Drugging the gut microbiome (Nature Biotechnology)

March 2015

Using conventional drug discovery and novel synthetic biology approaches, some investigators and companies are mining our resident microbes and their metabolites for targets in small-molecule drug programs. Ken Garber reports.

In January, Janssen Biotech, a Johnson & Johnson company, announced that it would license a gut microbe cocktail developed by Boston-based Vedanta Biosciences, for treating intestinal bowel disease (IBD). The deal, potentially worth $241 million to Vedanta, is the largest yet announced in the microbiome space. But Janssen is also developing gut microbiota–derived therapies on its own—not whole organisms, but small molecules. “Drug development could have many different flavors,” says Scott Plevy, a Janssen vice president and its IBD group leader. “It might be, if we’re lucky, that an active metabolite or even a gene product of a microbe could be a drug in and of itself.” Janssen is also identifying targets for its own novel compounds. “Through characterization of host receptors that interact with these metabolites, we could make a small molecule to these host receptors,” Plevy says. “We’ve been studying several.”

Small-molecule drug discovery is the microbiome’s new frontier. Last May, in a wide-ranging talk on the pharmaceutical industry, Tadataka Yamada, chief medical and scientific officer at Takeda Pharmaceuticals in Osaka, pointed out that the human microbiome is a massive untapped source of drug targets. “The human genome gave us 20,000 targets,” he said. With the microbiome, “we’re talking about several million microbial targets.” This is not hyperbole—based on data from the US Human Microbiome Project, the total number of bacterial genes in the microbiome exceeds the number of human genes by at least two orders of magnitude. “This is an area that the pharmaceutical industry has to get into,” Yamada says.

Obviously, not every gene product will be a worthwhile target, and no one knows whether drugging the microbiome will lead to useful therapies. But recent work suggests that no part of human physiology is untouched by commensal microbes. Proper immune system maturation relies on gut microbiota, and solid links to inflammatory disease, diabetes, cardiovascular disease and neurological disorders are continuously emerging. “We don’t want to oversell it, but everywhere we look there is some connection,” says Michael Dority, program administrator for the Host Microbiome Initiative at the University of Michigan in Ann Arbor.

But the sheer scale of the microbiotic superorganism presents a massive combinatorial problem for biologists trying to understand it and companies seeking to drug it. Each healthy person harbors at least 160 species of gut bacteria, according to data from the European Union and China’s Metagenomics of the Human Intestinal Tract (MetaHIT) project. Species act not just singly but in concert, so the number of possible microbial combinations that could be altering human biology is unfathomably vast, even before considering strains within species.

But, in the past few years, researchers have used elegant microbiome comparisons to identify several microbial biological pathways that appear important for human health and that contain actionable drug targets or potential drugs. In the process, they’ve established a discovery blueprint for the rest of the field. “It’s still very early for developing therapeutics from the microbiome,” says Michael Fischbach, a chemist at the University of California, San Francisco (UCSF). “But it’s moving very quickly.”

Targeting cardiovascular disease

Biotech companies in the microbiome space have so far focused mostly on fecal transplants , and the popular press has fixated on these stool enemas and “crapsules.” But pharma is seeking small-molecule drugs, in part because systemic delivery will be unnecessary, thus vastly simplifying drug design and avoiding toxicity. “Those medicines don’t have to have any bioavailability at all,” says Yamada. If you make drugs that don’t ever enter the body but stay in the gut cavity to have their effects, it’s probably an ideal drug from the standpoint of safety.”

“In the future we will drug the microbiome, not with antibiotics but with nonlethal drugs,” says Stan Hazen, chair of cellular and molecular medicine at the Cleveland Clinic Lerner Research Institute. Hazen is doing his best to make this happen, by targeting the gut microbiome to treat cardiovascular disease.

Hazen’s studies are now classics in the microbiota field, but he didn’t start out looking at microbes. An unbiased metabolomic survey of heart disease patient plasma revealed that the presence of the obscure small-molecule trimethylamine N-oxide (TMAO) was strongly associated with risk of heart attack, stroke and death. Suspecting that TMAO or a precursor might be a microbial cleavage product, Hazen found that, in animals, microbiota are required for the production of circulating TMAO originating from food containing the abundant dietary lipid lecithin1. High TMAO  in turn boosted plaque development in arteries, by generating cholesterol-packed foamy macrophages that stick to blood vessels.

Hazen later showed that gut microbes also act on l-carnitine, an abundant nutrient in red meat, to generate the TMAO precursor trimethylamine (TMA), which is then processed by liver enzymes into TMAO, increasing the risk of cardiac events. “Incredibly elegant, beautiful work,” says Sarkis Mazmanian, a Caltech microbiologist.

Hazen later showed prospectively that high plasma TMAO levels conferred a 2.5 times greater risk of future (three-year) heart attack, stroke or death in patients undergoing cardiac catheterization, and that stable heart failure patients who had high levels of both TMAO and the established prognostic marker brain natriuretic peptide have a 50% five-year mortality rate.

Hazen is working on a TMAO diagnostic, but that’s not his main goal. “We need to develop a ‘statin’ for this pathway, and that’s really what we’re trying to do,” says Hazen. In 2014, three different microbial enzymes systems were identified that convert lecithin or carnitine (or both) to TMA. Chemists in Hazen’s laboratory are looking for small-molecule inhibitors of all of these enzymes. “We’re actively pursuing a lot of candidates already,” says Hazen, who has received research funds from Takeda, Abbott and four other companies.

In theory, probiotics (preparations of specific microbes) or prebiotics (dietary substances that promote specific microbe growth) could be used to build a gut microbial community that does not convert lecithin or carnitine to TMA. But, Hazen cautions, the in vivo effects of probiotics and prebiotics are unpredictable because they alter the entire microbial community. “It’s a huge black box,” he says. “That’s why I actually think that the drugs approach is a more scientifically predictable and tractable approach.”

Nor are fecal transplants the answer for most chronic human diseases, Hazen says. This, despite a recent trial showing fecal transplants to be superior to vancomycin alone for treating Clostridium difficile enterocolitis2. “I just don’t see fecal transplants, even if they’re in a pill form, as something that’s going to take off,” Hazen says. “Except for a life-threatening disease like C. difficile infection, where you’ve got a real sudden dysbiosis, where there’s an abnormal pathogenic composition that needs to be changed. But for chronic degenerative diseases, we’re going to need to be much more precise in how we modulate the pathways.”

Small molecules, from the drug company point of view, also make financial sense. The microbiome “is a space where people have really tried hard to understand a business model that will work,” says Yamada. “The easiest business model is a small-molecule business model.”

A discovery blueprint

Mazmanian’s group did its own metabolomic survey in a mouse model of autism, knowing that people with autism spectrum disorder often have intestinal problems, and that dysbiosis of gut microbiota has been reported in people with autism. The microbial metabolite that jumped out of their survey, 4-ethylphenylsulfate (4EPS), then was fed to normal mice, which developed anxiety-like behavior similar to that seen in the autistic mice3. Mazmanian’s group is now trying to understand what 4EPS targets in the host—perhaps a receptor. “It’s very early work, and at this stage we don’t even have a candidate,” Mazmanian says.

Mazmanian thinks this general approach could generate many such mechanistic links between gut microbiota and disease states, and eventually druggable targets. (His laboratory is now beginning work on Parkinson’s disease, another condition associated with intestinal problems.) Automated metabolomic systems and a growing catalog of standards have made metabolite identification much more feasible and productive in the last few years, Mazmanian says. And metabolomics is likely to be more revealing than sequencing, “because the metabolites are likely the business end of many microbes,” he says. “There is only so much you can infer from sequence data.”

Second Genome, a S. San Francisco biotech, takes the metabolomics approach of Hazen and of Mazmanian a step further. The company’s discovery platform compares healthy versus disease microbiota not just in terms of metabolites but also at the level of gene, transcript and gene product, as well as host gene expression. “To be able to understand really well what the bacteria are doing, you do need three to four different views,” says Second Genome CSO Karim Dabbagh.

Second Genome’s human samples are intestinal mucosal biopsies, not stool. Stool can only approximate what’s happening at the key interface between microbe and host—the half-millimeter-wide intestinal mucus layer that shields the body from the teeming mass of microbes in the lumen of the colon. (Disruption of the mucus layer leads to inflammatory bowel disease, or IBD.) A mucosal biopsy “has the fingerprint of both the host expression and the bacterial community at the site of activity,” says Second Genome CEO Peter DiLaura.

Second Genome is collaborating with Janssen Biotech on microbiome drug discovery in IBD. Janssen brings IBD and immunology expertise to the table, whereas Second Genome applies its bioinformatics platform.

Janssen’s Plevy cites polysaccharide A (PSA) as a potential active metabolite drug. In 2005, Mazmanian and Dennis Kasper, a microbiologist at Harvard Medical School, showed that PSA, from the common gut microbe Bacteroides fragilis, directed the proper maturation of the mouse immune system4, and three years later they showed that PSA protects animals from experimental colitis. PSA also protects mice from experimental autoimmune encephalomyelitis, an established model of multiple sclerosis.

Mazmanian has shown that PSA works by inducing functional T-regulatory cells, or Tregs, the CD4 (helper T cell) lineage that shuts down effector T cells and helps prevent autoimmunity. He and Kasper founded a company, Symbiotix Biotherapies, to develop PSA as an oral therapeutic, and the company recently partnered with a undisclosed major pharma to move PSA forward in IBD.

Metabolite of the moment

Besides their induction of Tregs through PSA, gut microbes promote Tregs on a more general level. In 2011, Andrew MacPherson at the University of Bern showed that altered Schaedler flora (a standard mix of important mouse commensal bacteria) induce Tregs, and in 2013 Kenya Honda, an immunologist at RIKEN in Yokohama, gave a 17-strain cocktail of human clostridia to mice, inducing functional Tregs and protecting against inflammatory bowel disease (IBD)5. The biotech company Vedanta licensed an optimized version of Honda’s cocktail to Janssen Biotech in January for development against IBD. And since August 2013, three different groups have reported that short-chain fatty acids—common bacterial metabolites generated from dietary fiber—induce functional Tregs.

Those Treg studies are feeding a surge of interest in short-chain fatty acids (SCFAs). “This is an area that has really heated up in the last year,” says Mazmanian. The fermentation of complex polysaccharides (starches and fiber from plants) in the colon generates SCFAs, which are then absorbed into the bloodstream. Although SCFAs have long been known to maintain healthy intestinal walls and have wider metabolic effects that protect against diabetes and heart disease, these microbial metabolites, some argue, are the key to overall human health. In February 2014, two papers reported that SCFAs (from fiber) downregulate inflammation6,7. Might a lack of dietary fiber lead directly to autoimmune and inflammatory diseases?

That’s the view of Justin Sonnenburg, a Stanford microbiologist. “A reduction in short-chain fatty acid production… is what happens when you get rid of dietary fiber, and [leads to] increasing inflammatory responses of the host immune system,” he says. “And it’s this simmering state of inflammation that the Western immune system exists in that’s really the cause of all the diseases that we’ve been talking about.”

Sonnenburg argues that our gut microbiota evolved to cope with the fiber-rich diet of Neolithic subsistence farmers, and that the modern, relatively fiber-deficient Western diet is out of balance with the gut microbial community. Less fiber means less substrate for SCFA-producing bacteria. For example, a 2010 study comparing the gut microbiota of children in a rural Burkina Faso village with those of European children found that the important SCFAs propionate and butyrate were almost four times more abundant in Burkina Faso fecal samples than in Europeans’. “You can just imagine that if you get rid of these important regulatory molecules, and the immune system becomes a little bit proinflammatory across a large population, you’re going to see increases in things like cancer, heart disease, allergies, asthma and inflammatory bowel disease,” Sonnenburg says.

But how best to deliver butyrate and other SCFAs to take advantage of their anti-inflammatory benefits? Mazmanian takes two fiber pills a day to feed his microbes, which convert the fiber to SCFAs. “I don’t think it’s a viable business plan to develop short-chain fatty acids as pharmaceuticals,” says Mazmanian. “You can’t get composition of matter [patent claims], obviously.” But other delivery methods may be both medically and economically viable. For example, Vedanta’s microbial cocktail includes butyrate-producing bacteria. Such cocktails “are going to be a very powerful type of agent for delivering microbial-produced small molecules to the exact niche in the body where they’re meant to act, in the right amounts, and with the appropriate pharmacokinetics,” writes Vedanta COO Bernat Olle in an email. “They will be able to do that in ways that will be hard to mimic with oral small-molecule formulations.”

Such small-molecule, SCFA-mimicking drugs are coming, though. Janssen is now characterizing host receptors that interact with microbial metabolites as potential drug targets. Such metabolites include SCFAs, which bind to the G protein–coupled receptors GPR41, GPR43 and GPR109A. These receptors are expressed on Tregs and other immune cells and have a direct anti-inflammatory effect in rodents. Potent small-molecule  GPR109A agonists have been described by Merck and GlaxoSmithKline, and Amgen has reported GPR43 agonists. Plevy cautions that simply targeting one receptor might not be enough to be effective in inflammatory diseases. In humans, individual SCFAs “are likely interacting with multiple receptors,” he says. Still, he says, a SCFA receptor “would be an extremely logical target” in IBD.

Plevy won’t specify which targets Janssen is going after. Nor will Second Genome disclose the target of a lead compound that the biotech company is now independently testing in phase 1 in IBD. The space is becoming competitive as other pharmas move aggressively into microbiome small-molecule therapeutics. “There’s a lot of companies out there taking similar approaches,” says Plevy. “I wouldn’t be surprised to see a flurry of press releases from many large pharma companies working with smaller biotechs in this space.”

Janssen is also working with Second Genome and the University of Michigan to evaluate the microbiome as a marker of disease susceptibility or activity in genetically susceptible hosts. “Any change in the environment which correlates with increasing incidence of human diseases, I think it’s a viable hypothesis that we’re going to find changes in the microbiome that may either underlie or exacerbate those conditions,” Plevy says.

Treasure hunt

Meanwhile, some researchers and companies are beginning to systematically explore the vast numbers of small molecules produced by gut bacteria, not just to plumb mechanism but to identify drug targets and to obtain natural product drug leads. “The microbiota is producing hundreds or thousands of little drug-like compounds,” says Sonnenburg. “We don’t [know] what their chemical structures are and…we don’t know what pathways and receptors they’re binding to in our biology.” The few known examples suggest an enormous potential reservoir of bioactive small molecules. PSA protects from inflammatory disease; the Escherichia coli metabolite colabactin, which induces double-strand DNA breaks, contributes to colon cancer; and the gut microbial metabolite deoxycholic acid indirectly promotes liver cancer8. “There are probably 1 × 1030 bioactive small molecules in the biosphere,” said University of British Columbia microbiologist Julian Davies at a 2013 talk. “These are compounds we can look at. They’re unlimited. We just have to find them, we just have to identify them, based on what they can do for us.”

Natural product drug discovery from gut microbes, in principle, should be more productive than from traditional sources like soil or water bacteria, because gut bacteria already inhabit people. These microbes “have evolved in an environment where they would likely produce molecules that have learned to interact with, or evolved to interact with, their host,” says Mazmanian. To date, only one microbiota-derived molecule has made it to market: Redwood Pharmaceuticals’ linaclotide, a homolog of the enterotoxin peptides produced by E. coli that cause diarrhea. The US Food and Drug Administration approved it in 2012 for some chronic constipation indications.

One powerful approach to finding gut microbial metabolites is to search bacterial genomes for biosynthetic gene clusters (BCGs). Bacterial genes are organized into operons, or co-regulated gene clusters, including those dedicated to synthesizing small-molecule metabolites. UCSF’s Michael Fischbach has recently reported the results of the first systematic survey of BCGs in the human microbiome9. “What excited us was the prospect that these molecules might be some of the most important agents of action by which microbes influence the host, and each other,” Fischbach says. Using metagenomic sequence data from the Human Microbiome Project, Fischbach’s group identified more than 3,000 BCGs commonly found in healthy individuals. To their surprise, they found BCGs for thiopeptides, a class of antibiotics, widely distributed across microbiota genomes—the first time human commensal bacteria had been shown to produce drug-like molecules to protect from pathogens.

The BCG survey was just a first step. Fischbach’s group purified one of the thiopeptides and determined its chemical structure, a very laborious process for this particular molecule. A more systematic approach, he says, is needed. “You can’t do [discovery] in a one-off way,” Fischbach says. “Because then you have to get pretty lucky. But if you do it… thousands at a time, then you’re likely to find interesting molecules.”

Fischbach is now pursuing a “refactoring” synthetic biology approach to small-molecule discovery. Basically, he recodes the native organism’s BCG so that it can be cloned into a laboratory-friendly host like E. coli or Bacillus subtilis, and so its BCG genes will be expressed there. That’s necessary because native organisms are difficult to culture, or hard to do genetics in, or don’t express certain metabolites under the artificial conditions of laboratory culture. “All those things could get you into trouble and could hang up a systematic analysis,” he says. “The only way to really systematize it is to use synthetic biology to turn it into a pipeline, by taking the native host out of the equation.”

There have been many other refactoring efforts, mostly futile. “I would say those fail at least 95% of the time,” says Fischbach. “Nobody really understands why.” But Fischbach says he and collaborator Chris Voigt of the University of California, Berkeley, are making progress. “Right now a lot of work has to be done just to make it medium-throughput,” Fischbach says. “I’m quite hopeful that it will be made higher throughput.” The US National Institutes of Health and Defense Advanced Research Projects Agency both have major refactoring initiatives. And Fischbach is working with “a couple of companies,” but declines to name them.

Sonnenburg says the microbiota refactoring work is setting “the foundation of an entire field that’s just taking off.” He and Fischbach have an even more expansive vision—the application of synthetic biology not just to BCGs but to other specialized bacterial gene clusters, and the employment of such bioengineered cells in the body as diagnostic sensors and eventually drug delivery systems. “Microbes are like little mass specs [spectrometers] in our gut; they can sense every little nuance because their survival depends on it,” Sonnenburg says. “We know that our Bacteroides species that we study is exquisitely sensitive to inflammation, because it has to defend against the oxygen radicals that are produced. And so we can just tap into that genetic circuit and say, don’t just defend yourself but also secrete [anti-inflammatory cytokine interleukin] IL-10 when you sense inflammation.”

Therapeutic testing of such genetically engineered bacteria to enhance the gut microbiota is coming. ActoGeniX, based in Ghent, Belgium, already is clinically testing live Lactococcus lactis engineered to secrete therapeutic proteins, and the company is now developing microbiome modulators. Vedanta and Seres Health are developing defined communities of microbes as therapies for various diseases, and, to Fischbach, the next logical step is to define their molecular output as well. “The molecules produced by a community are one very important way that they influence host biology,” he says. “To me, it’s worth optimizing whatever set of molecules that is. I could hardly imagine that we would leave that to chance in the future.”


  1. 1. Wang, Z. et al. Nature 472, 57-63 (2011).
  2. 2. van Nood, E. et al. N. Engl. J. Med. 38, 407-415 (2013).
  3. 3. Hsiao, E.Y. et al. Cell 155, 1451-1463 (2013).
  4. 4. Mazmanian, S.K. et al. Cell 122, 107-118 (2005).
  5. 5. Atarashi, K. et al Nature 500, 232-236 (2013).
  6. 6. Trompette, A. et al. Nat.Med. 20, 159-166 (2014).
  7. 7. Chang, P.V. et al, Proc. Natl. Acad. Sci. USA 111, 2247-2252 (2014).
  8. 8. Yoshimoto, S. et al. Nature 499, 97-101 (2013)
  9. 9. Donia, M.S. et al. Cell 158, 1402-1414 (2014)


Table 1: Companies developing microbiota-based therapies


Therapy Indication Stage
OpenBiome Banked processed stool from screened anonymous donors for FMT C. difficile infection Commercial (nonprofit) distribution
Seres Health SER-109 (orally delivered mixture of bacterial spores meant to mimic a healthy gut community) Recurrent C. difficile infection Phase 3
Coronado Biosciences TSO (CNDO-201) (Trichuris suis ova, nonpathogenic helminth) Plaque psoriasis, autism Phase 2
Osel Lactin V (live biotherapeutic product, delivered intravaginally) Bacterial vaginosis, recurrent UTI Phase 2 complete
Rebiotix (formerly MicroBex) RBX2660 (five-microbe suspension, enema administration) Recurrent C. difficile infection Phase 2
Second Genome SGM-1019 (small molecule against undisclosed target) Inflammatory bowel disease Phase 1
Symbiotic Health Encapsulated fecal-derived bacteria C. difficile-associated diarrhea Pilot trial complete
Vedanta/Janssen Biotech (J&J) VE-202 (cocktail based on 17 commensal Clostridia species) Allergic and inflammatory conditions Preclinical
GT Biologics Thetanix and Rosburix live biotherapeutic products (LBPs) Pediatric Crohn’s disease and ulcerative colitis Preclinical
AvidBiotics Avidocins, retargeted R-type bacteriosins as specific bactericidal agents C. difficile related diseases; E. coli related UTI Preclinical
Symbiotix Biotherapies, undisclosed pharma collaborator


Orally delivered PSA (polysaccharide A) Autoimmune, allergic & inflammatory diseases Preclinical
Second Genome/Janssen Biotech (J&J) Microbial metabolites, gene products or small molecules against host targets Inflammatory bowel disease Discovery
ActoGeniX Genetically engineered bacterial delivery of targeted microbiome modulators Colon cancer, C. difficile infection Strain development & in vitro proof of concept
Cleveland Clinic Small-molecule inhibitors of bacterial TMA synthesis Cardiovascular disease Discovery
Cipac Therapeutics FMT encapsulation or reconstitution for fecal transplant C. difficile infection, Crohn’s disease Undisclosed
Enterome BioScience (Paris) EP-8018 small-molecule antagonist of bacterial fimbrial adhesion Crohn’s disease Preclinical
UTI, urinary tract infection; FMT, fecal microbiota transplantation.  PSA, polysaccharide A



Comments are closed.