Diversification of the vacAs1m1 and vacAs2m2 Strains of Helicobacter pylori in Meriones unguiculatus

The bacterium Helicobacter pylori exhibits great genetic diversity, and the pathogenic roles of its virulence factors have been widely studied. However, the evolutionary dynamics of H. pylori strains during stomach colonization are not well-characterized. Here, we analyzed the microevolutionary dynamics of the toxigenic strain vacAs1m1, the non-toxigenic strain vacAs2m2, and a combination of both strains in an animal model over time. Meriones unguiculatus were inoculated with the following bacteria: group 1-toxigenic strain vacAs1m1/cagA+/cagE+/babA2+; ST181, group 2-non-toxigenic strain vacAs2m2/cagA+/cagE+/babA2+; ST2901, and group 3-both strains. The gerbils were euthanized at different time points (3, 6, 12, and 18 months). In group 1, genetic alterations were observed at 6 and 12 months. With the combination of both strains, group 3 also exhibited genetic alterations at 3 and 18 months; moreover, a chimera, vacA m1-m2, was detected. Additionally, four new sequence types (STs) were reported in the PubMLST database for H. pylori. Synonymous and non-synonymous mutations were analyzed and associated with alterations in amino acids. Microevolutionary analysis of the STs (PHYLOViZ) identified in each group revealed many mutational changes in the toxigenic (vacAs1m1) and non-toxigenic (vacAs2m2) strains. Phylogenetic assessments (eBURST) did not reveal clonal complexes. Our findings indicate that the toxigenic strain, vacAs1m1, and a combination of toxigenic and non-toxigenic strains acquired genetic material by recombination. The allelic combination, vacAs2m1, displayed the best adaptation in the animal model over time, and a chimera, m1-m2, was also identified, which confirmed previous reports.

The bacterium Helicobacter pylori exhibits great genetic diversity, and the pathogenic roles of its virulence factors have been widely studied. However, the evolutionary dynamics of H. pylori strains during stomach colonization are not well-characterized. Here, we analyzed the microevolutionary dynamics of the toxigenic strain vacAs1m1, the non-toxigenic strain vacAs2m2, and a combination of both strains in an animal model over time. Meriones unguiculatus were inoculated with the following bacteria: group 1-toxigenic strain vacAs1m1/cagA+/cagE+/babA2+; ST181, group 2-nontoxigenic strain vacAs2m2/cagA+/cagE+/babA2+; ST2901, and group 3-both strains. The gerbils were euthanized at different time points (3, 6, 12, and 18 months). In group 1, genetic alterations were observed at 6 and 12 months. With the combination of both strains, group 3 also exhibited genetic alterations at 3 and 18 months; moreover, a chimera, vacA m1-m2, was detected. Additionally, four new sequence types (STs) were reported in the PubMLST database for H. pylori. Synonymous and nonsynonymous mutations were analyzed and associated with alterations in amino acids. Microevolutionary analysis of the STs (PHYLOViZ) identified in each group revealed many mutational changes in the toxigenic (vacAs1m1) and non-toxigenic (vacAs2m2) strains. Phylogenetic assessments (eBURST) did not reveal clonal complexes. Our findings indicate that the toxigenic strain, vacAs1m1, and a combination of toxigenic and nontoxigenic strains acquired genetic material by recombination. The allelic combination, vacAs2m1, displayed the best adaptation in the animal model over time, and a chimera, m1-m2, was also identified, which confirmed previous reports.

INTRODUCTION
Helicobacter pylori, a well-known member of the human microbiota, has a global distribution that is related to the migration of Homo sapiens over the past 60,000 years (Morelli et al., 2010;Moodley et al., 2012). H. pylori is a Gram-negative spiral bacterium that is associated with the development of peptic ulcers as well as some types of gastric lymphomas and gastric adenocarcinomas in humans (Mbulaiteye et al., 2009;Testerman and Morris, 2014).
The evolution of distinct genetic prototypes in H. pylori is linked to different human ethnic groups worldwide, supporting the presence of genetic mechanisms that have permitted rapid adaptation in human populations (Linz et al., 2007). High mutation rates and frequent inter-strain exchanges of genetic material that occur during infection are responsible for the extreme variation and genetic diversity among H. pylori strains (Blaser and Atherton, 2004;Suerbaum and Josenhans, 2007). Additionally, transmission mainly occurs through direct humanto-human contact, and single or multiple strains of H. pylori can colonize and recolonize a host to increase its variability (Brown, 2000;Frenck and Clemens, 2003). Consequently, mutations, inter-strain genetic exchange and the mode of transmission appear to account for the capacity of H. pylori to colonize different habitats in the stomach, and its indirect and direct interactions with the human host trigger different selective pressures that regulate the presence of strains in this changing habitat (Raymond et al., 2004;Kivi et al., 2007;Costa et al., 2009;Secka et al., 2011). Researchers have hypothesized that selective pressures that determine the presence of H. pylori in the stomach operate on three types of bacterial genes (e.g., genes that affect intrabacterial mutations, DNA uptake, repair and recombination; genes that favor bacteria-bacteria interactions; and genes that influence bacterial properties, such as adherence and immune responses that modulate interactions with the host) (Gangwer et al., 2010;Tanih et al., 2011).
The dynamics of H. pylori genotypes during stomach colonization are unknown because the successful establishment of these strains is an inadvertent process. A model that potentially explains the genotypic evolution of H. pylori in its human host assumes that strains with genotype vacA+/cagA+/babA+ are at a higher "fitness peak" (Montecucco and Rappuoli, 2001). These virulence genes encode proteins (i.e., VacA, CagA, and BabA) that help the bacteria to adhere and persist in the gastric epithelium by modifying and altering apical and cell junctions (Wroblewski et al., 2010); i.e., the vacAs1m1 allelic combination is capable of producing the VacA toxin (which induces vacuolation of gastric epithelial cells), whereas the vacAs2m2 allelic combination produces low amounts or none of the VacA toxin (Atherton et al., 1995;Letley and Atherton, 2000). Consequently, inactivation of any of these factors can shift the fitness of the strains. H. pylori can live in the stomach of an individual for many years, so it is possible that strains may emerge with vacA+/−, cag+/−, or babA+/− genotypes, and strains with the genotype cag-or cag+ can be isolated from the same patient. However, these strains likely cannot survive long because of the high recombination rates observed for this bacterium. Thus, only those bacteria that are efficient over long durations and engage in person-to-person transmission are thought to govern the evolution of H. pylori (Montecucco and Rappuoli, 2001;Prouzet-Mauleon et al., 2005).
Genetic alterations that are produced during the microevolution of H. pylori have not been studied because they can only be detected during the transition phase, i.e., after the passage of an in vitro strain (culture) to an in vivo (animal) setting or during colonization of a host that is not infected (Ferrero and Jenks, 2001). Animals that are infected (e.g., Rhesus monkeys, mice, and gerbils) with strains of known genotypes provide experimental models (Peek, 2008;Behrens et al., 2013;Linz et al., 2014) that can be used to follow the evolution of these strains in vivo, from the initial inoculation until the definitive establishment of the strain (Morelli et al., 2010;Linz et al., 2014). In the present study, we analyzed the evolutionary dynamics of the toxigenic strain, vacAs1m1, and the non-toxigenic strain, vacAs2m2, separately and together in an animal model over time.

Animal Model
The 8-week-old Mongolian gerbils (Meriones unguiculatus Hsd:MON, Harlan Teklad, Madison, WI, USA) used in this study were housed under specific pathogen-free conditions in plastic metabolic cages to prevent coprophagy under standard laboratory conditions (i.e., room temperature, 23 ± 2 • C; relative humidity 40-60%; and a 12-h light-dark cycle). Free access to a standard diet (special rodent food; Harlan Teklad, Madison, WI, USA) and sterilized tap water were provided. The Ethics, Biosafety and Scientific committees at the Health Institute approved the experiment.
The three groups of gerbils included five animals each one, ensuring that three animal would present H. pylori infection (Velazquez-Guadarrama et al., 2007). Gerbils were inoculated intragastrically with 500 mL NaHCO 3 (0.2 M), and 1 h later with a bacterial suspension of different genotypes of H. pylori [6 × 10 8 colony forming units (CFUs)/mL]. For 1 week, group 1 was inoculated with the toxigenic strain, vacAs1m1/cagA+/cagE+/babA2+; group 2 received the nontoxigenic strain, vacAs2m2/cagA+/cagE+/babA2+; and group 3 received both the toxigenic strain, vacAs1m1, and the nontoxigenic strain, vacAs2m2. The gerbils were fasted 18 h prior to the first inoculation until the end of the fifth inoculation. The H. pylori strains used in this study included reference strain 26695 (positive control for the vacAs1m1 genotype) and clinical strain 174F2 (positive control for the vacAs2m2 genotype). Control animals received saline alone. The gerbils were euthanized at 3, 6, 12, and 18 months by cervical dislocation under anesthesia to harvest the stomach. Assuming that the H. pylori generation time is at least 3 h (Jiang and Doyle, 2000;Joo et al., 2010), the generation numbers (G) achieved at these months were 654 G, 1309 G, 2617 G, and 3926 G, respectively. The stomach was dissected along the greater curvature and washed with phosphate-buffered saline (PBS pH 7.4, 0.01 M). It was then divided longitudinally into parts and macerated with Brucella broth (BD BBL) in a final volume of 200 µL. Next, 10 µL was used for a urease test, and 10 µL (STOCK) was inoculated in Casman agar plates (BD BBL, Sparks, MD, USA) with or without antibiotics (3 mg/mL vancomycin, 5 mg/mL trimeptoprim, and 2 mg/mL amphotericin B). Additionally, 170 µL was used for serial dilutions (1:10, 1:100, 1:1,000, and 1:10,000) on Casman agar plates with and without antibiotics. The plates were grown under microaerophilic conditions (5% O 2 , 5% CO 2 , 85% N 2 , and 10% humidity) at 37 • C for 7-14 days. The control group was compared with the infected groups.

Isolation and Identification of H. pylori
Bacterial isolation was performed using 10 presumptive colonies of H. pylori for each dilution of the different generations. Bacterial identification was based on colony morphology, Gram staining, and tests for urease, catalase, and oxidase. H. pylori was stored at −70 • C in 1.5 mL Brucella broth (BD BBL) supplemented with 10% fetal bovine serum and 25% glycerol.

Detection of Virulence Genes by PCR
Genomic DNA was extracted from a section of the stomach of gerbils, and colonies of H. pylori were isolated using a Wizard Genomic DNA Purification kit (Promega, Madison, WI, USA) according to the manufacturer's instructions with slight modifications of the incubation times. The DNA was quantified in an Epoch Microplate Spectrophotometer (BioTek, software Gen5 TM , Winooski, VT, USA), and the DNA integrity was evaluated by electrophoresis using 1% agarose gels. H. pylori was identified in gastric tissue and isolates based on the presence of the glmM gene (Smith et al., 2004). The vacA (s1, s2, m1, and m2), cagA, cagE, and babA2 genes were amplified by PCR using the conditions described by Atherton et al. (1995), Mizushima et al. (2001), and Kauser et al. (2004). Amplification was performed in a reaction volume of 25 µL Master Mix (Promega) containing 100 ng bacterial DNA, 2.5 mM MgCl 2 , 10 mM dNTPs, 2 U Taq DNA polymerase, 20 pmol each primer and nucleasefree water in a Thermo Hybrid thermal cycler (PCR Express, Emeryville, CA, USA). The PCR products were separated by electrophoresis using 1% agarose gels at 80 V, followed by staining with ethidium bromide and imaging under UV illumination (ChemiDoc Transilluminator, Bio-Rad, Hercules, CA, USA). DNA from reference strain 26695 was included as a positive control.

Bioinformatics and Phylogenetic Analyses
The sequences of the seven loci were aligned using ClustalX v2 (Larkin et al., 2007), edited with Seaview v4.2.5 (Gouy et al., 2010) and FinchTV V.1.4.0 Software (Geospiza, Inc.), and compared with those of all known alleles from H. pylori deposited in the PubMLST database 1 . Each strain was defined based on the presence of alleles of the seven genes (allelic profile); every allelic profile was defined as a sequence type (ST) (Feil et al., 2004;Vázquez and Berrón, 2004). To establish the open reading frame of each protein, the nucleotide sequences of the different STs from each housekeeping gene were translated into amino acid sequences using the translate tool in ExPASy 2 . To determine the presence of synonymous and non-synonymous mutations in different positions of the seven housekeeping genes, we used DnaSP v5.10 3 (Librado and Rozas, 2009).
The clonal relationship among strains of H. pylori was determined using the PHYLOViZ platform 4 . PHYLOViZ infers evolutionary descent patterns among allelic profiles using the goeBURST algorithm and a full Minimal Spanning Tree (MST)-like approach (Francisco et al., 2009). The phylogenetic relationships among the strains was determined using the eBURST algorithm 5 , which subdivided large multilocus sequence typing (MLST) datasets into non-overlapping groups of related STs or clonal complexes to discern the location of the most parsimonious isolates within groups or clonal complexes from the predicted founder (Feil et al., 2004). In addition, the eBURST algorithm explores the diversification of strains and can provide evidence for the emergence of clones of particular clinical relevance.

RESULTS
Helicobacter pylori was identified in all infected groups by endpoint PCR. However, H. pylori strains were isolated from groups 1 and 3. The genotypes derived from the toxigenic strain (vacAs1m1/cagA+/cagE+/ babA2+; ST181) in group 1 exhibited genetic alterations at 6 and 12 months (1309 and 2617 G, respectively), and we also observed the emergence of new clones [vacAs1m1/cagA+/cagE−/babA2−; ST2902 (1309 G) and    (Figure 1). Among the seven housekeeping genes that were analyzed, five genes in ST2902 and ST2093, atpA, efp, mutY, ppa and trpC, had more synonymous mutations. However, the proportions of each gene at 6 and 12 months were similar. The trpC gene exhibited the most variation (Figure 2).
The housekeeping genes of the STs in group 3 exhibited many mutational changes in both the toxigenic and the non-toxigenic 6 http://www.ncbi.nlm.nih.gov/genbank strains ( Figure 5). The number of mutated genes and the number of mutations within each gene were different in the STs. However, the genotype identified at 3 months did not exhibit mutational changes in the housekeeping genes, i.e., the ST was similar to that of the non-toxigenic strain, and the number of synonymous and non-synonymous mutations was similar to that of the toxigenic strain (ST181). All non-synonymous mutations occurred outside of the active site and other functionally important sites of the proteins (e.g., signature motif and substrate-binding domains); some synonymous mutations (15.29%) were identified in these regions.
The phylogenetic analysis of 1993 isolates deposited in the PubMLST database for H. pylori along with those isolates of H. pylori obtained from the animal model in the present study exhibited an overview of clonal complexes (Figure 6). Clusters of related isolates and individual unlinked STs are shown as a singletree eBURST, establishing the definition of category zero for seven shared alleles. ST2904 and ST2905 displayed a 'double-link' that featured six alleles in common. Moreover, ST2901, ST2902, and ST2903 are individual unlinked STs.
FIGURE 5 | Microevolution of STs was identified in group 3 (toxigenic strain vacAs1m1 and non-toxigenic strain vacAs2m2), as defined by PHYLOViZ (goeBURST algorithm) for strains of H. pylori isolated from the M. unguiculatus animal model. ST181 and ST2901 correspond to the reference 26695 (vacAs1m1 toxigenic strain) and clinical 172F2 (vacAs2m2 non-toxigenic strain) strains of H. pylori, respectively. Each line represents a different allele with mutational changes. The number of non-synonymous and synonymous mutations are indicated in parentheses. ST2901 was identified at 3 months * (chimera, Figure 3); the data are presented for both original strains. The reference strain 26695 (ST181) donated the vacAm1 allele, which originated the chimera and clinical strain 174F2 (ST2901) alleles of the housekeeping genes. The trpC gene (ST2904 and ST2905) was the only housekeeping gene with mutational change.
pathogens, the host appears to impose a selective pressure that drives variation within the bacterium (Thompson et al., 2004). In the present study, we infected M. unguiculatus with the following H. pylori strains with known virulence genotypes and STs: reference strain 26695 (vacAs1m1/cagA+cagE+bab2A+; ST181) and clinical strain 172F2 (vacAs2m2/cagA+cagE+bab2A+; ST2901). Our findings revealed genetic alterations of the introduced genotypes throughout the course of infection in an animal model. Group 1 exhibited genetic alterations in the cagA, cagE, and babA2 genes (Figure 1). Several previous studies have observed similar changes or alterations. Solnick et al. (2004) studied genetic alterations in H. pylori J166 and found replacements of babA with babB, suggesting that babA undergoes selective pressure early during the course of infection and that changes in babA might represent a crucial adaptation to the host stomach environment. In addition, Linz et al. (2014) reported that alterations within the babA promoter could occur as early as 1 week post-infection, also they reported mutations in the cag-PAI genes in J166 strain after 2 and 6 months in a macaque model, where the function of cag-PAI was apparently abolished, implying adaptation by mutation and recombination during early stages of the infection.
More than a decade ago, several authors described the potential for genetic rearrangements and recombination events among H. pylori strains during chronic infections in different parts of the stomach (Blaser and Berg, 2001;Blaser and Atherton, 2004). We identified different genotypes at 6 and 12 months post-infection in the group 1 (1309 and 2617 G, respectively) that exhibited alterations in the cagA, cagE, and babA2 genes. In these genes, the incorporation of genetic material by recombination between strains (e.g., cagA-cagEand babA2-) was evident. The presence of these genotypes in group 1 after 6 and 12 months strongly suggested that mechanisms of recombination or mutation had occurred over long periods of colonization in our animal model, resulting in the acquisition of this genetic material (Figure 1). The same authors suggested that such genetic variation can lead to the development of different strains, genotypes or subclones within the same host. Our data documented mutational changes in housekeeping genes and virulence genotypes; thereby, we FIGURE 6 | Population "snapshot" of H. pylori and STs found in strains of H. pylori that were isolated from the M. unguiculatus animal model. Clusters of related isolates and individual unlinked STs found in the MLST database for H. pylori are presented as a single-tree eBURST to define category zero of seven shared alleles. Unions link isolates that correspond to clonal complexes. Primary founders (blue) are located in the center of the group, and founders of subgroups are shown in yellow, as shown for ST2904 and ST2905 (red circle). ST181, ST2901, ST2902, and ST2903 are marked; the labels for the other STs (http://pubmlst.org//helicobacter/) have been removed for clarity.
identified the following new STs: ST2902, ST2903, ST2904, and ST2905 derived from ST181 (group 1) and ST181 and ST2901 (group 3). In group 2, which was infected with strain ST2901 alone, the infection was identified at different periods of time by PCR. However, isolation any H. pylori colonies was not possible, and we still do not know why this was the case. Generally, genotype vacAs2m2 is reported less frequently in infected patients, probably because of its need for more nutrients for its own development in vitro or might suggest to be a genotype unsuitable to infect, it can enter to a viable but nonculturable stage (coccoid forms). The coccoid forms of H. pylori are less virulent, less likely to colonize and induce inflammation (Mazaheri Assadi et al., 2015). Meanwhile, Pagliaccia et al. (1998) observed m1 and m2 alleles of the VacA cytotoxin, which can recognize different receptors on gastric epithelial cells in humans; however, similar findings have not been previously published for gerbils.
Global studies have shown that high degrees of polymorphism in housekeeping genes are associated with changes in the third nucleotide position of codons, and many of the observed variation in STs are synonymous changes (Achtman et al., 1999). We also observed a greater number of synonymous mutations, in accord with other studies (Achtman et al., 1999;Morelli et al., 2010;Secka et al., 2011). In our study, the trpC gene (ST2903) was the most variable and uniquely introduced stop codon. However, these changes were unlikely to be deleterious for the micro-organisms because they were isolated of animal model. Martincorena et al. (2012) have suggested an evolutionary optimization of the mutation rate to reduce the risk of deleterious mutations.
Phylogenetic studies have indicated a clear separation between sequences of the middle region; the m2 sequence contains an insertion of 21-25 amino acids that was not present in the m1 sequence (van Doorn et al., 1998;Gangwer et al., 2007). In the present study, we observed that group 3 (toxigenic strain vacAs1m1; ST181 and non-toxigenic strain vacAs2m2; ST2901) ( Figure 3) had a vacAs2m1-m2/cagA+cagE+babA2+ genotype with the presence of a natural chimera of the middle region of vacA at 3 months (654 G). To confirm the presence of the m1-m2 alleles of vacA in the chimera, each allele present in the four strains was sequenced. The identity percentages were 99% (m1) and 97% (m2), which were similar to those described by Tanih et al. (2011) who reported identity percentages of 87-99 and 89-98% for m1 and m2, respectively, in their chimeras. Our chimeras were aligned with other chimeras that have been previously reported (Figure 4), revealing the region proposed by Pan et al. (1998), in which recombination occurs between the m1 and m2 alleles in a region with limited homology (14 nt). Natural chimeras in strains of H. pylori have rarely been reported, suggesting that H. pylori strains with intact m1 or m2 of vacA provide favorable functional properties and, therefore, exhibit a selective advantage compared with strains containing m1-m2 chimera sequences (Ji et al., 2000). In the present work, we observed instability of the vacA m1-m2 chimera; moreover, the allelic combination, vacAs2m1, exhibited the best adaptation in the animal model over time. Our results demonstrate the combination of genotypes among the vacAs1m1 and vacAs2m2 strains in an animal model. Studies have shown that allelic combination of vacAs2m1 causes less damage to the host (Atherton et al., 1995). Once H. pylori is established in the stomach, it may or may not evolve to a vacAs1m1 genotype; however, this genotype is the most frequently noted in adult patients and is associated with duodenal and gastric ulcers and gastric cancer (Atherton et al., 1995(Atherton et al., , 1997Miehlke et al., 2000).
The high rate of mutations between STs results in high genetic diversity that reflects a long evolutionary history of various strains of H. pylori. In this study, ST2904 and ST2905 exhibited seven different housekeeping genes derived from the initial ST2901. Thus, we identified four new STs that were reported in the database PubMLST of H. pylori. Among these STs, three had alleles that had been previously reported in the database: ST2901 with the mutY1504 allele reported in Ireland, and ST2904 and ST2905 each with an atpA1708 allele reported in Brazil. The phylogenetic analysis conducted with 1993 isolates (PubMLST H. pylori; Figure 6) provided results that were consistent with previous studies (Suerbaum et al., 1998;Feil and Spratt, 2001;Hanage et al., 2006;Turner et al., 2007), indicating that H. pylori forms a non-clonal population, presents a high mutation rate that generates a large number of alleles and that there is a high rate of recombination among these alleles.
The microevolutionary history of H. pylori infection in humans reveals remarkable genetic diversity within this bacterium, which is mainly generated by point mutations and recombination (intragenic or intergenic). The high variability of H. pylori is thought to maximize its ability to adapt to the changing environment of the host gastric habitat, consequently facilitating chronic colonization. This study provides evidence for processes of recombination between genotypes, the emergence of new clones and patterns of evolutionary non-clonal descent among H. pylori strains obtained from an animal model. Our findings suggest that the recombination process in H. pylori in the host results from the adaptation of the bacterium to the host.

AUTHOR CONTRIBUTIONS
SM-El, GZ, and NV-G conceived and planned the study. SM-El and NA-R performed experiments and generated the database. RL maintained and provided care for the animal model. SM-El, GZ, and NV-G analyzed and interpreted data. PV-M helped with the animal model. SM-Es and FA-H reviewed and corrected the manuscript. The manuscript was prepared by SM-El, GZ, and NV-G. All authors revised and agreed on the final version of the manuscript.