
Listeria infection of Peyer’s patches and villous intestine. Loops were inoculated with EGD at doses of ca. 109, 108, and 107 CFU/loop. Intestinal tissue samples (Peyer’s patches or pieces of villous intestine of similar size) were recovered at 3 and 24 h postinoculation and processed to determine the number of gentamicin-protected bacteria. Data points and error bars represent the mean and standard deviation of log10 CFU per sample (mean of four rats for each point).
Histopathological and electron microscope study. (i) Villous intestine (Fig. 4).At early stages (≤3 h) of infection with EGD (dose of ca. 109 CFU per loop), myriad bacteria were seen in the intestinal lumen, apparently embedded in the mucus layer. There were also many clusters of 10 to 20 bacteria stuck to mucus released by goblet cells. Numerous bacteria were present in the lumen, but very few were seen associated with the epithelial cell surface. A moderate inflammatory reaction was observed, with leukocytes and erythrocytes in the intestinal lumen. This reaction was probably related to the burden of intraluminal listeriae, as it occurred even in areas where the epithelial layer was free of bacteria. No listeriae were observed crossing the epithelial barrier by TEM, although sections from different tissue samples collected from five animals were examined. Challenges with high bacterial concentrations (ca. 1010 CFU per loop) were no more successful for TEM analysis. SEM was not performed. At 24 h, small foci of infection were found in 1 of 100 to 200 intestinal villi. These foci were located in the chorion, at the tips of villi. The overlying epithelium never contained bacteria. A typical lesion was populated by a few inflammatory cells, exclusively consisting of mononuclear cells, and contained no more than 10 to 20 bacteria. Local inflammatory reaction was moderate.
Histological and immunohistological analysis of villous intestine 1 h after inoculation with EGD. (a) Immunofluorescence labeling of listeriae. Bacteria are abundant in the intestinal lumen, apparently embedded in mucus; note that very few bacteria are close to the epithelial cell surface. (b) Gram-Weigert staining. Aggregates of bacteria are stuck to mucus released by goblet cells; no bacteria are seen associated with the epithelial cell surface. Bars = 10 μm.
(ii) Peyer’s patches (Fig. 5).Histopathological analysis was consistent with the quantitative data obtained by counting gentamicin-protected bacteria. At early stages of infection (≤3 h), a few bacteria were observed associated with the FAE. The lack of significant interaction of listeriae with the FAE was also demonstrated by SEM (data not shown). Gram staining produced very few pictures of bacteria crossing the epithelial layer (one or two pictures per follicle and per section). However, as we used normal and not germ-free animals, it is possible that the organisms seen crossing the FAE were not listeriae. We did not find any bacteria with the same morphology as listeriae in the process of entering cells in TEM studies, although about 2,000 epithelial intestinal cells and 500 M cells were examined carefully. Similar results were obtained using a challenge of 1010 CFU per loop. Thus, L. monocytogenes does not penetrate Peyer’s patches very efficiently.
Histological, immunohistological, and TEM analysis of Peyer’s patch infection with EGD. (a and b) Gram-Weigert staining. (a) Six-hour infection (a few bacteria are seen in the dome area); (b) 24-h infection (many bacteria can be seen in the follicular tissue; note the major inflammatory reaction consisting of both mononuclear cells and neutrophils); bars = 5 μm. (c and d) Twenty-four-hour infection, double-fluorescence labeling of listeriae (c) and the macrophage-related antigen ED1 (d). A very large number of bacteria infect the follicular tissue except the germinal center; bars = 20 μm. (e and f) Twenty-four-hour infection, TEM of listeriae located within a mononuclear cell (e) and listeriae inside a neutrophil (f); bars = 1 μm.
As early as 6 h postinoculation, discrete, small foci of infection were detected in the dome area, just beneath the FAE. The bacteria were located inside mononuclear cells at this stage. At 24 h, the follicular tissue was infected with large numbers of bacteria. These bacteria were restricted to the dome area and the interfollicular region. They were never seen in the germinal center. The intensity of lesions varied greatly between the follicles of a given Peyer’s patch. The inflammatory reaction consisted of purulent to pyogranulomatous lesions. The underlying epithelial sheet was intact or had small erosions but, in some cases, was detached completely. Closer analysis by TEM showed that both mononuclear cells and neutrophils contained listeriae.
(iii) Infection with lower doses.Histopathological lesions were analyzed 24 h after inoculation with 108 and 107 bacteria per loop. No lesions were found in villous intestine (200 intestinal villi examined for each dose) with either dose. The intensity of lesions in Peyer’s patches was related to the dose used for inoculation. Only 20% of lymphoid follicles contained bacteria with the lowest dose. Bacteria were few in number and were associated with a few inflammatory cells, mainly mononuclear cells. The proportion of infected lymphoid follicles was about 50% for inoculum of 108 bacteria per loop. Infective foci were larger, and inflammatory cells consisted of both mononuclear cells and neutrophils. With either dose, listeriae were seen exclusively in the dome area and the interfollicular region.
Role of virulence factors.The roles of inlAB,hly, and actA in the intestinal step ofListeria infection were analyzed by studying deletion mutants for these genes in the ligated loop system. These mutants translocated to MLN, liver, and spleen as efficiently as their parental strains, as expected from the results obtained with L. innocua CLIP1162 (Fig. 6). Only thehly mutant did not grow within these organs (Fig. 6). There were no significant differences in early bacterial counts in Peyer’s patches and villous intestine between the actA andhly mutants and their parental strain (Fig. 6). However, the ability to grow in both Peyer’s patches and villous intestine of thehly mutant and, to a lesser degree, the actAmutant, was affected (Fig. 6). The inlAB mutant, BUG949, appeared 5- to 10-fold less invasive than its parent, but the difference did not reach a significant P value. Coinfection experiments were carried out to compare BUG949 and its parent, BUG600, in more detail. Intestinal loops were inoculated with a 1:1 mixture (ca. 109 CFU per loop) of BUG949(pAT18) and BUG600(pAT28) or of BUG949(pAT28) and BUG600(pAT18). Bacterial counts in tissue samples (Peyer’s patches, villous intestine, MLN, liver, and spleen) were determined for each strain, using erythromycin and spectinomycin as selective markers. Contrasting with previous results with separate animals, there was no difference between the inlAB mutant and its parent at any time point and in any tissue or organ (Table4).
Role of Listeria virulence factors. Loops were inoculated with the mutants and their parental strains at a dose of ca. 109 CFU/loop, and the numbers of bacteria in Peyer’s patches, villous intestine, MLN, liver, and spleen were determined at 3 and 24 h postinoculation. The mean and standard deviation of log10 CFU per organ are shown (mean of four rats for each point).
Coinfection experiments with BUG600 and BUG949 derivatives expressing differential selective markersa
The histopathology of Peyer’s patches at 24 h was consistent with quantitative data. The inlAB mutant gave lesions that were indistinguishable from those produced by its parental strain (data not shown). The infectious foci were infrequent and smaller with theactA mutant, suggesting that this mutant did not disseminate in the follicular tissue (Fig. 7). No foci of infection were observed in the specimens obtained from loops inoculated with the hly mutant (data not shown).
Histology and immunohistology of Peyer’s patches 24 h after infection with the actA mutant. (a) Immunofluorescence labeling of bacteria. The foci of infection are less numerous and smaller than with EGD; they are restricted to the dome area. (b) Gram-Weigert staining. Discrete foci of bacterial replication inside mononuclear cells can be seen (arrows). Bars = 20 μm.
DISCUSSION
The sequence of events leading to the invasion of the host byL. monocytogenes via the intestine was examined in a rat ligated ileal loop system. We used an in vivo adaptation of the gentamicin survival assays developed for studying bacterial invasion of cell monolayers. Bacteria were kept inside ligated loops for 1 h to allow bacterial invasion. The loop were then deligated, washed, and treated in situ with gentamicin to eliminate the extracellular bacteria present in the gut. These experimental conditions were aimed at preventing nonspecific phenomena that might have interfered withListeria invasion. L. monocytogenes produces several factors that are highly toxic to cells and are potent proinflammatory agents (28). Therefore, the presence of large numbers of listeriae inside the intestinal lumen for long periods, a situation which is probably rare in cases of human listeriosis (see below), may result in profound epithelial damage. Epithelial lesions may, in turn, artificially promoteListeria invasion by rendering the basolateral surface of intestinal cells accessible to bacteria. It has been shown that entry of L. monocytogenes into polarized monolayers of Caco-2 intestinal cells is greatly increased by disrupting intercellular junctions (14). In experimental shigellosis, the invasion of the intestinal mucosa by shigellae is promoted by the inflammatory reaction triggered by the few organisms that cross the intestinal barrier via M cells (25). The recruitment of neutrophils in situ makes bacterial invasion possible by destroying the cohesion of the epithelial barrier and rendering the basolateral surface of intestinal cells accessible to intraluminal shigellae. These findings are highly relevant as shigellosis is essentially an acute inflammatory bowel disease. In contrast, listeriosis is mainly a systemic infection transmitted by the intestinal route rather than an intestinal disease per se.
We found no evidence of a preferential involvement of the FAE, and more specifically of M cells, in the passage of L. monocytogenesacross the epithelial barrier. The presence of Peyer’s patches in ligated loops did not affect the rate of Listeriatranslocation to deep organs. The levels of uptake of listeriae by Peyer’s patches and villous intestine were similar and low. Gentamicin-protected listeriae in either organ 3 h after inoculation with 109 CFU amounted to ca. 50 to 250 CFU per cm2 of tissue. Much higher values were recently reported for Shigella flexneri, which initially invades the intestinal mucosa via M cells in experimental shigellosis (25,37). Counts of gentamicin-protected shigellae in Peyer’s patches reached ca. 15,000 CFU/cm2 of tissue 2 h after the inoculation of rabbit intestinal loops with ca. 5 × 109 CFU (25). TEM analysis of the early stage ofListeria infection was consistent with the very low numbers of listeriae recovered from tissues after gentamicin treatment. Despite the examination of many grids prepared from both villous intestine and Peyer’s patches, we did not find any organisms resembling listeriae inside absorptive epithelial cells or M cells.
The passage of L. monocytogenes across the epithelial sheet does not seem to require any of the virulence factors involved in infection of epithelial cells in vitro. Mutants with deletion ofinlAB, hly, and actA, the main virulence genes so far identified in L. monocytogenes, did not have reduced translocation. L. monocytogenes and the nonpathogenic related bacterial species L. innocuadisseminated to deep organs with similar efficiencies. Consistent with these findings, L. monocytogenes was recovered from MLN, liver, and spleen as early as 15 min after inoculation. Thus, the translocation process may occur without previous intraepithelial replication. Our results are in full agreement with recent studies in mice that failed to detect Listeria-specific antigens within intestinal epithelial cells after oral challenge (21) but are in apparent contradiction to a previous electron microscopy study of guinea pigs by Racz et al., showing replicating listeriae inside absorptive intestinal cells (29). This discrepancy may be due to differences in experimental procedures: Racz et al. used starved, opium-treated guinea pigs; enteric lesions were not observed in nonpretreated control animals.
Our results suggest that L. monocytogenes uses nonspecific mechanisms to cross the epithelial sheet. Virtually all oxygen-tolerant bacterial species are able to translocate from the gastrointestinal tract to MLN and other extraintestinal sites in various animal systems (4). Intestinal bacterial growth is the major mechanism promoting bacterial translocation (34). E. colicontinuously seeds the MLN as long as it maintains a population of at least 108 CFU per g of cecum in monoassociated gnotobiotic mice (5). The number of E. coli translocating to the MLN decreases when E. coli populations are reduced in the cecum and small intestine, by introducing an antagonistic indigenous microflora (5). Not all bacterial species translocate at the same efficiency. Enterobacteriaceae andPseudomonas aeruginosa translocate 5 to 10 times more efficiently than gram-positive organisms such as Staphylococcus epidermidis and Lactobacillus brevis (4,34). This is consistent with the low degree ofListeria translocation that we detected despite the use of high bacterial doses and the large reduction in the number of listeriae translocating to deep organs when inocula were reduced to less than 108 bacteria per loop. The mechanisms of translocation promoted by bacterial overgrowth are unknown. Both intracellular and paracellular passages of microorganisms have been suggested (4).
The ability of L. monocytogenes to replicate within Peyer’s patches was a major finding of this study. The rate of bacterial growth during the first 24 h was 10 to 20 times higher in Peyer’s patches than in villous intestine regardless of dose. Histology was consistent with the results of bacterial counts. Infection of Peyer’s patches was detected throughout the experiments and consisted of multiple, progressively confluent infectious foci, whereas infectious foci of the villous intestine were infrequent and small. At early stages of infection, listeriae were found exclusively within mononuclear cells in both tissues. This finding suggests that a subpopulation of mononuclear cells in Peyer’s patches, and more precisely in the subepithelial dome, is highly permissive to L. monocytogenes. Both resident macrophages and dendritic cells are plausible candidates. There is a population of long-living, major histocompatibility complex class II-negative, sialoadhesin-positive cells expressing macrophage markers in the subepithelial dome of Peyer’s patches in rodents (7). These cells have a deactivated phenotype, and their microbicidal activity is thus likely to be low.
Cells of the dendritic lineage may also be key target cells in Peyer’s patches. A population of dendritic cells that form a dense layer just beneath the dome epithelium has recently been identified in murine Peyer’s patches (19). These cells are thus well positioned for interaction with listeriae passing into Peyer’s patches. The ability of this population to internalize bacteria was not evaluated, but other studies have demonstrated that cells of the dendritic lineage may naturally have, or acquire, phagocytotic functions (30). A recent study has shown that L. monocytogenes invades mouse dendritic cells in vitro and develops a complete infection cycle within these cells, with escape from the phagocytosis vacuole and cell-to-cell spread (17). Invasion of dendritic cells could result in either cell death or persistent infection. Dendritic cells may thus be a reservoir for L. monocytogenes. Infected dendritic cells may also contribute to the spread of L. monocytogenes into the host as a result of their status as antigen-presenting cells. Indeed, dendritic cells home to intestinal and extraintestinal lymphoid tissues to induce a primary T-cell response after contact with antigens in the periphery (35).
These data suggest two possible pathways for Listeriatranslocation. Passage across the epithelial sheet is common to both pathways and does not require any Listeria-specific virulence factor. This is an inefficient process directly related to amount of bacteria in the gut. After crossing the epithelial sheet, listeriae either spread via the lymph and blood to distant tissues in a few minutes or are taken up by mononuclear cells in the subepithelial region. The number of listeriae inside mononuclear cells seems to depend on the killing activity of the cell population encountered by these organisms. In the chorion of intestinal villi, nonpermissive cells may be involved, resulting mostly in abortive infection. In Peyer’s patches, listeriae may invade weakly bactericidal mononuclear cells, either resident macrophages or dendritic cells, and replicate almost without restriction.Listeria virulence factors involved in intracellular growth and cell-to-cell spread are necessary for this step. Listeriae may subsequently enter lymph or blood vessels and gain access to other intestinal and extraintestinal tissues. It is unknown whether bacteria are transported by mononuclear cells or travel extracellularly (3).
A large outbreak of gastroenteritis due to L. monocytogenesin chocolate milk has recently been reported (8). The most common symptoms in the affected people were diarrhea and fever. No infection of extraintestinal sites was observed. The chocolate milk implicated in the outbreak contained very high levels of listeriae, and the median dose ingested was estimated to be as high as 2.9 × 1011 CFU per person. This may explain why infection presented as a gastrointestinal illness in this outbreak whereas gastrointestinal symptoms are not typical in most cases of listeriosis. These data are consistent with our results, as we found that gross intestinal lesions developed only after inoculation with very large doses of listeriae (≥109 CFU per loop) in our system. The mechanisms by which L. monocytogenes causes diarrhea in humans are unknown. The association of fever with diarrhea and the demonstration of high levels of serum antibody to listeriolysin O in infected individuals suggest an invasive intestinal process (8). This is also consistent with our experimental results.
Peyer’s patches are usually regarded as the “Achilles heel” of the mucosal barrier because of the capacity of M cells to actively take up particles, including viruses, bacteria, and protozoan organisms (33). Little attention has been paid to the lymphoid tissue, which is another area of weakness in the gut. This tissue contains cell populations that may be used by a large variety of pathogenic organisms to establish local or general infections. For example, some viruses that cause systemic diseases, such as reovirus, cross the epithelial barrier through M cells, replicate in M-cell-associated mononuclear cells, and then enter the host circulatory system (2). Infected B cells transport mouse mammary tumor virus to the mammary gland (23). Human immunodeficiency virus type 1 may initially encounter its CD4+ target cells in lymphoid tissue present in the rectal mucosa (1). L. monocytogenes and its virulence factors may exploit mononuclear cells, macrophages, or dendritic cells, which are dedicated to antigen processing and presentation but not to bacterial killing. This possibility is currently being studied by our group.
ACKNOWLEDGMENTS
We thank M. Leborgne and G. Pivert for technical assistance, G. Milon for helpful discussions, P. Cossart and T. Chakraborty for the gift of strains, and R. Fournier for help with the manuscript.
B.P. received financial support from the Institut National de la Santé et de la Recherche Médicale. This work was supported by the University Paris V and the CEE (grant BMH4CT96 0659/RA03813).
FOOTNOTES
- Received 3 September 1997.
- Returned for modification 22 October 1997.
- Accepted 22 November 1997.
- Copyright © 1998 American Society for Microbiology
