Linking dietary patterns with gut microbial composition and function

ABSTRACT

Emerging insights have implicated the gut microbiota as an important factor in the maintenance of human health. Although nutrition research has focused on how direct interactions between dietary components and host systems influence human health, it is becoming increasingly important to consider nutrient effects on the gut microbiome for a more complete picture. Understanding nutrient-host-microbiome interactions promises to reveal novel mechanisms of disease etiology and progression, offers new disease prevention strategies and therapeutic possibilities, and may mandate alternative criteria to evaluate the safety of food ingredients. Here we review the current literature on diet effects on the microbiome and the generation of microbial metabolites of dietary constituents that may influence human health. We conclude with a discussion of the relevance of these studies to nutrition and public health and summarize further research needs required to realize the potential of exploiting diet-microbiota interactions for improved health.

Diet as a major driver of microbial composition and function

The gut microbiota, which encompasses the trillions of organisms residing in the human gastrointestinal tract, is emerging as a major factor in the development of disease. These organisms assist with digestion, protect against invading organisms, and regulate metabolism and immunity.^1,2 A disruption in these microbial functions has been associated with local and systemic disease and development of autoimmune disorders. Proposed mechanisms range from generation of bioactive metabolites to inducing systemic low-grade inflammation via Toll-Like receptors (TLR).^3,4 While genetics, mode of delivery at birth, physical environment, age, stress, and other factors can influence the dynamics of the gut microbiota, diet may be the single most important driver of gut bacterial composition and function.⁵ In as little as 24-hourstransient changes in gut microbial composition and functional adaptations can be observed in response to dietary changes.^6,7 In addition, habitual dietary patterns drive the establishment of stable, dominant microbial networks.⁷ These networks, or enterotypes, have become a convenient model for classifying an individual's gut bacterial profile, and may eventually serve as an indicator of disease risk or of microbial plasticity and the ability to respond to targeted dietary interventions. Dominance of plant or animal-based foods in the diet appears to be a predictor of enterotype classification, and these dietary patterns can typically be described as high fiber (plant-based foods) or high fat and protein (animal-based foods). Therefore, we have focused this review on the broad effects of these dietary patterns and their associated macronutrient composition on gut microbial community dynamics and function in adult humans. In addition, we will review interactions between the gut microbiota and phytochemicals. This is an exciting emergent area of research that may shed light on the widely variable bioavailability and bioactivity of these plant-based compounds. As we continue to uncover the importance of a healthy microbial community and the far reaching impacts of microbial dysbiosis on human health, the ability to optimize the gut microbiome through dietary strategies is critically needed.

Plant vs. animal-based diets

Numerous studies have shown that vegetarians compared with omnivores following a western diet exhibit a lower risk for various chronic diseases including obesity, hypertension, dyslipidemia, coronary heart disease, stroke, diabetes, and certain cancers. For example, non-Hispanic white, Hispanic, and African-American vegetarians compared with their same race/ethnicity omnivorous counterparts exhibit less obesity and hypertension and have more favorable lipid profiles.^8-12 Longitudinal studies have shown vegetarians to exhibit lower mortality rates for cardiovascular disease and certain cancers in comparison to the general population.^13,14 While vegetarians likely benefit from other health-related behaviors associated with lower disease risk, experimental studies that attempt to control for such confounders still show that changing from a western to a vegetarian dietary pattern reduces several disease risk factors.¹⁵ Identifying reasons for the lower disease burden among vegetarians has been the source of both study and speculation. Lower intakes of total fat, saturated fatty acids, dietary cholesterol, and animal proteins among vegetarians are thought to be important features of this dietary pattern linked to lower risk for chronic disease. However, there are many differences in the vegetarian and omnivorous dietary patterns beyond the distribution of these dietary macronutrients. In comparison to those consuming a western diet, vegetarians tend to ingest more dietary fiber, phytochemicals, potassium, magnesium, ascorbate, folate, and omega-6 polyunsaturated fatty acids, and have lower intakes of sodium, iron, and vitamin B-12.¹³ While these differences in dietary macro and micro-nutrient intakes can directly influence chronic disease risk, consumption of a plant-based versus western-type dietary pattern also results in distinct gut microbial communities,⁷ which in turn, via microbial metabolites may influence processes that are linked to chronic disease pathology.

While the effect of an exclusively vegetarian diet on the gut microbiota and microbial metabolites is not well-studied, animal models and human international studies have provided important insight into plant-based vs. animal-based dietary patterns. For example, among 60 mammalian species studied, microbiota clustered according to dietary patterns, and host bacterial diversity was highest among herbivores, followed by omnivores, and then by carnivores.¹⁶ Several observational studies of humans residing in different geographical areas with widely differing dietary patterns have also revealed important findings. The composition of the gut microbiota of children living in the African village of Burkina Faso was found to be distinct from that of children in Florence, Italy.¹⁷ While genetics, environment and other factors could account for these differences, the microbiota of unweaned children from both populations clustered together, suggesting that diet was the primary driver. The diets of the African children were largely vegetarian, with significant contributions of calories from millet and sorghum; this dietary pattern was associated with almost 50% greater enrichment of Bacteroidetes (Kruskil-Wallis, P = 4.80 × 10⁻⁴), including the unique appearance of the genera Prevotella and Xylanibacter, both of which are well equipped to harvest energy from indigestible plant polysaccharides. Likewise, Ou et al.¹⁸ found that among rural African (NA) adults compared with African Americans (AA), bacteria capable of polysaccharide degradation and fermentation to short chain fatty acids (acetate, propionate, and butyrate), including Prevotella (NA = 8.2 × 10^10, AA = 3.5 × 10¹⁰; P = 0.011) and Clostridium clusters IV (NA = 5.1 × 10⁹; AA = 2.9 × 10⁹; P = 0.032) and XIVa (NA = 9.5 × 10⁹, AA = 5.1 × 10⁹; P = 0.049) were considerably more abundant. On the other hand, among the African Americans, who exhibited greater meat protein consumption, there was an abundance of bacteria capable of proteolytic fermentation with greater generation of branched chain fatty acids (BCFA) from amino acid degradation (isobutyrate: NA = 1.22, AA = 1.73; P  = 0.02; isovaleric acid: NA = 0.33, AA = 1.49 ± 0.19 μmol/g feces; P  = 0.0002; Mann-Whitney U test). Finally, in a third study, children and adults from the Amazonas State of Venezuela and from Malawi, where maize is eaten in large quantities, showed markedly greater fecal bacterial diversity compared with that of US children and adults (Americans∼1,200 OTUs compared with between 1,400–1,600 OTUs in Amerindians and Malawians), as well as up to 2.5x increase in amylase activity required for starch digestion, most likely the result of their higher maize intake.¹⁹ While these studies point to the important influence of dietary patterns on gut microbial composition, there are numerous non-dietary environmental factors that could also contribute to geographical differences in gut bacteria including hygiene and sanitation, soil composition, climate, etc. The nature of these international observational studies precludes controlling for such confounding factors.

Studies of vegetarians and non-vegetarians residing within similar geographic areas reduces confounding by other environmental factors. In general, these studies support findings from the international observational studies discussed above. Vegetarians compared with omnivores tend to exhibit greater bacterial diversity and richness and greater ratios of Prevotella to Bacteroides. For example, in a comparative study of vegans, lacto-ovo vegetarians, and omnivores, Zimmer et al.²⁰ used culture-based methods and found lower counts of Bacteroides spp., Bifodobacterium spp, E. coli, and Enterobacteriaceae (P = 0.001, P = 0.002, P = 0.006 and P = 0.008, respectively) in vegans compared with the non-vegetarians. Vegans were also found to exhibit lower stool pH compared with non-vegetarians (P = 0.0001), presumably a reflection of greater production of short chain fatty acids via carbohydrate fermentation.

The gut microbiota composition changes rapidly in response to dietary adaptations. Using an experimental protocol, David et al.⁶ found rapid and consistent shifts, indicated by increased b-diversity compared with baseline (q < 0.05, Bonferroni-corrected Mann-Whitney U test), induced by 5-day ingestion of an animal-based diet, while the plant-based diet produced less profound effects. The animal diet was high in fat (69% of total calories) and protein (30% of total calories) and almost entirely void of carbohydrates, including dietary fiber. It induced gut bacterial taxonomic shifts and transcriptional responses characteristic of carnivorous mammals, with higher concentrations of bile-tolerant bacteria (presumably due to the extremely high fat intake known to increase bile acid secretion) and greater proteolytic activity and amino acid fermentation as determined by ∼2-fold higher fecal concentrations of short branched-chain fatty acids (isovalerate and isobutyrate). The opposite occurred on the plant-based diet (26 g fiber/1,000 kcal), with gut bacterial and transcriptome responses in line with herbivorous mammals (e.g. increases in fecal short chain fatty acids—acetate (∼2-fold higher) and butyrate (∼1.5-fold increase)—reflecting carbohydrate fermentation), although the magnitude of this response was less pronounced than that produced in the opposite direction by the animal-based diet. Note that these were extreme and abrupt dietary changes, which may have contributed to the somewhat discordant findings relative to a recent comparative study of vegans and omnivores by Wu et al.²¹ The latter found few dietary group differences in gut bacterial composition, and despite large differences in dietary intake of fermentable substrates between groups, i.e. higher indigestible carbohydrate in the vegans, there were no differences in fecal short chain fatty acids. They suggest that individuals residing in westernized communities are constrained by a more ‘restrictive’ gut microbiota with a limited ability to engage in energy harvest via fermentation of indigestible carbohydrates.

Foods of animal origin contain higher amounts of choline and L-carnitine, which have been linked to higher risk for cardiovascular disease as a result of their conversion to trimethylamines by gut bacteria, absorption into portal circulation, and conversion to trimethylamine N-oxides (TMAO) in the liver. While the mechanism(s) by which TMAO increases cardiovascular risk needs further clarification, it has been shown that TMAO reduces reverse cholesterol transport and bile acid synthesis^22,23 potentially attenuating the normal route of intestinal cholesterol elimination. Vegans and lacto-ovo vegetarians have been shown to have negligible postprandial plasma TMAO concentrations in response to an L-carnitine meal challenge.²³ Thus it appears that the lower CVD risk associated with a plant-based diet could result, in part, from lower circulating TMAO.

While there are some inconsistent findings among dietary studies and there is significant inter-individual variability among study participants within similar dietary patterns, taken together the observational and experimental studies comparing plant and animal-based diets suggest there are substrate-induced differences in gut microbial composition and metabolites. In particular, diet-induced differences in microbial metabolites such as SCFA, BCFA, secondary bile acids, and products of protein degradation all have the potential to modulate the host environment for disease prevention or promotion. Likewise, certain members of the microbial community can promote or attenuate immune responses and inflammation. Thus, microbiome differences resulting from these dietary patterns may contribute to the lower chronic disease risk seen with plant-based vs. western diets. As these differences between plant and animal-based diets are likely driven by differences in macronutrient (fat, protein, fiber) composition and phytochemical intake, these components are discussed in more depth in the following sections.

Dietary fats and protein

Few human studies have specifically examined the role of dietary fats and protein consumption on the composition and function of the gut microbiota. Increases in a particular macronutrient are typically associated with decreases in other macronutrients unless the diet is strictly controlled, and the combined effects of these dietary changes are responsible for the resulting alterations in gut microbial populations. Despite the difficulties of studying macronutrient effects in isolation, there is evidence to support that dietary fat and protein consumption elicit both compositional and functional changes to the gut microbiome. The partitioning of individuals into enterotypes appears to be driven by whether their primary dietary patterns include high complex carbohydrate (Prevotella) or high fat/protein (Bacteroides) consumption.⁷ More specifically, the Bacteroides enterotype was most strongly correlated with reports of frequent consumption of animal protein and saturated fat. Wu et al.⁷ implemented a short-term feeding study (CAFÉ study) where some individuals were randomized to a high fat/low fiber diet (38% fat, 35% carbohydrate, 27% protein) for 10 d while others were given a high fiber/low fat diet (13% fat, 69% carbohydrate, 18% protein). Although specific taxa changes varied between individuals, the high fat diet slowed intestinal transit time by as much as 3 d. Metagenomic analysis suggested that functional shifts, including greater protein export (P  = 0.022) and lipoic acid metabolism (P  = 0.045), were also associated with the high fat diet. As mentioned in the previous section, David et al.⁶ showed a rapid shift in gut microbial community composition and increased populations of Alistipes, Bilophila, and Bacteroides with 5-day consumption of a high fat/protein diet, which they hypothesized to be a result of increased bile secretion. Specific effects of fat on the gut microbiota will undoubtedly be dictated by the types of fats consumed, as was demonstrated in a recent rodent study. Animals fed lard showed increases in similar microbial genera as described in the previous human study and displayed signs of metabolic dysfunction, while animals fed fish oil showed increased levels of lactic acid bacteria and were protected from metabolic dysfunction.²⁴

Diets high in fat interact in various ways with the gut microbiota to facilitate the translocation of bacterial lipopolysaccharides (LPS), which contribute to generation of chronic inflammation. LPS can be incorporated into lipid micelles formed during fat digestion, and certain gut microbes may be important in regulating this process. Studies in zebrafish suggest that gut bacteria in the phylum Firmicutes can increase the number and size of lipid droplets formed in the intestinal epithelia to facilitate lipid absorption.²⁵ Bacterial modification can also alter activity and re-absorption of bile acids, which are important in fat digestion.²⁶ Finally, these modified bile acids can act as agonists or antagonists of FXR, a central regulator of lipid transport and metabolism.²⁷

The effects of high protein consumption (without concurrent high fat) on gut bacteria are not well studied, but are of increasing importance because of the current popularity of high protein diets. Obese men consuming a weight maintenance diet (85 g protein, 116 g fat, and 360 g carbohydrate/d) and two high protein weight loss diets (medium carbohydrate: 139 g protein, 82 g fat, 181 g carbohydrate/day; and low carbohydrate: 137 g protein, 142 g fat, 22 g carbohydrate/day) in a crossover design showed a marked decrease in total bacteria (maintenance diet = 10.52 log10/gram; high protein diets = 10.30 log10/gram) and reduced proportions of select butyrate producing organisms, which likely reflects the reduced carbohydrates rather than the increased protein.²⁸ However, an examination of fecal metabolites from the high protein diets showed increased branch chain fatty acids (BCFA; ∼2.1 mmol/L in low carbohydrate diets vs 1.6 mmol/L in maintenance diet) Phenylacetic acid and N-nitroso compounds were about ∼3–5x higher in high protein diets, indicating a shift in microbial metabolism to protein fermentation. These metabolites, and the concurrent decrease in bacterial SCFAs resulting from reduced carbohydrate fermentation, may create a gut environment associated with inflammation and increased risk of colorectal cancer.²⁹ Although these metabolite changes could indicate a negative impact of high protein diets on the gut microbiota, the overall influence of high protein diets may be dependent on other host factors. For example, elite professional athletes had lower inflammatory markers, improved metabolism, and greater microbial diversity compared with both low and high BMI control males and the increased bacterial diversity was positively correlated with protein consumption (correlation coefficients = 0.24–0.43).³⁰ Decreased microbial diversity has been associated with colorectal cancer³¹ and other diseases.³² These seemingly opposing effects of high protein diets in the discussed studies suggest that the protein-diet interactions are modulated by factors such as host body composition and exercise intensity. The types and amounts of fats consumed in each of these studies were also likely important in the overall effects on the gut microbiota.

Carbohydrates and fiber

Carbohydrates as a chemical class have vastly different effects on gut microbiota, which is largely determined by the ability of the host to access the carbohydrate for energy. Relative to bacteria, humans possess a very limited number of enzymes for breakdown of dietary carbohydrates.³³ These enzymes function to break down complex molecules with appropriate monomeric linkages into simple mono and di- saccharides, which are absorbed in the small intestine. Designation of a carbohydrate as ‘dietary fiber’ is made when polymers of three or more units pass undigested from the small intestine into the colonic environment.^34,35 Gut microbiota rely on these recalcitrant fibers for energy, which they target for disassembly with a combined toolkit of thousands of enzymes.^36-39 Dietary fiber as a combined chemical group represents one of the most diversely different chemical classes.⁴⁰ Chemical structures include thousands of forms that vary in complexity and can be broken apart only with enzymes that pair appropriately with specific sugar composition, linkages and chain length.⁴¹ As such, dietary fiber represents enormous potential for modulation of gut microbiota based upon chemistry and accessibility of specific dietary fibers to particular microbial groups and consortia.⁴²

Recent research by Sonnenburg et al. sought to clarify the importance of microbiota-accessible carbohydrates (MACs) to gut microbial communities through prolonged deprivation.⁴³ Germ-free mice lacking resident microbiota were populated with a human fecal sample then assigned to either a high (LabDiet 5010) or low MAC (Harlan TD.86489) group. After seven weeks, mice on a low MAC diet had decreased abundance of 60% of the gut microbial taxa and nearly half of these taxa remained significantly less abundant six weeks after returning to a high MAC diet. The reduction of diversity primarily occurred in two bacterial taxa, Bacteroidales and Clostridiales. Further experiments in this same study revealed that the partial recovery of diversity when returning to the high MAC diet did not transfer to offspring, which experienced compounded diversity loss with each subsequent generation. Earlier investigations into effects of very low carbohydrate diets (approximately 20 g/day) on gut microbiota have suggested decreased abundance of butyrate-producing bacteria including Roseburia, Eubacterium rectale, and Bifidobacterium and a corresponding decrease in butyrate production.^28,44,45 Increased intake of dietary fiber has not been shown to have a bifidogenic effect, but has been associated with an increase in gut microbial richness and/or diversity especially in individuals with reduced diversity initially.^46,47 Additionally, dietary interventions that augment fiber intake have suggested altered gut microbial metabolism including increased glycan degradation⁴⁷ that may be dependent on the chemical composition of the fiber being consumed.⁴⁶ Similarly, overall gut bacterial gene diversity increased by 25% in obese and overweight subjects on a low calorie diet supplemented with soluble fiber.⁴⁸

Long-term patterns of dietary fiber consumption can also shape the overall bacterial community type. As previously discussed, enterotype assignment to the Prevotella group is associated with high-fiber diets.⁷ Similarly, native rural Africans showed an enrichment of Prevotella spp relative to Bacteroides-dominated African Americans, who consume less resistant starch and more meat and fat.¹⁸ O'Keefe et al. studied effects on gut microbiota when fat and fiber content of native African and African American diets were swapped for 2 weeks,⁴⁹ such that African Americans consumed high-fiber, low-fat rural African diets and vice versa. While changes in abundances of Bacteroides or Prevotella genera were not noted in either group, the high-fiber, low fat diets resulted in enrichment of bacterial genes for butyrate production and a decrease in genes for secondary bile acid synthesis.

Human microbiota changes associated with intake of specific dietary fibers with varying chemical characteristics were recently reviewed by Graf et al.⁵⁰ and also by Hamaker et al.⁴¹ This survey of current literature supports the idea that dietary fibers of varying chemical composition induce different changes to gut microbiota. For example, resistant maltodextrin and hydrolyzed arabinoxylans were associated with increased Bacteroides^51,52 after 21 or 24 d in two different crossover design studies in healthy adults. However, long chain inulin decreased Bacteroides and also Prevotella after 21 d in the same population type with a similar crossover design.⁵³ Therefore, when interpreting results across studies involving increased intake of dietary fiber it's important to make note of the types of fibers or foods consumed by study participants.

Starches

Dietary polysaccharides can be divided into starch and non-starch polysaccharides (Fig. 1). Humans are able to digest some starches, such as amylopectin and amylose, which starts digesting in the mouth with amylase found in saliva. Starches that cannot be broken down by human digestive enzymes enter the colon mostly unmodified from their ingested form and are subsequently fermented by gut microbiota. This component of dietary fiber is termed resistant starch and is categorized into four subtypes based on differing physicochemical properties. At least two human clinical trials investigating consumption of resistant starch have been conducted to date and both showed increases in Ruminococcus bromii and Eubacterium rectal.^54,55 One in vitro study investigated cooperative degradation of resistant starch by co-culturing R. bromii, E. rectale, Bacteroides thetaiotaomicron, and Bifidobacterium adolescentis. The results highlight R. bromii as a keystone species that initiates degradation of resistant starch and produces byproducts that are more easily used by other gut species.⁵⁶ Other results from this research varied widely and seemed to depend upon the type of resistant starch consumed.^41,50

Figure 1.

Humans only possess digestive enzymes for breakdown of a subset of dietary polysaccharides, which can be divided into starches and non-starch polysaccharides. Recalcitrant components are defined as dietary fiber and provide substrates for gut microbial fermentation in the colon.

Non-starch polysaccharides

Dietary fiber that does not fit within the definition of starch is termed non-starch polysaccharides and includes polyfructans, cellulose, hemicellulose, and pectins. These fibers are typically complex and diverse with some categories having both soluble and insoluble components. Oligosaccharides that resist human digestion are made up of three to ten monosaccharide units, are frequently grouped together as polyfructans, and are named according to the dominant sugar type. Galactooligosaccharides (GOS) are best known for being an important component of human breast milk.⁵⁷ Clinical trials investigating GOS consumption in infants⁵⁸ adults,^52,59 and older (age > 50) populations all showed increased Bifidobacterium, which was replicated in vitro.²⁸ Some results suggest that Bacteroides and Clostridium may also be reduced with GOS, while F. prausnitzii may be increased. However, these results could not be confirmed across studies due to inter-individual variability and inconsistency of technical approaches.^50,60 Fructooligosaccharides vary by fructan chain lengths from two up to nine, which can greatly influence study outcomes.⁵⁰ In vitro co-culture experiments, including different species and strains of Bifidobacterium and butyrate-producing Roseburia spp, suggest that the species and strains present in the microbial community are also important.^61,62 Bacterial cross-feeding likely plays a role in FOS degradation and further research is needed to determine how these relationships determine bifidogenic and/or butyrogenic outcomes. Xylo-oligosaccharides, predominated by xylose monomers, have also shown bifidogenic and butyrogenic effects relative to resistant maltodextrin ⁶³ Longer chains of fructooligosaccharides contain chains of up to sixty fructose monomers and are classified as inulin. Most studies involving inulin included co-administration of other fibers.^64-66 However, one human clinical trial involving consumption of 10 g/day longer chain inulin extracted from globe artichoke (Cynara scolymus) increased Bifidobacterium spp (2.8-fold; P < 0.05), Lactobacilli/Enterococci (2.4-fold; P < 0.01), and Atopobium spp (2.8-fold; P < 0.05) and decreased Bacteroides/Prevotella spp (1.77-fold; P < 0.05).⁵³ Results from in vitro studies ^67-69 and animal models ^70-72 suggest that cross-feeding of the gut microbial community occurs during inulin degradation and suggest that E. rectale may be a primary degrader and that some butyrate-producing bacteria (R. intestinalis and A. caccae) can utilize smaller inulin fragments.⁷²

Cellulose and hemi-cellulose

Cellulose has a particularly recalcitrant structure involving hydrogen bonding and microfibrils that interact with proteins and pectin to limit fermentation.^73,74 Fermentability of dietary celluloses in the human gut varies by both food source⁷⁵ and gut bacterial composition⁷⁶ Gut bacteria that degrade cellulose can be divided into methane-producers, predominately Bacteroidetes, and non-methane producers, predominately Firmicutes. The species that degrade cellulose during digestion depends upon the specific cellulose structure being consumed and may include Clostridium spp, Eubacterium spp, Ruminococcus spp^77-79 and Bacteroides spp⁸⁰ Both cellulose and hemi-cellulose are components of most plant cell walls and hemi-celluloses can be further divided into arabinoxylans, xyloglucans, β-glucans, glucomannans, and galactomannans.⁴¹ The molecular size of hemi-celluloses plays a role in their effects on the gut bacterial community. As such, these components are often hydrolyzed to smaller and more consistently sized component fragments before supplementing. Two human studies have examined effects of consumption of hydrolyzed arabinoxylans, which form arabinoxylooligosaccharides. Both studies had a crossover design and were conducted over a period of 3 weeks, but effects on gut microbiota were different. One study, which used fluorescent in situ hybridization (FISH) to investigate microbial changes, showed increased Eubacterium rectale, Roseburia/Eubacterium, Faecalibacterium prausnitzii, and Bacteroides spp The other study used real time polymerase chain reaction (RT-PCR) and found increases in Bifidobacterium with neither study finding decreases in any of the targeted bacterial groups. Relative to arabinoxylans, limited research has been done investigating effects of β-glucans on gut microbiota. One in vitro study found that hydrolyzed β-glucans favored growth of the Bacteroides-Prevotella group.⁸¹ So far, animal models have primarily been used to investigate pectin influence on gut microbiota. Studies supplementing 6.5–7% pectin in rats increased abundance of Bacteroides spp after 3 weeks of citrus pectin,⁸² but increased abundance of Anaeroplasma, Anaerostipes, and Roseburia with apple pectin.⁸³ Consuming apple pectin also decreased abundance of Alistipes and Bacteroides spp⁸³ These results suggest that the structure of pectins is also a determinate of the effect on gut microbiota and that it varies according to the food source.^82,83

Whole grains

The human diet rarely, if ever, includes chemically distinct fiber structures, but rather includes a complex mix of these structures as a part of the food we eat. As such, it is important to consider the effect of consuming these fibers in unique forms and combinations found in various foods and food products. Studies involving various cereal grains in humans show them to be bifidogenic. For example, two different crossover design studies lasting for three weeks showed increased Bifidobacterium spp, one using whole grain maize⁵³ and the other using whole grain wheat cereal.⁸⁴ Consuming maize cereal also increased abundance of Atobium spp and wheat cereal increased abundance of Lactobacilli/Enterococci. Two different studies, one supplementing with whole grain barley ⁵⁴ and the other with heat-stabilized rice bran⁸⁵ observed increased abundance of Bifidobacterium. A follow-up randomized clinical trial supplementing heat-stabilized rice bran (30 g/day) in colorectal cancer survivors did not find increased abundance of Bifidobacteria, but rather saw the largest increases in Bacteroides (20-fold increase; P = 0.043) and Lachnobacterium (23-fold increase; P = 0.039)⁴⁶ among many other changes, and may have reflected lower baseline populations of Bifidobacterium spp in this population. Bifidogenic effects were also observed in a study where healthy adults consumed apples daily for 2 weeks⁸⁶ or in different study where bananas were consumed daily for 60 d.⁸⁷ Decreases in abundance of the Clostridium group were noted in studies supplementing rice bran,⁴⁶ chickpeas,⁸⁸ apples,⁸⁶ mushrooms⁸⁹ and low flavonoid fruit and vegetables.⁹⁰ In contrast, increased abundance of Clostridium was observed in one human study supplementing wheat bran⁵³ after three weeks and another supplementing navy bean powder⁴⁶ after four weeks. The types of fibers contained in these foods combined with a summary of changes on gut microbiota listed in Supplemental Table 1.

Identifying effects of dietary fiber on gut microbiota is complicated by differences in chemical structures, other food components, host variation, and study methodologies. Regardless, some similar patterns are observed across studies including frequent increases in Bifidobacterium spp, Ruminococcus spp, Eubacterium spp and Faecalibacterium prausnitzii. The metabolites produced by fermentation of these fibers/whole foods often dictate whether observed microbial changes will impart positive or negative health effects as well as overall state of energy balance in the host. For example, SCFAs are a major by-product of fiber fermentation, and although they are critical to intestinal health and can positively influence host metabolism, they can also contribute to increased energy harvest from the diet.⁹¹ Additionally, polyphenols and other phytochemicals can be bound in ligno-cellulosic matrices and bacterial fermentation may alter their bioavailability and bioactivity, as discussed in the following section. Future research should focus on more detailed characterization of specific fibers found in whole plant foods and associations with microbial changes including these products of microbial metabolism.

Phytochemicals

“Phytochemical” is a term used to describe chemical compounds naturally occurring in plants which don't satisfy the definition of macro or micro-nutrients. In addition to the indigestible polysaccharides previously discussed, plants provide phytochemicals that include phenolic compounds, terpenoids, glucosinolates, chlorophylls, βlains, amines and other chemical compounds. One class of phytochemicals that has been studied extensively is the polyphenols. Phenolic/polyphenolic compounds are the second most abundant class of plant secondary metabolites and they act as antibiotics and antioxidants, as well as modulating numerous host processes via action as signaling molecules. While they are generally poorly absorbed in the host gastrointestinal tract, interactions with gut bacteria, particularly hydrolysis of glycosides, may increase their bioavailability.⁹² This section will be devoted to the findings on phenolic compounds and their impact on gut microbiota and metabolome, and will include findings from pre-clinical and in vitro studies due to the paucity of human intervention studies to date.

Monophenols and aromatic acids

Thymol and carvacrol, isomeric monoterpene phenols found in thyme have been studied for their potential antibiotic activities. Studies in broiler chickens indicate that thymol and carvacrol reduce the numbers of Campylobacter jejuni,⁹³ and Salmonella, spp, Escherichia coli and Clostridium perfringens.^94-97 Moreover, these studies reported that commensal bacteria were largely unaffected by these compounds. In vitro studies have demonstrated that thymol and carvacrol are not metabolized by human gut microbiota, suggesting they are likely to be absorbed in the upper gastrointestinal tract ⁹⁸ or may be excreted in the feces.

Chlorogenic acid, an aromatic acid, is present in high quantities in coffee and tart cherries. As with many other phenolic compounds, Bifidobacterium and, to a lesser extent, Lactobacillus have been reported as potentially involved in bioconversion of chlorogenic acid.^99-102

Flavonoids

A detailed review of bacterial species involved in flavonoid conversion has been published recently.¹⁰³ Among the flavonols, quercetin has been the most extensively studied. It appears that a wide array of bacteria (E. coli, Bacteroides fragilis and several lactic acid bacteria) are able to degrade quercetin.¹⁰⁴ Furthermore, studies in rats indicated that quercetin intake relieved the gut microbial dysbiosis induced by high fat diet consumption, possibly leading to reduced weight gain.¹⁰⁵

Isoflavonoids

Isoflavones have received some attention for their purported health benefits and their natural abundance in soy and other Fabaceae. Isoflavones have been shown to be converted by gut bacteria to equol, which reportedly has more potent health benefits than the parent compounds.¹⁰⁶ Specifically, Aldercreutzia equolifaciens, Slackia isoflavoniconvertens, Slackia equolifaciens, and Lactococcus garvieae have been identified as equol producers from daidzein, but only S. isoflavoniconvertens has been shown to produce equol from daidzein and genistein in vivo.¹⁰⁷ Daidzein can also be converted to O-desmethylangolensin, and this metabotype has been shown to be potentially associated with obesity.^108,109 Isoflavones are mostly present in glycoside forms, but aglycone forms are released by gut bacteria, which may increase their bioavailability in host tissues. In particular, the less common deglycosylation of isoflavone C-glycoside (puerarin and apitregin) is performed by Streptocococcus and Enterococcus strains.

Stilbenoids

Resveratrol is naturally found in several foods (peanuts, grapes, berries), but has mainly attracted interest as purportedly being responsible for the cardio-protective effects of red wine. However, there is currently insufficient evidence to support health claims. ^110,111 Early and recent reports indicate that resveratrol stimulates growth of Bifidobacterium and Lactobacillus in mice gut microbiota.^112,113 Moreover, Slackia equolifaciens and Adlercreutzia equolifaciens, have also been shown to produce dihydroresveratrol from trans-resveratrol.¹¹⁴

Tannins

Urolithins were the first microbial metabolites of dietary polyphenols, namely ellagitannins, to be attributed with more health beneficial properties than their parent polyphenol.^115-117 A study showed that elagitannins from pomegranate stimulates Bifidobacterium and Lactobacillus (while inhibiting growth of Bacteroides, Clostridia and Enterobacteriaceae) in stool slurries, but those genera did not produce urolithins.¹¹⁸ In the companion in vivo study, it was found that Akkermansia are drastically more abundant in individuals producing urolithins compared with non-producers suggesting a role for this genus in the breakdown of ellagitannins.¹¹⁹ However, the specific bacterial genera and metabolic pathways involved in conversion of elagitannins to urolithins are still unknown.¹²⁰

The other group of tannins that has been studied extensively is proanthocyanidins. Again, there is little information as to which groups of bacteria are involved in proanthocyanidin bioconversion, but Bifidobacterium and Lactobacillus increases have been reported in response to pure or anthocyanidin-rich fruit fermentation in vitro and in animal models.^121,122 Diverse microbial metabolites from proanthocyanidins have been reported including phloroglucinol and benzoic acid derivatives,¹²¹ gallic, syringic and coumaric acids.¹²³

Polyphenol-rich foods

A significant portion of the research on the effect of dietary polyphenols on gut microbiota and metabolome has been conducted with polyphenol-rich foods rather than specific polyphenol classes. Tea contains a mixture of polyphenolic compounds dominated by catechins. There is ample evidence that tea polyphenols exert strong antibacterial (pathogens), antitoxin and antiviral effects.^92,124 There is less information on potential positive modulation of gut microbiota, but there are reports of bifidogenic effects,¹²⁴ as well as reversal of diet-induced dysbiosis.^125,126 However, it was also reported that long-term consumption of green tea did not alter the human gut microbiota,¹²⁷ suggesting that the polyphenolic profile of the tea determines the gut microbiota response. Common metabolites from tea catechins include conjugated catechins, valerolactones, valeric acids and other phenolic acids.^124,128,129 Coffee has been shown in vitro to stimulate Bifidobacterium,^100,130,131 and to lead to the production of dihydrocaffeic acid, dihydroferulic acid, and 3-(3′-hydroxyphenyl)propionicacid.⁹⁹ Cocoa and chocolate consumption have also been shown to increase Lactobacillus and Bifidobacterium counts in animal models,^132,133 and human intervention studies.^134,135

Berries including strawberries, raspberries, and blackcurrants have all been shown to increase Lactobacillus and Bifidobacterium levels and lead to the production of elagitannins or proanthocyanidins derivatives.^136-138 Pomegranate appears to have a very similar impact to red berries on the gut microbiota and metabolome.^{118,119,139-141} Freeze-dried mango pulps prevented decreased Bifidobacterium counts observed in the cecal microbiota of mice subjected to a high-fat diet.¹⁴² Increase in Lactobacillus and Bifidobacterium counts were also observed in mice that were fed apples enriched in flavonoids; ¹²² however, this effect could arguably be due to the high amount of dietary fiber as previously discussed. Citrus contains large amounts of hesperetin, naringenin, and ferulic acid. These compounds are fermented to different hydroxyphenyl propionic acids by human gut microbiota in vitro ¹⁴³ and in vivo.^144,145

Understanding the interactions between dietary phytochemical compounds is a burgeoning area of study that promises to increase our understanding of the beneficial health effects of consuming plant-based foods. The reviewed studies clearly indicate bi-directional effects between dietary phytochemicals and gut microbial communities. Because of their poor bioavailability, the health relevance of many polyphenols has been questioned, but their effect on the intestinal environment and gut microbiota is worth further examination. The microbial metabolites of these compounds may have additional bioactivities that have not yet been explored and their interaction with the gut bacteria may increase host access to the aglycones and other metabolites. Properly controlled human intervention studies and observational studies incorporating valid phytochemical biomarkers are needed to determine the importance of these interactions in human health.

Conclusion

Diet is arguably one of the most important forces shaping the gut microbiota. Dietary interventions and targeted nutritional therapies, such as medical foods and dietary supplements, hold great promise for preventing and treating microbiota-associated diseases. However, much research is needed before these possibilities can be fully realized. Effects of specific nutrients, such as differential effects of mono- vs. polyunsaturated fats or various types of fiber, need to be assessed in clinical models. Also, while the enterotype concept is useful for classifying baseline gut microbiota populations, how these enterotypes influence disease risk and response to diet needs to be clarified before enterotyping can have clinical relevance. Individuals are likely to show varying responses to gut-targeted therapies depending on their baseline microbiota characteristics. Human genetic polymorphisms are also emerging as an important factor in determining diet and microbiota interactions, with recent research revealing that the fucosyltransferase 2 (FUC2) gene in women, which is referred to as “secretor” status, influences oligosaccharide content of the breastmilk and subsequent colonization of the microbiota of their infants.¹⁴⁶

The effects of non-nutrient dietary components are another area that need additional study. Phytochemical-microbiota interactions may be an important contributor to the health promoting properties of many plant-based foods. Another important consideration is the impact of “novel” food ingredients on the gut microbiota. For example, artificial sweeteners were introduced a few decades ago and are generally considered attractive alternatives to sugar for individuals with diabetes or those trying to lose weight. However, recent studies suggest that these compounds may unfavorably alter the gut microbiota in ways that encourage insulin resistance and weight gain.¹⁴⁷ Findings such as these suggest there may be a need to reassess the criteria used to evaluate the safety of food additives and ingredients. These and other questions regarding the microbiota-host-diet interactions are likely to be answered in the near future with a new emphasis in the field on integrating “omics” data sets and sophisticated gene and metabolite modeling for a systems-level approach.

Abbreviations

BCFA

branched chain fatty acids

FOS

fructo-oligosaccharide

GOS

galacto-oligosaccharides

LPS

lipopolysaccharide

MAC

microbiota accessible carbohydrates

SCFA

Short chain fatty acids

TLR

Toll-like receptor

TMAO

trimethylamine oxide

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

TLW is supported by funding from the International Life Sciences Institute (ILSI-North America) and Colorado Agriculture Experiment Station.