There is growing evidence that our normal resident gut microorganisms, termed commensal microbes, can affect human health. Promoting beneficial commensal microbes through a type of nutritional supplement called a prebiotic is an area of intensive scientific and medical research. However, trying to harness a diet with the desired effect is challenging because the gut microbial community (also known as the microbiome) is highly complex, and because dietary responses are modulated by multiple hereditary and non-hereditary factors. Writing in Nature, Delannoy-Bruno et al.1 fill an essential gap in our mechanistic understanding of diet–microbiome interactions by focusing on dietary fibre, a family of substances of pronounced physiological virtues that are predominantly metabolized by commensal microbes. This provides a sequel to the team’s previous work on the development of microbiome-targeting foods2–6.
To characterize the effect of dietary-fibre supplementation in overweight individuals, Delannoy-Bruno and colleagues used germ-free mice — animals raised and maintained in a sterile environment that are therefore devoid of the usual resident microbes of any sort. The team colonized the gut of each of nine mouse groups with the microbiome of one of nine women classed as obese. The mice were continuously fed a low-fibre, high-fat diet, coupled with periodic fibre supplementation (Fig. 1). Their microbiome was characterized to assess gene content (the level of particular genes present) before, during and after each episode of fibre supplementation. Building on previous work by this team5, which identified fibres that promote the growth of certain intestinal bacteria (those of the genus Bacteroides, which are less prevalent than normal in obese individuals), Delannoy-Bruno et al. chose three types of fibre that were fed sequentially to the mice. Each supplementation cycle was followed by a ‘washout’ period to enable intestinal clearing of the fibres, thereby allowing the authors to discern the effect of every individual type of fibre as distinct from the other fibres.
Human-microbiome-inoculated mice showed pronounced compositional shifts in their gut microbes in response to fibre supplementation. To identify fibre–microbe interactions and subtract irrelevant ‘noise’ in the data arising from normal fluctuations in the microbiome profile, the authors used a feature-reduction approach termed higher-order singular value decomposition. This revealed that exposure of the microbial population to a particular fibre resulted in a greater presence of genes that encode proteins needed for the metabolism of that fibre. For example, consumption of cellulose-containing pea and orange fibre led to a higher representation of genes encoding the β-glucosidase enzymes that hydrolyse such fibre, probably owing to the proliferation of bacteria that can use cellulose as an energy source. Interestingly, in mice inoculated with human microbiomes, similar signatures of fibre-responsive genes became highly abundant through the expansion of various bacterial taxa including, but not limited to, Bacteroides. The bacterial taxa that rose to dominance shared a competitive advantage in their response to the presence of the respective fibre. This highlights that a shift in microbial gene abundance, rather than a shift to a particular community structure in terms of the species present, is a common denominator in the response pattern to dietary interventions across individuals.
The authors then examined whether these findings have relevance to human biology. They characterized the microbiome of 12 overweight or obese study participants before, during and after pea-fibre dietary supplementation. To minimize the effect of microbiome variations stemming from dietary differences between individuals, participants were given meals consisting of a uniform high-saturated-fat, low-fibre diet (equivalent to the type of diet the mice were placed on). The microbial gene responses to fibre supplementation in humans largely resembled those observed in mice.
The authors also tested whether consumption of multiple fibre types results in a greater microbiome shift than was observed upon consumption of pea fibre alone as a supplement. For this study, a group of 14 individuals received a diet supplemented with a combination of two and then four types of fibre. The results indicate that the more types of fibre an individual consumes, the greater the rise in the number of microbial genes present that are involved in fibre metabolism. Changes in fibre-responsive enzymes closely correlated with changes in the plasma levels of proteins that are associated with specific metabolic and immune functions, as revealed by the analysis of blood samples from participants.
The emerging concept of precision nutrition proposes that human dietary responses are specific to food components, and that these components might be tailored to the individual to optimize their dietary response. Such responses are quantifiable and driven by a number of variables, including features of the food components (such as protein, carbohydrate and fat), the individual’s physiological traits, and gut microbiome composition and function. Indeed, it has been suggested that the gut microbiome is a key modulator of physiological food responses and associated differences in such responses between individuals7,8. Current US dietary guidelines recommend that healthy adults ingest at least 14 grams of fibre per 1,000 calories consumed9, with no individualization specified in terms of the fibre type or intake in grams. Delannoy-Bruno and colleagues’ findings provide a framework that paves the way for optimizing an individual’s fibre consumption, based on fibre type and their microbiome fibre-degradation signatures.
The authors’ results underscore the considerable plasticity of the gut microbiome and its amenability to manipulation by prebiotics. Fibre-degradation capacity in individuals with habitual low levels of fibre consumption is not necessarily irreversibly diminished, but rather might be recovered by increasing their fibre consumption through combinatorial fibre selection10–13. Furthermore, the findings by Delannoy-Bruno et al. demonstrate that dietary responses might be mainly determined by the levels of relatively sparse microbiome genes (that become predominant after a dietary intervention), stressing the importance of studying diet–microbiome interactions at the level of genes. Different microbial communities in individuals contributed to these distinct functional capacities, highlighting the possibility that microbiome function (genes), rather than species composition, might correlate with personalized human physiological responses to food.
In some situations, the metabolic and immune-system characteristics of germ-free mice harbouring human microbiomes might affect the validity of using these animals as human surrogates in an experimental context14,15. Nevertheless, the fibre-degrading properties of the mice analysed by Delannoy-Bruno et al. were strikingly similar to those of humans. This suggests that human microbiome colonization in germ-free mice might be a relevant tool for studying the causal drivers of diet–microbiome interactions and their effect on the mammalian host.
Such studies would greatly benefit from the development of computational-analysis tools that can identify temporal trajectories of microbial genes in response to dietary interventions, in both human donors and mice who receive microbial transplants. In this study, the authors used one such tool, higher-order singular value decomposition, which uncovered a profound microbiome gene response to dietary-fibre supplementation despite the modest sample size. Further refinement of such analytical pipelines, by the expansion of microbiome genome-level databases of gene functions and the incorporation of protein-level features, will probably aid further decoding of the contributions of the dietary–microbiome axis to host physiology.
Delannoy-Bruno and colleagues’ findings provide valuable mechanistic insights into the microbial contributions to human dietary responses. This will probably lead to long-term, randomized clinical trials that assess causal links between distinct food ingredients, microbiome modulation and downstream health-related outcomes for humans. Indeed, this research team recently reported2 that a data-driven dietary intervention targeting the microbiome helped to promote growth in undernourished children. The advances made by Delannoy-Bruno et al. bring us closer to the integration of precise microbiome engineering with evidence-based dietary sciences.
Nature 595, 32-34 (2021)
Delannoy-Bruno, O. et al. Nature 595,91–95 (2021).
Chen, R. Y. et al. N. Engl. J. Med. 384, 1517–1528 (2021).
Chen, R. Y. et al. N. Engl. J. Med. 383, 321–333 (2020).
Mostafa, I. et al. BMC Public Health 20, 242 (2020).
Patnode, M. L. et al. Cell 179, 59–73 (2019).
Gehrig, J. L. et al. Science 365, 139 (2019).
Zeevi, D. et al. Cell 163, 1079–1094 (2015).
Berry, S. E. et al. Nature Med. 26, 964–973 (2020).
US Department of Health and Human Services and US Department of Agriculture. 2015–2020 Dietary Guidelines for Americans (HHS/USDA, 2015).
Sonnenburg, E. D. et al. Nature 529, 212–215 (2016).
Eilam, O. et al. mBio 5, e01526-14 (2014).
Healey, G. et al. Br. J. Nutr. 119, 176–189 (2018).
Deehan, E. C. et al. Cell Host Microbe 27, 389–404 (2020).
Al-Asmakh, M. & Zadjali, F. J. Microbiol. Biotechnol. 25, 1583–1588 (2015).
Lundberg, R., Toft, M. F., August, B., Hansen, A. K. & Hansen, C. H. F. Gut Microbes 7, 68–74 (2016).
E.E. is the is the scientific founder of DayTwo and BiomX.