Multimodal interactions of drugs, natural compounds and pollutants with the gut microbiota

Summary

The gut microbiota contributes to diverse aspects of host physiology, ranging from immunomodulation to drug metabolism. Changes in the gut microbiota composition are associated with various diseases as well as with the response to medications. It is thus important to understand how different lifestyle and environmental factors shape gut microbiota composition. Beyond the commonly considered factor of diet, small-molecule drugs have recently been identified as major effectors of the microbiota composition. Other xenobiotics such as environmental or chemical pollutants can also impact gut bacterial communities. Here we review the mechanisms of interactions between gut bacteria and antibiotics, host-targeted drugs, natural food compounds, food additives and environmental pollutants. While xenobiotics can impact bacterial growth and metabolism, bacteria in turn can bioaccumulate or chemically modify these compounds. These reciprocal interactions can manifest in complex xenobiotic-host-microbiota relationships. Our review highlights the need to study mechanisms underlying interactions with pollutants and food additives towards deciphering the dynamics and evolution of the gut microbiota.

Introduction

The physiological importance of the gut microbiota − the largest microbial community in the human body − is apparent both in its form and function. A typical adult microbiota harbours an estimated 3.8 x 10¹³ bacteria, outnumbering the host somatic cells and contributing circa 200 g to the body weight¹. At the genetic level, the gut microbiota contains approximately 30 times more genes compared to the human genome2, representing a vast metabolic capacity. Indeed, the gut microbiota contributes to various host functions, including food digestion, detoxification, endocrine function, neurological signalling, immune modulation and pathogen defence^3–6. The essential role of gut bacteria in these functions is reflected in the associations between the gut microbiota composition and diseases such as inflammatory bowel disease⁷, cancer⁸, obesity, type 2 diabetes, cardio-metabolic diseases³ and neurological disorders⁹. The strong link to physiology and health argues for treating the gut microbiota as an organ of the human body (Box 1).

Box 1.

The gut microbiota as an organ/supra-organ/second liver

The human gut microbiota is often considered as an organ or supra-organ, as it is essential for host functions, such as detoxification, endocrine function, neurological signalling, immune modulation and pathogen defence^3–6 and is involved in pathology and physiology⁴³. The gut microbiota is often referred to as second liver, as it is involved in metabolic processes and can contribute to activation, deactivation or toxification of xenobiotics^33–36.

Xenobiotics

Xenobiotics are compounds foreign to the human body. These include drugs, pollutants, toxins, and food additives and contaminants. It has been suggested that throughout a lifetime, an individual is exposed to around 10,000 to 100,000 different xenobiotics, at varying concentrations⁴⁴.

Alpha diversity

Alpha diversity refers to the bacterial species diversity within an individual microbiota or within an individual sample.

Macronutrients

A macronutrient represent the nutritive/main component of our food that is used by the body for energy, such as protein, carbohydrates and fat.

Micronutrients

A micronutrient refers to chemical elements or substances only required in trace amounts for healthy growth and development, such as vitamins or minerals.

Bacteriophage (Phage)

Viruses that can infect bacteria and can thereby contribute to bacterial abundance and metabolism. Most phages in the gut replicate through incorporation of the phage genome into the bacterial genome, leading to formation of prophages. These prophages may be induced by DNA-damage, antibiotics or other xenobiotics, leading to expression of the phage genome⁴⁵.

The interactions between this microbe-filled supra-organ (Box 1) with the rest of the body is largely determined by its microbial composition, predominantly through secreted metabolites. In recent years the impact of microbial metabolites on host physiology has been described by various reviews, showing how e.g. levels of short-chain fatty acids associate with diseases such as type 2 diabetes and neurological diseases^10,11. Around 68 % of the gut metabolome can be explained by the microbiota composition¹², with many of the bacteria-linked metabolites detected in the blood¹³. Given that, it is critical to understand how various intrinsic and extrinsic factors impact gut bacteria and their metabolism. While several such factors have been identified, the extent and the mechanistic basis of their impact is poorly understood. Known intrinsic factors include host genetic makeup, age, sex, and bowel movement^14,15 (FIG. 1a).

Figure 1. Reciprocal interactions between the gut microbiota and xenobiotics.

A) Intrinsic factors shown to influence gut microbiota composition include host-GI physiology, sex, age, and host genetics^14,15. B) extrinsic factors shown to influence the gut microbiota composition include alcohol intake, medication, diet, and other environmental factors^14,16,17. C) Ingested xenobiotics can affect the gut microbiota composition by inhibiting or promoting the growth of bacterial species. In turn, xenobiotics may metabolise xenobiotics and thereby alter their activity in the host. D) Metformin, a common treatment for type 2 diabetes has been shown to affect the gut microbiota composition. This altered composition has been shown to be at least in part responsible for the antidiabetic effect observed by metformin treatment^21,54. E) The pesticide glyphosate, which may be ingested due to food contamination, has been shown to alter the gut microbiota composition⁷⁸.

Extrinsic influences include diet, alcohol consumption, medications, and other environmental factors^14,16,17 (FIG. 1b). Although environmental factors have been estimated to outweigh genetic differences in explaining gut microbiota variation^18,19, the determination of effect sizes remains difficult due to numerous confounding factors¹⁸. Nevertheless, medications have been shown to contribute substantially to microbiota variation²⁰. A good case in point is metformin, a widely used medication in type 2 diabetes treatment; the microbiota variation associated with the drug was found to be more pronounced than that associated with the disease²¹. Further, in contrast to most intrinsic factors, extrinsic factors such as diet and medications can be modified and adapted. Therefore, understanding the interactions between environment and gut microbiota will be critical for designing strategies to improve health and precision medicine.

The interactions between environmental factors such as diet and medications and the gut microbiota are reciprocal. Diet and drugs may modify microbial composition and in return, gut bacteria may chemically transform these compounds (FIG. 1c); examples include digoxin used for treating cardiovascular disease^22,23 and levodopa for Parkinson’s²⁴. The impact of environmental pollutants and other xenobiotics is also indicated by several studies²⁵. These include pesticides^26–28, heavy metals^29,30 and nitrosamines^31,32. The gut microbiota has thus been linked to the activation, inactivation and toxicity of xenobiotics^33–36.

Xenobiotic exposure occurs on a day-to-day basis through skin contact, inhalation, or ingestion. We may ingest xenobiotics voluntarily, through supplementation or medication, or involuntarily through food and water contamination. Ingested compounds can be metabolized and modified by the gut microbiota before or after absorption³⁷. Orally administered or ingested substances are either absorbed in the small intestine and metabolized in the liver, or pass through the small intestine into the colon, where they can be metabolized by the gut microbiota, which acts like a second liver (Box 1). In addition, previously absorbed substances or substances administered via a different route of exposure can circle back to the intestine via biliary secretion and then interact with the gut microbiota^38,39. The metabolism of xenobiotics through gut bacteria may be performed by many different bacterial groups and species and therefore depend on the gut microbiota composition^37,40,41. Given the vast variety and number of xenobiotics that are potentially encountered by the gut microbiota (>25,000; FIG. 2; Supplementary Table 1), understanding the extent and the nature of xenobiotic − microbiota interactions will be crucial in unravelling the mechanistic basis of microbiota variation, dynamics, and evolution across and within individuals.

Figure 2. Overview of ingested xenobiotics.

Through our diet, we consume a vast number of natural compounds. The FooDB − the largest resource for food constituents, chemistry and biology − lists 797 different foods containing 15,750 detected natural compounds and a further 55,176 expected and predicted compounds⁶⁰. Besides the natural compounds present in our food, we also consume many other xenobiotics on a day-to-day basis, many of which yet remain unknown. The shown chart is not comprehensive and is intended to provide an overview of the breadth of different xenobiotics that may be ingested regularly. The information shown is compiled from multiple different databases, which can be found in supplementary table 1. In addition, not everyone will be exposed to all of the here mentioned xenobiotics throughout their lifetime and exposures will vary depending on lifestyle, occupation and geographical location.

In this review, we discuss the influence of widely encountered xenobiotics on the gut microbiota composition and highlight known modes of xenobiotic − gut bacteria interactions. The compounds are grouped into four classes: (1) antibiotics, (2) host-targeted drugs, (3) natural food compounds and food additives, and (4) environmental pollutants and chemicals. Interactions revealed by multiple experimental approaches cover in vivo relevance as well as mechanistic insights. We focus on direct interactions and molecular mechanisms where possible, and span a much broader range of xenobiotic compounds than previously reviewed^25,42.

Xenobiotics and gut microbiota composition

Collateral damage by antibiotics

Antibiotics have strong short-term effects on the gut microbiota and cause bacterial cell damage across numerous bacterial strains^42,46. Both in vitro and human studies investigating the effects of broad-spectrum antibiotics rifaximin, ciprofloxacin, metronidazole and vancomycin showed their profound short-term effects in reducing total microbial biomass⁴⁷ and alpha diversity and richness (Box1)⁴⁸. In addition, these studies showed a large shift in gut microbiota composition with a decrease in Bacteroides-Firmicutes ratio, reduced abundance of several genera, including Bacteroides and Faecalibacterium, and a substantial reduction of butyrate-producing genera from 27 % to 0.3 % relative abundance^47,48, which demonstrated the scale to which antibiotics might disrupt the gut microbiota.

Besides short-term effects, antibiotics have also been shown to be associated with long-term changes in gut microbiota composition. Ciprofloxacin and clindamycin have been shown to reduce species richness and alter microbial composition, including a decrease in Bacteroides and an increase in Faecalibacterium and Alistipes, lasting for up to 12 months⁴⁹. The recovery of the gut microbiota to pre-intervention levels hereby depends on the antibiotic compound and host-specific effects⁵⁰. The negative impact of antibiotics on commensals can open a niche for low abundant species, and sometimes even for pathogens^51,52. A study in healthy men treated with a cocktail of meropenem, gentamycin and vancomycin observed blooms of enterobacteria and other pathobionts, and depletion of Bifidobacterium species and other butyrate producers, which are thought to be protective against pathogens⁵³. These studies highlight that the damage antibiotic treatment has on the microbiota may further negatively impact the host through reduced production of beneficial metabolites or through the opening of a niche for pathogens.

Specific changes induced by human-targeted drugs

Compared to antibiotics, human-targeted drugs are a larger and more diverse group of medications. The exposure for some of the drugs, like antidiabetics and antidepressants, is often over a much longer period than to antibiotics. As the development of host-targeted drugs does not focus on bacteria, their effect on the gut microbiota composition is more unexpected and less studied. One of the few examples is that of metformin (FIG. 1d), which has been shown to influence microbial species abundance and metabolic pathways in the gut microbiota, possibly via metal homeostasis regulation⁵⁴. Indeed, type 2 diabetes patients treated with metformin show a significantly different gut microbiota composition with an increase of Escherichia spp. and a decrease in Intestinibacter spp. compared to untreated patients, a difference that was previously erroneously linked to the disease state²¹. Further, this altered microbiota composition has been shown to be, at least in part, contributing to the antidiabetic effect of metformin. For example, faecal transplants from metformin treated patients to germ-free mice caused an improvement in glucose tolerance compared to mice receiving faecal transplants from type 2 diabetes patients before metformin treatment, but mechanisms remain unclear⁵⁴. Besides this well-known example, a recent meta-analysis showed that 17 drug categories were associated with individual microbial taxa, with proton-pump inhibitors, antidiabetic drugs (metformin) and laxatives showing the strongest association with relative abundance of microbial taxa^47,55. The effect of other host-targeted drugs on the gut microbiota have also been noted, including antipsychotics⁵⁶, statins⁵⁷, beta-blockers and ACE-inhibitors⁵⁸. This growing evidence highlights that a wide range of host-targeted drugs can alter the gut microbiota composition in a drug-dependant manner and hence need to be considered for their role in microbiota dynamics and drug response.

Effect of natural food components and food additives

Not just macronutrients, but also micronutrients influence gut microbiota composition (Box 1)⁵⁹. Beyond these, the number of compounds naturally occurring in foods is enormous. The FooDB − the largest resource for food constituents, chemistry and biology − lists >55,000 natural compounds⁶⁰. Among the naturally occurring xenobiotics that have been shown to impact gut microbiota composition are phytochemicals. Herbal extracts of olive, camomile, cinnamon, and ginger were shown to influence gut microbiota composition. Common effects among these extracts were the inhibition of Escherichia and Shigella, which have been associated with gastrointestinal disorders, and the favoured growth of the butyrate producers Coprococcus and Butyricimonas⁶¹. In addition, single natural xenobiotic compounds, such as curcumin, have also been shown to alter the abundance of several gut bacterial species. Consumption of curcumin, a primary ingredient in turmeric, was associated with an increase in Clostridium and Bacteroides and a reduction in Blautia and Ruminococcus species⁶². Tea polyphenols increased Bifidobacterium species, which are associated with probiotic function, and reduced Bacteroides-Prevotella⁶³, whereas berberine, an alkaloid found in several plants, reduced the abundance of Bacteroides uniformis⁴⁷.

In addition to natural food ingredients, food additives may also interact with and shape gut microbiota composition. Artificial sweeteners and emulsifiers are some of the most widely used food additives. The commercially available sweeteners Splenda, saccharin and sucralose^64–67 and the emulsifiers carbomethylcellusolse and polysorbate 80^68–70 have been shown to promote changes in gut microbiota composition and dysbiosis in mice. These changes include a decrease in Firmicutes and an increase in Bifidobacterium by sucralose⁶⁷, an increase in Proteobacteria and Escherichia coli overgrowth by Splenda⁶⁵, and an increase in Bacteroidetes and decrease in Clostridiales by saccharin, which had been associated with diabetes mellitus⁶⁶. In addition, both carbomethylcellusolse and polysorbate 80 promoted inflammation, reduced microbiota diversity and increased Bacteroidales while decreasing Clostridiales relative abundance^68,69.

Impact of pesticides and other pollutants

Humans unintentionally consume a large number of environmental contaminants such as pesticides, environmental chemicals, processing contaminants, mycotoxins and heavy metals⁷¹ (FIG. 2). To date the effects of these contaminants on the human microbiota have not been thoroughly investigated. Nevertheless, the exposure to various environmental pollutants have been associated with changes in gut microbiota composition²⁵.

Pesticides, including insecticides, fungicides, and herbicides, are widely used in food production and therefore represent a major group of food contaminants. In 2018, the Food and Drug Administration (FDA) and European Food Safety Authority (EFSA) detected 212 and 358 different pesticide residues in foods respectively^72,73. In addition, in 2020 the UK Expert Committee on Pesticide Residues in Food detected 123 residues in 2460 food samples, 42 % of which contained at least one pesticide residue⁷⁴. A recent study screening levels of 186 pesticides in a UK cohort (n=130) detected pyrethroid and/or organophosphorus insecticides in all urine samples. For example, the pyrethroids cypermethrin and permethrin were detected in >96 % of samples and the organophosphate diethylphosphate was detected in 75 % of samples. Glyphosate, the only herbicide detected, was present in 53 % of samples75, highlighting that the host, and thus likely the microbiota exposure to pollutants is widespread and should be systematically investigated. Various groups of pesticides have been shown to affect gut microbiota composition and microbiota-related metabolites, including short-chain fatty acids, bile acids and trimethylamine²⁸. A study in mice reported that interaction with the gut microbiota may be contributing to the adverse effects observed for organophosphates, such as impaired glucose tolerance⁷⁶.

Chlorpyrifos and glyphosate are among the most widely used organophosphates and have both been linked to changes in the gut microbiota composition in rodents^26,27,77. Even at low concentrations relevant for dietary exposure (1 − 3.5 mg/kg/d), chlorpyrifos and glyphosate exposure led to alterations in gut microbiota composition^26,27. Another recent study investigated the effect of glyphosate on the gut microbiota and metabolome in rats using a multi-omics approach (FIG. 1e). Glyphosate not only increased the abundance of Eggerthella spp., Shinella zoogleoides, Acinetobacter johnsonii, and Akkermansia muciniphila, but also caused changes in gut metabolite levels suggesting an inhibition of the shikimate pathway in gut bacteria, which is the primary mechanism through which glyphosate acts as a herbicide⁷⁸. As the shikimate pathway is not present in human and animal cells, glyphosate is generally assessed as safe. However, the effect of the inhibition of this pathway in the human gut microbiota on host health remains unclear. Nevertheless, these findings suggest that biochemical pathways in the gut microbiota need to be considered when assessing the adverse health effects of pesticides and other environmental chemicals.

An important class of environmental pollutants are N-nitrosamines due to their carcinogenic and toxic potential. N-nitrosamines are a diverse class of chemicals including over 300 different compounds⁷⁹. The total N-nitrosamine exposure has been estimated to be between 1900 and 25,000 ng/day depending on lifestyle choices, with tobacco-product consumption causing the highest exposure (~ 22,000 ng/day), followed by food (~ 1900 ng/day), alcoholic beverages (~ 1000 ng/day) and drinking water (~ 120 ng/day)⁷⁹. In addition, in recent years nitrosamine levels in medications exceeding the acceptable daily intake of 26.5 ng/day, have led to recall of >1400 products by the FDA, e.g. metformin and the blood pressure medication valsartan^80,81. While high concentrations in drinking water (mg/L) of various N-nitrosamines have been shown to be carcinogenic in rodent models^82,83, the human exposure to single N-nitrosamines is usually orders of magnitudes lower (ng/L), and thus has not been extensively studied⁷⁹. A study in male rats supplemented with an N-nitrosamine-mixture at environmentally relevant doses showed an appreciable shift in composition towards an obesogenic microbiota profile. This included an increased Firmicutes/Bacteroidetes-ratio and Alistipes and Ruminococcus enrichment³². In addition, supplementation with the nitrosamine dichloroacetonitrile altered microbiota composition and microbial metabolites, including metabolites involved in sphingolipid signalling pathway and fatty acid biosynthesis³¹. These results suggest that N-nitrosamines may affect the microbiota composition and host health at concentrations relevant for human exposure (ng/L).

Mycotoxins are toxic fungal secondary metabolites and are found as contaminants in food. More than 300 different mycotoxins have been identified, with some being regularly found in foods, such as aflatoxins, ochratoxins, fumonisins, patulin, zearalenone and trichothecenes⁸⁴. A recent literature review and analysis of EFSA and other large global survey data estimated the prevalence of mycotoxins in foods to be >60 % and 20 % of the food samples to exceed the European Union regulatory limits⁸⁵. A study in mice demonstrated that ochratoxin treatment (210μg/kg body weight) reduced microbiota diversity and increased Lactobacillus abundance, which has been shown to play a role in ochratoxin detoxification⁸⁶. In addition a study exposing mice for 9 months to doses relevant for estimated human intake (10μg/kg body weight) also showed changed abundance in several bacterial phyla, families and genera, including a decrease in the genera Bifidobacterium and an increase in Odoribacter, which have been associated with gut inflammation⁸⁷. These findings demonstrate that mycotoxins need to be considered regarding their interactions with the gut microbiota and in relation to host health.

Heavy metals are also common contaminants in food and drinking water. The human gut microbiota has been shown to contribute to the control and absorption of these toxic metals⁸⁸. For example, the gut microbiota has been shown to protect mice from arsenic toxicity. Compared to antibiotic-treated or germ-free mice, mice with a gut microbiota accumulated less arsenic in organs and excreted more arsenic in their stool³³, showing the possible importance of the gut microbiota in mediating heavy metal toxicity. In turn, heavy metal exposure has been shown to impact the gut microbiota diversity and composition⁸⁸. Cadmium, arsenic, copper, lead and aluminium exposure have all been shown to alter the gut microbiota composition at both phylum and genus levels, and associate with reduced diversity^29,30. In addition, zinc has also been shown to not only alter the gut microbiota composition in mice, but also increase susceptibility to C. difficile infection and severity of disease⁸⁹, highlighting the importance of understanding the impact heavy metals have on our gut microbiota.

Residues of veterinary antibiotics and medications may also contaminate food, especially animal products. In 2018 the EFSA found that out of 354,517 samples 0.3 % were non-compliant and exceeding the permitted limits. These non-compliant samples contained veterinary medications such as antibacterials and substances with anabolic effects⁹⁰. Although these residues are ingested in much lower concentrations than medications ingested for therapeutic use, they may also affect the gut microbiota composition⁹¹ and warrant further investigation.

Several other food contaminants, including food contact material, plant toxins or industrial chemicals can find their way into our food during food production, packaging, or preparation. As highlighted by the above examples, diverse classes of contaminants have the potential to alter the gut microbiota composition in vivo, stressing the need to consider microbiota-mediated impact on health in their safety assessment.

Xenobiotics − bacteria interactions mechanisms

The human cohort and animal studies have demonstrated numerous associations between the intake of xenobiotics and microbiota composition and function. However, understanding the underlying mechanisms is critical for leveraging these findings in a health context and requires in-depth investigation at species and strain level. Six major modes of xenobiotic − gut bacteria interactions have been described so far and are reviewed below (FIG. 3, FIG. 4, Table 1).

Figure 3. Xenobiotic − gut bacteria interactions.

There are six known modes of interaction between xenobiotics and the gut microbiota: (1) bacterial growth inhibition by xenobiotics^92,96, (2) bacterial growth promotion by xenobiotics^97,98, (3) change in natural bacterial metabolites by xenobiotics^93,98, (4) impact of xenobiotics on the virome and virulence of bacteria^106–108, (5) xenobiotic metabolism by bacteria^35,36,40 and (6) xenobiotic bioaccumulation by bacteria^97,98,. By inhibiting or promoting the growth of certain bacterial species xenobiotics contribute directly to shaping the gut microbiota composition, which in turn determines the function of the gut microbiota. In addition, gut bacteria can metabolize^40,41 or bioaccumulate ^97,98 and thereby change their activity or toxicity. This metabolism or bioaccumulation depends on the composition of the gut microbiota and the produced metabolites can in turn influence gut microbiota composition. All interactions may occur in various combinations and are dependent on gut microbiota composition. (For more xenobiotic-bacteria examples see Table 1)

Figure 4. An overview of xenobiotic − gut bacteria interactions investigated at the species level.

A) Growth promoting or growth inhibiting effects and effects of xenobiotics on virulence of bacteria. B) Xenobiotic biotransformation or bioaccumulation by gut bacterial strains. Growth promotion and xenobiotic bioaccumulation, in contrast to growth inhibition and biotransformation, have only recently started being assessed, and, thus, are yet underrepresented. The research to date also focusses mainly on antibiotics and host-targeted drugs, leaving other xenobiotics such as environmental pollutants underrepresented. The information in this diagram was compiled from 8 systematic screens (Supplementary table 2). The numbers next to xenobiotic classes indicate the number of compounds in each class, whereas the numbers next to bacterial phyla indicate the number of bacterial species for which a mechanistic xenobiotic-species relationship has been demonstrated experimentally.

Table 1. Examples of xenobiotic-gut bacteria interactions.

Interaction Type	Example	Mechanism	Ref.
Growth inhibition	B. vulgatus and B. uniformis & erythromycin (antibiotic)	Erythromycin causes cell shape defects (blebbing, cytoplasmic shrinkage, cell lysis)	124
Growth promotion	B. thetaiotaomicron & Alternaria toxins (mycotoxins)	Mycotoxins may be metabolised and used as carbon source by specific bacteria *	97
Change in native metabolites	Clostridium saccharolyticum & duloxetine (antidepressant)	Binding of duloxetine to metabolic enzymes.	98
Xenobiotic impact on virome and virulence	E.coli & carboxymethylcellulose and polysorbate-80 (emulsifiers)	Mediation of virulence factors flagella, type 1 pili and long polar fimbriae	107
Xenobiotic metabolism	B. thetaiotaomicron & diltiazem (oral calcium channel blocker)	Deacetylation of both diltiazem and diltiazem metabolites in the gut is bt4096-dependent.	40
	Eggerthella lenta & digoxin (cardiac drug)	Reduction of digoxin to dihydrodigoxin through cgr operon.	22
Xenobiotic bioaccumulation	Clostridium saccharolyticum & duloxetine (antidepressant)	After uptake duloxetine binds to proteins in bacteria. Differences in strain bioaccumulation may depend on different uptake and efflux systems.	98

^*Proposed mechanisms

Growth inhibition by xenobiotics

The most frequently reported interaction mode is bacterial growth inhibition by xenobiotics. In a large-scale in vitro experiment, 78 % of 144 antibiotics showed anti-bacterial activity against at least one of the 40 tested gut bacterial strains, many of which are common members of the human gut, such as Bacteroides vulgatus and B. uniformis⁹². Most antibiotics affected several strains with no specificity pattern identified. The same study found that 24 % of 835 human targeted drugs were active against at least one of the 40 tested strains⁹². The human targeted drugs each affected a smaller number of strains compared to antibiotics, highlighting their specific effect as opposed to broad collateral damage of antibiotics. This suggests human-targeted drugs targeting more specific pathways within the bacterial cells, which may only be present in specific strains. Yet, human targeted drugs were found to affect more abundant species supporting their community-scale impact observed in cohort studies. As an example, metformin has been shown to inhibit E. coli growth by inhibiting folate metabolism⁹³. In addition, the antipsychotic medication chlorpromazine and the antidepressant amitriptyline have been shown to inhibit ArcB-mediated efflux⁹⁴.

Food-associated xenobiotics have also been shown to inhibit bacterial growth. The artificial sweeteners sucralose, saccharin and acesulfame potassium inhibited the growth of E. coli strains HB101 and K-12⁶⁷. Further, an in vitro study looking at the effects of seven spice extracts (black pepper, cayenne pepper, cinnamon, ginger, oregano, rosemary, turmeric) on 88 different bacterial strains, found that the growth of all four tested Ruminococcus and some Fusobacterium and Clostridium species were inhibited by at least some of the tested spices⁹⁵.

Other food contaminants with adverse effects on the growth of gut bacterial species include dyes and mycotoxins. The dyes Sudan I, II, III, IV and Para Red showed growth inhibition for at least one of the 11 tested gut bacterial strains⁹⁶. A recent study demonstrated that a complex extract of Alternaria mycotoxins, toxic fungal metabolites, was able to inhibit the growth of diverse bacterial strains, including the abundant strain B. vulgatus⁹⁷. These examples indicate that anti-gut bacterial activity is likely a common characteristic for a wide range of xenobiotics.

Growth promotion by xenobiotics

An in vitro study investigating the effects of seven spice extracts on 88 bacterial strains found that the growth of all 33 tested Bifidobacteria and Lactobacillus strains were promoted by at least one of the spice extracts⁹⁵. In addition, another study found that some drugs may also promote the growth of certain bacterial species, such as B. uniformis by monekulast and Lactococcus lactis by loperamide⁹⁸. The latter is an interesting observation as L. lactis is a probiotic and loperamide is used for diarrhea treatment. A complex Alternaria mycotoxin extract also promoted the growth of B. caccae and B. thetaiotaomicron at low concentrations (0.5 μg/mL) ⁹⁷. For drugs and toxins growth inhibition is more frequently reported than growth promotion: Among 375 drug-bacteria interactions, 15 showed growth inhibition and only 5 promoted growth⁹⁸, whereas among 56 Alternaria-bacteria interactions at different concentrations, 17 showed growth inhibition at high (50 μg/mL) or low (0.5 μg/mL) concentrations and only 2 showed growth promotion at low concentrations ⁹⁷. There are also comprehensive reviews suggesting growth promoting effects of different natural xenobiotics ⁹⁹ or food additives ¹⁰⁰, however, most studies are performed on complex microbial communities, which makes it difficult to extrapolate single species growth effects. In addition, mechanisms underlying growth promotions are less well studied and understood. One possible explanation could be that xenobiotics serve as a nutrient source for bacteria¹⁰¹. For example, Klebsiella pneumoniae and Escherichia fergusonii have been shown to utilize the antibiotic chloramphenicol as a carbon source¹⁰². Taken together, growth promotion by xenobiotics is less frequently reported than inhibition (FIG. 4) yet may play a critical role in a community context when the concerned bacterium is central to the community ecology or function.

Modulation of bacterial native metabolism by xenobiotics

Beyond affecting growth, some xenobiotics have also been shown to alter the natural metabolism of gut bacteria. Despite the increasing evidence highlighting the importance of bacterial metabolism, there is still limited knowledge of how bacterial metabolism effects drug treatment and host health ¹⁰³. Several studies reported a change in faecal metabolome after xenobiotic administration ^31,78, yet only very few studies investigated this on a single strain level. Studies investigating the mechanisms behind metformin action reported that metformin not only altered bacterial folate and methionine metabolism ⁹³, but also altered central pathways, including upregulation of the tricarboxylic acid cycle and downregulation of glycolysis and arginine degradation ¹⁰⁴. In addition, a recent study performed metabolite profiling in six bacterial strains incubated with the antidepressant drug duloxetine. Of these, Clostridium saccharolyticum, Lactiplantibacillus plantarum, E. coli and L. lactis showed significant changes in extracellular metabolite abundances. Particularly C. saccharolyticum showed a large shift in its exo-metabolome, including changes in amino acids and nucleotide related pathways. This was also shown to be relevant for community composition through metabolic cross-feeding, as demonstrated by the indirect increase of Eubacterium rectale⁹⁸. Beside drugs, the natural xenobiotic trigonelline, an alkaloid found in coffee, also showed potential in altering bacterial metabolism. Incubation of Citrobacter freundii with trigonelline in choline-enriched media significantly reduced trimethylamine production, which can be converted to the pro-atherosclerotic metabolite trimethylamine N-oxide in vivo ¹⁰⁵. Overall, recent studies bring forward modulation of native bacterial metabolism as one of the fundamental modes of xenobiotic-bacteria interaction.

Xenobiotic impact on virome and virulence

The impact of xenobiotics on the gut virome and the virulence of gut bacteria is a mode of interaction that has been receiving increasing attention in recent years. The gut virome is dominated by bacteriophages (Box 1), which are viruses that can infect bacteria, thereby altering bacterial abundance and metabolism⁴⁵. A recent study showed that several medications, including the antibiotic mitomycin, the chemotherapeutic fludarabine and the cardiac drug digoxin, lead to prophage induction in bacterial isolates from the human gut, demonstrating the potential of phage-mediated effects to alter the gut microbiota composition¹⁰⁶. In addition, the emulsifiers carboxymethylcellulose and polysorbate-80 have been shown to promote chronic inflammation by directly inducing genes responsible for the mediation of virulence in adherent-invasive E.coli, such as the virulence factors flagella, type 1 pili and long polar fimbriae¹⁰⁷. Dietary trehalose has also been shown to increase the disease severity and virulence of the C. difficile strains RT027 and RT078, possibly via increased toxin production¹⁰⁸. Also, the artificial sweeteners saccharin, sucralose and aspartame have been shown to increase the ability of the model gut bacteria E. coli NCTC10418 and E. faecalis ATCC19433 to adhere to, invade and kill host epithelium in vitro, which is thought to be related to increased expression of virulence factors¹⁰⁹. These examples demonstrate the importance of taking not only the microbiome but also the virome into account when investigating the relationship between xenobiotics, gut microbiota, and the host.

Xenobiotic metabolism by gut bacteria

Enzymatic modification of drugs by gut bacteria (biotransformation or metabolization) is a long known and widely recognised interaction mode ¹¹⁰. A well-known recent example is the cardiac drug digoxin, which demonstrates microbiota-dependent interpersonal effectiveness in drug response ^22,23. In vitro and mouse studies have demonstrated that only specific strains of Eggerthella lenta, e.g. DSM2243, are responsible for the inactivation of digoxin in the gut ²², providing mechanistic insight into how the microbiota composition contributes to the varying effectiveness of digoxin treatment between patients. In addition, chemotherapeutic drugs, such as the fluoropyrimidine 5-fluorouracil or camptothecin, have been shown to have varying effects on C. elegans depending on their bacterial diets^111,112. For example, worms fed E. coli HB101 showed a 80-times higher minimum inhibitory concentration on egg hatching compared to worms fed E. coli BL21G (8μM vs. 0.1μM)¹¹¹. Other examples of microbial drug metabolism are the activation of the cancer prodrug irinotecan or the deactivation of the cancer drug gemcitabine. Irinotecan causes gastrointestinal toxicity, even if given intravenously, due to β-glucuronidase activation or re-activation in the gut¹¹³. Gemcitabine on the other hand has been shown to be deactivated by gammaproteobacteria, most of which belong to the common families of Enterobacteriaceae and Proteobacteriaceae¹¹⁴.

Recent studies indicate that transformation by gut bacteria is highly prevalent. An in vitro study screening 76 bacterial strains against 271 drugs found that around two thirds of these drugs were significantly reduced by at least one bacterial strain, and that each strain was able to metabolize between 11 and 95 drugs. In addition, they identified and validated 30 microbiota-encoded enzymes that converted 20 drugs to 59 metabolites, whose abundance in gut microbial communities partially explained the communities drug metabolising capacity ⁴⁰. Another study using personalized gut microbiota cultures found that 57 out of 438 screened drugs spanning 28 pharmacological classes were metabolised, including 45 novel drug-microbiota interactions ⁴¹. These studies suggest that the gut microbiota is involved in the metabolism of a range of drugs and that by determining an individual’s drug metabolizing capacity one could predict the response to a specific drug or therapy. A recent study investigating the therapeutic efficacy of methotrexate, a common drug prescribed for rheumatoid arthritis, not only demonstrated a correlation between pre-treatment microbiota and response to treatment, but was also able to predict the response to methotrexate using machine learning approaches¹¹⁵. While some chemical similarities, such as amide-, nitro-, azo-or urea groups, have been identified between drugs metabolised by gut bacteria⁴⁰, datasets to date are not large enough to enable generalisation.

Similar to therapeutic drugs, environmental contaminants may also be metabolized by the gut microbiota, which could affect their bioavailability or toxicity in the host ³⁷. In vitro incubation of Lactococcus lactis, Lactobacillus fermentum and E. coli demonstrated the potential of these strains to degrade the pesticide chlorpyrifos within 15 days to 70, 61 and 16 % respectively. Interestingly, each of these strains produced different metabolites, indicating that multiple pathways may be involved in microbial chlorpyrifos degradation ³⁵.

Mycotoxins, which are fungal secondary metabolites and found as contaminants in food, are also reported to be biotransformed by gut bacteria. A study investigating the metabolism of masked mycotoxins (i.e., mycotoxins conjugated to sugars, glutathione or sulphates) in human faecal samples demonstrated that these were efficiently hydrolysed in the faecal samples, which may lead to higher mycotoxin exposure, as unmasked mycotoxins are absorbed more efficiently ¹¹⁶. Another study investigated the metabolising capacity of 14 different bacterial strains to metabolise masked mycotoxins or to degrade free mycotoxins (i.e., deoxynivalenol, H2 and HT2 toxin). Several bacterial strains were shown to harbour these capacities, including B. adolescentis, an abundant strain in the human gut ¹¹⁷.

Different dyes, such as Sudan I, II, III, IV and Para Red, have also been shown to be metabolised by the gut microbiota. An in vitro study showed that 33 out of 35 tested bacterial species were able to degrade at least one of these dyes. This suggests that the ability to reduce these types of dyes are widely spread among human gut bacterial species ³⁶. In addition, a recent study demonstrated that the azo dye Red 40, one of the most used food colorants worldwide, was shown to promote colitis in mice with dysregulated IL-23 expression in a microbiota dependent manner. The abundant bacterial strains E. faecalis and B. ovatus were shown to degrade Red 40 in vitro, and were also shown to contribute to Red 40 induced colitis in mice¹¹⁸. Another common group of food contaminants are polycyclic aromatic hydrocarbons (PAHs), which can be introduced during high-temperature cooking. PAHs have been shown to have toxic, carcinogenic and estrogenic effects in the human body. It has long been established that PAHs can be activated by cytochrome P450 enzymes in the human body to more reactive metabolites, which is thought to contribute to their carcinogenicity¹¹⁹. In addition to this, an in vitro study investigated the metabolising capacity of the gut microbiota for different PAHs in a simulated system of the human intestinal tract ¹²⁰. Only microbial digested PAHs, and unmodified compounds, showed estrogenic activity indicating that the gut microbiota also contributes to the bioactivation of PAHs in the human body ¹²⁰.

The examples in this section demonstrate the potential of the gut microbiota to metabolise a wide variety of compounds and thereby alter their bioavailability, efficacy, and toxicity in the host.

Xenobiotic bioaccumulation by gut bacteria

Bioaccumulation of xenobiotics by gut bacteria is only recently being appreciated. A study investigating the effect of 14 gut bacterial strains on a complex Alternaria mycotoxin extract showed different tendencies of these mycotoxins to associate with bacterial cell pellets. Alternariol, alternariol monomethylethyl ether and altersetin were not only found to be significantly depleted by many bacterial strains, but also to accumulate within the bacterial pellets. This accumulation was correlated with the lipophilicity of these compounds and was observed more often for gram-negative bacteria, such as B. vulgatus and B. thetaiotaomicron ⁹⁷. Another study investigating drug-bacteria interactions found that 17 out of 29 interactions in which the drug was depleted, was not caused by biotransformation by bacteria, but by bioaccumulation. For example, duloxetine, a widely used antidepressant, was accumulated by several bacterial species, including Streptococcus salivarius, Clostridium saccharolyticum and E. coli⁹⁸. This finding challenges the view that the main mode of bacteria-drug interaction is biotransformation. Also, some drugs showed potential to be either bioaccumulated or degraded depending on the bacterial strain, and most bacterial strains were capable of both interaction types depending on the drug⁹⁸. These findings could explain results from other studies showing drug depletion without metabolite detection ⁴¹. Bioaccumulation thus needs to be considered in deciphering the effect of xenobiotics at species and community, and host level.

Xenobiotic mixtures

To date, most studies investigating microbiota-xenobiotic interactions focus on one compound at a time, a condition which does not occur in vivo. Medications are often taken in combination with food containing a large number of natural compounds and contaminants. As all ingested compounds will be co-present in the gut, it is important to determine not only single, but also combination effects of xenobiotics, as the effect of a single compound on a single strain or community may not be the same as the effect of a complex mixture on a single strain or community.

Several studies have shown that different dietary and environmental factors can influence drug action indicating combination effects ⁴⁴. For example, bisphenol A, a chemical used in plastic food packaging, has been shown to decrease the efficacy of anti-cancer drugs, such as rapamycin ^121,122. Also, the activation of the cancer drug fluoropyrimidine by gut bacteria has been shown to be regulated by dietary cues, such as pyrimidines or vitamins B6 or B9 in C. elegans^111,112. In addition, an increasing number of studies show that the combination of xenobiotics has varying effects on the gut microbiota compared to single substances. The inactivation of digoxin by specific E. lenta strains has been shown to be inhibited by arginine, indicating that an increase in protein consumption may reduce in vivo digoxin reduction ²². The combination of chemicals in glyphosate formulations have also been shown to have a different effect on gut bacteria, including probiotic L. rhamnosus strains, than the pesticide glyphosate alone ⁷⁸, suggesting a synergy between glyphosate and formulation compounds. Another study tested combinations of 79 compounds, including antibiotics, host-targeted drugs, and food additives on 6 bacterial strains, and found both synergistic and antagonistic effects, the latter being more frequent ¹²³. Here antagonism was shown between drugs targeting different cellular processes, such as the food additive vanillin, which has been shown to increase the levels of aerobic respiration control protein ArcA, and the antibiotic chloramphenicol, an inhibitor of protein synthesis, in E. coli. The less frequently observed synergies only occurred between compounds targeting the same process, such as vanillin, effecting the multidrug transporter MdfA, and the antibiotic spectinomycin, a further inhibitor of protein synthesis, in E. col¹²³. A search among the Prestwick drug library, containing FDA- and EMA-approved drugs, for antagonists of erythromycin, which is bactericidal to the common gut bacteria B. vulgatus by causing cell shape defects such as blebbing, cytoplasmic shrinkage and lysis, identified that dicoumarol and tolfenamic acid are able to rescue B. vulgatus without affecting the killing capability of erythromycin on pathogens ¹²⁴. Further, an in vitro study found that 45 out of 186 bioactive compounds showed synergistic effects with antibiotics, such as triclosan, on E. coli MC1061 growth ¹²⁵. These results bring forward the possibility of reducing adverse effects of antibiotics and other drugs using xenobiotic combinations ¹²⁴, but also stress the need to systematically study xenobiotic combinations to better understand the extent of and mechanisms through which xenobiotics shape the microbiota and vice versa.

Conclusion and Future Perspectives

A wide range of xenobiotic compounds that can interact with the gut microbiota are consumed as part of diet or as medication. Several of these xenobiotics have been shown to influence bacterial physiology and thereby the gut microbiota composition, while the bacteria in turn modify xenobiotic activity and toxicity. Many of these interactions, e.g. digoxin²² or fluoropyrimidine¹¹¹, have been shown to be variable across individuals depending on the gut microbiota composition. Further research of these inter-individual variations will advance the knowledge and enable development of personalised approaches for dietary or medical interventions.

In addition, most studies investigating environment-microbiota-host interactions have yet focussed on one environmental factor at a time, such as diet, drugs or environmental pollutants ^40,41,95. To gain a complete understanding of environment-microbiota-host interactions, it will be important to simultaneously account for multiple contributing factors. Previous reviews have described and summarised system-level approaches that can be used to help elucidate drug-mircobiome-host interactions⁴². For example, it has been shown that in vitro and in vivo experiments in combination with computational strategies can be combined to disentangle host and microbial contributors for drug metabolism, an approach that could be adapted for other xenobiotic compounds³⁸.

Another important aspect that needs to be considered when conducting or comparing studies is xenobiotic concentration. The effects of xenobiotics on single strains or the complete microbiota depend on the concentration at which they are tested. For example, a comprehensive drug screen showed that more species were inhibited at higher drug concentrations ⁹². In addition, growth inhibitory or promoting effects of spice extracts ⁹⁵ or Alternaria mycotoxins 97 were dependant on the concentration. B. thetaiotaomicron, for example, exhibited growth promotion at low concentrations (0.5 μg/mL) and growth inhibition at high concentrations (50 μg/mL) ⁹⁷, underscoring the critical importance of considering concentrations in assessing the xenobiotic effects. The observation that more bacterial species are inhibited at higher concentrations may be due to multiple factors, including changes in osmolarity or pH, or due to changes in compound uptake and binding to macromolecules. It is important to note that “high” or “low” concentrations are context and compound dependent. Besides this, the concentrations used in studies are also model dependent. For human or animal studies concentrations will usually be higher (mM range) than for in vitro cell studies (μM range).

While testing at concentrations relevant for the gastrointestinal tract would be ideal, this information is rarely available. The trend to include metabolomics analysis of faecal samples, urine and plasma in microbiota studies ^12,78 would thus become more beneficial, as this would provide an idea of the concentration of specific compounds and also about the concentration range between individuals and time points. Additionally, screening of wastewater and wastewater-based epidemiology provide one possibility to monitor trends on drug consumption or environmental pollutants on a community wide basis¹²⁶.

As more mechanistic data on xenobiotic-bacteria interactions become available, machine-learning approaches could help predict reciprocal xenobiotic-bacteria interactions given the information on the chemical structures and bacterial genomes. While some databases already cover >13,000 xenobiotic-gut bacteria interactions, >11,000 of these interactions concern pharmaceuticals ^127,128. To enable the successful implementation of machine-learning approaches a large dataset covering diverse xenobiotics classes (FIG. 1) and diverse bacterial species is needed. Indeed, there are many more xenobiotic classes than could be reviewed here due to lack of studies. Future research should put a strong focus on substances that are commonly found in foods and medications and are regularly ingested, but also consider other substances and compounds that enter the body via inhalation, dermal absorption, and injection.

While many experimental approaches to determine microbiota-xenobiotic interactions exist, each comes with its own limitations, making translation of experimental findings into clinical or health advisory settings challenging (FIG. 5). Large scale human studies (top-down approach) are needed to establish associations between xenobiotics and the gut microbiota. These represent the physiological conditions and are directly relevant for both functional studies and applications. However, the complexity of the study design and relatively low control over the conditions mean that many confounding factors exist, such as lifestyle and reciprocal host-microbiota interactions. It is thus often difficult to establish causal links or pinpoint underlying mechanisms. In vitro studies (bottom-up approach) on the other hand with cultured gut bacterial strains can reveal molecular mechanisms, but it is challenging to extrapolate to in vivo conditions, as in vitro studies may not always be relevant in the in vivo contect. In addition, many gut bacterial strains are present in very low abundances and are difficult to cultivate, which hinders their identification and examination¹²⁹. Animal models, such as mice, rats or C. elegans, allow an intermediate approach, where microbiota-host interactions can be investigated in vivo in controlled settings. However, translation of the findings from animals to humans may yet be difficult, as animal models of disease were not developed with the gut microbiota in mind130. In future, ingestible measurement and reporting devices, such as magnetic hydrogels containing living bacteria, or micro-bio-electronic devices, could provide information on in situ xenobiotic concentrations and thus shed light on xenobiotic-microbiota-human interactions under real-life conditions^131,132. Understanding the possibilities and limitations of each model is crucial to enable linking of different study types to deepen our understanding of the gut microbiota. We envisage that emerging mechanistic insights would enable merging of top-down and bottom-up approaches towards model-guided prediction of xenobiotic-microbiota interactions.

Figure 5. Models used to investigate environment − microbiota − host interactions.

A) Bottom-up approaches based on in vitro studies using cultured gut bacterial strains can help reveal molecular mechanisms. On the other hand, in vitro studies may not always be relevant in the in vivo context. Additionally, the abundance of many strains is very low and cultivation can prove difficult, hindering strain identification and examination¹²⁹. B) Animal models, including mice, flies and C. elegans, present an intermediate approach, allowing the investigation of microbiota interactions in vivo under controlled settings. Translation of findings from animal models to humans may yet be difficult, as animal models of disease were not developed with the gut microbiota in mind, leading to less informative results¹³⁰. C) Investigation based on human cohorts, so called top-down approach, is directly relevant for both functional studies and applications. However, the complexity of the study design and relatively low control over the conditions means that many confounding factors exist, such as lifestyle and reciprocal host-microbiota interactions. It Is thus often difficult to establish causal links or pinpoint underlying mechanisms.

The xenobiotic concentrations used in studies often vary across the spectrum of bottom-up and topdown studies. In animal model studies, concentrations used are usually higher (mM range) than those used in the in vitro studies (μM range). While a wide range of concentrations can in principle be used in animal models, typically higher concentrations are chosen to ensure physiological readouts are possible and to restrict the number of animals to an ethically acceptable degree. Concentrations in human studies are constrained by the cohort access (in the case of observational studies) or by safety concerns (in the case of interventional studies). Further, while in human and animal models the compounds pass through host digestive processes before they reach the bacteria in the gut, in in vitro studies the bacteria are directly exposed to the compounds.