|Home | About | Journals | Submit | Contact Us | Français|
Animal models are important tools for studies of human disease, but developing these models is a particular challenge with regard to organisms with restricted host ranges, such as the human stomach pathogen Helicobacter pylori. In most cases, H. pylori infects the stomach for many decades before symptoms appear, distinguishing it from many bacterial pathogens that cause acute infection. To model chronic infection in the mouse, a human clinical isolate was selected for its ability to survive for 2 months in the mouse stomach, and the resulting strain, MSD132, colonized the mouse stomach for at least 28 weeks. During selection, the cagY component of the Cag type IV secretion system was mutated, disrupting a key interaction with host cells. Increases in both bacterial persistence and bacterial burden occurred prior to this mutation, and a mixed population of cagY+ and cagY mutant cells was isolated from a single mouse, suggesting that mutations accumulate during selection and that factors in addition to the Cag apparatus are important for murine adaptation. Diversity in both alleles and genes is common in H. pylori strains, and natural competence mediates a high rate of interstrain genetic exchange. Mutations of the Com apparatus, a membrane DNA transporter, and DprA, a cytosolic competence factor, resulted in reduced persistence, although initial colonization was normal. Thus, exchange of DNA between genetically heterogeneous H. pylori strains may improve chronic colonization. The strains and methods described here will be important tools for defining both the spectrum of mutations that promote murine adaptation and the genetic program of chronic infection.
The Gram-negative human stomach pathogen Helicobacter pylori colonizes 50% of the world's population, and chronic infection leads to gastritis, peptic ulcer disease, gastric cancer, and gastric mucosa-associated lymphoid tissue lymphoma in a subset of infected patients (1). H. pylori infection is restricted to humans and nonhuman primates, is usually acquired in childhood, and persists throughout the lifetime of the host.
Several H. pylori strains have been established that will infect mice (2–5). The first and most widely used mouse-adapted strain, SS1, colonizes mice for at least 8 months (3), but it is a poor model for genetic studies, as it is quite difficult to transform, for unknown reasons (6). Strain NSH57 is genetically tractable, but infection wanes after 4 weeks (2). Despite these limitations, mouse models of infection have been used to identify virulence factors required for initial stomach colonization, including flagellar motility and chemotaxis (7, 8), a urease that neutralizes pH around the bacterium in the acidic lumen of the stomach (9), adhesins that bind the bacterium to the epithelial surface (10), and bacterial cell shape (11). There have also been two screens in small animals for genes required for short-term colonization (2, 12). Despite the importance of chronic colonization to disease progression, little is known about the genetic program that promotes long-term infection.
Sequencing of multiple isolates of H. pylori and comparative genomic studies demonstrate a wide diversity of gene content and alleles between clinical isolates (13–15), and most strains are naturally competent (16). Components of the Com apparatus (ComB2 to ComB4 and ComB6 to ComB10) form a type IV secretion system (T4SS) that is similar to the T4SS originally identified in Agrobacterium tumefaciens and mediates DNA uptake from the environment (17). Several proposed models explain the selective pressure maintaining natural competence in a variety of organisms, including H. pylori: (i) a source of templates for the repair of DNA (18), (ii) a source of energy or nucleotides (19), and (iii) adaptation to a hostile environment through genetic exchange (20). In prior work, we found that natural competence is not required during initial colonization (21), but the role of natural competence during chronic infection is unclear.
H. pylori strains often harbor a second T4SS, the Cag apparatus, and its presence increases the risk of the infected person's developing more severe gastric symptoms (22). The Cag T4SS translocates the effector protein CagA into gastric epithelial cells and induces mammalian cells to secrete proinflammatory cytokines, such as interleukin 8 (IL-8) (23, 24). Although H. pylori harboring the Cag T4S system can successfully infect humans for decades, Cag T4SS activity is thought to limit H. pylori colonization of the mouse (25). This work presents the development of an H. pylori strain for the study of chronic infection in mice. Using this new strain, we found that natural competence promotes stomach colonization during chronic infection.
H. pylori strains were grown on solid horse blood agar (HB) plates containing 4% Columbia agar base (BD Bioscience), 5% defibrinated horse blood (HemoStat Laboratories), 0.2% β-cyclodextrin (Sigma), vancomycin (10 μg ml−1; Sigma), cefsulodin (5 μg ml−1; Sigma), polymyxin B (2.5 U ml−1; Sigma), trimethoprim (5 μg ml−1; Sigma), and amphotericin B (8 μg ml−1; Sigma) at 37°C in a trigas incubator equilibrated with 10% oxygen, 10% carbon dioxide, and 80% nitrogen. For liquid culture, H. pylori was grown in brucella broth (BD Biosciences) containing 10% fetal bovine serum (BB10; HyClone) with shaking in the trigas incubator. For antibiotic resistance marker selection (chloramphenicol acetyltransferase gene [cat]), bacterial media were additionally supplemented with chloramphenicol (15 μg ml−1; Sigma), and for sucrose sensitivity counterselection (levansucrase gene [sacB]), bacterial media were additionally supplemented with sucrose (60 mg ml−1; Fisher). When bacteria were cultured from mouse stomachs, bacitracin (200 μg ml−1; Fisher) was added.
Strains used in this study are listed in Table 1. All strains described are derivatives of NSH57 (2), which is a mouse-adapted derivative of the sequenced human clinical isolate G27 (26). ΔcomB10::cat (HpG27_37) (21) and ΔdprA::cat (HpG27_315) (27) mutants in the NSH57 strain background were described previously and had 70% of the coding sequence (from G27 genomic positions 39460 to 40238) replaced with the cat cassette (comB10::cat) or 39% of the coding sequence (from genomic positions 345017 to 345326) replaced with the cat cassette (dprA::cat). The same mutations were introduced into the MSD132 strain background by natural transformation to generate strains MSD146a, MSD146b, MSD177a, and MSD177b. MSD132cat contains the cat cassette inserted between positions 204322 and 204768 in the G27 genome, a previously characterized neutral locus located between the divergently transcribed genes HPG27_186 and HPG27_187 (28). The MSD132cat insertion construct was created using PCR with splicing by overlapping extension (SOEing) (29) of the products of primers oMSD1131, -1132, -1431, and -1432 (Table 2). MSD132tac was created in the same manner with the cat gene inserted in the reverse orientation. NSH57 bearing the MSD85 allele of cagY (HPG27_486) was constructed by allelic exchange. A cat sacB insertion cassette, conferring chloramphenicol resistance and sucrose sensitivity, was constructed using PCR SOEing of the products of primers oIEC148, -149, -150, and -151 (Table 2) and replaced the region of the NSH57 cagY gene from genomic positions 524989 to 525379 to generate strain IEC42 (ΔcagY). Strain IEC42 was then transformed with genomic DNA isolated from MSD85 and a clone, IEC51 (cagY*), in which the cat sacB cassette at cagY had been replaced with MSD85 DNA was selected based on sucrose resistance and chloramphenicol sensitivity. PCR and Sanger sequencing of strain IEC51 using primers oIEC90, oIEC91 (PCR), and oIEC92 (sequencing) were used to confirm the allelic exchange.
Human AGS cells (ATCC CRL-1739), derived from a patient with gastric adenocarcinoma, were seeded at 5 × 104 cells/well on 24-well plates for 16 h before infection. H. pylori cells from mid-log-phase liquid cultures were added at a multiplicity of infection of 10:1. Supernatants were collected at 0 and 10 h, and release of IL-8 was assayed using an enzyme-linked immunosorbent assay (ELISA) following the manufacturer's protocol (GE Healthcare). At 0 and 10 h, infected cells were washed twice with phosphate-buffered saline (PBS) plus 10 mM sodium orthovanadate and resuspended in SDS loading buffer. CagA tyrosine phosphorylation and total CagA were detected by immunoblotting as described previously (24).
The region of the cagY gene from nucleotides 524297 to 526942 was amplified from genomic DNA of strains NSH57, MSD69, MSD83, MSD85, MSD86, MSD132, and IEC51 (Fig. 1B and Table 1) using PCR primers oIEC90 and oIEC91. The FHCRC Genomics Shared Resource sequenced amplified products using primer oIEC92 with ABI's BigDye terminator cycle sequencing reagent on an ABI 3730xl DNA analyzer (Applied Biosystems). Oligonucleotide primer sequences are listed in Table 2.
Female C57BL/6 or FvB/N mice 24 to 28 days old were obtained from Charles River Laboratories or Jackson Laboratories and certified free of endogenous Helicobacter infection by the vendor. The mice were housed in an Association for the Assessment and Accreditation of Laboratory Animal Care-accredited facility in sterilized microisolator caging and provided with irradiated PMI 5053 rodent chow and acidified, reverse-osmosis-purified water ad libitum. The Institutional Animal Care and Use Committee approved all manipulations. H. pylori infection and recovery from the stomach were performed as described previously (30). To determine numbers of CFU, dilutions of mouse stomach homogenates were plated on the appropriate selective medium. The competitive index was determined by dividing the CFU ratio of the two clones at each time point by the starting CFU ratio, and 20 colonies was considered the limit of detection. For experiments analyzed by quantitative PCR (qPCR), mouse stomachs were disrupted and diluted in brucella broth, and the supernatant was plated on nonselective medium. When >20 but <200 colonies were isolated, the mixture of colonies was grown overnight on a fresh plate to amplify for preparation of genomic DNA. All strains, including ΔcomB10::cat and ΔdprA::cat strains, showed equivalent growth rates in culture (data not shown). When >200 colonies were isolated, H. pylori cells were collected directly into nucleus lysis buffer. Genomic DNA was prepared using the Wizard genomic DNA kit (Promega) by the manufacturer's protocol, except that cells were heated in nucleus lysis buffer at 80°C for 5 min. Stomach tissue for histopathology was fixed in 10% formalin for 3 days. Tissue was prepared and stained using standard procedures by the FHCRC Experimental Histopathology Shared Resource.
qPCR was performed on genomic DNA in a standard reaction using iTAQ SYBR green (Bio-Rad) on an ABI Prism 7900HT sequence detection system (Applied Biosystems). A standard curve was prepared for each oligonucleotide primer set and used to determine the limit of detection and the relative quantity of each clone. Oligonucleotide primers are listed in Table 2. Clones were detected as follows: MSD132cat with oMSD1415 and oMSD1416, MSD132tac with oMSD1415 and oTHS7095, the ΔcomB10::cat mutant with oMSD1416 and oMSD1425, and the ΔdprA::cat mutant with oMSD1416 and oMSD1438. The competitive index is defined as the ratio of mutant to MSD132cat. The colonization load of each strain was obtained by multiplying the relative quantity of each clone by the total colonization load, measured as CFU.
The competitive index in each experiment was compared to 1, the expected value if all strains infected with equal efficiency, using the signed rank test in SAS v.9.2 (SAS Institute Inc.). Colonization loads of mutant compared to the MSD132cat wild type were compared using one-way analysis of variance (ANOVA) on log-transformed data using GraphPad Prism version 6.00 for Windows (GraphPad Software, La Jolla, CA).
Our lab has established strain NSH57 as a model for H. pylori infection in mice (2). This strain derives from the sequenced human clinical isolate G27 and survives for 1 to 4 weeks in the C57BL/6 mouse stomach, but the H. pylori burden drops after 4 weeks of infection and can be cleared in a subset of mice at 8 weeks (Fig. 1A). To develop a model of H. pylori chronic infection, strain NSH57 was introduced into mice, recovered and enumerated 4 to 9 weeks later, and the output clones were reintroduced into mice for another round of selection (Fig. 1B). In the first round, FVB/N mice were used, as they are slightly more permissive to colonization by H. pylori. Colonies recovered from these mice were then selected in C57BL/6 mice five more times by serial passage, resulting in increased colonization load (Fig. 1C).
A single colony (MSD132) was isolated from the final round of selection and colonizes the mouse stomach for at least 7 months (Fig. 1D). After 7 months, H. pylori were found throughout the glandular stomach of all five mice, were mostly associated with the epithelial surface (Fig. 1E), and were rarely found in the glands. No significant inflammatory cell infiltration was observed in the stomach epithelium of these mice (data not shown).
Previous studies have reported that H. pylori strains lacking the Cag T4SS survive longer in the mouse stomach than do unrelated cag+ strains (4, 25). The parent strain NSH57 has an active Cag T4SS (2), as do round 1 and 2 strains, which retained the ability to induce IL-8 (Fig. 2A) while producing a 10-fold increase in the bacterial load after 8 weeks of infection (Fig. 1C). In round 3, strains MSD85 and MSD86 were isolated from the same mouse (Fig. 1B): MSD86 retains the ability to induce IL-8 and to translocate CagA, whereas this capacity is lost in MSD85 (Fig. 2A and andC).C). After 8 weeks in mice, 20-fold more CFU/g tissue were found for MSD85 than MSD86 (data not shown), so MSD85 was chosen for further selection.
CagY is a surface-exposed component of the Cag T4SS and undergoes extensive recombination, usually producing recombinants with the correct open reading frame (31). Given this plasticity, we first checked whether cagY was mutated in MSD132 using the published G27 genome sequence (26) as a reference and found a single base insertion (+T at position 525134). This insertion creates a stop codon corresponding to amino acid 462, of 1,903 amino acids. As CagY is required for Cag T4SS activity (32), this mutation may explain the loss of IL-8 induction observed for MSD85. To verify the importance of this sequence change, we used allelic exchange (33) to place the MSD85 allele of cagY in the NSH57 strain (cagY*). The cagY* strain showed diminished IL-8 production (Fig. 2B), and no detectable CagA tyrosine phosphorylation (which requires T4SS-dependent translocation into the host cell cytosol) during infection of AGS gastric epithelial cells (Fig. 2C), similar to an NSH57 cagY deletion mutant (ΔcagY). Thus, the T insertion mutation is sufficient to abolish Cag T4SS activity, though we cannot rule out the existence of additional cag pathogenicity island (PAI) mutations in strain MSD85 and its derivatives. MSD85 and all strains from rounds 4 to 6 harbor this early stop codon, whereas MSD86 and strains from rounds 1 and 2 do not. Strains from rounds 1 and 2 have significantly improved chronic colonization of the mouse stomach compared to NSH57 (Fig. 1C), showing that the loss of Cag T4SS activity is not the only factor required for murine adaptation.
One-week colonization of the mouse stomach is often analyzed with a competition assay. One of the competing clones, usually the mutant, harbors the cat gene, conferring chloramphenicol resistance, while the other clone is not marked (30). In our chronic colonization model, MSD132 harboring the cat gene at a neutral locus (28) (MSD132cat) is compromised for colonization compared to MSD132 after 8 weeks, with an average competitive index (CIave) of 0.15 (range, 0.021 to 13). While this defect is not significantly different than the expected null result of a CIave of 1 (P = 0.13), we were concerned that cat might have a detrimental effect on mutant colonization, so we explored a new experimental design where both the experimental and control strains carry an insertion of cat into the genome. Primers specific to the unique junction between cat and the genome were designed for each clone, and quantitative PCR was used to assess the competitive index. To validate this method, we first tested MSD132cat against a strain where the cat marker was inserted at the same locus but in the reverse orientation, MSD132tac. After 8 weeks of infection an average competitive index of 0.78 (range, 0.0082 to 16) was observed, and this was not significantly different from 1 (P = 0.49). These results indicate that although expression of cat may confer a slight competitive disadvantage during chronic colonization, comparisons between strains that both contain cat are robust.
With a robust assay in hand, we wanted to investigate persistence phenotypes for genes not required for initial colonization. We chose to test natural competence because our previous study showed that natural competence is not required for colonization of the mouse stomach after 1 week (21), yet most clinical isolates are naturally competent (16). Natural competence was retained during mouse adaptation, as assessed by capacity for natural transformation of MSD132, which we utilized to generate the isogenic strains described here. Consistent with our prior studies using the NSH57 strain background, the ΔcomB10 mutant infected all mice and showed colonization loads similar to those of MSD132cat after 1 week (geometric mean colonization loads, 16,000 CFU of the ΔcomB10 mutant versus 7,800 CFU of MSD132cat per stomach). However, after 8 weeks, two independently generated ΔcomB10 mutant clones showed attenuated colonization compared to MSD132cat (Fig. 3A). As the transformation defect of natural competence mutants limits our ability to introduce exogenous DNA for complementation studies, we tested independent clones to control for possible secondary mutations influencing the observed colonization phenotype. For one clone, no ΔcomB10 mutant could be detected in one mouse, and in seven of the nine remaining mice the colonization load was lower for the ΔcomB10 mutant (geometric mean colonization loads, 5,500 CFU of the ΔcomB10 mutant versus 32,000 CFU of MSD132cat per stomach; P < 0.01). For the second clone, the ΔcomB10 mutant could be detected in all the mice, but all but one mouse showed lower colonization loads of this mutant (geometric mean colonization loads, 7,900 CFU of the ΔcomB10 mutant versus 51,000 CFU of MSD132cat per stomach; P < 0.001). These results suggest that natural competence plays an important role during chronic H. pylori infection.
In order to confirm that natural competence is required for chronic colonization, the role of DprA was tested. DprA is a cytosolic protein that binds DNA cooperatively with RecA, possibly to promote recombination into the genome (34). The ΔdprA mutant takes up DNA through the Com apparatus (35) but fails to recombine it into the genome (27), making it an ideal test of the hypothesis that natural competence may be a source of energy or nucleotides during chronic infection. Similar to ΔcomB10 (21), two independent ΔdprA mutant clones showed no significant defect in competition with wild-type strain MSD132cat after 1 week of infection (Fig. 3B) (geometric mean loads, 8,300 and 12,000 CFU of the ΔdprA strain versus 9,000 and 5,100 CFU of MSD132cat per stomach). At 8 weeks postinfection, 2/11 mice had no detectable ΔdprA cells for one clone and 1/11 had no detectable ΔdprA cells for the other clone. However, the remaining mice showed similar colonization loads for ΔdprA and MSD132cat cells. After 12 weeks, 4/12 mice had no detectable ΔdprA cells and four of the six remaining mice showed lower loads of ΔdprA cells (geometric mean colonization loads, 2,000 CFU of the ΔdprA mutant versus 12,000 CFU of MSD132cat per stomach; P < 0.001) These data show that natural competence promotes chronic colonization in a mouse model and suggest that natural competence has a role distinct from acquiring nucleotides or energy.
In this study, we report the development of a genetically tractable H. pylori strain, MSD132, that colonizes the mouse stomach for at least 28 weeks. This strain has lost the ability to translocate the CagA protein into host cells and stimulate the production of IL-8, a major mediator of the inflammatory response. Loss of Cag T4SS activity has been proposed to contribute to adaptation to the mouse stomach (25). However, during the selection of strain MSD132, significant increases in chronic colonization occurred prior to loss of Cag T4SS activity, suggesting that a variety of mutations accumulate during selection. These mutations may be combined through natural competence, thus contributing to long-term colonization of the mouse stomach.
H. pylori usually colonizes people at an early age, and infection persists throughout the lifetime of its host (36). In humans, gastritis is evident in the vast majority of H. pylori-infected people, even in patients without symptoms (22). In contrast, little inflammation is observed in mice colonized with the MSD132 strain after 7 months. One possibility is that loss of the Cag T4SS in strain MSD132 causes a decreased inflammatory response, resulting in higher colonization loads and longer survival in the mouse stomach. However, this change cannot be the only source of adaptation to the mouse, as a 10-fold increase in the bacterial burden occurred during rounds 1 and 2, prior to the acquisition of this mutation. Although the apparent lack of immune stimulation is a shortcoming of the model, most severe gastric symptoms in humans arise after decades of infection (22), suggesting that there is much to learn about the H. pylori genetic program of chronic colonization, even in the absence of a severe inflammatory reaction.
Our studies in the mouse using multiple independent mutants in multiple genes show that natural competence promotes chronic colonization. The increased time needed to observe a significant decrease in load for the ΔdprA mutant compared to the ΔcomB10 mutant (12 weeks versus 8 weeks) may result from a more severe defect in natural transformation for the ΔcomB10 mutant than for the ΔdprA mutant. While we have been unable to detect natural transformation in vitro for ΔdprA mutants, another group reported limited natural transformation (35). Alternatively, the difference may reflect stochastic differences in the time required to accumulate mutations that are the substrates of genetic exchange.
Natural competence may be maintained in a population as a source of templates for the repair of DNA (18), a source of energy or nucleotides (19), or to allow adaptation to a hostile environment through genetic exchange (20). Our previous studies have shown that cells lacking competence have wild-type sensitivity to DNA-damaging agents, suggesting that natural competence is not a major contributor to DNA repair (21). In this study, we also tested the idea that natural competence may be a source of energy or nucleotides. In cells lacking DprA, DNA enters through the Com apparatus (35) but is destroyed before it can be recombined (27). As ΔdprA mutants show attenuation of chronic colonization similar to that of ΔcomB10 mutants, the uptake of DNA as an energy source must not be the main function of natural competence during stomach colonization.
We favor the hypothesis that natural competence promotes adaptation to a changing environment during chronic colonization. It is clear that mutations arise during long-term colonization of the mouse stomach, as a mixture of cagY+ and cagY mutant strains were isolated from the same mouse in round 3. Beneficial and detrimental mutations may arise from inherent errors in replication or other environmental insults, creating a heterogeneous population of bacterial cells dominated by subpopulations of organisms with varied fitness. Competent organisms then import random fragments of DNA from these subpopulations, which can purge deleterious mutations and may occasionally combine two beneficial mutations into the same genome, thus producing an organism that is more fit than either parent strain.
Further studies using whole-genome sequencing approaches may address the spectrum of additional mutations that promote mouse stomach colonization. In addition, MSD132 will be a useful model strain to define the broad genetic program and the role of natural competence during chronic infection.
This project was supported by Public Health Service grants DK080894 (M.S.D.), AI054423 (N.R.S.), and NRSA T32 007270 from NIGMS (I.E.C.).
The contents of this article are the responsibility of the authors and do not necessarily represent the official views of the NIH.
We thank Sue Knoblaugh for histopathology analysis, Sarah Talarico for statistical analysis, Abigail Mazon for strain construction, and members of the Salama lab for helpful discussions.
Published ahead of print 31 October 2012