Author response:
Public Reviews:
Reviewer #1 (Public review):
Summary:
The authors are trying to characterize the sources of H. pylori in an island system. They find that, like the humans, the bacteria are admixed, but there is no correlation within the island between human ancestry and bacterial ancestry.
We thank the reviewer for their consideration of our work.
Strengths:
The study has taken particular care to characterize the humans from which isolates were obtained. Thus, it is a particularly convincing demonstration of a bacterial "melting pot".
Weaknesses:
The GWAS is highly confounded by population structure. In particular, there is a large group of strains that lack the cag pathogenicity island, and also differ in frequencies of other genes. So it's not clear that these differences are other than in cag status.
We ran the mentioned GWAS using the Cabo Verdean European lineage as controls group and the European gastric cancer lineage as cases to investigate the genomic differentiation between the Cabo Verdean European lineages and gastric cancer-associated lineages. Our objective was not to replicate previously reported associations or to identify novel H. pylori loci associated with gastric cancer. Therefore, we will revise the manuscript to make our objective clearer and remove statements that suggest direct associations with gastric cancer. In addition, because we employed a linear mixed model that accounts for population structure through the estimation of a genome-wide kinship matrix, we will include the genomic inflation factor. These measures will provide readers with a clearer assessment of the extent of population structure influencing the association analysis.
Figure 1 seems very inconclusive.
We agree that Figure 1 shows overlap between H. pylori-seropositive and seronegative distributions; in addition, its visual interpretation might be also influenced by a number of outlying observations. Nevertheless, the figure illustrates differences in pepsinogen I and II concentrations and pepsinogen I/II ratios between H. pylori-seropositive and seronegative individuals. To improve clarity, we will include statistics in the plot, and we will revise the text to provide a more complete description of the key findings shown in Figure 1. If the reviewers and editor consider that these descriptive data are not central to the main conclusions of the study, we would be happy to move Figure 1 to the Supplementary Material, where it can provide supporting context without detracting from the presentation of the principal results.
Reviewer #2 (Public review):
Summary:
This study investigates the population structure and ancestry of Helicobacter pylori in Cabo Verde, where the human population has mixed West African and European ancestry. The authors combine a population survey, serum markers, bacterial genome analysis, and paired human-bacterial ancestry data. They report high H. pylori seropositivity, several distinct bacterial groups, and limited correlation between human ancestry and bacterial ancestry. They also identify one European-derived bacterial group that appears to have undergone a recent expansion and carries fewer well-known virulence-related genes.
The study is interesting, and the dataset is valuable, especially because population-based H. pylori genomic data from Cabo Verde and West Africa are limited. The results provide useful information on bacterial diversity, historical migration, and host-bacterial ancestry. However, some of the main conclusions are stronger than the evidence currently supports, particularly the claims of host adaptation, increased transmission, and reduced virulence.
We thank the reviewer for their comments and careful consideration of our work.
Strengths:
(1) A major strength is the study population. Participants were recruited from the general population and not only from patients with gastrointestinal disease. This gives a broader view of H. pylori diversity in Cabo Verde than studies based only on hospital patients.
(2) The number of participants tested for H. pylori antibodies is substantial, and the authors also obtained a relatively large number of bacterial genomes. The combination of human and bacterial genomic information is another important strength. This allows the authors to directly examine whether human ancestry is related to the ancestry of the colonising bacteria.
(3) The population genetic analyses are extensive. The authors use several different approaches, and these generally support the existence of African-derived and European-derived bacterial groups in Cabo Verde. The identification of two low-diversity European-derived groups is also interesting and suggests a relatively recent expansion.
(4) The addition of new strains from Ghana and Portugal improves the reference dataset. The results may help future studies of H. pylori population structure in Africa, Europe, Cabo Verde, and populations affected by historical Atlantic migration.
(5) The finding that human ancestry and bacterial ancestry are only weakly related in this population is potentially important. It suggests that the long-term relationship between human and bacterial ancestry may be less stable in recently admixed populations.
Weaknesses:
The main weakness is that several biological conclusions are based on indirect evidence. The genomic results support recent expansion of one bacterial group, but they do not directly show that this expansion was caused by adaptation to local hosts or by increased transmission. Founder effects, population history, geographic clustering, household transmission, or random expansion may also explain the pattern. The wording should therefore be more cautious.
We thank the reviewer for underlining the limitations of inferring the causes of lineage expansion from indirect genomic patterns alone. We will review the manuscript to more carefully reflect the uncertainty surrounding the drivers of this lineage expansion. We will also expand the Discussion to outline analytical approaches that may help distinguish between demographic and selective scenarios in a highly recombining species (where high levels of recombination may have obscured local signatures of selection). Where feasible, we will explore additional analyses of the genomic data to assess the extent to which the observed patterns are consistent with these alternative hypotheses.
The conclusion of reduced virulence is also not fully supported. The expanded lineage often lacks the cag pathogenicity island and carries less virulent forms of vacA, which suggests lower virulence potential. However, this does not prove that the strains cause less gastric damage or lower disease risk. There are no endoscopic or histological data, and serum pepsinogen values are only indirect markers.
We will review our manuscript to adopt a more cautious description of these strains. However, we emphasize that, as show in in Figure 4 – source data3, we could not find strains belonging to this expanded lineage carrying an active cagPAI, or a virulent vacA allele. Our GWAs analyses show high divergence between these strains and gastric cancer strains both at a genome-wide level, and at previously identified virulence genes (as sabA and BabA, in addition to the ones already mentioned above). Finally, serum pepsinogen serum analysis, although indirect, have been shown to compare with to the “gold standard” method, histopathological biopsy microscopy (ex: Telaranta-Keerie et al 2010, 10.3109/00365521.2010.487918; Kitamura et al 2015, 10.1111/jgh.12987; Miftahussurur et al, 2020, 10.1371/journal.pone.0230064; please see more about this in our following comment). Taken together these results provide minimal support for the hypothesis that the expansion of this lineage is driven by increased virulence. We will review the manuscript in order to reflect this.
The description of the study population as having limited gastric inflammation is too strong. Serum pepsinogen measurements are useful for estimating gastric atrophy, but they do not directly measure the degree of histological gastritis. In addition, participants were recruited independently of symptoms, but this does not mean that they were all asymptomatic.
The epidemiological estimate is based on antibody testing. This measures seropositivity and cannot clearly distinguish current from previous infection. Therefore, terms such as active infection or colonisation should be used carefully.
As mentioned above, serum pepsinogen serum analysis has been widely shown to reliably detect both chronic and atrophic gastritis (e.g. Miftahussurur et al, 2020, 10.1371/journal.pone.0230064). We adopted conservative cut-offs for serum pepsinogen values after a careful review of the literature in our analysis. However, we acknowledge that these cut-offs may vary between populations. Therefore, we agree with the reviewer that a more precise assessment would have included a sensitivity and specificity analysis of serum pepsinogen measurements against histopathological biopsy microscopy in a subset of the individuals. Taking this is into consideration, we will review the manuscript to include this discussion and to adopt a more cautious description of these results. Although we acknowledge that seropositive does not distinguish active from past infection in the Discussion, we will bring this discussion into the Results section as well.
The proposed new West-Central African bacterial group is based on a small number of reference strains from Ghana and Nigeria. The result is interesting, but broader sampling from African countries is needed before this group can be considered firmly established.
We thank the reviewer for pointing the need to assess higher African diversity, which is an issue that we also mention in the Discussion. Although the new West-Central African group comprised fewer isolates than the other comparison populations, the clustering pattern is unlikely to be explained solely by sample size because the analysis was based on a normalised chromosome-painting coancestry matrix, which reduces the influence of unequal donor numbers. In our review, we will provide additional analyses to confirm the observed clustering pattern: 1) we will provide fineSTRUCTURE MCMC tree assignments; 2) we will repeat the Chromopainter/ fineSTRUCTURE analyses under random downsampling of the larger groups; 3) as suggested in further reviewer recommendations, we will repeat the Chromopainter/ fineSTRUCTURE analyses after removing the new Ghanaian sequenced strains.
The interpretation related to the trans-Atlantic slave trade is plausible, but the data mainly show patterns consistent with known historical migration. They do not directly demonstrate when or how the bacterial lineages moved.
We thank the reviewer for highlighting this point. The chromosome-painting analysis identifies shared ancestry and gene flow between populations but does not directly estimate the timing of those events. However, several lines of evidence suggest that the majority of the strains have been present in Cabo Verde for an extended period rather than representing recent introductions. First, the isolates were obtained from individuals who, as well as both of their parents, were born in Cabo Verde. Second, Cabo Verdean African strains show an excess of European ancestry relative to their putative parental African populations. And vice-versa: Cabo Verdean European strains exhibit an excess of African ancestry compared with their putative parental European populations. To further investigate this question, as also suggested in reviewer recommendations, we will explore the feasibility of identifying clonal or near-clonal relationships between Cabo Verdean and assess whether dating approaches such as BactDating can provide additional insights into the timescale over which these lineages have diversified and admixed within Cabo Verde.
The gastric cancer comparison may also be affected by bacterial population structure. Differences between the Cabo Verdean lineage and gastric cancer strains may reflect ancestry or lineage differences rather than disease association alone.
We thank the reviewer for highlighting this point. As outlined in our response to Reviewer 1, the primary objective of this analysis was to investigate genomic differentiation between the Cabo Verdean and gastric cancer-associated lineages. Therefore, the strong contribution of ancestry and lineage effects is central to the interpretation of this comparison. We will make this clearer in our final manuscript.
Overall, the study achieves its main aim of describing H. pylori diversity and ancestry in Cabo Verde. The evidence is strong for the population structure and ancestry findings, but less strong for the proposed mechanisms of adaptation, transmission, and reduced disease-causing potential. The work will be useful to the field, but the main conclusions should be stated more carefully.
We thank the reviewers and the editor for their time and effort in assessing the manuscript. We will submit a revised version that carefully addresses these points and incorporates the suggested changes.
