Animal models of Campylobacter jejuni colonization and disease and the lessons to be learned from similar Helicobacter pylori models
2001; Oxford University Press; Volume: 90; Issue: S6 Linguagem: Inglês
10.1046/j.1365-2672.2001.01354.x
ISSN1365-2672
Autores Tópico(s)Viral gastroenteritis research and epidemiology
Resumo1. Summary, 57S 2. Introduction, 57S 3. Criteria for animal models, 58S 4. Naturally acquired human campylobacteriosis and volunteer studies, 58S 5. Naturally acquired campylobacteriosis in animals, 59S 6. Experimental challenge of laboratory animals, 59S 6.1 Non-human primate models, 59S 6.2 Ferret models, 59S 6.3 Pig models, 60S 6.4 Rodent models, 60S 6.5 Rabbit models, 61S 6.6 Chicken model of colonization, 61S 7. Can we learn anything from animal models of Helicobacter pylori? 62S 7.1 Manipulation of the bacterium, 62S 7.2 Use of related bacterial species and their natural hosts, 63S 7.3 Manipulation of the host, 63S 8. Conclusions, 64S 9. Acknowledgements, 64S 10. References, 64S The mechanisms by which Campylobacter jejuni induces disease in human beings remain unknown. Identification of campylobacter virulence factors requires appropriate animal models. Several in vivo models of disease have been described. Models in non-human primates are closest to the disease in humans but are excluded on ethical grounds. An oral ferret model, inducing mild diarrhoea, has proved promising in the investigation of virulence factors but may be difficult to use widely. The same is true for piglet models. Rabbit models reported generally involve surgical intervention and abnormal routes of administration, and are consequently of limited accessibility and usefulness. Several mouse models have recently been described; one model with orally challenged SCID mice induced diarrhoea. However, the frequency of disease was low. Moreover, immune-compromized mice would have restricted usefulness for immunological studies. The use of alternative challenge routes in mice, such as via the nasal mucosa, may be of value if the effects are reproducible. In summary, models of campylobacteriosis remain unavailable. Acceptable models of colonization have been developed; the most frequently used is the orally challenged one-day-old chick. Identification of bacterial factors important in the colonization of chickens may lead to the development of targeted intervention strategies to reduce contamination of the food chain by this pathogen. However, the value of models of avian colonization to investigate human infection has yet to be established. Comparison of animal models of the related pathogens C. jejuni and Helicobacter pylori has been informative. Although faced with similar problems, acceptable animal models of several disease manifestations of H. pylori infection are now available. This experience suggests that appropriate models of campylobacteriosis can be developed. Approaches involving the manipulation of both the pathogen and host are suggested which may enable the virulence of C. jejuni to become detectable in animal models in the near future. Campylobacter jejuni and Helicobacter pylori have many common morphological, biochemical and physiological features. The relationships between these bacteria and their hosts are, in many respects, comparable. They colonize the gastrointestinal tracts of human beings, both inducing clinical disease in some individuals, and acting like commensals in others. On the other hand, the diseases, ecological niches and pathophysiology of these infections are quite different. The importance of C. jejuni and H. pylori infections in human beings has been recognized for about the same length of time: since 1977 (66) and 1984 (49), respectively. Nevertheless, considerably more progress has been made on understanding the molecular basis of the latter than the former. One reason for this has been the development and extensive use of a spectrum of animal models of H. pylori colonization and disease. These models have recently been comprehensively reviewed (30; 20; 44). In contrast, there has been a paucity of similar models for C. jejuni (29). In this review the current status of animal models of C. jejuni infection, and what lessons can be learned to develop further in vivo models using H. pylori as a paradigm, will be discussed. Acceptable animal models of bacterial pathogens, such as C. jejuni, are essential to: a) confirm the identification and enable the characterization of putative bacterial virulence factors b) validate in vitro models of pathogenic mechanisms c) investigate the role of host mechanisms in the induction of clinical symptoms d) determine host immune responses and investigate surrogates of protective immunity e) measure and test the efficacy of therapeutic or prophylactic treatments such as vaccines or antibiotics Optimally, in vivo models, in order to be reproducible and verifiable, should utilize small laboratory animals, which are specific-pathogen free, genetically homogenous, of known immune status and widely available. Obviously the model should mimic the human disease as closely as possible. The route of challenge, infective dose, clinical symptoms, colonization characteristics and pathophysiology (79) are all important. Such characteristics should mimic those in human beings, if possible, without the need to compromize host defences. Because the model is likely to be required to investigate host immunity, either as a surrogate of vaccine development or as an indicator of immunopathological events, a well-characterized immune system is preferable. Similarly, if the model is to be used to investigate therapeutic agents, then the pharmacokinetics of these agents should be similar to that in human beings. Moreover, in the postgenomic era, now relevant for both C. jejuni and H. pylori, animal models will be an essential component of strategies to understand functional genomics. In particular such studies will be used to: a) screen defined bacterial mutants for reduced colonization and virulence properties b) positively select bacterial genes expressed upon infection (in vivo expression technology; IVET) c) generate biomass for transcriptosomes and proteomes to detect mRNA and all polypeptides expressed during infection d) produce immunological reagents to identify bacterial components that are antigenic during infection Thus, for current research purposes, models which can detect quantitative differences in the virulence and colonization potential of wild-type strains and their mutants would be optimal. With many bacterial pathogens the choice of animal model is predictable from the observation of naturally occurring disease. The clinical features of human campylobacteriosis are complicated. In developing countries, where it is assumed that exposure is frequent, infection rarely causes disease symptoms except in children under 2 years of age. However, in industrialized countries, asymptomatic infection is rare and C. jejuni infection is generally characterized by 1–3 d of prodromal fever, followed by 5–7 d of acute diarrhoea with watery or bloody stools (72). This epidemiological picture indicates that prior exposure and the subsequent host immunological response can significantly affect the clinical outcome of infection. Such naturally acquired immune responses appear to protect from disease but not necessarily from colonization. Campylobacter-associated enteritis is frequently accompanied by fever and severe abdominal pain. Normally the infection is self-limiting and excretion is terminated within 10–14 d. Because of this, little information is available about the histopathology of such infections in man. Susceptible individuals, especially those who are immunocompromized, may develop chronic infections. Occasionally, severe complications, including bacteraemia, extra-intestinal infections and abortion, can arise apparently from host failure to limit the infection to the gut. Immunopathological sequelae are also recognized including arthropathies and neuropathies. Campylobacteriosis is now recognized as a common antecedent to the polyneuropathy, Guillain–Barré Syndrome (2). Other postinfectious symptoms, potentially requiring investigation in animal models, have been reported, including irritable bowel syndrome (50). Without doubt, human volunteer studies provide the best model of the human disease. Oral challenge of human volunteers with as few as 800 cfu organisms can induce diarrhoea (9). Such studies have confirmed the virulence of at least one of the laboratory strains, 81176, in regular use for in vivo models. These studies indicate that there is no clear relationship between dose and symptoms. Volunteers reported a spectrum of disease symptoms, from asymptomatic or mild to dysenteric illness. The severity of illness appears to be strain related. In ill individuals histopathology, using sigmoidoscopy biopsies of the rectal mucosa, indicated inflammatory cell infiltrate, with neutrophils in the crypts and lymphoid cells in the muscularis mucosa. The induction of protective immunity from disease on rechallenge, at least by homologous bacteria, was confirmed in these volunteer studies (9). These results also suggest that the immune response may not protect from colonization, confirming the epidemiological observations from developing countries. With the recent attempts to develop whole cell vaccines against C. jejuni (65), this human volunteer model has become crucial (59). Such a model is essential for testing vaccine toxicity and efficacy, and detecting immunological responses. Campylobacters can be isolated from the faeces of a wide range of wild, domestic and laboratory species. However, such colonized animals rarely demonstrate clinical signs of disease. The reason for this paradox is unknown. This lack of disease may reflect infection with strains that lack appropriate virulence factors and are therefore non-pathogenic, or the development of protective immunity following frequent exposure, as in humans in the developing world, or a lack of susceptibility, for example, as a consequence of the absence of appropriate host receptors for toxins. Nevertheless, outbreaks of campylobacter-associated disease can occur in some animals (52; 29). Campylobacter-associated enteritis was reported in captive breeding groups of non-human primates (62). Diarrhoea in companion animals, especially puppies and kittens, is not infrequent and is a well-recognized source of human infection (72). Enteritis associated with campylobacter infection in young domestic animals, such as piglets, calves and lambs, has also been reported but the organism is also found in healthy animals (67). Descriptions of naturally occurring disease in small mammals of laboratory status are rare but endemic campylobacteriosis has been reported in hamsters and ferrets (52; 29) and recently in rats (54). The rarity of these events suggests as yet unrecognized physiological components of the disease. Although birds, especially poultry, are naturally colonized by huge numbers of organisms, disease in avians has rarely been convincingly described, excepting occasional outbreaks of hepatitis which appear to be associated with C. jejuni infection. One further exception to this is campylobacter-associated death and enteritis in young ostriches (78). A wide range of laboratory animals have been challenged with C. jejuni in attempts to identify reproducible models of disease and/or colonization. Many of these models have been reviewed previously (52; 29). These early studies clearly indicated that, following oral challenge, colonization was relatively easy to establish in most laboratory mammals, especially in young animals such as infant mice, puppies and piglets. This experimental colonization was usually chronic and resulted in long-term intermittent excretion. Moreover, colonization, regardless of extent, was almost always asymptomatic. Over the last 10 years or so efforts have concentrated on the development of models which demonstrate a disease outcome from infection. Several such models have now been described with non-human primates, piglets, ferrets, rabbits and mice. The closest model to the human infection is oral infection of non-human primates, especially Macaca nemestrima (63). Challenged infant M. nemestrima, but not M. fascicilaris, developed vomiting and diarrhoea, with blood in their stools. Most animals had a bacteraemia. The onset and duration of disease was similar to that in human beings. The gross pathology and histology of these infections was also similar to that seen in human beings with acute campylobacter-associated colitis. Experimental infection induced acquired, and apparently protective, immunity. This model has been used to test the safety and immunogenicity of whole-cell, killed campylobacter vaccines (5), providing evidence for human clinical trials. Although this is apparently a good model of the naturally occurring human disease, there are sufficient problems to preclude the use of non-human primate models. In particular, the availability of such animals for experimental purposes, the facilities and skills required to house these animals, the variability in their immune status and ethical considerations, all prevent routine use. One promising model has been developed in 3–6-week-old ferret kits challenged orally with C. jejuni (31; 7, 8). In a modified version of this model, at about 24 h postchallenge, animals excreted greenish mucoid stools, frequently with occult blood, occasionally accompanied by anorexia, dehydration and bacteraemia (19). These symptoms were self-limiting and the infection induced acquired immune responses which appeared to be partly protective. The model was used to assess the virulence potential of mutants of pspA, a gene associated with the expression of a C. jejuni pilus-like appendage (19), and cheY (85). Recently the oral ferret model was used to compare the virulence of C. jejuni strains 81–176 and NCTC 11168, the genome sequence strain (4). The results of this study indicated that NCTC 11168 was poorly virulent, which may be due to the absence of certain gene(s). Thus this model appears to be an extremely useful in vivo tool in the study of virulence factors. However, the relationship to human disease remains debatable. The use of tincture of opium to suppress peristalsis, the high doses (1010–1011 cfu), and the short-lived symptoms of soft, but not diarrhoeic, stools raises a number issues. More importantly, for most laboratories availability for routine testing, for example mutants, is restricted due to price, seasonal breeding and a lack of campylobacter-free ferret breeding colonies. International importation into the United Kingdom is subject to rabies control legislation. This, plus the general paucity of ferret immune reagents and the generally subjective analysis of the diarrhoea, means that the model requires considerable skill to establish and interpret. Because of the similarity between the human and porcine gastrointestinal tracts, pigs have frequently been used as animal models of human enteric infections. Although C. jejuni and the closely related organism, C. coli, colonize piglets, the outcome of infection is largely dependant on the status of the animal. Experimental infection of normal piglets was generally asymptomatic while colostrum-deprived, neonatal animals developed diarrhoea which was occasionally bloody and mucoid and lasted for up to 6 d (74; 73; 3). Gross lesions and an abnormal histopathology are mainly confined to the large intestine. In experimentally infected gnotobiotic piglets, clinical and histopathological effects were also observed. These effects included inflammation and oedema of the caecum and colon with watery diarrhoea from about day 2 to day 12 (56; 10; 75). These results suggest that neonatal piglets, in the absence of maternally derived mucosal antibodies and, possibly, competing flora, provide a useful model of campylobacteriosis. The symptoms and histopathology appear to be consistent with human disease. However, too few studies have been undertaken to assess the reproducibility of the model. Moreover, a comparison of wild-type strains or mutants with this model has not yet been reported. Many attempts have been made to induce disease in campylobacter-infected rodents. Although oral inoculation of C. jejuni induced reproducible, albeit short-term, colonization of the gastrointestinal tracts of both adult and infant normal mice, disease was rarely observed. However, 36 reported diarrhoea in about 10% of C.B-17 scid beige mice orally challenged with fresh clinical isolates of C. jejuni. This was accompanied by colonic inflammation. However, 86 were unable to detect lower intestinal lesions in similarly challenged SCID mice, although they did observe antral gastritis. This lack of reproducibility may be a reflection of the genetic background of the scid mutation but seems to indicate that the disease symptoms observed in the former report are not just a consequence of the lack of host immune competence. Thus such a model may be difficult to standardize. This oral mouse diarrhoea model requires reproduction in other laboratories before it can be successfully used to test bacterial virulence properties. Also, with such a low incidence of disease, it is doubtful whether attenuated virulence, due to gene mutation, would be easily detectable. Although the natural route of transmission for C. jejuni is oral challenge, alternative routes to enhance virulence potential have been considered. Intranasal inoculation of adult mice caused death within 6 d postchallenge (6). The degree of lethality varied with the dose, mouse strain and campylobacter isolate used. At doses of 5 × 109 cfu, there was over 70% mortality of adult BALB/c mice with C. jejuni strain 81176. Organisms from mice challenged with this strain were recoverable from a variety of intestinal and extra-intestinal sites, suggesting a systemic infection, and splenic colonization persisted throughout the experiment. Thus, this appears to be a model of acute infection involving extra-intestinal colonization and invasion. It is also a model of immune protection (6) and was used to assess vaccine efficacy (45). The advantages of being able to work with such immuno-competent mice are considerable; however, the relationship of this model to the human disease remains questionable. The abnormal route, high dose and high mortality are a poor reflection of human disease symptoms. Moreover, the reproducibility of this model in other laboratories has yet to be established. Intraperitoneal challenge of adult mice with up to 1 × 109 cfu bacteria generally resulted in disseminated infection with recovery from liver and spleen but no obvious symptoms of disease even in different strains of mice (76). Supplementation of the challenge dose with iron dextran enhanced virulence (40). Such treatment induced a dose–response observed in terms of diarrhoea, weight loss, splenomegaly and mortality. These results were reproducible in another laboratory (Nielsen, E., pers. comm.) and differences in virulence were observable between bacterial strains. However, the relevance of this model is debatable, especially as challenge with heat-killed organisms induced similar, albeit significantly weaker, symptoms (Nielsen, E., pers. comm.). There were several early reports of disease induced by experimental challenge of hamsters with C. jejuni (37, 38). The outcome of infection was diarrhoea and entercolitis resembling the disease in humans. However, this outcome was only detected either by surgical inoculation into the caecum or by oral administration after purging followed by treatment with antibiotics, cimetidine and sodium bicarbonate. The oral administration of C. jejuni to rabbits resulted in transient colonization but no evidence of disease even in neonatal animals (28). However, adult rabbits are an appropriate size for animal models in which the bacteria can be injected directly into an intestinal loop. The removable intestinal tie adult rabbit diarrhoea (RITARD) model, originally developed to investigate the pathogenesis of Vibrio cholerae and Escherichia coli (68), involves the transient physical blockage of normal peristalsis by generating a temporarily sealed intestinal segment. C. jejuni held in this segment for about 4 h (13) induced a mucoid, occasionally bloody, diarrhoea within 3–5 d postchallenge. Acute inflammatory lesions and death were also frequently observed. The model detected qualitative differences between strains, wild types and mutants (77). Because the intestinal tie is temporary, the infected animals can be utilized to investigate longer-term pathological effects and may be allowed to induce detectable immune responses. Such studies have indicated the development of immunologically mediated protection against rechallenge (60). An alternative to the RITARD model is the rabbit ileal loop test (RILT). In this model sections of ileum (5 cm long) were ligated into loops. The loops (up to four per animal) were inoculated with bacterial suspension. The loop contents were harvested 18 h postchallenge. The resulting pathology comprised an acute inflammatory reaction (27). The RILT model primarily detected fluid accumulation which appears to be associated with host factors such as elevated levels of cyclic AMP, prostagladin E2 and leukotriene B4 (26). Thus this model may mimic some of the acute, host-related responses to infection. Both of these rabbit models require considerable surgical skills and postoperative care and are consequently not widely available. The relationship between the observations in these models to human disease is unclear; the route of administration is artificial, the effects of surgical trauma are unknown and the severe pathological effects occasionally observed poorly reflect human disease. However, they appear, in particular, to be models of host fluid secretion and acute inflammatory responses as well as translocation across the intestinal epithelial barrier via M cells. Because large numbers of organisms can be held in a confined space, the RILT model has also been useful for biomass production to investigate in vivo expression of antigens (55). Given its optimal growth conditions, it is generally considered that C. jejuni has evolved to preferentially colonize the avian gut. As this colonization is both extensive and asymptomatic the organism appears to act as a commensal in this host. Consequently, the chick has become an important model for the investigation of bacterial colonization factors. One-day-old chicks, orally challenged with as few as 30 cfu of fresh isolates of C. jejuni, achieved maximal ceacal colonization within 3 d at levels of up to 1 × 1010 cfu per g caecal contents (80; 17). Older birds, up to 8 weeks of age, were equally susceptible to experimental colonization. This colonization was chronic and only started to decline after 6–7 weeks or so, postchallenge (53). Although the main site of colonization was the caecum, organisms were recovered from throughout the gastrointestinal tract as well as the spleen and liver, indicating that some systemic infection occurred. The level and extensiveness of colonization in some chick models varied depending on the genetic lineage of the birds (71), the challenge strain (70; 87) and the degree of laboratory adaptation of the strain (Cawthraw et al. 1996). Colonized birds elicited systemic and mucosal antibody responses (14) and when cured of colonization by antibiotic treatment, these birds had a statistically significant reduction in susceptibility to rechallenge, suggesting that this immune response was partly protective (15). Challenge chick models have also been developed for testing vaccine efficacy. The results indicated that vaccination with killed whole cell or subunit vaccines may induce partial protection (41; 83; 58). However this efficacy was not reproducible (15). One problem with killed or subunit vaccine candidates was the inability, or inefficiency, of adjuvants to elicit effects at avian mucosal surfaces. Interestingly, antibodies present in newly hatched chicks (14), presumably maternally derived, did not appear to provide detectable protection from experimental challenge. To date, the oral chick model has proved remarkably useful for the assessment of the colonization potential of defined mutants and variants. One C. jejuni strain, 81116, produced a dose–response curve of colonization thus providing a quantitative model. This model has been particularly useful for detecting the compromized colonization potentials of deletion or insertion mutations (16). Mutants so far tested in this chick model have included flaA and flaB, sod, cat and regX (80; 12; 57; 33). Where a mutation reduced the colonization potential, the effects were observed either over the whole range of doses or, for some, over a restricted range. The chicken provides a model of campylobacter colonization in the avian gut. The relationship between this and mammalian gut colonization is unknown. Experimental challenge of chickens of all ages appears to induce an identical outcome to that observed during natural infection, when C. jejuni acts like an avian gut commensal. In studies of over 30 strains tested from a variety of sources (human, chicken, ostrich and water), even at levels of 1 × 1010 cfu per g caecal contents, colonization is asymptomatic regardless of strain (Newell, unpublished data). The reason why such colonization levels do not usually induce disease in birds is unknown. Possible reasons include host-specific expression of bacterial virulence factors or lack of host-specific receptors for such factors. Nevertheless, there are some reports of disease associated with experimental chicken infection, including diarrhoea, hepatitis and occasional death (64; 61; 82; 42). Moreover, there are occasional reports of associations between natural campylobacter infections and avian disease. Both a wasting syndrome leading to death in naturally infected young ostriches (78; 69) and symptoms of polyneuropathy in infected chickens (47) have been described. Such symptoms may be related to age, immune status, genetic background and maintenance conditions of the chickens; however, these conditions are not yet reproducible. The Helicobacter genus comprises a diverse group of bacterial species colonizing various mucosal sites throughout the gastrointestinal tracts of a number of different hosts. However, the helicobacter of major interest is H. pylori which colonizes the human gastric epithelial mucosa. This colonization is associated with chronic inflammation of the gastric epithelium (81). In about 10% of colonized individuals this may progress to more serious disease, including peptic ulceration (18), atrophic gastritis, gastric carcinoma (25) and, rarely, mucosa-associated lymphoid tissue (MALT) lymphomas (84). Under normal circumstances H. pylori appears to have a host-specificity for human and non-human primates. However, this organism has been ‘encouraged’, usually by repeated in vivo passage, to colonize the gastric epithelium of other orally challenged animals such as mice, dogs, piglets and gerbils. In addition, complementary animal models utilizing other gastric Helicobacter species, either in their own respective hosts or modified for artificial hosts, have been developed. Examples of such models include H. mustelae in ferrets and H. felis in mice. All these models have recently been extensively reviewed (30; 20; 44; 51). Each model has its problems. With the exception of non-human primates, excluded as previously mentioned, all these models poorly reflect the spectrum of human diseases observed. Nevertheless, most of the models have been used for the identification and screening of virulence factors, investigation of host immune responses, testing therapeutic agents and developing vaccines (20; 30; 44; 51). So what can we learn from the development and use of these models which would be useful in similar campylobacter research? Human-derived bacterial strains of H. pylori colonized piglets, which were gnotobiotic or barrier-raised and colostrum-derived (21; 24). This colonization was gradually enhanced by repeated in vivo passage (1). Similarly, repeated in vivo passage also enabled the adaptation of an H. pylori strain, SS1, to colonize the gastric mucosa of mice (46). Strain SS1 appeared to have been derived from a mixed inoculum of homogenates from the gastric tissue of several infected patients. Because multiple in vivo passages were required, and the adaptation was phenotypically stable, it seems likely that the organisms adapted to each new host by the sequential selection of several spontaneous mutations during the lifetime of the chronic infection. Host-related selection and adaptation therefore appears to be a property of H. pylori and may be reflected in the extreme genomic diversity of this bacterial species. C. jejuni is also genotypically highly diverse and may have evolved similar mechanisms for host adaptation. Although infection with this organism in human beings is usually acute, in avians the colonization is more chronic perhaps allowing time for host-specific adaptation. Thus multiple in vivo passaging may improve the ability of C. jejuni strains to colonize otherwise resistant hosts. Single in vivo passages of laboratory-attenuated strains of C. jejuni through chicks enhanced colonization by up to 105-fold (17). This enhancement was stable over multiple in vitro passages suggesting that it was a consequence of genetic selection. However, enhanced colonization in this model did not appear to be concomitant with enhanced virulence sufficient to cause disease, at least in chickens. In mammals the situation may be different. For example multiple in vivo passages of C. jejuni apparently enhanced virulence in intraperitoneally challenged infant mice, leading to acute disease and death (40). Given the growth conditions, such as antibiotic-containing media and high temperatures, usually applied to campylobacters during their routine detection and recovery from clinical material, it is not unlikely that selective pressures are imposed and that bacterial regulation of certain physiological processes is altered. C. jejuni expressed novel antigens during colonization (55) suggesting that in vivo passage may induce the up-regulation of the expression of appropriate colonization factors which may have become down-regulated during laboratory in vitro growth. Such bacterial components may account for the evidence that fresh isolates are more virulent than laboratory strains in the SCID mouse model (36). Specialized in vitro growth conditions may need to be developed to ensure the expression of appropriate bacterial factors. In summary, methods to enhance both colonization and virulence should be explored further. Such methods include the in vivo passage of C. jejuni strains through appropriate animal models, the use of strains isolated by non-selective methods, such as filtration with minimal laboratory culture, and challenge with tissue homogenates and/or feacal material containing organisms. Early studies screened many domestic and wild animals for evidence of environmental reservoirs of H. pylori. Although these studies demonstrated the host specificity of H. pylori for the gastric mucosa of humans and non-human primates, a multitude of other helicobacters were discovered. These fell into two groups, H. pylori-like and H. felis-like (20), on the basis of morphology and pathophysiology. These alternative bacterial species, and their respective hosts, have provided a wealth of information related to general concepts of helicobacter colonization and virulence. One of the most successful models used the H. pylori-like organism, H. mustelae, in the ferret (32). Even though the pattern of colonization varied from that of H. pylori in human beings, this model has been extensively used for vaccine studies. H. felis has the experimental advantage of a wider host range than the H. pylori-like organisms. Although background natural colonization with this or similar organisms restricted its use in animal models such as cats and dogs, H. felis has been extensively investigated in laboratory mice. In this model the disease was more severe than that induced by H. pylori but, nevertheless proved valuable for testing vaccine and treatment protocols and to investigate ulcerogenic and carcinogenic processes (20). There are also a multitude of Campylobacter species other than C. jejuni. Many of these colonize the intestinal mucosa and can be associated with enteric disease. C. fetus subsp. fetus can cause enteritis in human beings but is frequently associated with systemic infections in immunocompromized individuals. In pregnant women such infections may lead to abortion. In sheep this organism can also traverse the intestinal epithelium, inducing systemic infection leading to abortion in pregnant ewes. A similar outcome of infection is occasionally observed with C. jejuni in pregnant sheep. C. fetus subsp. fetus-induced abortion has an experimental model which has been used for investigation of the role of S-layer proteins (34). C. fetus subsp. fetus may therefore be used to model some aspects of C. jejuni diseases, especially those associated with invasiveness and systemic infection. C. jejuni subsp. doylei is frequently isolated in severe paediatric infections in developing countries (43). About 27% of all C. jejuni subsp. doylei isolates recovered were from blood cultures, in comparison with only 10% of C. jejuni subsp. jejuni, biotype 1. This data suggested that C. jejuni subsp. doylei is more invasive than most C. jejuni subsp. jejuni. Similarly C. upsaliensis is also a common cause of bacteraemia (43; 11). Whether this enhanced virulence of these alternative campylobacters is reflected in animal models has yet to be investigated and could provide much needed insight into the pathogenic mechanisms involved in tissue invasion. It has become clear, from animal model studies of H. pylori, that diseases associated with this infection are attributable to both bacterial and host factors. The host factors appear to apply at the species and strain levels. Different host species challenged with the same strain of Helicobacter spp. developed significantly different histopathologies (22). This stimulated helicobacter research workers to be much more adventurous with potential animal models. Recently enormous success has been achieved with Mongolian gerbils, which when challenged with H. pylori developed chronic/active antral gastritis, gastric ulcers and adenocarcinoma (39). Although 28, used C. jejuni to investigate many of the then available laboratory animals, it may now be timely to revisit this approach. Differences in host response are even observable between strains or populations of animals. Some murine strains, challenged with H. felis, developed severe gastritis (responder strains such as C57Bl/6), while others developed only mild or moderate gastritis (non-responder strains such as BALB/c). Too few mouse strains have been tested with C. jejuni to indicate whether this is an appropriate direction for research but it is clear that some mice strains were more susceptible to this infection than others (76). It seems likely that host immunological response contributes to disease manifestation during Helicobacter infections (51). Thus, the use of immunologically manipulated animals is enabling an understanding of the disease processes involved. At the moment such studies are really only feasible in mice where the genetic and immunological tools are available. Apart from some preliminary studies in SCID mice (32; 36) few attempts have been made to investigate the effects of immune compromization on C. jejuni colonization and virulence. Such effects may even be amplified by inclusion of some of the bacterial strain manipulation processes suggested above. Although mice are now frequently used for Helicobacter colonization studies their usefulness for virulence studies is constrained by the poor mimicry of the human disease. One notable difference between the human and animal infections is the lack of adherence of H. pylori to mouse gastric tissue, possibly as a consequence of the absence of appropriate bacterial receptors on murine tissues. To overcome this deficiency, transgenic mice were developed which expressed putative polysaccharide-based receptors. These mice apparently developed more severe disease (35). This introduces the option of manipulating host tissues to incorporate those defined features of the infection not naturally observable. Should C. jejuni-inducible enteritis be confined to only those animals expressing specific receptors, for example for toxins, the transgenic approach may provide a further option for manipulating defective animal models. A further option for manipulating the mouse model is the use of foetal human or non-human primate gastrointestinal sections xenografted into immunocompromized mice. This approach was successfully used to investigate Helicobacter colonization (48). Previous studies showed that xenografts of human intestinal tissues could be utilized for the investigation of bacterial pathogenesis (23). Although foetal human, or even non-human primate, intestinal tissue has a very restricted availability, many xenographed mice can be derived from individual intestinal tissue samples. As the tissue can be maintained for some time, this approach may provide a suitable model for the final analyses of putative virulence factors of C. jejuni. Despite over 20 years of investigation, no acceptably reproducible animal model of human campylobacteriosis has been developed. This has severely hampered attempts to identify and characterize potential virulence factors and to build on the enormous genetic resource provided by the recently published genome sequence. Nevertheless, it is clear from a comparison with similar studies undertaken to develop models for helicobacteriosis that we have not yet exhausted all possibilities. Further approaches to manipulate both the organism and the host should be explored. Models of C. jejuni colonization are available. The most useful at present is the oral chick model which is already being utilized to investigate putative colonization factors and will become a major tool for studies of functional genomics. Although the relevance of avian colonization to that in the human being has yet to be established, such studies may provide the information necessary to understand colonization in the chicken and enable appropriate intervention strategies to be developed to reduce campylobacter contamination in the food chain. I thank the Ministry of Agriculture Fisheries and Foods, GB, for financial support and Mr S Cawthraw, Dr G Manning and Dr C Thorns for helpful comments.
Referência(s)