Microbiome as an Immunological Modifier

Abstract

Humans are living ecosystems composed of human cells and microbes. The microbiome is the collection of microbes (microbiota) and their genes. Recent breakthroughs in the high-throughput sequencing technologies have made it possible for us to understand the composition of the human microbiome. Launched by the National Institutes of Health in USA, the human microbiome project indicated that our bodies harbor a wide array of microbes, specific to each body site with interpersonal and intrapersonal variabilities. Numerous studies have indicated that several factors influence the development of the microbiome including genetics, diet, use of antibiotics, and lifestyle, among others. The microbiome and its mediators are in a continuous cross talk with the host immune system; hence, any imbalance on one side is reflected on the other. Dysbiosis (microbiota imbalance) was shown in many diseases and pathological conditions such as inflammatory bowel disease, celiac disease, multiple sclerosis, rheumatoid arthritis, asthma, diabetes, and cancer. The microbial composition mirrors inflammation variations in certain disease conditions, within various stages of the same disease; hence, it has the potential to be used as a biomarker.

1. Introduction to the Human Microbiome

The term “microbiome” was introduced by the Nobel Laureate Joshua Lederberg and McCray in 2001, and it is used interchangeably with the term “microbiota” [1]. The two terms, however, have distinct meanings: Microbiota is defined as the entirety of microbes (bacteria, archaea, fungi, and viruses) that are found in a particular environment at a particular time [2], whereas the term microbiome refers to the collection of genomes encoded by the microbiota [2].

Microorganisms inhabit numerous human body sites including epithelial barriers and body fluids such as the skin, mouth, saliva, nose, vagina, vaginal secretion, seminal fluid, mammary glands, and the largest number residing in the gut due to the bioavailability of nutrients [3]. It is estimated that collectively the human gut microbiota are at least as numerous as the total cells in the human body by ten to one [4, 5], consisting of about 100 trillion microbes that represent around 5000 distinct species and make up to 2 kg of the human body weight [3]. The human microbiome is often referred to, as “our second genome” due to its size and sheer vastness [5]. In particular, the gut microbiota is now considered an “organ” of its own and is nicknamed “our second brain” due to its role in various immunological, neuronal, and endocrine responses [6].

Studies of the diversity in the human microbiome initially started in 1680, when Antoine van Leeuwenhoek compared his own oral and fecal samples using self-made microscopes [7]. He observed that the two habitats had distinct microbes and this observation led him to compare samples from healthy and diseased individuals where he also noted striking differences in their microbial communities [7, 8]. However, owing to the limitation of the in vitro microbial cultivation, the complex and dynamic nature of our microbiota was not fully recognized up until recently. This was facilitated by the advent of high-throughput sequencing technologies and the undertaking of large-scale metagenomics endeavors such as the Human Microbiome Project (HMP) that was initiated by the National Institutes of Health in the USA [9] and the Metagenomes Human Intestinal Tract (MetaHIT) project developed by the European Commission [10]. These projects have expanded our understanding of the Human microbiome. The MetaHIT study reported gut metagenomes from stool samples of predominantly healthy cohort of 124 European adults [11]. Whereas, in an effort to define the healthy human microbiome, the HMP researchers recruited 242 volunteers (129 males, 113 females) and sampled tissues from 15 body sites in men and 18 body sites in women. By incorporating several complementary techniques and analyses including 16S ribosomal RNA (rRNA) gene sequencing, whole-genome shotgun sequencing (WGS) and aligning the assembled sequences to the reference microbial genomes [12, 13] they were able to define the microbiome composition of each body site they sampled from [9]. They showed that, various body sites show a tenfold difference in the estimated microbial richness, with the gut being the richest ecosystem [9]. They also confirmed that a high interindividual variation of the microbiome exists and concluded that a healthy microbiome can be achieved by maintaining a balance between all the factors that contribute to the microbiome makeup (will be discussed later).

The most extensive research has been conducted with the gut ecosystem, in order to assess the impact of the gut microbes on human health and predisposition to various diseases [11, 14]. The gut microbiomes described as part of the MetaHIT cohorts indicated that a large proportion of the microbial genes are shared between individuals of this cohort and that more than 99% of these genes are bacterial, representing 1000–1150 bacterial species [11]. A healthy gut microbiota is mainly composed of strict anaerobes dominated by two main phyla: Bacteroidetes and Firmicutes [13].

In a healthy individual, many other body sites are occupied by microbial communities although not as well studied as the gut [13]. One of the largest human-associated microbial habitats is the skin surface, and is colonized primarily by Corynebacterium, Propionibacterium, and Staphylococcus [15]. The composition of the microbial communities shows more similarity “within habitats” as compared to “between habitats.” For example, samples collected from the oral communities show similar composition between different individuals, whereas compared with other communities within the same person, the oral ecosystem looks vastly different [13]. The oral and salivary microbiomes mirror the gut microbiota in terms of complexity and diversity [13].

An ecologically stable vaginal microbiome is dominated by species of Lactobacillus (L. crispatus, L. iners, L. jensenii, or L. gasseri) or by other microbes including Gardnerella vaginalis and Atopobium spp., among others and is classified in community state types [16]. This microbial diversity is affected by race, ethnicity, pregnancy, hormonal changes, sexual activities, and hygiene practices, among others [16–18]. The vaginal microbiome plays an important role in the prevention of genital and urinary tract infections and in prevention sexually transmitted diseases, and multiple studies have shown that an abnormal vaginal microbial composition is associated with an increased risk of HIV infection [19].

There are body sites particularly difficult to characterize, due to the low microbial load in healthy individuals, for example, the lung microbiota has a low microbial density, especially in the absence of an infection or pulmonary disease, as under normal conditions, the human lungs harbor approximately 2.2 × 10³ bacterial genomes per cm² [20]. It is also worth-noting that the identification of these microbial genomes remains tedious due to the technical challenges associated with the sampling from such sites. Similarly, habitats such as mammary glands, breast milk [21], placenta [22], and amniotic fluid [23] are of great interest due to their potential role in establishment of the early infant microbiome, but most studies have relied on traditional culture dependent approaches and/or single-cell techniques, such as FISH or microfluidics to study their microbiomes. Development of better sampling techniques facilitating the use of larger-scale high-throughput metagenomics studies are thus required to establish the exact composition and functionality of these challenging microbial habitats [23].

2. Universal Properties of the Human Microbiome

Based on the HMP and other research studies the “universal” properties of a healthy human microbiome have been identified. Dr. Lita Proctor, the coordinator of the HMP, pointed out few of these properties [13]. First, unlike the human genome, the human microbiome is not inherited but acquired upon birth. Newborns are described as “microbe magnets,” where the microbiota of the babies born vaginally resembles closely the mother’s vaginal microbiota, whereas those born via cesarean section (C-section) derive their microbiota from the mother’s skin [13] or the surrounding environment [24]. A second universal property is that, the adult microbial community composition is dictated by specific body sites, each body habitat such as gut, skin, oral cavity and urogenital tract, has a distinct microbial composition, regardless of the host gender, age, weight, or any other factor. HMP 16S rRNA data revealed clustering of specific microbial taxa within each of the body sites, and this clustering was driven by various factors including temperature, humidity, and pH, among others [25]. The final universally applicable property is that the gut microbiome changes over the lifetime, advances in metagenomics techniques have facilitated our understanding of the microbiome composition and its variation from infancy to old age. The ELDERMET project reported that the microbiomes in elderly people (aged 65 and over) are quite distinct from the microbiomes of middle-aged adults [26].

3. Inception of the Microbial Colonization

One of the most essential processes in the human life is the initial colonization of the infant’s gut by microbes, as this sets the stage for the establishment of a stable microbiome later in life [27]. Microbial contact during prenatal life and the inoculum transferred during birth and breastfeeding imprint the infant’s microbiota and the immune system [24, 28]. The gut microbiota of neonates is low in diversity and is dominated by two major phyla such as Proteobacteria and Actinobacteria, while in the weeks following birth, the microbiota becomes more diverse and dominated by Firmicutes and Bacteroidetes [11, 28–30]. Between one to two years of life, the microbial succession begins to vary, coinciding with the transition to a more diverse type of diet [29]. By the end of the second year of age, the taxonomic composition of the gut microbiome stabilizes and converges toward a characteristic adult microbiome [28]. Therefore, the first 2 years of life represents a vital period for any diet-based alteration to potentially impact the human health and immunity in later life.

4. Major Factors Influencing the Development of the Microbiome

Several pre and postnatal factors can significantly alter the composition of the gut microbiota including genetics; the mode of delivery at birth; the method of infant feeding; the use of medications, especially antibiotics; and diet, among others.

4.1. Birth Mode

Numerous studies have shown that the mode of delivery (vaginally or by C-section) has a strong influence on early gut colonization [24, 28]. Vaginally delivered infants are exposed to their mother’s vaginal and fecal microbiota and have an abundance of Lactobacillus and Bifidobacterium species [24]. Whereas, the very first microbial inoculum received by the infant born via a C-section is distinct; and these newborns become quickly colonized by the bacteria acquired from the skin and the surrounding hospital environment [24]. It is also suggested that the microbial composition in those infants may be disturbed for the first few months in life [31, 32]. Analysis of the meconium in the first few days after birth revealed that the microbial communities of the infant’s gut mirror either those of the mother’s vagina (Lactobacillus, Prevotella, or Sneathia) in the case of vaginal delivery, or the mother’s skin microbiota (Staphylococcus, Corynebacterium, and Propionibacterium) in the case of a C-section [24]. Vaginally delivered infants also share the higher proportion of their gut microbial 16S gene sequences with their own mothers as compared to other mothers [32].

4.2. Feeding Mode

The mode of feeding is another factor that influences the development of the microbiota in the infant’s gut. Not only is mother’s milk the major source of bacteria colonizing the infant’s gut, but it is also considered a major determinant and contributor to the child health status [33]. Analysis of breast milk samples revealed the presence of Streptococcus and Staphylococcus genus, and those are considered among the early colonizers of the infant’s gut [34]. Additional phyla such as Bifidobacterium and Lactobacillus were detected in the breast milk samples and also in the gut of breastfed infants in higher counts as compared to the formula-fed infants [35–37] indicating a major role of breast milk in seeding the gut microbiota early in life. The gut microbial diversity increases with the infant’s age and during the gradual shift from milk to solid food where a higher abundance of Bacteroides and Clostridium species in the toddler’s gut starts to be observed.

4.3. Antibiotics

The use of broad-spectrum antibiotics in infants and young children has negative effects on the composition of the gut microbiota, which results in significant drops in taxonomic richness, diversity and evenness of the gut microbial communities [38, 39]. Prenatal/or maternal antibiotic exposure is shown to increase the abundance of many pathogenic communities (pathobionts) such as

Staphylococcus, Streptococcus, Serratia, and Parabacteroides [40].

Early antibiotic exposure in neonates (such as Ampicillin and Gentamicin administered within the first 48 h of birth) results in significantly lower levels of Bifidobacterium and Lactobacillus when compared to untreated controls [41].

4.4. Preterm Birth

Compared to full term born infants, the preterm born infant’s microbiota has a reduced microbial diversity and a decreased abundance of bacteria such as Bifidobacterium and Bacteroides species [42]. In addition, potentially pathogenic bacteria such as Enterococcus species together with Proteobacteria species increase proportionally as a result of the reduction of the beneficial microbes [41]. Preterm birth is also associated with inflammatory intestinal disorders such as Neonatal Necrotizing Enterocolitis which has been linked to microbial dysbiosis and an increase in Proteobacteria is considered a predictive marker for the disease [43].

4.5. Host Genetics

The contribution of the host genetic makeup in shaping the early microbiota remains a subject of many speculations, as defining the host genotype–microbiota interaction in humans remains complicated. However specific bacteria found in the gut microbiota may be influenced in part, by the genetic makeup of the host, impacting as a result our metabolism and ultimately the health status.

Family studies with varying degrees of relatedness along with the twin cohorts’ studies provided a great input on the role of genetics in the microbiome composition. The microbial diversity is similar within family members as compared to unrelated individuals and monozygotic twins share certain the expression of certain heritable taxa compared to the dizygotic twins [44]. However, the overall variance in the microbiome composition is also affected for a few percent by host genetics and mostly determined by environmental factors [45]. A recent study in southern China, showed that the major influence on microbiome composition was due to the geographical localization in different districts and that these differences could simply be accounted by environmental or life style differences among districts [46]. Detailed genome-wide association studies are required to provide a greater insight into the role of the genetic makeup in the microbiome composition and in the mechanism underlying the microbiota–host genetic interactions.

4.6. Environmental Factors

Differences in lifestyle due to the geographical location, religion and culture are among the factors impacting the development of an early gut microbiota [47]. Infants from the Northern part of Europe have higher abundance of Bifidobacterium, Clostridium and Atopobium species in their gut microbiome, while infants living in southern Europe have higher percentages of Eubacteria, Lactobacillus and Bacteroides [36]. Similar differences in the gut microbial composition have been reported between German and Finnish infants [36], and between Swedish and Estonian infants [48]. Family dynamics may also act as a relevant environmental factor, where having siblings usually results in a higher percentage of Bifidobacterium as compared to single infants [49]. Exposure to household pets was shown to play a role in shaping the gut microbiome of infants [50]. An emerging spectrum of environmental factors including maternal or infant’s exposures to toxins/pollutants and xenobiotic substances are important to be considered and warrant more studies in order to determine their role in the development of the early gut microbiota.

5. Tools Used to Study the Microbiome

Recent advancements in the development of sequencing technologies and bioinformatics tools, fostered the growth of new omics technologies used to study the microbiome composition and its interaction with the host. Omics technology helps to explore the microbiome at various levels including gene expression, gene abundance, protein expression and microbial metabolites [51]. This knowledge is an essential tool used to study the role of the microbiome in many diseases and pathological conditions using various models (described later).

5.1. 16S rRNA Gene Sequencing

This is a gene-specific technique, where the bacterial taxonomic profile of a sample is characterized by sequencing specific hypervariable regions of the 16S rRNA gene [52]. Amplicons are generated using universal primers used to anneal with the conserved regions of the 16S rRNA gene sequences [43]. Those hypervariable region sequences are highly polymorphic, therefore allowing the classification of the bacterial taxa from phyla to species levels [52]. Illumina MiSeq and ThermoFisher Ion Torrent Personal genomic instruments are the widely used sequencers with a capacity to generate 300–400 bp reads [53]. Sequence reads can be analyzed using the traditional way of operation taxonomic units-based clustering approach using QIIME pipeline or calling amplicon sequence variants using DADA2 pipeline [54, 55]. The other available tools used to detect the taxonomic profiles using 16S data are LEA-seq, 16S based oligo typing, Long-read 16S, UNOISE2, and Deblur [56–59]. GreenGenes and SILVA are the most commonly used databases to profile the taxonomic differences among various samples [60, 61]. Eukaryotic organisms (such as fungi and protists) also play an important role in the microbiome. They can be analyzed in a similar manner to 16S studies by either looking at the hypervariable regions of the 18S rRNA or the internal transcribed spacer (ITS) regions between rRNA genes [62–74].

5.2. Metagenomics: Which Microbes Are There?

This technology is based on the shotgun sequencing of the whole genomic DNA and is used to study all the genetic materials in the microbiota [51]. Metagenomic study provides information about the taxonomic profile, functional composition and gene abundance of the microbiota [75]. It can also provide information up to species/strain level compared to 16S rRNA gene sequencing [75]. One challenge of metagenomic shotgun sequencing is that of retrieving enough data pertaining to the microbiota from low biomass samples. These low biomass samples are host tissues which present a low concentration of microbes relative to host somatic cells. Because shotgun sequencing is nontargeted, and the size of a mammalian genome is three orders of magnitude larger than that of a bacterial genome, the sequencing effort for a low biomass sample may end up being dominated by the host DNA.

Although metagenomics can explore the functional abundance (relative abundance of the microbial genes), it lacks information about the microbial activity (gene expression in live microbes) [75]. Therefore, recent studies are using metatranscriptomics (RNA-based) along with metagenomics (DNA-based) to explore the functionality of the active/viable microbes and their role in the host physiology [51]. Metagenomic sequencing reads can be taxonomically classified using, basically, two approaches, namely, the alignment of reads to reference bacterial genomes, or a comparison of snippets, or “words,” of genome of a fixed small size (known as k-mers) to the ones found in reference bacterial genomes. The HUMAnN pipeline [76] infers presence and relative abundance (quantity) of species by alignment to precomputed clade-specific genes, which are genes exclusive to a species or other higher-level taxonomic levels. Metabolic reconstruction using hits to reference are performed using ASGARD [77]. Classification of reads coming from species which are not in a reference database can be done by k-mer spectrum analysis using Kraken [78] and by assembling the reads into contiguous sequences, called contigs, followed by naive gene prediction and classification using a functional database like InterPro [79].

5.3. Meta transcriptomics: What Are Those Microbes Doing?

The advent development in the field of RNA-sequencing has provided more exciting opportunities in the field of transcriptomics, especially microbial transcriptomics. Metatranscriptomics reflect the functional activity of microbial communities by assessing the community-wide gene expression [80]. Sample processing for metatranscriptomics is laborious compared to the metagenomics library preparation. After total RNA extraction, bacterial rRNA is removed using specific magnetic probes of rRNAs. The prokaryotic RNA content is only 1–5% of the total RNA content and is enriched with mRNA that is converted into cDNA before proceeding with the RNA sequencing protocol [81]. Sequence reads are mapped to reference genomes, and the expressed genes are identified based on the sequence reads covering these regions. Raw Sequence data is filtered using Trimmomatic and mapped using BOWTIE and GEM [82]. The differentially expressed genes are identified using CuffDiff [83]. HUMAnN and MG-RAST are the most extensively used pipelines for metatranscriptomics analysis [82, 84–86].

This approach can give a panorama for assessing the microbial activity of the human microbiome, the antisense RNAs and small noncoding RNAs, and facilitates a deeper understanding of the host–microbiome interactions in abnormal and diseased states [81].

5.4. Metabolomics

Defined as the analysis of the metabolomic profiles using untargeted and targeted liquid chromatography–mass spectrometry (LC-MS) methods or nuclear magnetic resonance spectroscopy (NMR); metabolomics is used to explain the mechanism of microbial bioactivity [53]. Of all the previously described omics approaches, the metabolomics approach acts as a bridging gap between the microbiome and host physiology. Metabolomic profiles can be analyzed with biofluids such as plasma, serum, urine, tissue extracts, and fecal supernatant [87, 88]. The output of the metabolomics data allows correlations between microbiota, their metabolites and metabolic routes using multivariate analysis. It can also be used as a diagnostic and prognostic marker for the diagnosis and treatment of various diseases.

6. Microbiota Derived-Metabolites and Small Molecules

After completion of the HMP [9], Cimermancic et al. used an advanced computational approach with a moving aim from “who’s there?” to “what are they doing?” and identified more than 14,000 biosynthetic gene clusters in the human microbiome [89]. Of which, 3118 genes were present in one or more of the 752 sequenced samples [9]. While microbes are well known to produce diverse natural molecules mainly enzymes and antimicrobial agents, the recent genome mining have discovered numerous novel unexpected biosynthetic gene clusters and distinct natural molecules, which contribute to the microbial social and competitive interaction with other organisms [89–92].

The microbiota produces a range of molecules from various chemical classes including low molecular weight metabolites, to high molecular weight ribosomally synthesized posttranslationally modified peptides and proteins (RiPP), short-chain fatty acids (SCFAs) and diverse lipopolysaccharide (LPS) [92]. SCFAs contain acetate (C2), propionate (C3), and butyrate (C4) and are mainly produced by Butyrivibrio, Clostridium, Faecalibacterium, and Bifidobacterium species [93]. Vitamins like vitamin B and vitamin K are produced by various gut microbes including Propionibacterium freudenreichii, Salmonella enterica, Listeria innocua, and Enterobacter species, among others [94–96].

Those metabolites possess a wide range of activities; for example, microcin a narrow-spectrum antibacterial RiPP produced by Escherichia coli inhibits the growth of other closely related microbes; while the LPS and glycolipids released by the gut microbiota target various host immune cells; and the SCFAs produced by Bacteroidetes and Firmicutes stimulate a range of host processes such as epithelial cell proliferation, energy metabolism, host immune system (discussed later). We have summarized some of the natural molecules or metabolites produced by different microbiota species that mediate microbe-microbe interaction and host–microbe interactions in Table 1.

Table 1.

Selected microbial associated metabolites from the human microbiota

Metabolite	Related bacteria	Body site	Potential functions
Antimicrobial
Ruminococcin A, microcin, clostridiolysin S, zwittermicin, cytolysin, epidermin	Ruminococcus gnavus, Escherichia coli, Clostridium sporogenes, Bacillus cereus∗, Enterococcus faecalis, Staphylococcus epidermidis, Streptococcus pyogenes	Gut	Inhibit the growth of pathogens, promote commensal colonization in gut
Lactocillin	Lactobacillus gasseri	Vagina	Antibacterial peptide, inhibit the growth other pathogens
Epidermin	Staphylococcus epidermidis	Skin	Antimicrobial peptide
SA-FF22	Streptococcus pyogenes	Skin/oral	Antimicrobial peptide
Streptolysin
Tilivalline	Klebsiella oxytoca	Gut	Peptide, cytotoxic
Cereulide	Bacillus cereus	Gut	Cytotoxic
Staphyloxanthin	Staphylococcus aureus	Skin	Virulence factor
Mycolactone	Mycobacterium ulcerans	Skin	Polyketide, cytotoxic
Immunomodulatory
α-Galactosylceramide	Bacteroides fragilis	Gut	Essential for myelin synthesis, Stimulate host immune systems
Short-chain fatty acids: propionic acid, acetate, butyrate, indolepropionic acid	Bacteroides spp., Clostridium spp., Eubacterium spp., Butyrivibrio, Faecalibacterium spp., Clostridium sporogenes	Gut	Modulate host immune system, Energy to epithelial cells
Vitamins: Vitamin K, vitamin B12, biotin, folate, riboflavin, etc.	Bifidobacterium spp., Propionibacterium freudenreichii, Salmonella enterica, Listeria innocua, Enterobacter spp.	Gut	Sources of vitamins, stimulate immune system, Exert epigenetic effects
Lipids: LPS, peptidoglycan, acylglycerols, triglycerides, Polysaccharide A	Bifidobacterium spp. Roseburia, Lactobacillus, Clostridium spp., Bacillus fragilis	Gut	Induce inflammation, Enhance immune system, influence gut permeability
Polyamines: Putrescine, cadaverine, spermine	Campylobacter jejuni, Clostridium saccharolyticum	Gut	Modulate immune system, anti-inflammatory and antitumoral effects
Bile acids: Cholate, muricholate, etc.	Lactobacillus spp., Bifidobacterium spp., Clostridium spp., Bacteroides	Gut	Influence intestinal permeability
Muramyl dipeptides and tripeptides	Fusobacterium nucleatum	Oral	Modulate immune system
Siderophore
Corynebactin	Corynebacterium spp.	Skin	Peptide, Siderophore
Staphyloferrin B	Staphylococcus aureus	Skin	Peptide, Siderophore
Neurotransmitter
D-lactate Ammonia Acetyl-choline, Gamma-amino Butyrate, acetylcholine, Norepinephrine, serotonin, Dopamine	Lactobacillus spp., Enterobacter spp., Bacillus spp.	Gut	Influence blood-brain barrier, Neurotoxic
	Lactobacillus spp., Bifidobacterium spp. Escherichia spp., Streptococcus spp., Bacillus spp.	Gut	Influence human hormones, Neurotransmitter, Influence human behaviors
Unknown functions
Skatole, p-cresol, phenyllactic acid, listeriolysin S, clostridiolysin S, δ-aminovaleric acid, β-aminoisobutyric acid	Clostridium spp., Bifidobacterium spp., Listeria monocytogenes, C. sporogenes, Clostridium spp.	Gut	Unknown
Mutanobactin, coproporphyrin III, corynomycolic acid	Streptococcus mutans, Propionibacterium acnes, Corynebacterium spp.	Mouth/skin	Unknown

7. Microbiome–Host Interactions: A Bidirectional Relationship

The recent developments of the high-throughput technologies and availability of the germ-free mice or humanized mice models have shed light into the interaction of the microbiota and/or its mediators with the immune system, and contributed to a better understanding of the molecular mechanisms governing these interactions [97–100]. These models have provided important insights into how the microbiota influence the host immune signaling in mammalian cells [101–103]. Most studies focused on the interaction between the gut microbiota and the immune system, since the gut is the most studied ecosystem. These studies showed that the microbiota–immune cell interaction begins at birth, where the microbiota educates the immune system, and in turn the immune system shapes the composition of the gut microbiota [104]. This interaction can be local or systemic.

7.1. Local Interactions

A great example of the local interaction of the microbiome with the immune system is represented in the gut (Fig. 1). The intestinal mucosa is protected by an array of physical, biochemical barriers and immunological elements. The physical barrier is made of the mucus layer consisting of a glycosylated gel composed of mucin molecules that are secreted by Goblet cells, the intestinal epithelial cells (IECs) connected by tight junctional complexes for protection against pathogen translocation. The microbiota is considered a part of the physical intestinal mucosal barrier as well [105]. The biochemical barriers are made of bile and gastric acid, defensins, lectins, and lysozymes secreted by Paneth cells [106]. The immunological elements include the antimicrobial peptides which are mainly produced by Paneth cells and act against a wide array of invading pathogens [107]. In addition, the secretory IgA is produced by the plasma cells in the lamina propria and is transported to the lumen, where it blocks invading pathogens by preventing their attachment to the epithelial cell receptors [108]. All immune cells residing in the lamina propria such as macrophages, neutrophils, dendritic cells (DCs), IECs, T and B lymphocytes, T-regulatory cells (T_Regs) and plasma cells orchestrate an immune protection against invading pathogens and an immune tolerance for the microbiota (Fig. 1) [108–111].

Fig. 1.

Integrative view of the gut microbiome. MAMPs microbial associated molecular patterns, PAMPs pathogens associated molecular patterns, SCFAs short chain fatty acids, DC dendritic cells, Il interleukin, Treg regulatory T-cells, Th T-helper

Pathogenic microbes and the microbiota interact with the immune cells through similar conserved ligands that signal through various host pattern recognition receptors (PRRs) such as the Toll-like receptors family (TLR) [112–114]. Those receptors recognize various microbe-associated molecular patterns (MAMPs) including lipopolysaccharides, peptidoglycans, flagellin, and formylated peptides [115, 116]. Innate immune cells have TLRs that recognize MAMPs leading to the activation of Nuclear factor-Kappa B (NFκB) which triggers the cytokine production [117]. Moreover, the upregulation of these cytokines against the PAMPs [112] leads to T-cells activation, a function that is facilitated by the gut microbiota that acts as an intermediate connector between the innate and the adaptive immunity (Fig. 1) [112, 113].

7.2. Systemic Interactions

Beneficial microbes such as Lactobacillus and Bifidobacterium promotes the upregulation of potential transmembrane proteins to protect the tight junctions of IECs and reduces gut permeability [118, 119]. Impairment in the barrier leads to an increase in the gut permeability and promotes translocation of bacteria through the intestinal barriers (Fig. 1) [120]. Continuous exposure of bacteria and bacterial LPS to the mucosal immune cells leads to the secretion of cytokines as a primary local immune response which in turn activates the vagus nerve, the spinal afferent neurons [121], and as a result the central nervous system (CNS) [122, 123]. Gut microbiota indirectly affects the CNS by releasing their metabolites in the form of neurotransmitters, for example, Lactobacillus reuteri and Bifidobacterium infantis produces histamine and serotonin, respectively [124]. Serotonin synthesis is mainly dependent on the availability of tryptophan, an essential amino acid acquired through diet [125]. In patients with depression, a reduction in the levels of tryptophan was observed [126]. Increased levels of tryptophan have been observed as a result of the oral administration of B. infantis in rats [127], but whether similar effect will be observed in patients with depression is yet to be described.

Dietary tryptophan, an essential amino acid is metabolized by the gut microbiota in order to produce natural ligands including indole-3-aldehyde and indole-3-propionic acid [128]. Through these metabolites, the gut microbiota is able to regulate the activity of the astrocytes (microglial cells of brain) by activating the astrocyte aryl hydrocarbon receptors (AHR) [129].

Moreover, the dietary tryptophan and nicotinamide are involved in regulating the mammalian target of rapamycin pathway (mTOR), which is involved in cell proliferation, survival, protein synthesis and transcription among other functions. In an experimental mouse model with Angiotensin Converting Enzyme 2 (ACE-2) mutation, reduced activity of the mTOR pathway is observed in epithelial cells which results in increase severity of colitis [130]. However, when the mice diet is supplemented with tryptophan, the colitis symptoms are reduced [130].

8. Mediators of the Microbiome–Host Interactions

The continuous cross talk between the microbiota and our immune system is mediated by a wide array of signaling mechanisms that involve many classes of microbial products [131] including microbiota-derived nucleic acids, metabolites, and small molecules:

8.1. Microbiota-Derived Nucleic Acids and the Immune System

The microbial DNA is structurally diverse from the human DNA and shows a remarkable sequence diversity between microbial species [132]. Microbial DNA utilizes its unique properties such as the higher frequency of CpG or CG dinucleotides and the presence of unmethylated cytosine as compared to being methylated as in the mammalian DNA [133–135]. The unmethylated CpG motifs in the microbial DNA are sensed by TLR9, which is expressed in IECs, plasmacytoid dendritic cells (PDCs) and B cells in humans [136]. IECs express TLR9 on both the apical and basolateral sides, and distinct responses are triggered depending on which side the stimulus is coming from (Lee et al. 2006). Stimulation of TLR9 on the apical side of the IECs is known to inhibit their inflammatory response and instead it induces tolerance toward other MAMPs [137]. However, the basolateral stimulation, leads to the induction of cell signaling pathways including the mitogen activated protein kinases (MAPKs), NFκB and the production of proinflammatory cytokines, interferons, and chemokines [138, 139]. In general, the apical stimulation of TLR9 is known to improve the barrier functions of the gut mucosa [140].

8.2. Microbiome Associated Metabolites-Host Interactions

Inside the GI tract, the gut microbiota facilitates digestion, fermentation of foods and produces a number of gastrointestinal metabolites, while the IECs orchestrate the intestinal defense mechanisms and mediate the interaction between the microbial mediators and our immune system [141]. A key example of those microbial metabolites includes the SCFAs, vitamins, neurotransmitters, microbial cell components such as the cell envelope or capsular components which can also act as microbial mediators to stimulate the immune functions as discussed in the previous section [128, 142–144]. Taken together >50% of the fecal and urinary metabolites are either derived from or modified by the gut microbiome [145] aiming to modulate the host cellular functions [146–148].

Microbial metabolites stimulate the host immune response through several ways [92, 149, 150], (a) by providing energy to the intestinal cells, (b) by manipulating the immune cells transcriptional programming, (c) by driving immune modulation through unknown mechanism.

Mounting evidences indicate that the SCFAs serve as a key energy source for the IECs and can influence their metabolism and integrity [149]. It was shown that, in germ-free mice, in absence of butyrate, the IECs suffer an altered energy metabolism which leads to autophagy activation due to the nutrient starvation in those cells [149, 151]. However, the monoinoculation of the germ-free mice with butyrate producing bacteria, was shown to rescue those cells from autophagy confirming the important role of butyrate in producing energy in the IECs [149, 151].

In tandem of being a direct energy source, the microbial metabolites may also tune the transcriptional programming of immune cells through the epigenetic landscape in order to control the immune function and the gut microbial composition [149]. One example is the butyrate concentration in the gut which can influence the transcription of mucin-related genes in the goblet cells (Fig. 1) and in turn modulate the transcription of many proinflammatory mediators such as IL-6, NOS2, IL-12 [152, 153]. The gut microbial composition can also be influenced by other microbial metabolites through the activation of the Nlrp6 inflammasome signaling pathway resulting in the secretion of proinflammatory cytokine, IL-18 and synthesis of some antimicrobial peptides [154, 155]. These mediators prevent the growth of the invading pathogens and maintain the gut tolerance for the commensal microbes.

Part of the protective effect of the microbial metabolites is mediated by their capacity to modulate directly or indirectly the function and the regulatory network of the host immune cells [150]. For example, the recognition of microbial derived butyrate and nicotinic acid (NA) by the G protein-coupled receptor GPR109A, expressed on the epithelial cells, macrophages, and dendritic cells, is critical for development of T_Reg cells [156]. Through the GPR109A receptor, both butyrate and NA can trigger IL-18 and IL-10 secretion, which leads to the development of T_Reg from naïve CD4+ cells and fine-tune the functionality of the epithelial cells to provide protection against pathogens [156, 157]. On other hand, altered gut microbiota results in a weakened immune cell response against infection, deregulated metabolism and compromised immune homeostasis [158]. The interaction of microbial mediators with the immune cells are just beginning to be revealed, with different microbial metabolites being sensed through different receptors in order to modulate cell programming, metabolic pathways, epigenetic modification [155]; however, the mode of action remains to be defined.

9. Models Developed to Study the Microbiome–Host Interactions

As discussed previously, the host and the microbiome are in a mutualistic relationship [159]. To study this interaction, many models were developed and used, most of them focus on modeling the gut environment as it is the most studied body site [160–162]. Some of these models are described below:

9.1. In Vitro Models

The human gut environment plays a vital role in the metabolism of nutrients directly and indirectly through the gut microbiota [163]. In vitro models of multicellular cell lines can be cocultured with lymphocytes and monocytes to mimic the gut ecosystem [161]. Developing those cell-culture models (described below), in combination with a microbial consortium can help fill the gap in understanding the gut–microbial interactions.

9.1.1. Caco-2 Cell Line

It was established in 1970 from gastrointestinal tumors, widely used as an intestinal epithelial barrier model in microbiological, pharmacological and nutritional fields [164]. Caco-2 is mainly used as a model to study the interaction of commensal or pathogenic bacteria and their metabolites with the cells [164]. This model was used as a three-stage culture system, to assess adherence of the gut microbiota to the epithelial cells and measure cytokine expression [165]. This study revealed that Atopobium species as the predominantly adherent organism, and Bifidobacterium longum showed a significant modulation of the expression levels of IL-4, IL-10, TGF-β2, IL-1α, and tumor necrosis factor alpha (TNF-α) [165]. Caco-2 cells were also used to show a positive modulation of the gut microbiota after using prebiotic oligosaccharides that result in proinflammatory cytokines reduction [166].

9.1.2. HT-29 Cell Line

HT-29 cell line was isolated in 1964 from a human colon adenocarcinoma by Fogh et al., [167]. These enterocytes-like cells have the capability to produce cytokines such as interleukins and TNF-α, which makes it an important model to study the host–microbiome interactions [168]. A coculture model of Caco-2/HT29 cells was used to assess the immune functions and measure the improvement in the gut barrier using prebiotic fibers [169]. Caco-2 and HT-29 cells were also used to study the role of a bacterial metabolite, butyrate, in activating the activator protein-1 (AP-1) which is a potential target for cancer therapy [170]. In another study, HT-29 cells were used as a colon cancer model to study the metastatic therapeutic effect of nisin, a bacteriocin derived from Lactobacillus lactis [171].

9.1.3. IPEC-1 and IPEC-J2

IPEC cell lines (IPEC-1 and IPEC-J2) are porcine IECs isolated from the jejunum of a unsuckled neonatal piglet [172]. IPEC-1 cells were isolated from the ileal and jejunal tissue while the IPEC-J2 cells were isolated from the jejunum [172]. It is a unique and nontumorigenic cell line that can mimic the human physiology. It can be used as a supreme to study epithelial transport, interactions with the enteric bacteria, and to assess the effects of probiotic microorganisms [173]. IPEC-J2 was used to identify the role of butyrate in the epithelial cell integrity against LPS-induced damage [174]. This study confirmed that butyrate restored the LPS induced impairment in gut barrier via activation of Akt-mediated protein synthesis pathway by promoting claudin expression [174]. In another study, IPEC-J2 was used as a model to study the protective effect of Lactobacilli against enterotoxigenic E. coli (ETEC) which destruct the gut barrier. L. reuteri showed a substantial protection against ETEC by conserving tight junction structures of IPEC-J2 [175].

9.1.4. IEC-6 and IEC-18

IEC-6 cells were isolated from the rat small intestine and are considered a good model to study bacterial adhesion [176]. Adhesion of multiple Lactobacillus strains was tested using Caco-2 and IEC-6 cells and results showed that L. plantarum shows a higher ability of adherence [162]. In another study, L. reuteri showed a protective effect against LPS-induced inflammation [177]. Butyrate induced a higher expression of the Heat shock protein 25 in those cells reflecting its protective effect [178]. Moreover, the LPS derived from enteropathogenic E. coli and four types of Bifidobacteria species are shown to induce autophagy in those cells [179], which renders this model very useful to study the probiotic effect of the gut microbiota in various diseases.

9.2. Ex Vivo Models

Ex vivo means out of live beings, which can represent as a model between in vitro and in vivo [180] using human or animal cells, tissue, or organs to evaluate the host–microbiota interactions.

9.2.1. InTESTine™ System by TNO

TIMs (TNO intestinal Model) are model systems used to mimic the gut, TIM-1 simulates the stomach, Tiny-TIM simulates the small intestines and TIM-2 model simulates the physiological conditions of the colon [181]. TNO recently developed InTESTine™, a new in vitro intestinal model using healthy porcine intestinal tissue. It contains a mucus layer which enables to utilize in experiments involving interaction of single microbe or a microbial consortium. Till date, there is no publications using this system, but still it provides a novel way to study the host–microbiome interactions.

9.2.2. Ussing Chamber

An Ussing chamber is an apparatus to measure epithelial membrane properties. It is widely used to quantify transport and barrier functions in a variety of living tissues. The main advantage of this chamber is the ability to study different segments of the intestines which makes it a powerful ex vivo model [168]. It contains electrodes to measure the voltage and short-circuit current in order to determine both cell permeability and ions transport. This system is mainly used to study host–microbiota interaction via microbial toxins and microbiota [168]. The effect of Enterococcus faecalis on a colonic mucus mounted on Ussing chamber was studied to understand the role of bacterial invasion; the study revealed that an adhesion product of E. faecalis promotes its translocation in the colon mucus [182]. The probiotic role of Lactobacillus plantarum has also been studied using this model [183].

9.2.3. Stem Cell-Based Organoids

Intestinal organoids are cultured from adult intestinal stem cells or induced pluripotent stem cells, which represent the new sophisticated gut model [184, 185]. Directed differentiation methods allow the cell-specific responses, assessing the intestinal barrier functions and the host–microbe interactions [186]. Human enteroids have been widely used to study host–pathogen interactions including enterohemorrhagic E. coli, enterotoxigenic E. coli, enteroaggregative E. coli, and Cholera toxin [187–189] as well as host interactions with L. rhamnosus GG, L. acidophilus, and Bacteroides thetaiothaomicron [190–192].

The apical to basolateral polarity conserved even after exposure to the microbes and their components is considered the main advantage of the ex vivo models; however, the short-term survival is the main disadvantage of these models.

9.3. In Vivo Models

Animal model-based research provide opportunities to manipulate, engineer, and study the host–microbiome interplay with a certain level of experimental control which is not feasible in human recruited experiments [160]. Below are few models used for microbiome studies:

9.3.1. Fruit Fly

Fruit fly, Drosophila melanogaster has been widely used as a model system to study the host-innate immunity and microbial pathogenesis [117, 193]. Around 5–30 different bacterial species are found in the fruit fly and the predominant microbial members of the Drosophila gut are aerotolerant Acetobacteraceae, Lactobacillales, and Enterobacteriaceae families [194]. Due to their similarity to the mammalian gut and our ability to culture these bacteria in the laboratory, the availability of various mutants, its affordable price renders Drosophila is considered a very useful model to study host–microbe mechanisms associated with innate immunity modulation, probiotic role and diet [195–197].

9.3.2. Zebrafish

Zebrafish is a small freshwater fish mainly used as a vertebrate model that is simple to study host–microbe interactions. It has a diverse microbiota with gamma proteobacteria and fusobacteria species being the most abundant members of the zebrafish’s gut [198]. A study between conventionally raised zebrafish and germ-free zebrafish indicated several new insights into the host–microbe interactions related mechanisms [199].

9.3.3. Mice

Laboratory mouse has been widely used in understanding the role of gut microbiota in host physiology, such as angiogenesis, metabolism of nutrients, cognitive health, hepatic and immune functions [92, 200–203]. Almost 99% of mouse genetics are shared with human host and the gut microbiome members from phylum to family levels [204]. These characteristics made lab mouse as prime choice of animal model to evaluate host–microbe interactions.

The most commonly used inbred mice strain is C57BL/6.129, C3H and BALB/c are the other inbred strains used in microbiome studies. The other concept of mice model is to use mouse with less complex predefined microbial consortia. The germ-free mice colonized with a defined microbial cocktail (isolated from conventional mice) to produce mice with standardized gut microbiota are called gnotobiotic mice models, which can serve as an ideal tool in host–microbe interactions [205]. This defined cocktail consists of eight aerobic bacteria, were E. coli, var. mutabilis, Streptococcus faecalis, Lactobacillus acidophilus, Lactobacillus salivarius, group N Streptococcus, Bacteroides distasonis, a Clostridium sp., and a fusiform bacterium; anaerobes were excluded due to difficulties in cultivation [206].

10. Microbial Dysbiosis and Immune Dysregulation

A healthy individual maintains a unique balance between the microbiome, immune system for protection from invading bacteria [207]. However, in some cases pathogenic bacteria increases in numbers replacing the beneficial bacteria (microbial dysbiosis) leading to a significant impact on our health [208].

The cross talk between the host microbiota and our immune system begins at birth [24, 28]. In order to achieve homeostasis, the intestinal microbiota and our mucosal immunity interact constantly as summarized in Fig. 1.

Various chronic immune-mediated and metabolic diseases such as inflammatory bowel disease, celiac disease, multiple sclerosis, rheumatoid arthritis, asthma, diabetes, and certain types of cancers are associated with dysbiosis in the gut [207–209]. The availability of high-throughput techniques has significantly advanced our capacity to sequence the microbiome and demonstrated variable degrees of dysbiosis in the above diseases [207–209]. For example, subjects with colorectal cancer (CRC) suffer from microbial dysbiosis when compared to normal controls [210, 211]. CRC adenoma is characterized by a high proportion of pathogenic bacteria, such as Pseudomonas, Helicobacter, and Acinetobacter and relatively lower abundance of beneficial bacteria, such as butyrate-producing bacteria [212]. As the disease advances from adenoma to carcinoma, a significant increase in Bacteroides massiliensis, Bacteroides ovatus, Bacteroides vulgatus, and E. coli was observed [213]. The promotion of inflammation due to bacterial dysbiosis partially explains the induction of tumorigenesis [214, 215].

11. Microbiome and Cancer

The microbiome composition modulates metabolism, inflammation, immunity and other functions of the organism. It also affects cancer initiation, progression and response to therapy [216]. Genetically transformed cells require a favorable microenvironment to form tumors. The microbiome modulates the tumor microenvironment and tumor progression both locally on the epithelial barriers in which it resides and systemically.

Based on epidemiological studies, a role of the microbiome and dysbiosis (i.e., a change in microbiome composition associated to disease) has been proposed for gastrointestinal and lung cancer. The only bacterial species recognized by the International Agency for Research on Cancer (IARC) as a group 1 human carcinogen is Helicobacter pylori that is associated to stomach cancer [217]. Other bacterial species were observed within the neoplastic lesions or positively correlated with cancer risk in epidemiologic studies. The association of bacteria with cancer may not be an evidence of their direct role in carcinogenesis but rather be a biomarker for coregulated or keystone bacterial species modulating the microbiome composition. Also, the presence of changes in the microbiome composition in the cancer patients may be induced by the presence of the tumor rather than being a cause of it. None of the bacteria associated with human cancer but H. pylori has been identified with certainty as a human carcinogen by showing that their elimination from the host prevent disease [218]. However, mouse and human studies suggest that several bacterial species (e.g., Fusobacterium nucleatum, enteropathogenic Escherichia coli, enterotoxigenic Bacteroides fragilis, Enterococcus faecalis, Streptococcus gallolyticus, Helicobacter hepaticus, and Salmonella enterica) may play a role in carcinogenesis [214, 215, 218]. The microbiome modulates locally inflammation and immunity at all epithelial barrier surfaces, where commensals and host cells are in proximity [219]. In addition, various bacterial products including toxins and metabolites, are genotoxic or target proliferation and repair of epithelial cells [216].

11.1. Colorectal Cancer

Various species of Fusobacterium (F. nucleatum), orally associated anaerobic gram-negative bacteria that form biofilms in the mouth, can colonize the colon and are enriched in human colonic adenomas and adenocarcinomas [214, 215]. Various mechanisms have been proposed for the procarcinogenic effect of F. nucleatum: (1) recruitment of tumor-promoting myeloid cells; (2) modulation of autophagy resulting in chemoresistance; (3) binding of Fap2 protein to the TIGIT inhibitory receptor resulting in inhibition of NK and T cells activity; and (4) association of FadA adhesin to E-cadherin, thus activating β-catenin/Wnt signaling in epithelial cells [214, 215, 220, 221]. Bacterial biofilms, in some patients containing Fusobacteria, are often present in carcinomas of the ascending colon and produce the polyamine metabolite N1,N12-diacetylspermine that favor epithelial proliferation, diminish E-cadherin expression and activate STAT3 and IL-6 [222, 223]. Fusobacteria and their associated microbiome have also been observed in liver metastases of colon carcinoma patients and remain associated to the tumors derived from these metastases when they are passaged several times as xenographs in immune deficient mice [224]. Elimination of cancer-associated Fusobacteria by antibiotics treatment reduces tumor xenograph growth [224].

E. coli is also over-represented in the gut microbiome of some patients with colitis and colon cancer [225]. The colibactin toxin from E. coli strains expressing the pks pathogenicity island impedes SUMOylation of p53 and favors cellular senescence associated to the secretion of tumor promoting proinflammatory and growth factors [225]. Enterotoxigenic Bacteroides fragilis (ETBF), is a low abundance commensal of the human intestine that when overexpressed becomes a pathobiont causing diarrhea and associated with inflammatory bowel disease and cancer [218]. ETBT toxin degrades E-cadherin and activates β-catenin enhancing proliferation of epithelial cells. It also permeabilizes the mucosa and induces STAT3 phosphorylation in both epithelial cells and inflammatory cells. In animals it induces by expansion of both Tregs and Th17 cells and favors colon carcinogenesis [218]. Whether the microbiome affects cancer on other epithelial surfaces including the skin, the oropharyngeal cavity, the lungs, and the urogenital tract remains to be determined. In addition to bacteria and viruses, the presence of fungi producing carcinogenic acetaldehyde has been associated the oral and esophageal cancer observed in patients with autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED), a syndrome due to mutations in the AIRE gene [226]. Other members of our microbiome such as protists and helminths are also able to orchestrate gut immunity and could play a role in gastrointestinal cancer [227].

11.2. Tumors in Tissues Not Colonized by the Microbiome

The microbiome may also affect the development of tumors in sterile tissues. For example, certain bacteria modulate estrogen metabolism via β-glucuronidases and β-glucuronides and regulate endometrial and breast cancer through a noninflammatory pathway [228]. Microbial β-glucuronidases are also involved in the metabolism of xenobiotics and can produce genotoxic and procarcinogenic metabolites of heterocyclic aromatic amines from burned meat or environmental pollutants [229]. A systemic effect of the microbiome on carcinogenesis has been shown in mouse models. For example, Helicobacter hepaticus colonic infection results in an increased incidence of prostate and mammary carcinoma in APC^min/+/Rag2^−/− mice and also promotes chemical and viral liver carcinogenesis [230, 231]. The variable incidence of tumors in genetic models of carcinogenesis observed has been attributed to diversity of microbiome composition in different animal facilities (e.g., thymic lymphoma in mice genetically deficient for the ataxia telangiectasia mutated (Atm) gene [232]. Recently, the presence of a microbiome within solid tumors that originate or reside in sterile tissues, such as pancreatic cancer, lung cancer, and colorectal carcinoma distant metastases has been demonstrated [224, 233–236]. The presence of proteobacteria and other bacterial phyla such as Bacteroidetes and Firmicutes was reported for human and murine pancreatic cancer where they can affect the response to chemotherapy and the local immunological responses within the tumors by suppressing innate and adaptive immune response and preventing effective response to immune checkpoint inhibitor therapy [233–235]. Variovorax and Acidovorax species were found to be increased in lung carcinoma lesions and in particular that Acidovorax is preferentially associated with p53 mutated squamous cell carcinoma of the lung [236].

11.3. Bacteria as Cancer Drugs

Bacteria can also be used as cancer drugs. The New York surgeon William Coley in the late XIX century, based on clinical observations that acute infections are in some case associated with tumor regression, successfully treated patients with soft tissue sarcoma patients with “Coley’s toxins,” a preparation of Streptococcus pyogenes and gram-negative Bacillus prodigiosus (Serratia marcescens) [237]. Other bacteria have then been used for cancer treatment (i.e., Corynebacterium parvum and the streptococcal preparation OK-432); bacille Calmette–Guérin, an attenuated Mycobacterium bovis strain is still used for local therapy of superficial bladder carcinoma [238]. Salmonella, Escherichia, and Clostridium species accumulate in tumors when delivered systemically and are being developed as anticancer drugs [239]. The hypoxic and necrotic areas present in large tumors are poorly accessible to drugs and also resistant to radiation induced DNA damage [239]. When spores of obligate anaerobic bacteria such as Clostridium spp. are administered, they germinate and deliver antitumor activity only when they reach the hypoxic tumor tissues [240, 241]. Conversely, well oxygenated smaller tumors and metastases are more susceptible to the antitumor effect of facultative anaerobes (i.e., Salmonella and Escherichia genera) [242]. Ideal anticancer bacterial preparations need to be selectively toxic for tumor cells, able to reach and proliferate in tumor area not well accessed by traditional therapies, maintained under control with a sufficient latency by the immune response, and genetically modifiable to engineer their tropism [239]. Bacteria-derived vectors can also be used to deliver cytokines, antigens, toxins, antibodies and antitumor genes [239].

12. Microbiome Impact on Cancer Drug Metabolism and Efficacy of Treatment

The microbiome regulate the metabolism and pharmacokinetics of certain enteral or parenteral anticancer drugs [243]. Intestinal host and bacterial enzymes regulate absorption and bioavailability of certain oral drugs [244]. Gut microbial enzymes induce drug bio-transformation prevalently by catalyzing reduction and hydrolysis [245]. The drug metabolism can also be indirectly affected by the ability of drugs and other xenobiotics to modify the composition and gene expression of the gut microbiome [246]. Although the gut microbiome is known to be involved in the metabolism of more than 40 drugs, this effect has been described only for a few anticancer drugs: (1) nitroreduction of the radiation sensitizer misonidazole; (2) hydrolysis of the antimetabolite methotrexate; and (3) deconjugation of the liver detoxified form of the topoisomerase I inhibitor irinotecan (CPT-11) [247].

The gut microbiome modulates in the intestine and liver the expression of genes encoding molecules involved in xenobiotic metabolism and sensing [248, 249]. Germ-free are characterized by enhanced xenobiotic metabolism than conventional mice [248, 250]. Intestinal microbiome diversity may explain the heterogeneous response of patients to drugs and their toxicity [251].

12.1. Microbiome and Chemotherapy

The pharmacokinetics, efficacy, and toxicity of chemotherapy drugs can be modulated by the microbiota [252]. In mice, tumor associated γ-proteobacteria produce cytidine deaminase and degrade the drug gemcitabine [233]. The antibiotic ciprofloxacin reestablished sensitivity to gemcitabine in mice with a colon carcinoma tumor experimentally colonized with E. coli [233]. Proteobacteria have been described to be present in many human pancreatic ductal adenocarcinoma [233].

The chemical structure and activity of chemotherapy drugs can also be modified by direct interaction with bacteria. Thirty different drugs were incubated in vitro with either nonpathogenic gram-negative Escherichia coli or gram-positive Listeria welshimeri: the activity of ten of them was decreased whereas the activity of six other was improved [253].

Oxaliplatin and cisplatin have severely reduced antitumor activity when used in germ-free mice or antibiotics-treated mice [254]. The gut microbiome is endowing the antitumor activity of these platinum compounds by licensing myeloid cells in the tumor to produce, reactive oxygen species (ROS) needed for drug-induced DNA damage [254]. Genetic deficiency for the Cybb gene encoding the gp91phox chain of NADPH oxidase two or treatment with either the ROS scavenger N-acetyl cysteine or antibodies depleting myeloid cells make the mice resistant to the antitumor effect of oxaliplatin [254]. Cisplatin antitumor activity is restored by the administration of Lactobacillus acidophilus to antibiotic-treated mice [255]. Acetovanillone, an inhibitor of NADPH oxidases, protects mice from nephrotoxicity induced by cisplatin, suggesting that similar mechanisms may be involved both in the therapeutic and toxic effect of cisplatin [256].

The alkylating chemotherapy drug cyclophosphamide (CTX) increases intestinal mucosa permeability in mice allowing gram-positive bacteria (e.g., L. johnsonii, L. murinus, and Enterococcus hirae) to translocate to the mesenteric lymph nodes [257]. The translocated bacteria potentiate the antitumor immune response evoked by the immunogenic tumor cells killed by CTX [257]. Thus, the antitumor effect of CTX is dramatically reduced in germ-free mice and in antibiotics-treated mice [257].

12.2. Microbiome and Immunotherapy

Cancer immunotherapy has represented a major improvement in cancer treatment in the last few years. Many patients, however, fail to respond and not all type of tumors respond to immunotherapy. Thus, there is an urgent medical need to increase the proportion of cancer patients responding to immunotherapy. Recent data in animals and in clinical trial claim that the composition of gut microbiome maybe an important variable determining immunotherapy efficacy [254, 258, 259] and raise the intriguing prospect that targeting the microbiome in the patients may enhance the response to immunotherapy.

Already over 10 years ago it was shown that the gut microbiome was required for effective adoptive T cell therapy of cancer in mice that were preconditioned with total body irradiation (TBI) [260]. Translocation to the mesenteric lymph nodes of bacteria and activated dendritic cells following mucosal damage induced by TBI was required for expansion and antitumor toxicity of the transferred CD8⁺ T cells [260]. More recently, it was shown that the gut microbiome was necessary for antitumor effect of intratumorally injected TLR9 agonist CpG oligonucleotide (CpG-ODN) [254]. CpG-ODN treatment induces production of proinflammatory cytokines including TNF and IL-12 followed by TNF-mediated hemorrhagic necrosis [261]. In germ-free or antibiotic treated mice, the production of proinflammatory cytokines by tumor associated myeloid cells in response to CpG-ODN is impaired, preventing the TNF-mediated necrosis and the subsequent antitumor immune response [254]. In lymphoid organ phagocytes from germ-free mice, genes encoding proinflammatory factors have been shown to have an epigenetically closed chromatin conformation and to be unresponsive to microbial stimulation [262]. Thus, colonization of mice with the commensal microbiome primes myeloid cells for responses to inflammatory stimuli by altering their chromatin conformation. This priming is reminiscent of the mechanisms involved in the phenomenon of trained resistance in innate effector cells [262, 263]. The presence of the gram-negative Alistipes and the gram-positive Ruminococcus genera in the fecal microbiota positively correlated with the amount of TNF produced by tumor associated myeloid cells in response to CpG-ODN. Conversely, the Lactobacillus genus including L. murinus, L. intestinalis, and L. fermentum species known to be anti-inflammatory when used as probiotics negatively correlated with TNF production [254]. Administration of Alistipes shahii and L. fermentum by oral gavage to mice enhanced and decreased TNF production, respectively, by tumor-associated myeloid cells [254]. Thus, antibiotics by depleting the gut microbiome reverse the priming of myeloid cells for response to CpG-ODN while administration of bacterial strains that were positively and negatively correlated with TNF production increased or decreased TNF production, respectively.

Recently, the composition of the gut microbiome was demonstrated to modulate the efficacy of immune checkpoint inhibitors (ICI) therapies (anti-CTLA-4 and anti-PD1/PD-L1) both in animals and clinical trials [258, 259, 264–266].

12.3. Immune Checkpoint Blockers: Anti-CTLA4

It was shown that the absence of gut microbiome in antibiotic-treated or in germ-free mice abolishes the antitumor activity of anti-CTLA-4 [258]. The presence in the microbiome of certain bacterial species, mostly of the Bacteroides genus, was reported in mice to correlate with anti-CTLA-4 therapy efficacy [258]. Gavage of B. thetaiotaomicron or B. fragilis to microbiome-depleted mice activated dendritic cells and induced a Th1 response resulting in efficient antitumor activity of anti-CTLA-4 therapy [258]. Interestingly, treatment with both B. fragilis and Burkholderia cepacia decreased anti-CTLA-4-induced subclinical colitis while enhancing the anticancer response [258]. Thus, by targeting the microbiome it may be possible to dissociate the favorable effect on anticancer therapy from therapy-induced toxicity. Treatment of mice with antibiotics that spare Bacteroidales and Burkholderiales (e.g., vancomycin) improve the efficacy of anti-CTLA-4 therapy [258, 267]. A recent study reports the first series of ICI-associated colitis successfully treated with fecal microbiota transplantation.

The fecal microbiome of the melanoma patients responding to anti-CTLA-4 showed abundance of certain Bacteroides species [258]. When germ-free mice were colonized with this favorable microbiome showed enrichment for B. thetaiotaomicron or B. fragilis they start responding to anti-CTLA-4 [258]. In another trial, the presence of the Bacteroidetes phylum was identified as a prognostic factor for reduced intestinal toxicity [268]. A different study also identified in melanoma patients a microbiome pattern that was associated with both anti-CTLA-4 tumor response and colitis [269]. In this study, the response correlated with Faecalibacterium and other firmicutes while Bacteroidales correlated with poor antitumor response and less toxicity [269].

12.4. Immune Checkpoint Blockers: Anti-PD1/PD-L1

The anticancer activity of anti-PD1/PD-L1 is not fully dependent on the presence of the gut microbiome, but it can be enhanced by the presence of certain bacterial species [259, 265, 266]. Mice from different vendors sustain the growth of B16 melanoma at different rates and anti-PD-L1 therapy is more effective in the mice with slow tumor growth that are characterized by a higher number of tumor-infiltrating CD8⁺ T cells suggesting a stronger antitumor immunity before treatment [259]. Cohousing of the different mice transmit the ability to slow the growth of the tumor and the responsiveness to anti-PD-L1 suggesting a role of the microbiome [259]. Indeed, responsiveness correlated with the Bifidobacterium genus (B. breve, B. longum, and B. adolescentis) [259]. The therapeutic activity of anti-PD-L1 in poorly responsive mice could be enhanced by a commercial probiotic preparation of various Bifidobacterium species (including B. breve and B. longum) [259]. These results suggest that Bifidobacterium spp. in the fecal microbiome of mice sustain the development of antitumor immunity that, when enhanced by anti-PD-L1, prevents cancer progression.

The role of the microbiome in anti-PD1 therapy has been recently evaluated in clinical trials treating patients with melanoma and, lung and renal carcinoma [264–266, 270]. While the oral microbiome composition did not distinguish responder and nonresponder patients, the analysis of the fecal microbiota by shotgun metagenomic sequencing showed that certain enterotypes and microbial species in the fecal microbiome were associated with a favorable response [264]. In one of these studies the fecal microbiome of the responder patients had a higher alpha-diversity, possibly indicating a healthier and varied microbiome that better sustained the establishment of cancer immunity [265]. Surprisingly, however, the sets of fecal bacteria identified in these trials as associated with favorable response to anti-PD1 were completely not overlapping. Increased representation of Akkermansia muciniphila and a trend for a higher representation of Enterococcus hirae characterized lung and renal cancer patients that responded to anti-PD1 [266]. In melanoma patients, Faecalibacterium species were reported in one study to be associated with anti-PD1 response [264] while another study identified Bifidobacterium longum, Collinsella aerofaciens, and Enterococcus faecium [265]. Combined analysis of the data of the three studies with a common bioinformatic platform failed to identify any bacterial species associated with response in all three studies and did not resolve these discrepancy [271]. However, strong support for a role of patients’ microbiome in anti-PD1 response was provided by the finding in all these studies that after colonization of antibiotic-treated or germ-free mice with patients’ fecal microbiomes response to anti-PD1/PD-L1 was observed in those mice that have received the microbiome from responder patients but not in those that have received the microbiome from nonresponder patients [264–266]. Furthermore, the poor response to anti-PD1 of mice that have been associated with the microbiome from nonresponder patients could be rescued by colonization with some of the bacterial species shown to be associated with favorable response (i.e., A. muciniphila alone or combined with E. hirae) [266]. Notably, mice and patients with what has been identified in the different studies as a favorable microbiome had a higher cytotoxic CD8+ T cell infiltration in the tumor microenvironment at treatment baseline suggesting the presence of a preexisting antitumoral immunity that would allow a response to anti-PD1/PD-L1 therapy [264–266]. A further evidence for a role of the microbiome composition in the patients’ response to anti-PD1 was provided by the observation in a few clinical trials in different geographical localization that the microbiome alteration induced by antibiotic treatment of the patients within a month before initiation of the treatment resulted in a lower likelihood of response to the anti-PD1 therapy [266, 272].

Noteworthy, although these studies pinpointed the importance of the microbiota in response to therapy, they identified different bacterial players. These discordant results may be in part explained by the following: the heterogeneity of microbiota composition due to many factors including genetics, lifestyle, and previous therapies; the small patient cohorts studied from geographically distant populations; the different cancer types studied; and variability in criteria for therapy response. Also, most likely the observed regulation by the microbiota of anti-PD1 therapy is not mediated by a single bacterial species but rather is dependent on multiple changes in the microbiota ecology and metabolism that together stimulate cancer immunity. The species or group of species that have been highlighted in these studies could be considered just biomarkers of the ecological changes and, because of the small-sized and heterogeneous cohorts, different bacterial species may have reached significance.

13. Microbiome as a Biomarker, a Target for Cancer and Cancer Therapy Efficacy

Although it has been known for many years that the microbiome and/or specific bacterial species affect tumor formation and progression, the ability of the microbiome composition to modulate cancer therapy effectiveness and toxicity has been a recent finding based on both preclinical and clinical evidence.

The presence of certain bacteria can be used as biomarkers for cancer diagnosis and they could be targeted for cancer progression. The presence of H. pylori in the stomach is considered a risk factor for development of gastric cancer and its eradication is considered efficacious in cancer prevention although it may lead to dysbiosis, metabolic alteration and possibly susceptibility to other cancers of the upper gastrointestinal tract [273–275]. Presence and amount of Fusobacterium nucleatum has been proposed as a possible diagnostic and prognostic marker in colorectal cancer [214, 215, 276]. Recently it has been suggested that microbiome dysbiosis of the tongue coat could be used as a diagnostic marker for liver and pancreatic carcinoma [277, 278].

The recent evidence of the role of the microbiome in modulating cancer therapy has raised the expectation that targeting the microbiome may represent a novel modality to improve therapy efficacy particularly for the new immunotherapy protocols. Ideally, we should aim to modify the microbiome from a procarcinogenic and immunosuppressive composition to one that would enhance antitumor immunity while preventing therapy- and cancer-associated toxicity and comorbidities. However, formidable roadblocks remain. Most data supporting the role of the microbiome in modulating cancer therapy are from experimental mouse models and relatively few and contradictory data exist in humans. Mouse physiology and immunity are not perfect models for human findings and laboratory mice live in a very different environment than patients living in open communities. Recently, it has been suggested that mice colonized with microbiome from wild or pet mice might better replicate the patient response than conventional laboratory mice [279, 280]. Transfer of human microbiome in mice has been used extensively to characterize the mechanisms by which responder patients’ microbiome enhance the response to immunotherapy [258, 264–266]. However, the human microbiome in mice does not perfectly mimic the donor microbiome, it is relatively unstable, and the physiological and immune response of the mice to the human microbiome may not be identical to that of the patients.

The results of the clinical trials in patients treated with anti-PD1 have identified bacterial species that might be associated with response to the therapy (although stringent analysis excluding false discovery rate may limit the significance of these associations [271]) but unfortunately different species have been identified in the various trials [264–266]. In part these contradictory results may be due to the large variability of the composition of the human microbiome among individuals and geographical differences in microbiome composition. For example, it has been shown that while the microbiome composition can be used to predict susceptibility to metabolic disease within Chinese populations living in same district but not across districts [46]. Thus, a major challenge remains to identify reliable microbiome-related biomarkers for prediction of response and stratification of patients across clinical centers. It should also be considered that the microbiome may modulate cancer therapy response not because of a single bacterial species but due to a particular ecology and metabolism of the gut microbiome to which a large number of bacterial species and functions may participate and together affect anticancer immunity. Thus, the bacterial species that have been found in different clinical centers to correlate with response may just be the tip of the iceberg of complex ecological changes that reached statistical significance due to small-sized and heterogeneous cohorts analyzed.

Because of our inability to identify with accuracy the composition of a microbiome that would sustain cancer, it remains difficult to plan therapeutic approaches to target the microbiome in order to improve immunotherapy response. Fecal transplant trials have been planned or initiated to treat patients undergoing immunotherapy. A case series has been reported of immune checkpoint inhibitors-associated colitis successfully treated with fecal transplant from an unrelated healthy donor [267]. However, in the case of fecal transplant for enhancing response to therapy, the inability to identify a favorable microbiome composition has led to clinical protocols based on the transfer of microbiome from cancer patients that have successfully responded to anti-PD1 therapy (ClinicalTrials.gov Identifiers: NCT03353402, NCT03341143 and NCT03772899). If we would be able to identify a favorable microbiome composition for immunotherapy response, the use of healthy balanced fecal microbiome from healthy donors rather than possibly dysbiotic fecal microbiome from sick and treated patients would seem to be preferable. Also, whereas in fecal transplant for opportunistic infections (e.g., Clostridium difficile), most of the preexisting recipient microbiome is depleted by antibiotics treatment [281], the data showing that antibiotics treatment before immune checkpoint therapy may decrease the response rate [266, 272] would suggest to avoid this treatment in fecal transplant aimed to enhance therapy response.

Rather than a crude effort using total fecal transplant with the risk of transferring bacteria that could act as pathogen or pathobionts, it would be ideal to characterize mechanisms of microbiome enhancement of cancer therapy shared by different bacterial species and to identify ecologically balanced consortia of commensal bacteria that favor a clinical response and could be utilized in any clinical center. One approach is to develop standardized microbial ecosystem therapeutics (MET) composed of a pool of bacterial isolates from healthy donors’ fecal microbiome, selected for having minimal antibiotic resistance and lacking any known virulence determinants; this pool of isolates is then formulated into an ecosystem and tested in vitro using chemostat technology [282]. A clinical trial is planned to use one of these preparations orally (MET-4) in solid tumor patients receiving anti-PD1/PD-L1 therapies (ClinicalTrials.gov Identifier: NCT03686202). In a more targeted approach, a consortium of human commensals from healthy human feces was assembled that in mice induce interferon-γ secreting CD8⁺ T cells and enhance host resistance against Listeria monocytogenes infection and the therapeutic efficiency of immune checkpoint inhibitors in syngeneic tumor models [283]. On the basis of these findings, a clinical trial has been announced using VE800, a lyophilized preparation of eight bacterial strains with immunostimulating properties together with anti-PD1 therapy in patients with advanced or metastatic cancer (https://www.vedantabio.com/news-media/press-releases/detail/2492). In a more restrictive approach, clinical trials (ClinicalTrials.gov Identifiers: NCT03775850 and NCT03595683) have been initiated in patients with different type of advanced cancer and treated with immune checkpoint inhibitors using EDP1503, an orally delivered monoclonal microbial product (a Bifidobacterium strain) that correspond to one of the bacterial species identified in clinical studies to correlate with favorable response to anti-PD1 [265].

14. Future Directions and Conclusions

The microbiome research has come a long way in a short time. However, despite the remarkable work carried out in the field, there are still huge gaps in our understanding of how the microbiota shapes our host immunity and how the immune system regulates our microbiota. These questions will require innovation of more tools and new approaches. Regardless of the pending questions, a number of ongoing clinical trials are using microbiota with the aim of improving the outcomes of various chronic diseases (https://clinicaltrials.gov).

As discussed in the preceding sections, the microbiota profoundly affects our host immune system and thus has a great potential to be used as a new biomarker. The microbial composition can mirror inflammation variations in certain disease conditions, within various stages of the same disease; hence, it has the potential to be used as a biomarker.

Using microbes as early diagnostic biomarkers will facilitate disease intervention at the onset of the disease and will help monitor the progress of any therapeutic strategy. The understanding of the microbiome’s influence on the immune responses is vital as we move forward in this era of precision medicine. It is also important to understand in more depth, the complex factors that affect the gut microbiome and how they interact, in order to develop modern strategies to manipulate the microbiome aiming to delay disease onset, disease progression, or improve therapeutic responses.

Moving forward, more efforts should be made to integrate human and microbial multiomics data (such as genetics, transcriptomics, epigenetics, proteomics, and metabolomics) using current bioinformatics and technological approaches. Special attention should be given to standardize our clinical assessment used to classify patients as well as to use universal collection and storage of biological specimens. Well-planned longitudinal studies and integrated analyses of multiple dimensions of data in diverse populations are vital to complete the association and impact on treatment outcomes.