Pathogenesis of enteric Campylobacter infection.

Abstract

1. Summary, 45S 2. Introduction, 45S 3. Campylobacter biology, 45S 4. Clinical aspects and epidemiology, 46S 5. Campylobacter virulence factors, 46S 5.1 Motility and chemotaxis, 46S 5.2 Adhesion and invasion, 48S 5.3 Toxins, 48S 5.4 Iron acquisition, 49S 5.5 Surface polysaccharide structures, 49S 5.6 Oxidative stress defence, 50S 5.7 Heat shock response, 50S 6. Regulation of Campylobacter virulence, 51S 6.1 Iron-responsive regulation, 51S 6.2 Two-component and other regulatory systems, 51S 7. Conclusions, 52S 8. Acknowledgements, 52S 9. References, 52S Campylobacter jejuni and related species are important human pathogens, causing acute human enterocolitis, and they are the most common cause of food-borne diarrhoea in many industrialized countries. Previous infection with certain strains of C. jejuni is also linked with the development of the neurological disorder Guillain–Barré syndrome (GBS). Despite its importance as a human pathogen, relatively little is understood of the mechanisms of C. jejuni-associated disease. The recent release of the complete genome sequence of C. jejuni strain NCTC 11168, together with new strategies for random and directed mutagenesis, have allowed a better insight into some of the genetic determinants of C. jejuni virulence. In this review current knowledge on the pathogenesis of C. jejuni infection is summarized, and possible directions of future research indicated. Members of the genus Campylobacter colonize the gastrointestinal tract of a broad range of animals. They are commensals in most animals, but notably are associated with disease in humans. The human pathogens C. jejuni and Campylobacter coli are causative agents of acute human enterocolitis, and are the most common cause of food-borne diarrhoea in many industrialized countries (85; 14). There is also a link between infection with particular serotypes of C. jejuni and subsequent development of GBS (58). Despite its importance as a human pathogen, our understanding of the mechanisms of Campylobacter-associated disease is still relatively poor. Focusing on C. jejuni and C. coli, the objective of this review is to summarize current progress on understanding the pathogenesis of Campylobacter infection. Campylobacters are small, spirally curved, Gram-negative rods with a polar flagellum at one or both ends of the cell. The cells are highly motile. They are catalase and oxidase positive, and urease negative. Campylobacters are microaerophilic, requiring an oxygen concentration of 3–15% and a carbon dioxide concentration of 3–5%. The temperature range for growth of the thermophilic Campylobacter species C. jejuni and C. coli is 34–44°C, with an optimal temperature of 42°C, which probably reflects an adaptation to the intestines of warm-blooded birds. Campylobacter spp. are fastidious organisms that require complex growth media, and are unable to ferment carbohydrates. Campylobacter jejuni and C. coli have a genome of approximately 1600–1700 kilobases (kb), which is relatively small compared with that of enteropathogens such as Escherichia coli, which have a genome of approximately 4500 kb (13; 61). The G + C content of Campylobacter DNA is on average 30–35%, which can cause problems when cloning these A + T rich sequences in E. coli (41; 95). Several strains contain plasmids, and bacteriophages have also been described (86). Campylobacter jejuni and C. coli are naturally transformable, and conjugation has also been used for the introduction of DNA into C. jejuni and C. coli (97; 95). Campylobacters, normally spiral-shaped, have been reported to change into coccoid forms on exposure to atmospheric oxygen levels or other stresses. These coccoid forms have been coined viable non-culturable (VNC), and this form has been suggested to be a dormant state required for survival under conditions not supporting growth of Campylobacter, e.g. during transmission or storage (77). However, the existence and infectivity of a VNC form of Campylobacter is controversial, and it will always be difficult to establish a VNC status with no ‘contaminating’ spiral organisms (39; 56; 36; 11; 90). Thus, the role of the VNC state in Campylobacter transmission and colonization remains to be elucidated. Infection with enteric campylobacters usually leads to diarrhoea, which in industrialized countries most often manifests as inflammatory diarrhoea with severe cramping in young adults. In developing countries there is a high rate of asymptomatic carriage, and milder clinical symptoms of watery, non-inflammatory diarrhoea are usually seen in young children (41; 85). In industrialized countries the infection is often seasonal, and targets mostly young adults and young children, but is also found in older people. In developing countries the infection is mostly restricted to children, and does not show clear seasonality. This difference is believed to principally reflect earlier acquisition of immunity in patients in developing countries following multiple exposure to campylobacters and not the existence of geographical variation in strain (patho) types. This is also confirmed by the observation that Campylobacter illness in travellers is usually similar to that of the country of origin, and not the country of visit (41). Infection of immunologically naïve subjects with C. jejuni and C. coli is generally followed (after 1–7 d) by acute abdominal pain, often accompanied by fever and general malaise. The symptoms progress to profuse diarrhoea which later contains mucus and blood. The disease is usually limited to a period of 5–8 d, but may continue longer and bacterial shedding often persists after clinical symptoms have ended. Complications following Campylobacter infection are uncommon, but an association with certain neurological disorders is noteworthy and of interest in pathogenesis. Campylobacter jejuni infection has been implicated as a frequent antecedent of GBS (50; 58). Campylobacters are considered to be part of the normal intestinal flora of a wide range of domestic and wild animals. Faecal contamination of meat often occurs during slaughtering, and human infection is usually acquired through the consumption of undercooked contaminated meat or other cross-contaminated food products. However, Campylobacter spp. can also be acquired from untreated milk or contaminated water and possibly from pets with diarrhoea (41), and the infectious dose can be as low as 500–800 organisms (5). The continuing rise in incidence of Campylobacter infection in industrialized countries is thought to be related to changes in eating habits rather than reflecting increased awareness or better diagnostic tools. When colonizing the intestines, enteric campylobacters are predicted to express several putative virulence factors. These systems will be reviewed in this section, and are summarized in Fig. 1, together with the colonization phase in which they are predicted to be expressed (Fig. 1). Overview of the different phases of Campylobacter colonization of the intestine. Putative virulence factors discussed are indicated, together with the phase(s) in which these are thought to be expressed Colonization of the intestine requires the ability to move into the mucus layer covering the intestinal cells. Campylobacter motility is conferred by the polar flagella, and combined with their ‘cork-screw’ form allows them to efficiently penetrate this mucus barrier (60; 53; 82). The flagellum of C. jejuni consists of an unsheathed polymer of flagellin subunits, which are encoded by the adjacent flaA and flaB genes (63). The flaA and flaB genes show a very high degree of sequence identity (95%) (33; 63), and are subject to both antigenic variation and phase variation (10; 35). The flagellin genes are independently expressed from their own promoters; the flaA gene is expressed from a σ28 promoter, whereas the flaB gene is expressed from a σ54 promoter (33; 63; 1; 62). The flaA gene is expressed at much higher levels than the flaB gene, and the C. jejuni flagellum consists normally of FlaA protein. However, mutational studies have shown that a C. jejuni flaA– flaB+ mutant produces short, truncated flagella that leads to a non-motile phenotype (99; 100). However, a C. jejuni flaA+ flaB– mutant showed slightly decreased motility, indicating a role of FlaB in flagellar function (101). The flagella of C. jejuni were tested successfully as part of a subunit vaccine in mice, indicating the importance of flagella in the pathogenesis of C. jejuni (52). Several accessory proteins are involved in export, assembly and expression of flagella. Campylobacter jejuni contains several genes encoding proteins homologous to E. coli proteins involved in flagellar expression. In the annotation of the C. jejuni NCTC 11168 genome (66), 36 open reading frames (ORFs) were assigned (putative) functions in flagellar biosynthesis, export or assembly. However, experimental evidence to prove their role in flagellar biosynthesis or motility is still lacking for most of these open reading frames. Mutational studies have shown that lack of expression of the flhA and flhB genes leads to the complete cessation of expression of flaA or flaB (57; 55), while a pflA mutant leads to paralysis of the flagella and loss of motility (110). A mutation in the C. coli hook protein gene flgE lead to the absence of flagella but, unlike E. coli, not to repression of flaA or flaB expression (42). Several aflagellated mutants have been used to demonstrate the importance of flagella to C. jejuni colonization and pathogenesis (59; 101). The C. jejuni and C. coli flagella are also post-translationally modified through glycosylation (15). Several genes involved in glycolysation of flagellin were identified: the pgl locus in C. jejuni 81-176 (83), the ptmAB genes in C. coli VC167 (32), and the neuB2 and neuB3 genes in C. jejuni NCTC 11168 (54). An interesting gene not annotated as involved in flagellar regulation is the product of the Cj1024c gene of C. jejuni NCTC 11168, which shows significant homology to the Helicobacter pylori flgR regulatory protein (78). This protein coordinately regulates the expression of H. pylori genes involved in motility (78). Chemotaxis is the ability to detect and move up or down chemical gradients. Both motility and chemotaxis are essential for C. jejuni colonization, as non-chemotactic mutants were unable to colonize the intestine in animal models (84). Campylobacter jejuni is attracted to mucins, L-serine and L-fucose, whereas bile acids are repellants (38). The regulatory gene cheY was identified as the affected gene in a motile, non-invasive mutant of C. jejuni (111), and is thought to be involved in general modulation of C. jejuni virulence genes. Separately, the cheY gene was also identified using a polymerase chain reaction (PCR)-based approach for the identification of response regulators (see section 6.2) and in this study CheY has been shown to affect the rate of flagellar motor switching (J. Marchant, B.W. Wren and J.M. Ketley, unpublished results). A C. jejuni cheY mutant was more adherent and invasive than the wild-type strain, but unable to colonize ferrets or cause disease (111). Several other components of the chemotaxis system of C. jejuni have been identified in the genome sequence, including cheA, cheV and cheW genes. Other genes putatively involved with chemotactic responses are 10 chemoreceptor genes which contain methyl-accepting chemotaxis protein domains (66). An important feature in C. jejuni pathogenesis is its binding and entry in host cells, which was reviewed by 104. Upon infection, C. jejuni crosses the mucus layer covering the epithelial cells and adheres to these cells, and a subpopulation subsequently invades the epithelial cells. The invasion of epithelial cells can lead to the mucosal damage and inflammation often seen in Campylobacter infection; it is not clear whether inflammation has a direct role in epithelial damage and/or diarrhoea. In vitro and in vivo experiments have demonstrated that C. jejuni is capable of invading epithelial cells, although the invasive ability of strains differs (20; 44). Fresh clinical isolates tend to invade at higher frequencies, and continued in vitro passaging reduces the invasiveness of strains (45). Strains are also capable of translocating across cell layers (20; 48), but transcytosis maybe a direct consequence of host cell invasion (7). The first C. jejuni determinants identified to be involved in adherence and invasion were the flagella (100; 29). Adhesion and invasion are dependent on both motility and flagellar expression, as C. jejuni mutants showing reduced motility due to paralysed flagella showed reduced adherence, and absence of invasion (110). This indicated that, while flagella are involved in adherence, other adhesins must be involved in subsequent internalization. Invasion is also thought to be involved in triggering inflammation, as invasion of C. jejuni was required for induction of several markers of inflammation (19), such as the important proinflammatory cytokine interleukin 8 (37). Adhesion by bacterial pathogens is often mediated by fimbrial structures. Campylobacter jejuni was long thought not to produce fimbriae, but one unconfirmed report described the production of fimbrial-like appendages by C. jejuni and C. coli when the bacteria were grown in the presence of bile salts (16). The fimbrial subunit gene has not yet been identified, and the C. jejuni NCTC 11168 genome sequence does not contain a clear candidate gene. However, while non-fimbriated C. jejuni mutants showed no defects in in vitro adherence and invasion or ferret colonization, there was a significant reduction in disease symptoms in the ferret animal model (16). The pilus-like appendage has been shown to be an artifact of the presence of bile salts in growth medium (27). Other adhesins identified are the PEB1 and CadF proteins. The PEB1 protein is encoded by the peb1A locus. Campylobacter jejuni mutants in peb1A showed significantly decreased adherence as well as decreased invasion of HeLa cells, and were also significantly affected in their colonization ability in a mouse model (68). The CadF protein was identified as a fibronectin-binding protein (43); cadF mutants were unable to bind fibronectin, and did not colonize newly hatched leghorn chickens (112). Invasion of C. jejuni has been tested in vitro using several cell lines of intestinal origin, such as INT-407, HEp-2 and differentiated Caco-2 cells (45; 100; 20). Differentiated Caco-2 cells form tight junctions and produce apical surface enzymes. Campylobacter jejuni invasion is both dependent on de novo synthesized C. jejuni proteins as well as host cell signal transduction (47; 106). Co-cultivation of C. jejuni and INT-407 cells lead to the production and secretion of at least eight proteins, including the CiaB protein (47). A C. jejuniciaB mutant adhered equally to the wild-type strain, but was not internalized, and did not secrete any of the other seven proteins (47). The authors have suggested that CiaB is a member of a type III secretion system (47); however, this similarity has been challenged (65). Processes in the host cell are also heavily involved in internalization of C. jejuni. With several inhibitors of host protein kinases, in particular tyrosine kinases, it was shown that protein phosphorylation is important in C. jejuni internalization (106). Also caveolae were implicated in C. jejuni invasion, probably via endocytosis (106). So far all (random) mutagenesis techniques have only allowed the identification of motility-related genes as genetic determinants involved in C. jejuni invasion (flaA, flaB, pflA, cheY), and the C. jejuni genome sequence did not show clear candidates for invasion-related genes (66). Although invasion of host cells can undoubtedly contribute to the cytopathic effects associated with C. jejuni infection, the level of bacterial invasion may be too low to be solely responsible (see section 5.2; 104). Therefore a lot of attention has been focused towards the characterization of Campylobacter toxic activity. Several groups reported the presence of either cytotoxins, enterotoxins or both (98). However, apart from the cytolethal distending toxin (CDT, see below; 71), genes encoding toxins have not yet been isolated (98). The genome sequence of C. jejuni NCTC 11168 contains only the cdt genes, two genes containing haemolysin domains and a phospholipase (pldA) which caused the haemolytic activity of C. jejuni reported previously (30). It is likely that strain and assay differences are responsible for the differences in the range of reported toxic activities (98). Cytolethal distending toxin activity causes certain cell types (such as HeLa cells and Caco-2 cells) to become slowly distended, which progresses into cell death. Although all C. jejuni and C. coli strains tested contain the cdt genes, there is a profound variation in CDT titres (71; 21, 22). While most C. jejuni strains have a relatively high CDT activity, C. coli strains show mostly low activity (71); the cause of this strain difference has yet to be elucidated. CDT cytotoxicity is caused by a G2 phase cell cycle block in the host cell through blocking of the CDC2 kinase involved in entry into mitosis (102). The C. jejuni CDT is encoded by a three-gene operon (cdtABC), and isogenic C. jejuni cdt mutants lost all CDT activity (102; 72). CDT involvement in diarrhoea was proposed to disturb the survival or maturation of crypt cells into functional villus epithelial cells and cause a temporary erosion of the villus and a subsequent loss of absorptive functions (102). A C. jejuni cdtB mutant was unaffected in enteric colonization abilities in adult severe combined immunodeficient mice, but demonstrated impaired invasiveness into blood, spleen and liver tissues (72). The ability to acquire the essential nutrient iron from the host contributes to bacterial pathogenesis. The concentration of free iron in host tissues is too low to support bacterial growth, as iron is complexed into haem compounds and to transferrin (in serum) and lactoferrin (at mucosal surfaces); this iron limitation constitutes a non-specific host defence. Campylobacter jejuni can utilize a relatively low number of iron compounds (23). It does not produce siderophores, but is able to use the siderophores ferrichrome and enterochelin produced by other organisms (70). It is also able to use haem compounds, which might be released at the site of inflammation (23; 70). Gram-negative ferric iron acquisition systems usually consist of an outer membrane (OM) receptor, which transports the iron compound over the OM, a periplasmic binding protein and an inner membrane (IM) ABC transporter consisting of a permease and ATP-binding protein, while ferrous iron transport is usually accomplished by a single IM protein. Energy for the transport over the OM is transduced through the TonB–ExbB–ExbD complex, whereas transport over the IM is energized by ATP hydrolysis (8). Campylobacter jejuni expresses several ferric iron acquisition systems upon growth in iron-restricted conditions. So far, a haemin/haemoglobin uptake system (chuABCD) (76) and an enterochelin transport system lacking an OM receptor (ceuBCDE) (75) have been identified and demonstrated to be involved in iron acquisition; both are present in all strains tested to date. There is also the cfrA gene which encodes a homologue of an OM receptor for which no substrate is known; cognate periplasmic and IM transporter systems have not yet been identified for this receptor (34). The cfrA system is only present in some strains, a situation similar to that of the ferrichrome uptake system (fhuABD) identified in another strain (26). It might be expected that strains will contain either cfrA or fhuABD, but this awaits further investigation. A fifth ferric iron uptake system has been identified in the C. jejuni NCTC 11168 genome sequence (66), and consists of two divergent operons, one encoding a periplasmic binding and IM transport system, and the other encoding an OM receptor and TonB–ExbB–ExbD complex. The function of these operons is currently being investigated (76). The C. jejuni genome also encodes one putative ferrous iron transport system, a homologue of the E. coli FeoB protein. Escherichia coli and H. pylorifeoB mutants were unable to colonize the intestine and stomach respectively in a mouse model (80; 91). This indicates that while the FeoB protein might not be essential under in vitro conditions, it could have an important role in colonization by campylobacters. The outer membrane constituents lipo-oligosaccharide (LOS) and lipopolysaccharide (LPS) form a major component of the Gram-negative outer membrane, and are important virulence factors involved in serum resistance, endotoxicity and adhesion. Lipooligosaccharide is composed of two regions, a lipid A molecule joined to a core oligosaccharide, and LPS additionally contains an O-chain consisting of repeating oligosaccharide. Campylobacter jejuni strains always express LOS, but on Western blots probed with strain-specific antibodies it was also shown that some strains had an O-chain-like ladder (25; 103). Two groups reported the cloning of a C. jejuni operon containing genes thought to be involved in LOS/LPS biosynthesis (25; 103), as expression of these genes in E. coli led to the production of an O-antigen recognized by antisera to C. jejuni LPS (49). However, none of the mutants created in this region had a clear effect on C. jejuni O-chain production, but some mutants showed different mobility of the core antigen. This locus was involved in general protein glycosylation in C. jejuni (83). These unexpected results with the LOS/LPS cluster were later explained by the description of another large cluster of genes, which were involved in production of a capsular antigen (40). It was subsequently shown that C. jejuni produced a separate LOS antigen, and had an O-chain connected to a phospholipid carrier (24). The C. jejuni surface polysaccharide structures and flagella have been shown to be sialylated, which is thought to be responsible for the ganglioside mimicry leading to GBS (58). Guillain–Barré syndrome is a serious autoimmune disorder of the peripheral nervous system, and is one of the most common causes of acute flaccid paralysis. It is characterized by a rapidly progressing weakness of the limbs and respiratory muscle and, although also mostly self-limiting, may lead to severe neurological deficit. Guillain–Barré syndrome is linked to infection with particular heat-stable serotypes of C. jejuni; formation of autoantibodies is thought to be responsible for the demyelination leading to GBS (51; 4; 17). Three genes were characterized which encoded proteins homologous to the NeuB protein, a key enzyme in the production of polysialic acid (54). One of these genes (neuB1) was involved in the sialylation of the LOS molecule, whereas the other two genes (neuB2 and neuB3) were involved in the sialylation of the flagellin protein. Furthermore, there was heterogeneity between strains in the large cluster of genes encoding sugar synthetases and transferases involved in the biosynthesis of ganglioside mimics (28). A C. jejuni strain of serotype O:19, associated with GBS, had two genes encoding sialyltransferases, whereas strain NCTC 11168 (serotype O:2) had only one copy, which showed much lower sialyltransferase activity (28). There have also probably been internal genetic rearrangements and heterologous DNA uptake, leading to the strain differences. To assess the role of these gene clusters, further analysis of the composition of these loci in strains of different serotypes will be required as an analysis of gene content variation has revealed differences between strains (64). Campylobacters are microaerophilic bacteria, which means that they have to deal with toxic oxygen metabolites produced during normal metabolism, during transmission or when in contact with the host immune defences. Campylobacter jejuni and C. coli share the same oxidative stress defence systems, which can be divided into superoxide stress defence and peroxide stress defence (81). The main component of the C. jejuni superoxide stress defence is the superoxide dismutase (SOD) protein SodB (69; 74). This protein, encoded by the sodB gene, is an iron-containing SOD that is involved in the removal of superoxides by converting these to hydrogen peroxide. While not affected under standard in vitro growth conditions, a C. jejunisodB mutant showed significantly decreased intracellular survival in INT-407 cells (69), and a C. coli sodB mutant showed decreased survival in air and model food systems, and was also affected in its colonization of 1-day-old chicks (73). The peroxide stress defence consists mainly of two proteins, the catalase (KatA) and alkyl hydroperoxide reductase (AhpC, also named Tsa or TsaA) proteins (31; 2). Catalase converts hydrogen peroxide to water and oxygen, and thus removes the hydrogen peroxide produced by SOD or other metabolic processes. Surprisingly, a C. colikatA mutant, which was hypersensitive to hydrogen peroxide (31), was not affected in aerobic survival or its ability to colonize chickens (73). The AhpC protein reduces alkyl hydroperoxides to alcohols, while being oxidized itself. The AhpC protein was important in aerobic survival of C. jejuni, as a C. jejuni ahpC mutant showed a significantly decreased aerobic survival, dying about 3 h earlier than wild-type cells (2). Neither the C. jejuni ahpC mutant or the C. jejuni and C. coli katA mutants were affected under standard in vitro growth conditions (31; 2). An accessory component of the peroxide stress defence may be one of the two ferredoxin proteins of C. jejuni (FdxA) (93). The fdxA gene is divergently orientated to the ahpC gene, and is probably coregulated with ahpC (see section 6.1). A C. jejuni fdxA mutant showed decreased aerobic survival similar to that of the C. jejuni ahpC mutant, but was not affected in specific oxidative stress defence. A model has been proposed in which the FdxA protein is used to reduce the oxidized AhpC protein, thus recycling it. Further investigation is necessary to determine any direct role of peroxide stress defence in colonization and pathogenesis. Campylobacter jejuni and C. coli must be able to respond to a change in temperature, as they can be found in the avian gut, where the normal temperature is 42°C, as well as temperatures in human hosts (37°C) and during transmission in water, milk or on meat (4°C or varying temperatures). The thermal stress response of bacteria is mostly carried out by the induction of the expression of heat shock proteins (HSPs). These HSPs have an important function in thermotolerance as well as in the response to other stresses by acting as chaperones to promote the folding of most cellular proteins and proteolysis of potentially deleterious, misfolded proteins. Several HSPs have been identified in C. jejuni, including the GroESL, DnaJ, DnaK and ClpB proteins (46; 87, 88, 89). However, a role in C. jejuni pathogenesis has only been demonstrated for the DnaJ protein, as a C. jejuni dnaJ mutant was unable to colonize chickens (46). The importance of the C. jejuni thermal stress response is also indicated by the link between thermoregulation and chicken colonization through the RacR regulatory protein (6). The need for a good response to varying environmental temperatures is also apparent from the presence of three homologues of temperature-responsive regulators in the C. jejuni genome. While C. jejuni is lacking σ32, which regulates expression of HSPs in E. coli (9), it does contain a homologue of the HcrA protein, a repressor of HSPs in Bacillus subtilis (88). Campylobacter jejuni also contains a homologue of the HspR protein, which is a repressor of HSPs in H. pylori (79). Campylobacter jejuni HspR regulates the DnaK chaperone protein, and a C. jejuni hspR mutant was affected negatively in its growth at 44°C (M.J. Emery and J.M. Ketley, unpublished results). There is also thermoregulation by the RacRS two-component regulatory system (6). Which system regulates which HSPs, and whether cross-regulation is taking place, is currently under investigation. Iron is an essential nutrient for all living organisms, but is also capable of generating toxic oxygen metabolites. Therefore iron homoeostasis is of vital importance to the cell. This iron homeostasis is usually regulated in response to the intracellular ferrous ion concentration by repressing the uptake of iron (see section 5.4) through the repressor protein Fur (18). In host tissues the free iron concentration is mostly too low to allow bacterial growth. This iron restriction, a non-specific host-defence mechanism, is used by several bacterial pathogens as a signal for the coordinated expression of virulence factors through the Fur protein. Genes regulated in response to iron usually include toxins, haemolysins and iron acquisition genes, but can also include (virulence) genes not related to iron metabolism (18). Campylobacter jejuni contains two genes encoding Fur homologues, designated fur and perR (92; 105). The C. jejuni fur gene shows unusual characteristics, as it is the first bacterial fur gene not to have its own promoter, but is expressed from two promoters of upstream genes as part of multicistronic messengers which contain the housekeeping genes lysS and glyA (12; 94). The C. jejuni Fur protein is the major iron-responsive regulator, as a C. jejuni fur mutant was not able to regulate the expression of all known iron acquisition systems, and also grew significantly slower under standard in vitro growth conditions (96). However, there was still iron-responsive gene regulation in a C. jejuni fur mutant, indicating the presence of a second iron-responsive regulator. This second iron-responsive regulator was shown to be the PerR protein, which regulates the expression of the peroxide stress defence proteins AhpC and KatA (see section 5.6). A C. jejuni perR mutant was hyper-resistant to peroxide stress inducers. In this study a C. jejuni Fur PerR double mutant was also described, in which all iron-responsive gene regulation was abolished (92). However, the role of Fur and PerR regulation in C. jejuni pathogenesis needs to be further addressed using animal models. Two-component regulatory systems are widely spread in bacteria, and have an important role in signal transduction of environmental stimuli. They usually consist of a histidine protein kinase (HPK) sensor that is located in the IM with a cytoplasmic kinase domain site, and a response regulator (RR) that is phosphorylated by the HPK. The phosphorylated RR interacts with the promoters of its target genes and regulates their expression coordinately (67). With a PCR-based approach several putative RR genes were isolated from C. jejuni (109). For two of these the function is known; these are the cheY (see section 5.1) and the racR genes (111; 6). The RacR protein was demonstrated to be involved in thermoregulation and in colonization (6). A C. jejuni racR mutant showed a decreased growth rate at 42°C but not at 37°C compared with the wild-type strain. The expression of several proteins was affected in the racR mutant, although not all proteins showed thermoregulation. Finally, the racR mutant showed a significantly reduced colonization of chickens, indicating an important role for RacR regulation in colonization (6). The publication of the C. jejuni genome sequence allowed the identification of other RRs and HPKs (66). Not counting the chemotaxis genes (cheY–cheA), there are seven RR genes and five HPK genes in total. Of these there are four RR–HPK couples (Cj1223c–1222c, Cj1227c–1226c, Cj1261–1262 (RacRS) and Cj1491c–1492c), three orphan RRs (Cj0355c, Cj0643 and Cj1024c, the putative FlgR) and one orphan HPK (Cj0793) (66). A study of RRs and HPKs in H. pylori showed that, although not contiguous on the genome, the orphan RR homologous to Cj1024c and the orphan HPK homologous to Cj0793 formed a couple (3). The function of all these regulators and sensors is still mostly unknown, and is hampered by the inability to construct mutants in some of these genes. This is a situation similar to that in H. pylori, where some two-component regulatory systems could not be mutated (3). Several other putative regulators were identified in the genome sequence, but most of these await experimental evidence to confirm a putative regulatory role. Exceptions to this are the HspR homologue (see section 5.7), the Fur and PerR proteins (see section 6.1), the sigma factors σ28, σ54 and σ70 (107), and the regulator of the stringent response LytB (108). During the past 30 years Campylobacter spp. have become the focus of several research groups around the world. Improved diagnostic methods have demonstrated its importance in human disease, and it has been recognized as a major public health burden in industrialized countries. Despite the attention it has received, many questions remain to be answered. Targets of future Campylobacter research will include further elucidation of its pathogenic mechanisms, including the identification of invasion factors and elucidation of the roles of inflammation and toxin production by C. jejuni. A systematic search for C. jejuni factors involved in colonization and invasion should allow the identification of factors important for pathogenesis of C. jejuni infection. The availability of the C. jejuni NCTC 11168 genome sequence, and the development of (semi) random mutagenesis strategies, should allow further insight in the role of specific genes in C. jejuni virulence. However, to complement the chicken colonization model there is a need to develop a animal model of diarrhoeal disease. Finally, several other Campylobacter species, especially the recently recognized human pathogen Campylobacter upsaliensis, require further research and application of the techniques developed for C.jejuni. A.H.M. van Vliet is supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO grant 901-14-206) and J.M. Ketley by the Wellcome Trust, Royal Society, Biotechnology and Biological Sciences Research Council and the Department of Health. Present and past collaborators are acknowledged for their valuable input.