Cross-Domain and Viral Interactions in the Microbiome

Bacteria and archaea are present along the gastrointestinal tract, with the greatest density present in the colon, and have received much research attention due to their roles in digestion and immune function (24). Unsurprisingly, given the largely anaerobic environment of the gastrointestinal tract, the gut microbiota are primarily facultatively or strictly anaerobic (25, 26). The specific taxonomic composition can vary significantly between individuals, impacted by different lifestyles, diets, and ages, although generally, they are fairly stable over time within the same individual (27, 28). Insights from metagenomics have led to the conclusion that rather than a set of specific taxa comprising a “core microbiota,” there may instead be core functions that can be provided by different bacterial taxa in different individuals (27, 29, 30). However, metatranscriptomics suggests that there is still interindividual variation in transcription levels, which is intermediate between the highly idiosyncratic taxonomic composition and the more conserved functional capacity (31, 32). In the human gut, there appears to be a core metatranscriptome composed largely of housekeeping genes, with a much larger variable metatranscriptome of specialized pathways, suggesting that gut community transcription is context-specific and adaptive to the individual environment (32).

Despite this variation, sequencing, particularly large-scale efforts including the Human Microbiome Project and Meta-HIT, has revealed some common patterns of bacterial composition (27, 33). The human gut is generally colonized by hundreds of species-level bacterial taxa, which typically are dominated by members of only a few phyla; Firmicutes, Bacteroidetes, and Proteobacteria are most abundant, with Actinobacteria and Verrucomicrobia making up smaller proportions (Fig. 1) (25, 27, 30, 34). It should also be noted that the gut microbiota is not a single, homogenous community but instead displays significant three-dimensional organization. First, the gut is comprised of several unique environments, in particular the stomach, the small intestine (divided into the duodenum, jejunum, and ileum), and the large intestine (colon), each of which has different properties and harbors its own community (35). To date, the vast majority of research has focused on the colon due to the comparative ease of obtaining fecal samples and the fact that it contains by far the highest density and numbers of bacteria (24). Second, even within a given compartment, bacteria may differ along the transverse axis, with different populations found in the lumen versus the mucosa (35).

Despite the difficulties in studying the stomach and small intestine, techniques including endoscopy and biopsy have allowed profiling of these microbial communities. In general, the microbial community becomes increasingly anaerobic along the gastrointestinal tract, with the stomach and small intestine containing a greater proportion of facultatively aerobic taxa than the largely anaerobic colon (36). Work in the stomach has frequently focused on the species Helicobacter pylori, given its close association with the gastric mucosa and public health relevance as an organism linked to gastric ulcers and cancers (37–40). However, the presence and levels of H. pylori vary between individuals, and a combination of culturing and amplicon sequencing techniques has revealed that other genera can be found in the gut despite the harshly acidic conditions. While exact findings have differed, Streptococcus has been consistently observed in relatively high proportions, along with genera including Prevotella, Lactobacillus, Rothia, Veillonella, and Propionibacterium (41–45). Additionally, while the stomach lumen certainly contains transient microbes from the mouth and nose, the gastric community was shown to be distinct from either of these groups (43).

The small intestine also contains a distinct community of bacteria, typically containing the genera Streptococcus and Veillonella; other frequently encountered taxa include Escherichia, Clostridium, Turicibacter, and Lactobacillus (46). Like the stomach, it is less hospitable to bacterial life than the colon, with a faster transit time, higher acidity, more antimicrobial molecules, and an influx of bile acids, and therefore, it is less densely populated along most of its length (46). Studies of effluent from subjects with ileostomies suggest that the small intestinal bacterial microbiota tends to be more temporally variable than that of the colon, likely due to an increased short-term sensitivity to dietary intake given the small intestine’s primary role in host nutrient absorption (47). Indeed, metatranscriptomics indicates that the maintenance of the small intestinal bacteria is driven by the rapid uptake and utilization of simple carbohydrates, which could make this population particularly sensitive to the composition of ingested food (48). In particular, the genus Streptococcus expressed genes for these functions at high levels, matching their high relative abundance in the population (48). However, the community is not necessarily consistent along the entire small intestine, and there is evidence that the bacterial composition becomes more similar to that of the colon in the terminal ileum (46).

The colon is more diverse, densely colonized, and anaerobic. Firmicutes and Bacteroidetes make up the majority of bacteria, although Proteobacteria, Actinobacteria, and Verrucomicrobia are typically present in lower proportions. Within these phyla, a number of commonly prevalent bacterial families may be identified, including Bacteroidaceae, Clostridiaceae, Prevotellaceae, Eubacteriaceae, Ruminococcaceae, Bifidobacteriaceae, Lactobacillaceae, Rikenellaceae, Verrucomicrobiaceae, and Enterobacteriaceae (1, 30, 34, 46). Interestingly, within the phylum Bacteroidetes, individuals tend to be dominated by either Bacteroides or Prevotella based on their diet and lifestyle. Studies show that urban subjects eating a “Western” diet high in protein and fat tend to be dominated by Bacteroides, while members of rural communities eating more plant-based, fiber-rich diets are dominated by Prevotella (49). Additionally, the archaeal genus Methanobrevibacter, which feeds on metabolites from other gut microbes and produces methane, is typically found in the human colon and is highly active; along with other less dominant methanogenic archaea, these organisms drive bacterial metabolism by removing hydrogen from the local environment and thereby making polysaccharide fermentation more thermodynamically favorable (1, 30, 32, 34, 50–53).

Furthermore, it has become increasingly recognized that the gut microbial community displays a transverse organization, with a distinct composition in the lumen relative to the mucosa (25, 46, 51, 54). One reason for this is that luminal and fecal samples contain long-term residents alongside transient bacteria and DNA from the digesta, which is less true for mucosal communities. Another reason is that the intestinal mucus, composed of highly glycosylated mucin proteins, provides a distinct niche for certain microbiome members. In the colon, a continuous mucus barrier covers the epithelium, organized into a dense inner layer that blocks most bacteria and a loose outer layer adjacent to the lumen; in the small intestine, there is only a single layer, and it is patchier than in the colon (46, 55, 56). The outer layer is home to a number of bacteria, including primarily mucolytic species such as Akkermansia muciniphila, mucolysis-capable species such as Bacteroides thetaiotaomicron and some Bifidobacterium species, and nonmucolytic (and even asaccharolytic) species that can feed on downstream metabolites from this process (46, 54, 57). There is also an oxygen gradient in the intestines, with high oxygen concentrations at the epithelium relative to the largely anaerobic lumen (58). This gradient tends to favor an enrichment of species that are more aerotolerant closer to the epithelium, including facultative anaerobes and those possessing mechanisms, such as catalase and superoxide dismutase, to deal with oxidative stress (58, 59). Finally, some bacteria have adaptations for penetrating the mucus layers and coming into close contact with the epithelium, such as the segmented filamentous bacteria (SFB) (sometimes known as “Candidatus Arthromitus” or “Candidatus Savagella”), while others can shelter in the crypts of the small intestine or the folds of the proximal colon (59–61). As a result of these factors, a number of studies in humans and animal models have found that the communities and transcripts of the lumen or feces are distinct from those associated with the mucus and/or epithelium in the same individual (46, 51, 57–59, 62–65). Interestingly, even species found in both the lumen and mucus may behave differently based on their location, with work demonstrating differential transcriptional profiles observed between luminal and mucus-associated members of the same species (57).

In contrast to the bacteria in the colon, the oral community displays relatively low interindividual variation (known as beta diversity) but has comparably high levels of diversity within any given individual (or alpha diversity) (30). The oral community is frequently dominated by members of the genus Streptococcus but also contains Prevotella, Veillonella, Haemophilus, Neisseria, Corynebacterium, Actinomyces, and Rothia, among others; it may also contain archaea, including Methanobrevibacter (1, 27, 66). However, like the gastrointestinal tract, there are several distinct regions within the mouth, including the gingiva, tongue, and teeth, that harbor somewhat distinct communities (27, 66). Similarly, the skin does not harbor a single unified bacterial community, and the composition depends on the characteristics of the site sampled, for example, dry skin, oily (sebaceous) skin, or moist skin (1, 27, 67–69). In contrast to the gut, the skin is dominated by Actinobacteria, followed by Firmicutes, Proteobacteria, and Bacteroidetes; common genera include Staphylococcus, Propionibacterium, and Corynebacterium (1, 27, 68–70). In particular, the lipophilic genus Propionibacterium is associated with sebaceous sites (68–70). Additionally, the skin is colonized by the Thaumarchaeota phylum of archaea, possibly involved in ammonia oxidation (14, 71–73). Finally, the vaginal bacterial community is an interesting demonstration of the fact that grouping bacteria at higher taxonomic levels can hide the diversity at lower levels. The vaginal community in most individuals is dominated by the genus Lactobacillus, giving it an apparent low diversity at this level, but the species and strains present are diverse and variable (27). However, recent work has also revealed that a significant subset of individuals possess a more diverse vaginal bacterial microbiota, including Gardnerella, Atopobium, Megasphaera, Streptococcus, and Prevotella (74, 75).

As might be expected given their residence in the gastrointestinal tract, the gut bacterial and archaeal microbiota play an important role in digestion and metabolism. Collectively, gut bacteria possess the ability to extract energy from a wide variety of molecules that are indigestible by the host alone. Generally, these molecules are plant-derived polysaccharides, including fibers and starches, which are broken down into metabolites that can be used by the host or other microbes (76–78). Indeed, metatranscriptomics studies indicate that carbohydrate transport and metabolism are highly expressed functions across individual microbiomes, despite taxonomic variation (79). The importance of including such molecules in the diet is highlighted by studies that suggest that in their absence, gut microbes may instead overdigest the mucus layer, potentially allowing epithelial access to pathobionts (80–82). Additionally, nonfermentative members of the microbiota may form cooperative metabolic networks with the fermenters; for example, methanogenic archaea in the gut remove excess hydrogen from the local environment, driving fermentation by increasing the thermodynamic efficiency of the process (52, 53). In fact, metatranscriptomics studies have indicated that methanogens are particularly active relative to some other members of the gut microbiota (31).

The importance of the gut microbiota in harvesting energy from food can be demonstrated by studies in germfree animals, which lack any microbiota and display metabolic differences from their conventionally raised counterparts. Germfree mice are leaner than conventional mice despite consuming more food on a standard diet, but they lose this phenotype when they are colonized with the gut microbes of their conventional counterparts (83, 84). This effect arises from the reduced capacity of germfree mice to extract energy from food, thereby decreasing the caloric intake from the same amount of food, as well as the ability of the gut bacteria to promote fat deposition by the host (84). Another study found that when on a high-fat diet, germfree mice actually consumed less food than conventional mice, while also displaying increased lipid excretion and less-efficient food utilization. Together, these effects resulted in lower weight gain than in conventional mice, suggesting a degree of resistance to the ill effects of the high-fat diet (85). Recent work further confirms this observation, as germfree mice on a high-fat diet were shown to gain less weight, deposit less epididymal and mesenteric fat, excrete more triglycerides in the stool, and absorb significantly less lipid into the bloodstream than conventional mice; together, these data suggest that gut microbes play an important role in lipid digestion and absorption (86, 87).

Further studies have demonstrated that not all microbiomes are equal. For example, the bacterial microbiota of genetically obese mice have been shown to be more efficient at extracting dietary energy than those of their lean littermates. Obese mice showed an enrichment in bacterial genes for indigestible polysaccharide breakdown, produced more short-chain fatty acids (SCFA), and had lower fecal energy content, suggesting a greater ability to extract energy from their food. Furthermore, transferring microbiota from an obese mouse to a germfree mouse resulted in a significantly larger body fat percentage increase than transferring microbes from a lean mouse (88). In fact, transplanting fecal microbiota from humans has a similar effect; mice given microbiota from an obese human gained more weight and fat than mice given microbiota from the donor's lean twin (89). Interestingly, cohousing both types of mice together led the obese-transplant mice to resemble their lean-transplant counterparts in both bacterial microbiota and body composition, suggesting that the low-fat, high-fiber diet that the mice were provided with selected for the lean-associated microbes (89).

In fact, it is widely recognized that diet is a major factor that influences the makeup and function of the gut microbiota. For example, researchers comparing the gut bacterial microbiota of children in urban Italy and rural Burkina Faso found dramatic differences, including a high prevalence of fiber-digesting taxa and a significantly reduced Firmicutes-to-Bacteroidetes ratio in the African children compared with their European counterparts. Those authors attribute these differences to the high-fiber, low-animal-protein diet of the African cohort, which promotes the growth of bacterial taxa capable of digesting dietary fibers and starches (90). More experimental studies have further demonstrated the importance of diet in the makeup of the gut microbiota. One study found that a high-fat diet led to a reduction in the abundance of Bacteroidetes and increases in the abundances of Firmicutes and Proteobacteria, even in an obesity-resistant mouse model (91), while another found that weight loss in obese humans was associated with increases in the abundances of Bacteroidetes (92). Diet may also interact with microbes in the small intestine to regulate lipid absorption; recent work has found that colonizing germfree mice with jejunal microbiota from mice fed a high-fat diet increases their capacity for lipid absorption even on a low-fat diet, while transferring microbes from mice fed a low-fat diet did not have the same effect (87). Strikingly, Turnbaugh et al. found that switching mice from a low-fat, plant-rich diet to a high-fat, high-sugar diet could change microbial composition and metabolism in as little as a day (93), and the same group demonstrated alterations to the human gut microbiota after only 4 days on a plant-based or animal-based diet (94). As such, there is significant research interest in the microbial contributions to obesity and metabolic disorders, as well as in whether the gut microbiota present a therapeutic target to treat or prevent these conditions. However, these efforts have been generally complicated by conflicting results and difficulty in finding a consistent signature of metabolic disruption across experiments (95, 96).

In addition to simply liberating more energy from the diet, gut bacteria produce important metabolites that may promote host health. Many gut bacteria produce vitamins, particularly vitamin K and several B vitamins, although the amount absorbed by the host relative to the microbiota is unclear. More importantly, many of the gut-resident bacteria produce SCFA, primarily butyrate, acetate, and propionate, as end products of fermentation of undigested fiber, starches, and plant polysaccharides in the colon; in contrast, branched-chain fatty acids, including isobutyrate, methylbutyrate, and isovalerate, can also be produced as amino acid metabolism by-products (77). Acetate is produced by many enteric microbes, including the mucolytic A. muciniphila, Bacteroides species, and Bifidobacterium species. It enters peripheral circulation and is the primary SCFA detectable in blood, and some functions include serving as a fuel source for the liver and muscles and being used in the synthesis of molecules such as cholesterol (77, 97). Acetate can also be used by other gut microbes to produce butyrate (77). Propionate, produced by microbes including members of Bacteroidetes and the Negativicutes class of Firmicutes, is almost wholly metabolized in the liver and has impacts on gluconeogenesis (77, 97). Butyrate, which has received a significant amount of research interest, is primarily produced by members of Firmicutes, such as Faecalibacterium prausnitzii. It is the primary fuel source for the colonic epithelium and has been implicated as an anti-inflammatory influence that helps to maintain intestinal homeostasis (97, 98). The concentrations of SCFA decline along the length of the colon, reaching 70 to 140 mM in the proximal colon and 20 to 70 mM in the distal colon, and SCFA also form a concentration gradient from the lumen outwards; furthermore, they are present at different molar ratios, with acetate being the most abundant, followed by butyrate and propionate at approximately similar fecal levels, although this likely does not accurately represent the ratios in the colon itself due to differences in absorption (77, 96, 97, 99, 100). In addition to their role in host metabolism, SCFA are implicated as important signaling molecules mediating interactions between the gut bacteria and the host immune system, as described in more detail below.

In addition to their role in metabolism, the human microbiota play an important role in the immunity of the host, which must be able to differentiate between commensal and/or symbiotic microbes and potentially pathogenic bacteria. Therefore, there is an important balance that develops, involving a limitation of contact between the microbiota and the local mucosa in addition to immunoregulatory mechanisms, allowing beneficial microbes to persist while preventing autoimmunity or self-damage by the host. The contributions of commensal microbes to immunoregulation form an important part of the “old friends” (formerly “hygiene”) hypothesis (101–105). In short, this hypothesis posits that changes that have occurred in developed nations, including water sanitation, increased usage of antibiotics, higher rates of caesarean sections, more time indoors, and shifts to a low-fiber Western diet, have reduced early-life exposure to and colonization with helminths and beneficial microbes (old friends that humans coevolved with) that help to regulate the immune system, thereby leading to increases in autoimmune and allergic disorders in their absence (101–105). Here, we focus on the bacterial component of this hypothesis, but we discuss the contributions of helminths in a later section.

Humans are colonized with commensal microbes during and shortly after birth and must develop an immune system that can tolerate bacteria at many body sites without losing the ability to defend against pathogens. According to the old friends hypothesis, if there is insufficient exposure to diverse commensal or environmental microbes, it can lead to a failure to properly train immunological tolerance to harmless stimuli and subsequent overreactions to allergens or innocuous microbes. At the same time, many beneficial commensal microbes actively regulate the immune response, helping to prevent inappropriate immune activation to both the microbes themselves and other “bystander” antigens (101, 106, 107). Without this influence, particularly in early childhood, the risk for diseases of immune hyperreactivity, such as asthma, type 1 diabetes, multiple sclerosis, and inflammatory bowel disease (IBD), increases. In fact, adults who immigrate from low- or middle-income nations to high-income nations tend to retain protection against such disorders, but their children or those who immigrate when very young develop these diseases at higher rates more similar to those of indigenes of the new country (104, 108–114).

There are several mechanisms by which key members of the commensal microbiota modulate the immune response. First, the mere presence of gut microbiota is required for proper immune development; studies in germfree mice have revealed a number of immunological irregularities. For example, the microbiota are important for the development of the gut-associated lymphoid tissue (GALT), which allows the uptake and presentation of gut antigens to local immune cells. Accordingly, germfree mice have underdeveloped GALT compared to conventional animals. Specifically, they have small Peyer’s patches with fewer germinal centers, reduced numbers of CD4⁺ T cells in the lamina propria, and low levels of secretory IgA-producing plasma cells (115–118). They also show signs of a T_H2-biased immune system, even in peripheral locations such as the spleen, and a decreased ability to develop oral tolerance to ingested antigens (119–123). They also have increased accumulation of invariant natural killer cells in the colonic lamina propria, although they may be hyporesponsive (124, 125). Generally, such defects can be corrected by colonization with microbes at an early age, but not always in adulthood, supporting the importance of an early-life “critical window” for the microbiota to stimulate normal immune development (125–127). Even certain single species of bacteria can serve to normalize some aspects of immune function in germfree mice; for example, Bacteroides fragilis monocolonization can correct T_H2 bias and promote immunological balance (123), Bifidobacterium infantis can correct oral tolerance defects when administered to neonatal mice (128), and SFB (a lineage within the family Clostridiaceae) can direct balanced T-cell maturation comparable to that of a complete mouse microbiota (129).

The presence of the microbes is also important to the “education” of the adaptive immune system, training it to discriminate between innocuous commensals and harmful pathogens and thereby promoting tolerance of microbiota-derived antigens. A key component of this process is the development of forkhead box P3⁺ (FoxP3⁺) regulatory T cells (T_reg cells), which can suppress effector CD4⁺ T-cell subsets and thereby promote immune tolerance. Traditional T_reg cells arise from the thymus, with the objective of suppressing self-reactivity by the immune system (130–132); while these thymic cells (tT_reg cells) play a role in intestinal homeostasis, there is also an important role for naive T cells recognizing commensal antigens that are induced to differentiate into T_reg cells in the colon (iT_reg cells) (126, 133–136). This occurs in part through the action of tolerogenic CD103⁺ dendritic cells in the epithelium, which preferentially sample the luminal bacteria and favor the differentiation of naive CD4⁺ cells into iT_reg cells (137–141). As might be expected, germfree mice can display defects in their T_reg populations, although they do not lack them entirely (142, 143). The presence of colonic iT_reg cells with a diverse repertoire of receptors recognizing commensal antigens helps to prevent inappropriate responses to the microbiota and other bystander antigens, which have been implicated in the pathogenesis of IBD (144–146). This is thought to be particularly important during early life; the microbiota of humans is temporally unstable for the first several years and is theorized to provide a sampling window for the training and development of immunoregulatory responses (122, 144–146).

In addition to simply serving to educate the adaptive immune system, several types of commensal microbes have been found to actually direct certain immune responses, often promoting tolerance (147, 148). For example, the common gut microbe B. fragilis (of the phylum Bacteroidetes) has been found to activate development of T_reg cells and increase immunoregulatory cytokine production via the molecule polysaccharide A (PSA) (149–151). Accordingly, this molecule has been found to be protective against certain inflammatory diseases in mouse models (149–151). The related species B. thetaiotaomicron may be able to downregulate intestinal inflammation, even in the face of inflammatory challenge, by repressing host NF-κB signaling (152, 153); some other microbes, including Lactobacillus species and nonvirulent Salmonella strains, have demonstrated similar capabilities (153, 154). In addition, members of the class Clostridia (of the phylum Firmicutes) can induce the expansion of thymic T_reg cells and the development of colonic T_reg cells; this effect is at least partially mediated by the production of SCFA, particularly butyrate (140, 155–160). Specifically, the species F. prausnitzii has been found to be anti-inflammatory at least in part via its production of butyrate, inducing T_reg cells and anti-inflammatory cytokine production. Accordingly, it has been suggested to be protective against the development of IBD (161–163). Furthermore, colonization with altered Schaedler flora, a defined mix of eight commensal bacterial species that robustly colonize mice, including Lactobacillus, Clostridium, and Bacteroides species, has been shown to increase the levels of T_reg cells in the colonic lamina propria and promote intestinal immune homeostasis (164, 165). Finally, some Lactobacillus species have demonstrated an ability to drive T_reg development and subsequent interleukin-10 (IL-10) production (166). Contrarily, SFB have been found to associate closely with the mucosa and induce a T_H17 response; the T_H17 response is a generally proinflammatory pathway that can help to protect against bacterial pathogens but potentially contribute to autoimmune pathology (60, 148, 167).

The commensal bacterial microbiota also promote the function of the gut epithelial barrier, the integrity of which is important for preventing inappropriate immune activation and invasion by pathogens. A barrier of mucus, antimicrobial peptides, and secretory IgA serves to keep most microbes at a safe distance (168–173), although some are able to come into fairly close contact with the epithelium (61, 174). The commensal microbiota appear to serve as a stimulus for increased mucus production, as germfree animals have been observed to have impaired mucus production, which can be rescued via colonization with a normal microbiota or even administration of bacterial products, including lipopolysaccharide (LPS) (56, 175–178). The production of butyrate may contribute to this effect, as it has been demonstrated to promote epithelial production of the major mucus component mucin-2 (33, 179–181). Butyrate can also promote epithelial barrier function and integrity (182–184). Additionally, gut bacteria may stimulate the production of IgA and antimicrobial peptides (120). While some of these impacts may seem counterproductive to the gut bacteria, they ultimately help both host and microbiota by maintaining a tolerant, anti-inflammatory environment. Furthermore, some commensal bacteria may be able to use host immune factors such as IgA to aid them in stable gut colonization (185).

Finally, commensal bacteria and archaea provide resistance to host infection with pathogens, a phenomenon termed colonization resistance. Commensal microbes occupy the readily available niches of the sites that they colonize and stimulate the local immune system, preventing potential pathogens from effectively establishing infections. They can compete for nutrients, produce antibacterial or inhibitory molecules, or even kill other bacteria through type VI secretion systems; in contrast, commensal bacteria can also indirectly encourage resistance to pathogenic infection by promoting antimicrobial peptide production, epithelial barrier integrity, and T_H17 responses, as described above. In one example of such immune-mediated competition, the Gram-negative organism B. thetaiotaomicron can stimulate the production of the antimicrobial peptide RegIIIγ, which primarily acts against Gram-positive bacteria (174, 186, 187). Additionally, some interactions require a combination of interbacterial competition and host immune involvement; the probiotic Escherichia coli strain Nissle 1917 can antagonize Salmonella enterica colonization by competing for iron, but only when the host produces the innate immune molecule lipocalin-2 to limit bacterial iron availability (188). As might be expected, germfree mice or antibiotic-treated mice or humans are more susceptible to colonization with certain pathogens, including S. enterica, Clostridium difficile, Klebsiella pneumoniae, and pathogenic E. coli (174, 189).

Given the importance of the commensal bacterial microbiota to immune regulation and colonization resistance, there is interest in using probiotics (specific strains or cocktails of bacterial species) and/or prebiotics (food or nutrients, typically fibers, meant to foster the growth of beneficial bacteria) as therapeutic agents. Of greatest interest are lactic acid bacteria, which are generally well tolerated by humans and are often present in fermented foods. Lactobacillus and Bifidobacterium species are most commonly studied, although some other microbes, including Streptococcus, Lactococcus, Enterococcus, and E. coli Nissle 1917, have been studied as well (190). Both mouse and human studies have examined the potential for probiotics, sometimes in combination with prebiotics, to prevent or alleviate a wide variety of disorders, including antibiotic-related C. difficile infection, IBD, H. pylori infection, atopic disorders, and necrotizing enterocolitis in preterm infants, among others (191–194). In this work are potentially promising results, although many studies are of a small size or have methodological limitations, so it is difficult to draw robust conclusions in some cases. However, large and well-designed studies can demonstrate the potential of pre- and probiotics; for example, Panigrahi et al. included over 4,000 subjects in a randomized, double-blind, placebo-controlled study of a combined pre- and probiotic (“synbiotic”) that showed a reduced risk of sepsis in full-term infants in rural India (195).

The bacterial and archaeal microbiota, particularly within the gastrointestinal tract, perform a number of important functions beneficial to the eukaryotic host. Most directly, they play a major role in digestion, allowing the host to extract energy from dietary components that the host does not possess the capacity to break down. In doing so, the gut microbiota produces SCFA, including butyrate, which serves as a primary fuel source for the colonic epithelium. Furthermore, the microbiota and their metabolites have significant impacts on the development and function of the host immune system. They stimulate innate mechanisms to shield the gut epithelium, protect the host against pathogenic colonization, and direct adaptive immune cell populations, particularly T_H17 and T_reg cells; in fact, the lack of a diverse community in early life may contribute to the development of immunological disorders in the genetically susceptible. In return, the eukaryotic host provides its microbial passengers with a sheltered niche and an array of nutrients, maintaining a tolerant environment despite the huge numbers of nonself organisms found in and on its body.

Primarily using ITS sequencing, there have been a number of studies attempting to catalogue the resident gastrointestinal fungi and determine the composition of a core healthy mycobiome (Fig. 1). ITS sequencing of human feces has had somewhat variable results, although they generally support the conclusion that the gut mycobiome has relatively low diversity and varies significantly between individuals. In contrast to observations of the bacterial gut residents, the number of fungal operational taxonomic units (OTUs) detected pales in comparison to those seen for bacteria, and longitudinal sampling from the same individual over time indicates that the detectable fungi fluctuate significantly. This observation is supported by data from mice, which also have longitudinally variable gut mycobiota despite living under controlled conditions (202, 203).

Despite this, ITS sequencing studies have identified various fungal genera associated with the human gut. Most studies indicate that the healthy human gut contains Candida and Saccharomyces species, with other frequently detected taxa including Malassezia, Debaryomyces, Cladosporium, Aspergillus, Pichia, and Alternaria, among others (202, 204–206). There is also some research into gut-resident fungi in the context of human disease, and there is evidence that Candida itself and fungus-associated antibodies may be related to IBD. Specifically, anti-Saccharomyces cerevisiae antibodies (ASCA), which, despite the specificity of their name, can recognize antigens from other yeasts, including Candida, have been associated with IBD (207–210).

One issue is that there is significant variability between studies, in both the specific taxa detected and their relative abundances. Differential DNA extraction protocols, marker genes, primers chosen, analysis techniques, and sample populations may be to blame, in addition to inherent mycobiome variability. An issue acknowledged by many authors is that it is difficult to differentiate true gut colonizers from food-associated and environmental fungi that may simply pass through; this could potentially help to explain the high variability in fungal composition over time, as it may be influenced by recent diet and environmental exposures (202, 204). For example, while Saccharomyces is frequently found in the fecal mycobiota, S. cerevisiae, known more commonly as baker’s or brewer’s yeast, is used in the production of bread and beer and may present a confounding influence on analysis. The variable and uncontrolled diet of human subjects makes this a difficult question to answer. However, the studies that have been performed suggest that while dietary fungi can be detected at appreciable levels in the fecal mycobiome, they may be metabolically active while in the gut and therefore contributing to the local community, even if transiently (94).

There are also a number of fungi found in the human oral cavity, both on the mucosal surfaces and in subgingival plaque (199). A few groups have used ITS sequencing in humans to identify potential members of a core oral mycobiome, including Candida, Cladosporium, Alternaria, Aspergillus, Fusarium, Cryptococcus, and Aureobasidium (211–213). Unsurprisingly, those studies found that Candida was particularly common, reflecting previous results from culture-based studies (211). Ghannoum et al. also identified Saccharomycetales, Dothioraceae, Teratosphaeria, and Glomus as commonly occurring mycobiome members, while Dupuy et al. identified Malassezia, Irpex, Cytospora, Lenzites, and Sporobolomyces (212, 213). As in the gut mycobiome, differences in core taxa identified may be attributable to study populations, sampling type, extraction of fungal DNA, and methods of filtering and analysis, which were somewhat different between the two studies. Furthermore, food-associated and environmental contaminations are difficult to rule out. Finally, Ghannoum et al. were able to use community-level analysis to determine that subjects’ gender and ethnicity were associated with their mycobiota, with analysis indicating significant differences between samples from females, white males, and Asian males (213).

Fungi have also been identified on the skin, particularly the lipophilic yeast genus Malassezia (200, 214, 215). A comprehensive survey of the skin mycobiota sampled and analyzed the fungi at 14 skin sites of healthy adults (216). Most were dominated by Malassezia, with species differing between sites, although three foot-based sites showed a greater diversity of fungal genera including Aspergillus, Cryptococcus, Rhodotorula, and Epicoccum (216). Similarly, another group found that Malassezia species dominate much of the skin mycobiota, while other genera, including Candida, Meyerozyma, Rhodotorula, Trichosporon, Cladosporium, Aureobasidium, and Alternaria, could also be identified (217). Furthermore, another study demonstrated a similar composition of skin mycobiota in a non-Western population; samples from Chinese participants in Hong Kong were dominated by Malassezia at multiple sites, with other commonly occurring genera including Aspergillus, Penicillium, Candida, and Cryptococcus (218). Finally, recent work found that the skin mycobiome of children is more diverse and variable between individuals and that the domination by Malassezia becomes established by adulthood, potentially due to the activation of sebaceous glands that occurs during puberty (219, 220). There may also be some differences in facial carriage of Malassezia by gender, with women harboring fewer fungi, although this has been suggested to be attributable to the use of cosmetics rather than inherent differences between men and women (220–222).

When discussing the presence of fungi in the female reproductive system, much of the focus has been on vulvovaginal candidiasis, or “yeast infections.” However, the vagina may be colonized by a number of fungi, and opportunistic pathogens such as Candida colonize many women asymptomatically in the absence of an immune or vaginal bacteriome disruption (223, 224). In fact, Candida has been found in multiple studies to be the most frequent colonizer of the vaginal mycobiota, even in healthy individuals. In a study using an 18S rRNA gene clone library to identify fungal taxa in the vaginal microbiota, researchers identified a number of fungal OTUs, including Candida, Saccharomyces, Dothideomycetes, and Paecilomyces species, in healthy subjects (225). More recently, researchers used ITS sequencing in a large study of healthy women. They found that Candida species were the most prevalent, being present in 64.5% of subjects and having a mean relative abundance of 36.9% (226). In addition, they identified OTUs mapping to Cladosporium, Eurotium, and Alternaria present at low abundances, although their analysis was complicated by a large number of unspecified OTUs, which had a summarized relative abundance of 38.6% (226).

In general, work on the human mycobiome has been minimal in comparison to interest in the bacteriome. Much is still unknown regarding the “core” or healthy members of the fungal microbiota at various body sites, although it is clear that Candida species are commonly found across multiple mucosal body sites while Malassezia dominates much of the skin. Interestingly, studies identified the presence of several potentially pathogenic fungal genera, including Aspergillus, Fusarium, and Cryptococcus, suggesting that, like Candida, these taxa may be commonly present in healthy individuals without immune dysfunction. Authors of the studies discussed here are in general agreement that more work is required and that a number of steps should be taken to improve comparability between studies and increase the level of detail that can be obtained from sequencing. Similar to calls in the field of the bacteriome, a unified process for sample collection, DNA extraction, amplicon generation, marker gene sequencing, and sequence analysis has been identified as being potentially helpful in making comparisons between different studies. Particularly with regard to the oral and gut mycobiota, additional steps to minimize the influence of food-associated, noncolonizing fungi, such as sampling from mucosa rather than feces or saliva, may help to accurately assess the truly resident fungal taxa. Additionally, improved scope, accuracy, and curation of fungal sequence databases will be important to increase the number of OTUs that can be mapped to specific taxa, to decrease the number of potentially spurious results matching sequencing artifacts, and to prevent the classification of the same sequences with taxonomic names pertaining to different sexual states. In pursuit of this goal, Tang et al. recently developed the Targeted Host-Associated Fungi (THF) database, which has been curated and optimized for identification of human and murine gastrointestinal fungi using ITS1 sequencing, in an effort to improve future surveys of the mycobiome (227).

Given that fungi colonizing the human body cohabitate with the bacterial members of the microbiota, there is significant interest in how these species cooperate and compete in their shared niches. A significant proportion of existing work examines interactions between various bacterial species and Candida species, which have long been known to colonize the human host at multiple sites and to act as opportunistic pathogens under certain circumstances, making them a clear clinical and research priority. Existing research has revealed a diverse array of interactions, including competition, antagonism, growth enhancement, synergistic virulence, and others (Fig. 2). These interactions are frequently mediated by quorum-sensing molecules from both kingdoms and in some cases are dependent on the context or growth state of the organisms, including in vivo versus in vitro, in suspension versus in biofilm, or yeast cells versus hyphae.

Candida species are found as commensal organisms in the microbial communities of a large proportion of the human population, but their growth is typically limited by competition with other microorganisms and by the host immune system. When this dynamic, a form of colonization resistance, is disrupted, such as by antibiotic administration or immunosuppressive disorders or therapies, Candida species can bloom and cause local infections, including oropharyngeal, vulvovaginal, and cutaneous candidiasis or even the systemic infection candidemia (228). This is commonly characterized by a transition from yeast to hyphal growth, allowing the fungus to become invasive and damaging to tissues (228, 229). The host microbiota plays an important role in keeping this opportunistic pathogen in check, competing with it and inhibiting its expansion in a healthy microbiome. For example, laboratory rodents are not typically colonized by Candida albicans, but they become susceptible if raised under germfree conditions where they have no resident microbial community or if treated with broad-spectrum antibiotics that reduce the density and diversity of their native bacterial flora (203, 230).

While colonization resistance is mediated by a range of species inhabiting the host, bacteria of the genus Lactobacillus are particularly known for their antagonistic relationship with Candida species, especially in the vaginal microbiota, where Lactobacillus species frequently dominate (Fig. 2) (196, 231–233). Early studies showed that Lactobacillus was capable of displacing Candida from stomach epithelia in a murine model (234), and more recent work has shown that certain Lactobacillus species can interfere with Candida adhesion to vaginal epithelial cells via exclusion, competition, and displacement (235, 236). Some of this effect may be mediated by exopolysaccharide produced by some Lactobacillus species, which is structurally similar to that of C. albicans and is itself sufficient to antagonize Candida colonization via exclusion (235).

On a molecular level, there are a number of mechanisms by which lactobacilli can negatively impact the growth of Candida. A range of Lactobacillus species and their supernatants can inhibit growth, hyphal morphogenesis, and biofilm development (237). Lactobacillus rhamnosus can antagonize virulence factor production, biofilm capacity, and antifungal tolerance in a mixed biofilm with C. albicans, while lactobacilli more generally can compete with Candida for nutrients, particularly glucose (238, 239). Some species of Lactobacillus produce molecules toxic to Candida, including bacteriocins, biosurfactants, and hydrogen peroxide, which may help to antagonize Candida growth in shared environments (240–242). Additionally, while their name derives from their production of lactic acid, many lactobacilli also produce SCFA such as butyrate or acetate. This may contribute to the acidification of their environment, dampening the ability of Candida to grow or to undergo transition to the invasive hyphal phenotype (243); in addition, lactic, butyric, and acetic acids have been shown to have direct toxicity against C. albicans (244–246).

Lactobacilli may also interact with the host to improve defenses against Candida overgrowth or inflammatory damage. Lactobacillus species may benefit epithelial barrier integrity, including via impacts on mucus production and modification and on intercellular junctions (179, 247–251). They or the SCFA that they produce may also stimulate the production of antimicrobial peptides lethal to Candida, such as defensins or cathelicidin (235), as well as modulate the expression or activity of mucosal Toll-like receptors (TLRs) (252). Finally, lactobacilli may help to reduce inflammation resulting from Candida activity, as SCFA are known for their anti-inflammatory properties, and some species of Lactobacillus have been shown to promote regulatory T-cell differentiation (253).

It should be noted that many of these mechanisms are shared by other members of the native microbiota, and other bacteria contribute to colonization resistance, especially in the gut, where Lactobacillus makes up a relatively small proportion of the total flora (254). For example, SCFA are produced by other gut microflora, and species such as B. thetaiotaomicron and Blautia producta have been shown to confer colonization resistance to C. albicans in the mouse gut (255, 256). Additionally, the ability to antagonize Candida growth and colonization is likely species or even strain specific; some Lactobacillus species have demonstrated strong anti-Candida activity in a number of studies, while others have shown little or a lesser effect (237, 240, 242). As such, there has been significant research interest in identifying potentially probiotic strains of Lactobacillus, commonly including the species L. plantarum, L. crispatus, and L. rhamnosus, that may be useful in promoting resistance to Candida overgrowth in vulnerable populations (257–262).

However, the interaction between these two groups may not be entirely one-sided, and Candida may also be able to antagonize the growth of Lactobacillus in mixed communities. Most directly, one group found that C. albicans produces a quorum-sensing molecule, farnesol, that may interfere with membrane integrity in L. fermentum and potentially other lactobacilli (263). Another group analyzed the ability of multiple C. albicans secreted aspartate proteases (SAP) to generate antimicrobial peptides from hemoglobin, or hemocidins (264). Many of the tested SAP generated a wide range of these peptides, and these hemocidins demonstrated significant antimicrobial activity against L. acidophilus, a component of the human vaginal microflora (264). Given the presence of hemocidins as a major component of human menstrual discharge, those authors speculate that Candida may be able to regulate the bacterial composition of the microbiota in this niche via the production of these antimicrobial peptides from host hemoglobin (264, 265). In the gut, researchers found that inhibition of dectin-1 signaling, a C-type lectin pathway through which the host recognizes beta-glucans (components of the fungal cell wall), led to an expansion of L. murinus in a mouse model (253). Furthermore, they demonstrated that the lack of dectin-1 signaling led to a significant reduction in intestinal levels of antimicrobial peptides that suppress L. murinus growth. While researchers could not detect viable fungi in the specific-pathogen-free mouse hosts and speculated that dectin-1 was activated by food beta-glucan ligands, this study raises the possibility that gut-resident Candida or other fungi may provoke host responses that antagonize lactobacilli (253).

Researchers have also found that colonization with C. albicans after antibiotic administration altered the recovery of the digestive tract bacterial population in mice, enhancing the recovery of Bacteroidetes and biasing the recovery of lactic acid bacteria toward Enterococcus faecalis instead of Lactobacillus species (266, 267). Importantly, this could occur in the context of nonpathogenic colonization, indicating that even in the absence of overt invasive disease, the presence of Candida may be able to influence the bacterial makeup of the microbiome. Interestingly, E. faecalis has previously been linked with Candida, as they are frequently isolated together from nosocomial infections and E. faecalis may even be enriched in clinical samples containing Candida species (268–271). Additionally, researchers found that in a nematode model of polymicrobial infections, coinfection with C. albicans and E. faecalis led to reduced virulence, increased host survival, and decreased hyphal morphogenesis, suggesting that the two may together promote a commensal rather than an invasive lifestyle (272). The reduction in hypha formation was partially mediated by the E. faecalis quorum-sensing virulence regulator Fsr, one of many examples of quorum-sensing systems involved in mediating fungus-bacterium interactions; the bacteriocin EntV also appears to play an important role, reducing hyphal morphogenesis in both in vitro and in vivo models (272, 273).

While host-associated Candida will most frequently encounter the commensal microflora, as discussed above, it may also interact with bacterial pathogens. Given the potential for coinfections, there has been significant research interest in understanding the relationship between Candida and such bacteria. In many cases, including interactions with Pseudomonas aeruginosa, E. coli, Acinetobacter baumannii, and S. enterica, there is evidence of antagonistic relationships, particularly in a coculture setting (274). Frequently, the antagonism is mutual, with both bacteria and fungi inhibiting the growth or function of the other.

P. aeruginosa, a Gram-negative opportunistic pathogen of significant clinical importance due to high prevalence of antibiotic resistance and frequent implication in nosocomial infections, has multiple mechanisms of antagonizing C. albicans when the two are grown together (Fig. 2). For example, P. aeruginosa inhibits Candida hyphal morphogenesis in vitro via secretion of a homoserine lactone (HSL) quorum-sensing molecule. HSL resembles the Candida quorum-sensing molecule farnesol, which is used by the fungus to limit its own hyphal growth at high cell densities and has been shown to negatively impact membrane integrity in some bacterial species (263, 275, 276). A similar effect has been observed in another Gram-negative pathogen, Burkholderia cenocepacia, which also produces a farnesol-like signaling molecule that inhibits hyphal morphogenesis in Candida (277). Additionally, P. aeruginosa forms biofilms on C. albicans hyphae, which can lead to hyphal death through bacterial production of the virulence factors phospholipase C and phenazines (278, 279). Even at lower, nonlethal concentrations, phenazines can inhibit hyphal morphogenesis and biofilm formation (280). The farnesol produced by C. albicans, in turn, interferes with the production of the Pseudomonas quinolone signal (PQS) that regulates the production of phenazines such as pyocyanin, which may serve to help protect C. albicans from toxic effects of P. aeruginosa (281). Contrarily, the same group showed that farnesol could actually lead to the activation of downstream genes such as phenazines in strains of P. aeruginosa lacking the ability to produce PQS (282). Finally, P. aeruginosa may also be able to inhibit the growth of other fungi, including Aspergillus fumigatus, Scedosporium aurantiacum, and Lomentospora prolificans, although whether this occurs via the same mechanisms as its interactions with Candida is unclear (283, 284).

Similarly, there is evidence that E. coli, which is typically a human commensal but can cause disease with the acquisition of certain virulence factors, can negatively impact the growth of C. albicans (Fig. 2). Early studies indicated that E. coli could kill C. albicans when cocultured, an effect that was not observed when using heat-killed E. coli (285). Similarly, some clinical isolates of E. coli could inhibit the growth of C. albicans both in culture and in a gnotobiotic mouse model, and this effect could be traced to a diffusible factor produced by the inhibitory strains (286). More recently, Cabral et al. found that E. coli could kill C. albicans through a secreted, heat-labile factor dependent on low magnesium levels (287). The impacts of C. albicans on E. coli in coculture experiments are less clear. Some evidence indicates that C. albicans can enhance the growth of a commensal E. coli strain in culture via an iron-dependent interaction or that growth in a mixed biofilm can increase bacterial ofloxacin tolerance through interactions of E. coli with fungal beta-1,3-glucan (288, 289). Alternatively, C. albicans may negatively impact the short-term growth of E. coli in a biofilm, while farnesol produced by the fungus sensitizes E. coli to the antibiotic polymyxin B (263, 290).

The relationship between A. baumannii, a common cause of antibiotic-resistant nosocomial infections, and C. albicans is antagonistic in both in vitro and in vivo models (Fig. 2). A. baumannii has been shown to preferentially associate with the hyphal form of C. albicans and to demonstrate significant killing of hyphae but not yeast cells (291). As a result, A. baumannii was able to decrease the lethality of C. albicans infection in a nematode model, as well as biofilm formation on abiotic surfaces (291). It was later shown that hyphal adherence and killing were mediated by both the bacterial outer membrane protein OmpA and heat-labile secreted proteins (292). However, C. albicans is not defenseless in this interaction, and farnesol produced by C. albicans can inhibit the growth of A. baumannii (291). Farnesol interferes with a number of cellular functions in A. baumannii, including membrane integrity, cell division, biofilm formation, and motility (293). It also induced an upregulation of efflux pumps and OmpA, suggesting possible defense mechanisms of A. baumannii against C. albicans (293). Interestingly, recent work has revealed that A. baumannii uses OmpA to bind to the C. albicans protein Hyr1p and that antibodies raised against Hyr1p recognize several A. baumannii surface antigens and are protective against infection in a mouse model (294).

S. enterica serovar Typhimurium, a pathogen responsible for the eponymous diarrheal disease salmonellosis, also appears to be able to antagonize C. albicans, particularly in the hyphal form (Fig. 2). A nematode model demonstrated that during coinfection along with S. enterica, filamentation of C. albicans was reduced in a dose-dependent fashion (295). A similar effect could be achieved by growing Candida-infected worms in filtered stationary-phase S. enterica supernatants, implicating a secreted and growth-dependent molecule. Additionally, in a coculture model, S. enterica reduced the viability of fungal cells in planktonic culture, particularly at 37°C (human body temperature), and inhibited Candida biofilm formation in a dose-dependent manner (295). The bacteria were able to kill both yeast and hyphal cells, although hyphal killing was more pronounced. Later work implicated the S. enterica type III secretion system (T3SS) proteins SopB (an effector protein) and SipB (a translocase) in the killing of C. albicans, particularly hyphal cells (296). The bacteria responded to the presence of C. albicans by upregulating the production of sopB transcripts, while the inactivation of either gene led to significant reductions in S. enterica-mediated killing (296). It is likely that there are other mechanisms of S. enterica interfering with C. albicans growth, however, given that T3SS killing occurs via direct contact, while the previous study found that the bacterial supernatant could inhibit hyphal growth (295).

In some cases, fungi may be able to play a protective role against bacterial pathogens. For example, recent work demonstrated that the yeast Malassezia globosa secretes an aspartyl protease that cleaves a Staphylococcus aureus virulence factor (protein A) important in biofilm formation and immune evasion, potentially protecting the skin environment from bacterial pathogenesis (297). Additionally, gut fungi may be protective against C. difficile, an opportunistic pathogen that causes severe and debilitating diarrhea and frequently becomes a problem after broad-spectrum antibiotic treatment depletes competing bacterial flora and allows the pathogen to bloom. There is evidence that the yeast Saccharomyces boulardii may be able to both prevent and mitigate this bacterial infection during antibiotic treatment (298–300). Mechanistically, S. boulardii can increase the intestinal production of anti-C. difficile-toxoid IgA, reduce toxin binding by producing a protease that degrades C. difficile toxins, and promote anti-inflammatory pathways in the host (301–304). As such, S. boulardii is of interest as a potential probiotic organism to prevent C. difficile infection in susceptible populations or to treat recurrent, refractory existing infections (299, 305–307).

While there are numerous examples of antagonism between fungi and bacteria in culture models, in vivo work reveals that coinfection can frequently lead to synergistic virulence and worse outcomes for the host, even when it comes to some of those same antagonistic species (274). For example, in contrast to their relationship outside the host, infection models frequently demonstrate enhanced virulence occurring due to coinfection with C. albicans and P. aeruginosa (Fig. 2). Hospitalized patients with Candida colonization of the respiratory tract were more susceptible to pseudomonal ventilator-associated pneumonia, and patients colonized with Candida treated with antifungals had a decreased risk of pseudomonal pneumonia (308, 309). In murine models, rats given P. aeruginosa developed pneumonia only in the presence of C. albicans, while a mouse burn injury model demonstrated that infection with P. aeruginosa preceding infection with C. albicans led to significantly increased mortality compared to infection with either microbe alone (310, 311). Recently, a zebrafish swim bladder infection model was used to demonstrate the synergistic virulence of P. aeruginosa and C. albicans, with significant mortality which could be attributed to enhanced C. albicans epithelial invasion (312). Importantly, the two organisms may still exert antagonistic effects in the in vivo environment, but interactions with the host may be responsible for synergistic impacts of coinfection. Additionally, the context and type of infection may affect the relationship, with a mouse gut coinfection model indicating that C. albicans in fact has a protective effect against P. aeruginosa virulence by suppression of the siderophores pyoverdine and pyochelin (313).

Similarly, despite antagonism in culture, both in vivo and in vitro infection models of C. albicans and E. coli show increased virulence and mortality relative to either infection alone (Fig. 2). Intravenous infection by C. albicans followed by either E. coli or just its LPS led to increased mortality in a mouse model (314). Similarly, a synergistic effect is seen when mortality of intraperitoneal coinfection is compared to mortality of either microbe alone (315). LPS appears to be critical for this effect; multiple studies have indicated that LPS alone is sufficient to induce increased mortality during both intravenous and intraperitoneal C. albicans infections, in addition to showing that administration of anti-LPS antibody reduced mortality, endotoxemia, and bacteremia during coinfections of C. albicans with either E. coli or just its LPS (314, 316). The involvement of LPS suggests a role for the host immune response in this interaction. In vitro, a Caco-2 cell culture model demonstrated that coculture with C. albicans enhanced epithelial translocation of E. coli cells, even in the presence of secretory IgA (317). Similarly, coinfection of Caco-2 cells with C. albicans and enterohemorrhagic E. coli led to faster pathogen invasion, increased C. albicans virulence and hypha-associated gene expression, and greater severity of damage to cells (318).

C. albicans also demonstrates synergistic virulence in mixed infections with Staphylococcus aureus as well as a beneficial relationship in mixed-culture models (Fig. 2) (231). While S. aureus is ineffective at forming biofilms alone, in vitro it can form polymicrobial biofilms with C. albicans, specifically binding with a high affinity to hyphae (319, 320). Furthermore, in such a mixed-species biofilm, S. aureus demonstrates increased antimicrobial resistance, including resistance to vancomycin. Multiple studies have shown that this is at least partially due to physical protection of S. aureus by the extracellular matrix of the biofilm (319, 321); however, recent work also implicates low doses of the Candida quorum-sensing molecule farnesol in the development of antibiotic resistance in S. aureus due to an increase in the expression of efflux pumps (322). Importantly, these effects are context-dependent, as higher doses of farnesol can instead inhibit biofilm growth of S. aureus and sensitize it to antibiotics (320). On the in vivo side, mouse models also demonstrate that C. albicans can enhance the establishment of infection with S. aureus. Oral candidiasis facilitated both oral establishment and systemic dissemination of S. aureus, which could be impeded by antifungal therapy; furthermore, C. albicans enhanced the establishment of infection with S. aureus even when the two microbes were injected at different sites in a mouse model, suggesting the influence of the host immune response (323–325). Further work demonstrates that the two can exhibit synergistic virulence in a peritonitis model and lead to increased mortality compared to single infections (326). Given the frequent coisolation of S. aureus and C. albicans from human infections, ranging from burn wounds to lung infections, candidemia, and infections of implanted medical devices, the interaction between the two species is of significant clinical concern.

The relationships between Candida and Streptococcus species in the oral cavity are of particular clinical and research interest, as both colonize much of the human population, and their interactions may contribute to the pathogenicity of dental plaque biofilms (232, 327). Candida typically colonizes mucosal surfaces in the oral cavity but has been shown to associate with bacterial species, frequently Streptococcus, in oral biofilms on both mucosal and dental surfaces (328–330). Several studies indicate that Candida and Streptococcus species can have synergistic relationships in biofilm formation, although specific interactions may vary between the specific Streptococcus species involved (Fig. 2). On a global level, 16S rRNA and ITS gene sequencing of the oral microbiota showed that adults with higher levels of Candida had increased levels of saccharolytic, acidogenic bacteria, including Streptococcus species (331). In general, Streptococcus species may produce nutrients such as lactate and glucose that can be used by Candida, while Candida can relieve oxygen tension to allow growth of Streptococcus (231, 332).

In one example, mixed biofilms of C. albicans and Streptococcus gordonii on saliva-coated surfaces showed increased biomass and development of fungal hyphae compared to Candida-only biofilms (Fig. 2) (333). Mechanistically, while Candida uses the quorum-sensing molecule farnesol to repress its own hyphal morphogenesis at high concentrations, S. gordonii was able to relieve farnesol-induced hyphal repression, a capability that was dependent on the interspecies quorum-sensing molecule autoinducer 2 (AI-2) (333). Interestingly, the AI-2 molecule produced by another oral bacterial species, Aggregatibacter actinomycetemcomitans, has the opposite effect and inhibits hyphal growth and therefore biofilm formation of C. albicans (334). Additionally, while growth in a mixed biofilm increased total biomass, the quorum-sensing molecule competence-stimulating peptide (CSP) produced by S. gordonii favored the growth of C. albicans in a planktonic state rather than in a biofilm (335); the use of both AI-2 and CSP may allow the bacteria to modulate the amount of fungi in mixed communities by promoting either hyphal growth and planktonic dispersal from the biofilm, respectively.

The oral bacterium Streptococcus mutans is also capable of growing in mixed communities with C. albicans, with beneficial effects on its growth (336). This relationship appears largely dependent upon the presence of sucrose; while S. mutans binds poorly to C. albicans itself, it produces glucosyltransferase enzymes that bind to the surface of the fungi and produce glucans from sucrose. This creates a polysaccharide-rich extracellular matrix to which the bacteria themselves can bind (337, 338). There is evidence that C. albicans itself may promote this interaction; bacterial-fungal conditioned media increased the growth of S. mutans in a biofilm, specifically upregulating glucosyltransferase activity to enhance microcolony formation (336). It was later identified that farnesol achieved this effect at low concentrations, despite inhibiting the growth of S. mutans, and other bacterial species, at high concentrations (336).

However, S. mutans can repress hyphal morphogenesis in the yeast, which has been attributed to multiple mechanisms (Fig. 2). S. mutans CSP repressed hyphal growth, similarly to S. gordonii, while the bacterium also produces a fatty acid signaling molecule, dubbed Streptococcus diffusible signal factor or SDSF, that has a similar effect (339, 340). Interestingly, this effect may be overcome in a more complex community, as C. albicans showed enhanced hyphal formation and tissue invasion when grown with a combination of S. mutans, S. sanguinis, Actinomyces viscosus, and Actinomyces odontolyticus (341). Additionally, it was shown that a mixed biofilm with C. albicans strongly upregulated S. mutans quorum-sensing gene networks, through increased levels of the hexapeptide pheromone comX-inducing peptide (XIP); those authors speculate that proteases produced by C. albicans degrade proteins to produce XIP, thereby triggering the activation of the bacterial quorum-sensing networks (342).

S. mutans has long been known as an etiological agent of early childhood dental caries, and more recent work has found that C. albicans is also frequently found in carious lesions. Therefore, there is interest in the potential cariogenic synergy of these two organisms (343, 344). However, the effects of mixed growth on caries formation remains somewhat unclear. For example, it was shown that mixed-species biofilms have increased lactic acid production in vitro but that C. albicans actively increases the pH of the biofilm, which could theoretically reduce cariogenic potential (345). Contrarily, in an in vivo model, coinfection with S. mutans and C. albicans in a rat model led to significant increases in the viable populations of both organisms in plaque biofilms as well as increased and synergistic severity of dental caries on the smooth sides of the teeth (338). Finally, another study found that while both S. mutans and C. albicans were capable of generating occlusal caries (those at the interface between the upper and lower teeth) in a rat model, there was no enhancement of this effect when both organisms were inoculated together (346).

Streptococcus oralis has been more definitively implicated in pathogenic interactions with C. albicans in the oral cavity (Fig. 2). Both organisms may enhance the growth of the other in mixed biofilms on both abiotic and mucosal surfaces, and dual-species biofilms conferred a greater ability to invade both oral and esophageal mucosal tissue models on C. albicans (347–349). In vivo, the two organisms had a synergistic effect on infection in a mouse model; specifically, C. albicans promoted colonization with S. oralis, while coinfection promoted deep-organ dissemination of C. albicans and increased levels of oral candidiasis (350). A follow-up study examined the potential mechanism behind this effect and showed that the two organisms synergistically stimulate increases in the host protein calpain, which targets and cleaves the epithelial junction protein E-cadherin and allows for increased fungal paracellular invasion (351).

Overall, these observations indicate that several commonly occurring oral Streptococcus species have developed mutualistic relationships with C. albicans, resulting in enhanced biofilm formation, growth, or infective potential for one or both species. In vivo, it is likely that multiple species interact in more-intricate communities, with the production of multiple sets of quorum-sensing molecules, metabolites, and nutrients that may have more-complex interactions and implications for oral health.

Several studies of the human mycobiome have been performed in parallel with analyses of the resident bacteria, providing an opportunity for correlating the presence and abundance of fungal and bacterial taxa. One group was able to identify a number of such correlations in the gut microbiome, including strong negative correlations between Bacteroides and Candida and strong positive correlations between Syntrophococcus and Pichia, Anaerostipes and Fusarium, and Bryantella and Fusarium (204). Additionally, Candida and Saccharomyces were both positively correlated with the archaeon Methanobrevibacter and negatively correlated with the archaeon Nitrososphaera (204). A similar analysis of fungus-bacterium correlations in the gut microbiome found negative correlations between Penicillium and Faecalibacterium and between Cladosporium and Ruminococcaceae and a positive correlation between Botrytis and Rikenellaceae (206). While interesting, the potential implications of these associations have not been identified. Hoffman et al. suggested the existence of syntrophic “guilds” of bacteria, archaea, and fungi, with members cooperatively producing metabolites for others to use, but further work is required to experimentally assess the functional relationships between taxa across kingdoms (204).

Some work has also investigated bacterial-fungal networks in the oral microbiome. Increased levels of Candida have been associated with reduced bacterial diversity, with increased Bacillus and reduced Fusobacteria, Flavobacteria, and Bacteroidia abundances (331). As mentioned above, these microbiomes became dominated by saccharolytic and acidogenic bacteria, including Streptococcus species; this may be linked to the increased cariogenic potential of these communities (331). Additionally, an in vitro model suggested that the presence of Candida may have impacts on the proliferation of different species in an oral microbial community. Specifically, researchers found that inoculating saliva samples with C. albicans allowed the growth of anaerobic bacteria, including Veillonella, Fusobacterium, Prevotella, and Leptotrichia, even under apparently aerobic culture conditions, while samples that were not inoculated with the yeast were dominated by strictly or facultatively aerobic species (352). The authors of that study speculate that the rapid oxygen consumption by C. albicans may create microanaerobic niches for such bacterial genera to grow and that its presence in salivary biofilms may have important implications for microbial composition.

While fungi are the most common unicellular eukaryotes colonizing the body, organisms such as amoebozoans, trypanosomatids, apicomplexans, metamonads, and heterokonts can also infect human hosts (353). While there is comparatively less work exploring this topic, recent research has revealed that some of these organisms can interact with bacterial commensals or pathogens in a variety of ways. Infection with such organisms can itself alter the resident microbiota, the preexisting commensal community can alter the course of infection with these organisms, or, in some cases, both can occur (354). Here, we discuss the interactions of the microbiota at multiple body sites with the heterokont genus Blastocystis; the amoebozoan genus Entamoeba; the metamonad genera Giardia and Trichomonas; the apicomplexan genera Plasmodium, Toxoplasma, and Cryptosporidium; and the trypanosomatid genus Leishmania.

According to culture-independent surveys of the human gut, Blastocystis species are the most common nonfungal eukaryotes, with significant carriage rates even in industrialized nations (353, 355, 356). In a rural Mexican community, carriage of Blastocystis was found to be associated with increased bacterial alpha diversity and significant differences in bacterial community composition, including reductions in the abundance of Prevotella copri and increases in the abundances of several Clostridia lineages, including Ruminococcus bromii. Additionally, colonization was associated with increased levels of several fungal taxa as well as metabolic differences and reduced levels of intestinal inflammation markers (356). Similarly, another group studying Blastocystis in a French cohort found increased alpha diversity and enrichment in Ruminococcaceae associated with Blastocystis colonization. However, they also found that the Prevotellaceae abundance was increased rather than decreased in colonized individuals (357). Supporting this, a study in Denmark found that colonization with Blastocystis was associated with enrichments in Prevotella, alongside reductions in Bacteroides abundances (358), and a metastudy of subjects from multiple continents found that Prevotella copri was associated with Blastocystis colonization and that Bacteroides species and Ruminococcus gnavus were associated with uncolonized subjects (359). Overall, the specific impacts of Blastocystis colonization are somewhat unclear and may depend on the sociogeographic context and particular subtype of Blastocystis used (356).

Entamoeba histolytica causes the potentially lethal illness amoebic dysentery, a significant health issue in some parts of the world, although it has long been recognized that exposures are frequently asymptomatic; on the contrary, the related species Entamoeba dispar is nonpathogenic (360, 361). E. histolytica infects the gut and actually preys on the gut microbiota, although they are not strictly necessary for its growth, and its association with the gut microbiota may enhance its pathogenicity and protect it from oxidative stress (362–364). Both amoeba-induced dysentery and asymptomatic colonization with Entamoeba species have been associated with alterations to the gut microbiota. Quantitative PCR (qPCR) analysis revealed that amoebic dysentery reduced the abundance of numerous taxa while increasing the abundance of Bifidobacterium, while 16S rRNA gene sequencing demonstrated that asymptomatic colonization was associated with increased alpha diversity and increased abundances of several taxa but significant reductions in the abundance of Prevotella species, particularly P. copri (365, 366). Interestingly, a study of Bangladeshi children found that among those found to be colonized with E. histolytica, those that experienced symptomatic disease had higher levels of P. copri detected by qPCR, suggesting that higher levels of this taxon may predispose individuals to developing symptoms of amoebic infection (367).

On the other hand, the same group found that some gut microbes may be protective against E. histolytica in mouse models. First, mice treated with antibiotics to deplete their bacterial microbiota were more susceptible to E. histolytica infection and displayed a deficit in neutrophil recruitment to the gut (368). More specifically, mice colonized with SFB, which closely associate with the gut epithelium and provoke a T_H17 response, were protected from amoebic colonization and had an increased neutrophil presence in the gut. Intriguingly, this protection appears to be at least partially due to changes induced in bone marrow dendritic cells, as adoptive transfers of these cells from SFB-positive (SFB⁺) to SFB-negative (SFB⁻) mice was sufficient to confer protection, and gut-marrow signaling may be based on SFB-induced increases in levels of serum amyloid A (369).

Gut bacteria have also been studied in the context of infection with Giardia, a genus of metamonads that primarily colonizes the small intestine. While asymptomatic in many, it can cause acute diarrheal illness and recently has also been associated with the development of postdiarrheal irritable bowel syndrome (IBS). An accumulating body of work demonstrates bidirectional impacts between Giardia infection and the composition of the gut microbiota (370). For example, it was recognized that susceptibility to Giardia lamblia (also known as G. duodenalis or G. intestinalis) infection in mouse models was dependent on the source of the mouse, with Taconic Farms mice being resistant but Jackson Laboratories mice being susceptible. Antibiotic treatment eliminated resistance, but cohousing of the two groups conferred resistance to all mice, implicating the composition of the gut microbiota in this phenotype (371).

However, gut bacteria may contribute to symptom development, as germfree mice can be colonized by Giardia but suffer less intestinal pathology than conventional mice (372, 373). This may be due to a dampened host response, as germfree animals produce fewer Giardia-specific antibodies and have less immune cell accumulation in the lamina propria than their conventional counterparts (372). Similarly, antibiotic-treated animals have fewer activated CD8⁺ T cells in the lamina propria during infection (374). Furthermore, G. lamblia has been associated with small intestinal bacterial overgrowth during infection as well as decreased epithelial tight junction integrity, increased permeability and bacterial adhesion, and mucosal inflammation even after parasite clearance; this suggests a long-term disturbance of intestinal homeostasis due to infection, which could potentially contribute to the development of postdiarrheal IBS (375–377). Work in a Caenorhabditis elegans model has also suggested that exposure to G. lamblia may induce functional changes in commensal microbes, promoting intestinal colonization by bacteria and leading to lethality not observed during individual inoculations of either bacteria or parasites (378).

To more specifically profile the changes observed in the bacterial community, Barash et al. studied the effects of G. lamblia infection on the bacterial composition along the gastrointestinal tract in antibiotic-treated and naive mice (379). They found that antibiotic-treated mice were significantly more susceptible to Giardia colonization, displaying increased parasite burdens even at 2 weeks postinfection, and antibiotic-treated mice had more-widespread and longer-lasting community disruptions associated with Giardia infection; these included depletions of Clostridiaceae (Firmicutes) and enrichments of Moraxellaceae and Rhodocyclaceae (both Proteobacteria), particularly in the proximal small intestine. However, even untreated mice suffered disruptions to the composition of their gut bacteria (379).

However, there is some research suggesting that Giardia infection may actually reduce the incidence of diarrheal disease in areas with frequent gastrointestinal pathogen exposure, although the evidence is mixed and may be dependent on the specific coinfection (380–384). In support of such an effect for bacterial pathogens, in a mouse model, Giardia muris coinfection significantly attenuated pathology from the bacterium Citrobacter rodentium, reducing weight loss, colitis, pathogen load, and bacterial attachment while increasing host antimicrobial peptide production. Furthermore, coinfection of human intestinal epithelial cell monolayers with G. lamblia and enteropathogenic E. coli led to enhanced antimicrobial peptide production, and G. muris reduced the survival of both C. rodentium and E. coli in a coculture model (385). G. lamblia also appears to be able to antagonize host inflammatory responses, possibly by different mechanisms based on the genetic assemblage of the parasite. One study demonstrated that G. lamblia assemblage A secretes a cysteine protease that degrades the neutrophil chemoattractant IL-8 (CXCL8), even in the presence of a direct, inflammatory bacterial insult in the form of S. enterica serovar Typhimurium (386, 387). On the other hand, a study reported that G. lamblia assemblage B instead appeared to dampen inflammatory responses to enteroaggregative E. coli during protein malnutrition, reducing myeloid cell activation despite an increased number of these cells in the ileum (388).

Not all metamonads interact with bacteria in the gut, however, and there is evidence that the human genital tract pathogen Trichomonas vaginalis may have a mutually antagonistic interaction with Lactobacillus species in the vaginal microbiota. Epidemiologically, some recent studies have demonstrated a relationship between trichomoniasis and non-Lactobacillus-dominated vaginal microbiota, which can be associated with the disease bacterial vaginosis but also occurs asymptomatically in a subset of women (75, 389–393). Experimentally, early work recognized that T. vaginalis could have a negative impact on the growth of Lactobacillus acidophilus in coculture experiments (394). More recently, others have found that T. vaginalis can reduce the numbers of lactobacilli but not other vaginal bacteria associated with epithelial cells (395), while a number of Lactobacillus species were seen to inhibit adhesion of T. vaginalis to epithelial cells and even promote displacement of the parasite in a contact-dependent manner (396, 397). Additionally, T. vaginalis appears to have an association with the vaginosis-associated bacterium Mycoplasma hominis, which can be taken up and survive within cytoplasmic vacuoles. Some T. vaginalis isolates are stably associated with the bacterium, and such isolates may increase the local inflammatory response, potentially increasing disease severity (398–402).

There is also evidence of interactions between apicomplexans and the microbiota. For example, Cryptosporidium parvum, an intestinal parasite, has a differential infective capacity in mice based on the presence of the microbiome; conventional mice are resistant to C. parvum colonization for several weeks, while germfree mice become heavily infected much more quickly (403, 404). However, mice treated with antibiotics to deplete their gut bacteria remained resistant to infection, suggesting a more complex mechanism than colonization resistance by competition (404). As the increased susceptibility in germfree mice occurs in both immunocompetent and severe combined immunodeficiency models, which lack an adaptive immune system, the gut microbiota may confer resistance through nonspecific immune mechanisms (403).

The gut microflora also appears to have a role in mediating the host response to infection with another apicomplexan parasite, Toxoplasma gondii. This organism initially infects the small intestine, invading the epithelium to undergo differentiation from infective sporozoites to mobile tachyzoites. This epithelial damage allows increased bacterial penetration into the epithelium and interaction with the immune system, which can have both positive and negative consequences (354, 405). On the one hand, gut bacteria may play a supportive role in the mucosal immune response to T. gondii. In mice, the recognition of and immune response to T. gondii canonically occur through the action of dendritic cell Toll-like receptor 11 (TLR-11) and subsequent production of IL-12 (406). However, stimulation of other TLRs by the gut microbiota can also lead to IL-12 production, even compensating for the lack of TLR-11 in knockout mice, during oral T. gondii infection. This does not occur during systemic infection, in which the immune response is entirely TLR-11-dependent, suggesting that local interactions such as gut epithelium damage by the parasite are necessary for microbial IL-12 stimulation to occur. Gut microbes may thus serve as a molecular adjuvant of the mucosal immune response and may perhaps be even more important during human infections, given the lack of a functional TLR-11 homologue (406).

On the other hand, the gut flora can exacerbate intestinal inflammation and lead to significant ileitis and even death in mouse models (407, 408). Multiple groups have found that T. gondii infection results in inflammation only in the presence of the gut microbiota, as germfree or antibiotic-treated mice do not suffer these effects (407, 408). In particular, it appears that infection with T. gondii selectively expands Proteobacteria, particularly Enterobacteriaceae, which leads to the development of intestinal inflammation and pathology in this context (407, 408). There may be a positive-feedback loop exacerbating both inflammation and Enterobacteriaceae overgrowth; T. gondii infection in the presence of Enterobacteriaceae leads to intestinal damage and Paneth cell death, while the loss of the antimicrobial peptides produced by these cells may contribute to Enterobacteriaceae overgrowth (407). Additionally, some work suggests that T. gondii infection may lead to a loss of immunological tolerance to commensals, possibly due to increased mucosal exposure to gut microbes (409). Even after infection clearance and epithelial healing, mice had increased populations of apparently commensal-responsive CD4⁺ T cells. Furthermore, transferring bacterium-specific splenic T cells to a T. gondii-infected mouse results in significant proliferation and differentiation into effector cells, while this does not occur in uninfected hosts, demonstrating an increase in the frequency of immune cells responsive to the commensal microbiota during infection (409).

The host gut microbiota may also impact the severity and transmission of another apicomplexan-related disease, malaria, caused by parasites of the genus Plasmodium (410). Studies in mice have demonstrated that animals from different vendors, and therefore with different gut microbial compositions, were differentially susceptible to infection, with varied parasite burdens and mortality outcomes. Furthermore, this resistance or susceptibility could be transferred to germfree mice via cecal transplants (411). This resistance appeared to be related to an elevated humoral immune response, and resistance was associated with increased abundances of Lactobacillaceae, Bifidobacteriaceae, and Clostridiaceae while susceptibility was associated with higher levels of Bacteroidaceae, Prevotellaceae, and Sutterellaceae (411). Similarly, a study of the fecal bacterial microbiota of a Malian cohort found that a particular microbial assemblage, which included higher levels of Bifidobacterium, Lactobacillus, and Streptococcus and lower levels of Prevotellaceae and various Clostridia taxa, was associated with a reduced incidence of malaria (412). Furthermore, Plasmodium sporozoites possess certain surface glycans shared by some species of Enterobacteriaceae, and studies in both humans and mice suggest that antibodies against these glycans may be cross-protective against infection with Plasmodium (413).

Plasmodium infection may also cause alterations to the gut microbiota, and humans suffering from malaria may experience gastrointestinal symptoms (414–416). Work in a mouse model found that infection with Plasmodium berghei led to intestinal pathology in addition to dysbiosis, characterized by reductions in Firmicutes and increases in Proteobacteria and Verrucomicrobia abundances. In particular, Lactobacillaceae abundances were reduced during infection, while Verrucomicrobiaceae and Enterobacteriaceae abundances were increased; interestingly, alterations to the gut community could be observed before the onset of gastrointestinal pathologies (416). Additionally, infection of mice with Plasmodium yoelii leads to a reduction of colonization resistance against S. enterica (417). The microbiota may also play a role in the susceptibility of the mosquito host to carriage of the parasite. For example, the ability of Plasmodium to colonize and mature within its Anopheles mosquito vector is at least partially dependent on the midgut bacteria, via mechanisms including direct antiparasite antagonism and the ability to influence the insect’s immune system. In particular, the bacterium Serratia marcescens and some species of Enterobacter have been implicated in reducing parasite colonization (418–422).

Interactions between bacterial microbiota and unicellular eukaryotes may also take place in skin communities; several studies have suggested that cutaneous leishmaniasis, caused by trypanosomatids of the genus Leishmania, may interact with skin bacteria (422). Infection with Leishmania braziliensis in humans leads to dysbiosis of the skin bacterial microbiota, with reductions in the diversity and overgrowth of the genera Staphylococcus and/or Streptococcus; similar changes can be observed in a mouse model using Leishmania major ear infection (423). Interestingly, the specific changes observed may be dependent on symptom severity, with resolvable infections leading to Staphylococcus overgrowth and nonhealing lesions leading first to Staphylococcus but ultimately to Streptococcus dominance. Furthermore, parasite-associated dysbiosis could be transferred to naive mice by cohousing, and if these dysbiotic mice were then exposed to L. major, they displayed greater lesion severity and higher levels of inflammatory markers than mice with normal skin microbiota (423). The gut bacterial microbiota may also be affected during Leishmania infection, with a study demonstrating decreases in Gammaproteobacteria abundances in both healing and nonhealing mouse models, although such an effect requires more research (424).

The outcome of Leishmania infection in germfree animals is unclear, potentially dependent on the route of infection, parasite inoculation dose, or mouse strain. In one case using subcutaneous footpad infection, it was shown that germfree mice were able to mount an antiparasite response similar to that of conventional animals but that they failed to heal lesions and were more densely colonized than conventional mice or mice conventionalized postinfection (425, 426). However, in a study using an intradermal ear infection model, germfree mice were more permissive to growth of L. major but had lesions that were smaller and less necrotic (427). Similarly, subcutaneous tail and ear infection models with Leishmania amazonensis demonstrated that germfree mice had smaller lesions than those of conventional mice (422, 428). Additionally, the microbiota of the Leishmania infantum vector Lutzomyia longipalpis (sand fly) appears to be critical for the development of the parasite in the insect midgut; infection of L. longipalpis results in disturbance and loss of diversity in the bacterial microbiota, while depletion of the gut microbiota with antibiotics impeded the parasite’s ability to replicate and develop in the fly midgut (429).

Fungi may make up a relatively small proportion of the human microbiota, but their relationships with their bacterial neighbors and the host are not insignificant. Advances in sequencing technology have allowed a much greater understanding of the diversity of human-associated fungi than was possible with culture-dependent methods, and while fungal marker gene databases lag behind those for bacteria, progress is being made. It is becoming increasingly clear that fungi persistently colonize a range of body sites, form a variety of antagonistic or cooperative relationships with bacterial species, and can affect the course of disease during coinfections.

In particular, the ubiquitous human-associated fungus C. albicans demonstrates a range of interactions with bacteria. There are a number of examples of cooperative interactions between C. albicans and bacterial species, particularly in the oral environment. However, there are just as many examples of antagonism, including competition in shared niches and direct mechanisms of killing. Importantly, there are multiple examples in which bacterial species have an antagonistic relationship with C. albicans in an in vitro setting but demonstrate synergistic virulence in in vivo models, highlighting the important role that the host and possibly other species of bacteria and fungi may play in modulating some of these interactions.

Much of the interaction between bacteria and fungi is mediated by quorum-sensing molecules. Farnesol, a C. albicans quorum sensor that inhibits hyphal morphogenesis, is particularly important and has various effects depending on the concentration and context; it has fairly strong antimicrobial effects on multiple bacterial species at high doses but can have positive effects on bacterial fitness at low doses, including enhancing biofilm growth and increasing antibiotic resistance. Bacterial species themselves use a host of quorum sensors, including AI-2, CSP, Fsr, and farnesol mimics such as HSL, to regulate hyphal growth of C. albicans in mixed communities; in at least one case, C. albicans may actually induce a quorum-sensing system, specifically the XIP system in S. mutans. Additionally, a number of bacterial species have developed strategies to kill or suppress C. albicans, particularly in the hyphal form.

Moving forward, there are several directions to pursue. On a technical level, expansion and improved curation of fungal databases will be required to continue making progress, particularly in the arena of metagenomics and metatranscriptomics. This is particularly important to help overcome some of the limits of marker gene sequencing, which lacks detail and the ability to identify strain-specific differences but has been used in most studies to date. As has been found in bacteriome work, while descriptions of resident taxa are informative, this approach cannot accurately account for the contributions of specific genes, regulatory pathways, and/or metabolites to the gut community. Additionally, when studying relationships between fungi and bacterial pathogens, work comparing in vitro and in vivo models will be important to account for the impacts of the host and other microbiota, which may have a significant influence on the interaction. Finally, while next-generation sequencing studies have revealed a number of potential relationships between fungal and bacterial taxa in the gut microbiome, the biological implications of such relationships generally have yet to be revealed. Performing interventional studies to identify the impacts of diet or other factors, in parallel with in-depth metagenomics, metatranscriptomics, and/or metabolomics, may help to reveal the potential relevance of such observations and uncover functional networks of both fungal and bacterial species in the microbiome.

In addition to fungal colonization, human can be infected with a diverse array of other unicellular eukaryotes, which frequently demonstrate a parasitic relationship with their hosts. While the exact associations vary by organism, there is evidence for interactions between unicellular eukaryotes and the microbiota of the skin, gut, and genital tract. Infection with these organisms can alter the composition or function of the resident microbiota, commensal microbes may provide protection against infection, or resident microbes can contribute to the pathology of infection; in many cases, these interactions involve modulations of host immunity to exacerbate or protect against pathological responses. Furthermore, the microbiota of insect vectors can also interact with the parasites that they carry, either promoting or inhibiting development, which can have important implications for the spread and transmission of these diseases. Continuing work in this arena will benefit from the identification of specific microbes that confer protective effects, comparisons of results from mouse models with data from human subjects, and further elucidation of mechanisms via which unicellular eukaryotes, the microbiota, and the host interact.

There are three major groups of helminths that infect humans (Fig. 1); two of these groups fall into the phylum Platyhelminthes, commonly known as flatworms for their appearance (431, 433, 441, 442). First, there are trematodes, members of the class Trematoda, which are parasitic flatworms referred to as “flukes.” They can be broken up into two practical groups: Schistosoma species are blood flukes that primarily reside in the vasculature around the gut (S. mansoni and S. japonicum) or bladder (S. haematobium) in the human host, while tissue flukes, including Fasciola hepatica, Paragonimus westermani, and Echinostoma species, infect the liver, lungs, or intestines (443–445). Their life cycle includes a primary vertebrate host, where sexual reproduction occurs, and one or more intermediate mollusk hosts, where they reproduce asexually (431, 446). The second group of helminths implicated in human disease are the cestodes (Taenia species, Hymenolepis species, and Diphyllobothrium species), which are parasitic flatworms of the class Cestoda in the phylum Platyhelminthes. Cestodes are more commonly known as tapeworms and generally infect the gastrointestinal tract of vertebrate species (447). They frequently have a two-phase life cycle: an intermediate host consumes eggs passed from the primary host, hatched larvae penetrate into the muscle, the primary host eats such contaminated meat, and the larvae reach maturity in the gut and begin producing eggs (431, 433, 441, 442).

The largest group of clinically relevant helminths are the nematodes, or roundworms. Collectively, they comprise the phylum Nematoda and include a wide and diverse range of species with various habitats, reproductive strategies, and life cycles. Many of these organisms parasitize vertebrates, and several dozen infect humans (431, 433, 441, 442). Filarial worms, which include the causative agents of elephantiasis (Wucheria bancrofti, Brugia malayi, and Brugia timori) and river blindness (Onchocerca volvulus), use arthropods as intermediate hosts and are transmitted via insect bites (448). As such, they do not spend time in the gastrointestinal tract as a normal part of their life cycle, instead living in the lymphatic system and subcutaneous tissues (449). Most other pathogenic nematodes are categorized as intestinal helminths, as they are transmitted via ingestion of contaminated food or water and spend at least part of their life in the human host in the gut. Examples include pinworms (Enterobius vermicularis), whipworms (Trichuris trichiura), and Guinea worms (Dracunculus medinensis) as well as Ascaris and Trichinella species (433, 441, 442, 450). Hookworms (Necator americanus and Ancyclostoma duodenale) and threadworms (Strongyloides stercoralis) are also partially gut-resident, although these are transmitted through skin contact with contaminated soil (433, 441, 442).

Importantly, as many helminths spend a significant portion of their life cycles in the gastrointestinal tract of the human host, there is interest in their potential interactions with the native gut microbiota (Fig. 3). For gut-resident helminths, there is significant evidence that the presence of the intestinal microflora is important for successful infection, and some work has even implicated specific members of the microbiome. Conversely, there is also a body of work investigating the impacts that helminth infections have on the gut microflora of human and animal hosts; local interactions or systemic immunological responses to worm infection have the potential to influence the diversity, makeup, and function of the native gut community.

As mentioned above, helminth infection is thought to be an important part of the old friends hypothesis (102, 103, 105, 107, 435–437). There is evidence that over the thousands of years of coevolution with their human hosts, many helminths have adapted to modulate the immune system, thereby allowing them to persist in chronic infections. At the same time, the human immune system evolved in the context of these chronic infections and may have adapted to use helminthic immunomodulatory cues to regulate itself and prevent inappropriate activation and damage (103, 104). As modern living conditions have become more hygienic, with cleaner food and water supplies, exposure to helminths has become rare in developed nations. Given the negative impacts of many helminth infections, this has been generally beneficial; however, it is theorized that the lack of helminth infections may be associated with deficiencies in immune system regulation, particularly in the arena of immunological tolerance (451). Accordingly, individuals living in developed nations become more likely to develop disorders involving inappropriate immunological reactions to harmless or self antigens, including allergies, asthma, IBD, type 1 diabetes, and multiple sclerosis (Fig. 3) (437, 452, 453). On a more specific level, observational human studies have demonstrated an inverse relationship between helminth infection and markers of both atopy and autoimmune activity (454–457). For example, helminth infection was associated with reduced dust mite skin test reactivity in Gabonese schoolchildren, while a randomized controlled trial demonstrated that antihelminthic treatment reversed this trend, and the intensity of schistosome infection was negatively associated with antinuclear autoantibodies in a Zimbabwean population (456, 457).

It is believed that such effects occur through helminth-induced modulation of the host immune system. A striking feature of helminth infection in general is the activation of type 2 immunity, the host response against extracellular pathogens and allergens, which is characterized by the differentiation of T_H2 CD4⁺ T cells and the production of the interleukins IL-4, IL-5, and IL-13 (433, 458–460). Practically, this response results in enhanced epithelial mucus production, smooth muscle contraction, and fibrosis of granulomas, which can help to kill, expel, or seal off the parasite or its eggs and promote rapid healing of parasite-induced damage (Fig. 3) (458, 460). Additionally, activation of the T_H2 response downregulates the proinflammatory T_H1 and T_H17 pathways, which are commonly associated with viral, bacterial, and autoimmune diseases but can also contribute to pathological inflammation and tissue damage during some helminth infections (458, 460). However, a range of helminth species have developed mechanisms to dampen immune responses against them and thereby promote their own persistence, including activation of T_reg cells, regulatory B cells, tolerogenic dendritic cells, and alternatively activated macrophages, as well as inhibiting host responsiveness to the parasite’s own antigens; in particular, production of the regulatory cytokines IL-10 and transforming growth factor β (TGF-β) is implicated in the control of both proinflammatory T_H1/T_H17 pathologies and allergic T_H2-driven responses (Fig. 3) (430, 458–471). These changes allow some helminths to persist for years in the human host and may be protective against allergic and autoimmune disorders. Additionally, the immune system itself may have evolved to tolerate such immunomodulation rather than risking serious organ damage through a sustained immune response (104, 430, 458); for example, several studies of chronic filariasis and schistosomiasis indicate that those with greater disease pathology tend to have stronger immune responses, while those that maintain chronic, low-level infection demonstrate immune hyporesponsiveness (472–477).

Due to these broad immunological changes, deliberate helminth infection is being investigated as a treatment for several autoimmune diseases. Helminths, given their association with T_H2 polarization, promotion of an immunoregulatory state, and immune hyporesponsiveness, may be able to downregulate the self-inflammatory responses involved in autoimmune pathology. Furthermore, promising early observations indicating disease remission in naturally infected multiple sclerosis patients, which was reversed after antihelminthic treatment, have encouraged research interest in this area (478). There are several helminth species being investigated for this purpose, including Trichuris suis (pig whipworm) and N. americanus (hookworm) (479). A number of studies in animal models of autoimmune disorders have lent support to this approach (470, 480–483); human trials for several autoimmune disorders are ongoing and thus far have had mixed results (438, 484–487). However, helminth-associated immunomodulation can also negatively impact the immune response to secondary microbial infections. Studies have demonstrated potential helminth-related defects in response to an array of vaccines and infections, including tetanus, toxoplasmosis, tuberculosis, cholera, malaria, and infections by several viruses (488–500). In one example, researchers demonstrated that mice infected with the gastrointestinal nematode Heligmosomoides polygyrus or eggs of the trematode Schistosoma mansoni experience increased viral replication and even reactivation of latent herpesvirus via T_H2 cytokine responses, particularly IL-4 (501).

Importantly, the immune regulation mediated by helminths may impact or be impacted by the gut microflora, which occupy the same physical niche as many worms. Humans, gut commensals, and helminths have coevolved for millennia, likely leading to the development of bi- and tripartite interactions between host, bacteria, and worms. Here, we discuss evidence that host-associated microbes may impact helminth infection success, that helminth colonization may impact the diversity and composition of the gut microbiome, and that both can have impacts on immune function.

Perhaps unsurprisingly given the range of immunological impacts that helminths can have on their hosts, there is significant evidence that helminth infection is capable of altering the composition and function of the gut microbiota. While specific mechanisms remain, in many cases, unelucidated, it is likely that direct effects exerted by the helminths themselves, local immune responses to worms, general immunological shifts induced by infection, and other mechanisms may coincide to alter gastrointestinal conditions and allow for the proliferation of alternate gut microflora. However, there is little consensus on specific effects such as taxonomic shifts, given that studies have been conducted on a wide array of host and helminth species, and therefore, there is significant breadth but little depth to the field at the moment (502–504). Despite this, it seems clear that infections with a variety of helminths have the capacity to alter the composition of the host microbiome.

In an example of a potentially common impact of different helminth species, infections of rodents with the nematodes H. polygyrus (formerly Nematospiroides dubius), Trichuris muris, Nippostrongylus brasiliensis, and Trichinella spiralis may all promote enrichment in the Lactobacillaceae lineage of gut bacteria, particularly in the ileum, where the parasites reside. In one case, researchers analyzed the microbiota of healthy mice during H. polygyrus infection, finding an enrichment in ileal Lactobacillaceae across multiple experiments as well as an increase in the number of bacteria overall (505). Interestingly, this group also demonstrated that while the larvae of this helminth have a distinct microbial community, adult nematodes themselves come to harbor microbiota closely resembling that of their host ileal environment (505). In another study of H. polygyrus infection, it was shown that the abundance of Lactobacillus species in the mouse small intestine correlated positively with infection (506). Specifically, researchers found that the abundance of Lactobacillus increased during infection of the H. polygyrus-susceptible C57BL/6 strain of mice but not in the resistant BALB/c strain, where the helminth cannot take hold (506).

Similarly, multiple studies of mice infected with T. muris showed an association between infection and levels of this microbial lineage. Holm et al. found a significant enrichment in Lactobacillaceae, while a contemporaneous study by Houlden et al. found increased levels of Lactobacillales in infected animals; both studies also found reductions in diversity resulting from chronic T. muris infection (507, 508). Increases in levels of ileal Lactobacillaceae have also been observed in mice infected with N. brasiliensis, alongside increases in Coriobacteriaceae and decreases in levels of Peptostreptococcaceae, Clostridiaceae, Turicibacteraceae, and IL-17-inducing SFB (509). Similarly, mice infected with Trichinella spiralis displayed increased Lactobacillaceae and decreased Turicibacteraceae and Clostridiaceae abundances in the ileum as well as increased proportions of Ruminococcaceae and Lachnospiraceae in the colon (492). However, not all models show this trend in response to nematode infection; Trichostrongylus retortaeformis infection of rabbits did not lead to changes in Lactobacillaceae but resulted in decreased diversity, increased abundances of Leptonema and Desulfocella, and decreased abundances of Ruminococcus and Bacteroides (510). Beyond altering the intestinal bacterial populations, murine infection with gastrointestinal nematodes may have impacts on the virulence of bacterial pathogens. While they did not directly assess changes in the resident bacterial microbiota, Reynolds et al. demonstrated that H. polygyrus infection enhances colonization, pathogenicity, and invasion by S. enterica by altering the metabolic profile of the small intestine (511).

In addition to rodents, several studies have examined the impacts of infection with various nematodes on the microbiota of the abomasum (fourth stomach) of ruminant species. Infection with several related helminths has been shown to increase the abomasal pH due to reduced gastric acid production, and this pH shift was associated with an enrichment in anaerobic bacteria in the abomasa of sheep infected with Ostertagia circumcincta (512). Later, the same group investigated the impacts of infection with another related nematode, Haemonchus contortus, on the microbiome of the goat abomasum, finding significant differences in the beta diversities of the two groups and that 19% of OTUs were different between infected and uninfected goats (513). Specifically, infection was associated with increases in the abundances of several Veillonellaceae OTUs, decreases in the abundances of several Lachnospiraceae OTUs, and a mix of increases and decreases in the abundances of Bacteroidales OTUs. Additionally, functional potential analysis, using the 16S rRNA gene composition to predict the metagenomes present, revealed differential abundances of six gene families and 9 KEGG pathways between the two groups, indicating possible modifications of microbiome function (513).

The same group also demonstrated changes in the porcine microbiome after infection with the pig whipworm, T. suis, with these alterations persisting even after parasite clearance (514). Pigs were infected with worms for 53 days and then separated into worm-free pigs that had cleared infection and pigs still carrying significant worm burdens. The abundances of 48 out of 372 genera identified in the proximal colon microbiota of the pigs were significantly altered by infection, whether infection persisted or was cleared, suggesting long-lasting impacts from initial infection (514). Interestingly, the abundance of Campylobacter varied by worm status: compared with parasite-naive pigs, worm-heavy pigs harbored more Campylobacter bacteria, while pigs which had cleared infection harbored less (514). In another study from the same laboratory, researchers showed that a shorter infection with T. suis for 21 days was also sufficient to induce significant changes in the gut microflora, impacting approximately 10% of genera present in control pigs (515). Specifically, researchers observed significant increases in the abundances of Mucispirillum, Paraprevotella, and Desulfovibrio and decreases in the abundances of several Clostridiales, including Ruminococcus, Blautia, and Dorea (515). Additionally, metagenomics indicated that infected pigs showed an reduction in the abundance of KEGG pathways involved in carbohydrate metabolism and amino acid metabolism (515). Finally, a study of ponies infected by strongyles (nematodes of the families Strongylinae and Cyathostominae) also found reductions in the abundances of Clostridiales associated with infection; susceptible ponies that developed more-intense infections demonstrated reductions in the abundances of the commensal Clostridiales Ruminococcus, Lachnospiraceae, and Clostridium XIVa and increases in abundances of Pseudomonas, Campylobacter, and Bacillus relative to resistant ponies that developed less-intense infections (516).

Several studies have used animal models of IBD to investigate the impacts of helminth infection in the context of inflammation. For example, one study investigated the impact of helminth therapy in the nonhuman primate Macaca mulatta (rhesus macaque) with chronic idiopathic diarrhea (CID), a condition similar to human ulcerative colitis (517). Researchers treated five juvenile macaques with CID with the whipworm T. trichiura and found improved fecal consistency and weight gain in four subjects, despite a lack of establishment of patent infection. Analysis of the bacteria attached to the intestinal mucosa revealed a decrease in the absolute amount of bacterial attachment, an increase in Shannon diversity, and significantly different measures of beta diversity in posttreatment macaques; specifically, those researchers observed a significant increase in the abundance of the phylum Tenericutes in response to helminth treatment. Generally, they speculate that helminth treatment increased T_H2 responses, including mucus production and epithelial cell turnover, decreasing the attachment of immunostimulatory bacteria to the mucosa and improving symptoms (517). Similarly, researchers examined the impact of infecting mice lacking the IBD susceptibility gene nod2 with either H. polygyrus or T. muris (518). nod2^−/− mice typically display small intestinal abnormalities, including goblet cell defects and colonization by the bacterium Bacteroides vulgatus; after colonization with helminths, however, the goblet cell defect was corrected, and levels of B. vulgatus fell dramatically. The inhibition of B. vulgatus was associated with an increase in the abundance of the family Lachnospiraceae in the class Clostridiales and could be transferred to helminth-free nod2^−/− mice by cohousing. Those authors suggest that the increased mucus production occurring during helminth infection promotes the growth of Clostridiales over B. vulgatus (518).

While the majority of studies in animal models have been conducted using nematode parasites, there are some studies using other types of helminths. For example, while colonization of rats with the tapeworm Hymenolepis diminuta did not alter alpha or beta diversity, there were distinct differences in the cecal community composition affecting approximately 20% of the microbiome (519). Specifically, there were significant decreases in Turicibacter and significant increases in Peptostreptococcaceae abundances. They also exposed half of the animals to LPS challenge several days before analysis to assess the impacts of inflammatory challenge in the context of helminth infection. Interestingly, beta diversity was altered between LPS groups in uninfected animals, but there was no impact of LPS in helminth-infected rats (519). Further analysis showed that most of the changes observed resulted from shifts internal to the major phylum Firmicutes, with increases in the abundances of the class Clostridia and decreases in the abundances of the class Bacilli. Importantly, while the worm resides in the small intestine, these effects were observed in the cecum, indicating the potential for more than simply local effects of infection (519). Another group analyzed the effects of infection with the trematode liver fluke Opisthorchis viverrini on the microbiome of Syrian hamsters, a model for the bile duct cancer that this helminth can induce (520). That group found that hamsters infected with O. viverrini had an increased diversity of the gut microbiome, with increases in the abundances of Ruminococcaceae, Lachnospiraceae, and Lactobacillus and decreases in the abundances of Porphyromonadaceae, Erysipelotrichaceae, and Eubacteriaceae (520). Finally, a recent study using a mouse model of S. mansoni infection found significant reductions in alpha diversity associated with infection; additionally, researchers observed a significant expansion of the Verrucomicrobia member A. muciniphila, in addition to increases in the abundances of members of the Bacteroidales, Coriobacteriales, and certain Clostridiales and decreases in the abundances of Erysipelotrichia and other Clostridiales (521). Importantly, the microbiota was assessed both before and after the beginning of egg production (28 and 50 days postinfection, respectively), which demonstrated progressive differences in microbiome composition that specifically accompanied the onset of egg production and associated intestinal damage and inflammation (521).

There have also been some studies in humans, although these have generally been observational. One study used 16S rRNA gene sequencing to examine the gut microbiomes of rural Malaysians infected with one or more gastrointestinal helminth species (Ascaris species, Trichuris species, and hookworm) and compared them to those of uninfected controls, finding that helminth infection was associated with greater diversity of the fecal microbiome (522). Analysis of changes in the community’s predicted functional potential indicated an enrichment in the abundance of a number of metabolic pathways, including those involved in nucleotides and amino acids. Further analysis revealed that Trichuris infection specifically was strongly associated with enrichment in Paraprevotellaceae, and functional potential analysis indicated an enrichment in pathways associated with nucleotide metabolism, cell growth, and cell death (522). A later study by the same group studied a similar population before and after antihelminthic treatment and found that reductions in parasite burden were associated with decreases in Clostridiales and increases in Bacteroidales abundances, which supported their findings of helminth infection in nod2^−/− mice discussed above (518).

In a similar study, another group analyzed the impacts of infection with T. trichiura alone or in combination with Ascaris lumbricoides on the gut microbiome of Ecuadorean children. For children infected with T. trichiura alone and a subset of uninfected children, a follow-up was performed after curative antihelminthic treatment (523). Those researchers found no significant differences in microbial composition in children infected with T. trichiura alone and no differences after antihelminthic treatment. However, they observed a decreased proportion of Clostridia and reduced microbial diversity in children with mixed infections, suggesting a specific effect of A. lumbricoides infection on the microbiome (523). However, in the course of a larger study, Rosa et al. reanalyzed this data set using their own pipeline; they found associations between helminth infection and differences in various taxa, including increases in abundances of Eubacteria, Streptococcus, and the order Lactobacillales (524). In addition, they performed a study investigating the impacts of infection with A. lumbricoides, N. americanus, and T. trichiura in populations in both Liberia and Indonesia; a double-blind study of the impacts of antihelminthic treatment was performed on the Indonesian population to study the impact of worm clearance after 2 years (524). They found a number of associations with specific helminth infections as well as several taxa that were associated with helminth infection generally in both populations; for example, Lachnospiraceae incertae sedis were consistently negatively associated with helminths, while Desulfovibrionaceae, Olsenella, and Enterococcus were among taxa positively associated with infection. When studying the impacts of antihelminthic treatment after 2 years, they found that dewormed microbiota were more similar to infected than uninfected microbiota, suggesting that there may be lingering effects from helminth colonization (524).

Another group studied the impacts of gastrointestinal nematodes on the fecal microbiota of a Sri Lankan population; specifically, the authors examined individuals with current infection with A. lumbricoides, N. americanus, A. duodenale, and/or T. trichiura compared to those who were uninfected and those who had recently received antihelminthic treatment (525). While there was no observed difference in the alpha diversity of the communities, the authors found an increase in the beta diversity of the infected and treated groups. Additionally, there were some associations of specific taxa with infection status: infected subjects demonstrated increased abundances of Lactococcus, Akkermansia, and Verrucomicrobiaceae; uninfected subjects demonstrated an increased abundance of Leuconostocaceae; and treated subjects demonstrated an increased abundance of Bacteroides (525). Predictions of the functional potential of the communities indicated that helminth-infected subjects had downregulated ether lipid metabolism and apoptosis pathways and upregulated biotin metabolism pathways (525). Members of this group have also worked on several studies of experimental infection with N. americanus in human subjects, generally finding that infection was associated with minor increases in the microbial richness of the fecal and duodenal microbiota, particularly in the context of gluten challenge in subjects with celiac disease, but not with significant impacts on the community structure or abundance of specific taxa (526–528).

Other studies have studied the impacts of human infection with Schistosoma species on the composition of the gut microbiome. Such effects are particularly interesting, as this organism does not reside directly in the gut but resides in the vascular system. One study examined the impacts of S. haematobium, which resides in the urogenital vasculature and extrudes eggs into the bladder, on the fecal microbiomes of children in rural Zimbabwe, where the parasite is endemic (529). Those authors found 21 OTUs that varied with infection status, although only 5 were robust to multiple-hypothesis testing, all belonging to the genus Prevotella. Additionally, as all participants were treated with curative praziquantel at the time of sample collection, the authors performed follow-up at 12 weeks and found that treatment did not significantly impact the microbiomes of previously infected children at this time point (529). In a similar study, researchers studied the impact of infection with S. mansoni, which resides in the mesenteric vasculature and extrudes eggs into the gut, and praziquantel treatment on the microbiota of children in Cote d’Ivoire (530). They found that children with various levels of infection intensity were more likely to have an enrichment in the phylum Proteobacteria and specifically found an association between infection and the genus Klebsiella. In terms of praziquantel, treatment was associated with higher levels of the classes Bacilli and Erysipelotrichia after 24 h, and success of treatment was associated with higher levels of Fusobacteriales both before and after treatment (530).

In general, it seems clear that infection with a variety of both gut-resident and tissue-resident helminths can have impacts on the composition and function of the gut microbiota (Fig. 3). However, it is difficult to draw conclusions about specific changes, and in some cases, different groups have obtained contradictory results; for example, studies of humans naturally infected with gut-resident helminths have alternately found increased, decreased, and unchanged microbial diversity associated with infection (522, 523, 525). This may be due in part to the wide range of study conditions used. Across the studies, hosts include multiple animal models and various human populations, while helminths include species that inhabit the small intestine, large intestine, and vascular system. In addition, these studies utilize a multiplicity of analysis techniques, including different diagnosis strategies, sequencing platforms, 16S rRNA gene regions analyzed, databases used, and computational pipelines and analytics. Importantly, the majority of the human studies discussed here analyzed the fecal microbiota, frequently used as a proxy for the gut community due to the relative ease and lack of invasiveness of sample collection; however, it may miss changes that occur at the mucosal level, where gut-resident helminths are likely to reside, as well as differences between small and large intestinal communities. Despite such factors, some potential commonalities have arisen: several studies found an enrichment in Lactobacillaceae during infection of animal models with gut-resident nematodes, both human and animal studies found enrichment in potentially mucolytic (Mucispirillum, Akkermansia, and Prevotella) or sulfate-reducing (Desulfovibrionaceae, Desulfocella, and Desulfovibrio) bacteria, and several studies found evidence for microbiota disturbance persisting after helminth clearance.

Not only can helminths modify the native microbiota, but the presence or composition of the gut microflora may actually influence the course of infection with some helminths. Some of the first hints of such an effect came from studies of germfree animals. Experiments with germfree mice indicated that the presence of the gut microbiota has significant supportive effects on the course of infection with H. polygyrus. For example, infected conventional mice developed more adult worms, had longer infections, and showed higher mortality rates than germfree mice. In fact, as the infection progressed, the difference in adult worms harbored by the two groups grew wider, suggesting that the presence of the microbiota was required for efficient adult worm survival. However, the germfree mice also developed eosinophilia, intestinal wall thickening, and pronounced nodule formation in the small intestine in comparison to conventional mice; this suggests a strong immune response that may have contributed to reduced worm survival, possibly due to a lack of immunoregulatory influence from gut microbiota (531).

Follow-up studies from the same group added gnotobiotic mice monocolonized with a single bacterial species for comparison with conventional and germfree mice. For example, Chang and Wescott performed a study in which they used an unspecified Lactobacillus species isolated from conventional mice to create a monoassociated group (532). These mice appeared to display an intermediate phenotype between conventional and germfree mice in terms of the development of adult H. polygyrus worms, nodule development, and parasite egg production. Additionally, increasing the length of time between monoassociation and infection favored parasite survival, with parasite numbers similar to those in germfree mice at 2 weeks but increased parasite numbers in monoassociated mice at 6 weeks (532). Similarly, researchers used the nematode Trichinella spiralis to infect mice monoassociated with Bacillus mesentericus or Bacillus subtilis, which do not colonize the small intestine where this parasite resides, or P. aeruginosa, which can colonize both the small and large intestines (533). Mice monoassociated with P. aeruginosa displayed an intermediate phenotype, with researchers recovering more adult worms than from germfree mice but fewer worms than from conventional mice. On the contrary, mice monoassociated with Bacillus species were similar to germfree mice and contained low numbers of adult worms, suggesting that proximity between a bacterial species and the parasite is required for interaction. Both monoassociated mice and germfree mice also displayed eosinophilia relative to conventional mice, indicating a similar immune response (533). Finally, this group also demonstrated that the gut microbiota favors the development of another intestinal nematode, N. brasiliensis; while its larvae appear to migrate through the lungs at equivalent rates in conventional, monocolonized, and germfree mice, significantly more adult worms were recovered from the gut of conventional animals (534). The conclusions of this series of experiments were further supported by a study performed by a different group using the nematode Ascaridia galli to infect conventional, germfree, and monoassociated chickens. As in the above-described studies, rates of parasite establishment and survival were highest in conventional animals, lowest in germfree animals, and intermediate in animals monoassociated with either B. subtilis, Bacillus cereus, or Penicillium (535). Together, data from these studies suggest that even the presence of a narrow gut microbial community is able to promote parasite development and survival.

One issue with such studies is that complete abolishment of the microbial community from birth has significant effects on immune system development (115, 117, 143, 536, 537), and thus it is difficult to parse out impacts of the microbes themselves relative to immunological defects in germfree animals. Therefore, more recent work has sought to analyze links between members of the gut microflora and helminths in the context of a healthy, diverse microbial community. For example, in addition to links suggesting that helminth infection can promote Lactobacillus in mouse models, there is also evidence that Lactobacillus can promote the establishment of infection with H. polygyrus (506). Specifically, administration of Lactobacillus taiwanensis to a typically resistant mouse strain increased the levels of regulatory T cells and nematode establishment frequencies. Together with this group’s finding that H. polygyrus promotes Lactobacillus expansion, this indicates a self-reinforcing relationship in which the presence of Lactobacillus promotes H. polygyrus infection while the parasite increases the abundance of Lactobacillus (506). The authors of that study suggest that Lactobacillus-mediated T_reg expansion, potentially suppressing T_H2 immunity, may be the mechanism behind the increased susceptibility to H. polygyrus infection (506). In support of this finding, another group found that oral administration of either live or dead Lactobacillus bacteria before infection increased susceptibility to another mouse parasite, Trichuris muris, and led to reductions of both T_H1 and T_H2 cytokines (538).

There are a few studies that suggest that some species of helminth depend on the host microbiota in executing key stages of their life cycles. For example, Hayes et al. demonstrated that the murine nematode whipworm Trichuris muris may rely on host microbiota as a developmental cue (539). First, in vitro experiments showed that T. muris eggs did not hatch unless in the presence of bacterial or yeast cells. To test this in vivo, that group experimentally reduced the number of bacteria in mice with enrofloxacin before administration of T. muris eggs, finding a decreased number of established infections and a strong T_H2 immune response (539). Additionally, they performed the same experiment on mice lacking an adaptive immune system and found a similar reduction in parasite establishment in the gut, suggesting that acquired immunity and the enhanced T_H2 response are not solely responsible for this effect. It is possible that the requirement for the presence of microflora for hatching may be an adaptation evolved to ensure that parasite eggs hatch in the bacterium-lined cecum and colon (539). Similarly, another group found an altered course of intestinal schistosomiasis upon depletion of the gut microflora (540). After administering antibiotics and antifungal drugs to mice infected with S. mansoni, which resides in the venules surrounding the gastrointestinal tract and excretes eggs into the gut lumen, that group found that gut inflammation, intestinal granuloma formation, and egg excretion were all reduced compared to untreated mice. Interestingly, liver pathology and worm fecundity (egg production) were unchanged, suggesting specific local effects of the gut microflora that promote successful egg translocation (540). Furthermore, researchers have found that immunoregulation induced by helminth infection is at least partially influenced by the intestinal microflora. Mice infected with H. polygyrus showed reduced allergic airway inflammation in response to a dust mite extract compared with uninfected controls, but this effect was not seen in mice that were first treated with antibiotics to reduce the numbers of gut bacteria despite a similar number of adult worms establishing infection (541). Importantly, the effect could be restored by reestablishing a normal microbiota, via either oral gavage of cecal contents from or cohousing with untreated mice (541).

Finally, while most of this work focuses on the relationship between the host microflora and infectious nematodes, there are also cases in which microbes associated with the parasites themselves are relevant to the course of disease. Specifically, filarial worms harbor endosymbiotic bacteria of the genus Wolbachia, and a significant part of the pathology caused by these parasites occurs due to the host immune response to Wolbachia rather than to the nematodes themselves (542–547). Additionally, most filarial worms are unable to survive without their Wolbachia symbionts, presenting a target for antibiotic therapy rather than antihelminthic drugs to eliminate infection (543, 544, 546).

Generally, this work suggests that the gut microflora can play a significant role in the establishment and maintenance of helminth infections (Fig. 3). Gut-resident or gut-proximal helminths appear to survive and reproduce more efficiently in the presence of a diverse community of gut bacteria, although monoassociation experiments demonstrate that even microflora lacking diversity can be utilized by parasites to promote infection. Given that humans have been infected with helminths for most of their existence and the lack of germfree hosts in nature, it is possible that helminths have coevolved to take advantage of the presence of the gut microflora for their successful development. Potential interactions include the use of bacterial signals as helminth developmental cues, bacterium-mediated T_reg activation allowing for effective colonization despite T_H2 polarization, or even the consumption of bacterium-produced metabolites.

The great variety of helminth and host species used in the studies described here makes it difficult to draw specific conclusions about the effects of a given organism, but together, they paint a compelling picture of both bipartite and tripartite interactions between resident gut commensals, the host immune system, and helminth infections. The presence of gut microflora appears to support infection with certain helminths, including via direct interactions between the microbiota and helminths to promote life cycle progression or reproduction, immunomodulation to reduce the host response to helminth infection, and potentially other mechanisms. At the same time, infection with a variety of helminths appears to alter the composition and potentially the function of the gut microflora, although whether this occurs via local interactions between parasites and commensals, systemic immune changes favoring the proliferation of certain gut bacterial species, or a combination of the two remains unclear. There is some evidence that infection with helminths may impact the diversity of the gut microflora, although the directionality of this impact varied; however, whether or not overall diversity was altered, infection was frequently able to alter the abundances of a relatively large proportion of microbiome members.

These observations give rise to a number of interesting research directions. In terms of helminth infection prevention, there is evidence that microbiome depletion can reduce infection by or reproduction of certain helminth species. It would be useful to know how widespread this effect is across different clinically relevant helminths and to elucidate specific mechanisms, such as egg hatching in T. muris or egg excretion in S. mansoni, by which the course of infection is altered. Additionally, these effects have not yet been studied in humans, offering another point of study. There may also be specific bacterial taxa that interact with and promote infection with helminths, as seen in the mutually supportive relationship between H. polygyrus and Lactobacillus in mice. Identifying such interactions in other helminth infections could allow for the development of targeted interventions that might have lesser effects than microbiome abolishment, which can reduce microbial diversity and promote dysbiosis after recovery.

On the other side, the studies indicating the potential for helminths to alter the gut microbiome have implications for understanding the pathology of a number of neglected tropical diseases and for identifying potential side effects of helminth therapies for autoimmune disorders. For example, are helminths themselves directly responsible for the full extent of immunomodulation that occurs, or might they also modulate the gut microbiota to promote immunoregulatory bacteria? On a related note, does microflora alteration occur due to local interactions between gastrointestinal helminths and gut flora, as a consequence of a systemic immune alteration resulting from helminth infection, or both? While many of the immunological effects of helminth infection are well known, the exact mechanisms of the interactions between host, commensal, and helminth remain unclear. Studies of the impacts of non-gut-resident helminths, such as filarial worms and schistosomes, on the microbiota as well as profiling gut communities both proximal and distal to the site of gastrointestinal helminth infection may help to answer these questions.

Additionally, given that there is evidence from both human and animal studies that microbiome changes can persist even after parasite clearance, there is a question of what effect these changes might have on host health. If helminth infection alters the composition of the gut microflora, it leaves open the possibility of dysbiosis, inflammation, or opportunistic infections as potential sequelae, even after the parasite infection itself is treated. Especially given the interest in helminth therapy, it will be important to investigate and learn how to mitigate potential negative side effects of parasite-induced microbiome alterations; additionally, identifying specific helminth products or components that can promote beneficial immunomodulatory effects without requiring infection with live parasites is an important direction of research (548, 549). Finally, while many studies were able to identify changes in the composition and predicted function of the microbiota during helminth infection, full metagenomics and metatranscriptomics will be important in understanding the true functional differences in the microbiota, including nonbacterial members, under these conditions.

The approximately 10¹⁵ phages that inhabit the human intestine are predominantly dsDNA viruses in the Myoviridae, Podoviridae, and Siphoviridae families of the order Caudovirales and ssDNA viruses of the family Microviridae, constituting an estimated 1,200 viral genotypes (Fig. 1) (552–554, 564, 565). The Caudovirales and Microviridae phages have an average genome size of 30 kb and exhibit both lytic and temperate lifestyles, although temperate phages dominate the human gut (552–554). The temperate lifestyle often leads to the integration of phages into bacterial hosts as either episomes or chromosomal prophages. An analysis of prophages in both Gram-positive and Gram-negative bacteria found prophages with lengths of 10 kb or more in 71% (40/56) of the bacteria examined (566). These prophages are replicated along with bacterial DNA and expand the genetic repertoire of the bacterium in which they reside, accounting for up to 20% of the bacterial genome (567). Multiple environmental stressors, including reactive oxygen species (ROS), reactive nitrogen species (RNS), UV radiation, or antibiotics, can lead to the induction of the prophage; this results in a lytic cycle, replication, and escape of the phage progeny and bacterial host death (568). Thus, phages in the human virome have the ability to modulate bacterial populations through a lytic lifestyle or can potentially provide a bacterium with novel traits while integrated as a prophage (Fig. 4). Widespread lysis of commensal bacteria through prophage induction could potentially facilitate colonization by pathogenic bacteria in formerly colonized niches (Fig. 4). Findings outlining the composition of phage communities have also revealed interesting trends in the establishment and steady-state dynamics of the human intestinal virome.

Intriguingly, a recent study identified a novel type of bacteriophage termed crAssphage, a name derived from the cross-assembly tool used to assemble the first 97-kb crAssphage genome from metagenome samples. crAssphages are thought to be the most prevalent human-associated virus and have been found to comprise up to 90% of reads from human fecal virome metagenomes (569). The bacterial host range of these phages is still not understood, but early investigation suggests that crAssphages may be associated with the bacterial phylum Bacteroidetes, and they are thought to potentially form a family within the order Caudovirales (570). Initial studies found that 80% of the predicted crAssphage proteins did not have matches in reference databases, but recent efforts have had more success in characterizing these proteins (570). crAssphages have been propagated in vitro and imaged for the first time, presenting a novel opportunity to study these phages in the laboratory setting (571). Guerin et al. further identified novel crAss-like phages from metagenomic data sets and proposed taxonomic assignments for these bacteriophages (571). While these phages remain understudied, their widespread abundance in human microbiota metagenomic data sets suggests that crAssphages may have an important role as a constituent of the human virome.

Much like the development of the bacterial community of the human microbiome, phages are partially inherited from the mother, and populations undergo a maturation process over the lifetime, forming a steady-state community of phages known as the “phageome.” From birth to 2 years of age, the bacteriophage populations of infants transition from higher to lower diversity (572). During this time, there is also a shift in the relative abundance of phage populations, highlighted by a decrease in Caudovirales and an increase in Microviridae abundances (572). The changes in bacteriophage composition from infancy to adulthood are not well characterized and remain an area to be explored. In adulthood, phages of the families Myoviridae, Podoviridae, and Siphoviridae comprise a majority of the taxonomically identifiable viruses, while Microviridae make up a smaller but significant proportion (553, 554). Among healthy adults, the intestinal phageome exhibits high interpersonal variation, with individuals displaying distinct phage compositions. In contrast, in the same individuals, intrapersonal variation over time was comparably low. Ninety-five percent of viruses were retained over a 1-year study, similar to the gut bacterial population (28, 553). These findings suggest that humans harbor individual but highly stable intestinal phageomes.

It is well established that the bacterial and even fungal members of the microbiome play a key role in human health and disease, but the role of the virome is less understood. Attempts to characterize the virome of healthy individuals have naturally led to studies of how this community changes during disease states, including the role of phages and eukaryotic viruses during bacterial infection and host microbiome dysbiosis. Studies have revealed that the human virome is altered during disease states and plays a role in bacterial pathogenicity but can also be protective against bacterial infection.

While the role of bacteria has been examined in the context of inflammatory gut dysbiosis, we are just beginning to learn about the interplay between conditions such as IBD and the virome (573). For example, Manrique et al. analyzed metagenomic data sets of intestinal bacteriophages from 62 healthy subjects. Their study identified 155 bacteriophages that were determined to be part of the “healthy gut phageome,” although these represented only 4% of the estimated phage community. Of the 155 phages, 132 “common” bacteriophages were found in 20% to 50% of individuals, and a further 23 “core” bacteriophages were observed in >50% of subjects (565). Interestingly, this core phageome was depleted in subject groups with IBD, suggesting that the intestinal phageome may play a role in maintaining a healthy gut (565). However, this could also result from a reduction in bacterial host diversity common to IBD (574). Previously, studies of the enteric virome in IBD had observed that the gut mucosa of Crohn’s disease patients harbored higher numbers of virus-like particles (VLPs) (2.9 × 10⁹) of the order Caudovirales than healthy patients (1.2 × 10⁸) (556). Further metagenomic evaluations of IBD patients confirmed this observation, finding that the disease state is characterized by greater interpersonal variation in virome composition along with an increased abundance and richness of Caudovirales bacteriophages but less overall diversity of the virome (575–577).

Additionally, metagenomic analysis of both control and colitis-induced mice has revealed responses of the virome to inflammation in a more controlled setting. In mice with colitis, there was an expansion of phages that infect potential pathogens such as Proteobacteria, which are known to expand during gut inflammation and IBD (577, 578). Phages of Streptococcus and Alistipes bacteria were more abundant in the colitis group, but interestingly, the levels of the bacterial hosts did not change with colitis (577). This suggests that increased levels of a phage are not always dependent on elevated host abundance; instead, lysogenic phages may be induced by inflammation leading to replication and excision of phage. One of the most significant findings from this study is the overlap in phage metagenomes between the mouse model and human studies, suggesting that this murine model is suitable for studying phage dynamics of human IBD (576, 577). Together, data from these studies suggest that inflammatory disease in the gut can lead to a defined shift away from a healthy virome, and IBD-specific shifts in phages may reflect the expansion of pathobionts associated with disease or the induction of conditions that favor a lytic lifestyle.

Human disease provides an opportunity to study not only how the virome shifts in response to a perturbation but also how viruses can play an active role in facilitating and exacerbating bacterial infections. For example, the highly unique environment of the cystic fibrosis (CF) lung drives selection and rapid evolution of bacterial pathogens and their phages through various ecological pressures. P. aeruginosa is the most common pathogen of the CF lung, and the density of this bacterium is positively correlated with densities of its phages (579); the phages play a role in the ability of P. aeruginosa to survive and thrive within the CF lung, highlighting the importance of the human phageome during bacterial infection (579). A subset of phages, known as LES phages due to their identification in the Liverpool epidemic strain of P. aeruginosa, have been shown to integrate as prophages and confer fitness advantages to their hosts. Isolation of a polylysogenic strain of P. aeruginosa (LESB58) from a CF patient allowed researchers to determine the fitness advantages associated with the various prophages through competition assays in a rat lung infection model; by using signature-tagged mutagenesis of 3 of the prophages in the LESB58 genome, they found that a disruption of these prophage genes greatly reduced the competitiveness of strains in vivo (580). Moreover, growth of P. aeruginosa in the presence of lysogenic LES phages in an artificial sputum medium led to the development of mutations promoting fitness, frequently through prophage integration into genes such as those coding for type IV pilus machinery (581). This was in agreement with previous studies which showed that mutations in type IV pilus-associated genes increased the fitness of P. aeruginosa in the murine lung, potentially by providing immunity to superinfection by other phages that infect via this molecule (582).

Additionally, P. aeruginosa can utilize filamentous bacteriophages during the biofilm life cycle. For example, phages produced by the bacteria can associate with polymers to create a “liquid crystal” extracellular matrix that enhances adhesiveness and protection from desiccation and antibiotics (583). Counterintuitively, these phages can also contribute to the biofilm by inducing cell death in certain populations, specifically those at the center of microcolonies; this may serve as a form of directed cell death promoting the dispersal of small-colony-variant cells from the biofilm (584, 585). In fact, a strain of P. aeruginosa lacking the filamentous phage Pf4 formed abnormal and unstable biofilms, and mice infected with this strain survived longer that those infected with the wild-type counterpart with an intact phage (585). Phage production may be linked to the anaerobic growth mode of P. aeruginosa biofilms in vivo, as prophage genes were found to be strongly upregulated during anaerobic respiration of P. aeruginosa (586).

Phages can also play a role in facilitating the pathogenicity of ordinarily commensal bacteria by providing novel routes for infection. The bacterium Streptococcus mitis is typically an oral commensal in humans, but it can also cause infective endocarditis. Two proteins encoded by lysogenic phage SM1 play a direct role in the binding of S. mitis to platelets in the bloodstream, leading to endocarditis (587, 588). The phage proteins PblA and PblB attach to the cell wall of S. mitis, where they mediate binding of the bacteria to platelets; the enhanced platelet binding then increases the pathogenicity of S. mitis in an endocarditis mouse model (588). Viral metagenomic and 16S rRNA gene analyses have revealed that both the phage and its host are highly prevalent within the human oral cavity, and SM1 can be induced by dietary items such as soy sauce, leading to increased levels of PblA and PblB (589, 590).

Similarly, integrated prophages can confer traits that increase the virulence of many pathogenic bacteria. Pathogenic strains of Escherichia, Bacillus, Pseudomonas, and Burkholderia all contain greater numbers of prophages within their genomes than their benign relatives of the same species (591). In pathogenic E. coli strains, several phages, including Stx, Stx2, λ, and CP-933C, encode virulence factors and toxins (592–594). The human pathogen Vibrio cholerae derives its virulence from two phage components, CTX and TCP, which arise from the phages CTXφ and VPIφ (595–597). However, sometimes phages can confer traits to bacterial symbionts that are beneficial to the host. One striking example is the tripartite interaction between the eukaryotic pea aphid Acyrthosiphon pisum, its hemocoel-dwelling bacterial symbiont Hamiltonella defensa, and the bacteriophage APSE-3. This aphid is often parasitized by the wasp Aphidius ervi, which deposits eggs within the aphid hemocoel that kill the host when they pupate (598). However, H. defensa can prevent development of the wasp eggs, sparing the aphid host, but only when it contains the APSE-3 prophage, which contains the gene for the toxin required to kill the wasp eggs (599).

Although phages make significant contributions to the pathogenicity of many bacteria, there is mounting evidence that the host eukaryotic virome may provide protection from bacterial infection through activation of the host immune system and direct interaction with pathobionts (600). Mouse models have been used to demonstrate the ability of latent murine herpesvirus (MHV68) or cytomegalovirus (CMV) infection to activate macrophages and increase basal immune function in a manner that provides protection from bacterial challenges with Listeria monocytogenes and Yersinia pestis (Fig. 4) (601). The activation of basal immune function to protect against bacterial pathogens was further examined by MacDuff et al. Their work showed that in immunodeficient mice, chronic MHV68 infection increased basal levels of gamma interferon (IFN-γ) and its effector molecules, allowing for rapid clearance of L. monocytogenes (602). Additionally, murine norovirus (MNV) infection was found to restore intestinal physiology and immune function in germfree or antibiotic-treated mice in an IFN-α-dependent manner, allowing the MNV-infected mice to cope with challenges from chemically induced intestinal injury and C. rodentium infection (603). This work highlights the ability of the mammalian virome to maintain intestinal health through activation of the host immune system (Fig. 4).

It has also been postulated that the virome can protect the host from bacterial infection by serving a barrier function. Phage concentrations are increased in the mucus layer of humans, mice, and marine invertebrates, which results from a weak adherence between Ig-like domains of phage capsids and the glycan residues of mucin. This layer then provides a site for lytic phage infection of bacteria, which protects the underlying epithelial cells. In an in vitro model, the presence of phage increased bacterial cell death and decreased epithelial cell mortality in a mucus-dependent manner (604). Further research into this model suggests that the interaction with mucin reduces diffusive motion, increasing the number of encounters with potential bacterial prey on the mucus (605). Together, the results present a model for phage-driven, non-host-derived immunity of mucosal surfaces termed by those authors the “bacteriophage adherence to mucus” (BAM) model (Fig. 4) (604–606).

The growing body of research on the virome during health and disease states is transforming our understanding about the role of viruses in the human body. Studying viral community changes during IBD has led to insights into a potential role for phages during gut dysbiosis and the possibility of identifying specific phages as biomarkers of disease. Furthermore, it is commonly observed that phages can facilitate the virulence of bacterial infections, but the role of eukaryotic viruses and phages as potential players in host immunity is now being appreciated.

Horizontal gene transfer (HGT) represents a route for the dissemination of genes between bacteria, including genes coding for favorable fitness-related traits, virulence factors, and antibiotic resistance. There are several ways in which phages are involved in HGT in the microbiome: transduction, by which phage mediates the transfer of bacterial DNA from one bacterium to another; specialized transduction, where prophage excision involves packaging of both the prophage and adjacent host genes; and lysogenic conversion, during which a prophage causes a phenotypic change in the host cell (607). Analyses of HGT events across metagenomic data sets have demonstrated that human-associated bacteria exhibit far more gene transfer than those bacteria not associated with humans. After examining gene transfer events in different ecological niches, Smillie et al. suggested that HGT is enriched in bacterial communities inhabiting specific body sites, such as the gut (608). The rich density of bacteria and phage within the human microbiome makes it a potential hot spot for phage-mediated HGT of virulence and antibiotic resistance genes (ARG).

Antibiotic resistance genes are present in bacterial populations within the human intestinal microbiome as well as the human gut virome (609, 610). In vitro studies of commensal and pathogenic bacteria have found that many species are capable of phage-mediated gene transfer (595, 611–614). While HGT is normally limited to transfer within a species, some phages are capable of transferring virulence and antibiotic resistance genes between bacterial species and even genera (Fig. 4) (612, 615). For example, strains of Lactobacillus gasseri contain multiple integrated prophages, which can transduce both other strains of L. gasseri and strains of L. acidophilus (611). This expands the possibilities for the transfer of genes to opportunistic pathogens residing within the human microbiome, resulting in more-challenging bacterial infections. The transfer of virulence-, resistance-, and fitness-enhancing genes via phage has also been shown to occur in hosts infected with S. aureus, E. coli, P. aeruginosa, and S. enterica serovar Typhimurium (568, 579, 616). In one example in a moth larva model, S. aureus phage φ11 was released from a subpopulation of lysogenized cells, and these phages then infected neighboring susceptible bacteria that contained a resistance gene. These newly infected bacteria underwent cell lysis, and some of the resulting φ11 phage particles packaged the resistance gene of the lysed neighbors and transduced members of the lysogenic population, thus providing them with novel resistance (616). Those authors termed this process “autotransduction” and postulated that it may increase the fitness of a bacterial population through the acquisition of beneficial genes of neighboring populations (616).

Many phages carrying virulence and resistance genes reside in bacteria as prophages. Some species of bacteria carry prophages that can undergo spontaneous induction into lytic growth; for example, L. gasseri contains phage that can undergo spontaneous induction at fairly high levels, which could contribute to HGT in human gastrointestinal and vaginal microbiota (611). However, in many species, induction of prophages to excise from their host and infect other bacteria occurs through environmental or host triggers, including UV radiation, antibiotics, hydrogen peroxide, ROS, and, more broadly, SOS response activation (Fig. 4) (568, 617–619). Recent work has shown that inflammation-induced production of ROS triggers the SOS response, leading to subsequent prophage induction and HGT of a phage-encoded virulence factor (568). Within the mouse gut, S. enterica serovar Typhimurium triggers inflammation, which leads to the production of ROS and RNS; these trigger the S. enterica SOS response, in turn activating the tum antirepressor of prophage induction, thus allowing for the lytic induction of the SopEφ prophage. The new phage progeny then infect naive S. enterica and through lysogenic conversion transfer the phage-encoded virulence factor SpoE (568). Therefore, the host inflammatory response to bacterial infection may contribute to horizontal gene transfer by phages in the microbiome (Fig. 4).

Antibiotics are powerful therapeutic tools that have profound effects on the human microbiome, including the induction of prophages leading to HGT and the potential spread of antibiotic resistance (Fig. 4) (620, 621). For example, the Stx-encoding prophages in pathogenic Shiga toxin-producing E. coli (STEC) were shown to be induced by treatment with the fluoroquinolone antibiotic ciprofloxacin (622). In mice, ciprofloxacin triggered not only prophage induction but also transmission of Stx phage to naive bacteria. Ciprofloxacin treatment increased Stx transductants and Stx production, resulting in more-severe STEC infection and ultimately a significant increase in mortality of the mice (622). This represents one of the challenges of treating enteric infections, as antibiotics can exacerbate ongoing infections by inducing prophages and spreading virulence. Later experiments in the murine intestine showed that antibiotic treatment can increase potential phage-bacterium interactions within the microbiome as well as enrich the enteric population of phages for carriage of traits, including resistance to the antibiotic used as well as to other antibiotics (621). Naive bacteria that were infected ex vivo with phages from antibiotic-treated mice obtained higher levels of resistance than phages from non-antibiotic-treated mice, suggesting that antibiotic resistance in the intestinal phageome can be enriched by drug treatment and has the potential to provide functional benefits to members of the microbiome through HGT (621). However, it is possible that the contribution of phages to ARG transfer is somewhat overestimated due to use of lax metagenomic data annotation for ARG databases, bacterial DNA contamination, and unverified resistance annotations (623).

There is no doubt that phages play a role in the mobilization of virulence and resistance genes through HGT. Through transduction and lysogenic conversion, phages are able to disseminate genes and modulate the functional capacity of the microbiome. Intriguing research suggests a role for phages in liberating DNA for the process of natural transformation through bacterial lysis, further implicating the role of phages in HGT (624). Factors including antibiotics and host inflammation have also been shown to impact phage-bacterium interactions within the context of the microbiome (Fig. 4). Together, these findings reveal the importance of phage-mediated HGT within the microbiome and its impact on host health. Analyzing metagenomic data sets for HGT events has been shown to be a powerful approach for understanding the dynamics of gene flow in the context of the human microbiome, and future studies should aim to utilize these analytical techniques to better understand the role of bacteriophage in HGT within the human microbiome (608, 625).

Since Félix Hubert d’Hérelle coined the term “bacteriophage” in 1917, researchers have been experimenting with utilizing phages for the treatment of bacterial infections in humans (626). In a summary of this early work, d’Hérelle articulated enthusiasm for the possibility of treating deadly bacterial infections such as V. cholerae, septicemic Staphylococcus, and Y. pestis. Concluding his report, d’Hérelle recognized that extensive research will be required to fully demonstrate the efficacy of phage therapy (626). Almost one century later, the field of phage therapy finds itself in a similar position: excited over the possibility of a novel treatment for bacterial pathogens but in need of more clinical trials in order to understand the efficacy and safety of bacteriophages as treatments. The discovery of antibiotics almost completely diverted attention away from phage therapy research in the United States, although bacteriophages have still been used and developed as therapies in Eastern Europe (627). However, the current antibiotic resistance crisis is weakening our antibiotic arsenal, and focusing more attention on the potential utilization of bacteriophages may be beneficial in clinical practice to combat multidrug-resistant (MDR) bacterial infections (628, 629). Recent studies have begun to evaluate the safety, efficacy, and possible role of phage therapy in treating bacterial infections.

While there are currently no FDA-approved bacteriophage treatments for bacterial infections in humans, there has been research into efficacy and safety in animal models as well as a limited number of cases involving humans. Animal models have shown efficacy when utilizing lytic bacteriophage treatments as prophylaxis or following symptoms for infections involving K. pneumoniae, Enterococcus faecium, V. cholerae, E. coli, C. difficile, S. aureus, and P. aeruginosa (Fig. 4) (630–636). In these studies, treatment with lytic phage showed positive health outcomes that could match or exceed those of traditional antibiotic therapies or enhance treatment efficacy when combined with antibiotics. Furthermore, research has shown that phage therapy can succeed in treating antibiotic-resistant Enterococcus infections in a mouse model, supporting its potential benefit in a world of increasing resistance to current drugs (631, 637).

While results from animal models have suggested that bacteriophages are able to successfully treat bacterial infections, there is far less evidence for phage-based treatments in humans. When used to treat acute V. cholerae infection in humans, bacteriophages displayed no clinical efficacy, compared to the traditional antibiotic therapy (tetracycline), which was successful in treating the infection (638). Additionally, more recent clinical applications of phage therapy were unsuccessful in treating E. coli gastrointestinal infections in humans while also highlighting the need for more highly targeted phage cocktails to be used in treatment (639). However, despite these failures, phage therapies are generally considered to be safe, and studies have reported no adverse effects related to phage treatment (640–643). Another benefit observed was that unlike antibiotics, which have a broad effect range and can cause intestinal dysbiosis, phage therapies are much more targeted and can preserve the integrity of the host microbiome (632, 642). Despite the lack of widespread use, in the United States there have been recent case reports in which experimental phage therapy was administered to patients as a last resort when infections were not responsive to antibiotic therapy. In these limited reports, administration of phage cocktails was associated with resolution of serious bacterial infections (644, 645). In the United Kingdom, a controlled study of patients with drug-resistant P. aeruginosa ear infections showed significant improvement in patients receiving bacteriophage therapy compared to placebo (641). While these cases showcase the potential power of phage therapies, controlled clinical studies are still required before such treatments become mainstream.

The use of bacteriophage therapy is now evolving beyond the idea of using lytic phages to kill bacteria. Instead, researchers are devising ways of using phages to sensitize bacteria to antibiotics in order to enhance their efficacy and circumvent certain mechanisms of resistance. Pioneering work by Lu and Collins utilized a genetically modified lysogenic phage of E. coli that was able to modulate antibiotic efficacy in vivo through the regulation of gene networks involved in antibiotic sensitivity (646). Researchers increased the lethality of several antibiotics by suppressing the bacterial SOS response through a phage-borne lexA3 gene, thus opening the possibility of utilizing engineered lysogenic phages as adjuvants to increase antibiotic efficacy in treating bacterial infections (646). Although not intended for use in vivo, Edgar et al. proposed a method to resensitize bacteria to antibiotics using engineered phages: to combat drug resistance developed through mutations in bacterial genes, phages modified to contain the sensitive version of the resistance genes were used to lysogenize bacteria (647). In this case, a sensitive gyrA gene was introduced via phage to a gyrA mutant E. coli strain that was resistant to the quinolone nalidixic acid. Upon lysogeny with the phage carrying a sensitive gyrA gene, the E. coli strain became more susceptible to nalidixic acid (647). In a similar vein, a separate group used phages to deliver a CRISPR-Cas system to target and destroy antibiotic resistance plasmids within bacteria to increase the effectiveness of antibiotic therapies (648). Additionally, Chan et al. found that exposing MDR P. aeruginosa to both bacteriophage and antibiotics can help to overcome resistance mechanisms (649). Their MDR P. aeruginosa strain attained resistance through an efflux pump, which is the same system that the lytic phage OMK01 uses to bind to and infect the bacteria. Losing a component of the efflux pump in the bacteria made the bacteria susceptible to tetracycline but resistant to OMK01; when the efflux system was functional, the bacteria were tetracycline resistant but also sensitive to phages (649). This trade-off displays the potential use of combined antibiotic and phage therapy to target MDR bacteria in novel ways. Together, these findings represent a novel method for utilizing phage to enhance the efficacy of antibiotics for treating bacterial infections.

Although phage therapy provides a novel method of killing compared to that of antibiotics that may circumvent problems with MDR bacterial infections, bacteria also possess defenses against phages that present potential roadblocks to developing phage therapies. Bacteria can develop a number of resistance mechanisms against phages, including prevention of phage attachment and subsequent DNA injection, restriction-modification and CRISPR-Cas systems to destroy phage genomic DNA, and abortive infection by which an infected bacterium sacrifices itself to prevent phage replication and further infections. These systems, along with strategies employed by phages to circumvent bacterial defenses, are described in detail elsewhere (650–652). The results of research done in animal models as well as some intriguing case reports show promise for the future of phage therapy, although more controlled and targeted human trials are needed. Furthermore, utilizing phages as an engineerable tool to treat MDR infections could play an important role in combating the antibiotic resistance crisis.

Study of the virome is limited and mainly focused on the intestine and phages. Unlike bacterial and fungal studies, there is no universal marker or phage equivalent for the 16S rRNA gene or ITS markers. This makes characterization of viral communities difficult and heavily dependent on metagenomic sequencing, which is limited by the low number of viral sequences in databases; for example, there are 2,040 complete Caudovirales genomes in the NCBI database. Thus, a majority of sequences from these metagenomic studies cannot be classified and are from unknown viruses, the “viral dark matter.” The recent identification of crAssphages as the most prevalent component of the human virome is an example of how much of this field remains to be explored and presents an opportunity to uncover the role of these bacteriophages in the context of the human microbiome. In addition, challenges remain in collecting viral DNA without introducing contamination and in obtaining sufficient quantities for metagenomic analysis, and culture-based work is extremely limited due the unknown host range of most phages. In addition to taxonomic characterization, there is a substantial knowledge gap in understanding the interactions between the virome and the microbiome and how these interactions may impact human health and disease. Research has shown that gut dysbiosis can lead to a shift away from a core phageome and that phages are capable of increasing pathogen virulence and drug resistance through horizontal gene transfer. However, it also appears that the human virome may be able to help prevent bacterial infection and maintain health. The role of phages in the clinical setting has recently generated interest to help combat the growing threat of antibiotic resistance, leading to the development of new and creative strategies to potentially utilize phages in the treatment and prevention of MDR bacterial infections.