Human gut microbiome: A primer for the clinician

Abstract

The human host gets tremendously influenced by a genetically and phenotypically distinct and heterogeneous constellation of microbial species—the human microbiome—the gut being one of the most densely populated and characterized site for these organisms. Microbiome science has advanced rapidly, technically with respect to the analytical methods and biologically with respect to its mechanistic influence in health and disease states. A clinician conducting a microbiome study should be aware of the nuances related to microbiome research, especially with respect to the technical and biological factors that can influence the interpretation of research outcomes. Hence, this review is an attempt to detail these aspects of the human gut microbiome, with emphasis on its determinants in a healthy state.

The present review provides a perspective on the gut microbiome for the clinician interested in conducting related research. The technical factors such as sample collection and storage, type of sample, DNA extraction, sequencing and bioinformatic platform should be uniform for all samples in a given study. Similarly, biological confounders such as diet, age, genetics, geography and environment should be considered in interpreting results of microbiome related study.

Introduction

Humans are inhabited by a second genetically distinct organ, the gut microbiome, often called the “second metabolic organ”. ¹ The microbial cells and genes are believed to outnumber the human cells and genes by an order of 10 and 100, respectively, although two studies have challenged this proportion and have equated the number of bacterial and human cells. ² , ³ The gut microbiome exerts enormous influence on human health, ⁴ including competition with pathogens through niche exclusion ⁵ and production of antimicrobial peptides, participating in metabolism and energy harvest with thousands of enzymes, ⁶ education/development of the immune system, ⁷ bile salt metabolism, synthesis of vitamins, neurotransmitters, and other metabolites, as well as xenobiotic degradation. ⁸ The human gut is inhabited by all three domains of life: Archaea, Bacteria, and Eukarya, although bacteria dominate the gut environment. The change in quality and quantity of the gut flora or dysbiosis has been associated with several inflammatory and metabolic disorders, including inflammatory bowel disease, ⁹ , ¹⁰ metabolic syndrome and obesity, ¹¹ liver diseases, and neurological disorders. ¹² However, one needs to understand the basics of the healthy human gut microbiome and its determinants before understanding and planning studies on the interaction of the gut microbiome and these disorders. This review describes the functional and structural characteristics of the human gut microbiome and discusses the factors related to its variability and determinants.

Vocabulary of human microbiome

The terms used to define various aspects of the human microbiome are used interchangeably and are quite confusing. We therefore start with a description of terminology related to microbiome so that the rest of this review can be better understood. ¹³ , ¹⁴ While α‐diversity is a measure of microbiome diversity within the sample, β‐diversity indicates the difference in taxonomic diversity between different samples. α‐diversity represents the number and relative abundance of the different species; a high α‐diversity is associated with a healthy gut microbiome, whereas a reduced α‐diversity is associated with various disease states such as Clostridioides Difficile colitis, inflammatory bowel disease (IBD), metabolic syndrome, and liver disease. Samples with similar α‐diversity might have different relative abundances of various taxa, and this difference is captured by β‐diversity, represented by separate clustering of different samples (Fig. 1 and Table 1).

Figure 1.

Multi‐omics terminologies related to the analysis of gut microbial composition and function.

Table 1.

Vocabulary used for describing the microbiome composition and that used during microbial analysis

Measures	Definitions
α‐Diversity	Measure of variability of the microbiome diversity within a sample, reflected by the richness (number of species within a sample) and evenness (relative abundance of different species within a sample) of bacterial species. Various indices for α‐diversity include Shannon index: Measures both richness and evenness with more weightage on richness Simpson index: Measures both richness and evenness with more weightage on evenness Chao: Nonparametric measure of species richness, giving more weightage to low‐abundance species Abundance‐based coverage estimator: Nonparametric measure of species richness Faith's phylogenic diversity: Measure that also incorporates phylogenetic difference between the species
β‐Diversity	Represents the difference in taxonomic diversity between different samples, and can be expressed with (weighted) or without (unweighted) considering the relative abundance of individual taxa Non‐phylogeny‐based: Takes into account abundance of various taxa within samples being compared, for example, Bray–Curtis, Euclidean, or Jaccard distance matrices Phylogeny‐based: Considers relative phylogenetic distances between various taxa and also Unweighted Unifrac, which considers only the presence or absence of taxa Weighted Unifrac, which also takes into account the relative abundance information of various taxa
Operational taxonomic unit (OTU)	DNA sequences with a definite level of similarity (>95%, 97%, or 99%)
Amplicon	Target sequence or gene that is amplified
Amplicon sequence variants	Refers to single DNA sequences recovered from a high‐throughput marker gene analysis. Provides finer sequence resolution than OTU‐based analysis (at the level of single nucleotide change) and is more reproducible, precise, and comprehensive.
Assembly	Alignment and merger of short DNA sequences to form longer DNA fragments
Contig	Contiguous DNA sequences formed by assembly of short DNA fragments
Scaffold	Longer continuous DNA sequences formed by assembly of contigs
Binning	Grouping of DNA sequences or contigs on the basis of their similarity with further assignment into taxa
Annotation	Assignment of functional categories to genes/sequences by mapping to reference genomes
Core microbiome	Microbiome that is present in a definite percentage (50–80%) of the population at a fixed level of abundance (0.01–0.1%)

Analyzing the human microbiome

Interpreting the microbiome depends upon two major factors: technical factors related to sample collection, processing, and analysis; and biological factors related to age, genetics, diet, drugs, geography, and environment. Knowledge of the technical factors is equally important for a clinician for optimum planning and execution of a microbiome‐related study. ¹⁴

Technical factors

Types of samples for microbiome study

There is significant variability between mucosal and fecal flora, which has important implications for any gut microbiome study. ¹⁵ The stool sample reflects the luminal microbiome composition, is easy to collect and useful for longitudinal studies where multiple samples are required at different time points, and can be easily obtained from healthy individuals. However, fecal samples are not exactly representative of the intestinal flora, do not reflect the physiological variations occurring across the intestinal chemical, nutrient, and oxygen gradient, and cannot capture the variability because of differences in intestinal segments. Endoscopic biopsy samples can be used to investigate the mucosal microbiota across different segments of the gastrointestinal (GI) tract, as they more closely reflect the intestinal microbiota composition. However, it is prone to bias because of bowel preparation, contamination due to GI luminal fluid in the biopsy channel, lack of sufficient material for multi‐omics studies, and contamination with the host DNA. Because of the invasive nature of biopsy, these samples are not suitable for longitudinal studies and studies on healthy individuals. Multiple studies in healthy controls and across various disorders including irritable bowel syndrome (IBS), IBD, and colorectal cancer have demonstrated the variability between the two sampling locations, ¹⁶ with greater compositional difference in the fecal sample (between healthy controls [HCs] and disease) in IBS (as compared to mucosal) versus higher difference in mucosal sample in IBD (as compared to fecal). ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ Further, a study in HCs has demonstrated that fecal samples do not reflect the composition and metagenomic function of mucosa‐associated microbiota distributed across multiple sites of the intestine. ²² The specific methodology and sample type related to gut microbiome studies in various intestinal diseases depend upon the research question (Table 2).

Table 2.

Specific methodology for microbiome studies related to inflammatory bowel disease, irritable bowel syndrome, and colorectal cancer ^†

Disease type	Questions that need to be answered	Study design	Type of sample	Marker gene versus metagenomics	Functional studies
Inflammatory bowel disease	Identification of disease‐associated microbiota	Cross‐sectional	Fecal; Mucosal if region‐specific difference and host bacterial interaction need to be studied	Both can be done with fecal sample; Marker gene for mucosal sample	For identification of differences in bacterial gene expression and metabolites produced
	Development of biomarker for disease development	Longitudinal with healthy asymptomatic individuals at risk for disease development	Fecal	Both can be done	To describe specific functional feature associated with disease development
	Development of biomarker for predicting outcomes and response to treatment	Longitudinal in diseased individuals	Fecal	Both can be done	To describe specific functional feature associated with disease variability
	Mechanistic studies to evaluate effect of microbiota on host physiology	Gnotobiotic mice models Patient‐derived organoid models to evaluate interaction of microbiota/metabolites
Irritable bowel syndrome	Identification of disease‐associated microbiota	Cross‐sectional	Fecal; Mucosal if region‐specific difference and host bacterial interaction need to be studied (e.g. small intestinal for patients with IBS)	Both can be done with fecal sample. Marker gene with mucosal sample	For identification of differences in bacterial gene expression and metabolites produced
	Microbiota stability between IBS and controls	Longitudinal	Fecal	Both
	Mechanistic studies	Germ‐free and gnotobiotic mice models, manipulation of gut microbes and microbial metabolites related to IBS‐associated physiology
Colorectal cancer (CRC)	Development of biomarkers for risk stratification and CRC screening	Longitudinal	Fecal	Both depending upon the depth of sequencing queried	Not required
	Detection of adenomas/CRC during surveillance	Longitudinal	Fecal	Both depending upon the depth of sequencing queried	Not required
	Identification of bacteria associated with adenoma formation and CRC	Cross‐sectional	Mucosal	Marker gene	Metabolomics may reveal functional differences between normal tissue and CRC

^† Other important aspects for well‐conducted microbiome study: Robust clinical metadata, uniform sample collection and bioinformatics pipeline, rigorous statistical testing, power calculation, and correction for multiple hypothesis, and adjustment for other covariates such as diet, drugs, ethnicity, and environment.

Collection, transport, processing, and storage of fecal samples

Collection and transport

The sample needs to be collected in a clean container or a clean plastic sheet over the toilet seat. The subsequent transfer to the laboratory depends upon the mode of storage and the study design. ²³ The collected specimen needs to be transported to the laboratory as soon as possible, at room temperature within 4 h; if longer, then it should be done at 4°C within 24–48 h. With a DNA stabilizer solution (RNA later, 95% ethanol, Omnigene‐Gut R), the sample can be transported over a longer time; however, preservatives can affect the metabolites in the fecal sample and such samples cannot be sub‐sampled for culture or used for fecal microbiota transplantation (FMT).

Storage

In the laboratory, samples should be stored at −80°C (such samples remain viable for up to 2 years) (Fig. 2). ²⁴ However, before storage, samples should be aliquoted as per the study design and requirement, as this prevents unnecessary freeze–thaw cycles, which are detrimental to the microbial content in the fecal sample. Sample consistency (as per the Bristol stool chart) also needs to be noted down before storage, but there is no consensus on fecal sample homogenization before aliquoting.

Figure 2.

Means of fecal sample collection, transport, processing, and storage for microbial analysis.

DNA extraction

It is one of the most important factors influencing gut microbiome composition. Available kits differ on their method of bacterial lysis (enzymatic, chemical, mechanical), with most utilizing a combination. Different DNA extraction protocols introduce an inherent bias, thus generating variable results, and hence it is recommended to employ a uniform protocol for the study. ²⁵

Sequencing and bioinformatics

Purified DNA after extraction is sequenced on various types of platforms and further channelized through various bioinformatic pipelines to get a meaningful output in decoding the gut microbiome. ²⁶ Of the different sequencing platforms available, Illumina‐based Miseq is the most commonly used platform, although other platforms, such as Ion Torrent PGM, Pacific Biosciences, NanoPore, Roche 454 GS, and FLX Titanium, are also available. ²⁷ Importantly, for a given study, the same sequencing platform should be used for all the samples. Though the advent of next‐generation high‐throughput sequencing has revolutionized the sample analysis workflow, the large amount of data produced poses significant analytical challenges, which, however, have been mitigated through advancements in bioinformatics platforms and algorithms. Different commercial and open‐source bioinformatic pipelines are available, which differ in their statistical approaches, computational requirements, data handling, and classification accuracy. The classification algorithm can be composition‐ or comparison‐based depending upon whether they compare sequence features (such as GC content) or homology‐based sequences to a reference database. BLAST homology search is one of the commonly used comparison‐based method, but recently hybrid methodologies that combine both approaches have been employed, one of the popularly used tools being MetaPhlAn2. Reference databases are equally important and are used for mapping the sample reads to known sequences for taxonomic classification. Several such databases exist, such as SILVA, Greengenes, and the Ribosomal Database Product (RDP) ²⁶ (Fig. 3).

Figure 3.

Comparison of analytical techniques for 16SrRNA analysis and shotgun metagenomics sequencing for gut microbial analysis.

Sample analysis

The gut microbiome composition can be evaluated either by culturing the organisms or by molecular techniques that either detect specific sequences (marker gene analysis) or sequence the entire genome (metagenomics) (Table 3 and Fig. 3).

Table 3.

Methods of compositional and functional analysis of the gut microbiome

	Description	Advantages	Disadvantages
Marker gene analysis	Based on targeting an amplicon of one gene (marker: DNA sequence that identifies the genome that contains it without the need to identify the entire genome) that is present in every member of population, and is different between individuals with different genomes.	Convenient for taxonomic classification; Cheaper than metagenomics; Easy to perform and interpret	Low resolution at species level; Difficulty in detecting low‐abundant taxa
16S rRNA	For bacterial analysis. Located in the 30S subunit of prokaryotic ribosome. Nine variable regions, each flanked by highly conserved DNA sequence, which provides primer sites for amplification
Internal transcribed spacer region	For fungal analysis
18S rRNA	For fungal and parasitic analysis
Meta‐genomic shotgun sequencing	Total extracted DNA is fragmented and randomly sequenced. Reveals functions of encoded microbial DNA. Taxonomic classification is achieved through comparison with previously annotated genes	Higher resolution than marker gene analysis Functional analysis possible	Requires high levels of expertise, computational overheads, and high sequencing costs
PICRUSt	Phylogenetic Investigation of Communities Using Reconstruction of Unobserved States	Can assign functional pathways to 16srRNA‐based genes, based on their mapping to previous databases
Quantitative microbial profiling	Combination of flow cytometric bacterial cell counts with qPCR targeting the 16srRNA gene	Gives absolute counts of microbiome in a given sample, rather than relative abundance

Culture‐based techniques

The advent of molecular techniques in the late 20th and early 21st century phased out anaerobic culture techniques ²⁸ as most species remained uncultured. Although culture‐based techniques detect the “real” organism and can inform about the exact abundance of a particular species, they are limited by poor sensitivity and can define only 20–40% of the intestinal bacterial community.

Molecular techniques

• Marker gene analysis (Table 3)

Marker gene analysis is based on the detection of a specific sequence or gene (marker) of an organism, which is present in all organisms of the same type and different from others. 16S rRNA‐based sequencing (for bacteria) has revealed more complex gut microbiota than culture‐based techniques, with newer sequences and a larger number of species, most of them being assigned to three major phylogenetic linkages (Bacteroides, Clostridium coccoides, and Clostridium leptum groups). ²⁹ Marker gene sequencing began with Sanger sequencing, which is a labor‐intensive, slow, and expensive technique, which gradually was replaced with next‐generation sequencing (NGS), a cost‐effective technique with massive parallel sequencing throughput brought about by amplification of 16S rRNA genes using primers containing adaptors. Marker gene analysis is convenient for taxonomic classification, easy to perform and interpret, and cheaper than metagenomics. However, it is not appropriate for functional analysis, provides low resolution at the species level, and is difficult to detect low‐abundant taxa.

• Metagenomics (Table 3)

Shotgun sequencing was further developed to sequence the entire bacterial DNA (metagenomics) and involves the random sequencing (e.g. whole‐genome shotgun sequencing) of the total extracted bacterial DNA and then matching the sequences with previously annotated genes and pathways for taxonomic and functional analysis. ³⁰

Metagenomics provides higher resolution than marker gene analysis and its functionality can be predicted, but it requires high levels of expertise, computational overheads, and sequencing costs.

Culturomics

Though the gut bacterial diversity was revealed largely through metagenomics, most of these bacteria remained uncultured until culturing techniques resurfaced through “culturomics.” Described initially by environmental microbiologists, culturomics incorporates multiple culture conditions, matrix‐assisted laser desorption time‐of‐flight (MALDI‐TOF) mass spectrometry, and 16S rRNA sequencing for the identification of bacterial species. ³¹

Functional analysis

Functional potential or exact functionality can be assessed indirectly through the metagenomics approach (what can they do?), through the measurement of gene expression (metatranscriptomics, what is being done?), or through the quantification of the proteins (metaproteomics) or metabolites produced (metabolomics) (what is the end result?). Analysis of functional potential is possible either through metagenomic or 16SrRNA sequencing followed by function prediction through specialized pathways (PICRUSt, Table 3) against references databases such as the Kyoto Encyclopedia of Genes and Genomes (KEGG) and the MetaCyc database, or functionally annotated orthologous groups such as eggNOG. Metatranscriptomics, through analysis of gene expression (RNA sequencing), informs about the active bacterial species present, providing more functionally relevant picture of the gut microbiome than metagenomics. Metaproteomics and metabolomics, through techniques such as liquid chromatography‐mass spectroscopy (LC‐MS), provide a picture of the actual phenotype by measuring what exactly is produced by the microbiota. Metabolomics is the most sensitive technique with respect to functional resolution of the microbiome, although 90% of measured metabolite features may be unknown. ³²

Characteristics of healthy human gut microbiome

Composition of the human gut microbiome

The journey of the human intestinal microflora began with the landmark discovery of Theodor Escherich, who in 1885 described the properties of bacterial population in infant feces, termed “bacterium coli commune,” currently known as Escherichia coli. ³³ The approximate bacterial load in the human intestine (2 × 10¹¹ bacterial cells per gram of feces) and the number of species (~100) were described initially by culture‐based techniques, with Bacteroides, Eubacterium, Clostridium, and Ruminococcus, as the predominant genera. ³⁴ Large‐scale 16S rRNA‐based analysis of mucosal and fecal flora revealed novel sequences belonging to archeal (Methanobrevibacter Smithii) and bacterial phylotypes (Firmicutes, Bacteriodetes, Proteobacteria, Actinobacteria, Fusobacteria, and Verrucomicrobia in decreasing order of abundance). ³⁵ Metagenomic analysis further expanded this complexity, increasing the number of predicted bacterial genes to 0.5 million in the meta‐HIT study of 124 Europeans. ³⁰ , ³⁶ Similarly, the taxonomic alignment of these genes into species or phylotypes increased to 1000. The composition matched previous reports, with >99% genes belonging to bacteria and archea, with Bacteriodes and Firmicutes having the highest abundance. Less than 0.1% was of eukaryotic or viral origin. The Human Microbiome Project (HMP), which commenced in 2007, characterized the microbiome of 242 screened and phenotyped North Americans, covering five major body areas (oral cavity and oropharynx, nasal cavity, skin, gut, and vagina). Oral and stool microbiota had the highest α‐ and β‐diversities. The microbiota at each body habitat exhibited relationships with various driving physical factors such as oxygen, pH, moisture, host immunological factors, as well as inter‐microbial relationships such as mutualism or competition. The recently published expanded dataset of HMP (whole metagenome sequencing of 1631 new samples) characterized new bacterial species, eukaryotes, archaea, and viruses, demonstrating the co‐occurrence of nonbacterial organisms with bacterial species. ³⁷ , ³⁸ An integrated catalog of human gut microbiome was established in 2014 by combining 1267 samples across three continents, and covering strains with diverse occurrences, frequencies, and abundances ³⁹ (Fig. S1).

Functional structure of the human gut microbiome

The gut microbiome is enriched in pathways for metabolism of plant polysaccharides (resulting in the production of short‐chain fatty acids [SCFAs; acetate, propionate, butyrate] and gases), methanogenesis, synthesis of essential amino acids and vitamins, detoxification of xenobiotics, and deglucuronidation of bile salt. ⁶ , ⁴⁰

The gene catalog of the meta‐HIT study has been mapped to ~19 000 different functions, which can be classified into the “minimal gut genome” (functions necessary for bacterial survival) and the “minimal gut metagenome” (involved in homeostasis of entire ecosystem: metabolism of plant polysaccharides and synthesis of amino acids and vitamins). ³⁶

Stability and variability of the human gut microbiome

Inter‐individual variability and intra‐individual stability

In contrast to more than 99.5% genetic similarity between different individuals, the gut microbiome of every individual is personalized, being significantly different from those of other individuals. ²⁹ Though the percentage of Bacteridetes and Firmicutes per individual varies from 10% to 90%, their combined percentage remains at about 95%. In the meta‐HIT study, even for the ubiquitous species (present in >90% individuals) the variability ranged from 12‐ to 2187‐fold, ³⁶ and in HMP, the within‐subject variation over time was much smaller than the between‐subject variation; similar findings have been replicated in the expanded HMP. ³⁷ , ³⁸ Further, several longitudinal studies including 2–37 individuals over a time period as long as 5 years have suggested the temporal stability of the gut microbiome. ⁴¹

Species versus functional variability

In the expanded HMP, the species‐level dynamics was more personalized than at the pathway level, with a greater between‐subject species variability than functional variability. ³⁸ Another study on six adult twin pairs and their mothers suggested a core microbiome at the gene level rather than taxonomy, and >93% functional characteristics were shared between individuals, who otherwise shared less than 0.5% bacterial species. ¹¹

Development and determinants of healthy microbiota

Aging

Infant microbiome

Though controversial, the fetus is considered sterile in‐utero, and the infant acquires the microbiota during and after birth, gradually progressing to dense colonization with aging and reaching an adult‐like configuration by 3 years of age. ⁴² The mode of delivery (vaginal vs caesarean) is the first determinant of an adult‐like microbiome composition, ⁴³ the other factors being the mode of feeding (breast vs formula feed), change in diet (milk‐based to complex plant polysaccharides), ⁴⁴ , ⁴⁵ and other environmental exposures such as antibiotics (Fig. S2). One important age‐related change in the functional pathways involves vitamin biosynthesis, with the infant microbiome being enriched in de novo folate biosynthesis pathway and adult microbiome characterized by pathways metabolizing dietary folate and synthesis of vitamin B12, B7, and B1. ⁴⁶

Adult microbiome

Puberty is associated with a major physiological transition related to sexual maturation, which is reflected in the microbiome composition with divergence into a gender‐specific one. ⁴⁷ The influence of gut microbiome on the susceptibility to autoimmune disease in females was explained in a non‐obese diabetic (NOD) mice study, which showed that the higher susceptibility of female than male mice to disease was diminished in germ‐free conditions and that fecal transfer from male to female mice prior to disease onset protected against the development of disease. ⁴⁷ , ⁴⁸ Age‐specific changes at this stage include a decline in aerobes and facultative anaerobes and an increase in anaerobes and microbial diversity. Adolescent microbiome is less complex and different from that of adults; the adult microbiome is most stable, complex, and resilient to change, with a core microbiome (present in >50% adults) at the functional level. ⁴⁹ The adult microbiome is influenced by a multitude of factors including genetics, diet, geography, environment, and others, which are described in detail below.

Microbiome in the elderly

Transition to elderly‐hood (>65 years) is associated with loss of physical function and decline in functional capacity of organs related to immunity (immunosenescence: loss of naïve CD4+ T cells and increase in the NFkB pathway), growth, metabolism, and energy homeostasis. ⁵⁰ It is also reflected in the gut microbiome composition, as evidenced by the dominance of the phylum Bacteriodetes (57% population) as the core microbiome and reversal of the Bacteriodetes to Firmicutes ratio (Fig. S3) in comparison with the microbiome in a healthy young adult. ⁴² , ⁵¹ The microbiome has also been linked with diet (low‐fat/high‐fiber vs high‐fat/low‐fiber), residential pattern (community dwelling, outpatient, short‐term rehabilitation, long‐stay rehabilitation), and health status, with community‐dwelling population harboring a significantly different and more diverse microbiome than people in long‐stay care (Fig. S4). ⁵² A study examining centenarians (>100 years of age) revealed an interesting association between microbiota and longevity, with the microbial signature of these individuals being paradoxically associated with health‐associated taxa such as Bifidobacterium, Christensenellaceae, and Akkermansia. ⁵³

Genetics

Twin studies

Small low‐powered initial studies have demonstrated statistically insignificant greater similarity between monozygotic (MZ) than dizygotic (DZ) twins, ¹¹ , ⁵⁴ which was further confirmed by larger studies. ⁵⁵ , ⁵⁶ Taxonomically, the family Christensenellaceae (order Clostridiales) was the most inheritable.

Linkage studies

Linkage studies, by associating specific genetic polymorphism with the gut microbiome composition in disease or health states, have further strengthened host genetics–gut microbiome association. ⁵⁷ IBD‐specific loci in Crohn's disease, such as loss‐of‐function polymorphism in the FUT2 gene and the NOD2 risk allele, have been correlated with the modulation of energy metabolism in the gut microbiome and relative abundance of Enterobacteriaceae. ⁵⁸ Further, the host quantitative trait loci (QTL) associated with genes related to immunity have linked them with the relative abundances of specific microbial taxa. ⁵⁹ , ⁶⁰

GWAS studies

Three large genome‐wide association studies (GWAS) have characterized the gene–microbiota association from three cohorts: Germany, Canada, and the Netherlands. ⁶¹ , ⁶² , ⁶³ These studies linked several loci/single‐nucleotide polymorphism (SNP) (9–53) with the relative abundance of microbial taxa and pathways, with the strongest association for C‐type lectin molecules: LCT gene locus with Bifidobacterium genus, LINGO2 with Blautia and VDR (vitamin D receptor) gene. The overall contribution of host genetics to β‐diversity in these studies ranged from 10.4% to 33%. Further, the analysis of human contamination reads in the data from HMP and expanded HMP ⁶⁴ , ⁶⁵ also correlated the taxonomic and functional composition of the gut microbiome with host genetic variation.

Metagenomics

Metagenomic analysis of 250 adult twins from the Twins UK Registry also demonstrated a greater degree of microbial SNP sharing in MZ than DZ twins. ⁶⁶ Further, the SNP similarity between MZ twins decreased with decades of living apart, highlighting the impact of environmental influences. However, a recent genotype and microbiome (metagenomics and 16S rRNA gene sequencing) analysis demonstrated minimal contribution of the host genetics (only 1.9%) to the microbiome variability, challenging the concept of genetic influence on the gut microbiome. ⁶⁷

Diet

Diet is one of the most important environmental factors that shapes the gut microbiome, the evidence being derived from both observational and interventional studies.

Observational studies

The diet–microbiome interaction starts as soon as the baby is colonized, with Bifidobacterium dominating the gut microbiome of breast‐fed infants, and Atopobium, Bacteriodetes, and Enterobacteriacae being relatively abundant in formula‐fed infants. ⁶⁸ The influence of maternal antenatal and postnatal high‐fat diet on the infant microbiome composition was also demonstrated in a recent study. ⁶⁹ The dietary pattern (low‐fat/high‐fiber plant‐based diet vs high‐fat/low‐fiber “Western diet” rich in animal proteins) has a major impact on the gut microbiome composition in adults, as demonstrated by a significant increase in bacterial diversity from carnivore to omnivore to herbivore. ⁷⁰ Similar findings have been replicated in several studies that compared different dietary patterns across different populations from different or the same geography (Table 4). ⁴⁶ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹

Table 4.

Effect of diet and geography on the gut microbiome across several populations

Author/year/children versus adults	Regions studied	Diets compared	Results	Overall remarks
De Filippo et al. (2010)/Children	Rural village in Burkina Faso	Vegetarian, rich in starch, fiber, plant polysachharide	Rich in Actinobacteria and Bacterodetes, genus Prevotella, Xylanibacter, Treponema More SCFAs	Overall higher bacterial richness in African than Italian population. Dominance of diet over geography
	Town in Italy	Rich in animal fat, protein, sugar and starch	Rich in Firmicutes, Proteobacteria, and Enterobacteriaceae genus Allistepes, Bacteriodes
Yatsuneko et al. (2012)/Children versus adults	USA	Western diet rich in animal fat and proteins	Least diversity. Enriched in Bacteriodetes. Rich in glutamine and other amino acid degradation, simple sugar degradation, vitamin and lipoic acid biosynthesis, bile salt metabolism, protein export	Higher temporal instability of children than adult microbiomes Dominance of age, diet, and geography
	Amerindian population of Venezuela	Ancient diet: corn and cassava	Rich in Prevotella, Vitamin B2 biosynthetic pathway, starch degradation, glutamate synthase
	Rural Malawian	Agricultural diet: maize, fruits, vegetables
Ou et al. (2013)/Adults	African‐Americans	Rich in animal fat, protein, sugar, and starch	Predominance of Bacteriodes More gene encoding secondary bile acid production Higher fecal secondary bile acids	Dominance of diet
	Rural Africans	Vegetarian, rich in starch, fiber, plant, polysachharide	Higher diversity Predominance of Prevotella More butyrate producers Higher fecal SCFAs
Schnorr et al. (2014)/Adults	Hazda hunter‐gatherers, Tanzania (Foragers)	Wild foods: meat, honey, baobab, berries and tubers, and game meat	Rich in Prevotella, Succinivibrio, Treponema and unclassified members of Bacteroidetes, Clostridiales and Ruminococcaceae, Proteobacteria. Absent Bifidobacterium	Higher diversity in Hazda than Italians Sex‐related divergence due to difference in dietary composition Dominance of diet and gender
	Bolgona, Italy (modern lifestyle)	Mediterranean diet: abundant plant foods, fresh fruit, pasta, bread and olive oil. Low dairy, poultry, fish and red meat	Higher abundance of Bifidobacterium, Firmicutes (Blautia, Ruminococcus, and Faecalibacterium)
Obregon Tito et al. (2014)/Adults	Matses: Hunter‐gatherer population in Peru	Gathered tubers, invasive plantains, fish, low dairy product	Enriched for Proteobacteria Spirochaetes, Cyanobacteria, Tenericutes, and Euryarchaeota	Higher diversity in Mastes and Tunapoco than Oklahoma; Dominance of diet
	Tunapuco, a traditional agricultural community from the Andean highlands	Stem and root tubers, fruits, guinea pig, pork, lamb, and infrequently cow cheese, rice, and bread	Enriched for Proteobacteria, Spirochaetes, Bacteroidetes, Prevotella
	Oklahoma, a typical US university community	Processed foods including canned fruits and vegetables, bread, and prepackaged meals	Enriched for Actinobacteria, Firmicutes, Bacteriodes, Ruminococcus, Blautia, Dorea
Gomez et al. (2016)/Adults	BaAka rainforest hunter‐gatherers	Ancient hunter diet	Increased abundance of Prevotellaceae, Treponema, and Clostridiaceae. Increased abundance of predicted virulence, amino acid, and vitamin metabolism functions	Progressive change in microbiome diversity, composition, and function from hunter to agricultural to urban population
	Bantu neighbors	Agricultural diet	Enriched in Firmicutes and Bacteriodes. Intermediate abundance of Prevotella, Clostridiaceae, and Treponema
	USA Americans	Western diet	Enriched in Bacteriodes Increased abundance of predictive carbohydrate and xenobiotic metabolic pathways
Morton et al. (2015)/Adults	Pygmy hunter‐gatherers	Hunter ancient diet: Gathered tubers, invasive plantains, fish, low dairy product	Higher Proteobacteria Succinivibrio, Treponema, and Ruminobacter Lower Lachanospiracae	Presence of Entamoeba, location, subsistence, and ancestry as factors determining microbiome compositon. Parasites had the highest dominance Low Shigella and Escherichia in all three
	Bantu farmers	Agricultural diet: grown cereals, vegetables, and meat	High Firmicutes, Ruminococcus, and Treponema
	Bantu fishing population	Fishes, meat, dairy products	Lowest Bacteriodes and highest Prevotella and Bifidobacteria
Zimmer et al. (2012)/Adults	Germany	Vegetarians versus vegans versus omnivores	Bacteroides, Bifidobacterium, Escherichia coli, Enterobacteriaceae low in vegans. Total microbial counts similar; stool pH lowest in vegans	Effect of diet seen
Wu et al. (2016)/Adults	Urban USA	Vegans versus omnivores	Difference in plasma metabolome Similar fecal SCFAs No difference in gut microbiome	No effect of diet seen
Das et al. (2018)/Adults	Ballabgarh rural	Predominantly vegetarian Cooking oil: ghee	High alpha and low beta diversity High Firmicutes and Proteobacteria, low Bacteroidetes High Parabacteroides, Blautia, Brevundimonas, Pelomonas, Megamonas, Collinsella	Effect of diet, cooking oil and geography Functional pathways High in Ballabgarh: membrane transport, carbohydrate metabolism, lipid metabolism, ion channels, and signal transduction and xenobiotic metabolism pathways High in Leh: vitamin biosynthesis, energy metabolism and anti‐inflammatory pathways
	Ballabgarh urban	Predominantly vegetarian Cooking oil: ghee	High α‐ and β‐diversity High Firmicutes, low Bacteroidetes High Lactobacillus, Bacteroides, Vibrio, Eggerthela and Pseudomonas, Collinsella
	Leh rural	Predominantly nonvegetarian Cooking oil: sunflower oil	Low α‐ and β‐diversity High Bacteroidetes, low Proteobacteria High Prevotella, Roseburia, Faecalibacterium, and Lachanospiraceae

Enterotypes

The effect of long‐term dietary pattern in a cohort of 39 individuals assigned the gut microbiome into three clusters or enterotypes (Bacteroides, Prevotella, and Ruminococcus) based on the dominance of specific bacterial taxa enriched for specific gene functions. The Bacteroides enterotype was associated with a Western‐type diet high in proteins and fat, and the Prevotella enterotype was associated with a diet rich in plant polysaccharides. ⁸⁰ However, the enterotype concept was challenged subsequently ⁸¹ , ⁸² following a microbial survey of 200 individuals, showing minimal clustering into Bacteriodes and Prevotella enterotypes, and another survey showing a continuum of Bacteroides abundance across samples rather than distinct clustering. ⁸³ The concept has been further revisited with the accumulation of data and re‐analyses providing a balanced approach toward this understanding. ⁸⁴ Additionally, with the advent of quantitative microbiome profiling (combining amplicon‐based qPCR with flow cytometric enumeration of microbial cells), a 10‐fold variation in the microbial loads of healthy individuals was observed, which was related to enterotype differentiation, with the identification of a low cell count of Bacteroides enterotype (Bact 2, characterized by low proportion of Fecalibacterium and high proportion of Bacteroides), and was correlated with systemic inflammation and disease states. ⁸⁵ , ⁸⁶

Interventional studies

The effect of dietary interventions on the gut microbiome was demonstrated initially in mice studies, which documented a decline in the Bacteroidetes/Firmicutes ratio, increase in Proteobacteria, and a rapid shift in gut microbial composition and functional pathways on high‐fat diet in normal and humanized mice (germ‐free mice populated with human gut microbiota). ⁸⁷ , ⁸⁸ Further, a study of 5 genetically different inbred (wild, MyD88_/_, NOD2_/_, ob/ob_/_, Rag_/_) and >200 outbred mouse strains demonstrated a reproducible and reversible alteration in the gut microbiota on high‐fat/high‐sugar diet across all inbred mice strains independent of their genotype. ⁸⁹

Human studies on diet‐induced weight loss in obese individuals (with high Firmicutes to Bacteroidetes ratio) have demonstrated improvement in the microbial gene richness, increase in the Bacteriodetes/Firmicutes ratio, reduction in butyrate‐producing organisms, increased fecal branched‐chain fatty acids, and decline in fecal SCFAs. ⁹⁰ , ⁹¹ In another study, there was rapid (within 1 day of diet change), reproducible, and reversible alteration in the gut microbiome in response to rapid switches from an animal‐based to a plant‐based diet. ⁹² However, these effects of short‐term dietary perturbations are short‐lasting, and long‐term dietary pattern is the major determinant of the gut microbiome, as demonstrated by Wu et al. who documented that the gut microbiome remained stable on short‐term dietary perturbations (high‐fat/low‐fiber vs low‐fat/high‐fiber diet). ⁹³

Geography and environment

The effect of geography on the gut microbiome includes the effects of genetics and ethnicity, diet and lifestyle, and environment and culture, with diet and lifestyle being the most important. The influence of these factors on the gut microbiome has been assessed by analyzing three major subgroups: population resembling Paleolithic society represented by the hunter‐gatherer population (primitive lifestyle with diet consisting of tubers, nuts, honey, wild game); population resembling Neolithic society represented by rural agricultural societies (thriving on cultivated crop, dairy, and domestic animals); and modern population represented by the urban towns of European and North American populations (on Western high‐protein/high‐fat/low‐fiber refined diet). In general, the gut microbiome proceeds from highest diversity in the foraging population to lowest in urban population, with agricultural society falling in between (Table 4). Gupta et al. recently described the concept of geographically conserved core microbiome, which refers to the set of genera commonly found in a specific body site of all populations irrespective of their geographic location. ⁹⁴ Regarding the gut microbiome, there are 25 genera that are common in all populations across 12 countries, although the relative abundance of these genera might vary. In a recent study, we investigated the gut microbiome of three healthy Indians communities residing at high and low altitude areas (urban and rural). The gut bacterial composition displayed specific signatures and was observed to be influenced by the topographic location and dietary intake of the individuals. The gut microbiome of individuals living at high altitudes was observed to be significantly similar, with a high representation of Bacteroidetes and a low abundance of Proteobacteria; in contrast, the gut microbiome of individuals living in low altitude areas harbored higher numbers of Firmicutes and Proteobacteria and was enriched with microbial xenobiotic degradation pathways. ⁷⁹ The predicted functional diversity of high‐altitude and low‐altitude rural microbiome was higher than that of the low‐altitude urban microbiome.

Other factors

Season and temperature

Seasonal changes in the gut microbiota occur primarily because of the dietary modifications, and have been documented mainly in populations that are dependent upon the environment for their diet. ⁹⁵

The effect of temperature was demonstrated in an elegant study, which showed that adaptation to cold temperature changed the microbiota composition to become sufficiently resistant to cold and resembled the microbiota configuration of obese/high‐fat‐diet‐fed mice: high Firmicutes to Bacteroidetes ratio and low abundance of Akkermansia muciniphila, which increased the ability of the microbiota to harvest energy. ⁹⁶

Pregnancy

The effect of pregnancy on the gut microbiome was demonstrated in a study of 91 pregnant women, which characterized the gut microbiome in all trimesters. The microbiome of the first trimester resembled that of healthy, non‐pregnant women, and the microbiome of third trimester was characterized by increased abundance of Proteobacteria and Actinobacteria, high inter‐individual variability, and lower richness within each woman with respect to the first trimester. ⁹⁷

Drugs

In an elegant study, ciprofloxacin was found to decrease the overall diversity, richness, and evenness of bacterial composition in three individuals and impact the abundance of approximately one‐third of bacterial taxa, which for some did not reverse even after 6 months of ciprofloxacin withdrawal. ⁹⁸ The effect was even reproducible after a second course of ciprofloxacin. ⁹⁹ Further, antibiotic use, through alterations in gut microbiome, can have long‐lasting influences on the host physiology, as demonstrated by an increase in the incidence of allergic and inflammatory disorders such as asthma and IBD in children exposed to antibiotics ¹⁰⁰ and mice studies documenting the transition of the gut microbiome toward an inflammatory phenotype, both in the parent mice and in fecal transfer experiments in the germ‐free mice. ¹⁰¹ , ¹⁰² Drugs other than antibiotics can also impact the gut microbiome as evidenced a study on drug screen of >1000 marketed drugs, of which 24% drugs influenced the gut microbiome. ¹⁰³ Proton pump inhibitors, because of their acid‐suppressing properties, can favor the survival of oral bacteria in the distal segments of GI tract and can modify the flora of all GI segments. ¹⁰⁴

Human gut virome and mycobiome

As compared to “bacterial” microbiome, the human gut virome and mycobiome remain relatively unexplored. Though a detailed description on the characteristics of virome and mycobiome is out of the scope of this review, we provide an introductory primer on the characteristics of these microbial populations in the human gut.

The human gut virome consists primarily of bacteriophages and prophages along with a smaller proportion of eukaryotic viruses. The number of virus‐like particles (VLPs) matches the number of bacterial cells. Viruses are most difficult to characterize because of the necessity of a eukaryotic or prokaryotic host, absence of conserved genes, and lack of matches in reference databases. ¹⁰⁵ The first description of uncultured virome in human feces was published in 2003, and the majority of phage sequences were temperate phages. ¹⁰⁶ Like the microbiome, the virome is unique for each individual, being influenced by diet and the environment, temporally stable, and dominated by Caudavirales and Microviridae. ¹⁰⁷ Recent studies have also characterized the virome in Crohn's disease and ulcerative colitis, ¹⁰⁸ and virome characteristics have also been correlated with response to microbial manipulation in Clostridioides Difficile colitis. ¹⁰⁹

Human gut “mycobiome” has primarily been explored through culture‐based techniques and, recently, with the marker gene analysis, which targets the internal transcribed spacer (ITS) sequence in the fungus. ¹⁰⁵ Mycobiome research still remains unclear on the standardization of analytical techniques including the fungal DNA extraction, sequencing, metagenomics (no metagenomic study on fungal composition to date), and bioinformatic pipelines including reference databases. ¹¹⁰ Fungal diversity is significantly less than bacterial diversity (10⁵–10⁶ fungal cells as compared to 10¹¹–10¹² bacterial cells), although the fungal cells are 100‐fold larger in volume than bacterial cells and fungal diversity is considered uniform across the GI tract. ³ Like for bacteria, a core fungal microbiome of 10 genera has been established, the major phyla being represented by Ascomycota and Basidimycota. Similarly, early life and dietary influences on the mycobiome have also been reported, and progress is being made in understanding the fungal–bacterial relationships, be it mutualism, commensalism, parasitism, or competition. The mycobiome also influences the gut immune system, and several IBD susceptibility genes have been involved in fungal recognition and response to fungi. ¹¹¹

Conclusion and future perspective

Lederberg in 2001 coined the term “microbiome”, which was meant to include the collective genome of the resident microbes associated with any habitat in the human body, and the definition of healthy and dysbiotic microbiome was based on defining the characteristics of these resident microbes. ¹¹² However, the microbiome census does not only include taxomony or “who is present”, but also the functional repertoire or the annotated genes (“what can be done”), the expressed genes or RNA analysis (transcriptomics; “what is being done”), and the synthesized metabolites and proteins (“what is the end result”). This complexity surrounding microbiome analysis and significant variation (even in health states) across individuals, populations, and geography make it difficult to establish a uniform definition of a healthy microbiome, ¹¹³ and there could be multiple healthy microbiome configurations instead of a perfectly healthy microbiome. ¹¹⁴ In terms of taxonomy, because of large inter‐individual differences, the concept of a healthy core structural microbiome is gradually vanishing, together with the realization of the concept that healthy taxa are individual‐ and context‐specific, such as Akkermansia, which is positively correlated with health in metabolic disorders but negatively in multiple sclerosis. ¹¹⁵ The diversity of the microbiome is better correlated with health states, and a highly diverse microbiome is more stable and resilient (capacity to return to homeostatic state in response to external influences) to perturbations, which further characterizes the healthy microbiome, an ecological state that remains temporally constant even after being disturbed by known and unknown factors. ¹¹⁶ Functionally, the microbiome is more similar between individuals, and there have been consistent functional associations with health and disease states. Further, the upcoming concept of microbial ecology, which has expanded the definition to include the host influence on the microbiome, necessitates the need to incorporate the concepts of community ecology into the field of microbiome science. ¹¹⁷ It considers the host as the foundation species for the microbiome with its vast influence on the microbial habitat, nutrition, metabolism, and immune function. ¹¹⁸ The host shapes the microbial community toward the dominance of species that are beneficial for the host, the concept entertained as the “germ–organ theory”, ¹¹⁹ the prime example being the maintenance of epithelial hypoxia thorough oxygen consumption via mitochondrial oxidative phosphorylation, which facilitates the dominance of anaerobes over facultative and obligate aerobes. Therefore, the definition of a heathy microbiome would include the beneficial microbial species and functions and the host component of epithelial hypoxia that maintains these beneficial microbes. Through the phenomenon of colonization resistance, the beneficial microbiome inhibits the harmful species through nutrient competition and production of antibacterial metabolites, and in this way provides a nonspecific immunity towards the pathogens, called “microbiota‐nourishing immunity”. ¹²⁰

Microbiome science has been advancing at a considerable pace, and so has been the advancement in our knowledge on what constitutes a healthy microbiome. Dissecting out these intricacies of microbial contributions to health and disease states would lead to novel strategies to manipulate the microbiome for disease prevention and therapy.

Supporting information

Figure S1. Phylogenetic distribution of bacterial kingdom with the major phyla associated with human gut microbiome (representative example for each phylum and its corresponding class, order, family, genus and species is provided).

Figure S2. Determinants and succession of the infant gut microbiome.

Figure S3. Physiologic transition of the microbial composition and function from adulthood to elderly population.

Figure S4. Determinants of the microbial signature in the elderly population.