Previews Can Your Microbiome Tell You What to Eat? Jairam K.P. Vanamala,1,2 Rob Knight,3,* and Timothy D. Spector4 1Department of Food Science and Center for Molecular Immunology and Infectious Disease, The Pennsylvania State University, University Park, PA 16802, USA 2The Pennsylvania State Hershey Cancer Institute, Penn State Milton S. Hershey Medical Center, Hershey, PA 17033, USA 3Departments of Pediatrics and Computer Science & Engineering, University of California, San Diego, La Jolla, CA 92093, USA 4Department of Twin Research and Genetic Epidemiology, King’s College London, London WC2R 2LS, UK *Correspondence: [email protected]
Prospects for using the gut microbiome for personalized medicine are substantial since the gut microbiome is known to modulate metabolism and varies substantially among individuals. Zeevi et al. (2015) demonstrate that the gut microbiota can be used to predict individualized blood glucose responses to particular foods, which differ between individuals.
Studies of dietary effects have been plagued by individual variability. Huge amounts of time and money have been invested in looking for the sources of these effects on health or on the human genome. However, we’re all nearly identical in terms of our human genomes— up to 99.9% the same, whereas our microbiomes have over a hundred-fold greater gene count and might only be 10% similar from one person to the next. Consequently, we may have been looking in the wrong place for variation in diet response, as the true source of variation may be in our microbial genes rather than our human genes. Existing studies of diet on the gut microbiome in humans have generally seen modest effects over the short term (Wu et al., 2011), except for truly extreme shifts in diets such as all meat and cheese (David et al., 2014). However, at the population level, differences in long-term diet have a major impact (Wu et al., 2011). In fact, in both the American Gut Project (a crowdfunded volunteer survey of over 7,000 adults) and in the TwinsUK project (of over 2,000 UK adult twin volunteers), diet, and particularly the diversity in the number of kinds of plants, was shown to have a massive impact, comparable to drugs and disease. However, the major challenge with traditional long-term diet interventions is that they take a long time for subjects to see a difference, and many people abandon them while being discouraged during the lag phase. The ability to test reliably for changes in our gut microbes could change this. Zeevi et al. (2015) take a dramatic step forward in addressing this challenge at
the individual rather than the population level and can help to explain our individual differences. They show that by measuring microbiome responses in 800 people using 16S rRNA and shotgun metagenomic profiling to assess taxonomy and function, respectively, and collecting finegrained data on blood glucose levels, they can reveal how the gut microbiome in individuals can make a food healthy for one person and unhealthy for another. Intriguingly, their machine learning models only work well when microbial biomarkers are included, reinforcing the message that we need to include our gut microbes in predictive models of our health (Figure 1). Geneticists have long recognized the importance of host genes in our response to food. Classical twin studies showed that weight gain in response to overfeeding or calorie restriction was highly variable between people but strongly determined by genes (Bouchard et al., 1990). Similarly, our choice and preference for which foods we ingest (e.g., how much we like burgers, garlic, or salads) is also strongly genetically determined (Pallister et al., 2015). What is not yet clear is how these large individual variations could be explained by our microbes—unless our genes determine which microbes we have. Individual microbial species might even secrete metabolites that increase our preference for certain foods, which, in turn, keep their gene pool alive. Zeevi et al. also found some fascinating diet-microbe interactions that can be explained in terms of known interactions and that point the way toward a more
960 Cell Metabolism 22, December 1, 2015 ª2015 Elsevier Inc.
comprehensive view of microbes and health. For example, authors mentioned that the average post-prandial glucose response (PPGR) to rice and potatoes was relatively high, whereas that for dark chocolate was relatively low, in agreement with published data. In food crops such as rice and potatoes, non-nutrient bioactive components such as polyphenols co-occur along with macronutrients (carbohydrates, protein, fats) and micronutrients (vitamins and minerals). Historically, farm-to-fork operations have been focused on maximizing yield, storability, and processing. Although these variables are important, content and composition of polyphenols in the food are rarely considered even in today’s nutritional studies. Emerging evidence suggests that dietary polyphenols alter carbohydrate metabolism in humans in several ways: they attenuate postprandial glycemic responses and glucose uptake fasting hyperglycemia, and improve acute insulin secretion and insulin sensitivity (Hanhineva et al., 2010). A recent clinical trial showing clear long-term health benefits of the Mediterranean diet over low-fat diets (Estruch et al., 2013) may result from higher levels and variety of natural polyphenols in Mediterranean foods. Recent studies indicate that dietary polyphenols contribute to maintenance of gut health by increasing species diversity and stimulating growth of beneficial bacteria (i.e., faecalibacterium, lactobacilli, and bifidobacteria), which are low in type 2 diabetic patients, while inhibiting pathogenic bacteria such as C. perfringens and C. histolyticum. Thus, there is tremendous potential to make a
Refined Carbohydrates Low Polyphenols
Complex Carbohydrates High Phytonutrients
also very high. A range of drugs from acetaminophen to cyclophosphamide have now been shown to have differential toxicity or efficacy depending on the host microbiome. With improved wearable sensors for a range of physiological conditions and perhaps blood, fecal, or urine metabolites, especially coupled to microbial readouts performed in real time, we can imagine a future where routinely measuring and controlling our microbiomes allows us to redefine our relationships to our diets and health.
Gut Bacterial Diversity
REFERENCES Post Prandial Glucose Response
Bouchard, C., Tremblay, A., Despre´s, J.P., Nadeau, A., Lupien, P.J., The´riault, G., Dussault, J., Moorjani, S., Pinault, S., and Fournier, G. (1990). N. Engl. J. Med. 322, 1477–1482.
Risk for Type 2 Diabetes
David, L.A., Maurice, C.F., Carmody, R.N., Gootenberg, D.B., Button, J.E., Wolfe, B.E., Ling, A.V., Devlin, A.S., Varma, Y., Fischbach, M.A., et al. (2014). Nature 505, 559–563.
Figure 1. Modulation of Gut Bacterial Diversity with Food Approach to Prevent Type 2 Diabetes Food patterns that provide both complex carbohydrates and greater levels of phytonutrients such as polyphenols can increase gut bacterial diversity and reduce postprandial glucose response. Such communities may protect against type 2 diabetes. The personalized nutrition approach pioneered by Zeevi et al. may help us understand which of these types of features will apply to everyone and which will need to be applied to specific individuals.
difference in population-level health by selecting staple crops that are naturally high in polyphenolic content and establishing farm-to-fork operations that retain these key compounds and their healthbenefiting properties (Vanamala, 2015). Indeed, purple-fleshed potato cultivars contain approximately 2–3 times greater levels of polyphenols compared to their white potato counterparts. Recently, a human clinical study showed that the low glycemic index of colored potatoes could be related to their polyphenol content (Ramdath et al., 2014). Thus, it is critical to consider not only the fiber and micronutrient content of foods but also the phytonutrient content to improve our dietary recommendations. Last year, using twins and 16S rRNA gene sequencing to read out the microbiota, we showed that human gene variation is associated with differences in
abundance of gut microbes (Goodrich et al., 2014). The increased correlations in identical twins and the resulting heritabilities were modest, but will likely be shown to be greater when more comprehensive metagenomic sequencing is used. Nearly 100 gene variants have been associated with body weight, but although this trait is 70% heritable, human gene variants explain < 5% of the variability. Some of this missing heritability could be due to interactions with our microbes. Studies of humans in the future must therefore account for both host and microbe genetics. Using identical twins discordant for diets or microbial species may be an ideal way of mirroring mouse diet experiments in the most important model—humans. Although the study by Zeevi et al. (2015) focuses on diet, the microbial and metabolic variability in response to drugs is
Estruch, R., Ros, E., Salas-Salvado´, J., Covas, M.I., Corella, D., Aro´s, F., Go´mez-Gracia, E., Ruiz-Gutie´rrez, V., Fiol, M., Lapetra, J., et al.; PREDIMED Study Investigators (2013). N. Engl. J. Med. 368, 1279–1290. Goodrich, J.K., Waters, J.L., Poole, A.C., Sutter, J.L., Koren, O., Blekhman, R., Beaumont, M., Van Treuren, W., Knight, R., Bell, J.T., et al. (2014). Cell 159, 789–799. Hanhineva, K., To¨rro¨nen, R., Bondia-Pons, I., Pekkinen, J., Kolehmainen, M., Mykka¨nen, H., and Poutanen, K. (2010). Int. J. Mol. Sci. 11, 1365– 1402. Pallister, T., Sharafi, M., Lachance, G., Pirastu, N., Mohney, R.P., MacGregor, A., Feskens, E.J., Duffy, V., Spector, T.D., and Menni, C. (2015). Twin Res. Hum. Genet. Published online September 28, 2015. http://dx.doi.org/10.1017/ thg.2015.69. Ramdath, D.D., Padhi, E., Hawke, A., Sivaramalingam, T., and Tsao, R. (2014). Food Funct. 5, 909–915. Vanamala, J. (2015). Crit. Rev. Food Sci. Nutr., 0, http://dx.doi.org/10.1080/10408398.2015.1028023. Wu, G.D., Chen, J., Hoffmann, C., Bittinger, K., Chen, Y.Y., Keilbaugh, S.A., Bewtra, M., Knights, D., Walters, W.A., Knight, R., et al. (2011). Science 334, 105–108. Zeevi, D., Korem, T., Zmora, N., Israeli, D., Rothschild, D., Weinberger, A., Ben-Yacov, O., Lador, D., Avnit-Sagi, T., Lotan-Pompan, M., et al. (2015). Cell 163, 1079–1094.
Cell Metabolism 22, December 1, 2015 ª2015 Elsevier Inc. 961