Our results are consistent with a non-industrial gut harboring a more resilient ecology with respect to SCFA production, while the industrial gut ecology would be vulnerable to disruption of such pathways, yet the pattern is complex and nuanced. The increased gene abundance in non-industrial populations and overall ratio of acetate:butyrate:propionate generally agrees with previous studies of SCFAs5,12. Similarly, the higher genus-level diversity of bacteria encoding acetate, compared to the other SCFAs, is expected and matches studies that have documented the taxa that encode different SCFAs13,17,19. The overall high richness, high diversity at Hill numbers 1 and 2, and high Gini-Simpson indices found in non-industrial populations at the genus level indicates a highly diverse and evenly distributed production of SCFAs. From an ecological perspective, uneven production of SCFA dominated by a few bacteria in industrial gut microbiomes means lower functional diversity and less redundancy, which ultimately leads to an expectation of decreased resilience. In other words, this study finds that industrial gut microbiomes are at a higher risk of reduced SCFA production because SCFA synthesis is dominated by only a few genera. Given the lower resilience, factors that disrupt the gut ecology are expected to have a more extreme consequence to those living an industrial lifestyle.
While there is an overall trend of increased genus-level functional diversity and redundancy for SCFA production in non-industrial populations, variation exists when examining the SCFAs and populations individually. At the genus-level, the pastoral and rural agricultural populations have increased richness of genera encoding genes involved in acetate and butyrate synthesis, while there is similarity across the different lifestyles for genus richness for propionate encoding taxa. Although hunter-gatherers have similar, or lower, genus richness as industrial populations, they have significantly higher diversity at Hill number orders 1 and 2 and Gini-Simpson indices for butyrate and propionate. Additionally, the pastoralists have a generally similar profile to the industrial populations for acetate and propionate Hill number diversity, as well as similarity to the industrial populations in species PD, which may be linked to this pastoralist group having a diet similar to some industrial populations; namely, a diet high in dairy and red meat consumption, coupled with few dietary sources of plant-derived fibers23. This paints a complex picture. Non-industrial populations have a high diversity of genera encoding butyrate synthesis, and butyrate production is spread more evenly across genera in non-industrial populations than in industrial populations. Hunter-gatherers and rural agriculturalists have significantly greater evenness of propionate production, even though they have fewer number of total genera encoding this SCFA. Finally, the richness and evenness of genera encoding acetate is similar between industrial and non-industrial populations. Ecologically, we would expect the industrial populations to be less resilient for production of butyrate and propionate when faced with a shift in taxonomic composition, while non-industrial populations may be only marginally more resilient for acetate production compared to industrial populations. Intriguingly, SCFA relative abundance does not appear to correlate to resilience profile. Acetate and butyrate are significantly more abundant in non-industrial populations but only butyrate shows much stronger resilience profile for non-industrial populations. Additionally, propionate is slightly more abundant in industrial populations, although not significantly, yet our results indicate greater resilience in non-industrial groups for propionate production. This indicates that measuring only total gene, and/or molar, abundance is not enough to make statements about metabolic processes in the human microbiome; rather, ecological approaches are necessary to understand diversity in functional potential of the human microbiome.
The increased species-level alpha diversity in industrial populations initially runs counter to the genus-level results but the genus and species level results ultimately yield similar interpretations after accounting for ecology and ascertainment bias, as discussed below. The substantially higher species richness in industrial populations is striking; however, the differences in PD between industrial and non-industrial populations are not nearly as extreme. This means that the high species richness in the industrial populations is driven by species that are closely phylogenetically related. Indeed, we observed SCFA producing genera found at high abundance in industrial populations (Bacteroides and Clostridium) to have up to nine species encoding SCFAs, while highly abundant non-industrial genera only have one or two species. Therefore, what first appears to indicate high species-level ecological resilience in SCFA production in the industrial populations is actually the result of closely related species performing the same function. It follows that closely related species may be prone to changes in abundance or even elimination after certain types of ecosystem shift events. For example, narrow-spectrum antibiotics33 and exposure to various xenobiotic compounds that lead to variable bacterial metabolic responses34 are events that can affect a limited range of bacteria and lead to shifts in microbial abundance and metabolic activity. While this result has ecological implications, it is also likely the result of historical trends of microbiology research. Bacterial taxa at high abundance in non-industrial gut microbiomes have not been a focus of microbiological isolation and species identification until recently; therefore, we expect more species to be identified from non-industrial gut microbiomes in the future35. Additionally, classification of bacteria into distinct genera and species is undergoing a revolution in the genomic era36 meaning that the high number of species classified to Bacteroides and Clostridium may ultimately be reclassified to different genera. Nevertheless, the fact that we observe a large jump in species richness, but only a minor increase in species PD, in the industrial gut microbiomes suggests that the high industrial species richness is driven by closely related species and therefore, results in the same interpretation as the genus richness results: diversity is high in non-industrial populations.
Ascertainment bias extends to the databases used to identify taxa and genes: fewer genes were identified in non-industrial populations and a smaller proportion of these genes can be linked back to bacteria at every taxonomic level, in non-industrial gut microbiomes. In some cases, such as butyrate synthesis genes, less than 10% of genes are identified to species for non-industrial populations, while over 50% of such identifications were possible for industrial populations. A decreased ability to identify the genus and species encoding SCFA synthesis genes in non-industrial populations means that the ecological metrics underestimate the true ecological diversity of these genes. Moreover, the drop-off in classification from the genus to the species level was significantly greater in non-industrial populations compared to industrial populations. This drop-off means a much lesser ability to identify species compared to genera in non-industrial populations, which helps explain why species diversity was substantially lower in non-industrial populations. Nevertheless, the statistically significant differences observed at the genus-level send a strong signal of the high functional diversity, and potential resilience, of SCFA synthesis genes in non-industrial gut microbiomes.
The metagenome-wide poor performance in terms of gene identification and classifying SCFA genes to genera and species indicates a bias in reference databases that underrepresents diversity in non-industrial gut microbiomes, which is unsurprising. Bias is expected because the vast majority of human gut microbiome studies have used samples from industrial populations. There is an immense challenge in including non-industrial communities in biomedical research, including recruiting research participants, sustaining longitudinal sampling, building culturally appropriate community relationships, and even securing transport of samples35. This has resulted in comparatively few metagenomic studies of human gut microbiomes from non-industrial settings35. Nevertheless, our data demonstrate the extent of this bias and how it can hinder more in-depth study of human gut microbiome health. Given this sizable ascertainment bias favored industrial populations, the non-industrial populations are likely even more diverse, more resilient, than our databases can sufficiently characterize, making our genus-level results even stronger. Without a serious investment to include such populations, the characterization of microbiomes will remain naive to the ecological breadth of the core, healthy, human gut. Imagine studying forest ecology, with only city parks at your disposal. This has been, overwhelmingly, the analogous practice of human microbiome research.
The relative lack of microbiome studies with non-industrial populations means an underrepresentation of not only metagenomic data and genome annotation but also fewer opportunities for cultivation and validation of novel species of bacteria. This ultimately leads to an inequality in the depth to which researchers can describe microbiome samples from non-industrial communities, compared to industrial microbiomes, as diverse groups of novel taxa may be grouped into a single group of “unknown” or “unclassified” bacteria35. Similarly, an incomplete picture of microbial functional potential means that genes may be misidentified or even unannotated completely. Unknown taxa and misidentified genes may be playing key roles in ecological and metabolic processes but researchers are unable to confidently identify them, let alone make statements about their importance in a microbial ecology35. Recent human gut microbiome metagenome studies from diverse populations will undoubtedly improve database representation but the number of studies and metagenomic samples from non-industrial populations still pales in comparison to industrial gut microbiomes26,35,37,38.
Limitations in annotating the full extent of microbial diversity impacts health research. Recently proposed ‘Microbiota Insufficiency Syndrome (MIS)’2 postulates that, while the microbiome has adapted to industrialization, these adaptations are maladaptive to human health. The decreased phylogenetic diversity and loss of specific taxa (e.g. Prevotellaceae, Succinivibrionaceae, and Spirochaetaceae) observed in industrial gut microbiomes may contribute to the increase in non-communicable chronic diseases found at higher prevalence in industrial populations. The root cause of MIS in industrial populations is undoubtedly multifactorial; however, diet is suggested to play a major role2. This syndrome is compelling and we postulate that this insufficiency precisely rests on the stability of functional capacity. Our findings of decreased resilience in industrial populations, as well as species-level diversity driven by a few closely related species, fits in well with MIS. Low resilience in SCFA production may ultimately manifest itself as altered colonocyte function and/or autoimmune disruptions (both symptoms of MIS) due to a decrease in SCFA bioavailability after a group SCFA-producing bacteria were wiped-out during an ecological shift, such as antibiotic or xenobiotic exposure. Similar to MIS, diet is likely to play an important role in SCFA resilience. The non-industrial populations studied in this paper consume much more fiber than industrial populations, on average3,5,14,25,26, and microbial fermentation of dietary fibers is a major source of SCFAs in the human digestive tract39. A diet poor in dietary fiber means less substrate for microbial fermentation and therefore less SCFA production and also higher competition for that fiber, potentially resulting in competitive exclusion and less microbial diversity. Nevertheless, if we are unable to fully characterize and annotate non-industrial gut microbiomes then we will be unable to paint a complete picture of MIS. Currently, we have confidence that there is a wealth of undiscovered resilience in non-industrial gut microbiomes. Once we describe the extent of this diversity/resilience, through increased sampling and focus on partnerships with research institutes in industrializing countries, we will have a more complete picture of MIS and possibly develop therapeutic approaches to combat non-communicable chronic diseases related to the human gut microbiome.
Improved sampling, metabolic profiling, and annotation will not only improve our understanding of SCFA resilience, but it will also permit more detailed picture microbiome wide resilience. Our work shows the value of focusing on specific SCFA genes, due to their importance in human biology and previously reported variation in SCFA molar abundance between industrial and non-industrial populations31,32; however, future work will undoubtedly add to our findings. One avenue for future work is through analyzing SCFA molar concentrations in fecal samples in a longitudinal setting and comparing these results to predicted SCFA resilience from metagenome panels. Unlike genomic data, where we can infer about SCFA production potential via taxonomic diversity, one-time measures of fecal SCFA molar concentrations will not inform about future resilience because SCFA molar concentrations carry no information about which taxa produce each SCFA. Longitudinal SCFA concentration and metagenomic data from non-industrial populations, or animal models, is necessary to inform about SCFA resilience and production in diverse lifestyles. Another avenue for future work is to focus resilience analysis on other microbiome functions of interest, such as resilience of antibiotic resistance genes and amino acid biosynthetic pathways. These valuable studies would be valuable for comparing microbiome resilience dynamics for different functions, with the caveat that there is sufficient genomic annotation data to yield interpretable results.
Lack of sample diversity is not unique to human microbiome research, as human genetics research has been grappling with this very issue for decades. In 2009, 96% of individuals included in human genome-wide association studies (GWAS) claimed European ancestry, as compared to 78% in 201940. Thus, while there have been improvements, GWAS clearly fail to reflect the breadth of human diversity. Incorporating diverse populations in human genome and microbiome research has the potential to greatly benefit the scientific community’s understanding of human biology and develop treatments that are based on human diversity rather than European-ancestry genetics and microbiomes. A key component of increasing representation in genetics and microbiome studies is that these studies are designed as partnerships with minority and/or indigenous communities in a manner that builds both trust between the community and researchers, as well as facilitates the ability for the sample donors to exercise their rights on how data are treated and shared41.
Source: Ecology - nature.com
