Repeated out-of-Africa expansions of Helicobacter pylori driven by replacement of deleterious mutations

Abstract

Helicobacter pylori lives in the human stomach and has a population structure resembling that of its host. However, H. pylori from Europe and the Middle East trace substantially more ancestry from modern African populations than the humans that carry them. Here, we use a collection of Afro-Eurasian H. pylori genomes to show that this African ancestry is due to at least three distinct admixture events. H. pylori from East Asia, which have undergone little admixture, have accumulated many more non-synonymous mutations than African strains. European and Middle Eastern bacteria have elevated African ancestry at the sites of these mutations, implying selection to remove them during admixture. Simulations show that population fitness can be restored after bottlenecks by migration and subsequent admixture of small numbers of bacteria from non-bottlenecked populations. We conclude that recent spread of African DNA has been driven by deleterious mutations accumulated during the original out-of-Africa bottleneck.

Fig. 1. Ancestry and migration history of hpEurope isolates.

a Painting profiles of hpEurope isolates and their putative ancestral populations from Africa and Asia showing proportion of each genome (horizontal bar) painted by each of five ancestral donor populations (circles). hpEurope isolates are grouped by country of isolation, with bars to the right indicating the H. pylori population each strain is assigned to. The representative isolates from each donor population are grouped by population, with countries of isolation listed for each group. b Genetic drift profiles for hpEurope subpopulations, shown separately for each ancestry component. c Schematic summarizing the migration and admixture history of the hpEurope subpopulations. Source data are provided as a Source Data file. The map in a is from https://commons.wikimedia.org/wiki/File:World_map_blank_gmt.png.

Fig. 2. Within- and between-population divergence.

a Within population dN/dS (y axis) plotted against dS (x axis). Small dots show pairwise distances; larger solid dots indicate population means. b dN/dS, calculated to the H. acinonychis outgroup, plotted against dS for isolates (semi opaque points) and populations (solid points), excluding hpAfrica2 isolates. The triangle indicates the genome from Ötzi. Source data are provided as a Source Data file.

Fig. 3. Genetic ancestry of hpEurope subpopulations as a function of mutation score.

a Average ancestry in chromosome painting analyses plotted against mutation score (mutation frequency in African populations minus mutation frequency in Asian populations) in bins of 0.02. Regression lines were calculated separately for positive and negative mutation scores from the unbinned data. b Mean of the regression slopes pseudovalues with 95% confidence intervals estimated using a gene-by-gene jackknife for excess Asian and excess African mutations, respectively. For each point, the average was calculated over 840 occurrences. Source data are provided as a Source Data file.

Fig. 4. Simulations of populations under bottlenecks and admixture.

a Within population dN/dS and b dN/dS calculated to the ancestor measured at generation 8000. Semi opaque points show pairwise distances; solid points indicate population means. c Population average fitness shown during the generations of the simulation, with bottleneck starting at generation 5000 (horizontal line) and admixture at generation 8000 (arrow). d Proportion of bottleneck and non-bottleneck ancestry in the bottleneck population, for the generations after the beginning of admixture. Sites with unknown ancestry are shown in gray. Each arrow corresponds to the migration of one strain. e Ancestry painting in a sample of genomes from the bottleneck population, in the generations subsequent to admixture. Only the first 20 kb of each genome is shown. f Average bottleneck population ancestry, at generation 8300 plotted against mutation score (frequency in the non-bottleneck population minus the frequency in the bottleneck population before the admixture begins). Error bars show the standard error on the mean (for each bin, from −1 to 1, the sample size is 1369, 344, 453, 639, 1264, 914, 666, 381, 241, 251, 1839). Source data are provided in the github repository https://github.com/EliseTourrette/Hpylori/tree/main/HpEurope (DOI: 10.5281/zenodo.7130003).