Positive Selection of Mutations in the Helicobacter pylori katA 5′ Untranslated Region in a Mongolian Gerbil Model of Gastric Disease

ABSTRACT

To evaluate potential effects of gastric inflammation on Helicobacter pylori diversification and evolution within the stomach, we experimentally infected Mongolian gerbils with an H. pylori strain in which Cag type IV secretion system (T4SS) activity is controlled by a TetR/tetO system. Gerbils infected with H. pylori under conditions in which Cag T4SS activity was derepressed had significantly higher levels of gastric inflammation than gerbils infected under conditions with repressed Cag T4SS activity. Mutations in the 5′ untranslated region (UTR) of katA (encoding catalase) were detected in strains cultured from 8 of the 17 gerbils infected with Cag T4SS-active H. pylori and none of the strains from 17 gerbils infected with Cag T4SS-inactive H. pylori. Catalase enzymatic activity, steady-state katA transcript levels, and katA transcript stability were increased in strains with these single nucleotide polymorphisms (SNPs) compared to strains in which these SNPs were absent. Moreover, strains harboring these SNPs exhibited increased resistance to bactericidal effects of hydrogen peroxide, compared to control strains. Experimental introduction of the SNPs into the wild-type katA 5′ UTR resulted in increased katA transcript stability, increased katA steady-state levels, and increased catalase enzymatic activity. Based on site-directed mutagenesis and modeling of RNA structure, increased katA transcript levels were correlated with higher predicted thermal stability of the katA 5′ UTR secondary structure. These data suggest that high levels of gastric inflammation positively select for H. pylori strains producing increased levels of catalase, which may confer survival advantages to the bacteria in an inflammatory gastric environment.

INTRODUCTION

Helicobacter pylori are Gram-negative bacteria that persistently colonize the human stomach in half of the world’s population (1–3). H. pylori colonization of the stomach results in a chronic mucosal inflammatory response. Gastric inflammation does not cause symptoms in most individuals, but H. pylori colonization of the stomach is a risk factor for peptic ulcer disease and gastric cancer (1, 2).

The gastric mucosal inflammatory response to H. pylori is characterized by increased levels of reactive oxygen species (ROS) and reactive nitrogen species (RNS). H. pylori utilizes several mechanisms to resist the effects of ROS and RNS. These include production of antioxidant enzymes such as superoxide dismutase (SOD), which converts superoxide radicals into hydrogen peroxide (4, 5), and catalase (katA), which converts hydrogen peroxide into water and oxygen (6–8). Other systems used by H. pylori to overcome ROS and RNS stress include three peroxiredoxins (alkyl hydroperoxide reductase [AhpC], thiol peroxidase [Tpx], and bacterioferritin comigratory protein [BCP]) (9–13), thioredoxins (either Trx1 or Trx2 coupled with a thioredoxin reductase, TrxA) (14), NADPH quinone reductase (MdaB) (14, 15), and NapA (16–18). H. pylori mutant strains with null mutations in these genes have a reduced ability to colonize the mouse stomach (5, 11, 14, 15, 18). The use of multiple enzymes to counteract ROS and RNS stress likely reflects the substantial impact of oxidative and nitrosative stress conditions on H. pylori in the gastric niche.

H. pylori strains isolated from unrelated individuals exhibit a high level of genetic diversity, including variations in gene content and variations in the nucleotide sequences of individual genes (19, 20). The high level of allelic diversity is attributable to both a high mutation rate and the natural competence of H. pylori, which facilitates horizontal gene transfer and intraspecies recombination (21, 22). Phase variation provides an additional mechanism for genetic variation (23–26).

In addition to the marked genetic variation observed among H. pylori strains isolated from unrelated individuals, H. pylori strains undergo genetic changes within individual stomachs (27–36). Selection of specific mutations has been observed over time within individual human hosts and in association with colonization of new human hosts (28–33, 37). H. pylori genetic diversity is proposed to contribute to resilience of the bacterial population upon initial colonization and when subsequent changes in the gastric environment are encountered. The use of animal models provides a valuable tool for analyzing the effects of environmental conditions on selection of H. pylori variants (35, 38–43). For example, previous studies have shown that dietary composition (such as a high-salt diet or low-iron diet) (39–41) can influence the selection of specific mutations.

One of the most striking genetic variations among H. pylori strains is the presence or absence of a chromosomal region known as the cag pathogenicity island (PAI) (44, 45). Colonization of the human stomach with H. pylori strains containing the cag PAI is associated with increased severity of gastric inflammation and higher gastric cancer risk, compared to colonization with strains lacking the cag PAI (44). The cag PAI encodes a type IV secretion system (Cag T4SS) that is required for the delivery of CagA (a secreted bacterial oncoprotein) and other bacterial constituents into gastric cells (46–48). Numerous cellular alterations potentially leading to malignant transformation occur upon entry of CagA into host cells (49–52). Previously, we employed a TetR/tetO system (53) to regulate the H. pylori cagUT operon, which encodes two proteins that are essential for Cag T4SS function. The use of this system in a Mongolian gerbil model of H. pylori infection allowed control of Cag T4SS function in vivo (54). The administration of doxycycline in animal chow resulted in derepression of cagUT expression in vivo, whereas cagUT expression remained repressed in the absence of doxycycline administration. Animals infected with the genetically modified H. pylori and fed doxycycline had higher levels of gastric inflammation and an increased incidence of gastric cancer compared to infected animals fed a drug-free diet (54).

We hypothesized that prolonged severe gastric inflammation results in the selection of H. pylori variants that are optimally adapted for growth in an inflammatory environment. In the current study, we used whole genome sequencing to analyze H. pylori strains isolated from animals in the previous study that had varying levels of gastric inflammation (54). We show that strains isolated from animals with severe gastric inflammation commonly contained single nucleotide polymorphisms (SNPs) in the 5′ untranslated region (UTR) of the katA gene, which were not detected in the parent strain. The SNPs in the katA 5′ UTR gene resulted in elevated katA transcript levels, increased katA transcript stability, increased catalase enzymatic activity, and resistance in vitro to the bactericidal effects of hydrogen peroxide. Taken together, these results suggest that a highly inflammatory gastric environment provides a strong positive selection for H. pylori strains that are most fit for growth within this type of gastric niche.

RESULTS

H. pylori strains isolated from experimentally infected Mongolian gerbils contain SNPs in the katA 5′ UTR.

Previously, we demonstrated the utility of a TetR/tetO system for controlling Cag T4SS activity in a Mongolian gerbil model of H. pylori infection and gastric cancer (54). Specifically, gerbils were infected with an H. pylori strain modified to contain both a TetR repressor in an intergenic locus nonessential for gastric colonization and tetO sites upstream of the cagUT operon (encoding components of the Cag T4SS required for T4SS activity), described further in Materials and Methods (53, 54). Derepression of cagUT expression occurs in the presence of anhydrotetracycline or doxycycline (53, 54). To control cagUT expression in vivo, gerbils were fed either chow containing doxycycline (to derepress Cag T4SS activity, i.e., Cag ON) or drug-free chow (i.e., Cag OFF) (Fig. 1A, B) (54). Increased severity of gastric inflammation was observed at 3 months postinfection in H. pylori-infected gerbils that were fed doxycycline-containing chow compared to infected animals that were fed a drug-free chow (54). In the current study, we used whole genome sequencing to analyze H. pylori strains isolated from 17 of the animals that were fed a diet containing doxycycline and 17 animals that were fed drug-free chow (Fig. 1B). We sought to identify genetic features in the H. pylori strains cultured from gerbils (output strains) that were different from those in the strain introduced into the gerbils (input strain), using criteria described in Materials and Methods.

FIG 1.

Schematic illustrating the experimental design. (A) Gerbils were infected with VM202-203, an H. pylori strain in which the expression of cagUT is under the control of the tet operator. Infected animals were fed the appropriate diets (chow containing doxycycline or drug-free chow) beginning 1 week prior to H. pylori infection, and these diets were continued for 3 months. Infections were carried out by oral gavage on day 0 and day 2 and animals were euthanized 3 months postinfection. (B) H. pylori output strains cultured from animals fed chow containing doxycyline (n = 17) or drug-free chow (n = 17) were analyzed by whole genome sequencing.

In comparison to the genome sequence of the strain introduced into the gerbils, the genomes of the output H. pylori strains cultured from the gerbils contained numerous substitution mutations, as well as insertions and deletions. Many of these alterations were detected in only a small proportion of the sequence reads for each output strain or only in a small number of output strains and therefore were not studied further. Conversely, 8 of 34 output strains contained SNPs immediately upstream of the katA open reading frame (ORF) (Fig. 2A), and these substitutions were detected in 100% of sequence reads. The SNPs were located downstream of the katA transcriptional start site, within the 5′ UTR of katA (55, 56). Five unique nucleotide changes, designated SNPs 1–5, were identified in strains isolated from individual animals (Table 1, Fig. 2A and B).

FIG 2.

katA 5′ untranslated region (UTR) single nucleotide polymorphisms (SNPs) identified in this study. (A) Nucleotide sequences of the katA 5′ UTR in the input H. pylori strain (VM202-203) and gerbil output (OP) strains. The 5 different SNPs identified in this study (highlighted in red) are found immediately downstream of the katA transcriptional start site (55, 56). Highlighted in yellow is the nucleotide sequence of the 5′ UTR of katA in the VM202-203 input strain. RBS, ribosomal binding site. (B) RNA-fold software was used to predict the RNA secondary structure of the katA 5′ UTR in the input strain. Blue font indicates the locations of the nucleotide SNPs (RNA) based on the predicted structure of the katA 5′ UTR of the input strain. SNPs 1, 2, and 4 were identified in output strains OP1, OP2, and OP6, respectively. SNP 3 was identified in multiple output strains (OP3, OP4, OP5), and SNP 5 was also identified in multiple output strains (OP7 and OP8). The nucleotide changes (NC) for each SNP and the output strains harboring the SNPs are shown. OP9 is a representative output strain that does not contain nucleotide changes in the katA 5′ UTR. The katA transcriptional start site (TS) is defined as position 1. SNP1 contains a G3C substitution, in which the guanine at position 3 of the katA 5′ UTR is replaced with a cytosine. SNPs 2, 3, 4, and 5 contain A5G, C8T, A9G, and G17C substitutions (Table 1).

Table 1.

Summary of output strains analyzed in this study

Strain	Doxycycline	Disease state	Inflammation score	katA 5′-UTR SNPa	SNPb
OP1	Yes	Adenocarcinoma	8	Yes	G3C
OP2	Yes	Dysplasia	8.5	Yes	A5G
OP3	Yes	Gastritis	3	Yes	C8T
OP4	Yes	Dysplasia	9	Yes	C8T
OP5	Yes	Dysplasia	8.5	Yes	C8T
OP6	Yes	Gastritis	3	Yes	A9G
OP7	Yes	Dysplasia	9	Yes	G17C
OP8	Yes	Gastritis	3	Yes	G17C
OP9	No	Normal	0	–	–
OP10	No	Gastritis	1	–	–
OP11	No	Normal	0	–	–
OP12	No	Gastritis	1.5	–	–
OP13	No	Normal	0	–	–
OP14	No	Normal	0	–	–
OP15	No	Gastritis	1	–	–
OP16	No	Gastritis	1	–	–
OP17	No	Normal	0.5	–	–
OP18	No	Normal	0	–	–
OP19	No	Normal	0	–	–
OP20	No	Normal	0.5	–	–
OP21	No	Normal	0	–	–
OP22	No	Normal	0	–	–
OP23	No	Gastritis	3	–	–
OP24	No	Gastritis	1	–	–
OP25	No	Gastritis	1.5	–	–
OP26	Yes	Gastritis	2	–	–
OP27	Yes	Gastritis	3	–	–
OP28	Yes	Gastritis	2.5	–	–
OP29	Yes	Gastritis	5	–	–
OP30	Yes	Gastritis	1	–	–
OP31	Yes	Gastritis	1	–	–
OP32	Yes	Dysplasia	9	–	–
OP33	Yes	Adenocarcinoma	8	–	–
OP34	Yes	Adenocarcinoma	10	–	–

^aThe presence of single nucleotide polymorphisms (SNPs) in the katA 5′ untranslated region (UTR) detected by whole genome sequencing, as described in Materials and Methods. (–) Indicates that the katA 5′ UTR is identical to that of the VM202-203 input strain used to infect the Mongolian gerbils; no SNP was identified in these strains.

^bThe nucleotide present in the input strain is followed by the location of the SNP and nucleotide present in the output strain (OP). Position of the SNPs is relative to the katA transcriptional start site (position 1) (55, 56).

As a first approach for analyzing this region, we examined the katA 5′ UTR of the input strain using the RNA structure prediction program RNAfold (Institute for Theoretical Chemistry, University of Vienna, Vienna, Austria) (57). As shown in Fig. 2B, nucleotides 3 to 27 of the 5′ UTR of katA (input strain) are predicted to form a stem loop structure. SNPs 1-5 are all predicted to be located within this stem-loop structure (Fig. 2B).

Correlation of katA 5′-UTR SNPs with diet, disease, and inflammation scores.

All the strains containing a katA 5′-UTR SNP were isolated from animals fed a diet containing doxycycline (which derepresses Cag T4SS activity) (Table 1). Therefore, we examined in greater detail the association of katA 5′-UTR SNPs with diet, disease, and inflammation scores (Fig. 3). SNPs in the katA 5′ UTR were detected in 8 of 17 output strains from animals fed a doxycycline-containing diet and none of the 17 output strains derived from animals fed a drug-free diet (Fig. 3A; odds ratio = infinite; P = 0.0027, Fisher’s exact test). As shown in Fig. 3B, strains containing a katA 5′-UTR SNP were associated with higher inflammation scores compared to strains in which a 5′-UTR SNP was absent (P = 0.0005, Mann Whitney U test). In addition, katA 5′-UTR SNPs were more likely to be found in output strains from animals with more severe disease (dysplasia or gastric cancer) than in output strains from animals with less severe disease (Fig. 3C). Specifically, katA 5′-UTR SNPs were identified in 5 of 8 (62.5%) output strains from animals with dysplasia or gastric cancer and 3 of 26 (8%) animals with less severe disease or normal histology (Fig. 3C; odds ratio = 11.4 [1.4, 121.9]; P = 0.0085, Fisher’s exact test). In an analysis limited to infected animals receiving a doxycycline-containing diet, there was a statistically nonsignificant trend (P = 0.2370, ANOVA with Tukey’s multiple comparison test) toward higher inflammation scores in animals from which the corresponding H. pylori isolates contained a katA 5′-UTR SNP than in animals from which the H. pylori isolates lacked such SNPs (Fig. S1 in the supplemental material). Taken together, the results indicate that selection for katA 5′-UTR SNPs occurred mainly in animals with more severe gastric inflammation and more severe gastric disease.

FIG 3.

katA 5′-UTR SNPs are found in output strains from animals with higher gastric inflammation scores. (A) Animals received either doxycycline-containing chow (to derepress cagUT expression in vivo) or drug-free chow. Seventeen H. pylori strains were isolated from gerbils fed a diet containing doxycycline (Cag ON), and 17 were from gerbils fed a drug-free diet (Cag OFF). Eight of 17 strains from animals receiving doxycycline contained a katA 5′-UTR SNP; 0 of 17 strains from animals on the drug-free diet contained a katA 5′-UTR SNP (P = 0.0027, Fisher’s exact test). (B) H. pylori strains harboring a katA 5′ UTR mutation were associated with higher gastric inflammation scores compared to strains in which katA 5′-UTR mutations were absent (***, P = 0.0005, Mann Whitney U-test). Inflammation scores from animals receiving a doxycycline-containing diet are shown in red, while inflammation scores from animals fed a drug-free diet are shown in blue. Horizontal lines indicate mean values. (C) The association between gastric disease state and the presence or absence of a katA 5′-UTR SNP is shown. None of the 10 strains isolated from gerbils with normal gastric histology contained a katA 5′-UTR SNP. Only 3 of the 26 strains isolated from gerbils with mild gastric disease (normal gastric histology or mild gastritis only) contained a katA 5′-UTR SNP. In contrast, a katA 5′-UTR SNP was present in strains isolated from 5 of the 8 animals with severe disease (dysplasia/cancer) (P = 0.0085, Fisher’s exact test).

katA 5′-UTR SNPs are associated with higher catalase enzymatic activity and catalase transcript levels.

To evaluate whether the katA 5′-UTR SNPs influence catalase enzymatic activity, we analyzed catalase enzymatic activity in output strains containing 5 different SNPs (output strains OP1, OP2, OP3, OP6, and OP7; Table 1) compared to output strains in which katA 5′-UTR SNPs were not identified. As shown in Fig. 4A, the levels of catalase activity in the SNP-containing strains were significantly higher than those of control strains lacking a 5′-UTR SNP. Consistent with this, we also observed increased katA transcript levels in the SNP-containing strains compared to the control strains in which the 5′-UTR SNPs were absent (Fig. 4B). In contrast, there was little or no difference in the expression of the housekeeping gene gyrB (DNA gyrase subunit B) in the two groups of strains (Fig. 4C). These results indicate that strains containing the katA 5′-UTR SNPs have increased catalase enzymatic activity and increased katA transcript levels compared to strains in which these SNPs are absent.

FIG 4.

Increased catalase activity and katA transcript levels are observed in output strains containing a katA 5′-UTR SNP. (A) Comparison of catalase activity in strains containing a katA 5′-UTR SNP with catalase activity in strains in which these SNPs are absent. Dots in A, B, and C represent the mean results from at least 4 independent experiments for each strain. Catalase activity analysis in each experiment was performed in triplicate for each biological sample. Horizontal lines indicate mean values for all strains combined. (B and C) katA and gyrB transcript levels in strains containing a katA 5′-UTR SNP and strains in which no katA 5′ UTR was identified. The results are based on a minimum of 4 independent experiments. The 5 analyzed strains with SNPs were output strains 1, 2, 3, 6, and 7 (Table 1). The 5 analyzed strains in which SNPs were absent (i.e., input katA 5′ UTR was present) were output strains 9, 30, 31, 32, and 33 (Table 1). Results for catalase activity (A) and transcript levels (B and C) are reported as relative levels in comparison to output strain OP9, an output strain in which the katA 5′ UTR does not contain a SNP. **, P < 0.01, Mann-Whitney U test.

Output strains harboring katA 5′-UTR SNPs exhibit a fitness advantage under conditions of hydrogen peroxide stress.

We next explored whether increases in katA transcript levels affected the ability of H. pylori output strains to survive two types of oxidative stress: hydrogen peroxide (eliminated by catalase) or cumene hydroperoxide (not eliminated by catalase) (13, 15, 58, 59). As shown in Fig. 5, hydrogen peroxide inhibited growth of H. pylori strains OP9, OP32, and OP33, which lack katA 5′-UTR SNPs. In contrast, H. pylori output strains containing katA 5′-UTR SNPs A5G (OP2), C8T(OP3), or G17C (OP7) were able to grow in the presence of hydrogen peroxide. Cumene hydroperoxide (1 mM) also had an inhibitory effect on H. pylori growth, but its effects were uniform for all six of the H. pylori output strains tested (Fig. 5). These results provide evidence that the katA 5′-UTR SNPs confer a fitness advantage to H. pylori strains in the presence of hydrogen peroxide.

FIG 5.

Output strains with katA 5′-UTR SNPs exhibit increased resistance to hydrogen peroxide. Output strains containing katA 5′-UTR SNPs (output strains OP2, OP3, and OP7; Table 1) and output strains that did not contain a SNP in the 5′ katA UTR (output strains OP9, OP32, and OP33; Table 1) were tested for susceptibility to the bactericidal activities of hydrogen peroxide and cumene hydroperoxide. H. pylori cultures (at an optical density at 600 nm of 0.4) were treated for 10 min with either hydrogen peroxide (100 mM) or cumene hydroperoxide (1 mM) (6, 7, 13, 59). As controls, H. pylori cultures were treated with water or DMSO (diluents for hydrogen peroxide and cumene hydroperoxide, respectively). The cultures were then serially diluted and aliquots (5 μl) were added to blood agar plates. The figure shows the testing of 1:625 dilutions and observation of H. pylori growth after 5 days. Higher concentrations of cumene hydroperoxide (5 mM and 10 mM) were also tested, but there was no growth of any of the H. pylori strains under those conditions (data not shown).

Increased katA transcript stability in output strains containing katA 5′-UTR SNPs.

We hypothesized that SNPs present in the katA 5′ UTR might affect katA transcript stability, thereby altering katA transcript levels and catalase enzymatic activity. To test this hypothesis, we examined levels of katA mRNA at various time points following the addition of rifampin, an inhibitor of bacterial DNA-dependent RNA polymerase (60), to H. pylori broth cultures (61). We compared katA mRNA levels in output strains that contained katA 5′-UTR SNPs (output strains OP1, OP2, and OP3; Table 1) with the levels in output strains that did not contain a SNP in the 5′ katA UTR (output strains OP9, OP32, and OP33; Table 1). As shown in Fig. 6A, a time-dependent decrease in katA transcript levels was observed upon the addition of rifampin to all the broth cultures. The rate of katA decay was dependent on whether a SNP was present in the 5′ UTR of katA (Fig. 6A). The katA half-lives for output strains OP1, OP2, and OP3 with 5′-UTR SNPs (6.34, 9.64, and 7.32 min, respectively) were significantly longer than those of output strains OP9, OP32, and OP33 (2.9, 2.9, and 2.2 min, respectively) (P = 0.0057, one-way ANOVA with Dunnett’s multiple comparisons) (Fig. 6B). The presence of SNPs in the 5′ UTR in katA did not affect transcript stability of the control gene gyrB in the tested strains (Fig. 6C). These results suggest that katA 5′-UTR SNPs increase katA transcript levels by increasing katA transcript stability.

FIG 6.

katA transcripts are more stable in output strains containing katA 5′ UTR mutations. Transcription was inhibited by addition of rifampin to H. pylori cultures, as described in Materials and Methods. At the indicated time points, aliquots of the cultures were removed and frozen. Following RNA isolation, katA and gyrB transcript levels were quantified by RT-qPCR, as described in Materials and Methods. (A) katA levels at each time point were compared to the katA levels of the same cultures that did not receive rifampin treatment (i.e., the levels at time zero) (see Materials and Methods). Results from output strains harboring a katA 5′-UTR SNP (OP1, OP2, and OP3) are shown in blue, while results from output strains without a katA 5′-UTR SNP (OP9, OP32, and OP33) are shown in red. The results are based on a minimum of 4 independent experiments (mean ± SE). (B) The data from A were used to plot a katA transcript decay curve, and katA transcript half-lives were calculated as described in Materials and Methods. A one-way ANOVA with Dunnett’s multiple comparison was used to analyze statistical significance (*, P < 0.05; **, P < 0.01). (C) Analysis of the same samples for gyrB expression.

Introduction of SNPs into the katA 5′ UTR alters katA transcript levels, katA transcript stability, and catalase enzymatic activity.

We next sought to determine whether experimentally manipulating nucleotide sequences in the 5′ UTR of katA altered katA transcript levels. Specifically, we used a counterselection protocol (described in Materials and Methods) (40, 61–63) to introduce the desired SNPs into the katA 5′ UTR (Fig. 7A). In the first set of experiments, we introduced SNP1 (G3C), SNP2 (A5G), SNP3 (C8T), SNP4 (A9G), and SNP5 (G17C) (highlighted in orange in the schematic of Fig. 7A) into output strain OP9 (which contains a katA 5′ UTR identical to that of the input strain). SNP numbers are shown in Fig. 2 and OP strain numbers are described in Table 1. In parallel, as a control, we used the same protocol to genetically manipulate OP9 but retain the 5′ UTR nucleotide sequence originally present in the strain (Fig. 7A, highlighted in yellow). As controls, we also introduced two additional mutations (T24G and T24C), different from the SNPs identified in output strains, into the 5′ UTR of katA (Fig. 7A, B, highlighted in pink). The predicted location of T24 in the RNA structure (Fig. 2) is discussed subsequently. As shown in Fig. 7B, the levels of katA transcript were higher in manipulated strains containing SNPs (1, 2, 3, 4, or 5) in the katA 5′ UTR than in OP9 (labeled “input” or “input res”) (P < 0.05, Student’s t test). As expected, the manipulated control output OP9 strain in which the 5′ UTR of the input strain was restored showed little or no increase in katA transcript levels compared to the original output OP9 strain. The introduction of T24G and T24C did not affect katA transcript levels (Fig. 7B, pink bars). The expression of the housekeeping gene gyrB was similar in all the strains (Fig. 7C).

FIG 7.

Site-directed mutagenesis of the katA 5′ UTR results in altered katA transcript levels. (A) H. pylori output strain OP9, which contains a katA 5′ UTR identical to that of the input strain VM202-203, was genetically manipulated by inserting an aacC4-rpsL cassette into the 5′ UTR of the katA gene. The resultant strain containing the katA:aacC4-rpsL insertion was transformed with plasmids that either re-introduced the VM202-203 5′ katA UTR (yellow) or introduced nucleotide variations in the katA 5′ UTR corresponding to those found in the output strains OP1 (G3C), OP2 (A5G), OP3 (C8T), OP6 (A9G), or OP7 (G17C). UTRs harboring a SNP are represented in orange. OP9 control strains harboring mutations C1 (T24G) and C2 (T24C) in the 5′ UTR of katA were also generated (pink). RNA was isolated from the manipulated strains, and transcript levels of katA and gyrB were quantified by RT-qPCR as described in Materials and Methods. (B) katA transcript levels in the mutated strains. (C) gyrB expression in the mutated strains. Transcript levels relative to OP9 are reported for each mutated strain examined. The results are based on a minimum of 4 independent experiments (mean ± SE). *, P < 0.05, **, P < 0.001 by Student’s t test, significant differences in katA transcript expression between the indicated strains and the input strain. (D) Catalase enzymatic activity in the manipulated strains was analyzed using an Amplex red catalase kit (described in Materials and Methods). Catalase activity relative to OP9 is reported for each strain examined. Results represent the mean ± SE from at least 4 independent experiments for each strain. Catalase activity analysis in each experiment was performed in triplicate for each biological sample. *, P < 0.05, **, P < 0.001 by Student’s t test, significant differences in catalase activity between the indicated strains and the OP9 strain in which the input katA 5′ UTR was restored.

We also examined catalase enzymatic activity in the manipulated strains. As shown in Fig. 7D, catalase enzymatic activity was higher in manipulated OP9 strains containing SNPs 1, 2, 3, 4, or 5 in the katA 5′ UTR, compared to the OP9 parental strain (Fig. 7D). The control strains (C1 and C2, pink bars) harboring mutations T24G and T24C, respectively, exhibited catalase activities similar to that of the OP9 parental strain and the manipulated OP9 control strain in which the 5′ UTR of the input strain was retained (yellow bar). The increased levels of enzymatic activity in the manipulated OP9 strains (Fig. 7D), compared to either OP9, OP9 with restored input 5′ UTR, or control strains C1 and C2, correlated with increased katA transcriptional levels in the former strains (Fig. 7B).

A similar mutagenesis approach was next carried out with output strains OP1, OP2, and OP3 (Fig. 8A), which contain SNP1 (G3C), SNP2 (A5G), and SNP3 (C8T), respectively. OP1, OP2, and OP3 were previously shown to have higher levels of katA transcript levels than output strain OP9 (which contains the 5′ UTR of the input strain) (Fig. 4B). As shown in Fig. 8B, the introduction of the katA 5′ UTR present in the input strain into the SNP-containing output strains OP1, OP2, and OP3 (highlighted in yellow) resulted in levels of katA transcript that were similar to that observed in output OP9. In contrast, the reintroduction of the 5′ UTR containing SNPs 1 to 3 into OP1 (orange bar), OP2 (green bar), and OP3 (red bar), respectively, resulted in higher levels of katA transcript (Fig. 8B). Levels of the control gene (gyrB) did not vary among the strains (Fig. 8C). Together, these data support the hypothesis that the increased katA enzymatic activity and increased katA transcript levels observed in output strains (Figs. 4 and 6) are attributable to the presence of SNPs in the 5′ UTR of katA.

FIG 8.

Additional site-directed mutagenesis analyses of the katA 5′ UTR. (A) Output strain OP1, which contains a G3C SNP in the katA 5′ UTR, was genetically manipulated as shown in the schematic illustration. An aacC4-rpsL cassette was inserted into the 5′ UTR of the katA gene. The resultant strain (containing the katA::aacC4-rpsL fusion) was then transformed with plasmids that either reintroduced the 5′ UTR of OP1 (shown in green) or plasmids that introduced the 5′ UTR of the input strain (shown in yellow). A similar method was used for genetic manipulation of the katA 5′ UTR in output strains OP2 and OP3 (which contain A5G and C8T SNPs, respectively). The katA 5′ UTR of output strains OP2 and OP3 is shown in orange and red, respectively. RNA was isolated from the manipulated strains and transcript levels of katA and gyrB were quantified by RT-qPCR as described in the Materials and Methods. (B and C) katA and gyrB transcript levels in the mutated strains. Each column is color coded based on the katA 5′ UTR (A). Transcript levels presented in panels B and C are relative to OP9, an output strain with a 5′ UTR identical to that of the input strain VM202-203. The results are based on a minimum of 4 independent experiments (mean ± SE). *, P < 0.05 by Student’s t test, significant differences in katA transcript in each of the strain backgrounds, comparing strains in which the original SNP was restored with the corresponding strain background when the input 5′ UTR (yellow) was introduced. Green indicates OP1 5′ UTR; orange, OP2 5′ UTR; red, OP3 5′ UTR.

As our previous data (Fig. 6) showed that output strains harboring 5′-UTR SNPs exhibit increased katA transcript stability, we hypothesized that katA transcript stability would also be altered in the experimentally engineered strains. To test this hypothesis, we compared katA transcript stability in output strain OP9 harboring an experimentally engineered A5G SNP (SNP2) with transcript stability in a manipulated OP9 strain in which the input strain katA 5′ UTR was restored (Fig. 7A). As shown in Fig. 9A and B, the introduction of the A5G SNP into OP9 resulted in a katA transcript that was more stable (half-life = 8.5 min) than that in OP9 containing the input strain 5′ UTR (half-life = 2.1 min) (P = 0.0079, Mann-Whitney U test). These results are consistent with what was observed with transcript stability studies in the unmanipulated output strains (Fig. 6), where katA transcripts in the output strains containing the A5G SNP were more stable than those in output strains without the mutation. The transcript stability of the control gyrB gene (Fig. 9C) was unaffected by the genetic manipulation of the katA 5′ UTR. These results further support the hypothesis that higher steady-state levels of the katA mRNA transcripts in strains containing katA 5′-UTR SNPs are due to increased katA transcript stability accompanying these nucleotide changes.

FIG 9.

Increased katA transcript stability in H. pylori output strains harboring site-directed mutations in the katA 5′ UTR. Strain OP9 katA::gent-rpsL (Table 2) was genetically manipulated as shown in Figure 7A, resulting in the introduction of a guanine at position 5 of the katA 5′ UTR (A5G SNP2, found in output OP2) or restoration of an adenine at position 5 (i.e., A5, input UTR). To analyze transcript stability, H. pylori cultures were exposed to rifampin for varying lengths of time. At the indicated time points, aliquots of the cultures were removed and frozen. Following RNA isolation, katA and gyrB transcripts were quantitated by real time RT-qPCR. (A) katA levels at each time point were compared to katA levels of the same cultures at time zero, prior to the addition of rifampin. The results are based on a minimum of 4 independent experiments (mean ± SE). (B) Data from A was used to calculate the half-life of katA transcript levels, as described in Materials and Methods. Half-lives (mean ± SE) are shown for OP9 strains containing either the input katA 5′ UTR or containing SNP2 (**, P < 0.01, Mann-Whitney U test). (C) Analysis of gyrB in the same samples.

Specificity of nucleotide substitutions at position 5 of the katA 5′ UTR.

We next investigated whether the observed relationship between katA 5′-UTR SNPs and katA transcript levels was dependent on the specific SNPs identified in the output strains or whether alternate nucleotide substitutions at these sites might have the same effect. To investigate this topic, the 5′ UTR of katA was mutated to introduce nucleotide changes to position 5 (i.e., A5C, A5T) (Fig. 10A, blue asterisk) that were different from the A to G substitution (A5G) found in output strain OP2. Position 5 was chosen for detailed analysis because strains harboring the A5G SNP consistently exhibited higher katA transcript levels and catalase enzymatic activity than control strains (e.g., OP9) that did not contain a SNP in the 5′ UTR (Fig. 7). Moreover, among the SNP-containing strains (i.e., OP1 [G3C], OP2 [A5G], and OP3 [C8T]) that were analyzed for transcript stability (Fig. 6B), katA transcripts in H. pylori strains harboring the A5G SNP were observed to be the most stable. Three nucleotide changes (A5G, A5C, and A5T) were introduced into H. pylori OP9, which contains the 5′ UTR of the input strain. We then compared katA transcript levels in these strains with katA transcript levels in the original OP9. As shown in Fig. 10B, an approximately 4-fold increase in katA transcript was observed in a strain containing the previously identified A to G substitution (A5G found in SNP 2, segment 2, open bar, P < 0.05, Student’s t test). Smaller increases in katA transcript levels were observed in strains harboring A5C or A5T (segments 3 and 4, open bars) substitutions. As controls, katA transcript levels were monitored in H. pylori strains harboring SNPs to nucleotide 24 of the 5′ UTR. Nucleotide 24 is predicted to base pair with the adenine present at nucleotide 5 of the 5′ UTR (Figs. 2 and 10). The introduction of the T24C and T24G substitutions (in the absence of additional genetic modifications) did not affect katA transcript levels (Fig. 10B). As shown in Fig. 10C, expression of the control gene gyrB was unaffected by the nucleotide substitutions upstream of katA.

FIG 10.

Nucleotide sequence and secondary structure of the katA 5′ UTR are determinants of katA transcript levels. (A) RNAfold was used to generate a predicted RNA secondary structure for the katA 5′ UTR of the VM202-203 input strain. Mutations that disrupted the secondary structure were introduced into katA 5′ UTR (at position 5, denoted by *) (see supplemental Fig. S3). Compensatory mutations downstream of the introduced SNP were subsequently introduced at position 24 of the 5′ UTR [denoted by # (green)] to maintain the secondary structure (see supplemental Figure S4). Plasmids harboring the nucleotide changes at position 5 or position 24 of the 5′ UTR of katA were transformed into output strain OP9 (which harbors a katA 5′ UTR identical to the input strain VM202-203; see Figure 7A for experimental approach). (B and C) katA and gyrB transcript levels were quantified by RT-qPCR (see Materials and Methods) in the strains harboring the katA 5′ UTR changes (B and C, respectively). The open (white) bars show results for strains harboring single nucleotide changes to either position 5 or position 24 of the katA 5′ UTR. The closed (black) bars show results for strains harboring nucleotide changes to both position 5 and position 24. Results (mean ± SE) are based on a minimum of 4 independent experiments for each strain. (****, P < 0.0001), **, P < 0.01 by Mann-Whitney U test, significant differences between strains harboring a katA 5′ UTR change at position 5 and strains harboring both a 5′-UTR change at position 5 plus a compensatory mutation at position 24. ####, P < 0.0001, #, P < 0.05 by Mann-Whitney U test, significant differences between strains harboring the input katA 5′ UTR (i.e., no change in nucleotide at position 5, first open bar) and strains harboring either an A5G or A5C mutation.

Relationship between katA 5′ UTR stem loop structure and katA transcript levels.

We next used the multiple alignment tool in the program LocARNA (64) to predict a consensus RNA structure of the katA 5′ UTR based on sequences from seven commonly used H. pylori laboratory strains. This analysis predicts the presence of two stem-loop-forming structures, designated stem A and stem B, in the katA 5′ UTRs of these laboratory strains (supplemental Fig. S2). RNAfold analysis (57) also predicts the existence of stem A in the 5′ UTR of katA from the input strain (Fig. 2, Fig. 10A). Interestingly, all five of the SNPs described in this study localized to stem A (supplemental Fig. S2, Fig. 2). We postulated that nucleotide changes to position 5 may affect the secondary structure of the 5′ UTR and that these structural changes might account for the increased katA transcript levels in H. pylori strains containing A5G, A5C, and A5T substitutions (Fig. 10A). Using RNAfold analyses (57), we evaluated whether the secondary structure of the katA 5′ UTR was affected by these substitutions. As shown in supplemental Fig. S3, the A5G substitution at position 5 of the 5′ UTR (SNP2, Fig. S3B), as well as A5C and A5T substitutions (Fig. S3C and D), affected the predicted structure of stem A, resulting in predicted secondary structures that were different from that in the input katA 5′ UTR. These changes are likely due to loss of base pairing between the upstream and downstream nucleotides that form stem A.

To further evaluate if stem A was an important determinant of katA transcript levels, site-directed mutagenesis was performed to introduce compensatory mutations downstream of the nucleotide substitutions at position 5 (supplemental Fig. S4). For example, the double mutant T24C/A5G was generated when the thymine at position 24 of the 5′ UTR was replaced with a cytosine (T24C) in the katA 5′ UTR containing the A5G substitution (supplemental Fig. S4A). The introduction of the compensatory mutations resulted in complementary base pairing with the nucleotides substitutions at position 5, thereby restoring the predicted RNA stem loop structure of stem A in katA 5′ UTRs (supplemental Fig. S4). The restoration of stem A led to 5′ UTRs that were predicted to be more thermodynamically stable (lower ΔG free energy values) compared to the respective 5′ UTRs that did not contain the compensatory mutations (supplemental Fig. S4; for example, ΔG = −1.0 for A5C and ΔG = −6.3 for T24G/A5C). The mutations were next introduced into output OP9, which contains the katA 5′ UTR of the input strain, and katA transcript levels in the manipulated strains harboring these mutations were compared to katA transcript levels found in output strain OP9. As shown in Fig. 10B, the introduction of T24C into the katA 5′ UTR of H. pylori strains containing the A5G substitution (T24C/A5G, black bars) resulted in marked increases in katA transcript levels compared to the levels in H. pylori strains containing A5G alone (white bars). Likewise, increased katA transcript levels were observed in strains containing T24G/A5C and T24A/A5T alterations (black bars) compared to strains containing A5C and A5T alterations (white bars), respectively. The T24C and T24G substitutions alone did not affect katA levels (Fig. 10B). No changes in gyrB levels were observed in strains harboring these katA mutations (Fig. 10C). Taken together, these results suggest that both the primary nucleotide found at position 5 of the katA 5′ UTR and characteristics of the stem-loop structure of stem A are critical determinants of katA transcript levels.

DISCUSSION

H. pylori colonization of the human stomach results in gastric mucosal inflammation, and varying levels of inflammation severity are detected in different individuals. Similar variation in levels of gastric inflammation is observed among animals experimentally infected with H. pylori. In this study, we investigated the effects of gastric inflammation on H. pylori evolution within the stomach. The study was designed so that gastric inflammation was elicited in response to H. pylori Cag T4SS activity (54). All animals were infected with the same H. pylori strain, thereby avoiding potential biases that might be encountered with experimental approaches requiring the use of multiple strains. For example, a comparative analysis of wild-type and mutant strains might be complicated by differences in the presence of nonabundant genetic variants in the starting populations.

We hypothesized that severe mucosal inflammation might promote the selection of H. pylori strains harboring genetic variations that confer increased fitness in an inflammatory gastric environment. To test this hypothesis, we analyzed H. pylori strains isolated from experimentally infected Mongolian gerbils that displayed varying levels of gastric inflammation (54). Whole genome sequencing revealed many changes among the output strains (e.g., single nucleotide polymorphisms, insertions and deletions) compared to the input strain, most of which were not associated with the severity of gastric inflammation. Conversely, mutations within the 5′ UTR of the katA gene (encoding catalase) were commonly identified in H. pylori strains cultured from animals with high levels of gastric inflammation and rarely identified in strains cultured from animals with low levels of gastric inflammation. Five distinct SNPs in the katA 5′ UTR were identified in eight output strains. Increased katA transcript levels and increased catalase enzymatic activity were observed in H. pylori strains harboring the 5′-UTR SNPs, compared to strains in which the SNPs were absent. In addition, output strains containing the katA 5′-UTR SNPs were more resistant to killing by hydrogen peroxide.

To evaluate the functional specificity of the five nucleotide changes detected in the katA 5′ UTR, we introduced alternate nucleotide substitutions (e.g., A5T or A5C instead of the A5G SNP identified in one of the output strains). Strains possessing the A5T and A5C substitutions showed only minimal increases in katA transcript levels compared to levels in control strains that lacked 5′-UTR SNPs. These results suggest that the observed changes in katA transcript levels and catalase activity were dependent on specific katA 5′-UTR nucleotide substitutions and may explain why some specific SNPs in the katA 5′ UTR were positively selected in vivo and others were not.

The selection of mutations in katA has previously been reported in Mongolian gerbil models fed either normal, low-iron, or high-salt diets (39, 40). For instance, a katA 5′-UTR SNP different from those identified in the current study was previously identified in output strains isolated from Mongolian gerbils with substantial inflammation (inflammation scores of 6.5 to 12 on a 0 to 12 scale) (39). Output strains possessing this 5′-UTR SNP had increased katA enzymatic activity compared to the input strain where the SNP was absent (39). In a separate study (40), H. pylori strains isolated from experimentally infected Mongolian gerbils either contained an intact katA open reading frame (ORF) or a frameshift mutation within the katA ORF. Retrospective analysis of those results showed that output strains possessing catalase activity contained an intact katA ORF and were isolated from animals with high levels of gastric inflammation (average inflammation score 8.8 on a scale of 0 to 12) (40). In contrast, the strain harboring the katA frameshift mutation did not possess catalase activity and was found among animals with lower levels of inflammation (average inflammation score: 3.9) (P = 0.0095) (40).

Our current study provides evidence that the presence of specific SNPs in the 5′ UTR of katA increases katA transcript stability. Using the RNA secondary structure program LocARNA, a consensus structure containing two stem loops (stem A and stem B) was predicted for the katA 5′ UTR of seven commonly used H. pylori strains (supplemental Fig. S2). Stem A is predicted by RNAfold to be present in the 5′ UTR of the VM202-203 input strain, but stem B is not. Importantly, the SNPs identified in output strains were all localized to stem A, suggesting the importance of stem A as a determinant of katA transcript levels. Replacement of A5 in the katA 5′ UTR with guanine, thymine, or cytosine affected predicted stem A formation. The restoration of the secondary structure of stem A through the introduction of compensatory mutations at position 24 downstream of A5G, A5T and A5C resulted in predicted secondary structures with increased thermostability and led to a dramatic increase in the levels of katA transcript levels in these strains compared to strains that possessed only the A5G, A5T, or A5C substitutions. These results provide strong evidence that stem A is an important determinant of katA transcript stability and accumulation, leading to increased KatA levels and enzymatic activity. This finding does not preclude the possibility that the secondary structure changes in the 5′ UTR arising from the SNPs may have effects on ribosome binding site accessibility and translation. The importance of stem loop structures in the 5′ UTR in determining gene expression has previously been reported for H. pylori genes such as cagA and vacA (61, 65).

Traditionally, catalase-mediated protection against oxidative stress had been attributed exclusively to degradation of hydrogen peroxide. More recently, catalase mutants devoid of enzymatic activity have also been shown to offer protection against oxidative stress (8). The protection offered by these enzymatic null mutants occurs through oxidant quenching by the six Met residues present in catalase (8) and the subsequent repair of these oxidized residues by methionine sulfoxide reductase (66, 67). Our current results demonstrate increased catalase enzymatic activity in H. pylori strains possessing the katA 5′-UTR SNPs, which likely confers a selective advantage when the bacteria are challenged with hydrogen peroxide. In addition, it is possible that enzymatic-independent activities of KatA also contribute to oxidant quenching and may similarly confer a selective advantage. The latter view is supported by the observation that we repeatedly detected mutations in the katA 5′ UTR, whereas we did not detect mutations in other genes encoding proteins capable of degrading hydroperoxides (e.g., peroxiredoxin family enzymes such as AhpC).

We propose that the katA 5′-UTR SNPs detected in the current study confer increased fitness in an inflammatory gastric environment. This hypothesis is supported by the finding that the SNPs conferred a selective advantage to strains challenged with hydrogen peroxide in vitro. Retrospective analysis of H. pylori densities in gerbil gastric tissue did not reveal any significant differences when comparing strains harboring the katA 5′-UTR SNPs with strains lacking these SNPs (data not shown). However, it should be noted that levels of gastric inflammation were not identical in these two groups of animals, so the gastric environments were considerably different. The five distinct SNPs in the katA 5′ UTR detected in this study presumably arose through random mutation events and then were positively selected in individual animals. At present, it is unclear if the katA 5′ UTR is a “hot-spot” for mutation or if the results simply reflect the outcome of strong selective forces.

H. pylori-infected gerbils develop gastric inflammation similar to that which occurs in H. pylori-infected humans (68). However, the specific katA 5′-UTR SNPs detected in the current study have not been commonly detected in katA 5′ UTRs of H. pylori strains isolated directly from human stomachs (supplemental Fig. S2). Similarly, the SNPs detected in the current study were not detected in previous genomic analyses of H. pylori strains isolated from Mongolian gerbils (39, 40). There are multiple possible reasons that might account for the relatively high frequency with which katA 5′-UTR mutations arose in the current experiments. For example, the high rate of katA 5′-UTR mutations might reflect specialized properties of the H. pylori strain used in these studies, or administration of doxycycline may influence the gastric environment. In addition, there might be features of the gerbil gastric environment that differ from features of the human gastric environment.

Previous studies have shown that H. pylori evolves within the human stomach and in the stomachs of experimentally infected animals (27–36, 38–43), but there have been relatively few studies examining the effects of specific selective forces on H. pylori evolution. The present study suggests that gastric inflammation drives positive selection of SNPs in the 5′ UTR of the catalase gene to increase katA transcript levels. We propose that the increased catalase levels associated with these SNPs help with the detoxification of oxygen radicals and contribute to H. pylori survival in an inflammatory gastric environment.

MATERIALS AND METHODS

H. pylori strains and growth conditions.

H. pylori strain VM202-203 (54) was previously constructed by introducing the Tet repressor (TetR) into an intergenic region and tetO elements upstream of the cagUT operon in the gerbil-adapted H. pylori strain 7.13, a Cag T4SS-positive strain that has been shown to reproducibly cause cancer in a Mongolian gerbil model. In a recent study (54), we experimentally infected Mongolian gerbils with strain VM202-203 and controlled expression of the cagUT operon in vivo by feeding gerbils AIN-93M rodent chow (Bio-Serv) supplemented with either 10 mg/kg or 25 mg/kg doxycycline (Cag ON). In parallel, control animals were fed AIN-93 chow without doxycycline (Cag OFF). These diets began 1 week prior to infection with H. pylori. Two 0.5-mL orogastric doses of H. pylori (1 × 10⁹ CFU/mL of H. pylori VM202-203) were administered 2 days apart (Fig. 1A). At 90 days postinfection, the animals were euthanized and their stomachs were excised. H. pylori strains were isolated from each of the animals by plating homogenized stomach tissue onto selective medium. Specifically, H. pylori output pools (Table 1) were obtained by culturing on Trypticase soy agar plates containing 5% sheep blood (Hemostat Laboratories), vancomycin (20 μg/ml; Sigma-Aldrich), nalidixic acid (10 μg/ml; Sigma-Aldrich), bacitracin (30 μg/ml; Sigma-Aldrich), and amphotericin B (2 μg/ml; Sigma-Aldrich) to select for H. pylori growth. These plates were incubated at 37°C in a microaerobic chamber (BD GasPak EZ Campy container system) for 5 days. Multiple bacteria from a single plate were pooled to create a freezer stock.

In the present study, we analyzed pools of strains that were cultured from these animals (output strains). Specifically, we analyzed 17 pools of output strains cultured from 17 animals on a doxycycline-free diet, and 17 pools of output strains cultured from 17 animals fed a diet containing doxycycline (Fig. 1B, Table 1). Each pool represents an H. pylori population isolated from a unique animal. For routine growth of bacteria for genome sequencing and transcriptional analyses, H. pylori strains were grown on Trypticase soy agar plates supplemented with sheep blood, in modified Brucella broth containing 5% fetal bovine serum (BB-FBS), or on BB-FBS agar plates in room air supplemented with 5% CO₂ at 37°C. When necessary, BB-FBS agar plates were supplemented with kanamycin (12 μg/ml), chloramphenicol (5 μg/ml), gentamicin (10 μg/ml), or streptomycin (25 μg/ml). Escherichia coli strains were grown on Luria-Bertani medium. When necessary, the E. coli culture medium was supplemented with ampicillin (50 μg/ml), chloramphenicol (25 μg/ml), or streptomycin (25 μg/ml). H. pylori strains cultured from gerbils are listed in Table 1 and mutant strains generated in this study are listed in Table 2.

Table 2.

List of strains generated in this study

Strain name	Relevant characteristics
OP9 katA::gent-rpsL	OP9 containing katA::aacC4-rpsL, GentR
OP1 katA::gent-rpsL	OP1 containing katA::aacC4-rpsL, GentR
OP2 katA::gent-rpsL	OP2 containing katA::aacC4-rpsL, GentR
OP3 katA::gent-rpsL	OP3 containing katA::aacC4-rpsL, GentR
OP9 mut1	OP9, katA 5′ UTR containing SNP1 (G3C)
OP9 mut2	OP9, katA 5′ UTR containing SNP2 (A5G)
OP9 mut3	OP9, katA 5′ UTR containing SNP3 (C8T)
OP9 mut4	OP9, katA 5′ UTR containing SNP4 (A9G)
OP9 mut5	OP9, katA 5′ UTR containing SNP5 (G17C)
OP9 restored	OP9, restored input katA 5′ UTR
OP9 mut6	OP9, katA 5′ UTR containing T24G substitution
OP9 mut7	OP9, katA 5′ UTR containing T24C substitution
OP9 mut8	OP9, katA 5′ UTR containing A5C substitution
OP9 mut9	OP9, katA 5′ UTR containing A5T substitution
OP9 mut10	OP9mut2, katA 5′ UTR containing A5G and T24C substitutions
OP9 mut11	OP9mut8, katA 5′ UTR containing A5C and T24G substitutions
OP9 mut12	OP9mut9, katA 5′ UTR containing A5T and T24A substitutions
OP1 mut1	OP1, katA 5′ UTR containing input katA 5′ UTR
OP1 restored	OP1, katA 5′ UTR containing SNP1 (G3C)
OP2 mut1	OP2, katA 5′ UTR containing input katA 5′ UTR
OP2 restored	OP2, katA 5′ UTR containing SNP2 (A5G)
OP3 mut1	OP3, katA 5′ UTR containing input katA 5′ UTR
OP3 restored	OP3, katA 5′ UTR containing SNP3 (C8T)

H. pylori genome sequence analysis.

DNA samples were prepared from H. pylori with the Wizard genomic purification kit (Promega). H. pylori DNA samples were subjected individually to enzymatic fragmentation using the NEBNext dsDNA Fragmentase kit (NEB) according to the manufacturer’s instructions, with average fragment length of 600 bp (range of 400 to 1,000 bp). Libraries of DNA were prepared from purified fragmented DNA samples using the Jetseq library prep kit with unique indexes as per the manufacturer’s protocol (Meridian Biosciences, Inc., Cincinnati, OH, USA). Quantification of these library preps was performed with the Jetseq library quantification kit (Meridian Biosciences, Inc.), and the libraries were sequenced on a MiSeq sequencer using the 600V3 kit (Illumina Inc., San Diego, CA, USA). For the analysis, raw reads were quality trimmed and aligned to the reference sequence (H. pylori strain B8 [GenBank accession no. NC_014256.1], a closely related strain whose complete genome sequence is available) (69) using CLCbio Genomics workbench version 11. Sequence data were deposited in NCBI (Bioproject ID: PRJNA818266). Data pertaining to read counts and fold coverage are shown in Table S1. Alignment parameters were as follows: Mismatch cost = 2; Insertion cost = 3; Deletion cost = 3; Insertion open cost = 6; Insertion extend cost = 1; Deletion open cost = 6; Deletion extend cost = 1; Length fraction = 0.6; and Similarity fraction = 0.7. The alignment files exported from the CLCbio workbench in BAM format were then imported into a genomics analysis and visualization platform (VGAS) (70) for further coverage and SNP analysis. SNP reports were generated using a 10% cut-off, and genome-wide comparisons of SNPs in the output strains compared to the input strain were then performed.

To identify genetic changes that were subject to strong positive selection, we focused on SNPs (different from the input strain) that were detected in >90% of the sequence reads of output strains. In addition, we sought to identify SNPs that were present in output strains from animals with high-inflammation scores (>6) but not detected in output strains from animals with low-inflammation scores (<2). Using this approach, we detected SNPs in the katA 5′ UTR in seven output strains. One of these same SNPs was detected in a lower proportion of sequence reads from an additional output strain (23% of sequence reads from output 8).

RNA isolation and real-time PCR.

To prepare H. pylori RNA samples for quantitative real-time PCR (RT-qPCR), overnight broth cultures of H. pylori output strains were subcultured into fresh BB-FBS at an an optical density at 600 nm (OD₆₀₀) of ~0.3. Following 6 h of growth, the bacteria were pelleted and resuspended in RNAlater (Ambion) for 40 min. The cell suspensions were centrifuged at 3,500 × g, supernatants were decanted, and the pellets were stored at −70°C. Total RNA was prepared as previously described (40, 71) using TRIzol reagent. Contaminating DNA was removed by digesting the RNA with RQ1 RNase-free DNase (Promega), and the samples were subjected to a cleanup step using RNeasy columns (Qiagen). Each RNA sample was eluted in 100 μl of water. Next, cDNA synthesis was performed with 100 ng of purified RNA using the iScript cDNA synthesis kit (Bio-Rad). As a control, first-strand cDNA synthesis reactions without reverse transcriptase were carried out in parallel. The cDNA and control preparations were diluted 1:20 and used in real-time PCRs. Real-time PCR was performed with an ABI Real Time PCR machine (Step One Plus) with iTaq SYBR green (Biorad) as the fluorochrome. Transcript abundance was assessed by the ΔΔCT method (cycle threshold [CT]), with each transcript signal normalized to the abundance of the 16S rRNA internal control. The normalized transcript signal for each test biological sample was then divided by similarly normalized values from the reference samples to obtain a relative expression ratio (test sample vs. reference sample). The primers used for real-time analysis were as follows: 16S rRNA, 5′-GGAGTACGGTCGCAAGATTAAA-3′ and 5′-CTAGCGGATTCTCTCAATGTCAA-3′; katA, 5′-TCCACTTTGAAACCATGCAA-3′ and 5′-TACAATGCCCACTTCCATCA-3′; and gyrB, 5′-CGTGGATAACGCTGTAGATGAGAGC-3′ and 5′-GGGATTTTTTCCGTGGGGTG-3′.

mRNA stability assays.

These assays were performed as previously described (61). Briefly, overnight cultures of H. pylori strains were grown in BB-FBS and subcultured into fresh BB-FBS (starting at an OD₆₀₀ of ~0.2). Following 6 h of growth, rifampin (80 μg/ml; Sigma-Aldrich) was added to the culture to inhibit transcription. Aliquots of the culture (5 ml) were collected at 3, 6, 10, and 16 min after the addition of rifampin. The bacteria were pelleted and resuspended in RNAlater (Ambion) for 40 min. The cell suspensions were centrifuged at 3,500 × g, the supernatants were decanted, and the bacterial pellets were stored at −70°C. RNA extraction, cDNA synthesis, and real-time RT-qPCR were performed as described above to determine katA transcript levels for each strain at the given time points after rifampin addition. The half-life of the katA transcript was calculated as previously described (61, 65). Briefly, the natural logarithm of relative katA transcript levels was calculated and plotted against time. The slope of each line (i.e., half-life coefficient K) was then used in the formula T_1/2 = ln (2)/K to calculate the half-life (T_1/2) of katA mRNA for each strain.

Analysis of catalase enzymatic activity.

Catalase enzymatic activity of the H. pylori cultures was analyzed as previously described (40). Briefly, overnight broth cultures of H. pylori were inoculated into fresh BB-FBS broth and grown to an OD₆₀₀ of ~0.5 to 0.6. All cultures were normalized to an OD₆₀₀ of 0.1, and catalase enzymatic assays were performed with the Amplex red catalase kit (Life Technologies) per manufacturer’s instructions. The catalase activity in 2-fold serial dilutions of culture samples was compared to the catalase activity of purified catalase standards provided in the kit.

Susceptibility of H. pylori strains to hydrogen peroxide and cumene hydroperoxide.

One-day old cultures of H. pylori grown on plates were resuspended in BB-FBS medium. All cultures were normalized to an OD_600nm of 0.4. Aliquots of each sample (100 μl) were next added to a 96-well plate and treated with either 100 mM H₂O₂ or 1, 5, and 10 mM cumene hydroperoxide (6, 7, 13, 59) for 10 min. As controls, cultures were similarly treated with water (diluent for H₂O₂) or DMSO (diluent for cumene hydroperoxide). The treated cultures were next serially diluted in Brucella broth containing 5% FBS. Aliquots (5 μl) were then added to Trypticase soy agar plates supplemented with sheep blood, and the plates were incubated for 5 days in 5% CO₂ at 37°C.

Mutagenesis of the katA 5′ UTR.

To introduce unmarked mutations into the 5′ UTR of the katA gene of H. pylori output strains, a counterselection method using an aacC4-rpsL cassette was employed. This cassette confers resistance to gentamicin mediated by the aminoglycoside-3-acetyltransferase IV (aacC4) gene and susceptibility to streptomycin mediated by the intact rpsL gene from H. pylori 26695. As a first step in the counterselection mutagenesis, rpsL mutants were generated by transforming H. pylori VM202-203 with a nonreplicating plasmid containing a cloned H. pylori rpsL gene harboring an A-to-G mutation at codon 43 of rpsL (40, 62, 63). The resultant Lys (K)-to-Arg (R) amino acid substitution at position 43 of RpsL confers streptomycin resistance to H. pylori strains bearing this mutation. The Sm^r rpsL-K43R VM202-203 mutants were next transformed with pkat::aacC4-rpsL, a nonreplicating plasmid that allows the insertion of a aacC4-rpsL cassette (conferring gentamicin resistance) into the katA 5′ UTR ORF. To generate the aacC4-rpsL cassette, aacC4 (a kind gift of Sylvie Garneau-Tsodikova, University of Kentucky) was cloned into a previously described pAD-CAT vector (40, 72), generating pADgent-CAT, thereby allowing expression of aacC4 from the ureA promoter present in the vector. Next, the cat (chloramphenicol acetyl transferase) cassette from pADgent-cat was replaced with the rpsL gene of H. pylori. Briefly, pADgent-cat was digested with BamHI and SmaI to release the cat cassette. The rpsL gene was PCR-amplified from pJ261 (73) (a plasmid containing the rpsL gene) as a template, using primers 5′- AAGGATCCGATGCTTTATAACTATGGATTAAACAC-3′ and 5′-AAGGGCCCATCTATCCACTTTTCAATCTATATCATC- 3’containing BamHI (underlined) and SmaI (underlined), respectively. Ligation of the rpsL gene with the BamHI/SmaI digested pADgent plasmid yielded pADgent-rpsL. Plasmid pADgent-rpsL was next digested with kpnI/ApaI to release a 1.5 kb fragment containing the aacC4-rpsL cassette. For downstream cloning, the 1.5 kb kpnI/ApaI fragment was blunt-ended with DNA Polymerase I, Large (Klenow) fragment. To introduce the aacC4-rpsL cassette into the 5′ UTR of the katA gene, plasmid pkatA containing 500 bp upstream and 500 bp downstream of the katA ATG start site was synthesized (Genscript). Plasmid pkatA was digested with AflII (a unique restriction site in the katA 5′ UTR), blunt-ended with klenow, and ligated with the blunt-ended kpnI/ApaI 1.5 kb aacC4-rpsLcassette described above. After ligation with the aacC4-rpsL cassette, the resultant plasmid (pkatA::aacC4-rpsL), which is unable to replicate in H. pylori, was transformed into the rpsL-K43R mutants of H. pylori strains, and single colonies resistant to gentamicin (10 μg/ml) but sensitive to streptomycin (25 μg/ml) were selected. Introduction of the aacC4-rpsL cassette into the katA 5′ UTR was confirmed by PCR amplification and DNA sequencing.

SNPs identified in this study were introduced into plasmid pkatA. These mutated plasmids were then used to transform H. pylori strains harboring pkatA::aacC4-rpsL. Transformations were plated onto BB-FBS agar plates containing streptomycin (25 μg/ml), and streptomycin-resistant colonies were isolated. These streptomycin-resistant colonies result from recombination events in which the aacC4-rpsL insertion in the katA 5′ UTR was deleted and replaced with the desired mutations. The introduction of the desired mutations was confirmed by DNA sequencing.

Ethics statement.

The animal experiments discussed in this study were described in a previous study (54) and were approved by the Vanderbilt University Institutional Animal Care and Use Committee (protocol M1700055-00).

Genome sequence deposition.

H. pylori genome sequence data were deposited in NCBI (Bioproject ID: PRJNA818266).