|Home | About | Journals | Submit | Contact Us | Français|
Helicobacter pylori persistently colonizes the harsh and dynamic environment of the stomach in over one-half of the world's population and has been identified as a causal agent in a spectrum of pathologies that range from gastritis to invasive adenocarcinoma. The ferric uptake regulator (Fur) is one of the few regulatory proteins that has been identified in H. pylori. Fur regulates genes important for acid acclimation and oxidative stress and has been shown to be important for colonization of H. pylori in both murine and Mongolian gerbil models of infection. To more thoroughly define the role of Fur in vivo, we conducted an extensive temporal analysis of the location of, competitive ability of, and resultant pathology induced by a Δfur strain in the Mongolian gerbil model of infection and compared the results to results for its wild-type parent. We found that at the earliest time points postinfection, significantly more Δfur bacteria than wild-type bacteria were recovered. However, this trend was reversed by day 3, when there was significantly increased recovery of the wild-type strain. The increased recovery of the Δfur strain at 1 day postinfection reflected increased recovery from both the corpus and the antrum of the stomach. When the wild-type strain was allowed to colonize first, the Δfur strain was unable to compete for colonization at any time postinfection. However, when the Δfur strain was allowed to colonize first, the wild type efficiently outcompeted the Δfur strain only at early times postinfection. Finally, we demonstrated that there was a delay in the development and severity of inflammation and pathology of the Δfur strain in the gastric mucosa even after comparable levels of colonization occurred. Together, these data indicate that H. pylori Fur is most important at early stages of infection and illustrate the importance of the ability of H. pylori to adapt to its constantly fluctuating environment when it is establishing infection, inflammation, and disease.
Helicobacter pylori persistently colonizes the gastric mucosa of over one-half of the world's population (10) and has been classified as a class I carcinogen by the World Health Organization (2) due to its close association with the development of gastric adenocarcinoma and mucosa-associated lymphoid tissue (MALT) lymphoma. Fortunately, the majority of individuals infected with H. pylori manifest only subclinical gastritis. However, H. pylori has also been associated with gastric ulcers and duodenal ulcers in addition to the cancers mentioned above (16). Thus, H. pylori infection is the major risk factor for development of a broad spectrum of gastric diseases (24).
Given the harsh environment of the stomach, it seems remarkable that H. pylori is able to persistently colonize the gastric mucosa. In this dynamic gastric niche, H. pylori encounters fluctuations in pH (45) and in the availability of iron (8) and other nutrients (5). Thus, survival of H. pylori in the face of this tumultuous environment undoubtedly requires adaptive mechanisms that allow the bacterium to alter gene expression in response to changing environmental cues. Despite this requirement, H. pylori encodes a paucity of transcriptional regulators and two-component systems (64). Among the regulatory factors that have been identified, the ferric uptake regulator (Fur) has been shown to be necessary for adaptation to iron limitation (69), low pH (9), and oxidative stress (19, 26). In light of the small number of regulators, it is perhaps not surprising that Fur has such diverse roles in this pathogen.
Classically, Fur functions as a transcriptional repressor such that when iron is available, Fur binds to its ferrous cofactor and represses gene transcription. This classical regulation is used to control the expression of multiple genes in H. pylori, including amiE and frpB1, which have been implicated in acid tolerance and iron uptake, respectively (18, 68). Uniquely, H. pylori Fur is also capable of apo regulation, and in the absence of iron, apo-Fur can repress its target genes. sodB, encoding a superoxide dismutase, and pfr, encoding an iron storage molecule, have been shown to be regulated in this manner (19, 26).
Fur has also been shown to play a role in H. pylori colonization, although it is not essential (14, 30). In mice, there was a 2-log difference between the number of fur mutant bacteria recovered at 1 month postinfection and the number of wild-type bacteria recovered (14). Conversely, in the Mongolian gerbil model, our group previously showed that an H. pylori Δfur mutant displayed a lag in colonization fitness compared to the wild type; there was a 50-fold decrease in the number of Δfur mutant bacteria that were recovered at 3 days postinfection, but the levels of colonization were comparable by 14 days postinfection. Additionally, we found that when a Δfur strain was used with a wild-type strain in competition assays, the mutant had a 100-fold early competitive defect that was observed throughout later stages of infection (30). Together, these data suggest that fur plays a role in the early stages of colonization in the Mongolian gerbil, which is a robust small animal model for studying the development of H. pylori-induced carcinoma (29, 71).
Since Fur is able to regulate an array of genes important for the H. pylori stress response and has an effect on colonization, we sought to better define the specific role of Fur in vivo during colonization of the Mongolian gerbil stomach. As suggested by previous data, we found that Fur is most important during early time points during infection. Examination of the distribution of wild-type and Δfur strains during colonization of the stomach showed that the Δfur strain indiscriminately colonizes both the corpus and the antrum but is rapidly cleared from the acid-producing corpus of the stomach. Moreover, the surviving Δfur bacteria exhibit a lag in the ability to grow in the antral region. Finally, we observed attenuation in the development and severity of the host pathology induced by the Δfur strain.
The gerbil-adapted H. pylori strain 7.13 (29), the isogenic fur mutant DSM143 (30), and the isogenic Δfur/fur+ complementation strain DSM804 were used in this study. As previously described, DSM143 has a deletion/insertion that is approximately 194 bases into the fur coding sequence. This results in insertion of a kanamycin (Kan) resistance cassette into the middle of the fur coding sequence, which results in a null phenotype (30). Construction of DSM804 is described below. All strains were maintained as frozen stocks at −80°C in brain heart infusion medium supplemented with 20% glycerol and 10% fetal bovine serum (FBS). Bacteria were grown on horse blood agar (HBA) plates that contained 4% Columbia agar, 5% defibrinated horse blood (HemoStat Laboratories, Dixon, CA), 0.2% β-cyclodextrin (Sigma), 10 μg/ml vancomycin, 5 μg/ml cefsulodin (Sigma), 2.5 U/ml polymyxin B (Sigma), 5 μg/ml trimethoprim (Sigma), and 8 μg/ml amphotericin B. H. pylori liquid cultures were grown in brucella broth supplemented with 10% fetal bovine serum and 10 μg/ml vancomycin with shaking at 100 rpm. Microaerobic conditions (5% O2, 10% CO2, 85% N2) were generated with an Anoxomat gas evacuation and replacement system (Spiral Biotech) in gas evacuation jars. All H. pylori plate and liquid cultures were grown under microaerobic conditions at 37°C. The H. pylori strains used for infection of animals were grown under the liquid culture conditions described above for approximately 18 h prior to use. Where appropriate, cultures and plates were supplemented with 8 μg/ml chloramphenicol (Cm) and/or 25 μg/ml kanamycin (Kan).
The fur promoter and coding sequence were PCR amplified with primers FurF1-BamHI (GGATCCAAGGCTCACTCTACCCTATT) and FurCR1-XbaISM (TCTAGAGCCCTATCTAAGCTTCTCC), which contain a BamHI site and an XbaI restriction site, respectively. The 836-bp fragment was digested with these enzymes and subsequently cloned into similarly digested pRDX-C (17), yielding pRDX-C::fur. The pRDX-C plasmid carries a chloramphenicol resistance cassette flanked by upstream and downstream segments of the rdxA locus, which allows integration of cloned fragments into this locus (17). The fur-containing fragment was cloned into the multicloning site upstream of the resistance cassette to create p756. This plasmid was naturally transformed into the 7.13 Δfur (DSM143) strain, and transformants were selected on plates containing 25 μg/ml Kan and 8 μg/ml Cm. Proper integration of the fur locus in the resulting 7.13 Δfur/fur+ strain (DSM804) was confirmed by sequencing.
To confirm expression of fur in DSM804, as well as functional complementation of Fur-dependent regulation, RNase protection assays (RPAs) were performed. Briefly, DSM804 and H. pylori wild-type 7.13 and DSM143 controls were grown for 18 h in liquid cultures exactly as described above. One half of each culture was removed and used for RNA extraction, while iron was depleted in the other half by addition of 200 μM 2,2′-dipyridyl (DPP), an iron chelator. After 1 h of chelation, the cells were harvested and used for RNA extraction. RNA was extracted as previously described (63). To examine expression of the fur transcript, riboprobe templates for fur were constructed using primers Hp fur RPA F (GAGCGCTTGAGGATGTCTATC) and Hp fur RPA R (GTGATCATGGTGTTCTTTAGC) (15). To measure iron-bound and apo-Fur regulation, riboprobe templates were also generated as previously described for amiE and pfr, respectively (30). The resulting fur, amiE, and pfr amplicons were ligated into pGEM-T Easy (Promega), and riboprobes were generated by using a Maxiscript kit (Applied Biosystems) and 50 μCi [32P]UTP (Perkin Elmer). Then 1.5 μg of total RNA was used to perform RPAs with an RPA III kit (Applied Biosystems) as previously described (15). The resulting gels were exposed to phosphor screens. The screens were scanned using an FLA-5100 scanner (Fujifilm), and the intensities of protected bands were quantified with Multi-Gauge software (version 3.0; Fujifilm).
To confirm expression of the Fur protein, bacterial lysates were prepared from the wild-type 7.13, DSM143, and DSM804 strains grown in liquid cultures as described above. Protein concentrations were determined using the bicinchoninic acid (BCA) protein assay (Thermo Scientific), and equal concentrations of the samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using an 18% separating gel. The separated proteins were transferred to nitrocellulose membranes using a semidry transfer apparatus (Owl; Thermo Scientific) and then probed with a 1:200 dilution of rabbit polyclonal anti-H. pylori Fur antibody, which was prepared using the rabbit Quick Draw protocol and was produced by Pocono Rabbit Farm and Laboratory (B. Carpenter, H. Gancz, S. Benoit, S. Evans, P. S. J. Michel, R. Maier, and D. S. Merrell, submitted for publication). This was followed by addition of a 1:20,000 dilution of horseradish peroxidase (HRP)-conjugated bovine anti-rabbit IgG secondary antibody (Santa Cruz Biotechnology). Proteins were detected using a SuperSignal West Pico chemiluminescent substrate kit (Thermo Scientific/Pierce) and a LAS-3000 intelligent dark box with LAS-3000 Lite.
Six- to 12-week-old male Mongolian gerbils (Charles River Laboratories International, Inc., Wilmington MA) were fasted for 12 h prior to infection. Animals were then orogastrically infected with approximately 109 H. pylori cells. Two independent biological repeats of each infection experiment were performed as follows. For single-strain infections, gerbils were infected with either the wild-type or Δfur strain and then sacrificed at 1, 3, 5, 8, 14, or 30 days postinfection (9 or 10 animals for each group at each time point) or at 2, 4, 8, or 16 weeks postinfection (12 to 14 animals per group at each time point). As a control for the 2-, 4-, 8-, and 16-week single-infection time course, we also included a mock-infected group (8 or 9 animals per time point). At each time point, the glandular portion of the stomach of each animal was bisected. One half of the tissue was weighed and homogenized in brucella broth with a mechanical homogenizer (Tissue Tearer; Biospec Products Inc.), and the number of viable CFU was determined by plating portions on HBA plates. The remaining half of the stomach was embedded in paraffin, sectioned, and stained with hematoxylin and eosin. Sections were then scored blindly by a pathologist for acute and chronic inflammation using a scale from 0 to 3 corresponding to normal, mild, moderate, and marked inflammation according to the updated Sydney system (20). Since the acute and chronic inflammation scores were similar, they were averaged to produce a single combined score, which is described below. The sections were also scored blindly for the development of dysplasia and invasive adenocarcinoma by two pathologists. Dysplasia and invasive adenocarcinoma were diagnosed using guidelines described previously (12). Dysplastic mucosa consisted of elongated or branched, irregular glands with enlarged, crowded cell nuclei that appeared to be stratified. Invasive adenocarcinoma was defined as tubular or irregular structures composed of dysplastic epithelium that penetrated through the muscularis mucosa, infiltrating the submucosa.
For coinfections, animals were infected with a 1:1 mixture of wild-type and Δfur bacteria or wild-type and Δfur/fur+ bacteria, using a total input of approximately 109 CFU. Animals were sacrificed at days 1 and 3 postinfection, the glandular portions of the stomachs were excised, weighed, and mechanically homogenized, and portions were plated on HBA plates and HBA plates supplemented with either kanamycin or kanamycin and chloramphenicol. The relative number of wild-type bacteria was determined by subtracting the number of kanamycin-resistant Δfur colonies or chloramphenicol- and kanamycin-resistant Δfur/fur+ colonies from the total number of colonies on HBA plates without antibiotics. Competitive indexes (CI) were calculated by dividing the number of Δfur or Δfur/fur+ bacteria by the number of wild-type bacteria and then correcting for the actual input ratio of the strains. Two independent biological repeats of the coinfection experiment were performed, and a total of 10 animals were examined for each day for each coinfection group.
Gerbils were infected with approximately 109 Δfur bacteria and were subsequently superinfected with approximately 109 wild-type bacteria at 1, 3, 7, 14, or 28 days after the initial infection. Three independent biological experiments were performed, and a total of 10 to 14 animals were examined for each time point. Additionally, we conducted a reciprocal study in which the animals were initially infected with wild-type bacteria and were subsequently superinfected with the Δfur strain (5 animals per time point). In each case, animals were sacrificed 2 weeks after the second H. pylori dose, the glandular stomach of each animal was excised, weighed, and mechanically homogenized, and portions were plated on HBA plates and HBA plates supplemented with kanamycin to determine the relative numbers of wild-type and Δfur bacteria.
To determine the gross distribution of bacteria in the stomach, animals were infected with approximately 109 wild-type or Δfur bacteria. Animals were sacrificed at 1, 3, 5, 8, 14, or 30 days postinfection, and the glandular stomachs were excised. The data presented below are data from two independent biological experiments performed with a total of 10 animals per group at each time point. Each stomach was splayed, and the corpus and the antrum were then divided based on gross anatomical features (40). Each section was then weighed and mechanically homogenized, and portions were plated on HBA to determine the number of CFU. To ascertain the distribution of H. pylori in the mucus layer and H. pylori more intimately attached to the stomach tissue, animals infected with approximately 109 wild-type or Δfur bacteria were sacrificed on days 1, 3, 5, 8, 14, and 30 postinfection. Then the glandular stomachs were excised, weighed, repeatedly flushed with 1 ml of a 2.5% solution of N-acetylcysteine (NAC), and then weighed again. NAC flushing resulted in extraction of the mucus layer since NAC is able to disrupt disulfide bonds of glycoproteins in mucus due to the presence of its sulfydryl group (74). Consistent with a previous report (34), the concentration of NAC used in these studies was subinhibitory for growth and survival of 7.13 and DSM143 (data not shown). The mucus was recovered and plated on HBA, and the stomach tissue was mechanically homogenized and plated on HBA to determine the numbers of CFU. The data presented below are data from two independent biological experiments performed with a total of 10 animals per group at each time point.
Microsoft Office Excel 2003 was used for log10 two-tailed Student t test analysis of colonization data, and two-tailed Student t tests were used for analysis of the inflammation data. Fisher's exact test was used for analysis of pathology data (GraphPad 2005).
Gancz et al. previously showed that in the Mongolian gerbil model at 3 days postinfection the colonization density of a Δfur H. pylori strain is lower than that of the wild-type 7.13 strain (30). However, by 2 weeks postinfection, there was no perceptible difference in colonization between the two strains (30). This suggests that Fur is important in the early stages of colonization but is not essential at later stages of infection. To expand this initial observation, we performed single-strain infection experiments with biological repeats using a larger number of animals and adding additional time points that were not examined in the initial study. We monitored colonization of the wild-type and Δfur strains at 1, 3, 5, 8, 14, and 30 days postinfection. Similar to the previous findings, starting at day 3 we observed a reproducible decrease in the amount of the Δfur strain recovered compared to the amount of the wild-type strain recovered (Fig. (Fig.1).1). The difference was statistically significant at 3 days (P = 0.0002) and 8 days (P < 0.0001) postinfection. Also similar to the previous findings (30), the difference gradually decreased over time until there was no difference between the colonization of the wild-type strain and the colonization of the Δfur strain starting at day 14 (Fig. (Fig.1).1). Since Fur appeared to have an effect on early stages of colonization, we examined the effect of Fur at a time point even earlier than the time points assessed previously. Unexpectedly, at day 1 postinfection we observed a statistically significant increase in the number of Δfur bacteria recovered compared to the number of wild-type bacteria recovered (P = 0.0007) (Fig. (Fig.11).
Gancz et al. previously showed that when gerbils were infected simultaneously with wild-type and Δfur strains, there was a 100-fold defect in recovery of the Δfur at strain at days 6 and 20 postinfection (30). Given the unexpected increase in the number of Δfur strain bacteria recovered at day 1 compared to the number of bacteria recovered at day 3, we wondered whether a similar competitive defect could be observed at earlier time points. Thus, to examine earlier time points during colonization, we coinfected animals with the wild-type and Δfur strains and monitored the colonization at days 1 and 3 postinfection. We observed that at day 1 the CI was 0.113, which represented a 10-fold defect in competition for the Δfur strain. Similar to previous findings, the CI at day 3 was 0.016, which represented a 100-fold defect. The 10-fold increase in the value for the competitive defect from day 1 to day 3 for the Δfur strain suggests that the dynamics of infection change during the initial stages of colonization. Additionally, despite the fact that more Δfur bacteria were recovered at day 1 in the single-strain colonization experiment, our coinfection assays suggested that the increased recovery did not imply that there was increased fitness of the Δfur strain at this time point.
Even though Fur appears to be encoded by a single gene and the mutation that we constructed was not expected to result in any polar effects, we wanted to ensure that the changes that we observed were due to fur deletion and not to a random secondary mutation. To accomplish this, a fur complementation strain (DSM804) was constructed as described in Materials and Methods. RPA and Western blot analysis showed that DSM804 expressed the fur transcript, which resulted in accumulation of the Fur protein (data not shown). Additionally, DSM804 showed wild-type Fur-dependent patterns of expression of amiE and pfr (data not shown). Since the coinfection assay is considered one of the most sensitive ways to assess colonization fitness, we assessed whether the fur mutation could be complemented in trans by coinfection of DSM804 and the wild-type strain. Again, we monitored the colonization at days 1 and 3 postinfection. At day 1, the CI was 1.402, indicating that the complemented strain had colonization ability similar to that of the wild-type strain. Moreover, the CI at day 3 was 0.206, which represented a 5-fold defect; this minor defect was considerably less than the 100-fold defect observed for the Δfur strain. Together, our data suggest that expression of fur in trans results in almost complete restoration of the wild-type phenotype and suggests that the colonization defects of the Δfur strain observed were due to loss of Fur and not due to an undefined secondary mutation.
Fur appears to be most crucial during the early stages of establishment of a persistent infection but to be unessential for maintenance of infection after normal levels of colonization have been achieved. We therefore reasoned that if the Δfur strain was allowed to infect animals first and the animals were subsequently superinfected with the wild-type strain at various times postinfection, we should observe temporal differences in the ability of the wild-type strain to displace the Δfur strain. Groups of animals were inoculated with Δfur bacteria and then superinfected with the wild-type strain at 1, 3, 7, 14, or 28 days after the initial infection. Two weeks after superinfection, the animals were sacrificed, and the number of cells of each colonizing strain was determined by plating; the data were expressed as a competitive index (CI) as described in Materials and Methods. As shown in Fig. Fig.22 A, the wild-type strain was able to displace the Δfur strain during the first week of infection; we observed a 100-fold defect in the number of Δfur bacteria recovered at days 1 and 3 and a 10-fold defect at day 7 (Fig. (Fig.2A).2A). However, the ability of wild-type bacteria to displace the Δfur strain progressively diminished over time; by day 28, no wild-type bacteria were recovered from 8 of the 14 gerbils examined. To ensure that this phenomenon was not simply due to colonization dynamics in the model, we also performed the converse experiment. In this experiment, wild-type bacteria were allowed to colonize the animals first, and then the animals were superinfected with the Δfur strain. We found that the Δfur strain was not able to displace the wild-type strain at any time point (Fig. (Fig.2B).2B). Together, our results indicate that Fur plays a role in establishing an infection but that once colonization has occurred, Fur is no longer crucial for maintenance of the infection.
H. pylori colonization is typically more sparse in the corpus than in the antrum (7, 32, 60, 61); however, alterations in this distribution have been associated with both host factors (3, 60) and bacterial factors (62). Given the increased recovery of the Δfur strain that we observed at 1 day postinfection (Fig. (Fig.1)1) and the fact that mutation of other H. pylori genes has been shown to alter localization of bacteria in the gastric niche (62), we wondered whether Fur was important for localization in the stomach. Initially, immunohistochemical staining of our paraffin-embedded sections was used in an attempt to quantitate differences in localization. However, the number of bacteria visible per field in our sections was very low, and no overall difference between the wild type and the Δfur strain was observed in this analysis (data not shown). We reasoned that this incongruence could have been due to the combination of the small number of H. pylori enumerated in each section and the loss of the mucosal layer during tissue preparation and sample processing. Therefore, to more specifically examine the distribution of H. pylori associated with the mucosal layer and the bacteria intimately adhering to the stomach mucosa, we separated the mucus from the tissue layer as described in Materials and Methods and determined the number of each colonizing strain in each fraction. Parallel to the global stomach colonization data (Fig. (Fig.1),1), at day 1 postinfection we observed increased numbers of Δfur bacteria in the mucus layer (P = 0.048) (Fig. (Fig.3A).3A). By day 3 there was a 10-fold increase in the number of wild-type bacteria in the mucus layer, but there was little overall change in the number of Δfur bacteria at this site. This resulted in comparable numbers of wild-type and Δfur bacteria at this time point. After this, the number of wild-type bacteria in the mucus layer continued to increase until the peak number was observed at day 8. Conversely, the data for the Δfur strain showed that there was a decrease in colonization at day 5, which was followed by a much slower increase in the number of bacteria recovered, which peaked at day 14 (Fig. (Fig.3A3A).
Analysis of the number of bacteria intimately adhering to the stomach tissue showed that while more of the Δfur bacteria were found at day 1, the number of wild-type bacteria adhering to the tissue had dramatically increased by day 3 (P = 0.003). After this, more wild-type bacteria were attached to the tissue (day 5, P = 0.045; day 14, P = 0.006) (Fig. (Fig.3B).3B). Overall, we observed rapid growth of the wild-type bacteria in both the mucosal and stomach tissue layers (Fig. 3A and B). Conversely, for the Δfur strain there was a slight decrease in total colonization from day 1 to day 3, and there was a lag before growth started (Fig. (Fig.11 and 3A and B). Together, these data suggest that the colonization pattern for the Δfur strain in the stomach is different and that Fur is important in initiating growth of H. pylori in the gastric niche.
Since Fur regulates a diverse repertoire of genes (25, 30) and because we observed differences in the mucosal and adherence distributions, we also wondered if there was a difference in the distribution of the wild-type and Δfur strains between the corpus and the antrum. Thus, animals were infected with wild-type or Δfur bacteria for 1 or 3 days, and the corpus and the antrum were separated and examined for colonization as described in Materials and Methods. As expected (6), at day 1 postinfection the majority of wild-type bacteria were localized to the antrum, and few bacteria were present in the corpus (Fig. (Fig.3C).3C). This was in stark contrast to the distribution of the Δfur strain at 1 day postinfection; we observed that the Δfur strain was distributed throughout the corpus and the antrum (104 bacteria in each section) (Fig. (Fig.3C).3C). By day 3 postinfection there was a dramatic increase in the number of wild-type bacteria in the antrum, and there was a smaller increase in the corpus (Fig. (Fig.3C).3C). Conversely, we observed clearance of the majority of the Δfur strain from the corpus and less clearance of this strain from the antrum (Fig. (Fig.3C).3C). Overall, these findings illustrate the importance of Fur regulation for the ability of H. pylori to properly localize to its antral niche, where it is able to adapt and colonize.
Since Fur plays an important role in regulation of a diverse set of genes and consequently affects colonization, we reasoned that it might also play a role in development of host inflammation and injury. In order to evaluate acute and chronic inflammation, coded sections harvested from animals infected for 2, 4, 8, or 16 weeks were stained with hematoxylin and eosin and blindly graded using the modified Sydney system as described in Materials and Methods (20). Since the acute and chronic inflammation scores were similar (data not shown), they were averaged to produce a single combined inflammation score for each section. The same sections were also blindly examined for dysplasia and carcinoma. For comparison, mock-infected animals were treated with brucella broth and maintained for similar periods of time. No inflammation (Fig. (Fig.44 A and B) or other lesions were observed in any of the control animals (data not shown). We found that even though the levels of colonization of the wild-type and Δfur strains were similar by 2 weeks postinfection (Fig. (Fig.1)1) and despite early abundant colonization of the corpus by the Δfur strain (Fig. (Fig.3C),3C), the levels of inflammation in the corpus of the wild-type strain-infected animals were statistically significantly greater than the levels of inflammation in the corpus of the animals infected with the Δfur strain at weeks 4 (P = 0.0114), 8 (P = 0.0018), and 16 (P = 0.0186); wild-type levels of inflammation in the corpus were never observed in the Δfur strain-infected animals (Fig. (Fig.4A).4A). Similarly, we observed a lag in the ability of the Δfur strain to produce wild-type levels of antral inflammation (Fig. (Fig.4B).4B). There was more severe inflammation in wild-type strain-infected animals than in Δfur strain-infected animals at week 4 (P = 0.0003). However, the levels of antral inflammation produced by the Δfur strain progressively increased to the wild-type levels, and the levels were indistinguishable from the wild-type levels by week 16 (Fig. (Fig.4B).4B). These observations suggest that while Fur is not required for development of inflammation, it plays a role in the progression and severity of inflammation.
Since the level of inflammation is considered a risk factor for development of cancer (54), we sought to determine whether there was a parallel delay in the development and severity of premalignant and malignant lesions in Δfur strain-infected animals. Therefore, tissue sections were graded for the appearance of low-grade dysplasia and invasive adenocarcinoma as described in Materials and Methods. As expected based on the inflammation scores (Fig. 4A and B), by 4 weeks postinfection there was more severe gastric injury (P = 0.0013) in the wild-type strain-infected animals than in the Δfur strain-infected animals (Fig. (Fig.55 A). At 4 weeks postinfection, mock-infected animals showed no pathology (Fig. (Fig.5B).5B). Of the wild-type infected animals, 50% developed gastritis, 33% developed low-grade dysplasia, and 17% developed invasive adenocarcinoma (Fig. (Fig.5D).5D). Conversely, 25% of the Δfur strain-infected animals developed gastritis (Fig. (Fig.5C)5C) and 8% developed low-grade dysplasia by 4 weeks postinfection. Statistical analysis revealed that there were increased rates of development of gastritis (P = 0.0001), dysplasia (P = 0.0005), and cancer (P = 0.011) in the wild-type strain-infected animals in the period from 2 to 4 weeks, whereas for the Δfur strain-infected animals there were not any significant differences in the development of pathologies during the same period. Indeed, it was not until 8 weeks postinfection that the Δfur strain-infected animals exhibited a significant difference in the development of carcinoma (P = 0.0046). Whereas 93% of wild-type strain-infected animals developed invasive carcinoma by 16 weeks postinfection, only 64% of Δfur strain-infected animals developed similar pathology by the same time point. Together, these data indicate that while Fur is not essential for development of disease, it plays a crucial role in the rate of disease development and the overall disease severity.
Pathogenic bacteria must be able to adapt to fluctuations in their in vivo environments to successfully survive and colonize. Due to its low pH and overall harsh environment (45, 47), the stomach was once thought to be a sterile chamber that is not hospitable to bacteria. However, it is now clearly recognized that H. pylori is able to successfully colonize this niche. Indeed, it is estimated that over 50% of the world's population is colonized by H. pylori (10), which is remarkable given the dynamic nature of the gastric niche, the constant need to adapt to the fluctuating environment, and the fact that the H. pylori genome is predicted to encode a paucity of regulatory factors and two-component systems that aid in the process of adaptation (1, 56, 64). Because of this, it has been hypothesized that the regulators encoded by H. pylori have acquired additional functions to compensate for the lack of other adaptive systems. H. pylori Fur exemplifies this hypothesis as it regulates genes directly involved in iron uptake and storage (25, 30), as well as genes involved in acid acclimation (67, 68), nitrogen metabolism (68), and the oxidative stress response (19, 26). These increased regulatory functions, as well as the diverse H. pylori Fur regulon, suggest that Fur plays an important role in H. pylori adaptation (46, 67) and colonization. Indeed, in addition to the fur-mediated gerbil colonization defect observed in our study, previous work with the mouse model showed that there was a colonization defect of the H. pylori fur mutant (14). Moreover, it is important to note that Fur has also been shown to be an important factor for both colonization and virulence in a diverse range of bacteria. Some examples include the closely related organism Campylobacter jejuni, where a fur mutant exhibits diminished colonization (50), and Staphylococcus aureus, where a fur mutant exhibits attenuated abscess formation (33).
We found that despite the fact that during infection the initial levels of Δfur H. pylori were greater than those of the wild type, the Δfur strain attempts to colonize the body of the stomach and does not properly attach to the stomach tissue. This led to a dramatic reduction in colonization by day 3 postinfection (Fig. (Fig.3).3). After this, the surviving Δfur bacteria exhibited a lag in the ability to grow in the antrum, but the levels of colonization eventually reached the wild-type levels. The gastric mucosa is a dynamic environment where mucus is constantly secreted from the glands, and the overall conditions in the fed and fasting states are radically different (43, 48). Normally, H. pylori is found to predominantly colonize the antrum of the human stomach and is believed to reside primarily in the gastric mucus layer (38, 49). Previous work has shown that, as observed for humans, in the gerbil mucosa nearly all of the H. pylori cells are located in the juxtamucosal layer above the tissue surface (4). Proper localization of H. pylori to this site depends on environmental cues, and the distribution of H. pylori in the mucus layer is important for maintenance of infection. Indeed, Azevedo-Vethacke et al. previously showed that when gerbils infected with H. pylori were treated with proton pump inhibitors to increase the gastric pH, the distribution of H. pylori in the mucus layer was altered (4). This perturbation of the distribution of H. pylori in the mucosal layer was hypothesized to subsequently increase the potential for this organism to be cleared from the stomach (4).
Optimal localization in the stomach requires both motility and chemotaxis, both of which have been shown to be important for H. pylori colonization in numerous animal models (17, 28, 37, 44, 53, 62). Indeed, motility defects due to mutations in the flagellins FlaA and FlaB have been shown to affect H. pylori colonization (23, 35). Chemotaxis mutants also have a defect in colonization and persistence or are unable to colonize in certain models (28, 62). Consistent with these observations, H. pylori genes encode CheW, CheA, and CheY homologues, which affect flagellar direction and velocity, as well as three proteins that are hybrids of CheW and CheY (CheV proteins). Unlike the proteins of other organisms, these CheV proteins do not appear to be redundant with CheW, suggesting that these hybrid proteins have unique chemotactic features in H. pylori. CheV2 is upregulated in vivo (57), and based on transcriptional profiling it has been suggested that cheV2 expression is regulated by apo-Fur (25). When examined in the mouse model, cheV1, cheV2, and cheV3 mutants were all shown to be outcompeted by a wild-type H. pylori strain (42). Additionally, CheV3 was shown to be essential for colonization in the gerbil model (36), although it should be noted that the cheV3 transposon mutation used in that study may have had polar effects on the downstream genes cheA and cheW. Overall, given the important role of the CheV proteins in vivo, deregulation of CheV2 in the fur mutant may affect chemotaxis characteristics that are crucial for the initial localization of H. pylori. Furthermore, Fur has been shown to regulate expression of the flagellar biosynthesis genes fliP, flaB, and flgE (25). Two of these genes, flaB and flgE, have been shown to be upregulated during gerbil infection (25, 57). In a fur mutant, these motility and chemotaxis genes are expressed aberrantly, and this may be a factor that affects the ability of the Δfur strain to colonize the more proximal corpus of the stomach, as well as its inability to colonize the antral stomach epithelium to the same extent as the wild type during early stages of infection (Fig. (Fig.3C).3C). Thus, improper localization of the Δfur bacteria to the corpus and mucosal layers may be due in part to alteration of the expression of multiple motility and chemotaxis genes.
Although H. pylori colonizes the stomach, it is considered a neutrophile, and numerous bacterial components aid in its survival in an acidic niche (55). Given the fact that Fur regulation has been shown to be crucial for growth at an acidic pH (9), it is perhaps not surprising that genes that are involved in acid acclimation, including the aliphatic amidase gene amiE and the formidase gene amiF, are upregulated in vivo (57). The amidases provide alternative sources of ammonia by degrading amide substrates to ammonia and the corresponding organic acid (58, 59). The resulting ammonia then helps buffer the microenvironment and provides protection for H. pylori. Thus, the decreased acid acclimation capability of the Δfur strain may be linked to the crucial role of Fur in regulating these factors. Additionally, changes in the expression of motility and chemotaxis genes may result in changes in the orientation of H. pylori in the mucosal layer which could expose the Δfur strain to increased acid stress and could account for the increased clearance of the Δfur strain that we observed from days 1 to 3 (Fig. (Fig.1).1). Furthermore, the need to acclimate to acid varies in the corpus and antrum of the stomach; the acid-secreting corpus has been shown to have a broader pH range (pH 1.8 to 4.5) than the antrum (pH 1.6 to 2.6) (45). While wild-type H. pylori may be able to sense this gradient and preferentially colonize the antral site with a relatively low and stable pH, the ability of the Δfur strain to sense and properly acclimate to the pH gradient is likely compromised. This would account for the dramatic clearance of the Δfur strain from the corpus that we observed (Fig. (Fig.3C3C).
Even though Fur controls a diverse gene regulon in H. pylori (25, 30), it does not appear to be required for chronic infection (Fig. (Fig.11 and and2);2); Fur is important for establishing colonization, but it is not necessary for maintenance of infection. In fact, at later stages of infection we were unable to displace the Δfur strain with the wild-type strain in most animals (Fig. (Fig.2A).2A). While the reason for this is not completely clear, one possible explanation for the lack of wild-type superinfection is that the Δfur strain is able to properly localize over time and adapt to its gastric niche through utilization of other regulatory proteins. Indeed, CsrA (5), ArsRS (41), and NikR (14) have each been shown to be important for the adaptation of H. pylori to oxidative stress and low pH. Thus, after the Δfur mutant has successfully colonized an animal, these regulatory proteins may aid in adaptation to the gastric environment. Another possible explanation is that there are limited sites for bacterial colonization in the stomach. Once the Δfur strain occupies and colonizes the gastric sites, there may be few remaining sites that are available for wild-type colonization. Together, these findings suggest that, although Fur is important for the dynamics of infection, it is not necessary for persistent colonization of the stomach.
Compared to many other bacterial pathogens, H. pylori has a compact genome that is predicted to code for only about 1,600 proteins (1). Given this small genome size and the fact that this bacterium occurs in an intimate equilibrium with the host, it is perhaps not surprising that our finding that mutation of a single gene can affect H. pylori colonization and inflammation is not unprecedented. For instance, the acid-buffering urease (21, 22, 65, 72), the ferritin gene pfr (70), and the chemotaxis gene cheY (44) each have been shown to be important for colonization. Additionally, the level of inflammation in the host induced by a tlpB chemotaxis mutant has been demonstrated to be decreased despite levels of colonization that are similar to those of the wild-type strain (44). Thus, multiple H. pylori genes appear to play an intimate role in colonization and host-pathogen interactions.
Despite the fact that the wild-type and Δfur strains exhibited comparable levels of colonization by 4 weeks postinfection, we observed a detectable lag in the development and severity of pathology in the Δfur mutant-infected animals. The fact that both wild-type-infected animals and Δfur mutant-infected animals developed gastric adenocarcinoma indicates that, while Fur is not essential for disease, it does play a role in the time required to reach this endpoint. Once a Δfur strain reaches the proper site of colonization, there is a delay in its ability to evoke a host response (Fig. 4A and B). One possible explanation for this delay is altered expression of outer membrane proteins (OMPs). It is predicted that 4% of the H. pylori genome codes for OMPs (64), many of which are involved in adhesion (52, 73) and immune modulation (27, 66). One of these OMPs, HopZ, is Fur regulated and is involved in adhesion (30, 52). Thus, altered expression of hopZ could contribute to the defect in tissue adherence that we observed (Fig. (Fig.3B),3B), as well as subsequent changes in inflammation and injury (Fig. (Fig.5A).5A). In support of the role of adherence in cross talk with the host, increased adherence of H. pylori to cells has been shown to result in an increase in the host immune response (31). Thus, the observed preference of wild-type bacteria for the gastric tissue over the mucus layer (Fig. (Fig.3C)3C) suggests that Fur-regulated adherence genes, such as hopZ, may also be involved in the initial colonization of the stomach and the development of inflammation.
An important aspect of H. pylori-induced disease is the ability of some strains to deliver the virulence factor CagA into host cells. CagA is delivered into epithelial cells through a type IV secretion system, and CagA-positive strains have been shown to be associated with more severe gastric inflammation and gastric cancer (11, 39, 51). While the expression of cagA appears to be Fur independent, delivery of CagA and the resulting downstream effects require intimate interaction of H. pylori with host cells. Thus, decreased adherence by the Δfur strain may decrease the amount of CagA delivered to host cells. Furthermore, the type IV secretion system has also been shown to be important for the delivery of peptidoglycan to host cells, which has been reported to provoke inflammation through Nod1 signaling (13). Thus, in addition to altered CagA delivery due to the decreased bacterium-host contact, strains displaying altered adherence would also provide less exposure of the host to H. pylori peptidoglycan. Intimate contact between H. pylori and gastric epithelial cells is clearly important for host inflammation and disease. Our finding that there were delays in the development and severity of inflammation and gastric injury in our Δfur strain-infected animals undoubtedly was a consequence of the multiple effects of Fur regulation. Our findings also demonstrate that the mere presence or absence of Fur is not responsible for disease. Future studies examining which Fur-regulated genes are important for proper H. pylori tissue targeting and colonization of the gastric niche may provide insight into the development of prophylactic therapeutic targets.
This work was supported by NIH grants AI065529 (D.S.M.), CA082312 (A.D.), and DK58587, CA77955, and CA116087 (R.M.P.).
We thank K. Jones for her assistance with harvesting mucus samples, C. Olsen for her input concerning statistical methods and resources, A. Barnoy for help with figure preparation, and members of the Merrell lab for useful discussions.
The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the NIH or DOD.
Editor: S. R. Blanke
Published ahead of print on 26 April 2010.