ABSTRACT
Toll-like receptors (TLRs) are a major component of the pattern recognition receptor repertoire that detect invading microorganisms and direct the vertebrate immune system to eliminate infection. In chickens, the differential biology of Salmonella serovars (systemic versus gut-restricted localization) correlates with the presence or absence of flagella, a known TLR5 agonist. Chicken TLR5 (chTLR5) exhibits conserved sequence and structural similarity with mammalian TLR5 and is expressed in tissues and cell populations of immunological and stromal origin. Exposure of chTLR5+ cells to flagellin induced elevated levels of chicken interleukin-1β (chIL-1β) but little upregulation of chIL-6 mRNA. Consistent with the flagellin-TLR5 hypothesis, an aflagellar Salmonella enterica serovar Typhimurium fliM mutant exhibited an enhanced ability to establish systemic infection. During the early stages of infection, the fliM mutant induced less IL-1β mRNA and polymorphonuclear cell infiltration of the gut. Collectively, the data represent the identification and functional characterization of a nonmammalian TLR5 and indicate a role in restricting the entry of flagellated Salmonella into systemic sites of the chicken.
The pattern recognition receptors (PRRs) play a central role in the rapid initiation of host immune responses and the generic identification of an invading pathogen (36, 43) by recognition of pathogen-associated molecular patterns. Toll-like receptors (TLRs) have emerged as a major component of the vertebrate PRR repertoire. Upon activation, TLRs induce the expression of a wide range of immunoregulatory and effector molecules (41, 51) and maturation of immune cell types (1, 3, 11, 24, 50).
A range of TLR genes has been identified in nonmammalian vertebrates including chicken (10, 18, 32) and fish (6, 26, 37). To date, avian orthologues of TLR2 and TLR4 have been characterized and expressed sequence tags (ESTs) with sequence homologies to TLR1, -6, or -10; TLR3; TLR5; and TLR7 have been identified (34, 48; our unpublished results). Two chicken TLR2 (chTLR2) molecules (type 1 and type 2) were identified that lie in a tandem arrangement within a genomic region expressing conserved synteny to mammals (10, 18). The chTLR4 gene was also demonstrated to lie in a region of conserved synteny and has been associated with susceptibility to systemic infection with Salmonella enterica serovar Typhimurium in young chickens (32). Collectively, these data indicate that a range of distinct TLR genes, orthologous to the mammalian TLR repertoire, were present before the divergence of birds and mammals over 300 million years ago.
The observation that nonflagellated Salmonella enterica serovars (Gallinarum or Pullorum) typically cause more acute systemic infection than flagellated serovars (Typhimurium or Enteritidis) provoked our interest in chTLR5. Our working hypothesis was that TLR5-flagellin interactions contribute to the broad biology of Salmonella serovars in the chicken. We identified a chicken orthologue for TLR5, determined expression patterns in tissues, and isolated immune cell populations and cultured cells. Exposure of chTLR5+ cells to flagellin induced upregulation of chicken interleukin-1β (chIL-1β), and the differential biology of aflagellar and intact serovar Typhimurium in the chicken revealed a likely role for TLR5 in avian salmonellosis.
MATERIALS AND METHODS
Animals and Salmonella.Specific-pathogen-free inbred line 72 (White Leghorn) and Rhode Island Red chickens were bred at the Institute for Animal Health (IAH) and reared under conventional conditions. A spontaneous nalidixic-acid-resistant mutant of S. enterica serovar Typhimurium strain F98 (phage type 14) and serovar Typhumurium F98 fliM::Km are described elsewhere (4, 55). Bacterial stocks were stored in glycerol at −70°C and grown in Luria-Bertani broth at 37°C in an orbital shaking incubator at 150 rpm.
Database mining and BAC identification.Chicken EST databases (12, 53) were screened by BLAST with human TLR5 (huTLR5) sequences (NM003268). Primers were designed from a putative chTLR5 EST and used to generate amplification products from genomic DNA for hybridization screening of the Wageningen Bacterial Artificial Chromosome (BAC) library (14) with filters supplied by the Medical Research Council Human Genome Mapping Project Resource Centre, Cambridge, United Kingdom.
Positive BAC clones were sequenced (Lark Technologies, Saffron Walden, United Kingdom) and analyzed using the NIX program (http://www.hgmp.mrc.ac.uk/NIX ). Alignments were performed using Clustal W version 10 (52), and structural domains were predicted using SMART (http://smart.embl-heidelberg.de ).
Tissues, sorted cell populations, and cultured cells.Tissue samples were obtained from 8-week-old line 72 chickens. Splenic B, T-cell receptor gamma delta (TCRγδ), TCRαβ1, and TCRαβ2 cells were enriched using the AutoMACS system (Miltenyi Biotech, Surrey, United Kingdom) according to the manufacturer's instructions. Briefly, spleens were disrupted and lymphocytes were isolated using Ficoll-Paque Plus (Amersham Biosciences, Uppsala, Sweden). Splenocytes were labeled with anti-Bu-1 phycoerythrin, anti-TCRγδ (immunoglobulin G1 [IgG1]), or anti-TCRαβ2 (IgG1) at 200 μl/108 cells or with TCRαβ1 phycoerythrin at 500 μl/108 cells in phosphate-buffered saline-0.5% bovine serum albumin for 15 min on ice. Following washing, cell populations were isolated by positive selection (purity, >95%) by using antifluorochrome or rat anti-mouse IgG1-conjugated paramagnetic beads (Miltenyi Biotech). Monocyte-derived macrophages and heterophils were obtained from peripheral blood as described previously (28, 57).
Primary cultures of chicken kidney cells (CKC) and embryonic (e10d) fibroblasts (CEF) were prepared according to standard techniques. All cells were cultured in RPMI 1640 medium supplemented with penicillin-streptomycin, 7% fetal calf serum, and 3% chicken serum at 41°C under 5% CO2. The macrophage-like HD11 cell line (9) and the bursal lymphoma cell line DT40 (ATCC accession number CRL2111 ) were cultured in supplemented RPMI medium at 41°C under 5% CO2.
Treatment of cultured cells with serovar Typhimurium flagellin.Confluent monolayers of CEF, CKC, or HD11 cells in 24-well plates (Nunc, Roskilde, Denmark) were treated with purified serovar Typhimurium-derived flagellin (InvivoGen, San Diego, Calif.) for 6 h (19). Analysis was performed with three independent experiments, each including triplicates of flagellin-treated and mock-treated wells.
RNA extraction.Tissue samples were stored in RNAlater (QIAGEN Ltd. Crawley, United Kingdom) at −20°C before disruption by homogenization (Mini-bead beater; Biospec Products, Bartlesville, Okla.). Isolated cell subsets or cultured cells were disrupted directly in RLT buffer (QIAGEN) and frozen at −20°C until RNA extraction. RNA was extracted with the RNeasy Mini kit (QIAGEN) according to the manufacturer's instructions. Contaminating DNA was digested on column with RNase-free DNase 1 (QIAGEN) for 90 min at room temperature. The RNA was eluted with RNase-free water and stored at −80°C.
RT-PCR.cDNA was obtained using a reverse transcription (RT) kit (Promega, Southampton, United Kingdom) according to the manufacturer's instructions. All cDNA preparations were standardized by RT-PCR for β-actin with primers (forward, TGCTGTGTTCCCATCTATCG; reverse, TTGGTGACAATACCGTGTTCA; accession number L08165 ) and were free from genomic DNA contamination (data not shown).
PCR conditions were as follows: cDNA (1 to 2 μg), 200 μM deoxynucleoside triphosphates, 1× reaction buffer, 0.5 U of Taq polymerase (Promega, Southhampten, United Kingdom), and primers for chTLR5 (forward, TGCACATGTTTTCTCCTAGGT; reverse, CCACATCTGACTTCTGCCTTT) at 5 pM in a 50-μl final reaction volume with amplification (iCycler; Bio-Rad, Hemel Hampstead, United Kingdom) at 1 cycle at 95°C (2 min); 30 cycles at 95°C (30 s), 58°C (1 min), and 72°C (2 min); and 1 cycle at 72°C (10 min).
Quantitative analysis of cytokine mRNA.The mRNA levels of chicken IL-1β, IL-6, and 28S rRNA were quantified by real-time RT-PCR using the ABI Prism 7700 sequence detection system (TaqManR; PE Applied Biosystems, Warrington, United Kingdom) as described previously (27, 29).
Fold differences (R) in the expression of cytokine genes between samples (A and B), each relative to reference rRNA, were calculated from the CT values C (for the cytokine) and C′ (for rRNA) by using the equation ln R(A/B) = (CA -CB)/S -(C′A-C′B)/S′, where S and S′ are, respectively, the slopes of plots of the CT value against the natural logarithm of concentration for serial 10-fold dilutions of cytokine DNA and rRNA, assayed on the same plate. This calculation avoids assumptions about the efficiency of the PCR amplifications and reduces to the common ΔΔCT method in the case that both have perfect efficiency.
In vivo challenge.Groups of 20 1-day-old specific-pathogen-free Rhode Island Red chickens were challenged orally with 2.5 × 108 CFU of wild-type or fliM mutant serovar Typhimurium F98 strain or mock treated. Birds were sacrificed at 0, 9, 24, and 48 h postinfection (p.i.) for bacterial analysis in the spleen, liver, and cecal contents as described previously (58). Samples of small intestine and cecal tonsil (CT) were fixed in 10% formalin saline for histology and in RNAlater for cytokine mRNA analysis. All experimental procedures satisfied the requirements of local and national regulation and ethical review with appropriate licenses.
Nucleotide sequence accession number.The coding sequence for the chTLR5 was deposited in the EMBL database under accession number AJ626848 .
RESULTS
Identification and mapping of chTLR5.An EST with sequence homology to huTLR5 (BI066471) was identified and used to identify three BACs (Wag002L14, Wag019C29, and Wag501K15) which mapped (by fluorescent in situ hybridization) to the centromeric region 3q11-3q21 of chicken chromosome 3 (unpublished data). Analysis of the Wag002L14 sequence identified a region with homology to huTLR5. A chTLR5 open reading frame was identified in a single exon (as with mammalian TLR5) and confirmed by sequence of RT-PCR products.
Comparison of the chTLR5 amino acid sequence with huTLR and mouse TLR (moTLR) revealed the highest identity with TLR5 (50 and 46%), with higher identity in the TIR domain (70 and 69%). Amino acid sequences of chTLR5, huTLR5, and mTLR5 were aligned, and predicted domain analysis (31) showed conservation of the TLR5 LRR pattern (Fig. 1).
Amino acid sequence of chTLR5 protein and alignment with huTLR5 and moTLR5 proteins. Accession numbers were AJ626848 for chTLR5, NM003268 for huTLR5, and AF186107 for mTLR7. Alignment was performed using the ClustalX program and edited in the Genedoc program; shading was performed using the conserved mode (black shading for conserved residues and light grey shading for similar residues). The double continuous lines denote leucine-rich repeats, the single continuous line indicates transmembrane domain, and the dotted line represents the TIR domain.
Expression of chTLR5 mRNA and flagellin-induced cytokine production.ChTLR5 mRNA was expressed in a broad range of tissues, including those with minimal (e.g., kidney and liver) or substantial (e.g., spleen and gut) immune compartments (Fig. 2A). The strongest signals were obtained with colon, spleen, kidney, lung, heart, and testes. With the immune cell fractions, chTLR5 message was detected with all cell types examined, including monocyte-derived macrophages, heterophils, and B- and T-cell subsets (Fig. 2B). Although chTLR5 message was detected in all four of the cultured cell populations, the relative intensity of the RT-PCR products differed (Fig. 2C), with strongest signals from the stromal-cell-derived populations (CEF and CKC).
Expression of chTLR5 mRNA in tissues (A), immune cell populations (B), and cultured cells (C) by RT-PCR from 8-week-old line 72 chickens. Products were resolved by agarose gel electrophoresis and visualized with ethidium bromide. Data are representative of three PCR amplifications with independent samples. Small Int., small intestine.
Exposure of cultured CEF, CKC, and HD11 cells to Salmonella-derived flagellin led to the upregulation of chIL-1β mRNA (Fig. 3) 114-fold, 6-fold, and 3-fold, respectively. In contrast, chIL-6 mRNA was not significantly increased by exposure of the cultured cells to the flagellin preparation.
Exposure of chicken cell lines to flagellin induces upregulation of chIL-1β mRNA but not chIL-6 mRNA. Three adherent cell cultures (CEF, CKC, and HD11) plated in 24-well plates (4 wells/treatment) were exposed to 10, 100, or 200 ng of purified Salmonella-derived flagellin/ml or were mock treated for 6 h. Real-time RT-PCR (Taqman) was used to determine the levels of cytokine and 28S mRNA, with the latter used to normalize the cytokine data, depicted as the mean relative fold increase ± standard error, compared with mock-treated wells. An asterisk indicates a statistically significant difference (P < 0.05) between flagellin-treated and mock-treated wells. The data are representative of three independent experiments.
Flagella status affects serovar Typhimurium infection.To assess the importance of flagella-host interactions during avian serovar Typhimurium infection, we analyzed the bacteriology, cytokine induction, and pathology after challenge with wild-type or flagella-deficient fliM mutant bacteria. The numbers of wild-type or fliM mutant Salmonella in the cecal contents were equivalent at both 9 and 24 h p.i. (Fig. 4A). In contrast, the numbers of aflagellar fliM bacteria in the liver at 9 h p.i. were substantially higher (103/g) than those with wild-type serovar Typhimurium (undetectable). At 24 h p.i., the numbers of fliM mutant and wild-type bacteria in the liver were comparable, indicating that flagella deficiency only enhances the initial phases of systemic colonization.
Numbers of aflagellar (fliM mutant) and wild-type (WT) S. enterica serovar Typhimurium F98 cells in cecal contents and livers (A) and heterophil infiltration in the cecal wall (B) after oral infection of 3-day-old chicks (4/group). (A) Bars represent mean numbers of Salmonella CFU/g ± standard error on a logarithmic scale. Asterisks indicate statistically significant differences between flagellin-treated and mock-treated wells (P < 0.05). ND, not detected. (B) Heterophil infiltration was assessed microscopically in hematoxylin-eosin-stained sections of cecal wall at 9, 24, and 48 h p.i. +/−, relative degree of infiltration; -, level of infiltration in the uninfected chickens; +++, heavy infiltrate with numerous heterophils in the tissue.
To characterize the effect of flagella status on the induction of host responses, the levels of cytokine mRNA (chIL-1β and chIL-6) and inflammatory infiltrate in the gut were assessed. The CT represents a major site for the invasion of serovar Typhimurium, whereas fewer bacteria appear to enter the body in small intestinal sites (5). Greater increases in the amounts of mRNA for proinflammatory cytokines were detected in CT than in ileal tissue samples (Fig. 5). At 9 h p.i., CT chIL-1β mRNA levels reflected the flagella status of the Salmonella, less being detected with aflagellar bacteria. Small differences were evident with ileal samples at 9 h p.i., but these were not statistically significant. At 24 h p.i., levels of CT chIL-1β mRNA were elevated (∼4,500-fold) but there were no differences related to flagella status. Levels of chIL-6 mRNA were increased at 24 h p.i. (Fig. 5), with higher amounts induced by flagella-intact serovar Typhimurium (P < 0.05). Microscopic examination of the cecal wall revealed the rapid development of a polymorphonuclear (PMN) cell infiltrate (Fig. 4B) which was slower during infection with aflagellar serovar Typhimurium.
The levels of chIL-1β and chIL-6 mRNA in CT and small intestine (Small int.) of chicks challenged with aflagellar (fliM mutant) or wild-type (wt) S. enterica serovar Typhimurium F98. Bars represent the mean fold increases ± standard errors (4 birds/group) of cytokine mRNA by real-time PCR (Taqman) using uninfected birds as the fold denominator. *, statistically significant difference between infected and uninfected chickens (P < 0.05); **, statistically significant difference between chickens infected with wild-type and fliM mutant Salmonella (P < 0.05).
DISCUSSION
TLRs represent a major component of the vertebrate PRR system which affords the ability to detect and to differentiate between major groups of invading microorganisms. TLR-derived signals are important in the initiation and fine tuning of responses to mediate protective immunity. Innate recognition of Salmonella enterica is mediated by a range of TLR-pathogen-associated molecular pattern interactions, including TLR4-lipopolysaccharide (LPS), TLR5-flagellin, and TLR9-unmethylated CpG motifs. Indeed, involvement of TLR4 in the protection against systemic salmonellosis has been described for both mammals and chickens (32, 39, 40). With MOLF/Ei mice, TLR5 is linked with resistance to systemic salmonellosis (46) and flagellin-TLR5 interactions mediate Salmonella recognition by enterocytes (19, 20, 60, 61). In chickens, aflagellar serovars (Gallinarum or Pullorum) typically cause more severe systemic infection than flagella-intact serovars (Typhimurium or Enteritidis). These observations stimulated our work to characterize chTLR5 and to assess the role for flagellin-dependent interactions in defining the biology of Salmonella in the chicken.
In mammals, a repertoire of 13 TLRs has been described. With chickens, only chTLR2 and chTLR4 have been characterized in detail (18, 32), although EST sequences with similarity to mammalian TLR (TLR1, -6, or -10), TLR3, TLR5, and TLR7 have been reported (34, 48). A range of TLR agonists has been shown to stimulate the production of nitric oxide and cytokines and/or changes in cell surface molecules with cultured chicken cells. These agonists include LPS (15), lipoteichoic acid (17), polyinosinic-poly(C) (35) and unmethylated CpG DNA motifs (23).
The chTLR5 gene was identified and localized to the previously defined region of homology shared between the proximal region of chicken chromosome 3, human chromosome 1, and mouse chromosome 1 (21, 44). The level of amino acid identity for chTLR5 and huTLR5 is similar to that for chTLR2 and huTLR2 and that for chTLR4 and huTLR4 (18, 32). Comparisons of gene structure, genomic location, amino acid composition, and patterns of LRR support the premise that chTLR5 is an orthologue of mammalian TLR5. Moreover, chTLR5 mRNA is expressed with a similarly broad cell and tissue distribution to mammalian TLR5 (25, 59).
Bacterial flagellin is an agonist for mammalian (22) and fish TLR5 (54) and stimulated chTLR5+ cells to upregulate chIL1β mRNA but not nitric oxide or chIL-6 mRNA. Provocatively, the degree of flagellin-induced IL-1β mRNA correlated with the level of TLR5 RT-PCR product detected. Similarly, levels of chTLR4 correlated with the amount of LPS-induced nitric oxide in macrophages (15, 16). The failure of flagellin to induce chIL-6 mRNA was unexpected and contrasted with the exposure of CKC to live serovar Typhimurium (27) and the production of IL-6 by mammalian cells (20). With live Salmonella, CKC upregulated chIL6 mRNA after exposure to flagellated serovars (Typhimurium, Enteritidis, and Dublin) but not aflagellated serovar Gallinarum (27). However, our results suggest that the apparent correlation of CKC IL-6 response with flagella status is not mediated by flagellin-chTLR5 interactions. Exposure of CKC to live serovar Typhimurium or serovar Enteritidis failed to stimulate chIL-1β (27), suggesting that some serovars may subvert flagellin-TLR5 responses.
Analyses of epithelial-dominated CKC cultures are not necessarily representative of early in vivo interactions with enterocytes that express TLR5 only on the basolateral surface (19). Unfortunately, well-defined enterocyte-like cell lines are not available for the chicken and an in vivo approach using the aflagellar serovar Typhimurium F98 fliM mutant and parental wild-type F98 was adopted for subsequent studies. Consistent with previous studies (2), the numbers of cecal salmonellae were independent of flagella status at both 9 and 24 h p.i. Despite the comparable availability of both strains for invasion, higher numbers of aflagellar salmonellae were detected in the liver at 9 h p.i., indicating a competitive advantage in early systemic colonization. This difference was not seen at 24 h p.i., and the advantage of aflagellar status was short-lived and probably related to an ability to evade early host recognition.
TLR5 has been shown to play a significant role in the susceptibility of MOLF/Ei mice to systemic salmonellosis, independent of either NRAMP or TLR4 loci (46, 47). Interestingly, MOLF/Ei mice succumb to infection with lower bacterial loads than C57BL/6 (natural resistance-associated macrophage protein [NRAMP])s mice, indicating a complex interplay between bacterial load and disease outcome (45). A wide range of flagella-deficient strains of serovar Typhimurium have been examined for virulence in rodents either after intravenous, intraperitoneal, or oral routes with sometimes contradictory conclusions (13, 33, 42, 49). The early, enhanced systemic infection associated with aflagellar status has also been documented after oral challenge of mice, and the transient nature of the effect may represent reduced intramacrophage survival (42, 56). The transient nature of aflagellar advantage in chickens would not have been detected in studies with fliC serovar Enteritidis at 24 to 72 h p.i. (2, 38).
The reduced levels of CT chIL1β and chIL6 seen after challenge with aflagellar serovar Typhimurium are consistent with the reduced enteric PMN infiltration also noted with murine colitis (49). Other cytokines and chemokines are induced by intact Salmonella (7, 8, 58) and may also contribute to the reduced infiltrate and enhanced translocation of aflagellar bacteria from the gut. Flagellin-induced IL-1β upregulation would characterize the response of many cell types of immune or stromal origin. TLR5 is basolaterally orientated on enterocytes, but these cells can respond to serovar Typhimurium-induced translocation of flagellin (20) or as the result of subepithelial bacterial invasion. Hence, enterocyte-induced cytokine or chemokine response may represent a first line of defense against flagellated Salmonella. The delayed IL-6 response is probably a downstream event associated with differential immune cell recruitment due to the induced IL-1β response, which may also be TLR5 dependent. Indeed, the consequences of exposure to TLR agonists are dependent upon cell type even when the cells express a similar array of TLRs (25). A clear consequence of reduced proinflammatory cytokines in the CT was reduced PMN infiltration, an important factor in defining the chicken-Salmonella relationship (30). These observations support the hypothesis that the early phases of avian salmonellosis are defined, at least in part, by flagellin-TLR5 interactions.
ACKNOWLEDGMENTS
This work was funded by the BBSRC grant numbers 201/S15839 and 8/BFP11365 and BBSRC studentship 02/A1/S/08451.
We also acknowledge the assistance provided by the staff of the production units and experimental animal facilities and the cell culture core facility of the IAH.
FOOTNOTES
- Received 8 October 2004.
- Returned for modification 9 November 2004.
- Accepted 10 December 2004.
- Copyright © 2005 American Society for Microbiology