Segmented Filamentous Bacteria Prevent Colonization of Enteropathogenic Escherichia coli O103 in Rabbits

Abstract

Despite their distribution in the intestines of many mammals, including man, segmented filamentous bacteria (SFB) have not been found in rabbits, nor has any function been identified for these uncultivable microbes. New Zealand White rabbits were infected with rabbit enteropathogenic Escherichia coli O103 (REPEC O103) derivatives, followed up clinically, and randomly killed 1–4 days after inoculation. Intestinal tissue samples were examined by electron and light microscopy to search for SFB and to evaluate REPEC O103 colonization. Twelve of 21 rabbits showed SFB colonization on ileal absorptive villi. The presence of SFB was correlated with lack of REPEC 0103 ileal colonization (P < .01) and disease. Rabbits without SFB were always colonized by this pathogen. SFB appear to inhibit intestinal colonization by REPEC O103 and thus protect against REPEC 0103 disease. SFB colonization in rabbits is also described for the first time.

Segmented filamentous bacteria (SFB), recently classified as Candidatus Arthromitus, can colonize the ileal villi of a variety of mammals and birds during weaning [1–4]. To date, SFB cannot be cultivated but can be detected by light and electron microscopy [5]. They attach via a nipple-like holdfast segment but do not provoke inflammatory reactions, nor do they have cytotoxic effects on epithelial cells [6]. SFB are host-specific, sensitive to a wide range of antibiotics, and considered to be part of the indigenous nonpathogenic intestinal flora. They exhibit a complex life cycle and are randomly distributed within the population [7–9]. They are thought to stimulate the host immune response by elevating IgA levels in intestinal tissues and serum [10–13]. Garland et al. noted a possible protective role against infection with Salmonella enteritidis in rats [14]. However, interference with other enteropathogens has not been described, hence their role as potential antagonists needs to be determined. In addition, SFB have not been detected in rabbits [2].

Several investigators described ileal Peyer's patches (PP) as the preferential site for SFB colonization. Attachment to absorptive villi, domed villi, or both is species-specific, and occurs on the surface of absorptive enterocytes, M cells, and at tight junctions [1, 15–22]. Electron-dense attachment zones resulting from actin accumulation were seen beneath invaginated SFB [15, 23]. In some ways, these structures resemble those observed after attachment of enteropathogenic Escherichia coli (EPEC), in which actin rearrangement results in pedestal formation beneath bacteria [24–26]. The noninvasive organisms EPEC and rabbit EPEC, RDEC-1, as well as another rabbit EPEC strain, REPEC O103 (unpublished data), attach to ileal PP during early stages of infection [27]. Similarly, invasive pathogens such as Vibrio, Campylobacter, Yersinia, and Salmonella species initially attach to ileal PP before being presented to the underlying lymphatic tissue [27]. It is therefore of interest to determine whether, and in what way, SFB interfere with the colonization of these enteropathogens as they attach to the same site in the host.

In our laboratory, studies have been underway to investigate the pathogenesis of EPEC. In developing countries, EPEC is the leading cause of bacterial-mediated diarrhea [28–30]. Several virulence factors have been described recently, and animal models have become useful tools to study EPEC-mediated disease and to apply in vitro observations to the in vivo setting. Infection of weaned rabbits with rabbit enteropathogenic Escherichia coli, RDEC-1 (serotype O15:H-), and REPEC O103 (serotype O103:H2:K-) results in diarrhea and disease. Experimental infection with both rabbit EPEC strains in our laboratory resulted in unsuccessful colonization of a varying, yet large, number of rabbits. Investigators who experienced similar findings assumed that they might be witnessing a natural attenuation of the rabbit strains, which had inexplicably lost their level of virulence for rabbits [31]. REPEC O103 has been isolated as a highly pathogenic strain predominantly in rabbits in Spain and France [32, 33]. SFB have been reported predominantly in the United States, the United Kingdom, and the Netherlands. This inverse correlation led us to examine whether SFB may inhibit REPEC O103 infection.

Here we document the presence of SFB in rabbits. We also found that the presence of SFB correlated perfectly with the lack of disease in rabbits inoculated with REPEC O103. These results indicate that SFB provide a potential protective mechanism against gastrointestinal infections.

Methods

Bacterial strains

Rabbit EPEC strain 85/150 was kindly provided by Dr. Johan E. Peeters (National Institute of Veterinary Research, Belgium). It is a rhamnose-negative strain, serotype O103:K-:H2 (REPEC O103). Detailed characteristics of the strain have been described elsewhere [34, 35]. From this wild-type strain, we constructed nalidixic acid-resistant derivatives: REPEC O103 wild type and 2 mutants lacking the virulence genes espA and espB [35].

Rabbit infection and experimental design

Weaned female New Zealand White rabbits, aged 30–55 days and weighing between 0.5 and 1.6 kg, were obtained either from the R&R Rabbitry, Stanwood, Washington, or, specific-pathogen-free (SPF), from Charles River, St. Constant, Quebec, Canada. Before experimental inoculation, rabbit serum and stool samples were tested and confirmed free of REPEC O103 antibodies or strains, respectively. For infection, rabbits were anesthetized and inoculated via an orogastric tube with 10⁹ bacteria per rabbit, as described elsewhere [35]. Six rabbits were challenged with the wild type strain, 5 rabbits with the EspA mutant, and 5 rabbits with the EspB mutant. Five control rabbits received only PBS. Animals were watered and fed ad libitum before and after infection.

Fecal bacterial shedding of the inoculated strain was followed up daily until the rabbits were killed. For the specific identification of the test strains, fecal suspensions were spread on MacConkey plates containing nalidixic acid. To monitor shedding of other fecal microorganisms, suspensions were, in addition, spread on plain MacConkey, Luria-Bertam, brain-heart infusion, and inhibitory mold agar plates. Colony forming units (cfu) of varying phenotype were Gram's stained and evaluated microscopically. For light and electron microscopic studies of intestinal tissues, rabbits were randomly killed 1,2, 3, or 4 days after inoculation. Tissue samples from ileum, ileal PP, cecum, and proximal colon were removed for histologie processing.

Tissue preparation for light and electron microscopy

Animals were killed by intravenous injection of ketamine and an overdose of sodium pentobarbital. Intestinal tissues were excised immediately after death, thoroughly washed 3 times in PBS to remove nonadherent bacteria, and further processed as described elsewhere [35]. Specimens for light microscopy were stained for comparative studies with hematoxylin-eosin (HE), periodic acid-Schiff (PAS), Gram's stain, and silver stain (Steiner).

Tissue examination by electron microscopy

To evaluate the correlation between SFB establishment and failure of REPEC O103 colonization in the distal ileum, absorptive ileal villi were examined by scanning electron microscopy (table 1). Per tissue sample (i.e., per rabbit), a total of 10 villi allowing maximal accessibility for microscopic examination were randomly chosen and carefully screened on all sides. Visible attached SFB and adherent REPEC O103, respectively, were counted on each villus, and their mean colonization level per villus was calculated. The score for SFB attachment per villus was determined as follows: —, no SFB; +, occasional presence of SFB; ++, <1 x 10¹ SFB; + + +, <1 × 10² SFB; + + + +, <1 × 10³ SFB; and + + + + +, >1 × 10³ SFB. The score for REPEC O103 adherence per villus (cf. morphological features of the pathogen [35]) was determined as follows: —, no adherent bacteria; +, occasional presence of adherent bacteria; + +,<1 × 10¹ adherent bacteria; + + +, <1 × 10² adherent bacteria; and + + + +, <1 × 10³ adherent bacteria.

Table 1

Level of segmented filamentous bacteria (SFB) establishment in the distal ileum of rabbits and correlation between SFB and rabbit enteropathogenic 0103 Escherichia coli (REPEC O103) colonization after infection with REPEC O103 derivatives, evaluated by scanning electron microscopy.

Statistical analysis

In table 1, Fisher's exact test was used to determine whether the presence of SFB was correlated with the lack of REPEC O103 colonization in the distal ileum; P < .05 was considered statistically significant.

Results

SFB are present on the rabbit intestinal surface

Light and electron microscopic examination revealed SFB present on the intestinal tract in 12 of 21 rabbits (figures 1–3). Organisms were seen in rabbits of all age groups examined (32–60 days), as well as in controls and those infected with REPEC O103 derivatives. Filamentous bacteria varied in number and were found predominantly on the absorptive villi of the distal ileum (figure 1B and 1D), and in slightly lower numbers on the absorptive villi of the ileal PP (figure 1C, 1E, and 1F).They were found free in the lumen or were attached to enterocytes (figure 2). Organisms regularly covered the villus surface in a netlike array or, when present in high numbers, were arranged in brushlike structures (figure 1B). Sides and tips of absorptive villi were colonized equally. Domed villi of PP were always free of the organisms (figure 1E). In 3 rabbits, filamentous bacteria were seen occasionally in the cecum or proximal colon. These findings show that SFB colonize rabbits in a manner similar to that reported for other mammals. In the present study, SFB were detected in ∼50% of the rabbit population.

Figure 1

Scanning electron micrographs of ileal surfaces from different rabbits. A, Absorptive ileal villi (v) with normal appearance in a rabbit free of segmented filamentous bacteria (SFB). B, SFB heavily colonize tips and sides of absorptive villi (v), providing the intestinal surface with a tufted appearance. C, SFB penetrate between adjacent enterocytes (arrow) or disappear into openings (arrowhead). D, Four days after experimental inoculation, rabbit enteropathogenic Escherichia coli 0103 cannot be seen where SFB are established on the intestinal surface. Filamentous organisms penetrate the intestinal cell either at their surface (arrow) or between enterocytes (arrowheads). E, SFB colonize absorptive ileal villi (a) of the Peyer's patches. Domed villi (d, in the background), however, are free of the organisms. F, SFB exhibit a smooth surface (s) with rounded ends (arrow) or are divided by transverse septa into barrel-shaped segments (b) with presumably broken ends (arrowhead).

Figure 3

Histological sections of rabbit ileum colonized with segmented filamentous bacteria (SFB). A, Gram's stain: bacteria are in the lumen or attached to absorptive villi. They stain gram-negative or gram-variable. B, Silver staining (Steiner): SFB attach to both columnar epithelial cells (light brown) and goblet cells (dark brown). C, Hematoxylin-eosin staining: ileal cell association shows a normal appearance. Inflammatory reactions in response to SFB attachment are not detected. D, Periodic acid-Schiff staining: mucus formation on the villus surface (purple) is not affected by SFB attachment.

Figure 2

Transmission electron micrograph of segmented filamentous bacteria (SFB) on the rabbit ileal surface. SFB are illustrated in horizontal (small round organisms) and vertical or diagonal cross-sections (long and oval organisms, respectively). One SFB attaches between an absorptive enterocyte (e) and a goblet cell (g) (arrowhead). Transverse septa divide the filamentous organisms into barrel-shaped segments (b). Intrasegmental bodies can have a triangular form representing the nipple-like “holdfast” segment (h). Original magnification, ×5000.

Table 2 summarizes the species-specific sites of SFB attachment and age-related establishment within the distal ileum of mammals and chicken. In most species examined, SFB attach to ileal absorptive villi (chicken, tonsilla cecalis). In mice, rats, and pigs, attachment occurs in addition to the domed villi, whereas SFB in horses attach exclusively to domed villi. SFB colonize the intestine 2 weeks after birth and remain during weaning. However, time course studies showed that they cannot be found in older animals (dogs, chickens, and pigs).

Table 2

Species-specific distribution of segmented filamentous bacteria within ileum and ileal Peyer's patches

Morphological features of SFB

The morphological features of rabbit SFB are similar to those described for other species [10, 17, 19, 22]. The filaments varied in length, from ∼2–100μm, and in diameter, from 0.7–0.9μm. Short filaments usually exhibited a smooth surface, with the free end tapered and rounded (figure 1F). In contrast, longer filaments were divided by multiple transverse septa resulting in formation of 1.1 μm short, barrel-shaped segments. The free ends of these filaments often exhibited a broken appearance, apparently where segments had sheared. Upon light microscopic examination, organisms stained gram-negative or gram-variable, and PAS-negative (figure 3). Silver staining showed SFB attachment to both columnar epithelial cells and goblet cells. PAS and HE staining showed that the continuous mucus layer present on absorptive villi surfaces was still intact where SFB were present, and the morphology of the intestinal cell layer was not altered. As expected, inflammatory reactions to SFB attachment were not observed. Electron microscopic examination revealed that the filamentous organisms penetrated absorptive enterocytes, M cells, and goblet cells on their surface or where epithelial cells were adjacent to each other (figure 1C–1F). The attached ends appeared to embed between the microvilli that were pushed aside and sometimes disappeared in the depth where goblet cells had openings (figure 1C). Transmission electron micrographs illustrated the presence of intrasegmental bodies and the formation of triangular holdfast segments thought to intimately embody SFB in its host (figure 2) [19].

The present morphological findings indicate that rabbit SFB exhibit a complex life cycle and are related to SFB previously described in other species. The lack of immune response reactions to SFB attachment indicates that they are part of the indigenous intestinal young rabbit flora.

SFB inhibit REPEC O103 infection

In previous studies, we showed that REPEC O103 colonizes the distal ileum of weaned rabbits within 48 h after infection, before it spreads to the lower intestine (unpublished data). Within 72 h, this pathogen triggers characteristic attaching and effacing (A/E) lesions with effacement of intestinal microvilli in all the ileum, ileal PP, and the proximal colon, resulting in severe diarrheal disease. REPEC O103 derivatives lacking the virulence proteins EspA or EspB do not cause A/E lesions but are still able to adhere to and colonize the intestinal surface [35].

In this study, microscopic examination revealed that in each infection group (REPEC O103 wild type, ΔespA, or ΔespB), ileal absorptive villi were colonized by these bacteria in 4/6, 2/5, and 3/5 rabbits, respectively. Villi of the lower intestine were colonized in 6/6, 3/5, and 3/5 rabbits, respectively. In the affected wild type group, animals shed ⩽10⁹ cfu of the inoculated pathogen per gram of feces and started to experience diarrhea within 3 days after infection. In both mutant groups, fecal shedding of the inoculated pathogen reached 10⁴ cfu per gram of feces within the same time.

In the remaining animals (2/6, 3/5, and 2/5 rabbits, respectively), ileal villi, however, were not colonized by the pathogen. Instead, SFB were present in varying number. They covered sides and tips of the villi, sometimes exhibiting a dense tuft on the epithelial surfaces. With the exception of 2 rabbits in which SFB were only occasionally observed (wild type, 1/6; ΔespA, 1/5), the inoculated pathogens were never seen together with the filamentous organisms, nor were they hidden in the depth of the tufted arrangements. REPEC O103 strains eventually adhered to epithelial surfaces of the lower parts of the intestine in these rabbits. Fecal shedding, however, never reached >10² cfu per gram feces, and diarrheal symptoms did not develop. Evaluation of total fecal microbial content showed the presence of gram-positive and gram-negative coccoids and rods. Filamentous bacteria, however, could not be cultured.

Taken together, there is a significant correlation between the presence of SFB and the lack of colonization by REPEC O103 (P = .0032; see table 1). At the same time, successful colonization with the REPEC O103 derivatives only occurred when SFB were not found on the intestinal surface. These findings indicate that SFB play a protective role against REPEC O103 adherence and subsequent disease.

Discussion

In the present study, we showed the presence of SFB in the rabbit intestine. The lack of host defense mechanisms in response to SFB colonization, together with ultrastructural features, indicate that they are tolerated as part of the indigenous intestinal rabbit flora in the distal ileum. Unsuccessful colonization with REPEC O103 in the presence of SFB indicates that SFB antagonize the adherence of this enteropathogen, possibly in a competitive way, preventing REPEC O103 mediated disease and possibly other enteric infections.

SFB have been described in several vertebrate species, including humans. The failure of previous investigations to find them in rabbits might be due to the examination of only a small number of animals [2]. In addition, it is known that SFB colonization can be influenced by host characteristics and environmental factors, such as stress and diet, and therefore is not obligatory in all animals all the time [10].

Previous studies in rats have shown that SFB preferably colonize PP domed villi commencing 2 weeks after animal birth. Similar observations have been made in mice, pigs, and horses (table 2). In our study, however, we found that SFB in rabbits colonized exclusively the absorptive villi similar to that observed in dogs, cows, and fowl. It is unclear why SFB colonization is different among these animal species, yet it might be due to insufficient examination of intestinal tissues. However, in all cases, filamentous organisms only colonize the very short part of the distal ileum including the ileal PP and tonsilla cecalis (chickens), respectively, areas where lymph follicles are aggregated and where enteropathogens interact with host cells.

Our own studies, together with previous investigations, have shown that SFB mainly colonize the distal ileum including the ileal PP. Colonization occurs during the age of weaning when the immune system matures (table 2). In some species (dogs, chickens, and pigs) SFB were found exclusively during the first months after birth but not later (to date, time-course studies longer then the period of weaning were not performed in other species). Similar to this colonization pattern, EPEC (EPEC, RDEC-1, and REPEC O103) utilize ileal PP for initial adherence, probably for replication, before they spread into the lower parts of the intestine. These pathogens cause severe diarrhea in weanlings, whereas adults are much less or even not susceptible. It is known that bacterial antagonism is an important defense mechanism against microbial infections [36]. Indigenous bacteria can inhibit adherence and replication of enteropathogens, and, subsequently, disease, in a competitive way [37]. In addition, they can increase the bowel motility propelling pathogens quicker through the lumen [38]. In our laboratory, experimental REPEC O103 infections were at times impaired and inconsistent. SFB may actually prevent colonization of this pathogen, based on our findings in this study that REPEC O103 fail to colonize the ileum where SFB are present. SFB may help to consolidate the immune system during the age of weaning, but they may also prevent colonization of pathogens in a competitive and mechanical way. Since other enteropathogens, such as Salmonella, Shigella, Campylobacter, and Yersinia species, choose ileal PP for initial contact with the host before they are presented to the underlying lymphoid tissue, SFB may play a key role in host resistance to a variety of enteric pathogens.

Most REPEC O103 studies have been done in Europe, where this highly pathogenic microorganism was first isolated in 1988 from French and Spanish rabbitries. In contrast, to date there exist no reports about natural infections of North American rabbits with REPEC O103. In addition, our own studies, as well as work done by other groups, led us to assume that rabbit EPEC strains in North America were experiencing “an inexplicable loss of their high level of virulence” during experimental infections in North America [31]. At the same time, most studies about SFB establishment were done in the United States, and several more in the Netherlands, the United Kingdom, and Australia. Our suggestion that SFB prevent REPEC O103 colonization may explain the transient infection rate in North America. The finding that SFB in rabbits obtained from 2 different sources (conventionally kept rabbits from Seattle and SPF-rabbits from Ontario, Canada) indicates that SFB are widespread in North America.

We postulate that SFB in rabbits and, apparently, in other vertebrates may be potent antagonists of enteropathogens. Further studies should be done to evaluate the exact role SFB play in affecting infectious diseases, potentially further establishing a protective function for these filamentous organisms.

Acknowledgments

We thank Dr. Elaine Humphrey for her excellent technical advice and assistance with the electron microscopy studies; Dr. Johan E. Peeters for providing strain REPEC O103 (85/150); Julie Chow for her technical assistance with the histology studies; and Dr. Robert De-Wreede and the University of British Columbia Statistics Department for their help with the statistical analysis.

Animal experiments were performed in accordance with the guidelines of the Canadian Council of Animal Care and the University of British Columbia.

Financial support: Fonds zur Förderung der wissenschaftlichen Forschung, Austria (J1632-MOB); Naito Foundation/Japan; Howard Hughes Medical Institute International Research Scholar (HHMI 75197-526604); Medical Research Council of Canada (MT 13452).