Abstract

High individuality, large complexity and limited understanding of the mechanisms underlying human intestinal microbiome function remain the major challenges for designing beneficial modulation strategies. Exemplified by the analysis of intestinal bacteria in a thousand Western adults, we discuss key concepts of the human intestinal microbiome landscape, i.e. the compositional and functional ‘core’, the presence of community types and the existence of alternative stable states. Genomic investigation of core taxa revealed functional redundancy, which is expected to stabilize the ecosystem, as well as taxa with specialized functions that have the potential to shape the microbiome landscape. The contrast between Prevotella- and Bacteroides-dominated systems has been well described. However, less known is the effect of not so abundant bacteria, for example, Dialister spp. that have been proposed to exhibit distinct bistable dynamics. Studies employing time-series analysis have highlighted the dynamical variation in the microbiome landscape with and without the effect of defined perturbations, such as the use of antibiotics or dietary changes. We incorporate ecosystem-level observations of the human intestinal microbiota and its keystone species to suggest avenues for designing microbiome modulation strategies to improve host health.

INTRODUCTION

Microbial cells may outnumber our own cells in the body, presenting an example of nature's intricate evolutionary scheme of host–microbe symbiosis (Frank and Pace 2008; Ley et al. 2008; Sender, Fuchs and Milo 2016). This unique setup of human–microbe interactions and its resilience is of paramount importance for health. Starting from birth, various factors shape the composition and function of the human intestinal tract microbiome, defined as the entire habitat including all microbes, their genomes and surrounding environment (see Table 1) (Sekirov et al. 2010; Ottman et al. 2012).

Table 1.

Terms and definitions used in this review.

Terminologies Definitions Reference 
Active functional core The common functions that are translated and translated actively across human population and have an effect on the whole ecosystem. Kolmeder et al. (2012
Alternative stable states Resilient community states that are stable and resist change over ecologically relevant timescales. Fukami and Nakajima (2011
Alternative transient states The community states that are not necessarily stable and vary in structure and/or function. Fukami and Nakajima (2011
Bacterial phylogenetic core The assemblage of phylogenetically related bacterial taxa in the human intestinal microbiome. Rajilić‐Stojanović et al. (2009); Tap et al. (2009
Bimodal bacteria Bacteria observed to be present in either high or low abundances within a given population of individuals. The states may be driven by external factors and are not necessarily resilient to changes. Lahti et al. (2014
Bistable bacteria Bacteria that have two alternative stable states of low and high abundance, and unlike bimodal bacteria, the states are able to resist change. Faith et al. (2013); Lahti et al. (2014
Common core microbiota Community of microbes and their functions that are shared in majority of humans. Qin et al. (2010
Deterministic effect Assembly of communities in a deterministic manner which is a result of environmental factors and interactions between community members. Fukami and Nakajima (2011
Human intestinal microbiome The entire habitat (all microbes, their genomes and surrounding environment) in the human. Marchesi and Ravel (2015
Microbiota The assemblage (collection and combination) of microorganisms present in a defined environment Marchesi and Ravel (2015
Minimal intestinal metagenome The catalog of functions (coding capacity as assessed by metagenomics) involved in the homeostasis of the whole ecosystem, encoded across many species. Qin et al. (2010
Resilience The capacity of a community to return to the original state after perturbation. Paine, Tegner and Johnson (1998); Walker et al. (2004
Stability Tendency of the community to maintain a state of homeostasis and resist disturbances or to show resilience after disturbance. Scheffer et al. (2001
Stochastic effect Assembly of communities as a result of unpredictable disturbance, dispersal or birth–death events. Chase and Myers (2011
Terminologies Definitions Reference 
Active functional core The common functions that are translated and translated actively across human population and have an effect on the whole ecosystem. Kolmeder et al. (2012
Alternative stable states Resilient community states that are stable and resist change over ecologically relevant timescales. Fukami and Nakajima (2011
Alternative transient states The community states that are not necessarily stable and vary in structure and/or function. Fukami and Nakajima (2011
Bacterial phylogenetic core The assemblage of phylogenetically related bacterial taxa in the human intestinal microbiome. Rajilić‐Stojanović et al. (2009); Tap et al. (2009
Bimodal bacteria Bacteria observed to be present in either high or low abundances within a given population of individuals. The states may be driven by external factors and are not necessarily resilient to changes. Lahti et al. (2014
Bistable bacteria Bacteria that have two alternative stable states of low and high abundance, and unlike bimodal bacteria, the states are able to resist change. Faith et al. (2013); Lahti et al. (2014
Common core microbiota Community of microbes and their functions that are shared in majority of humans. Qin et al. (2010
Deterministic effect Assembly of communities in a deterministic manner which is a result of environmental factors and interactions between community members. Fukami and Nakajima (2011
Human intestinal microbiome The entire habitat (all microbes, their genomes and surrounding environment) in the human. Marchesi and Ravel (2015
Microbiota The assemblage (collection and combination) of microorganisms present in a defined environment Marchesi and Ravel (2015
Minimal intestinal metagenome The catalog of functions (coding capacity as assessed by metagenomics) involved in the homeostasis of the whole ecosystem, encoded across many species. Qin et al. (2010
Resilience The capacity of a community to return to the original state after perturbation. Paine, Tegner and Johnson (1998); Walker et al. (2004
Stability Tendency of the community to maintain a state of homeostasis and resist disturbances or to show resilience after disturbance. Scheffer et al. (2001
Stochastic effect Assembly of communities as a result of unpredictable disturbance, dispersal or birth–death events. Chase and Myers (2011

The human intestinal tract microbiota consists mainly of Bacteria, but also includes Archaea, microeukaryotes and many viruses, mainly bacteriophages. The dominant bacterial phyla are Bacteroidetes and Firmicutes, while Actinobacteria, Proteobacteria and Verrucomicrobia constitute minor phyla (Li et al. 2014; Falony et al. 2016). Archaeal representatives consist largely of methanogens, among which Methanobrevibacter and Methanosphaera are the most prevalent genera (Dridi et al. 2009). The microeukaryotes commonly observed in the human intestinal tract include Blastocystis spp. and fungal genera, such as Candida spp., Gloetenia/Paecilomyces and Galactomyces. Finally, various reports have detailed the presence of bacteriophages in the human intestinal tract (Minot et al. 2013; Waller et al. 2014). A recent deep genomic analysis of virus-like particles present in human fecal samples revealed that most bacteriophages belong to the Caudovirales order (Podovridae, Siphiridae and Myoviridae), some to the Microviridae and no eukaryotic viral sequences were detected (Manrique et al. 2016). The latter study revealed 64 healthy subjects to harbor a total of 44 bacteriophage groups that included a set of core bacteriophages shared among over half of the subjects, a common set that was present in 20%–50% of the subjects, and a set of bacteriophages that are either rarely shared or unique to an individual. While there are indications that bacteriophages may be involved in controlling the intestinal microbiota in health and disease, additional deep studies are needed to assess how these phages are involved in the host–microbe interactions (Reyes et al. 2010, 2015; Manrique et al. 2016).

Interactions between the host, the prokaryotic and eukaryotic microbes and their viruses take place at various levels. While there are marked differences in microbial composition and function along the entire length of the intestinal tract, most research has focused on the fecal microbiota as fecal sampling is a cost-efficient and non-invasive method. However, the compositional and functional attributes of the microbiota vary along the intestinal tract—most notably, the small intestine harbors a unique microbial ecosystem that has been characterized extensively (Zoetendal et al. 2012; Zoetendal and de Vos 2014). Moreover, the fecal microbiota may differ from that in the colonic mucosa but systematic studies are lacking because of the invasiveness of the sampling (Zoetendal et al. 2002; Donaldson, Lee and Mazmanian 2016). As bacteria are the dominant group in this ecosystem, they have received most attention in the recent studies (Lozupone et al. 2012; Li et al. 2014; Falony et al. 2016). After birth, the infant intestinal tract becomes colonized by bacteria largely gained from the mother (Favier, de Vos and Akkermans 2003; Koren et al. 2012). The programmed outgrowth, manifested as succession, and the composition of the adult intestinal microbiota depend on a variety of deterministic host factors, including age, diet, genetic-make-up, immune and health status, and gender and geographical location (Biagi et al. 2010; De Filippo et al. 2010; Claesson et al. 2012; Yatsunenko et al. 2012). Furthermore, stochastic factors such as colonization order or antibiotic exposure can also affect microbial colonization (Lederberg and McCray 2001; Harris et al. 2015; Korpela et al. 2016). Overall, this process leads to the development of a dynamic and complex web of interactions between microorganisms, the host and the environment (Costello et al. 2012; Ottman et al. 2012). Knowledge of the deterministic and stochastic factors affecting the host–microbiota homeostasis could help in defining the factors responsible for evolutionary selection and establishment of the symbiosis between human and microbes.

Bacteria form the dominant domain in the large intestinal microbiota and hence we focus here on the accumulated knowledge of this group. For convenience, the definition of technical terminologies used in this review is given in Table 1. We discuss the concepts of the bacterial compositional and functional core, community types and alternative stable states of the microbial ecosystem. We will further discuss how improved understanding of these features could be instrumental in microbial diagnostics, predicting changes or homeostasis and determining the efficacy of intervention strategies for microbiome modulation.

THE COMMON CORE MICROBIOME IN THE HUMAN LARGE INTESTINE

A major goal in human microbiome studies is to identify and characterize the bacterial taxa and functions that are shared by the majority of individuals. Viewing the bacterial community from a taxonomic perspective reveals an increasing level of complexity toward the lower taxonomic ranks from phylum level to species level. Moreover, the intestinal bacteria interact with each other and the host at various levels from single cells to populations of specific bacteria to the entire community. The complexity of such multi-scale interactions presents a major challenge in understanding community functioning and its impact on host health (Greenblum, Turnbaugh and Borenstein 2012; Greenblum et al. 2013). One way to approach this challenge is to focus on the most abundant bacterial species and specifically the prevalent ones that are shared by many individuals.

Coevolution of host and microbiome can be explained by ecological selection processes (Pianka 1970; Ley, Peterson and Gordon 2006). Host factors such as genetics, immune regulation and age represent top–down selection that plays a role in shaping the microbial community. In contrast, bottom–up selection results in the growth of microbes that have specialized functions, such as those associated with degradation of specific dietary components. Another concept is the r/K selection where organisms that have adapted to maximize the rate of growth are so-called r-strategists, whereas those adapted to compete and survive when resources are limited are known as K-strategists (Pianka 1970). However, applying these idealized concepts to the highly complex, dense and dynamical intestinal ecosystem is challenging (Freilich et al. 2010). The large intestine is known to have spatial heterogeneity in terms of bacterial species which occupy microhabitats (Donaldson, Lee and Mazmanian 2016). For example, the outer mucus layer is densely populated compared to the inner mucus layer (Johansson et al. 2008). The presence of microhabitats with variable population densities suggests that, rather than being present at one of the two endpoints, the bacterial community should be viewed along the r-K continuum that is governed by top–down and bottom–up selection processes.

The aforementioned factors result in the selection of bacteria belonging to specific phylogenetic clusters with adaptive and functional features characteristic for the ecosystem. This coexisting community of microbes and their functions, which are characterized by the minimal intestinal metagenome, have co-evolved with the host for mutual benefits and are prevalent in the population, constitutes the common core (Jalanka-Tuovinen et al. 2011; Li et al. 2014). The core microbiota may include specific keystone species that are important for maintaining an efficiently functioning ecosystem, and whose gain or loss may have profound influence on ecosystem structure and function through their effect on other community members (Paine 1966). In the intestinal tract ecosystem, examples of keystone species include bacteria that are part of the common core. Their functional capabilities include for instance breakdown of complex carbon sources to support the growth of the other core members (Ze et al. 2013; Trosvik and Muinck 2015). In the following section, we will discuss both the phylogenetic core and the minimal intestinal metagenome.

Phylogenetic core

Several studies focused on defining a human intestinal core microbiota in terms of phylogenetic composition (Hamady and Knight 2009). Longitudinal surveys of fecal bacterial 16S ribosomal RNA (rRNA) sequences in individuals for extended periods of time of up to more than 10 years have revealed that a significant fraction of bacterial phylotypes is continuously present in the large intestine indicating that it comprises a stable individual core in healthy adults (Faith et al. 2013; Rajilić-Stojanović et al. 2013). Cross-sectional comparisons of the fecal microbiota in a large population cohort have indicated the presence of a prevalent core microbiota, which comprises those bacterial taxa that can be detected in the majority of individuals (Zoetendal, Rajilić-Stojanović and De Vos 2008). In contrast, the individual core microbiota includes those taxa that are frequently detected within a particular individual; this can overlap with the common core, but also include individual-specific features.

Several studies have aimed to characterize the composition of the common core but their comparison is difficult since different criteria were used (Table 2). The discrepancy in categorizing the common core microbiota could be partially attributed to methodological differences in determining microbiota composition and defining the core. Importantly, the depth of sequencing and analysis, number of samples and taxonomic resolution are the factors that affect the analysis.

Table 2.

Overview of studies that aimed at identifying the compositional and functional core.

Sr. No. Core Compositional Functional Number of subjects Criterion—other comments Study/ year 
No Yes 154 Presence of OTU in all individuals—both obese and lean subjects included Turnbaugh et al. (2009
66 OTUs No 35 Presence of OTU in more than 50% individuals Tap et al. (2009
15 genus-like groups No 10 The study used phylogenetic microarray for the first time to demonstrate the common core Rajilić‐Stojanović et al. (2009
19 No Low number of subjects Claesson et al. (2009
406 in elderly subjects (at least 65 yr). 18 in subjects between 18 and 58 yr No 127 Criterion was presence of OTU in more than 50% individuals. Included ageing subjects Claesson et al. (2011
6% (V1–V3) to 8% (V3–V5) No 327 Criterion was presence of OTU in at least 95% of individuals Huse et al. (2012
Not studied Yes Criterion was proteome features present in all subjects Kolmeder et al. (2012
Variable No 104 Depending on the cut-offs selected for ubiquity and abundance Li, Bihan and Methé (2013
75 species 294 110 genes 124 The most common species were those present in ≥90% of individuals with genome coverage >1% Qin et al. (2010
    Common genes are those present in >50%  
10 290 phylotypes No 115 Core size depends on depth of analysis. Present in all individuals (phylogenetic microarray technology) Salonen et al. (2012
10 288 phylotypes No Present in all individuals at a detection threshold of 0.03% (phylogenetic microarray technology) Jalanka-Tuovinen et al. (2011
11 22 OTUs No 64 Observed at an average frequency of occurrence higher than 90% Zhang et al. (2014
12 43 OTUs No 20 Present in 15 out of 20 individuals Nam et al. (2011
13 9 genera, 30 OTUs No 314 9 genera present in all and at OTU level shared by at least 90% Zhang et al. (2015
14 17 genera No 4000 (Dutch, Belgian, UK and USA) Prevalence threshold was 95%. At varying low mean abundance the number of core genera detected was higher Falony et al. (2016
 14 genera  4000 plus 308 samples from Papua New Guinea Peru, and Tanzania   
 35 genera  In 1106 individuals from the Belgian Flemish Gut Flora Project cohort   
Sr. No. Core Compositional Functional Number of subjects Criterion—other comments Study/ year 
No Yes 154 Presence of OTU in all individuals—both obese and lean subjects included Turnbaugh et al. (2009
66 OTUs No 35 Presence of OTU in more than 50% individuals Tap et al. (2009
15 genus-like groups No 10 The study used phylogenetic microarray for the first time to demonstrate the common core Rajilić‐Stojanović et al. (2009
19 No Low number of subjects Claesson et al. (2009
406 in elderly subjects (at least 65 yr). 18 in subjects between 18 and 58 yr No 127 Criterion was presence of OTU in more than 50% individuals. Included ageing subjects Claesson et al. (2011
6% (V1–V3) to 8% (V3–V5) No 327 Criterion was presence of OTU in at least 95% of individuals Huse et al. (2012
Not studied Yes Criterion was proteome features present in all subjects Kolmeder et al. (2012
Variable No 104 Depending on the cut-offs selected for ubiquity and abundance Li, Bihan and Methé (2013
75 species 294 110 genes 124 The most common species were those present in ≥90% of individuals with genome coverage >1% Qin et al. (2010
    Common genes are those present in >50%  
10 290 phylotypes No 115 Core size depends on depth of analysis. Present in all individuals (phylogenetic microarray technology) Salonen et al. (2012
10 288 phylotypes No Present in all individuals at a detection threshold of 0.03% (phylogenetic microarray technology) Jalanka-Tuovinen et al. (2011
11 22 OTUs No 64 Observed at an average frequency of occurrence higher than 90% Zhang et al. (2014
12 43 OTUs No 20 Present in 15 out of 20 individuals Nam et al. (2011
13 9 genera, 30 OTUs No 314 9 genera present in all and at OTU level shared by at least 90% Zhang et al. (2015
14 17 genera No 4000 (Dutch, Belgian, UK and USA) Prevalence threshold was 95%. At varying low mean abundance the number of core genera detected was higher Falony et al. (2016
 14 genera  4000 plus 308 samples from Papua New Guinea Peru, and Tanzania   
 35 genera  In 1106 individuals from the Belgian Flemish Gut Flora Project cohort   

The aforementioned issues pertaining to the analytical definition of core microbiota have been addressed by using the HITChip (Human Intestinal Tract Chip), a phylogenetic microarray targeting 16S rRNA gene sequences of bacterial taxa that are reported to occur in the human intestinal tract (Salonen et al. 2012). This phylogenetic microarray represents a measurement platform with high reproducibility, robustness and dynamic range, allowing the incorporation of both abundant and rare taxa in microbiota profiling. However, while capturing the vast majority of intestinal microbial species, these and other phylogenetic microarrays lack the ability to detect novel taxa that can be assessed by next-generation sequencing techniques (Hazen et al. 2010; Roh et al. 2010; Tottey et al. 2013). Previously, the HITChip-based analysis of >1000 rare and abundant bacterial phylotypes from 130 genus-level groups in 115 adults revealed a common core of 290 phylotypes (Jalanka-Tuovinen et al. 2011; Salonen et al. 2012). We extended this methodology to analyze the common core in a cohort of 1006 non-compromised Western adults (Lahti et al. 2014). This provided a global view of the common core that revealed bacteria related to Faecalibacterium prausnitzii, Oscillospira guillermondii and Ruminococcus obeum as the top three taxa shared by all adults (Fig. S1, Supporting Information). Previously, we reported that the method used for DNA extraction affected the microbiome analysis and that mechanical extraction is to be preferred, as further confirmed by the International Human Microbiome Standards analysis (Salonen et al. 2010; Santiago et al. 2014). Hence, we separately analyzed the samples processed using a mechanical or enzymatic DNA extraction protocol. While no major difference in the inclusion or exclusion of major core bacteria was observed, differences in their ranked cumulative fractional abundances were observed (Fig. S2, Supporting Information). Limiting our analysis to the subset of 401 samples obtained by mechanical extraction using repeated bead-beating revealed that bacteria related to F. prausnitzii, R. obeum and Subdoligranulum variabile to be the top three taxa shared by all adults. Moreover, we found that 34 out of the 130 genus-like groups targeted by the HITChip are shared in 95% of the subjects at the minimum detection threshold of 0.1% relative abundance, with the highest prevalence being observed for bacteria belonging to the order Clostridiales (Fig. 1; see Supporting Information for methods). This estimate is similar in number to the recently published data by the Belgian Flemish Gut Flora Project (FGFP) on 1106 individuals at 95% prevalence (Falony et al. 2016). However, in the same study, additional inclusion of the subjects from the LifeLines-DEEP study cohort (Dutch nationals), as well as from UK and US studies, resulted in a reduced core consisting of 17 genera that could be observed in both cohorts and there was no specific threshold on relative abundance for identifying core bacteria. In comparison, our study cohort included subjects from across mainland Europe, UK/Ireland and the USA (see Supporting Information). The lower estimates of the core size in the FGFP cohort could be partially explained by technical issues, including differences in DNA extraction methods, variations in sequencing depth or 16S rRNA gene primer sequences. However, at genus level, the reported core taxa from these two studies are largely consistent (Table S1, Supporting Information). Furthermore, reducing the stringency in the prevalence threshold in our data set to 80% and 50% with a constant detection threshold of 0.1% resulted in an increased size of the common core from 34 to 44 and 55 genera at 80% and 50% prevalence, respectively (Fig. 1 and Table S1, Supporting Information). Thus, half the population studied here shared 42% of the 130 genus-like groups investigated. This further supports previous observations that the definition of the core microbiota is sensitive to the abundance and prevalence thresholds (Salonen et al. 2012). In addition, there is a large amount of shotgun metagenomics data available in public databases which in some cases can identify species and led to the metagenome species concept defining not yet cultured and genomically sequenced taxa (Sunagawa et al. 2013). This metagenomic approach is not affected by primer or PCR bias but to some extent is limited by the depth of sequencing and choice of DNA extraction method. Therefore, we further investigated the largest available metagenome catalog of 1267 human fecal samples to identify the core bacterial genera (Li et al. 2014). At a detection threshold set at 0.1%, only Bacteroides spp. were detected to be present in >90% of the samples (Table S1, Supporting Information). In order to overcome the influence of lower sequencing depth, we reduced the abundance threshold by 10-fold to 0.01% (still a conservative threshold) and identified eight genera namely Bacteroides, Eubacterium, Faecalibacterium, Alistipes, Ruminococcus, Clostridium, Roseburia and Blautia in more than half of the studied population (Table S1, Supporting Information). While a comparison of technologies is not the aim of the current review, we would like to highlight that the core microbiota includes a limited number of taxa that are commonly observed, irrespective of the technology used. As sequencing costs decline and new bioinformatics tools are being developed, shotgun metagenomics can be expected to become a standard in microbiome research and will lead to more uniformity between studies for comparisons. Hence, it will be imperative to revisit the core microbiota analysis in the future for more conclusive evidence. Studies incorporating strain-level comparison and identification of genomes without a reference genome will provide an even more detailed picture of the members, dynamics and functional aspects of the core microbiota (Qin et al. 2010; Li et al. 2014; Scholz et al. 2016). We hypothesize that different phylogenetic groups, which share important core functions in the intestinal ecosystem, co-exist and are part of the core. Analysis of the cumulative fractional abundance indicated that the common core bacteria are among the most dominant taxa in the Western adult population, including bacteria related to F. prausnitzii, R. obeum, S. variabile, Prevotella melaninogenica and Oscillospira guillermondii (Fig. S2, Supporting Information). Cultured representatives for the vast majority of the common core bacteria are available and have sequenced genomes with information regarding their general metabolic traits and their association with human health. To this end, an extensive survey of literature was done to identify the general metabolic traits of the common core genus-like groups, and information for a subset discussed in more detail in this review is highlighted in Table 3. For information on other common core genus-like groups, see Table S2 (Supporting Information). We observed that there is a lack of in-depth understanding of the physiology of most of the core bacteria and their potential role in improving the stability of the intestinal microbiome.

Figure 1.

Bacterial phylogenetic core in the human large intestinal microbiome. We use data and analysis methodology from our previous studies (Jalanka-Tuovinen et al. 2011; Salonen et al. 2012). The data set was filtered based on DNA extraction method i.e. we included only the samples processed with the repeated bead beating method (rbb) which has been shown to outperform other methods (Salonen et al. 2010).

Figure 1.

Bacterial phylogenetic core in the human large intestinal microbiome. We use data and analysis methodology from our previous studies (Jalanka-Tuovinen et al. 2011; Salonen et al. 2012). The data set was filtered based on DNA extraction method i.e. we included only the samples processed with the repeated bead beating method (rbb) which has been shown to outperform other methods (Salonen et al. 2010).

Figure 2.

Spearman correlation coefficient for KO profiles for 80 genomes representative of the top 50 genus-level bacterial taxa including the phylogenetic core. (A) The distribution of Spearman correlation values in the core taxa. (B) The representative core bacterial genomes and their correlation to other genomes (each point depicts an independent genome correlated to genome on y-axis). The genomic data were produced by the US Department of Energy Joint Genome Institute http://www.jgi.doe.gov/ in collaboration with the user community and were accessed using the IMG system.

Figure 2.

Spearman correlation coefficient for KO profiles for 80 genomes representative of the top 50 genus-level bacterial taxa including the phylogenetic core. (A) The distribution of Spearman correlation values in the core taxa. (B) The representative core bacterial genomes and their correlation to other genomes (each point depicts an independent genome correlated to genome on y-axis). The genomic data were produced by the US Department of Energy Joint Genome Institute http://www.jgi.doe.gov/ in collaboration with the user community and were accessed using the IMG system.

Table 3.

Selected core genus-like taxa in human large intestine. Prevalence, general metabolic trait and health associations of few key taxa in the human intestine. Microbiota profiling was done as described previously (Lahti et al. 2014). The analysis is based on 130 genus-like groups as defined by (Rajilić‐Stojanović et al. 2009). For a more detailed information on all the common core genus-like taxa can be found in the supplementary Table S2.

Sr. No. Genus-like taxa Prevalence General metabolic trait Health association(s) 
Faecalibacterium prausnitzii 100% Produce butyrate, formate and lactate (Duncan et al. 2002Decreased abundance in Crohn's disease and colon cancer (Kang et al. 2010; Chen et al. 2012
Bifidobacterium 99.50% Produce lactate and acetate by utilizing various oligosaccharides (Turroni et al. 2008Widely used in probiotic preparations for health benefits (Turroni et al. 2008
Akkermansia 96.26% Dominant mucin degrader (Belzer and de Vos 2012Indicator of a healthy metabolic profile in humans (Dao et al. 2016
Prevotella melaninogenica 89.53% Acetate producer (Wu et al. 1992One of the tipping elements (Lahti et al. 2014
Ruminococcus bromii 88.53% Amylolytic activity (Ze et al. 2012Keystone species and healthy effects via breakdown of resistant starch (Ze et al. 2012
Bacteroides fragilis 86.28% Few strains have ability to degrade mucin (Tailford et al. 2015Anti-inflammatory effects and also opportunistic pathogen (Sansonetti 2011
Sr. No. Genus-like taxa Prevalence General metabolic trait Health association(s) 
Faecalibacterium prausnitzii 100% Produce butyrate, formate and lactate (Duncan et al. 2002Decreased abundance in Crohn's disease and colon cancer (Kang et al. 2010; Chen et al. 2012
Bifidobacterium 99.50% Produce lactate and acetate by utilizing various oligosaccharides (Turroni et al. 2008Widely used in probiotic preparations for health benefits (Turroni et al. 2008
Akkermansia 96.26% Dominant mucin degrader (Belzer and de Vos 2012Indicator of a healthy metabolic profile in humans (Dao et al. 2016
Prevotella melaninogenica 89.53% Acetate producer (Wu et al. 1992One of the tipping elements (Lahti et al. 2014
Ruminococcus bromii 88.53% Amylolytic activity (Ze et al. 2012Keystone species and healthy effects via breakdown of resistant starch (Ze et al. 2012
Bacteroides fragilis 86.28% Few strains have ability to degrade mucin (Tailford et al. 2015Anti-inflammatory effects and also opportunistic pathogen (Sansonetti 2011

Minimal intestinal metagenome

The human intestinal microbiome is a functional organ, with complex overall effects on host health. Therefore, it is crucial to identify functions commonly associated with the microbiota in health and disease. Whole shotgun metagenomics is widely used for identifying functional potential of the intestinal microbiome. Initially, the metagenomic core was analyzed based on the DNA isolated from fecal samples of 18 individuals with European ancestry (Turnbaugh et al. 2009). Despite the limited sample size, this analysis suggested the presence of a minimal intestinal metagenome that was shared by almost all subjects in this study. In a first comprehensive metagenome study, the functional similarities of the fecal metagenome of 124 European individuals with 3.3 million unique open reading frames were reported and indicated that on average 38% of an individual's total gene pool is shared with others (Qin et al. 2010). In a recent study, the coding capacity of the large intestinal microbiome was addressed in further detail by studying the metagenome of 1070 individuals from three continents, i.e. Europe, America and Asia (Li et al. 2014). This study, which included both healthy and compromised subjects, identified around 9.9 million unique microbial genes within the intestinal microbiome, and the analysis indicated that this estimate is close to the saturated coverage of core gene content and functions. In case of the intestinal ecosystem, the minimal functional core can be expected to also contain more ecosystem-specific pathways, such as those involved in resistance to bile and ability to grow in an anoxic environment using fermentation or anaerobic respiration with terminal electron acceptors other than oxygen. In the first large-scale metagenomic study, more than 1000 clusters of genes that encode vital functions necessary for survival in the large intestine were identified (Qin et al. 2010). This would suggest that the genomes of core bacterial taxa would have evolved toward high functional redundancy. To assess whether the common core observed in this study also demonstrated functional redundancy, we analyzed the representative genomes of the top 50 bacterial taxa that are part of the phylogenetic core (Fig. 1). Since our analysis was performed at genus level, we included known intestinal genomes from different species targeted by the HITChip which resulted in selection of 80 bacterial genomes representative of 50 genus-like taxa (Table S3, Supporting Information) (Rajilić‐Stojanović et al. 2009). We first identified the KEGG orthologs (KO) abundances for each of the genomes using the Integrated Microbial Genomes (IMG) system and calculated the Spearman correlation coefficient (Markowitz et al. 2012) (Figure 2, see Supplementary data for details). From the inferred network, we observed that bacteria belonging to the phylum Bacteroidetes have high functional redundancy, whereas the phylum Firmicutes was comprised of a large number of more functionally diverse core bacteria (Fig. 3). Recently, analysis of the gyraseB encoding gene (gyrB) in fecal samples from humans and related primates (wild chimpanzees, bonobos and gorillas) indicated that Bacteroides and Bifidobacterium spp. have cospeciated with their mammalian hosts over thousands of generations suggesting strong symbiotic association of these taxa (Moeller et al. 2016). This would indicate that the host may also play a role in selecting functionally redundant species, which results in high competition for resources in the ecosystem but at the same time leads to higher ecosystem stability, which ultimately benefits the host (Coyte, Schluter and Foster 2015). However, the common core also includes taxa with specialized functional roles. A striking example is Ruminococcus bromii that was previously reported as a keystone species for its ability to degrade resistant starch and supporting growth of other bacteria capable of utilizing glucose, maltose, panose and isomaltose for growth (Ze et al. 2012). A similar role of providing simple carbon sources and amino acids for supporting growth of microbes in the vicinity of the mucus layer can be played by degraders of mucin, another important food source in the colon. The abundant representative of the Verrucomicrobia, Akkermansia muciniphila, another core member (Fig. 1), is considered a keystone species in this process and is capable of using mucus as sole carbon and energy source, while signaling to the intestinal mucosa (Belzer and de Vos 2012; Everard et al. 2013; Derrien, Belzer and de Vos 2016).

Figure 3.

Functional correlation network of 80 bacterial genomes representative of the top 50 common core bacteria. The Spearman correlation matrix is represented as a network in which each genome is a node and each correlation an edge; the width of the edges is proportional to the magnitude of the correlation (the higher the correlation the thicker the edge line). The nodes with correlation coefficients below 0.5 are not connected, and genomes were placed by a graph ‘spring’ layout algorithm in qgraph (Epskamp et al. 2012). The genomic data were produced by the US Department of Energy Joint Genome Institute http://www.jgi.doe.gov/ in collaboration with the user community and were accessed using the IMG system.

Figure 3.

Functional correlation network of 80 bacterial genomes representative of the top 50 common core bacteria. The Spearman correlation matrix is represented as a network in which each genome is a node and each correlation an edge; the width of the edges is proportional to the magnitude of the correlation (the higher the correlation the thicker the edge line). The nodes with correlation coefficients below 0.5 are not connected, and genomes were placed by a graph ‘spring’ layout algorithm in qgraph (Epskamp et al. 2012). The genomic data were produced by the US Department of Energy Joint Genome Institute http://www.jgi.doe.gov/ in collaboration with the user community and were accessed using the IMG system.

Metagenomic sequencing data can provide insights into the function, taxonomic position and strain-level diversity of both known and not yet cultured species (Sunagawa et al. 2013; Li et al. 2014; Nielsen et al. 2014; Jeraldo et al. 2016). However, the active functions within the intestinal microbiome cannot be identified based on DNA sequencing data alone but require analysis of transcripts and proteins. To this end, several studies have addressed the active functional capacity of the intestinal microbiota at the RNA and protein level, supporting the presence of an active functional core (Verberkmoes et al. 2009; Erickson et al. 2012; Kolmeder et al. 2012, 2015). Metatranscriptomics analysis of fecal samples from 10 American adults revealed alterations in gene-expression profiles after changes in dietary intake (David et al. 2014). Similarly, metaproteomic comparison of 29 non-obese and obese adults has shown larger differences in the active functional bacteria between the two groups than those observed at the compositional level (Kolmeder et al. 2015). While these studies are a step toward functional understanding of the human microbiome, the field is still young. Improved statistical techniques and clinically relevant predictions will play a key role in a wider adoption to identify changes of functional features characteristic of different disease states.

CLUSTERS, CONTINUOUS GRADIENTS AND ALTERNATIVE STABLE STATES

Several approaches have been employed to identify overarching community types, as well as alternative stable states in more specific taxonomic groups (Arumugam et al. 2011; Holmes, Harris and Quince 2012; Ding and Schloss 2014; Lahti et al. 2014). Moreover, longitudinal analysis of bimodal bacteria revealed temporal stability of the alternative states of low and high abundance, which are in some cases observable as ecosystem-level shifts (Lahti et al. 2014). One potential explanation for the observed bistability could be that these states are resilient to perturbations and stable over ecologically relevant timescales (Connell and Sousa 1983). Additionally, the transition from one stable state to an alternative stable state as a result of a disturbance by e.g. dietary change, may comprise a period which can be stable for a long time relative to the extreme end state, called the alternative transient state (Fukami and Nakajima 2011). In alternative transient states, the communities are in a variable, non-resilient state and vary in structure and/or function as a result of processes such as priority effects (time and order of species arrival), external conditions and deterministic processes (Fukami and Nakajima 2011). Thus, the intestinal bacterial community can be viewed as a landscape representing varying global and/or local levels of stability, species richness and diversity (Arumugam et al. 2011; Koren et al. 2013; Thaiss et al. 2014; Zoetendal and de Vos 2014). In this context, the core bacteria and their abundances can be considered as potential biomarkers that may indicate changes in the intestinal ecosystem.

Alternative community states

In an early approach to detect ecosystem-level structures in the intestinal microbiome, deep metagenome analysis suggested the presence of three clusters, termed enterotypes and characterized by co-occurring taxa named after their dominant bacterial groups i.e. Bacteroides, Prevotella and Ruminococcus (Arumugam et al. 2011). One can argue whether it is possible to infer community assembly rules for the complex and highly dynamic intestinal ecosystem based on discrete alternative community types or whether the intestinal microbiome should rather be viewed as a continuous dynamic ecosystem, which is in constant flux but existing in a microbiota–host homeostatic state (Jeffery et al. 2012; Knights et al. 2014). Recently, it was reported that the enterotype-like clustering observed in ordination analysis could be influenced by the initial abundances of Prevotella and Bacteroides in the samples that were analyzed (Gorvitovskaia, Holmes and Huse 2016). Moreover, it has been demonstrated that high levels of Prevotella and Bacteroides can co-exist in the host, suggesting that these types could be complementary rather than mutually exclusive (Lahti et al. 2014). In addition, a separate study showed that removal of these two taxa from analysis did not result in differentiation of communities into clusters, suggesting that the dominance of Prevotella or Bacteroides did not represent underlying differences in community structure (Gorvitovskaia, Holmes and Huse 2016). However, that study also found that the dominance of Prevotella or Bacteroides correlated well with dietary habits and lifestyle. Thus, it was suggested that the abundance of Prevotella and Bacteroides might be considered as a ‘biomarker’ of diet and lifestyle instead of using the term enterotypes (Gorvitovskaia, Holmes and Huse 2016). A different approach to clustering and classification of samples based on microbial profiling data uses a probabilistic approach, called the Dirichlet multinomial mixture (DMM) models (Holmes, Harris and Quince 2012; Ding and Schloss 2014). This approach found evidence for the presence of alternative community configurations in the human intestinal microbiome regardless of the extensive intra- and interpersonal variations (Holmes, Harris and Quince 2012; Ding and Schloss 2014).

Ecologists have also applied graph theory-based approaches to identify species interactions networks and modularity. In most studies, the co-occurrence and co-abundance network have been inferred from correlation-based and other similarity measures (Weiss et al. 2016). One such attempt incorporated sparse regression with a number of similarity measures to predict microbial co-occurrence and co-exclusion relationships (Faust et al. 2012). It was observed that the microbial community is organized into microcommunities which would indicate a trade-off between certain combinations of microbes that may represent alternative communities (Faust et al. 2012). Recently, other tools have been developed to handle sparse compositional data (Friedman and Alm 2012; Gevers et al. 2014; Kurtz et al. 2015). Compared to the standard correlation analyses, these new methods take better into account the sparsity, compositionality and high dimensionality of the taxonomic networks. One example is the use of CCREPE to identify Crohn's disease associated community network where it was observed that proinflammatory bacteria dominate and co-exclude other potentially beneficial taxa (Gevers et al. 2014). This seems to be an example of a community shift. A similar observation of variable modularity in microbial interaction network was made in inflammatory bowel disease (IBD) patients and non-IBD controls (Baldassano and Bassett 2016). At the population level, the two states had a different modular structure and highlighted the different community memberships. However, it is vital to remember that the tools and models for identifying co-occurrences and modularity in microbiome data still need improvement (Weiss et al. 2016). In all cases, these studies are observational and do not include interventions and hence inferences on causality cannot be made (de Vos and de Vos 2012). In general, more comprehensive longitudinal analyses and intervention studies will be needed to establish causal relationships.

Continuous gradients and bistability

The observed dynamics of the large intestinal ecosystem suggests the possibility for transitions between alternative and potentially stable states, with diet being one of the major driving factors (Wu et al. 2011; Cotillard et al. 2013; Salonen et al. 2014; Falony et al. 2016). For instance, it has been observed that a diet rich in proteins and animal fat is associated with the Bacteroides dominance, whereas carbohydrates have been linked to high abundance of Prevotella (Wu et al. 2011). The extremes of dietary habits point toward a Prevotella/Bacteroides trade-off driven by diet and suggesting a gradient in relative abundances between the two extreme states (high and low abundance) (Kelsen and Wu 2012). The joint population frequencies (represented as peaks and valleys) of the bistable taxa, Prevotella and Bacteroides exhibit a landscape distribution in the Western adult population (Fig. 4A). This highlights the population-level variation in these key taxonomic groups. These two taxa exhibit high degree of functional redundancy as based on their genomic content when compared to functions shared between populations belonging to the Firmicutes or Bifidobacterium spp. (Fig. 3). Previously, ecosystem-level models based on metagenomic and genomics data indicated that the habitat filtering (selection pressure inflicted by the habitat) may have greater impact than functional independence or cooperation on the community composition and thus, may result in selection of functionally redundant species (Levy and Borenstein 2014). It has also been proposed that the stability of intestinal microbiota is affected more by competing species than cooperating species (Coyte, Schluter and Foster 2015). A new hypothesis emerged from the analysis of genome-based models that showed the module-derived functional redundancy to be reduced in the Bacteroides spp. as compared to other genera, potentially linking this to a decreased resilience to perturbations (Vieira-Silva et al. 2016). Future studies aimed at mechanistic understanding of the interaction between these and other highly abundant or bistable taxa in the community and their potential role in ecosystemic transitions could provide the clues for microbiota modulation for instance based on dietary, antibiotic or other interventions.

Figure 4.

Two-dimensional kernel density (2D-kde) estimates. The abundance of bacteria (log10 transformed relative abundance data) mapped onto the x- axis and y-axis respectively, and the z-axis represents the kernel density estimate, in a 3D perspective plot showing the joint population frequencies of these two taxa demonstrating features similar to a landscape. (A) Mapping of P. melaninogenica and B. fragilis illustrates two distinct peaks that represent low B. fragilis and high P. melaninogenica population and low B. fragilis low P. melaninogenica population. (B) Mapping of P. melaninogenica and Dialister illustrates three distinct peaks that represent low Dialister and high P. melaninogenica population; low Dialister and low P. melaninogenica population; high Dialister and low P. melaninogenica population. The phylogenetic microarray data were obtained from previous study (Lahti et al. 2014).

Figure 4.

Two-dimensional kernel density (2D-kde) estimates. The abundance of bacteria (log10 transformed relative abundance data) mapped onto the x- axis and y-axis respectively, and the z-axis represents the kernel density estimate, in a 3D perspective plot showing the joint population frequencies of these two taxa demonstrating features similar to a landscape. (A) Mapping of P. melaninogenica and B. fragilis illustrates two distinct peaks that represent low B. fragilis and high P. melaninogenica population and low B. fragilis low P. melaninogenica population. (B) Mapping of P. melaninogenica and Dialister illustrates three distinct peaks that represent low Dialister and high P. melaninogenica population; low Dialister and low P. melaninogenica population; high Dialister and low P. melaninogenica population. The phylogenetic microarray data were obtained from previous study (Lahti et al. 2014).

To assess the dynamics of the large intestinal microbiota, it is necessary to study microbiota composition over different time scales and investigate the influence of perturbations. Compared to cross-sectional studies, analysis of time-series can provide information on intraindividual variations. Longitudinal analysis of larger populations gives insights into dynamics and stability of microbiota and could be informative in identifying key species that may play a role in ecosystem shifts (Faust et al. 2015). By applying a discrete time Lotka-Volterra model to characterize bacterial interaction networks, it was observed that a high abundance does not necessarily imply a large role in intestinal microbiota stability (Fisher and Mehta 2014). The identified keystone species (Bacteroides fragilis and B. stercosis) in this study had moderate abundances but showed disproportionate influence on the interaction network. However, this study was limited by the number of individuals (two) included in the study. Variations of the Lotka-Volterra model have been used widely in ecology for the understanding of two species interactions. It is well known that the use of antibiotics may have a large impact on the microbial community and makes it prone to invasion by pathobionts (Bokulich et al. 2016). A combination of Lotka-Volterra modeling and regression analysis in antibiotic-treated mice predicted that a microcommunity network was important in protection against Clostridium difficile invasion (Stein et al. 2013). It was observed that a compromised community showed a catastrophic shift into an alternative state, which was stable and did not revert back to the initial community state after removal of the perturbations. Future studies applying these and other models on larger human cohorts may give better insights and aid in understanding ecology of the large intestinal microbiome.

In the intestinal tract microbiota, there exists a dynamic state and consists of species with varying fluctuations in abundances (Caporaso et al. 2011; Flores et al. 2014; Thaiss et al. 2014). As there are thousands of species in the ecosystem, stability may arise in multiple states within a closed domain of possible alternative states also called ‘basins of attraction’ defined by a dynamic attractor (Scheffer et al. 2001; Rosenfeld 2008; Pepper and Rosenfeld 2012). The microbial community in a given individual may move from one attractor to another over time and may include transiently occurring species from external sources (e.g. food) (Fukami and Nakajima 2011; Lang, Eisen and Zivkovic 2014). In many studies, single time point fecal samples are analyzed and these may represent alternative transient states and consequently explain the high individuality observed in many studies (Flores et al. 2014). However, personalized patterns in the intestinal microbiota were detectable over 10 years of time, as shown by long-term monitoring of five healthy subjects (Rajilić-Stojanović et al. 2013). Dense sampling of the human intestinal microbiota has revealed diurnal oscillations in relative abundance of almost 10% of all bacterial taxa and revealed a repetitive pattern of fluctuations in microbiota community configurations specific for a given time of the day (Thaiss et al. 2014). These fluctuations highlight the dynamic nature of the intestinal microbiota and indicate that there exists low and high abundance states for most bacteria. When put in perspective of the alternative transient states, these observations reveal an interesting feature of the intestinal microbiota. It is possible that the daily factors such as diet result in changes in the microbiota structure and composition (introduction of transient bacteria through food and water) and thus result in alternative transient states. This has been experimentally verified by small studies of extreme diets (plant-based diet or animal-based diet) and a diet swap study of African Americans and Native Africans (David et al. 2014; O'Keefe et al. 2015). There can be several possible paths of alternative transient states that may reach in one of multiple states within a limited set of attractors. However, if there is a major shift in dietary habit, this might push the community on an alternative transient path that ultimately results in a catastrophic shift in the community. Investigation of individual bacterial groups revealed bistability (i.e. alternative states of either high or low relative abundance) and identified six genus-level groups that were not only bimodal in their population frequencies but also exhibited bistability, i.e. being predominantly present at either high or low relative abundance with less frequent observations and a reduced temporal stability of the intermediate abundances (Lahti et al. 2014). The intermediate abundance range (tipping point) exhibited reduced temporal stability within subjects (Lahti et al. 2014). These bistable groups varied relatively independently and hence were termed ‘tipping elements’. At a community level, the corresponding compositional states could be described as specific combinations of such bistable tipping elements. Similar to Prevotella melaninogenica and B. fragilis distribution, we were interested in observing the population frequencies of highly abundant bistable bacteria P. melaninogenica and a lower abundance bistable taxon such as Dialister spp. in the cohort of Western adults. We used these examples to illustrate the alternative stable states mapping of the two-dimensional kernel density estimates for bacteria related to P. melaninogenica and Dialister. This analysis revealed three areas with high kernel density (Fig. 4B). The less abundant Dialister group can exist in its high and low abundance states with P. melaninogenica low and high abundance states. Previously, the longitudinal assessment of bistability revealed that Prevotella groups exist in highly bistable states and that their abundance levels are more stable in individuals over 3 months compared to other bistable taxa (Lahti et al. 2014). Furthermore, it was observed that the intestinal microbiota composition of individuals with bistable taxa closer to the tipping point is more variable in time (Lahti et al. 2014). This indicates that the two extreme states are resilient, and changes in the ecosystem are more pronounced when bistable taxa have relative abundances closer to their tipping point.

POTENTIAL IMPLICATIONS OF COMMUNITY CHARACTERISTICS FOR MICROBIOME MODULATION STRATEGIES

In the past few decades, research investigating the ecology of the intestinal microbiome has revealed various features, such as community assembly, temporal dynamics and stability and host covariates associated with intestinal microbiota (Table 4). The knowledge of positive, negative or neutral effects of bacteria on the ecosystem may help in identifying targets for therapeutic modulation. Additionally, features such as resilience of the ecosystem and effect of perturbations need to be understood in detail. In the modification of an ecosystem, perturbation (random, episodic or programmed events) plays a major role (Paine, Tegner and Johnson 1998). The type of perturbation will affect the assembly of the community and also play a role in determining the recovery state. For example, the low and high Prevotella states are highly stable and do not easily change over a short period of treatment (Lahti et al. 2014). Furthermore, the initial state of a bacterial community has been reported to affect its recovery after antibiotic treatment (Raymond et al. 2015). It has also been observed that a low calorie dietary intervention could be helpful for obese individuals with lower bacterial richness (Tap et al. 2015). Richness of the large intestinal microbiome is commonly associated with higher resilience (Lozupone et al. 2012). Resilience and stability are the key features of the intestinal microbiome that need to be addressed while attempting modulation. Different alternative states may vary in both resilience (tendency of a system to return to a position of equilibrium when disturbed) and stability (tendency to maintain the state when subject to disturbance) (Walker et al. 2004). Detailed understanding and identification of potential markers of these features will aid in designing intestinal microbiome modulation strategies.

Table 4.

Representative studies investigating the ecology of the intestinal microbiome employing high-throughput sequencing approach (most recent in each category have been listed).

Sr. No. Names of the study Conceptual insights 
Temporal variability is a personalized feature of the human microbiome (Flores et al. 2014). Elucidates the temporal dynamics and stability of the microbiome including the human. 
Moving pictures of the human microbiome (Caporaso et al. 2011). First large-scale longitudinal sampling temporal dynamics and stability of the human microbiome. 
Transkingdom control of microbiota diurnal oscillations promotes metabolic homeostasis (Thaiss et al. 2014). Address the diurnal changes in microbiota and identifies hourly fluctuations in some bacterial genera. 
Cospeciation of gut microbiota with hominids (Moeller et al. 2016). Provides support to the theory of cospeciation of human intestinal tract microbiota along with hominids. 
Ecological modeling from time-series inference: insight into dynamics and stability of intestinal microbiota (Stein et al. 2013). Temporal dynamics and stability and influence of perturbation on community dynamics and stability. 
The long-term stability of the human gut microbiota (Faith et al. 2013). Addresses the question of long-term stability of microbial community using novel high-throughput sequencing method. 
Antibiotics, birth mode and diet shape microbiome maturation during early life (Bokulich et al. 2016). Elucidated the community assembly in early life and effect of perturbation on shaping the microbiome. 
Population-level analysis of gut microbiome variation (Falony et al. 2016). Population level insights into intestinal microbial community structure and composition with analysis of large number of co-founding factors. 
An integrated catalog of reference genes in the human gut microbiome (Li et al. 2014). Population level insights into intestinal microbial community function. 
10 Enterotypes of the human gut microbiome (Arumugam et al. 2011). Differences in community types/clusters in human intestinal microbiome based on partitioning around the medoid method. 
11 Dynamics and associations of microbial community types across the human body (Ding and Schloss 2014). Differences in community types/clusters in human intestinal microbiome on similar lines of Enterotypes paper but using DMM models. 
12 Linking long-term dietary patterns with gut microbial enterotypes (Wu et al. 2011). Stability of Prevotella enterotype from the Bacteroides enterotype was tested. 
13 Interpreting Prevotella and Bacteroides as biomarkers of diet and lifestyle (Gorvitovskaia, Holmes and Huse 2016Investigating data bias in detecting enterotypes and address the issues pertaining to it. 
14 Microbial co-occurrence relationships in the human microbiome (Faust et al. 2012). Community interactions networks, niche specialization and assembly of microbial community have been demonstrated. 
15 The treatment-naive microbiome in new-onset Crohn's disease (Gevers et al. 2014). Disease and healthy interaction network with respect to Crohn's disease are elucidated. 
16 Metagenomic systems biology of the human gut microbiome reveals topological shifts associated with obesity and IBD (Greenblum, Turnbaugh and Borenstein 2012). Metabolic network modularity of microbiome in health and disease state. 
17 Identifying keystone species in the human gut microbiome from metagenomic timeseries using sparse linear regression (Fisher and Mehta 2014). A novel method for identifying the keystone species based on longitudinal studies is reported. 
18 Ecology of bacteria in the human gastrointestinal tract—identification of keystone and foundation taxa (Trosvik and Muinck 2015Using network statistics and reverse ecology the authors identify foundation and keystone taxa in human intestinal microbiome. 
Sr. No. Names of the study Conceptual insights 
Temporal variability is a personalized feature of the human microbiome (Flores et al. 2014). Elucidates the temporal dynamics and stability of the microbiome including the human. 
Moving pictures of the human microbiome (Caporaso et al. 2011). First large-scale longitudinal sampling temporal dynamics and stability of the human microbiome. 
Transkingdom control of microbiota diurnal oscillations promotes metabolic homeostasis (Thaiss et al. 2014). Address the diurnal changes in microbiota and identifies hourly fluctuations in some bacterial genera. 
Cospeciation of gut microbiota with hominids (Moeller et al. 2016). Provides support to the theory of cospeciation of human intestinal tract microbiota along with hominids. 
Ecological modeling from time-series inference: insight into dynamics and stability of intestinal microbiota (Stein et al. 2013). Temporal dynamics and stability and influence of perturbation on community dynamics and stability. 
The long-term stability of the human gut microbiota (Faith et al. 2013). Addresses the question of long-term stability of microbial community using novel high-throughput sequencing method. 
Antibiotics, birth mode and diet shape microbiome maturation during early life (Bokulich et al. 2016). Elucidated the community assembly in early life and effect of perturbation on shaping the microbiome. 
Population-level analysis of gut microbiome variation (Falony et al. 2016). Population level insights into intestinal microbial community structure and composition with analysis of large number of co-founding factors. 
An integrated catalog of reference genes in the human gut microbiome (Li et al. 2014). Population level insights into intestinal microbial community function. 
10 Enterotypes of the human gut microbiome (Arumugam et al. 2011). Differences in community types/clusters in human intestinal microbiome based on partitioning around the medoid method. 
11 Dynamics and associations of microbial community types across the human body (Ding and Schloss 2014). Differences in community types/clusters in human intestinal microbiome on similar lines of Enterotypes paper but using DMM models. 
12 Linking long-term dietary patterns with gut microbial enterotypes (Wu et al. 2011). Stability of Prevotella enterotype from the Bacteroides enterotype was tested. 
13 Interpreting Prevotella and Bacteroides as biomarkers of diet and lifestyle (Gorvitovskaia, Holmes and Huse 2016Investigating data bias in detecting enterotypes and address the issues pertaining to it. 
14 Microbial co-occurrence relationships in the human microbiome (Faust et al. 2012). Community interactions networks, niche specialization and assembly of microbial community have been demonstrated. 
15 The treatment-naive microbiome in new-onset Crohn's disease (Gevers et al. 2014). Disease and healthy interaction network with respect to Crohn's disease are elucidated. 
16 Metagenomic systems biology of the human gut microbiome reveals topological shifts associated with obesity and IBD (Greenblum, Turnbaugh and Borenstein 2012). Metabolic network modularity of microbiome in health and disease state. 
17 Identifying keystone species in the human gut microbiome from metagenomic timeseries using sparse linear regression (Fisher and Mehta 2014). A novel method for identifying the keystone species based on longitudinal studies is reported. 
18 Ecology of bacteria in the human gastrointestinal tract—identification of keystone and foundation taxa (Trosvik and Muinck 2015Using network statistics and reverse ecology the authors identify foundation and keystone taxa in human intestinal microbiome. 

In various diseased conditions, there is a shift in the composition and function of the intestinal microbiota driven by specific bacteria (possibly pathobionts). These bacteria are responsible for stabilizing an alternative community state, which is unfavorable for host health by producing specific metabolites that deter growth/survival of other bacteria. For example, in an inflamed intestine, increased levels have been observed of proinflammatory bacteria, such as members of the Enterobacteriaceae, associated with increased levels of the proinflammatory lipopolysaccharide produced by this group of Gram-negative bacteria (Wang et al. 2007). The resulting inflammatory cascade includes an increased concentration of host-produced reactive oxygen species that hamper growth of indigenous anaerobic bacteria (Lupp et al. 2007). Thus, this aberration may represent an alternative stable state, the stability of which is maintained by an inflammatory milieu created by the host and influenced by members of the Enterobacteriaceae. Hence, the transition from higher densities of obligate anaerobes to proinflammatory facultative anaerobes would be critical for the establishment and/or maintenance of a state of increased inflammation. In such cases, a reduction in numbers of proinflammatory bacteria will be crucial for a shift toward an alternative state (may or may not be stable) no longer supportive of inflammation, and thus leading to re-establishment of anoxic conditions in the intestine. Along with the reduction in proinflammatory facultative anaerobes, it will be crucial to support the growth of bacteria capable of tolerating oxygen at low concentrations, such as lactobacilli and Akkermansia muciniphila that have anti-inflammatory effects for re-establishment of an anoxic milieu (Tien et al. 2006; Ouwerkerk et al. 2016).

The relative abundance of diagnostic indicator taxa, such as bacteria belonging to Enterobacteriaceae (increasing relative abundance) and A. muciniphila (decreasing relative abundance), was shown to indicate particular environmental conditions such as an inflamed intestinal milieu (Png et al. 2010). In addition, indicator taxa can be sensitive to perturbations and therefore might serve as an early warning sign of unhealthy shifts in the community. Such indicator taxa can be monitored as their abundance may reflect the efficacy of intervention strategies. Using the knowledge of drivers and indicators, it would be possible to detect diagnostic biomarkers, identify specific therapeutic microbes (single keystone species or mixed consortia) for treatment and suggest lifestyle changes (such as dietary habits) that will result in a ‘healthier’ intestinal bacterial community.

Evidence for the predictive nature of intestinal bacteria was recently presented when investigating obese subjects undergoing dietary interventions (Korpela et al. 2014). The responsiveness of the intestinal bacterial community to a specific dietary intervention was predicted by the baseline relative abundances of Clostridium clusters IV, IX and XIVa, and Bacilli. The members of several of the Clostridium clusters also include a number of the core phylotypes in the human intestine, most notably Clostridium cluster IV (C. leptum group, a major constituent of which is the Ruminococcaceae family) and Clostridium cluster XIVa (C. coccoides group, which resembles the Lachnospiraceae family) (Tap et al. 2009; Rajilić-Stojanović and de Vos 2014). Based on these findings, it is tempting to speculate that some core phylotypes may also serve as indicator species for responsiveness, and their relative abundance may have profound influence on the efficacy of a given intervention strategy. Another example is A. muciniphila, which is a part of the intestinal core, and has been reported to be indicative of a healthy metabolic profile. Furthermore, its abundance along with high gene richness was indicative of better clinical outcomes after calorie restriction diet in overweight/obese adults (Dao et al. 2016). Evidence supporting the importance of bacterial community structure in determining individuals’ responses to dietary modulation (in ecological terms; ecological disturbance) was provided by the findings of Salonen et al. (2014). It was observed that the dietary intervention had little effect on the intestinal bacterial composition of non-responders who were characterized by higher initial bacterial diversity. Thus, a higher baseline bacterial diversity represents a more resilient community, and increased responsiveness would be characteristic of an unstable community. This highlights that the baseline bacterial diversity could be one of the potential markers in predicting responsiveness to dietary modulation. However, it is important to acknowledge the effect of other, more specific, factors such as relative abundance of certain bacteria, e.g. bistable bacteria, in determining an individuals’ response to intervention. It will be interesting to investigate whether the bistable groups can serve as predictors for responsiveness or non-responsiveness to modulation strategies.

The functional core and the identification of bistability of specific metabolic processes could also provide clues for predicting efficacy of interventions. For instance, also the overall bacterial gene richness has been demonstrated to exhibit bimodal population frequencies (Le Chatelier et al. 2013). Western individuals have been observed to be separated into two groups based on gene count, i.e. low gene count (LGC) and high gene count (HGC), with low gene richness being associated with pronounced dys-metabolism and low-grade inflammation. Thus, LGC and HGC could represent two alternative states in which one state (LGC) represents a state that is indicative of increased risk for obesity (Cotillard et al. 2013). In this study, it was observed that the LGC group after dietary intervention had improved gene richness concomitant with improvement in the clinical phenotypes but showed little change in the inflammation status of these individuals. These functional alternative states, i.e. LGC and HGC, provide us with an opportunity to predict the efficacy of any approach used for modulation. This further highlights the importance of richness of the intestinal microbiome (both compositional and functional) as a factor in determining resilience to ecosystem changes. It is known that species richness determines the susceptibility to invasion by non-resident species and resilience to overall ecosystem disturbances (Levine and D'Antonio 1999). In case of the intestinal microbiome, resilience of a community with high functional richness adds an important dimension to our understanding of its complexity. Thus, there is a need for considering both species richness and functional gene richness during intervention studies aimed at microbiome modulation.

FUTURE PROSPECTS: APPLYING COMMUNITY LEVEL UNDERSTANDING FOR BENEFICIAL LANDSCAPING OF THE INTESTINAL MICROBIOME

In the past decade, numerous studies targeting the human intestinal microbiome have led toward remarkable advances in our understanding of the crucial role these microbes play in human health. Recognizing the importance of this microbial community, the intestinal microbiome is often called a ‘metabolic organ’ signifying its crucial functional role which has an effect on our health status (Bocci 1992). Importantly, this metabolic organ represents a functioning ecosystem, which is governed by ecological mechanisms. At present, clinically meaningful evidence for the potential application of modulating the intestinal microbiota for therapeutic gain has created considerable interest and enthusiasm (Smits et al. 2013; Li et al. 2016). Treatment of metabolic syndrome and Clostridium difficile infection by ecosystem-level cleansing restoration via fecal microbiota transplantation (FMT) has demonstrated the potential for translating microbiota research to clinical practice (Li et al. 2016). While FMT is promising with proven efficacy in several cases, the underlying mechanisms that result in beneficial effects on host health are unclear and will need further refinement. One of the approaches is targeting specific group(s) of bacteria to bring about the desired changes in the intestinal microbiome. The state of a dynamic system may be manipulated by triggering a switch to an alternative state, for instance with a strong but temporally limited, transient perturbation (‘pulse’, antibiotics), or longer impact (‘press’, diet) (Costello et al. 2012; Faust et al. 2015). These transient perturbations can also be combined with introduction of single bacteria or groups of bacteria that support or inhibit growth of specific bacteria resulting in desired changes in composition and function of the intestinal microbiome.

Managing invasion or overgrowth of pathobionts

In order to manage invasions or overgrowth of pathobionts, we first need to identify phylotypes that are common residents and their relative proportions in the intestinal ecosystem. The study of common core phylotypes and core functions is important in this aspect as this would form the baseline knowledge of ‘normal’ inhabitants. Comparison of this core healthy state with diseased states will assist in identifying invasive species (potentially harmful species that are not part of the core) and/or pathobionts (resident microbes with pathogenic potential). For instance, during IBD Enterobacteriacaeae are known potential pathobiont species that affect the numbers of anaerobic intestinal bacteria as mentioned above. In such cases, the use of therapeutic microbes combined with specific dietary interventions can be used to support growth of core phylotypes capable of fermenting complex polysaccharides to produce short-chain fatty acids (SCFA) and gas (CO2 and H2), which results in decreased intraluminal pH to establish an anoxic environment (Tien et al. 2006; Damaskos and Kolios 2008). Other examples of pathobionts in the human gastrointestinal intestinal tract are Helicobacter pylori, Fusobacterium spp., Desulfovibrio and C. difficile (Biagi et al. 2010; Rowan et al. 2010; Castellarin et al. 2012; Kamada et al. 2013; Martin and Solnick 2014). In most cases, the common observation in disease cases is the overgrowth of these pathobionts. In mouse models of antibiotic-mediated C. difficile infection, it was observed that a few bacteria confer protection and imbalance in their populations increases susceptibility to overgrowth by pathobionts (Stein et al. 2013). These few bacteria included Akkermansia, Coprobacillus and Blautia spp. that are all part of the core and their inhibition by antibiotics was proposed to support overgrowth of C. difficile assisted by Enterococcus spp. in mice (Stein et al. 2013). Thus, it could be possible that the knowledge of common core phylotypes combined with the low and high abundance states of pathobionts could help in designing specific approaches for managing invasion by pathogens or increased relative abundance of pathobionts. These approaches could include changes in the diet that support higher growth of core phylotypes and functions such as butyrate production which inhibits pathogen colonization (Scott et al. 2014).

Adapting intervention strategies for modulating the microbiome

The outcome of a given intervention strategy will depend on the individual's response to certain modulation attempts. The response of the individual to a specific intervention can be predicted and monitored by observing overall community descriptors such as species and gene richness and diversity, as well as abundance/presence/absence of key indicator taxa that may reflect the efficacy of intervention strategies. Indicator taxa could include bistable bacteria, and their shift toward the tipping point could provide clues for transition from diseased to healthy alternate stable states. Studying the transition between low and high abundance states or vice versa with an intermediate tipping point could provide clues for the critical time point for ecosystem restoration by modulating the microbiome toward a different, possibly healthier state. It is also possible that the shift from a diseased state may not necessarily lead to a healthy state but rather to a state different from either of the states. These alternative scenarios may be dependent on the strength of perturbation that resulted in shift to a diseased state. In any case, monitoring of microbiome composition and its dynamics over time should be incorporated in intervention strategies.

Maintenance of a healthy intestinal microbiome landscape throughout life span

Human health is interlinked with the human body and its associated microbiome, and thus, maintaining a healthy microbiome is essential. The characteristics of a healthy community composition have been debated, and could be potentially identified by observing the common core and bacterial communities associated with alternate stable states. The core phylotypes have co-evolved to provide crucial ecosystem services such as production of butyrate (energy source for colonic epithelial cells) and protection against invasive pathogens and outgrowth of endogenous pathobionts. The process of modulating microbiome can be initiated in early life by managing certain host associated and environmental factors like diet. The mode of delivery has an effect on the pattern of development of the intestinal bacterial community (Grölund et al. 1999; Cox et al. 2014). Caesarean delivery is associated with a reduced number of Clostridium species most of which are part of the core microbiota in adult life (Fig. 1) (Bokulich et al. 2016). However, another recent report did not observe associations between the delivery mode and adult microbiota composition (Falony et al. 2016). Maternal health status (obesity) and maternal stress have also been associated with an aberrant intestinal bacterial community in infants (Galley et al. 2014; Zijlmans et al. 2015). However, modifying the dietary habits, breast feeding and incorporating specific oligosaccharides have shown to aid in improving the abundance of anaerobic bacteria (Haarman and Knol 2006). Exercising regularly is known to have various health benefits. A recent study on professional athletes revealed that exercise could aid in increasing intestinal bacterial diversity (Clarke et al. 2014). In elderly subjects, the intestinal microbiome is characterized by reduced bacterial diversity (Biagi et al. 2010; Claesson et al. 2012). Reduced abundance of beneficial core bacteria, increased inflammation and a decrease in the concentration of SCFA's have also been observed in elderly subjects (Salazar et al. 2013). These deviations from a normal adult healthy state could be reduced by targeted modulation, which may include restoring the healthy core phylotypes (through probiotic and/or prebiotic interventions), reducing proinflammatory bacteria and increasing the SCFA production by incorporating dietary fibers in the daily diet.

CONCLUDING REMARKS

In the past few decades, we have gained significant insights into the composition, dynamics and function of our intestinal microbiome and its impact on our well-being. The Western adult population shares dozens of genus-level taxonomic groups that show high functional correlation. The intestinal microbiome should be viewed as a complex landscape, which is a mixture of common and individual characteristics. In addition to continuously changing gradients across bacterial abundance and function, this intestinal microbiome landscape also includes distinct features present as alternative stable states across varied compositional, structural and functional features. It is now clear that treatment for most of the disease conditions necessities careful consideration of the intestinal microbiome. A recent example is the application of a prediction algorithm based on dietary habits, physical activity and gut microbiota that allowed predicting blood glucose levels and designing personalized diets which were effective in controlling post-prandial blood glucose levels (Zeevi et al. 2015). While the field of microbiome research rapidly moves forward, there has to be a careful consideration of the confounding variables in the studies, knowing that correlation does not imply causation, technical bias between studies, differences in statistical models, human error and most importantly reproducibility (Ioannidis 2005; Falony et al. 2016). As the microbiome research moves into the next phase of diagnostics and therapeutics, overcoming these issues will be a major focus of future research.

SUPPLEMENTARY DATA

Supplementary data are available at FEMSRE online.

AUTHOR CONTRIBUTIONS

SAS designed and coordinated the manuscript preparation, and took main responsibility on the writing. LL assisted in generation of figures. FH, LL, HS and WMdV provided critical reviews and suggestions for the content, and contributed to the text. All authors read and approved the final manuscript.

FUNDING

This work was partly supported by the Gravitation grant SIAM (Soehngen Institute for Anaerobic Microbiology) Grant number 024-002-002, the European Research Council (ERC grant 250172—Microbes Inside), and the Academy of Finland (grant number 141130) to WMdV and by the Academy of Finland (grant 295741) to LL.

REFERENCES

Arumugam
M
,
Raes
J
,
Pelletier
E
et al
.
Enterotypes of the human gut microbiome
.
Nature
 
2011
;
473
:
174
80
Baldassano
SN
,
Bassett
DS
.
Topological distortion and reorganized modular structure of gut microbial co-occurrence networks in inflammatory bowel disease
.
Sci Rep
 
2016
;
6
:
26087
, DOI:
Belzer
C
,
de Vos
WM
.
Microbes inside—from diversity to function: the case of Akkermansia
.
ISME J
 
2012
;
6
:
1449
58
Biagi
E
,
Nylund
L
,
Candela
M
et al
.
Through ageing, and beyond: gut microbiota and inflammatory status in seniors and centenarians
.
PLoS One
 
2010
;
5
:
e10667
Bocci
V.
.
The neglected organ: bacterial flora has a crucial immunostimulatory role
.
Perspect Biol Med
 
1992
;
35
:
251
60
Bokulich
NA
,
Chung
J
,
Battaglia
T
et al
.
Antibiotics, birth mode, and diet shape microbiome maturation during early life
.
Sci Transl Med
 
2016
;
8
:
343ra382
Caporaso
JG
,
Lauber
CL
,
Costello
EK
et al
.
Moving pictures of the human microbiome
.
Genome Biol
 
2011
;
12
:
1
Castellarin
M
,
Warren
RL
,
Freeman
JD
et al
.
Fusobacterium nucleatum infection is prevalent in human colorectal carcinoma
.
Genome Res
 
2012
;
22
:
299
306
Chase
JM
,
Myers
JA
.
Disentangling the importance of ecological niches from stochastic processes across scales
.
Philos T R Soc B
 
2011
;
366
:
2351
63
Chen
W
,
Liu
F
,
Ling
Z
et al
.
Human intestinal lumen and mucosa-associated microbiota in patients with colorectal cancer
.
PLoS One
 
2012
;
7
:
e39743
Claesson
MJ
,
Cusack
S
,
O'Sullivan
O
et al
.
Composition, variability, and temporal stability of the intestinal microbiota of the elderly
.
P Natl Acad Sci USA
 
2011
;
108
:
4586
91
Claesson
MJ
,
Jeffery
IB
,
Conde
S
et al
.
Gut microbiota composition correlates with diet and health in the elderly
.
Nature
 
2012
;
488
:
178
84
Claesson
MJ
,
O'Sullivan
O
,
Wang
Q
et al
.
Comparative analysis of pyrosequencing and a phylogenetic microarray for exploring microbial community structures in the human distal intestine
.
PLoS One
 
2009
;
4
:
e6669
Clarke
SF
,
Murphy
EF
,
O'Sullivan
O
et al
.
Exercise and associated dietary extremes impact on gut microbial diversity
.
Gut
 
2014
;
63
:
1913
20
Connell
JH
,
Sousa
WP
.
On the evidence needed to judge ecological stability or persistence
.
Am Nat
 
1983
;
121
:
789
824
Costello
EK
,
Stagaman
K
,
Dethlefsen
L
et al
.
The application of ecological theory toward an understanding of the human microbiome
.
Science
 
2012
;
336
:
1255
62
Cotillard
A
,
Kennedy
SP
,
Kong
LC
et al
.
Dietary intervention impact on gut microbial gene richness
.
Nature
 
2013
;
500
:
585
8
Cox
LM
,
Yamanishi
S
,
Sohn
J
et al
.
Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences
.
Cell
 
2014
;
158
:
705
21
Coyte
KZ
,
Schluter
J
,
Foster
KR
.
The ecology of the microbiome: Networks, competition, and stability
.
Science
 
2015
;
350
:
663
6
Damaskos
D
,
Kolios
G
.
Probiotics and prebiotics in inflammatory bowel disease: microflora ‘on the scope’
.
Brit J Clin Pharmacol
 
2008
;
65
:
453
67
Dao
MC
,
Everard
A
,
Aron-Wisnewsky
J
et al
.
Akkermansia muciniphila and improved metabolic health during a dietary intervention in obesity: relationship with gut microbiome richness and ecology
.
Gut
 
2016
;
65
:
426
36
David
LA
,
Maurice
CF
,
Carmody
RN
et al
.
Diet rapidly and reproducibly alters the human gut microbiome
.
Nature
 
2014
;
505
:
559
63
De Filippo
C
,
Cavalieri
D
,
Di Paola
M
et al
.
Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa
.
P Natl Acad Sci USA
 
2010
;
107
:
14691
6
de Vos
WM
,
de Vos
EA
.
Role of the intestinal microbiome in health and disease: from correlation to causation
.
Nutr Rev
 
2012
;
70
:
S45
56
Derrien
M
,
Belzer
C
,
de Vos
WM
.
Akkermansia muciniphila and its role in regulating host functions
.
Microbial Pathog
 
2016
, http://dx.doi.org/10.1016/j.micpath.2016.02.005
Ding
T
,
Schloss
PD
.
Dynamics and associations of microbial community types across the human body
.
Nature
 
2014
;
509
:
357
Donaldson
GP
,
Lee
SM
,
Mazmanian
SK
.
Gut biogeography of the bacterial microbiota
.
Nat Rev Microbiol
 
2016
;
14
:
20
32
Dridi
B
,
Henry
M
,
El Khechine
A
et al
.
High prevalence of Methanobrevibacter smithii and Methanosphaera stadtmanae detected in the human gut using an improved DNA detection protocol
.
PLoS One
 
2009
;
4
:
e7063
Duncan
SH
,
Hold
GL
,
Harmsen
HJ
et al
.
Growth requirements and fermentation products of Fusobacterium prausnitzii, and a proposal to reclassify it as Faecalibacterium prausnitzii gen. nov., comb. nov
.
Int J Syst Evol Microbiol
 
2002
;
52
:
2141
6
Epskamp
S
,
Cramer
AO
,
Waldorp
LJ
et al
.
qgraph: Network visualizations of relationships in psychometric data
.
J Stat Softw
 
2012
;
48
:
1
18
Erickson
AR
,
Cantarel
BL
,
Lamendella
R
et al
.
Integrated metagenomics/metaproteomics reveals human host-microbiota signatures of Crohn's disease
.
PLoS One
 
2012
;
7
:
e49138
Everard
A
,
Belzer
C
,
Geurts
L
et al
.
Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity
.
P Natl Acad Sci USA
 
2013
;
110
:
9066
71
Faith
JJ
,
Guruge
JL
,
Charbonneau
M
et al
.
The long-term stability of the human gut microbiota
.
Science
 
2013
;
341
:
1237439
Falony
G
,
Joossens
M
,
Vieira-Silva
S
et al
.
Population-level analysis of gut microbiome variation
.
Science
 
2016
;
352
:
560
4
Faust
K
,
Lahti
L
,
Gonze
D
et al
.
Metagenomics meets time series analysis: unraveling microbial community dynamics
.
Curr Opin Microbiol
 
2015
;
25
:
56
66
Faust
K
,
Sathirapongsasuti
JF
,
Izard
J
et al
.
Microbial co-occurrence relationships in the human microbiome
.
PLoS Comput Biol
 
2012
;
8
:
e1002606
Favier
CF
,
de Vos
WM
,
Akkermans
AD
.
Development of bacterial and bifidobacterial communities in feces of newborn babies
.
Anaerobe
 
2003
;
9
:
219
29
Fisher
CK
,
Mehta
P
.
Identifying keystone species in the human gut microbiome from metagenomic timeseries using sparse linear regression
.
PLoS One
 
2014
;
9
:
e102451
Flores
GE
,
Caporaso
JG
,
Henley
JB
et al
.
Temporal variability is a personalized feature of the human microbiome
.
Genome Biol
 
2014
;
15
:
1
Frank
DN
,
Pace
NR
.
Gastrointestinal microbiology enters the metagenomics era
.
Curr Opin Gastroenterol
 
2008
;
24
:
4
10
Freilich
S
,
Kreimer
A
,
Meilijson
I
et al
.
The large-scale organization of the bacterial network of ecological co-occurrence interactions
.
Nucleic Acids Res
 
2010
;
38
:
3857
68
Friedman
J
,
Alm
EJ
.
Inferring correlation networks from genomic survey data
.
PLoS Comput Biol
 
2012
;
8
:
e1002687
Fukami
T
,
Nakajima
M
.
Community assembly: alternative stable states or alternative transient states?
.
Ecol Lett
 
2011
;
14
:
973
84
Galley
JD
,
Bailey
M
,
Dush
CK
et al
.
Maternal obesity is associated with alterations in the gut microbiome in toddlers
.
PLoS One
 
2014
;
9
:
e113026
Gevers
D
,
Kugathasan
S
,
Denson
LA
et al
.
The treatment-naive microbiome in new-onset Crohn's disease
.
Cell Host Microbe
 
2014
;
15
:
382
92
Gorvitovskaia
A
,
Holmes
SP
,
Huse
SM
.
Interpreting Prevotella and Bacteroides as biomarkers of diet and lifestyle
.
Microbiome
 
2016
;
4
:
1
Greenblum
S
,
Chiu
H-C
,
Levy
R
et al
.
Towards a predictive systems-level model of the human microbiome: progress, challenges, and opportunities
.
Curr Opin Biotechnol
 
2013
;
24
:
810
20
Greenblum
S
,
Turnbaugh
PJ
,
Borenstein
E
.
Metagenomic systems biology of the human gut microbiome reveals topological shifts associated with obesity and inflammatory bowel disease
.
P Natl Acad Sci
 
2012
;
109
:
594
9
Grölund
M-M
,
Lehtonen
O-P
,
Eerola
E
et al
.
Fecal microflora in healthy infants born by different methods of delivery: permanent changes in intestinal flora after cesarean delivery
.
J Pediatr Gastr Nutr
 
1999
;
28
:
19
25
Haarman
M
,
Knol
J
.
Quantitative real-time PCR analysis of fecal Lactobacillus species in infants receiving a prebiotic infant formula
.
Appl Environ Microb
 
2006
;
72
:
2359
65
Hamady
M
,
Knight
R
.
Microbial community profiling for human microbiome projects: Tools, techniques, and challenges
.
Genome Res
 
2009
;
19
:
1141
52
Harris
K
,
Parsons
TL
,
Ijaz
UZ
et al
.
Linking statistical and ecological theory: Hubbell's unified neutral theory of biodiversity as a hierarchical Dirichlet process
.
Proc IEEE
 
2015
;
1
14
, DOI:
Hazen
TC
,
Dubinsky
EA
,
DeSantis
TZ
et al
.
Deep-sea oil plume enriches indigenous oil-degrading bacteria
.
Science
 
2010
;
330
:
204
8
Holmes
I
,
Harris
K
,
Quince
C
.
Dirichlet multinomial mixtures: generative models for microbial metagenomics
.
PLoS One
 
2012
;
7
:
e30126
Huse
SM
,
Ye
Y
,
Zhou
Y
et al
.
A core human microbiome as viewed through 16S rRNA sequence clusters
.
PLoS One
 
2012
;
7
:
e34242
Ioannidis
JP.
.
Why most published research findings are false
.
PLoS Med
 
2005
;
2
:
e124
Jalanka-Tuovinen
J
,
Salonen
A
,
Nikkilä
J
et al
.
Intestinal microbiota in healthy adults: temporal analysis reveals individual and common core and relation to intestinal symptoms
.
PLoS One
 
2011
;
6
:
e23035
Jeffery
IB
,
Claesson
MJ
,
O'Toole
PW
et al
.
Categorization of the gut microbiota: enterotypes or gradients?
.
Nat Rev Microbiol
 
2012
;
10
:
591
2
Jeraldo
P
,
Hernandez
A
,
Nielsen
HB
et al
.
Capturing one of the human gut microbiome's most wanted: reconstructing the genome of a novel butyrate-producing, clostridial scavenger from metagenomic sequence data
.
Front Microbiol
 
2016
;
7
:
783
Johansson
ME
,
Phillipson
M
,
Petersson
J
et al
.
The inner of the two Muc2 mucin-dependent mucus layers in colon is devoid of bacteria
.
P Natl Acad Sci USA
 
2008
;
105
:
15064
9
Kamada
N
,
Chen
GY
,
Inohara
N
et al
.
Control of pathogens and pathobionts by the gut microbiota
.
Nat Immunol
 
2013
;
14
:
685
90
Kang
S
,
Denman
SE
,
Morrison
M
et al
.
Dysbiosis of fecal microbiota in Crohn's disease patients as revealed by a custom phylogenetic microarray
.
Inflamm Bowel Dis
 
2010
;
16
:
2034
42
Kelsen
JR
,
Wu
GD
.
The gut microbiota, environment and diseases of modern society
.
Gut Microbes
 
2012
;
3
:
374
82
Knights
D
,
Ward
TL
,
McKinlay
CE
et al
.
Rethinking “enterotypes”
.
Cell Host Microbe
 
2014
;
16
:
433
7
Kolmeder
CA
,
De Been
M
,
Nikkilä
J
et al
.
Comparative metaproteomics and diversity analysis of human intestinal microbiota testifies for its temporal stability and expression of core functions
.
PLoS One
 
2012
;
7
:
e29913
Kolmeder
CA
,
Ritari
J
,
Verdam
FJ
et al
.
Colonic metaproteomic signatures of active bacteria and the host in obesity
.
Proteomics
 
2015
;
15
:
3544
52
Koren
O
,
Goodrich
JK
,
Cullender
TC
et al
.
Host remodeling of the gut microbiome and metabolic changes during pregnancy
.
Cell
 
2012
;
150
:
470
80
Koren
O
,
Knights
D
,
Gonzalez
A
et al
.
A guide to enterotypes across the human body: meta-analysis of microbial community structures in human microbiome datasets
.
PLoS Comput Biol
 
2013
;
9
:
e1002863
Korpela
K
,
Flint
HJ
,
Johnstone
AM
et al
.
Gut microbiota signatures predict host and microbiota responses to dietary interventions in obese individuals
.
PLoS One
 
2014
;
9
:
e90702
Korpela
K
,
Salonen
A
,
Virta
LJ
et al
.
Intestinal microbiome is related to lifetime antibiotic use in Finnish pre-school children
.
Nat Commun
 
2016
;
7
:
10410
, DOI:
Kurtz
ZD
,
Müller
CL
,
Miraldi
ER
et al
.
Sparse and compositionally robust inference of microbial ecological networks
.
PLoS Comput Biol
 
2015
;
11
:
e1004226
Lahti
L
,
Salojärvi
J
,
Salonen
A
et al
.
Tipping elements in the human intestinal ecosystem
.
Nat Commun
 
2014
;
5
:
4344
, DOI:
Lang
JM
,
Eisen
JA
,
Zivkovic
AM
.
The microbes we eat: abundance and taxonomy of microbes consumed in a day's worth of meals for three diet types
.
PeerJ
 
2014
;
2
:
e659
Le Chatelier
E
,
Nielsen
T
,
Qin
J
et al
.
Richness of human gut microbiome correlates with metabolic markers
.
Nature
 
2013
;
500
:
541
6
Lederberg
J
,
McCray
AT
.
Ome SweetOmics–a genealogical treasury of words
.
Scientist
 
2001
;
15
:
8
8
Levine
JM
,
D'Antonio
CM
.
Elton revisited: a review of evidence linking diversity and invasibility
.
Oikos
 
1999
;
87
:
15
26
Levy
R
,
Borenstein
E
.
Metagenomic systems biology and metabolic modeling of the human microbiome: From species composition to community assembly rules
.
Gut Microbes
 
2014
;
5
:
265
70
Ley
RE
,
Lozupone
CA
,
Hamady
M
et al
.
Worlds within worlds: evolution of the vertebrate gut microbiota
.
Nat Rev Microbiol
 
2008
;
6
:
776
88
Ley
RE
,
Peterson
DA
,
Gordon
JI
.
Ecological and evolutionary forces shaping microbial diversity in the human intestine
.
Cell
 
2006
;
124
:
837
48
Li
J
,
Jia
H
,
Cai
X
et al
.
An integrated catalog of reference genes in the human gut microbiome
.
Nat Biotechnol
 
2014
;
32
:
834
41
Li
K
,
Bihan
M
,
Methé
BA
.
Analyses of the stability and core taxonomic memberships of the human microbiome
.
PLoS One
 
2013
;
8
:
e63139
Li
SS
,
Zhu
A
,
Benes
V
et al
.
Durable coexistence of donor and recipient strains after fecal microbiota transplantation
.
Science
 
2016
;
352
:
586
9
Lozupone
CA
,
Stombaugh
JI
,
Gordon
JI
et al
.
Diversity, stability and resilience of the human gut microbiota
.
Nature
 
2012
;
489
:
220
30
Lupp
C
,
Robertson
ML
,
Wickham
ME
et al
.
Host-mediated inflammation disrupts the intestinal microbiota and promotes the overgrowth of Enterobacteriaceae
.
Cell Host Microbe
 
2007
;
2
:
119
29
Manrique
P
,
Bolduc
B
,
Walk
ST
et al
.
Healthy human gut phageome
.
P Natl Acad Sci USA
 
2016
;
113
:
10400
5
Marchesi
JR
,
Ravel
J
.
The vocabulary of microbiome research: a proposal
.
Microbiome
 
2015
;
3
:
1
Markowitz
VM
,
Chen
I-MA
,
Palaniappan
K
et al
.
IMG: the integrated microbial genomes database and comparative analysis system
.
Nucleic Acids Res
 
2012
;
40
:
D115
22
Martin
ME
,
Solnick
JV
.
The gastric microbial community, Helicobacter pylori colonization, and disease
.
Gut Microbes
 
2014
;
5
:
345
50
Minot
S
,
Bryson
A
,
Chehoud
C
et al
.
Rapid evolution of the human gut virome
.
P Natl Acad Sci USA
 
2013
;
110
:
12450
5
Moeller
AH
,
Caro-Quintero
A
,
Mjungu
D
et al
.
Cospeciation of gut microbiota with hominids
.
Science
 
2016
;
353
:
380
2
Nam
Y-D
,
Jung
M-J
,
Roh
SW
et al
.
Comparative analysis of Korean human gut microbiota by barcoded pyrosequencing
.
PLoS One
 
2011
;
6
:
e22109
Nielsen
HB
,
Almeida
M
,
Juncker
AS
et al
.
Identification and assembly of genomes and genetic elements in complex metagenomic samples without using reference genomes
.
Nat Biotechnol
 
2014
;
32
:
822
8
O'Keefe
SJ
,
Li
JV
,
Lahti
L
et al
.
Fat, fibre and cancer risk in African Americans and rural Africans
.
Nat Commun
 
2015
;
6
Ottman
N
,
Smidt
H
,
De Vos
WM
et al
.
The function of our microbiota: who is out there and what do they do?
.
Front Cell Infect Microbiol
 
2012
;
2
:
104
Ouwerkerk
JP
,
van der Ark
KC
,
Davids
M
et al
.
Adaptation of Akkermansia muciniphila to the oxic-anoxic interface of the mucus layer
.
Appl Environ Microb
 
2016
;
82
:
6983
93
Paine
RT.
.
Food web complexity and species diversity
.
Am Nat
 
1966
;
100
:
65
75
Paine
RT
,
Tegner
MJ
,
Johnson
EA
.
Compounded perturbations yield ecological surprises
.
Ecosystems
 
1998
;
1
:
535
45
Pepper
JW
,
Rosenfeld
S
.
The emerging medical ecology of the human gut microbiome
.
Trends Ecol Evol
 
2012
;
27
:
381
4
Pianka
ER.
.
On r-and K-selection
.
Am Nat
 
1970
;
104
:
592
7
Png
CW
,
Lindén
SK
,
Gilshenan
KS
et al
.
Mucolytic bacteria with increased prevalence in IBD mucosa augment in vitro utilization of mucin by other bacteria
.
Am J Gastroenterol
 
2010
;
105
:
2420
8
Qin
J
,
Li
R
,
Raes
J
et al
.
A human gut microbial gene catalogue established by metagenomic sequencing
.
Nature
 
2010
;
464
:
59
65
Rajilić-Stojanović
M
,
de Vos
WM
.
The first 1000 cultured species of the human gastrointestinal microbiota
.
FEMS Microbiol Rev
 
2014
;
38
:
996
1047
Rajilić-Stojanović
M
,
Heilig
HG
,
Molenaar
D
et al
.
Development and application of the human intestinal tract chip, a phylogenetic microarray: analysis of universally conserved phylotypes in the abundant microbiota of young and elderly adults
.
Environ Microbiol
 
2009
;
11
:
1736
51
Rajilić-Stojanović
M
,
Heilig
HG
,
Tims
S
et al
.
Long‐term monitoring of the human intestinal microbiota composition
.
Environ Microbiol
 
2013
;
15
:
1146
59
Raymond
F
,
Ouameur
AA
,
Déraspe
M
et al
.
The initial state of the human gut microbiome determines its reshaping by antibiotics
.
ISME J
 
2015
;
10
:
707
20
Reyes
A
,
Blanton
LV
,
Cao
S
et al
.
Gut DNA viromes of Malawian twins discordant for severe acute malnutrition
.
P Natl Acad Sci USA
 
2015
;
112
:
11941
6
Reyes
A
,
Haynes
M
,
Hanson
N
et al
.
Viruses in the faecal microbiota of monozygotic twins and their mothers
.
Nature
 
2010
;
466
:
334
8
Roh
SW
,
Abell
GC
,
Kim
K-H
et al
.
Comparing microarrays and next-generation sequencing technologies for microbial ecology research
.
Trends Biotechnol
 
2010
;
28
:
291
9
Rosenfeld
S.
.
Why do high-dimensional networks seem to be stable?-A reflection on stochasticity of dynamically unstable nonlinear systems
Gauges
R.
et al
(ed.)
Fifth Workshop on Computation of Biochemical Pathways and Genetic Networks
 ,
University of Heidelberg
,
2008
,
101
12
Rowan
F
,
Docherty
NG
,
Murphy
M
et al
.
Desulfovibrio bacterial species are increased in ulcerative colitis
.
Dis Colon Rectum
 
2010
;
53
:
1530
6
Salazar
N
,
López
P
,
Valdés
L
et al
.
Microbial targets for the development of functional foods accordingly with nutritional and immune parameters altered in the elderly
.
J Am Coll Nutr
 
2013
;
32
:
399
406
Salonen
A
,
Lahti
L
,
Salojärvi
J
et al
.
Impact of diet and individual variation on intestinal microbiota composition and fermentation products in obese men
.
ISME J
 
2014
;
8
:
2218
30
Salonen
A
,
Nikkilä
J
,
Jalanka-Tuovinen
J
et al
.
Comparative analysis of fecal DNA extraction methods with phylogenetic microarray: effective recovery of bacterial and archaeal DNA using mechanical cell lysis
.
J Microbiol Methods
 
2010
;
81
:
127
34
Salonen
A
,
Salojärvi
J
,
Lahti
L
et al
.
The adult intestinal core microbiota is determined by analysis depth and health status
.
Clin Microbiol Infect
 
2012
;
18
:
16
20
Sansonetti
P.
.
To be or not to be a pathogen: that is the mucosally relevant question
.
Mucosal Immunol
 
2011
;
4
:
8
14
Santiago
A
,
Panda
S
,
Mengels
G
et al
.
Processing faecal samples: a step forward for standards in microbial community analysis
.
BMC Microbiol
 
2014
;
14
:
1
Scheffer
M
,
Carpenter
S
,
Foley
JA
et al
.
Catastrophic shifts in ecosystems
.
Nature
 
2001
;
413
:
591
6
Scholz
M
,
Ward
DV
,
Pasolli
E
et al
.
Strain-level microbial epidemiology and population genomics from shotgun metagenomics
.
Nat Methods
 
2016
;
13
:
435
8
Scott
KP
,
Martin
JC
,
Duncan
SH
et al
.
Prebiotic stimulation of human colonic butyrate-producing bacteria and bifidobacteria, in vitro
.
FEMS Microbiol Ecol
 
2014
;
87
:
30
40
Sekirov
I
,
Russell
SL
,
Antunes
LCM
et al
.
Gut microbiota in health and disease
.
Physiol Rev
 
2010
;
90
:
859
904
Sender
R
,
Fuchs
S
,
Milo
R
.
Are we really vastly outnumbered? Revisiting the ratio of bacterial to host cells in humans
.
Cell
 
2016
;
164
:
337
40
Smits
LP
,
Bouter
KE
,
de Vos
WM
et al
.
Therapeutic potential of fecal microbiota transplantation
.
Gastroenterology
 
2013
;
145
:
946
53
Stein
RR
,
Bucci
V
,
Toussaint
NC
et al
.
Ecological modeling from time-series inference: insight into dynamics and stability of intestinal microbiota
.
PLoS Comput Biol
 
2013
;
9
:
e1003388
Sunagawa
S
,
Mende
DR
,
Zeller
G
et al
.
Metagenomic species profiling using universal phylogenetic marker genes
.
Nat Methods
 
2013
;
10
:
1196
9
Tailford
LE
,
Crost
EH
,
Kavanaugh
D
et al
.
Mucin glycan foraging in the human gut microbiome
.
Front Genet
 
2015
;
6
:
81
Tap
J
,
Furet
JP
,
Bensaada
M
et al
.
Gut microbiota richness promotes its stability upon increased dietary fibre intake in healthy adults
.
Environ Microbiol
 
2015
;
17
:
4954
64
Tap
J
,
Mondot
S
,
Levenez
F
et al
.
Towards the human intestinal microbiota phylogenetic core
.
Environ Microbiol
 
2009
;
11
:
2574
84
Thaiss
CA
,
Zeevi
D
,
Levy
M
et al
.
Transkingdom control of microbiota diurnal oscillations promotes metabolic homeostasis
.
Cell
 
2014
;
159
:
514
29
Tien
M-T
,
Girardin
SE
,
Regnault
B
et al
.
Anti-inflammatory effect of Lactobacillus casei on Shigella-infected human intestinal epithelial cells
.
J Immunol
 
2006
;
176
:
1228
37
Tottey
W
,
Denonfoux
J
,
Jaziri
F
et al
.
The human gut chip “HuGChip”, an explorative phylogenetic microarray for determining gut microbiome diversity at family level
.
PLoS One
 
2013
;
8
:
e62544
Trosvik
P
,
Muinck
EJ
.
Ecology of bacteria in the human gastrointestinal tract—identification of keystone and foundation taxa
.
Microbiome
 
2015
;
3
:
1
Turnbaugh
PJ
,
Hamady
M
,
Yatsunenko
T
et al
.
A core gut microbiome in obese and lean twins
.
Nature
 
2009
;
457
:
480
4
Turroni
F
,
Ribbera
A
,
Foroni
E
et al
.
Human gut microbiota and bifidobacteria: from composition to functionality
.
Anton Leeuw
 
2008
;
94
:
35
50
Verberkmoes
NC
,
Russell
AL
,
Shah
M
et al
.
Shotgun metaproteomics of the human distal gut microbiota
.
ISME J
 
2009
;
3
:
179
89
Vieira-Silva
S
,
Falony
G
,
Darzi
Y
et al
.
Species–function relationships shape ecological properties of the human gut microbiome
.
Nat Microbiol
 
2016
;
1
:
16088
Walker
B
,
Holling
CS
,
Carpenter
SR
et al
.
Resilience, adaptability and transformability in social–ecological systems
.
Ecol Soc
 
2004
;
9
:
5
Waller
AS
,
Yamada
T
,
Kristensen
DM
et al
.
Classification and quantification of bacteriophage taxa in human gut metagenomes
.
ISME J
 
2014
;
8
:
1391
402
Wang
M
,
Molin
G
,
Ahrné
S
et al
.
High proportions of proinflammatory bacteria on the colonic mucosa in a young patient with ulcerative colitis as revealed by cloning and sequencing of 16S rRNA genes
.
Dig Dis Sci
 
2007
;
52
:
620
7
Weiss
S
,
Van Treuren
W
,
Lozupone
C
et al
.
Correlation detection strategies in microbial data sets vary widely in sensitivity and precision
.
ISME J
 
2016
;
10
:
1669
81
Wu
C-C
,
Johnson
J
,
Moore
W
et al
.
Emended descriptions of Prevotella denticola, Prevotella loescheii, Prevotella veroralis, and Prevotella melaninogenica
.
Int J Syst Evol Micr
 
1992
;
42
:
536
41
Wu
GD
,
Chen
J
,
Hoffmann
C
et al
.
Linking long-term dietary patterns with gut microbial enterotypes
.
Science
 
2011
;
334
:
105
8
Yatsunenko
T
,
Rey
FE
,
Manary
MJ
et al
.
Human gut microbiome viewed across age and geography
.
Nature
 
2012
;
486
:
222
7
Ze
X
,
Duncan
SH
,
Louis
P
et al
.
Ruminococcus bromii is a keystone species for the degradation of resistant starch in the human colon
.
ISME J
 
2012
;
6
:
1535
43
Ze
X
,
Le Mougen
F
,
Duncan
SH
et al
.
Some are more equal than others: the role of “keystone” species in the degradation of recalcitrant substrates
.
Gut Microbes
 
2013
;
4
:
236
40
Zeevi
D
,
Korem
T
,
Zmora
N
et al
.
Personalized nutrition by prediction of glycemic responses
.
Cell
 
2015
;
163
:
1079
94
Zhang
J
,
Guo
Z
,
Lim
AAQ
et al
.
Mongolians core gut microbiota and its correlation with seasonal dietary changes
.
Sci Rep
 
2014
;
4
, DOI:
Zhang
J
,
Guo
Z
,
Xue
Z
et al
.
A phylo-functional core of gut microbiota in healthy young Chinese cohorts across lifestyles, geography and ethnicities
.
ISME J
 
2015
;
9
:
1979
90
Zijlmans
MA
,
Korpela
K
,
Riksen-Walraven
JM
et al
.
Maternal prenatal stress is associated with the infant intestinal microbiota
.
Psychoneuroendocrinology
 
2015
;
53
:
233
45
Zoetendal
E
,
Rajilić-Stojanović
M
,
De Vos
W
.
High-throughput diversity and functionality analysis of the gastrointestinal tract microbiota
.
Gut
 
2008
;
57
:
1605
15
Zoetendal
EG
,
de Vos
WM
.
Effect of diet on the intestinal microbiota and its activity
.
Curr Opin Gastroenterol
 
2014
;
30
:
189
95
Zoetendal
EG
,
Raes
J
,
van den Bogert
B
et al
.
The human small intestinal microbiota is driven by rapid uptake and conversion of simple carbohydrates
.
ISME J
 
2012
;
6
:
1415
26
Zoetendal
EG
,
von Wright
A
,
Vilpponen-Salmela
T
et al
.
Mucosa-associated bacteria in the human gastrointestinal tract are uniformly distributed along the colon and differ from the community recovered from feces
.
Appl Environ Microb
 
2002
;
68
:
3401
7
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com