The United States Food and Drug Administration (FDA) is responsible for “protecting the public health by assuring the safety, efficacy, and security of human…drugs…[and] our nation’s food supply.” Yet these efforts have largely ignored the unintended consequences for the health of our human microbiome or in turn, the ability of the microbiome to modulate the impact of these interventions on host pathophysiology.

Members of the Turnbaugh Lab at the University of California, San Francisco (UCSF) are hard at work correcting this massive oversight. We focus on the human gut microbiome, due to its remarkable evolutionary, genetic, and enzymatic diversity. Our approach is interdisciplinary in nature and spans multiple scales, from molecules to organisms to ecosystems, with the long-term goal of revealing clinically relevant cellular and molecular mechanisms through which the microbiome shapes nutrition and pharmacology. To accomplish these goals we emphasise collaborative science, which is amplified by the UCSF team spirit combined with the diverse personal and scientific backgrounds of our lab members. Inspired by the microbiome itself, we prioritise working together as obligate but mutualistic symbionts; parasites and commensals need not apply.

In the area of nutrition, we have become fascinated by the changes in host and microbial metabolism that occur in states of carbohydrate restriction. Paired studies in humans and mice identified consistent impacts of low-carbohydrate diets that induce host ketogenesis on the gut microbiota. Surprisingly, we discovered that a major host-derived ketone body β-hydroxybutyrate (βHB) was sufficient to alter the gut microbiota leading to a decrease in TH17-inducing bacteria2, emphasising the broad impacts of ketogenesis on the host and microbiome. Furthermore, we found that elevated dietary arginine (i.e. protein) can rescue against gut bacterial TH17 activation and colitis in mice. In addition to changing the interaction between the immune system and the microbiome, diet can have important consequences for microbial interactions. For example, we recently showed that caloric restriction can perturb the human gut microbiota, allowing for the expansion of the enteric pathogen Clostridioides difficile and contributing to weight loss in mice.

Our studies of pharmacology have also revealed intriguing links between diet, the microbiome, and host immunity. For reasons we still do not fully understand, dietary arginine prevents the gut Actinobacterium Eggerthella lenta from inactivating the cardiac drug digoxin in vitro and in gnotobiotic mice. In humans, we demonstrated that both the pre-treatment and post-treatment microbiome predicts the anti-inflammatory effects of the drug methotrexate, used to treat rheumatoid arthritis. These results may also be diet-dependent given that methotrexate is an “anti-metabolite” drug, as are the fluoropyrimidines that are a mainstay of cancer therapy and similarly exert off-target effects on the gut microbiota.

A major goal of our current work is to test the physiological relevance of gut bacterial pathways for drug metabolism in mice. Colonisation of mice with isogenic E. coli strains that differ in a bacterial operon for the inactivation of the anti-cancer drug 5-fluorouracil led to a marked and significant impairment in drug efficacy. While these results were predictable from our in vitro experiments, the transition to mice has also prompted us to more broadly consider the alternative mechanisms through which gut bacteria could shape drug metabolism and disposition. For example, we found that azo dye “excipients” commonly added to processed food and drugs inhibit drug transporters in the intestine; bacterial metabolism rescues this inhibition, increasing drug absorption. These results emphasise the importance of understanding how the gut microbiota shapes expression and function of host drug transporters, both in the gastrointestinal tract and in other body sites.

Finally, we have realised that our methods for manipulating the human microbiome are woefully inadequate, especially given the broad and unpredictable effects of both food and drugs on complex microbial communities. The exquisite specificity of bacteriophage provides a potential solution, which could be enhanced by sequence-specific targeting with CRISPR-Cas or related tools. As a proof-of-concept, we were able to achieve strain- and gene-specific targeting of E. coli within the mouse gut, providing a valuable tool for adding or removing genes of interest. This success was built on a vast history of work on E. coli and the M13 bacteriophage, which we currently lack for most phage-bacterial pairs of interest. As a first step, we characterised the type I-C CRISPR-Cas system of E. lenta, which coupled to our expanding E. lenta plasmid and strain library has set the stage for phage-delivered genome editing in this species.

Taken together, these efforts emphasise the underappreciated role for the microbiome in nutrition and pharmacology. While data-driven recommendations remain elusive, we remain optimistic about the translational relevance of these studies, especially given the rapid expansion in the number of microbiome labs that continues to fuel innovation. In the future, the real FDA may harness this information to not only decide which food or drug to recommend, but also to facilitate methods to precisely manipulate the microbiome to enhance and expand the available dietary and pharmaceutical options for patients.