Strain genetic diversity

The bacterial strains were assigned to populations according to their multilocus DNA sequences: those from African hosts yielded only hpAfrica1, those from Spanish yielded hpEurope and those from Koreans yielded hspEAsia ( Table 1 ). However, Huitoto and Guahibo Amerindians yielded both hspAmerind and hpEurope strains, and Mestizos yielded hpEurope and hpAfrica1, but not hspAmerind. The least and most diverse strains of H. pylori populations were hspAmerind and hpEurope, respectively ( Figure 2A ). Nonetheless, when grouping the strains by host, strain diversity in Amerindians increased to the levels found in Spanish and Mestizo hosts ( Figure 2B ), consistently with the circulation of mixed strains ( Table 1 ) and with the remarkable mosaicism reflected in the multilocus sequences ( Figure 3 ). As stated previously [7], [26], the ancestry patterns of modern hpEurope strains revealed extensive recombination between the two ancestral populations ancestral Europe1 and ancestral Europe2. Spanish H. pylori ( Figure 3 ) further include components from ancestral hpAfrica1, which possibly reflects the role Africa has played in shaping the Spanish modern human gene pool. Traces of African bacterial ancestry were also detected in strains from Mestizos and Amerindians. The observed mosaicism reflects extensive recombination and results in lower genetic structure.

Open in a separate window

Figure 2

Pairwise genetic distances between H. pylori strains grouped by bacterial population (2A), or according to their human host (2B).

Differences in pairwise distances among strains in Fig. 2A were significant (Kruskal-Wallis test, p<2.2×10⁻¹⁶, Wilcoxon and Bonferroni; p<10⁻¹⁴) with a decreasing order hpEurope>hpAfrica1>hpEastAsia>hspAmerind. When grouped by the human host from which each strain was isolated (Fig 2B), strain diversity in Amerindians was as high as in Spanish and Mestizos (with no significant differences among them; Wilcoxon Pairwise comparison p>0.7), with a decreasing order Spaniards=Amerindians=Mestizos>Africans>Koreans. The waist of the dress-like box is the median with the waist side openings indicating the 95% interval for the median; the top represents the 3rd quartile and the bottom the 1st quartile. The interval in dashed lines represents a maximum of 1.5× interquartile range and the open circles are outliers. Permutation tests confirmed group differences in strain diversity.

Open in a separate window

Figure 3

Mosaic structure of the multilocus H. pylori sequences in representative strains.

The ancestral source of each polymorphic nucleotide is shown by a vertical line for each of the seven gene fragments in the multilocus analysis of 10 representative strains from each group (see the legend below Mestizo strains). Individual nucleotides were derived from ancestral Europe1 (grey), ancestral Europe2 (green), ancestral Africa1 (blue) and ancestral EastAsia (yellow). Nucleotides not assigned with >50% probability to any one population are indicated by white lines. African and European components can be observed in hpEurope strains from Spaniards and Mestizos, as well as in hpAfrica1 from Mestizos, while African hpAfrica1 strains and Amerindian hspAmerind strains tested were largely homogeneous.

Genetic divergence and convergence

Although most of the data analyzed in this study was already available, they had previously not been analyzed in terms of the associations between genetic diversity of humans and their indigenous microbial population. There are many genetic studies that confirm that human intrapopulation genetic diversity decreases with geographic distance from Africa, and that the Amerindians suffered a genetic bottleneck [12]–[17]. The finding that a low-diversity human population was associated with a low-diversity bacterium could be due to increased genetic drift (on a populations of low effective size), and/or selection by low diversity blood groups on the H. pylori population.

Genetic isolation and drift as well as selection may explain the radiation of humans into different groups, and the concomitant divergence of H. pylori into geographical types, whereby both host and bacteria show patterns of isolation by distance [26]. However, the modern world is increasingly bringing genetic influx into each of the human groups, both at the human and microbiome level. Indigenous microbes that once diverged during most of human history are currently in the process of converging, which erase phylogeographic signals. How this process is occurring is important because some human-evolved bacteria are relevant to human health (for example H. pylori is implicated in gastric diseases), and changes in the dynamics of its coevolution with humans might lead to changes in disease patterns.

Genetic diversity and fitness

There might be greater local genetic variation in parasite populations relative to their hosts. In this case, parasite populations might increase their genetic structure by locally adapting to their hosts [28]. Consistently, mosaic strains would be expected to be genetically more diverse and more generalist. This might explain why hpAfrica1 strains from Mestizos are more diverse than those from African hosts (Kruskal-Wallis p<10−14; Figure S1 ). Recombination obviously requires coexistence of different strains in a single stomach, which we indeed have found to be frequent in patients from Venezuela [29].

Diversity optimizes niche partitioning. In the gastric context, strain mosaicism (intra-genomic diversity) and strain diversity (inter-genomic variants) is likely to be a key factor in the hpEurope strains host range expansion (in both Mestizos and Amerindians). Meanwhile, hspAmerind strains seem to lack the diversity that would be required to survive in the high host variability brought by Mestizos. Displacement of hspAmerind by hpEurope strains could occur by two different means: 1) Strain competition during colonization, in which “generalist” strains will have better chance to colonize diverse niches than “specialist” strains [as in the case of O binder Amerindian strains described above, which would have less success in colonizing non-O blood group Mestizos [22]]; 2) Strain subversion by transformation, in which DNA from hpEurope strains is taken up and recombinant strains become increasingly European. The high component of European ancestry in the mosaic strains from Mestizos and Amerindians, and the relatively low Amerindian component in Mestizos ( Figure 3 ) support this second hypothesis.

The consumption of antibiotics and other drugs (such as proton pump inhibitors that reduce gastric acidity) might also be currently affecting the bacterial populations by increasing selection pressure in favor of particular variants, thus shaping the genome of indigenous microbes in modern humans. Modern human admixture results not only in increased human diversity but also in increased microbial diversity of the human microbiome. In the long term, however, one can predict that maximum genetic distances will be succeeded by homogenization of the genomic structure in strain variants (all mosaic or recombinant strains) and humans (all Mestizos) and decreased within population genetic distances brought by further recombination and lack of isolation, as illustrated in Figure 4 . Since no influx of genetic variation is expected from outside the global village, low genetic flow in both microbes and hosts will eventually result in decreased human and microbiome diversity.

Open in a separate window

Figure 4

Development of inter-strain genetic diversity over time as two populations recombine.

In the context of our hypothesis, a low diversity H. pylori population (P1) arose from co-evolution with the isolated Amerindian host population. With the introduction of new H. pylori strains (P2) the new population formed (P3) is more diverse than any of the source populations alone. Selection acts and strains recombine with the consequent homogenization of the population (P3′). The cycle is repeated when new populations (P4) arrive. Given time and isolation (no gene flow), population diversity will be reduced (P6). Based on their mosaic structure and high genetic distances, it seems that current H. pylori from Amerindians and Mestizos are in one of the intermediate states (P3′ or P5′). Arrows indicate introduction of new populations.

To our knowledge, Wirth et al. [30] were the first to analyze both human DNA and H. pylori sequences from 50 patients in Ladakh in northern India. Our results support the need of a study with a true large scale population-based approach involving human and microbe samples from the same hosts in order to better measure the extent in which human populations shape the genomes of H. pylori. As new technology optimizes cost and time of sequencing, there can be new studies performed in individual Homo sapiens and his individual microbiome, using more genes. These studies will be crucial to understand the coevolution of humans and their microbes.

Human mtDNA sequences

A total of 1,148 sequences of human mtDNA hypervariable segment I were retrieved from Genbank from published work (Table 2), from humans close to the hosts from which H. pylori strains had been cultured: non-Bantu Africans, Spanish, Koreans, Amerindians (Guahibo, Huitoto, Inuit) and South American Mestizos.

We are grateful to J. Bertrand Petit for providing Spanish sequences of human mtDNA. We thank Gladys Ramos for her help with Figure 3.