Epidemiology of susceptible and AR C. jejuni: human and animal studies

Campylobacter jejuni colonization of the host

Microbiota mediated colonization resistance

Early attempts to develop a murine model of C. jejuni colonization and colitis were retarded by an inability to colonize wild-type (WT) mice with C. jejuni or low-level C. jejuni colonization with subclinical disease (Chang and Miller, Reference Chang and Miller2006; O'Loughlin et al., Reference O'Loughlin, Samuelson, Braundmeier-Fleming, White, Haldorson, Stone, Lessmann, Eucker and Konkel2015). C. jejuni colonization resistance was abolished by infecting gnotobiotic mice; and persistent, high-level C. jejuni colonization was achieved by infecting immune deficient gnotobiotic mice (Chang and Miller, Reference Chang and Miller2006). Along with persistent C. jejuni colonization, immune deficient (i.e. SCID) gnotobiotic mice had marked inflammation of the cecum, colon, and stomach (Chang and Miller, Reference Chang and Miller2006). Other reports showed that both specific pathogen-free (SPF) C57BL/6 wild-type and congenic C57BL/6 interleukin-10 deficient (IL-10^−/−) mice were colonized by C. jejuni, but only IL-10^−/− mice were susceptible to colitis after infection with C. jejuni 11168 (Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007, Reference Mansfield, Patterson, Fierro, Murphy, Rathinam, Kopper, Barbu, Onifade and Bell2008a). In this work, it was also shown that presence of Helicobacter hepaticus or related mouse pathogens conferred immunological resistance to colonization with C. jejuni strains. The C57BL/6 IL-10^−/− mouse model has also been used to determine if the microbiota plays a critical role in the inflammation seen in C. jejuni-infected C57BL/6 IL-10^−/− mice.

Shifts in microbiota enhance colonization of C. jejuni

Inoculation of mice with human fecal material has been used to generate humanized microbiota mice (^Humicrobiota). In one model, ^Humicrobiota mice were generated by using a five-antibiotic cocktail (ampicillin, vancomycin, ciprofloxacin, imipenem, and metronidazole) to deplete the microbiota, followed by inoculation with either murine or human feces. Per oral C. jejuni infection of these mice resulted in clearance of C. jejuni in 2 days by murine microbiota mice. In contrast, mice with human microbiota remained colonized for 6 weeks and displayed exacerbated T cell, B cell, and pro-inflammatory cytokine responses in the colonic mucosa (Bereswill et al., Reference Bereswill, Fischer, Plickert, Haag, Otto, Kühl, Dasti, Zautner, Muñoz, Loddenkemper, Gross, Göbel and Heimesaat2011). However, murine microbiota controls were also pre-treated with antibiotics thus raising the question of whether this affected their immune responses.

Transplanted human fecal microbiota from young adults (^Humicrobiota) altered immune responses to C. jejuni infection in C57BL/6 mice when compared with congenic mice with conventional mouse microbiota (^Convmicrobiota) (Brooks et al., Reference Brooks, Brakel, Bell, Bejcek, Gilpin, Brudvig and Mansfield2017). ^Humicrobiota and ^Convmicrobiota mice had statistically significant differences between their microbial communities, although alpha diversity indices showed no differences in diversity between experimental groups. C57BL/6 mice carrying a stable ^Humicrobiota and experimentally infected with C. jejuni strains from a colitis patient and a GBS patient had higher C. jejuni colonization levels and colonic inflammation scores than congenic mice with ^Convmicrobiota. C. jejuni 11168, but not 260.94, elicited T_H1- and T_H17-associated anti-C. jejuni antibody responses. Notably, ^Humicrobiota mice displayed a T_H2 biased anti-C. jejuni antibody response independent of inoculation status. Finally, ^Humicrobiota mice infected with both C. jejuni patient strains had elevated GM1 anti-ganglioside antibody responses, but only those given strain 260.94 were significantly higher than conventional microbiota mice given the same strain. These data demonstrate that human microbiota alters host–pathogen interactions in infected mice increasing colonization and autoimmune responses in a C. jejuni strain-dependent manner. Thus, particular microbiota compositions are likely to enhance host susceptibility to GBS following C. jejuni infection.

Disease associated with C. jejuni: hemorrhagic gastroenteritis and IBD in human beings

C. jejuni causes a spectrum of diseases in people with some individuals colonized but asymptomatic. However, most patients ingesting C. jejuni in undercooked meat or unpasteurized milk develop mild-to-severe gastroenteritis targeting the colon, which is debilitating but self-limiting within 7–10 days (Vandenberg et al., Reference Vandenberg, Cornelius, Souayah, Martiny, Vlaes, Brandt and On2013). Lesions include colonic crypt distortion, crypt abscesses, mucin depletion, edema of the colonic lamina propria, and significant infiltration of granulocytes and mononuclear cells (Young and Mansfield, Reference Young, Mansfield, Ketley and Konkel2005). Infection initiated in the GI tract can become extra-intestinal, particularly in immune-compromised hosts (Karmali and Fleming, Reference Karmali and Fleming1979; Blaser et al., Reference Blaser, Glass, Hug, Stoll, Kibrya and Alim1980; Ketley et al., Reference Ketley, Guerry, Panigrahi, Newell, Ketley and Feldman1996). Damage resolves in most patients, but campylobacteriosis can be life-threatening in immune-compromised or HIV-infected people with persistence, systemic spread, and multi-organ damage (Young and Mansfield, Reference Young, Mansfield, Ketley and Konkel2005; Fernandez-Cruz et al., Reference Fernandez-Cruz, Munoz, Mohedano, Valerio, Marin, Alcala, Rodriguez-Creixems, Cercenado and Bouza2010). Based on CDC data, these infections are more likely to lead to death (CDC, 2011). AR C. jejuni have been suggested to cause more severe infections requiring lengthier hospitalizations when compared with susceptible infections (Moore et al., Reference Moore, Barton, Blair, Corcoran, Dooley, Fanning, Kempf, Lastovica, Lowery, Matsuda, McDowell, McMahon, Millar, Rao, Rooney, Seal, Snelling and Tolba2006) and, thus, represent an important public health concern. Macrolides and fluoroquinolones are the antibiotics of choice for treating C. jejuni infections, while tetracyclines are sometimes considered as alternatives to these drugs but are rarely used. In cases of bacteremia or multi-organ infection, intravenous amino glycosides are often used. Based on the rising incidence of resistance to these drugs, Danish scientists compared adverse health events associated with susceptible and AR C. jejuni infection in 3471 patients (Moore et al., Reference Moore, Barton, Blair, Corcoran, Dooley, Fanning, Kempf, Lastovica, Lowery, Matsuda, McDowell, McMahon, Millar, Rao, Rooney, Seal, Snelling and Tolba2006). Patients with quinolone-resistant strains had a 9.68-fold increased risk of adverse events within 30 days, while patients with erythromycin-resistant strains had a 5.51-fold risk of an adverse event within 90 days when compared with patients with susceptible strains of C. jejuni. More work is needed to understand whether this enhanced disease is due to AR alone or whether virulence attributes of these strains are enhanced in some other manner such as selection compounded by dysbiosis. Given the rising incidence of AR C. jejuni in people and animals, this is a key question to address keeping in mind that multiple interactions may be at work.

The inflammatory nature of Campylobacter enteritis, along with compatible endoscopic or histopathologic findings, can produce a clinical picture that resembles IBD (Farmer, Reference Farmer1990; Perkins and Newstead, Reference Perkins and Newstead1994). Fecal erythrocytes and leukocytes are present in the majority of campylobacteriosis cases whether or not the stools are grossly bloody. In several studies, investigators have shown that acute C. jejuni gastroenteritis is followed by an increased risk for IBD (Garcia Rodriguez et al., Reference Garcia Rodriguez, Ruigomez and Panes2006) (Ternhag et al., Reference Ternhag, Torner, Svensson, Ekdahl and Giesecke2008). Other closely related strains such as Campylobacter concisus have also been linked to the development of IBD (Hold et al., Reference Hold, Smith, Grange, Watt, El-Omar and Mukhopadhya2014).

T helper-1 based inflammation is associated with C. jejuni-induced hemorrhagic colitis

IL-10 is classified as an anti-inflammatory cytokine that downregulates host response to invasion by intracellular pathogens by inhibiting several key inflammatory regulators, including major histocompatibility complex II and T-cell co-stimulatory factors B7-1 and B7-2, and expression of interferon (IFNγ) (Moore et al., Reference Moore, Malefyt, Coffman and O'Garra2001; Ouyang et al., Reference Ouyang, Rutz, Crellin, Valdez and Hymowitz2011). Congenic C57BL/6 IL-10 deficient mice (C57BL/6 IL-10^−/−) but not their IL-10^+/+ counterparts were susceptible to colitis when infected by C. jejuni 11168 (Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007). C. jejuni 11168 successfully colonized the GI tract of C57BL/6 WT and IL-10^−/− mice; however, only IL-10^−/− mice developed inflammation of the colon and cecum. The cecum was the GI site with the highest level of colonization; and C. jejuni was isolated from most GI compartments (i.e. cecum, stomach, colon, and jejunum) or detected by C. jejuni specific (gyrA) PCR of tissue homogenates. All mice were colonized at comparable levels and colonization was necessary but not sufficient for GI lesions as only IL-10^−/− mice developed disease and lesions (Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007).

Colitis in IL-10^−/− mice was C. jejuni strain dependent (Bell et al., Reference Bell, Jerome, Plovanich-Jones, Smith, Gettings, Kim, Landgraf, Lefébure, Kopper, Rathinam, Charles, Buffa, Brooks, Poe, Eaton, Stanhope and Mansfield2013a; Malik et al., Reference Malik, Sharma, St Charles, Dybas and Mansfield2014), and genomic composition of the C. jejuni strain was an important factor in determining the disease outcome (Bell et al., Reference Bell, Jerome, Plovanich-Jones, Smith, Gettings, Kim, Landgraf, Lefébure, Kopper, Rathinam, Charles, Buffa, Brooks, Poe, Eaton, Stanhope and Mansfield2013a). To date the entire suite of genes required for C. jejuni colitis remains unknown; however, comparative genomics of available C. jejuni genomes and gene expression analysis of C. jejuni strains that caused colitis in C57BL/6 IL-10^−/− mice compared with those that did not yield 201 potential virulence genes, collectively called the C. jejuni virulome (Bell et al., Reference Bell, Jerome, Plovanich-Jones, Smith, Gettings, Kim, Landgraf, Lefébure, Kopper, Rathinam, Charles, Buffa, Brooks, Poe, Eaton, Stanhope and Mansfield2013a). Motility is a major determinant of C. jejuni pathogenesis. C. jejuni diminished motility and non-motile mutants colonized at rates 100 to 1000-fold less than the WT (Wassenaar et al., Reference Wassenaar, Zeijst, Ayling and Newell1993) thus variation in motility amongst strains played a role in infection outcomes. Further, an experiment in C. jejuni 11168-infected germ-free C57BL/6 mice showed that expression levels of 90 open reading frames (ORFs) were significantly up- or down-regulated in the mouse cecum at least 2-fold compared with in vitro growth (Bell et al., Reference Bell, Jerome, Plovanich-Jones, Smith, Gettings, Kim, Landgraf, Lefebure, Kopper, Rathinam, St Charles, Buffa, Brooks, Poe, Eaton, Stanhope and Mansfield2013b). Genomic content of these ninety C. jejuni 11168 ORFs was significantly correlated with the capacity to colonize and cause enteritis in mice. Differences in gene expression levels and patterns are thus an important determinant of pathotype in C. jejuni strains in this mouse model and more work is needed to reveal the function of many of these virulence factors.

Mouse C. jejuni disease models have demonstrated that most strains are invasive and elicit strong inflammatory responses particularly in the colon (Chang and Miller, Reference Chang and Miller2006; Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007; Reference Mansfield, Patterson, Fierro, Murphy, Rathinam, Kopper, Barbu, Onifade and Bell2008a). Many C. jejuni strains causing disease in these models have cytolethal distending toxin and produce effacing lesions that would be expected to release many self-antigens (Pickett et al., Reference Pickett, Pesci, Cottle, Russell, Erdem and Zeytin1996; Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007, Reference Mansfield, Patterson, Fierro, Murphy, Rathinam, Kopper, Barbu, Onifade and Bell2008a). Upon invasion, most C. jejuni strains are captured and processed by DCs in the lamina propria (Rathinam et al., Reference Rathinam, Hoag and Mansfield2008, Reference Rathinam, Appledorn, Hoag, Amalfitano and Mansfield2009). BM-DCs challenged with C. jejuni efficiently internalized and killed C. jejuni 11168 and significantly upregulated surface MHC-II, CD40, CD80 and CD86 demonstrating a mature phenotype. Infected BM-DCs secreted significant amounts of TNFα, IL-6 and IL-12p70. Maximal activation of murine BM-DCs required internalization of C. jejuni; attachment alone was not sufficient to elicit significant responses. Also, various strains of C. jejuni elicited different magnitudes of cytokine production from BM-DCs. TLR2, TLR4, MyD88, and TRIF also played a role in C. jejuni-induced inflammatory activation of murine DCs (Rathinam et al., Reference Rathinam, Appledorn, Hoag, Amalfitano and Mansfield2009). DC upregulation of MHC-II and costimulatory molecules after C. jejuni challenge was profoundly impaired by TLR2, TLR4, MyD88, and TRIF deficiencies. Similarly, C. jejuni-induced secretion of IL-12, IL-6, and TNFα was significantly inhibited in TLR2^−/−, TLR4^−/−, MyD88^−/−, and TRIF^−/− DCs compared with WT DCs. Furthermore, C. jejuni infection induced IRF-3 phosphorylation and IFN-β secretion by DCs in a TLR4-TRIF dependent fashion, further demonstrating activation of this pathway by C. jejuni. Importantly, TLR4, MyD88, and TRIF deficiencies markedly impaired Th1-priming ability of C. jejuni-infected DCs. These results showed that C. jejuni-induced signaling through TLR4-MyD88 and TLR4-TRIF axes mediated maturation and inflammatory responses of DCs. Finally, in a coculture system, C. jejuni -infected BM-DCs induced high-level IFNγ production from CD4 + T cells indicating Th1 polarization. This finding correlates with in vivo studies demonstrating Th1-associated IgG2b antibody responses in IL-10^+/+ and IL-10^−/− mice of C57BL/6 (Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007), C3H and non-obese diabetic (NOD) genetic backgrounds (Mansfield et al., Reference Mansfield, Patterson, Fierro, Murphy, Rathinam, Kopper, Barbu, Onifade and Bell2008a) and in C57BL/129 mice (Fox et al., Reference Fox, Rogers, Whary, Ge, Taylor, Xu, Horwitz and Erdman2004) challenged orally with C. jejuni. However, excessive innate or T-cell mediated inflammatory responses in the intestine triggered by DCs in the absence of immunoregulatory elements, like IL-10, contribute to immune pathology as evident in the C57BL/6 IL-10^−/− enteritis model with lesions indistinguishable from human Crohn's disease patients (Mansfield et al., Reference Mansfield, Bell, Wilson, Murphy, Elsheikha, Rathinam, Fierro, Linz and Young2007, Reference Mansfield, Patterson, Fierro, Murphy, Rathinam, Kopper, Barbu, Onifade and Bell2008a; Bell et al., Reference Bell, St Charles, Murphy, Rathinam, Plovanich-Jones, Stanley, Wolf, Gettings, Whittam and Mansfield2009).

Humphrey et al., infected four commercial breeds of broiler chickens with C. jejuni M1 strain and demonstrated that some chickens experience disease and inflammation (Humphrey et al., Reference Humphrey, Chaloner, Kemmett, Davidson, Williams, Kipar, Humphrey and Wigley2014). Infected birds of all four breeds mounted an innate immune response similar to that seen in mammals, which was controlled in most breeds by upregulation of IL-10. Prolonged inflammatory responses were seen in one chicken breed along with diarrhea and GI lesions. Despite this large body of experimental infection studies in animals including chickens, little is known about whether AR C. jejuni strains elicit enhanced inflammatory responses post infection.

Disease associated with C. jejuni: GBS and autoimmunity in humans

C. jejuni has also been linked to the peripheral neuropathies, GBS and Miller Fisher Syndrome, as well as IBD, irritable bowel syndrome (IBS), and reactive arthritis (RA). All are significant autoimmune conditions associated with recent Campylobacter infection with GBS considered the leading cause of acute neuromuscular paralysis worldwide (Willison and Plomp, Reference Willison and Plomp2008). A recent meta-analysis by Keithlin et al. (Keithlin et al., Reference Keithlin, Sargeant, Thomas and Fazil2014) estimated the incidence of GBS following C. jejuni infection to be 7 per 10,000, while rates of IBS and RA were higher at 40 per 1000, and 28.6 per 1000, respectively. Tam et al. (Reference Tam, O'Brien, Petersen, Islam, Hayward and Rodrigues2007) have estimated an excess risk of 60% for development of GBS following C. jejuni infection in the UK (Tam et al., Reference Tam, O'Brien, Petersen, Islam, Hayward and Rodrigues2007). In June 2011, 26 patients in the southwestern USA were stricken with GBS after ingesting C. jejuni-contaminated water, showing that these sequelae can also occur in outbreaks (Jackson et al., Reference Jackson, Zegarra, Lopez-Gatell, Sejvar, Arzate, Waterman, Nunez, Lopez, Weiss, Cruz, Murrieta, Luna-Gierke, Heiman, Vieira, Fitzgerald, Kwan, Zarate-Bermudez, Talkington, Hill and Mahon2014). Yet, no prior studies have determined whether AR C. jejuni strains or AR or susceptible strains from animals are more likely to result in GBS and other long-term sequelae relative to susceptible strains.

Several forms of GBS are recognized, but the most common form occurring after C. jejuni infection is acute motor axonal neuropathy (AMAN) where damage to motor neurons occurs at the nodes of Ranvier. Acute motor sensory axonal neuropathy (AMSAN) can follow after AMAN leading to acute inflammatory demyelinating polyradiculoneuropathy (Oh et al., Reference Oh, LaGanke and Claussen2001). C. jejuni infection is well known to precede the AMAN form of GBS when bacterial LOS mimics host nerve gangliosides, induces autoantibody production and causes subsequent nerve damage. This molecular mimicry between C. jejuni LOS structures and gangliosides found in the membranes of peripheral nerve cells is the hypothesized mechanism for GBS (van den Berg et al., Reference van den Berg, Walgaard, Drenthen, Fokke, Jacobs and van Doorn2014). Upon infection lipooligosaccharides on the C. jejuni outer membrane elicit the production of antibodies that cross-react with gangliosides, such as GM1 and GD1a on peripheral nerves (van den Berg et al., Reference van den Berg, Walgaard, Drenthen, Fokke, Jacobs and van Doorn2014; St Charles et al., Reference St Charles, Bell, Gadsden, Malik, Cooke, Van de Grift, Kim, Smith and Mansfield2017). In human beings and other mammals, gangliosides located at or near the node of Ranvier on peripheral nerves are the target for these cross-reactive antibodies. Once these antibodies bind to the axolemma at the node, complement binds, a membrane attack complex forms, and this transmembrane channel leads to the disappearance of voltage-gated sodium channels. Subsequently, this can lead to detachment of paranodal myelin, nerve conduction failure and in some cases Wallerian degeneration (McGonigal et al., Reference McGonigal, Rowan, Greenshields, Halstead, Humphreys, Rother, Furukawa and Willison2010). Thereafter, macrophages are called in to remove myelin and axonal debris can cause more damage, but this cleanup improves the process of tissue repair (Martini and Willison, Reference Martini and Willison2016). Kaida and Kusunoki (Reference Kaida and Kusunoki2010) have associated several C. jejuni-associated neurological syndromes with antibodies to gangliosides GM1, GD1a, GT1a, and GQ1b (Kaida and Kusunoki, Reference Kaida and Kusunoki2010). In one study (Nachamkin et al., Reference Nachamkin, Liu, Li, Ung, Moran, Prendergast and Sheikh2002), C. jejuni strains associated with GBS cases had a high likelihood of having LOS resembling ganglioside GD1a, which is plentiful in the peripheral nervous system of human beings and mice (Lehmann et al., Reference Lehmann, Lopez, Zhang, Ngyuen, Zhang, Kieseier, Mori and Sheikh2007). Identity between C. jejuni LOS variants and seven different gangliosides has been demonstrated (Gilbert et al., Reference Gilbert, Karwaski, Bernatchez, Young, Taboada, Michniewicz, Cunningham and Wakarchuk2002). Some of these variations in LOS structure have been traced to polymorphisms in particular genes such as cst-II (Koga et al., Reference Koga, Takahashi, Masuda, Hirata and Yuki2005); for example, the role of cst-II was verified when a cst-II knockout strain of C. jejuni was shown to be incapable of evoking anti-ganglioside antibodies in knockout mice that lack the corresponding ganglioside and thus treat it as a foreign antigen (Godschalk et al., Reference Godschalk, Heikema, Gilbert, Komagamine, Ang, Glerum, Brochu, Li, Yuki, Jacobs, van Belkum and Endtz2004).

There is considerable between-strain variation in the branched terminal oligosaccharide portion (‘outer core’) of C. jejuni LOS structure; this variation can be attributed to a variety of genetic mechanisms (Gilbert et al., Reference Gilbert, Karwaski, Bernatchez, Young, Taboada, Michniewicz, Cunningham and Wakarchuk2002), including lateral gene transfer (Gilbert et al., Reference Gilbert, Godschalk, Karwaski, Ang, van Belkum, Li, Wakarchuk and Endtz2004; Phongsisay et al., Reference Phongsisay, Perera and Fry2006) and phase variation (Linton et al., Reference Linton, Gilbert, Hitchen, Dell, Morris, Wakarchuk, Gregson and Wren2000). Phase variation has been shown (1) to occur during human infection (Prendergast et al., Reference Prendergast, Tribble, Baqar, Scott, Ferris, Walker and Moran2004), (2) to alter invasiveness in tissue culture (Guerry et al., Reference Guerry, Szymanski, Prendergast, Hickey, Ewing, Pattarini and Moran2002) and (3) to enhance virulence during infection in mice (Jerome et al., Reference Jerome, Bell, Plovanich-Jones, Barrick, Brown and Mansfield2011). Horizontal gene exchange, which occurs at significant rates in C. jejuni (Suerbaum et al., Reference Suerbaum, Lohrengel, Sonnevend, Ruberg and Kist2001; Fearnhead et al., Reference Fearnhead, Smith, Barrigas, Fox and French2005), has also been implicated in the generation of LOS diversity (Parker et al., Reference Parker, Horn, Gilbert, Miller, Woodward and Mandrell2005; Phongsisay et al., Reference Phongsisay, Perera and Fry2006). Extensive variation has also been demonstrated in the chromosomal region containing genes required for LOS synthesis (Gilbert et al., Reference Gilbert, Karwaski, Bernatchez, Young, Taboada, Michniewicz, Cunningham and Wakarchuk2002; Godschalk et al., Reference Godschalk, Heikema, Gilbert, Komagamine, Ang, Glerum, Brochu, Li, Yuki, Jacobs, van Belkum and Endtz2004; Knudsen et al., Reference Knudsen, Bang, Nielsen and Madsen2005; Parker et al., Reference Parker, Horn, Gilbert, Miller, Woodward and Mandrell2005). To date, five major ‘LOS classes’ have been identified depending on the genetic composition of the complex LOS locus (Parker et al., Reference Parker, Horn, Gilbert, Miller, Woodward and Mandrell2005). Godschalk et al. (Reference Godschalk, Bergman, Gorkink, Simons, van den Braak, Lastovica, Endtz, Verbrugh and van Belkum2006) identified three GBS-associated genetic loci within the LOS region; all three encode galactosyltransferases presumably involved in the synthesis of the terminal oligosaccharide (Godschalk et al., Reference Godschalk, Bergman, Gorkink, Simons, van den Braak, Lastovica, Endtz, Verbrugh and van Belkum2006).

Disease associated with C. jejuni: GBS in animals

Currently, rats and chickens serve as the main animal models of GBS and other peripheral neuropathies. Rats injected with myelin preparations mixed with complete Freund's adjuvant develop neurological disease and lesions. Rats show experimental autoimmune neuritis (EAN) after this immunization and exhibit mononuclear cell infiltration and demyelinated nerve fibers in peripheral nerves, e.g. the sciatic nerve (Archelos et al., Reference Archelos, Fortwangler and Hartung1997). Rats with experimental autoimmune polyradiculo-neuropathy and nerve biopsies of GBS patients both had T cells within the epi- and perineurium, expressing a TNFα-converting enzyme likely active in these demyelinating disorders (Kurz et al., Reference Kurz, Pischel, Hartung and Kieseier2005). However, myelin-injected rats do not model the AMAN form of GBS induced by C. jejuni (Nachamkin et al., Reference Nachamkin, Allos and Ho1998). Chickens acquire a form of peripheral neuropathy secondary to C. jejuni inoculation with GBS patient strains (Li et al., Reference Li, Xue, Tian, Liu and Yang1996). In chickens given a C. jejuni Penner serotype O:19 strain orally, 33% of the chickens became paralyzed within 12 days. In paralyzed chickens, early lesions included nodal lengthening and paranodal demyelination that was later followed by Wallerian-like degeneration and even paranodal re-myelination in some long-term survivors. Thus, chickens inoculated orally with GBS associated C. jejuni strains from patients can be considered a naturally occurring disease model for GBS. Work with patient samples, using rats and chickens, has helped to test the molecular mimicry hypothesis between C. jejuni LOS and peripheral nerve gangliosides as a mechanism for anti-ganglioside antibody induction.

C. jejuni strains from patients with GBS can produce neurological disease in mice after a single oral infection allowing the mechanisms controlling this type of autoimmunity to be studied in these models (Malik et al., Reference Malik, Sharma, St Charles, Dybas and Mansfield2014; St Charles et al., Reference St Charles, Bell, Gadsden, Malik, Cooke, Van de Grift, Kim, Smith and Mansfield2017). The main mechanism resulting in peripheral nerve dysfunction is molecular mimicry; some C. jejuni strains with sialylated outer core oligosaccharides that mimic host gangliosides (GM1/GD1a) on the peripheral nerves induced antiganglioside antibodies that attached to and damaged peripheral nerves in experimental trials. This was accompanied by peripheral neurological disease as measured on several phenotyping apparatuses. When the chloramphenicol resistant C. jejuni strain 260.94 was used for experimental infections, chloramphenicol treatment worsened the outcomes and intensified these autoimmune responses (St Charles et al., Reference St Charles, Bell, Gadsden, Malik, Cooke, Van de Grift, Kim, Smith and Mansfield2017). Contrasting T cell responses were produced after C. jejuni infection in C57BL/6 IL-10^−/− mice (Malik et al., Reference Malik, Sharma, St Charles, Dybas and Mansfield2014). C. jejuni strains with α−2,3 sialyation in the outer core of the LOS produced Th2 responses and induction of antiganglioside antibodies against the nerves. C. jejuni strains without this branching produced Th1 responses and caused inflammation in the colon. These GBS mouse models are available as test systems to develop new therapeutics and to test the role of AR C. jejuni in driving this autoimmune disease.

Interestingly transplanted human fecal microbiota enhanced GBS autoantibody responses after C. jejuni infection in C57BL/6 mice. Inoculating germ-free C57BL/6 WT mice with a mixed human fecal slurry provided a murine model that stably passed its microbiota over >20 generations. ^Humicrobiota conferred many changes upon the WT model in contrast to previous results, which showed only colonization with no disease after C. jejuni challenge. When compared with ^Convmicrobiota mice for susceptibility to C. jejuni enteric or GBS patient strains, infected ^Humicrobiota mice had (1) 10–100-fold increases in C. jejuni colonization of both strains, (2) pathologic change in draining lymph nodes but not colon or cecal lamina propria, (3) significantly lower Th1/Th17-dependent anti-C. jejuni responses, (4) significantly higher IL-4 responses at 5 but not 7 weeks post infection (PI), (5) significantly higher Th2-dependent anti-C. jejuni responses, and (6) significantly elevated antiganglioside autoantibodies after C. jejuni infection. These responses in ^Humicrobiota mice were correlated with a dominant Bacteroidetes and Firmicutes microbiota. Thus, this human microbiota also enhanced susceptibility to this autoimmune disease.

C. jejuni strain-dependent differences in disease outcomes

It has been suggested that colitis and GBS disease are C. jejuni strain dependent. Testing of many C. jejuni strains in the same inbred mouse model supports this hypothesis (Bell et al., Reference Bell, St Charles, Murphy, Rathinam, Plovanich-Jones, Stanley, Wolf, Gettings, Whittam and Mansfield2009). All C. jejuni strains did not produce similar disease when tested in mouse models and it has been suggested that the outcomes of hemorrhagic colitis and autoimmunity such as GBS are strain dependent (Malik et al., Reference Malik, Sharma, St Charles, Dybas and Mansfield2014). C57BL/6 IL-10^−/− mice infected with isolates from patients with colitis had significantly upregulated type 1 and 17 but not type 2 cytokines in the colon coincident with infiltration of phagocytes, T cells and innate lymphoid cells (ILCs). Both ILC and T cells contributed to the interferon-c, IL-17, and IL-22 upregulation, but in a time- and organ-specific manner. However, T cells were necessary for colitis as mice depleted of Thy-1þ cells were protected while neither Rag1-/- nor IL-10R blocked Rag1-/- mice developed colitis after infection. Depleting IFN-c, IL-17, or both significantly decreased colitis and drove colonic responses toward type 2 cytokine and antibody induction. However, C. jejuni strains from GBS patients induced mild colitis with mixed cytokine profiles associated with blunted type 1/17 but enhanced type 2 responses. It was only the type 2 antibody isotypes that cross-reacted with peripheral nerve gangliosides producing autoimmunity. Thus, contrasting T-cell responses contributed to either colitis or autoimmunity in C57BL/6 IL-10^−/− mice (Malik et al., Reference Malik, Sharma, St Charles, Dybas and Mansfield2014). This reflects the roles of Th1 responses in killing intracellular pathogens and of Th2 responses in the induction of autoimmunity (Nurieva and Chung, Reference Nurieva and Chung2010).

Arguments against strict strain dependent disease pathotypes include, (1) varied animal reservoirs such as poultry, swine, cattle, and fecal contaminated water are the most common sources of human infections, and (2) in one outbreak C. jejuni isolates from patients with colitis and GBS were genetically identical. These outcomes may be explained by recognizing that not all C. jejuni strains have ganglioside mimics on their surface, but have other glycans (Nachamkin et al., Reference Nachamkin, Liu, Li, Ung, Moran, Prendergast and Sheikh2002). Moreover, many individuals infected with C. jejuni bearing self-glycans will maintain tolerance and elicit protective responses against other surface molecules of the bacterium (Willison et al., Reference Willison, Jacobs and van Doorn2016), and the mechanisms controlling B-cell tolerance to T-cell-independent antigens are largely unknown.

C. jejuni within-host adaptation can enhance disease

C. jejuni's genome is not static during in vivo passage (Wassenaar et al., Reference Wassenaar, Geilhausen and Newell1998; Nuijten et al., Reference Nuijten, Berg, Formentini, Zeijst and Jacobs2000; de Boer et al., Reference de Boer, Wagenaar, Achterberg, van Putten, Schouls and Duim2002; Jerome et al., Reference Jerome, Bell, Plovanich-Jones, Barrick, Brown and Mansfield2011; Kim et al., Reference Kim, Artymovich, Hall, Smith, Fulton, Bell, Dybas, Mansfield, Tempelman, Wilson and Linz2012; Kivistö et al., Reference Kivistö, Kovanen, Skarp-de Haan, Schott, Rahkio, Rossi and Hänninen2014). A significant proportion of this genomic variation occurs in virulence-associated genes that are involved in the synthesis of antigenic structures including the LOS, flagella, and capsule that are involved in triggering immune responses and potentially aiding in immune evasion (Jerome et al., Reference Jerome, Bell, Plovanich-Jones, Barrick, Brown and Mansfield2011; Kivistö et al., Reference Kivistö, Kovanen, Skarp-de Haan, Schott, Rahkio, Rossi and Hänninen2014). Some genomic variants have direct links to biological outcomes, such as increased motility; thus C. jejuni adaptation does influence infection outcomes.

The first evidence for C. jejuni adaptation in vivo came from variability in pulse-field gel electrophoresis banding patterns following passage in chickens, where analysis of initially clonal isolates of C. jejuni revealed multiple banding patterns in recovered isolates, providing evidence that large-scale genomic rearrangements occurred during in vivo passage (Wassenaar et al., Reference Wassenaar, Geilhausen and Newell1998). Concurrently, Parkhill et al., Reference Parkhill, Wren, Mungall, Ketley, Churcher, Basham, Chillingworth, Davies, Feltwell, Holroyd, Jagels, Karlyshev, Moule, Pallen, Penn, Quail, Rajandream, Rutherford, van Vliet, Whitehead and Barrell2000 identified hypervariable regions in the C. jejuni 11168 genome consisting of homopolymeric tracts of nucleotides (Parkhill et al., Reference Parkhill, Wren, Mungall, Ketley, Churcher, Basham, Chillingworth, Davies, Feltwell, Holroyd, Jagels, Karlyshev, Moule, Pallen, Penn, Quail, Rajandream, Rutherford, van Vliet, Whitehead and Barrell2000). Since this discovery, our laboratory has shown that insertions or deletions in homopolymeric tracts of nucleotides allow C. jejuni 11168 to rapidly adapt during passage in mice (Jerome et al., Reference Jerome, Bell, Plovanich-Jones, Barrick, Brown and Mansfield2011). C. jejuni farm isolates also contained variants in homopolymeric tracts (Kivistö et al., Reference Kivistö, Kovanen, Skarp-de Haan, Schott, Rahkio, Rossi and Hänninen2014). In both cases, the majority of variants in homopolymeric tracts were found in the LOS, capsular, and flagellar genes. Collectively, these homopolymeric tracts were called contingency loci as they have higher rates of mutation than the rest of the genome. Mutations in contingency loci contribute to phase variation: the ability to turn gene expression on or off (Moxon et al., Reference Moxon, Paul, Martin and Richard1994); phase variation may directly impact pathogenesis by altering the expression of virulence factors including LOS, capsule, and flagella (Jerome et al., Reference Jerome, Bell, Plovanich-Jones, Barrick, Brown and Mansfield2011).

Variation in homopolymeric tract length can result in observable biological outcomes. Notably, the passage of C. jejuni in vivo led to the presence of antigenic ganglioside mimics on the LOS of C. jejuni 81–176 that initially lacked any ganglioside mimics (Prendergast et al., Reference Prendergast, Tribble, Baqar, Scott, Ferris, Walker and Moran2004). Site-directed mutagenesis of homopolymeric tracts in the cgtA gene (N-acetyl-galactosaminyl transferase) in C. jejuni 81–176 shifted the ratio of GM2 and GM3 ganglioside mimics and enhanced the invasiveness of the C. jejuni cgtA mutant compared with the WT strain (Guerry et al., Reference Guerry, Szymanski, Prendergast, Hickey, Ewing, Pattarini and Moran2002). The host environmental cues that drive evolutionary selection for phase variation are unknown, but it is known that this process allows for rapid adaptation to novel environments, increased diversity, and evasion of the host immune system (van der Woude and Baumler, Reference van der Woude and Baumler2004; Jerome et al., Reference Jerome, Bell, Plovanich-Jones, Barrick, Brown and Mansfield2011). Slipped-strand mutagenesis (Moxon et al., Reference Moxon, Paul, Martin and Richard1994; Zhou et al., Reference Zhou, Aertsen and Michiels2014) and the absence of several homologs of E. coli DNA repair genes contribute to the high incidence of phase variation in C. jejuni (van der Woude and Baumler, Reference van der Woude and Baumler2004). To our knowledge little has been done to understand the role of antibiotic selection on phase variability.

Other mechanisms of C. jejuni adaptation that affects disease

C. jejuni has other mechanisms of adaptation to novel environments in addition to variation in homopolymeric tracts. Several recent studies have demonstrated that single nucleotide variants outside of homopolymeric tracts contribute genetic diversity to C. jejuni strains (Cagliero et al., Reference Cagliero, Cloix, Cloeckaert and Payot2006; Mohawk et al., Reference Mohawk, Poly, Sahl, Rasko and Guerry2014; Thomas et al., Reference Thomas, Lone, Selinger, Taboada, Uwiera, Abbott and Inglis2014). In addition, phenotypic adaptation in C. jejuni has also been observed. In vivo passage conferred increased motility in mice and rabbits (Caldwell et al., Reference Caldwell, Guerry, Lee, Burans and Walker1985; Jones et al., Reference Jones, Marston, Woodall, Maskell, Linton, Karlyshev, Dorrell, Wren and Barrow2004), although this is not surprising because motility is a major factor in C. jejuni colonization (Nachamkin et al., Reference Nachamkin, Yang and Stern1993; Wassenaar et al., Reference Wassenaar, Zeijst, Ayling and Newell1993; Yuki et al., Reference Yuki, Taki, Inagaki, Kasama, Takahashi, Saito, Handa and Miyatake1993). It has also been shown that C. jejuni virulence increases following passage both in similar (Bell et al., Reference Bell, St Charles, Murphy, Rathinam, Plovanich-Jones, Stanley, Wolf, Gettings, Whittam and Mansfield2009) or divergent hosts (Kim et al., Reference Kim, Artymovich, Hall, Smith, Fulton, Bell, Dybas, Mansfield, Tempelman, Wilson and Linz2012), which may result from enhanced colonization potential (up to 10,000-fold increase) (Cawthraw et al., Reference Cawthraw, Wassenaar, Ayling and Newell1996), alterations in virulence-associated gene expression (Bell et al., Reference Bell, Jerome, Plovanich-Jones, Smith, Gettings, Kim, Landgraf, Lefébure, Kopper, Rathinam, Charles, Buffa, Brooks, Poe, Eaton, Stanhope and Mansfield2013a), or the presence of antigenic stimuli based on modifications of surface structures of C. jejuni (Prendergast et al., Reference Prendergast, Tribble, Baqar, Scott, Ferris, Walker and Moran2004). Intragenomic recombination can also occur during passage and even restore motility in non-motile flaA mutants (Nuijten et al., Reference Nuijten, Berg, Formentini, Zeijst and Jacobs2000) or enhance phage resistance (Scott et al., Reference Scott, Timms, Connerton, Loc Carrillo, Adzfa Radzum and Connerton2007). Intergenomic recombination is possible, but current data are limited to recombination between C. jejuni strains with highly similar gene content (de Boer et al., Reference de Boer, Wagenaar, Achterberg, van Putten, Schouls and Duim2002). Together, these results confirm the plasticity of the C. jejuni genome while establishing that adaptation of C. jejuni often affects loci with the potential to modulate host immune responses.