Investigating the Campylobacter jejuni Transcriptional Response to Host Intestinal Extracts Reveals the Involvement of a Widely Conserved Iron Uptake System

ABSTRACT Campylobacter jejuni is a pathogenic bacterium that causes gastroenteritis in humans yet is a widespread commensal in wild and domestic animals, particularly poultry. Using RNA sequencing, we assessed C. jejuni transcriptional responses to medium supplemented with human fecal versus chicken cecal extracts and in extract-supplemented medium versus medium alone. C. jejuni exposed to extracts had altered expression of 40 genes related to iron uptake, metabolism, chemotaxis, energy production, and osmotic stress response. In human fecal versus chicken cecal extracts, C. jejuni displayed higher expression of genes involved in respiration (fdhTU) and in known or putative iron uptake systems (cfbpA, ceuB, chuC, and CJJ81176_1649–1655 [here designated 1649–1655]). The 1649–1655 genes and downstream overlapping gene 1656 were investigated further. Uncharacterized homologues of this system were identified in 33 diverse bacterial species representing 6 different phyla, 21 of which are associated with human disease. The 1649 and 1650 (p19) genes encode an iron transporter and a periplasmic iron binding protein, respectively; however, the role of the downstream 1651–1656 genes was unknown. A Δ1651–1656 deletion strain had an iron-sensitive phenotype, consistent with a previously characterized Δp19 mutant, and showed reduced growth in acidic medium, increased sensitivity to streptomycin, and higher resistance to H2O2 stress. In iron-restricted medium, the 1651–1656 and p19 genes were required for optimal growth when using human fecal extracts as an iron source. Collectively, this implicates a function for the 1649–1656 gene cluster in C. jejuni iron scavenging and stress survival in the human intestinal environment.

human fecal extracts from 9 volunteers (human pools 1, 2, and 3 [HP1, HP2, and HP3]) were prepared by dilution of cecal/fecal material in sterile H 2 O and sterilization by filtration (Fig. 1A). Wild-type C. jejuni strain 81-176 was grown in Mueller-Hinton medium (MH) alone or MH supplemented with 30% CP, HP1, HP2, or HP3 to investigate the effects of extract exposure on in vitro growth and biofilm formation ( Fig. 1B and C). The 30% supplement condition was selected for these analyses based on limited extract volume and C. jejuni sensitivity to high osmotic stress (26). C. jejuni exponential growth rates were comparable under each condition for the first 12 h after inoculation, with doubling times ranging between 1.9 and 2.2 h (Fig. 1B). At 24 h, C. jejuni cells grown in medium alone showed reduced viability, as expected under this growth condition, while C. jejuni cells incubated in medium containing either chicken cecal or human fecal extracts maintained significantly higher viability (Fig. 1B). Growth in extracts did not impact C. jejuni cell morphology (data not shown). Biofilm formation was assessed for C. jejuni exposed to medium alone, and to medium containing 10% chicken cecal or human fecal extracts (Fig. 1C). A concentration of 10% extract was selected to conserve extract, as preliminary testing showed no difference between levels of C. jejuni biofilm formation in medium containing 10% or 30% extracts (data not shown). After 12 h, biofilm formation in medium containing extracts was comparable to or slightly higher than that in medium alone. However, at 24 and 36 h, C. jejuni cells grown in medium containing extracts exhibited lower biofilm abundance than cells grown in medium alone. One exception was C. jejuni cells grown in medium containing 10% CP at 36 h, which did not show a significant difference. The reduced biofilm formation in medium containing extracts was not caused by poor growth, since planktonic cell concentra- (B) C. jejuni growth in MH, MH plus 30% CP, and MH plus 30% HP1, HP2, and HP3 as measured by dilution plating. (C) C. jejuni biofilm formation over time with and without 10% extract as measured by the crystal violet assay (top) compared to the viable planktonic cell count (bottom). Error bars represent the standard deviation from 3 replicates. Statistical analysis for growth and biofilm formation was performed using the Student's t test with Welch's correction and compares MH plus extract conditions (CP, pink; HP1, light blue; HP2, medium blue; HP3, dark blue) to the MH control (black) for each time point. *, P Ͻ 0.05; **, P Ͻ 0.005; ***, P Ͻ 0.0005; ****, P Ͻ 0.0001. tions were comparable or higher under these conditions, similar to observations in shaking broth noted above.
C. jejuni exposed to host intestinal extracts responded by altering the expression of metabolism and iron uptake-related genes. We used RNA sequencing to assess the transcriptional response of C. jejuni upon exposure to extracts. C. jejuni cells were incubated in medium alone, medium plus 30% CP, and medium plus 30% HP1, HP2, or HP3. RNA was collected after 20 min to assess transient differences in gene expression and after 5 h to assess adaptive homeostatic differences. Sample data from the 20 different conditions were grouped as outlined in Table S1 in the supplemental material, and two different evaluation conditions were analyzed in detail: (i) exposure to medium containing extract versus medium alone (referred to as extract versus medium-discussed in this section) and (ii) exposure to human fecal extracts versus chicken cecal extract (referred to as human versus chicken-discussed in the next section).
C. jejuni cells exposed to medium containing either human or chicken extracts versus medium alone revealed 23 genes with Ͼ2-fold higher expression (Table 1) and 17 genes with Ͼ2-fold lower expression (Table 2). Of the genes showing higher expression after exposure to extracts, 13 were related to metabolism, nutrient uptake, and energy production (aspA, dcuA, dcuB, 0438, 0439, 0884, 0885, 1389 -1391, metC, purB-2, and 1570) (27)(28)(29), and 3 encoded hypothetical proteins with no annotated function (0204, 0440, and 1005). Also demonstrating higher expression was a 7-gene cluster (1649 -1655) that includes a homologue of an iron transporter in Escherichia coli (1649) (30), and p19 (1650), encoding a periplasmic iron binding protein involved in iron transport (25,31) (Table 1). Genes 0438 and 0439 showed the largest increase in expression both at 20 min (4.5-and 4.4-fold, respectively) and 5 h (11.8-and 12.1-fold, respectively). These genes (originally annotated by NCTC11168 locus tags cj0414 and cj0415) are required for gluconate dehydrogenase activity, allowing C. jejuni to use gluconate as an electron donor (32). Of the 17 genes showing reduced expression after incubation in medium supplemented with extracts in comparison to medium alone, one was a cell surface adhesin (peb3), 9 were involved in metabolism, nutrient uptake, and energy production (gltB, gltD, 0580, 0581, 0685, 0912, 0941, 0942, and 1386), 2 were involved in oxidative stress (herA and 1656), and 4 encoded hypothetical proteins with no annotated function (1006, 1184, 1185, and 1657) ( Table 2). The gene exhibiting the greatest decrease during incubation in extracts was peb3 at both 20 min (3.0-fold) and 5 h (6.2-fold). It encodes a cell surface glycoprotein adhesin that also imports phosphorylated compounds (33). PEB3 is a major antigenic protein during human infection, so rapid downregulation of this gene may be advantageous to host colonization (34,35). A list of differentially expressed genes in C. jejuni cultured in medium with either human or chicken extracts in comparison to medium alone at each sampling time is in Table S2 in the supplemental material, and a Venn diagram of these genes is presented in Fig. S1 in the supplemental material. C. jejuni exposed to human fecal extracts versus chicken cecal extract had higher expression of iron uptake and formate dehydrogenase genes. As noted above, we also compared C. jejuni gene expression during growth in medium supplemented with human fecal extracts to expression during growth in medium supplemented with chicken cecal extract. This identified 2 genes with Ͼ2-fold higher expression after 20 min and 12 genes with higher expression after 5 h (Table 3). No genes showing reduced expression were identified. The 2 genes with higher expression in human extracts after 20 min of exposure (fdhT, 2.4-fold; fdhU, 2.9-fold) increased further at 5 h (6.4-and 3.3-fold higher, respectively). FdhTU is involved in formate dehydrogenase activity and contributes to the invasion and intracellular survival of C. jejuni in intestinal epithelial cells (36,37). The remaining 10 genes with 2.5-to 3.4-fold higher expression after 5 h are known or hypothesized to be involved in iron uptake and/or utilization ( Table 3). The cfbpA, ceuB, and chuC genes encode parts of three different C. jejuni iron uptake systems: cfbpA encodes the periplasmic iron binding protein for the ferri-transferrin uptake system (38), ceuB encodes the periplasmic permease for the ferri-enterochelin uptake system (39), and chuC encodes part of the ABC transporter system for the heme uptake system (40,41). None of the other components in these iron uptake systems showed higher expression in human fecal extracts versus chicken cecal extract (41). The remaining 7 genes were 1649 -1655, which were also more highly expressed in C. jejuni cells exposed to extracts in comparison to those exposed to medium alone and are described in more detail in the following sections. Human fecal extracts contain higher total iron content than chicken cecal extract. Since the majority of the genes more highly expressed during growth in human fecal extracts were related to iron uptake, we hypothesized that cells grown in human fecal extracts were more iron starved than cells growing in chicken cecal extract. Quantification of the total iron present in extracts in all its forms was carried out using inductively coupled plasma mass spectrometry (ICP-MS). Unexpectedly, human extracts were found to contain approximately four times more iron than the chicken cecal extract (Fig. 2), suggesting the iron present is tightly chelated and requires specialized uptake systems.
Homologues of 1649 -1655 are widely conserved in the bacterial kingdom. The increased expression of the 1649 -1655 genes suggested that they are important in the C. jejuni response to extracts, especially human fecal extracts. The Escherichia coli homologue (FetM) of 1649 is a ferrous iron permease (30), and the periplasmic protein 1650 (P19) has been characterized as an iron transporter in C. jejuni (25), E. coli (30),  Bordetella (42), and Yersinia pestis (43). However, no prevalence or functional studies have been reported on the downstream genes (1651-1656) or homologues. As determined by conserved domain analyses and genome annotation, 1651-1655 were predicted to consecutively encode an inner membrane protein (1651), two inner membrane permeases (1652 and 1653), a cytoplasmic ATPase (1654), and a periplasmic thioredoxin (1655) (Fig. 3A). Also, one downstream gene (1656) overlaps with 1655 by 35 bp, encodes a second putative periplasmic thioredoxin (1656), and is assumed to be part of the same gene cluster (Fig. 3A). This set of 8 genes (1649 -1656) is carried directly downstream of a Fur box and has one putative primary transcription start site located 54 bases upstream of the 1649 start codon, suggesting that these genes are transcribed as part of an operon (Fig. 3A) (44). The gene upstream of 1649 (1648, encoding a hypothetical protein) ends 350 bp before the start of 1649, and the gene downstream of 1656 (1657, also encoding a hypothetical protein) is carried on the opposite strand. Each of these flanking genes has its own primary transcription start sites and is thus unlikely part of the proposed 1649 -1656 operon (44).
The organization of the operon and predicted gene products in each of the 33 bacterial species were examined. All 33 species carried homologues of the first six genes (1649 -1654) with the same gene order, similar gene lengths, and comparable spacing between genes (Fig. 3B). While most members of the Proteobacteria carried a predicted thioredoxin gene as the seventh gene in the cluster, members of the Actinobacteria, Firmicutes, Spirochaetes, and Synergistetes encoded a predicted flavin mononucleotide (FMN) binding protein. Although the thioredoxins may not be directly involved in iron uptake, both thioredoxins and FMN binding proteins can serve as a source of periplasmic or extracytoplasmic reduction potential to reduce ferric iron to the ferrous form for transport into the cell. In Proteobacteria without a thioredoxin encoded by the eighth gene, a gene coding for either a cytochrome or a hypothetical protein was present. The high genetic conservation of this system in multiple bacterial phyla suggests that, minimally, the first 7 gene products form one functional system, which is further investigated in the following sections.
The 1651-1656 genes are involved in iron acquisition. To determine if the 1651-1656 genes were involved in iron acquisition similar to 1649 and p19, deletion (⌬1651-1656) and complemented (1651-1656 C ) C. jejuni strains were constructed (Fig. 4). The p19 (1650) deletion mutant (⌬p19) and complemented (p19 C ) strains were used as controls since P19 is well conserved and is known to be involved in iron transport (25). The ⌬p19 and ⌬1651-1656 mutants each exhibited lower growth rates than the wild-type 81-176 strain or the corresponding complements during growth in MH medium, but reached comparable final viable cell concentrations at 32 h (Fig. 4D).
Iron depletion by addition of 20 M desferroxamine (DFO), an iron chelator, resulted in a similar initial growth rate for the ⌬1651-1656 mutant, a modestly reduced initial growth rate for the ⌬p19 mutant, and significantly reduced viability for both mutants after 12 h compared to the wild-type and complemented strains (Fig. 4E). Growth rates in MH for each mutant could be restored to wild-type and complement levels by supplementation with 100 M iron(III) citrate (Fig. 4F).
The 1649 -1656 iron uptake system is involved in the C. jejuni response to acid, streptomycin, and H 2 O 2 -mediated oxidative stress. C. jejuni wild-type strain 81-176, the ⌬p19 and ⌬1651-1656 mutant strains, and the respective complements were exposed to acid (pH 5), antibiotic, and oxidative stress (1 mM H 2 O 2 ) conditions to determine whether this iron uptake system is important for C. jejuni stress survival (Fig. 5). To test acid tolerance, cells were inoculated into MH at neutral pH or MH at pH 5 with and without supplementation with 100 M iron(III) citrate and grown for 24 h. The ⌬p19 and ⌬1651-1656 mutants exhibited a 2-to 3-log reduction in growth at pH 5 in comparison to neutral pH, while growth of the wild-type and complement strains was unaffected by the pH 5 condition (Fig. 5A). The acid sensitivity phenotype of the mutants was observed in media both with and without 100 M iron(III) citrate supplementation; however, iron supplementation partially restored growth of the mutants (Fig. 5A). Iron is more soluble at lower pH, which may contribute to the improved growth of mutants in acidic medium supplemented with iron(III) citrate.
To assess whether this iron uptake system was required for antibiotic resistance, the MICs of different antibiotics representing multiple classes (aminoglycosides, amphenicols, ␤-lactams, cationic peptides, fluoroquinolones, and macrolides) were screened for wild-type and ⌬1651-1656 deletion mutant strains using standard doubling dilution assays. Most classes showed Յ2-fold differences from wild type, except aminoglycosides, the most notable of which was streptomycin (STM) (data not shown). To test STM susceptibility in more depth, cells were inoculated into MH or MH supplemented with 100 M iron(III) citrate containing no STM or doubling dilutions of STM from 0.02 to 16 g/ml and grown for 48 h. In iron-unsupplemented MH, the MICs observed for the ⌬1651-1656 (2 g/ml) and ⌬p19 (2 g/ml) mutants were 8 times lower than the STM MIC for the wild-type and complemented strains (16 g/ml). Supplementation of MH with excess iron restored the MICs of the ⌬1651-1656 and ⌬p19 mutants (8 and 16 g/ml, respectively) to wild-type levels (16 g/ml). To expand our analysis of STM sensitivity, cell densities (optical density at 600 nm [OD 600 ]) were measured under increasing concentrations of STM both with and without iron supplementation ( Fig. 5B and C). A bimodal growth phenotype was observed for wild-type and complemented strains both with and without iron supplementation ( Fig. 5B and C). This bimodal phenotype consisted of STM-sensitive growth between 0 and 1 g/ml, where cell density decreased rapidly from 100% to Ͻ40% compared to growth with no STM, and then STM-tolerated growth between 1 and 8 g/ml. Unlike the wild-type and complemented strains, the ⌬1651-1656 and ⌬p19 mutants exhibited monomodal growth C. jejuni, Host Gut Extracts, and Iron ® under increasing STM concentrations in iron-limited MH, with both strains becoming undetectable at 2 g/ml STM (Fig. 5B). Iron supplementation restored bimodal growth for both mutants (Fig. 5C).
To examine if this operon had an effect on oxidative stress, wild-type strain 81-176, the ⌬p19 and ⌬1651-1656 mutants, and their respective complements were grown in medium alone or medium supplemented with 100 M iron(III) citrate for 6 h, at which time H 2 O 2 was added to a final concentration of 1 mM. Cell viability was monitored at 3 h and 6 h after H 2 O 2 addition. The mutants were more resistant to H 2 O 2 stress than the wild-type or complemented strains when grown in MH and in iron-supplemented MH, remaining viable at 3 h and 6 h post-H 2 O 2 addition, whereas the wild-type and complemented strains were no longer recoverable by 3 h (Fig. 5D and E).
The 1649 -1656 system is required for optimal iron acquisition from human and chicken extracts. To determine whether the 1649 -1656 system is involved in acquisition of iron found in extracts, we grew the wild-type and mutant strains in iron-limited MH (15 M DFO) or iron-limited MH supplemented with 10% chicken or human extracts. Growth in MH and iron-limited MH supplemented with 10 M iron(III) citrate was included as a control. The DFO concentration of 15 M was selected to allow optimal C. jejuni responsiveness to iron supplementation based on optimization testing, and the iron(III) citrate concentration of 10 M was selected to approximate the maximum amount of iron expected in the extract samples. After 24 h, wild-type and complemented C. jejuni strains showed reduced growth under iron restriction in comparison to the MH control, which was rescued by supplementation with human extracts or iron(III) citrate ( Fig. 6A; see Fig. S2A to F in the supplemental material). Notably, supplementation with chicken cecal extract was not sufficient to restore wild-type growth at 24 h, but did improve wild-type growth at 48 h in comparison to iron-limited medium alone ( Fig. 6A; Fig. S2C and F). These results suggest that wild-type C. jejuni can utilize iron present in the extracts and that more accessible iron is present in human fecal than in chicken cecal extracts. The ⌬p19 and ⌬1651-1656 deletion strains were also deficient for growth in iron-restricted MH. However, supplementation with human fecal extracts did not restore bacterial growth to levels observed in MH alone or iron-limited MH supplemented with iron(III) citrate ( Fig. 6B; Fig. S2A to F). This demonstrated that the 1649 -1656 system is required for optimal acquisition of iron from human fecal extracts.

DISCUSSION
The ability of C. jejuni to colonize both chicken ceca and human large intestines to high concentrations (7,8,12) is reflected in this study, where supplementation of MH with extracts prolonged C. jejuni survival in broth culture. The lower biofilm formation in extracts, which may provide additional nutrients, is consistent with studies indicating that C. jejuni biofilm formation is less abundant in richer media (62). We collected intestinal extracts using water as a diluent, which preferentially extracts polar compounds such as amino acids, carbohydrates, alcohols, and short-chain fatty acids (63). Accordingly, C. jejuni cells exposed to these extracts responded by modulating expression of many genes related to metabolism, nutrient uptake, energy production, and chemotaxis. Other diluents used to prepare fecal extracts, such as methanol, acetonitrile, and ethyl acetate, preferentially extract aromatic compounds and lipids (16,19,63,64). Future assessments may benefit from using different diluents to determine whether additional biologically active metabolites also impact pathogen response.
In this study, C. jejuni exposed to extracts, and in particular human fecal extracts, showed higher expression of genes related to iron uptake and utilization than C. jejuni in medium alone. A recent study of C. jejuni gene expression in human feces postcolonization also found that genes involved in iron acquisition were more highly expressed: notably, 6 out of 8 genes in the 1649 -1656 operon were identified (65). Iron is an essential micronutrient for most organisms, with many producing high-affinity iron chelating proteins and siderophores to scavenge iron. C. jejuni does not produce siderophores, but does possess at least 5 distinct systems to import iron from exogenous chelators (41). These systems transport iron bound to enterobactin (CfrA, CfrB, CeuBCDE [66,67]), heme (ChuABCDZ [40]), lactoferrin/transferrin (CtuA, CfbpABC, and ChaN [68]), ferrous ions (FeoB [69]), and potentially bound to rhodotorulic acid (Cj1658Ϫ1663 in strain 11168, which are homologues of 1649 -1654 in strain 81-176 [31]). C. jejuni globally regulates expression of the majority of its iron uptake systems through the iron-responsive Fur regulator (67,70,71). The specific increase in 1649 -1655 expression during exposure to human extracts and during human colonization, rather than a global response to low iron availability, suggests that other regulatory mechanisms target this system in addition to the Fur-based response. Furthermore, 1656, which shares an overlapping reading frame with 1655, actually showed a small decrease in expression during C. jejuni exposure to extracts, which may suggest even more complex regulation mechanisms at play.
It remains to be determined why the 1649 -1655 operon was more highly expressed during C. jejuni exposure to extracts, especially human extracts, which contain more iron than the chicken cecal extracts. One hypothesis is that the iron present in the extracts, particularly human fecal extracts, is chelated to a compound that is specifically recognized by this iron uptake system. Indeed, in medium depleted of iron, supplementation with human extracts was able to fully restore growth of wild-type C. jejuni but not of the ⌬p19 or ⌬1651-1656 mutant. The 1649 -1656 system was suggested to import iron bound to rhodotorulic acid, a fungal hydroxamate siderophore (reported as unpublished data in reference 31). However, other studies show that C. jejuni is unable to utilize iron bound to rhodotorulic acid for growth (72,73); therefore, other chelators must be considered.
In Yersinia pestis, mutations in fetMP (homologues of C. jejuni 1649 and p19) resulted in iron-deficient growth that could only be restored when the downstream genes y2367 to y2362 (homologues of C. jejuni 81-176 genes 1651-1655) were included in the complementation strain (43). Here, genes downstream of p19 (1651-1656) were also required for C. jejuni growth under iron-restricted conditions. Furthermore, the widespread distribution and genetic conservation of the 1649 -1656 cluster in bacteria with a variety of host preferences (e.g., humans, plants, cows, and bees) and ecologic niches (e.g., freshwater, seawater, soil, and thermal site) suggests that this entire system is used for iron uptake in many contexts. An analysis of the diversity and phylogenetic evolution of this iron uptake system, together with follow-up genetic and biochemical analyses, will allow better understanding of its role in bacteria.
Iron uptake in C. jejuni may be generally related to acid tolerance and antibiotic survival. Transcriptomic and proteomic studies have shown that this iron uptake system, in addition to several other iron uptake genes (cfbpA/B, ceuE, and chuZ), was more highly expressed upon C. jejuni exposure to acid stress (74,75). In addition, deletion of fur increased C. jejuni sensitivity to acidic conditions (76). Paradoxically, however, C. jejuni acid sensitivity caused by deletion of p19, 1651-1656, and fur was independent of iron availability in the growth medium, further indicating that additional regulatory mechanisms must be considered. Variability in iron availability and disturbance of bacterial iron homeostasis have also been shown to impact antibiotic efficacy in various bacteria (77). However, since the C. jejuni 81-176 strain has not been reported to carry streptomycin resistance genes, the mechanism of streptomycin tolerance and potential contribution of iron to this extended tolerance are unknown. The interplay between acid stress and iron homeostasis remains an interesting avenue for future research. Additional information pertaining to how C. jejuni responds to these stresses and the implications for resistance to streptomycin can be gained from genomewide expression analyses of C. jejuni under a combination of acid and iron stresses.
Unexpectedly, the 1649 -1656 system adversely affected the C. jejuni response to H 2 O 2 independent of the addition of iron. C. jejuni responds to oxidative stress and iron limitation by inducing expression of a common subset of genes (e.g., katA, sodB, and ahpC) that are controlled by both the Fur and PerR regulons (78,79). Thus, loss of the 1649 -1656 iron uptake system may cause constitutive cellular stress that elevated the baseline expression of genes contributing to oxidative stress tolerance. Alternatively, the presence of iron and copper binding proteins in the 1649 -1656 system may directly render C. jejuni more sensitive to H 2 O 2 -mediated cell death by formation of highly reactive and damaging intermediates upon exposure to H 2 O 2 (25,80). Exploring whether related conditions such as nitrosative stress yield similar results would be interesting, as would examining the effects of combining the ⌬1651-1656 or ⌬p19 mutations with mutations in genes known to be required for oxidative stress survival.
An updated model of the 1649 -1656 iron uptake system is proposed based on bioinformatic analyses, showing the presence of signal peptides, transmembrane regions, and protein domains (Fig. 7). The 1649 -1656 gene products are hypothesized to form one functional system based on comparable phenotypes observed between the ⌬1651-1656 and ⌬p19 mutants and conservation of the gene cluster in multiple bacterial phyla. Of note, the 1651 membrane protein has a periplasmic DUF2318 domain, encoding a conserved CxxC-x(13)-CxxC-x(14,15)-C motif (see Fig. S3A and B in the supplemental material). Conserved cysteines with regular spacing are often observed in iron-sulfur cluster binding domains, such as those seen in the Fer2 and Fer4 families (Fig. S3C and D). Furthermore, we found that genes encoding DUF2318-containing proteins almost always mapped to gene clusters with homologues of 1649 -1656 in different bacterial species, indicating association with this bacterial iron uptake system. Ongoing experiments aimed toward single gene deletions and domain or residue specific substitutions should help determine the identity of transported substrates, protein-protein interactions, and mechanism(s) of iron transport.
In sum, this study is the first to compare the bacterial transcriptional and functional responses to gut extracts from a disease-susceptible host (humans) and a zoonotic host (chicken). RNA sequencing identified genes that may be important for C. jejuni to colonize both chickens and humans, with subsequent biological analyses demonstrating that the 1649 -1656 iron uptake system was necessary for C. jejuni growth under iron-depleted conditions, impacted growth and survival under specific stress-related conditions, and was required for acquisition of iron from human fecal extracts. Given the system's conservation in a wide range of bacteria, the majority of which colonize and infect humans, this work should be broadly relevant to the general bacteriology and pathogenesis communities as well. Collectively, we showed that investigation of pathogen gene expression in response to exposure to the host metabolome, together with follow-up genotypic and phenotypic analyses, is not only an informative method of studying host-pathogen interactions but should improve our understanding of the mechanisms employed by pathogens to cause disease.

MATERIALS AND METHODS
Ethics statement. Written and informed consent was obtained from all human fecal sample donors as described in ethics application H14-00859, which was approved by the University of British Columbia Clinical Research Ethics Board in Vancouver, Canada.
Bacterial strains and growth conditions. C. jejuni strain 81-176, a pathogenic and common laboratory reference strain, was used in this study (9). The C. jejuni ⌬1651-1656 mutant strain and corresponding 1651-1656 C complemented strain were constructed as described in Text S1 in the supplemental material, using primers listed in Table S4 in the supplemental material. The ⌬p19 deletion and complemented (p19 C ) strains were obtained from reference 25. C. jejuni strains were grown in MH (Oxoid) broth or agar (1.5% [wt/vol]). MH was supplemented with antibiotics where appropriate: vancomycin (V; 10 g/ml), trimethoprim (T; 5 g/ml), kanamycin (Km; 50 g/ml), and chloramphenicol (Cm; 20 g/ml). Agar plates and standing cultures were incubated at 38°C under microaerobic and capnophilic conditions (12% CO 2 and 6% O 2 in N 2 ) in a Sanyo tri-gas incubator. Shaking broth cultures (including 15-to 18-h overnight cultures) were incubated microaerobically in airtight containers with the Oxoid CampyGen atmosphere generation system at 38°C and shaken at 200 rpm. Cell concentration was measured by dilution plating; the limit of detection for this method is 10 3 CFU/ml. Escherichia coli strain DH5␣ (Invitrogen) was used for cloning. E. coli was grown aerobically at 37°C and shaken at 200 rpm in Luria-Bertani broth (LB; Sigma) or incubated aerobically at 37°C on LB agar (1.5% [wt/vol]) supplemented with Cm (20 g/ml) and Km (50 g/ml) for selection, as required.

FIG 7
Model of the C. jejuni 1649 -1656 iron uptake system. We hypothesize that this iron acquisition system imports iron (yellow stars) that is bound to a specific, but as yet unidentified, chelator (light blue hexagons). Gray hexagons represent other iron chelators. Since the gene cluster does not appear to encode an outer membrane receptor, Chan et al. suggested that iron is transported across the outer membrane by an as yet unidentified receptor (gray box) before binding to P19 (dark blue) in the periplasm (25). During iron uptake, iron-bound P19 may interact with the periplasmic metal binding domains of 1651 (brown), followed by an iron reduction step involving the 1655 and 1656 thioredoxins (light and dark green), prior to being transported into the cytoplasm via the 1649 inner membrane transporter (black). The substrates transported by the 1651 membrane protein and the 1652 and 1653 permeases (pink and red), with the help of the 1654 ATPase (purple), are unknown. We hypothesize that the chelated form of iron acquired by this system is more abundant in human than in chicken intestinal environments.

ACKNOWLEDGMENTS
We acknowledge and thank all the members of the Gaynor lab and Charles Thompson for discussions and technical support, especially Emilisa Frirdich and Jenny Vermeulen for critical reviews and Reuben Ha for support with sample processing. We thank Neil Ambrose (Sunrise Farms) for providing us with chicken gut sacks and all volunteers for their donation of samples and time. We thank the core sequencing teams at the Sanger Institute for performing the RNA sequencing, and Mariko Ikehata for processing the ICP-MS samples.
This work was funded by Canadian Institutes of Health Research (CIHR) grants MOP-68981 to E.C.G. and MOP-49597 to M.E.P.M., CIHR grant GSM-136327 to M.L., Canada Graduate Scholarship-Michael Smith Foreign Study Supplement 2014-338587 to M.L, and Medical Research Council grant G1100100/1 to J.P. Sequencing was supported by Wellcome Trust grant WT098051 to J.P. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. The authors declare that they have no conflicts of interest with the contents of this article.