Addressing the Microbiome’s Role in Drug Discovery and Development
Cristiano Ruch Werneck Guimarães
Katia Sivieri
Miller Freitas
Stephani Saverio
Nintx – Next Innovative Therapeutics

he drug discovery and development field established over the years a range of in vitro and in vivo tools to model host-drug interactions and better estimate the pharmacodynamics (PD), pharmacokinetics (PK), and toxicity of drug candidates in humans. However, the microbiome’s role was generally ignored (see article in September Global Forum). The range of effects that the gut microbiome might have on administered drugs can produce unexpected outcomes on PK, PD, and toxicity. In turn, drugs may impact the gut microbiome and its interactions with the host cells (intestinal epithelial cells, immune cells, nerve cells, and enteroendocrine cells), perturbing normal body homeostasis. At the extreme end, such as with antibiotics, disruption to the microbiome can lead to dysbiosis and aggravate diseases.

This article summarizes emerging approaches that could be routinely used in preclinical development to investigate the complex drug-microbiome interactions and ultimately the drug-microbiome-host triad. A better understanding of these interactions can be instrumental in fully unraveling the mechanisms of action of drugs and their metabolites, identifying root causes of side effects, reducing attrition in clinical trials, and developing safer and more efficacious therapies.

Studying the Drug-Microbiome Interactions in Vitro

To focus on the bidirectional interaction between drugs and the gut microbiome, taking the host out of the equation, in vitro simulations are the method of choice. The more sophisticated multicompartment in vitro systems have between three and five bioreactors in sequence. A bioreactor refers to a vessel that provides a suitable reaction environment for living cells or enzymes to carry out cell proliferation and/or product formation. The systems with three bioreactors only simulate the ascending, transverse, and descending colon, while additional bioreactors, totaling five, include the simulation of the stomach and the small intestine (Figure 1). The pH, volume, transit time, and temperature are controlled in the different bioreactors. Digestive enzymes and bile are added to the simulated stomach and small intestinal compartments, where applicable, while inoculation with fecal microbiota from human donors (healthy volunteers or patients) only takes place in the bioreactors simulating the colon portions, as the concentration of microbes in the large intestine is 1000-fold greater than in the small intestine. These systems closely emulate the microbial composition and activity in different portions of the gut, being highly predictive of results obtained from fecal samples of in vivo animal and human studies.
Sophisticated multicompartment in vitro systems including a sequence of three to five bioreactors
Figure 1. Sophisticated multicompartment in vitro systems include a sequence of three to five bioreactors. Systems with three bioreactors only simulate the ascending, transverse, and descending colon.
By incubating a given drug or drug candidate with a specific human colon microbiota, the multicompartment in vitro systems can provide a window into the drug-microbiome interactions by allowing dynamic sampling over time in different consecutive regions of the simulated human colon. Through next-generation sequencing, one can understand how the gut microbiome is affected by drugs. Through metabolomics, one can follow the different drug metabolites (and metabolites of metabolites) produced exclusively by gut microbiota metabolism. These simulations may show how deep the drug-microbiome interactions can go. The shifts in gut microbiome composition are not necessarily unfavorable for the host, as is the case for metformin. This drug, efficacious in Type 2 diabetes with a mechanism of action not fully understood, seems to exert its effects via inhibition of liver gluconeogenesis (generation of glucose from noncarbohydrates), but recent evidence also points to the modulation of the gut microbiota through the growth of beneficial short-chain fatty acid producers. However, a recent study of over 1000 drugs demonstrated that about one-quarter of them showed antibacterial activity (even though most were not designed for that), affecting the growth rate of at least one out of the 40 bacterial strains evaluated. When testing microbial communities or single bacterial strains incubated with drugs, up to 65 percent of them were metabolized, suggesting that microbial drug metabolism is far more common than expected.

In summary, despite the absence of a physiological host environment, multicompartment in vitro systems can maintain a humanlike gut microbiota and specific activities observed in humans, and allow the study of drug-microbiome interactions without interference from the host. The system is even more predictive when the mucosal environment, in addition to the luminal environment, is incorporated by adding mucin to the model. In vitro models offer additional advantages as they are standardized and generate results with high reproducibility. Also, there are no ethical constraints. Finally, despite its important role in understanding drug-microbiome interactions, it is crucial to combine this technology, very useful for screening purposes, with lower-throughput approaches that more closely resemble the in vivo situation, as interactions between drugs and microbes identified in vitro need to be validated in the host context.

Studying the Drug-Microbiome-Host Triad in Vitro and in Vivo

The gut microbiota naturally produces peptides, proteins, antigens, signaling molecules, and a range of metabolites from fermentation that can interact with the host cells. When drugs are administered to an animal or patient, drug metabolites produced by the gut microbiota can also interact with the host. It is important to consider the drug-microbiome-host triad, which can be modeled at different levels of complexity. Several studies investigated the effect of samples from the in vitro simulations described above on enterocytes or enterocyte-like cells (including Caco-2, HT-29, T84, IEC-6, IEC-18, IPEC-J2, and IPEC-1 cell lines) and immune cells, such as the macrophage cell line U937, to study adherence, cytokine production, or gene expression.

An important advancement in the study of drug-microbiome-host triads was the development of organoids by differentiating stem cells into specific intestinal cell types, such as enterocytes, goblet cells, and Paneth cells, that formed spherical crypt-like structures. When combined with microfluidic technology, these cells form a gut-on-a-chip device that models the whole intestine. This type of in vitro technology holds a promising future by transforming our mechanistic understanding of the drug–microbiome–host triad.

In pre-clinical development, the ultimate goal is to understand the drug-microbiome-triad in a whole organism, generally in rodents, the standard model used in drug and microbiome research for many years. For example, in cancer immunotherapy, the gut microbiota plays an important role in host response to treatment. And research with mice has shown that the effect of PD1 and PDL1 checkpoint inhibitors against melanoma can be enhanced with increased abundance of Bifidobacterium species.

When it comes to drug metabolism, there are few strategies to separate host and microbiome contributions, such as the comparison of drug and drug metabolites’ systemic exposures between conventional and germ-free animals, conventional, germ-free, and humanized animals (colonized with human gut microbiota), and through combinatorial therapies, i.e., the drug under investigation with and without antibiotics. All of these efforts aim at teasing out the influence of the microbiome on the drug and drug metabolites’ PK, PD, and toxicity.

Call to Action

A successful drug discovery and development program transitions from in vitro assays to animal models to patients to the market. In every step of the way, assessments of PD, PK, and toxicity are made to ensure a drug candidate can turn into a safe and efficacious therapy. Up until now, all of the in vitro and in vivo models in pharmaceutical companies have acted as a surrogate of host-drug interactions only. Although important, the microbiome’s contributions have been mostly neglected. This is somewhat understandable, as next-generation sequencing and other advanced analytical techniques only became mainstream in the last decade. However, there is no more time to lose, as recent evidence suggests. For instance, the hepatotoxic metabolite of the oral antiviral drug brivudine, bromovinyluracil, is mainly produced by the gut microbiome. Clonazepam, an oral anticonvulsant and antianxiety drug, undergoes a complex metabolic pattern of host and microbial conversion impacting systemic drug and drug-metabolite exposure. The molecule p-cresol, produced by the gut microbiome, competitively inhibits enzymes in the liver, affecting acetaminophen elimination, leading to increased and potentially toxic levels of this drug.

It is imperative that the drug discovery and development field incorporates systematic approaches, such as the ones briefly described herein, to understand drug-microbiome interactions and the drug-microbiome-host triad, complementing the current in vitro and in vivo arsenal routinely employed to model host-drug interactions. This would (i) improve compound selection in preclinical development, (ii) better define specific patient populations, taking microbiome composition and activity into account, (iii) reduce or at least better understand interpersonal variability, and (iv) bridge the gap between pre-clinical and clinical results, ultimately leading to safer and more efficacious therapies. As added benefits, it would pave the way to repurposing nonantibiotic drugs with antimicrobial activity to develop nonresistant antibiotics and present a new perspective on the unknown origins of efficacy and side effects of many drugs in the market.

References available upon request.