Download Review - Jeffrey Gordon

Leading Edge
Review
Ecological and Evolutionary Forces
Shaping Microbial Diversity
in the Human Intestine
Ruth E. Ley,1 Daniel A. Peterson,1 and Jeffrey I. Gordon1,*
1
Center for Genome Sciences, Washington University School of Medicine, St. Louis, MO 63108, USA
*Contact: [email protected]
DOI 10.1016/j.cell.2006.02.017
The human gut is populated with as many as 100 trillion cells, whose collective genome, the
microbiome, is a reflection of evolutionary selection pressures acting at the level of the host
and at the level of the microbial cell. The ecological rules that govern the shape of microbial
diversity in the gut apply to mutualists and pathogens alike.
Our intestinal tract is a nutrient-rich environment packed
with up to 100 trillion (1014) microbes. The vast majority reside in our colon where densities approach 1011–1012
cells/ml, the highest recorded for any microbial habitat
(Whitman et al., 1998). Today, there are 6.5 billion humans
living on Earth. Together, we represent a gut reservoir of
1023–1024 microbial cells. This number is just five orders
of magnitude less than the world’s oceans, which contain
an estimated 1029 cells (Whitman et al., 1998). Therefore,
together with other mammals, the human gut constitutes
a substantial microbial habitat in our biosphere.
Because we are born germ free, the microbes that populate our intestinal tract must come from the outside. Microbial diversity on our planet is vast: although 55 divisions (deep evolutionary lineages) of Bacteria and 13
divisions of Archaea have been described to date (DeLong
and Pace, 2001; Rappe and Giovannoni, 2003; Rondon
et al., 1999), much diversity remains unexplored. The intestine is remarkable for its exclusivity: selection pressures whittle down the microbial diversity of the outside
world so that the gut microbiota in adults is dominated
by members of just two divisions of bacteria—the Bacteroidetes and Firmicutes—and one member of Archaea,
Methanobrevibacter smithii (Bäckhed et al., 2005; Eckburg et al., 2005). The gut microbial community presumably has strict requirements for membership: an arsenal
of enzymes to utilize available nutrients; cell-surface molecular paraphernalia to attach to the ‘‘right’’ habitat,
evade bacteriophages, and appease a reaction-ready immune system; genetic gadgetry for mutability to stay well
adapted; the ability to grow rapidly to avoid washout; and
the stress resistance needed when making the jump to
other hosts via a largely dry and toxic ‘‘ex-host’’ environment. Written into the microbiome is a survival guide for
microbes who wish to live in the gut environment.
Lederberg (2000) has emphasized the importance of
having a broad ecologic view of our relationships with microbes. In this view, we are seen as superorganisms composed of an amalgam of both microbial and H. sapiens
cells, where the survival of microbe and human is interde-
pendent. The gut is a natural laboratory for studying the
‘‘microevolution’’ of humans. Experimental and computational tools are now in hand to comprehensively define diversity in our gut microbiota, decipher the gene content
of the microbiome, and to explore the metabolome encrypted by this collection of microbial genes—a collection
that is estimated to exceed the number of our own human
genes by at least two orders of magnitude. The results of
this exploration promise to reveal the operating principles
that underlie beneficial host-microbial and microbial-microbial relationships, as well as new ecologic perspectives
and ‘‘ecogenomic’’ views of how pathogens arise and
function within our indigenous microbial communities.
In this review, we argue that the microbial diversity of
the human gut is the result of coevolution between microbial communities and their hosts. We suggest that the peculiar structure of microbial diversity in the human gut resulted from natural selection operating at two levels. Host
level, ‘‘top-down’’ selection on the community favors stable societies with a high degree of functional redundancy.
An opposing ‘‘bottom-up’’ force is selection pressure driving microbial cells to become functionally specialized (Figure 1). In addition to the selection pressures shaping microbes in the gut, we discuss factors that constrain
diversity. Finally, we present an ecologic view of pathogenic relationships: the challenges potential pathogens
face when encountering a microbiota where there is functional redundancy with virtually all niches and habitats
filled and the idea that microbial community structure
should be considered as a factor that can influence predisposition to specific diseases in certain host contexts.
Who Passes the Gauntlet? Patterns of Microbial
Diversity in the Gut
Definitions
Bacteria and Archaea in the gut multiply by binary fission.
They can ‘‘differentiate’’ genetically by a number of mechanisms, including lateral gene transfer via phage- and mobile element-mediated insertions, conjugal transfer, or uptake of naked extracellular DNA. Alterations to vertically
Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc. 837
Figure 1. Schematic Diagram of the Selection Pressures Operating at Different Levels in the Human-Microbial Hierarchy
Brown arrows indicate selection pressures and point to the unit under
selection (red). Black arrows indicate emergent properties of one level
that affect higher levels in the hierarchy. According to hierarchy theory,
higher levels place constraints on possible organizational solutions
at lower levels. Ecologic principles predict that host-driven (‘‘topdown’’) selection for functional redundancy would result in a community composed of widely divergent microbial lineages (divisions) whose
genomes contain functionally similar suites of genes. Another prediction is the widespread occurrence of, and abundant mechanisms for,
lateral gene transfer. In contrast, competition between members of
the microbiota would exert ‘‘bottom-up’’ selection pressure that results in specialized genomes with functionally distinct suites of genes
(metabolic traits). Once established, these lineage-specific traits can
be maintained by barriers to homologous recombination (Majewski
et al., 2000).
inherited genomes can come from mutations (Giraud
et al., 2001b) and chromosomal rearrangements. Within
this mutable context, the term ‘‘individual’’ is misleading
(single cells have been referred to as ‘‘dividuals’’ [Koestler,
1967]), and the term ‘‘species’’ is ambiguous. We use
‘‘species’’ here to refer to named types (e.g., Eubacterium
838 Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc.
rectale), and ‘‘phylotype’’ (phylogenetic type) to refer to
clusters of related 16S rRNA gene sequences characterized by levels of pairwise sequence identity (R97% ID is
the commonly used threshold used for a ‘‘species’’).
The gut ecosystem is very dynamic. Within a given
intestinal habitat (address), some microbial members
function as entrenched ‘‘residents’’ (autochthonous components), while others act more like hitchhikers (allochthonous members) from ingested food, water, and various
components of our environment. Relationships between
members of the microbiota and humans are frequently described as commensal (one partner benefits while the
other seems unaffected) rather than mutualistic (both partners derive benefit). In this review, we address the evolution of the entrenched residents. We favor the term mutualist for the permanent residents of the gut based on the
view that selection for stable communities favors beneficial relationships.
Low Levels of Deep Diversity
The most comprehensive and least biased enumerations
of microbial diversity within the mammalian gut have
come from sequencing 16S rRNA genes. These sequences are obtained directly from DNA extracted from
gut mucosal biopsies or feces using PCR primers targeted
to broad phylogenetic groups. The largest data sets consist of 13,335 16S rRNA sequences from three healthy
adult humans (Eckburg et al., 2005) and 5,088 sequences
from 19 adult mice belonging to the commonly used
C57Bl/6 inbred strain (Ley et al., 2005). In the study of
the human intestinal microbiota, bacterial and archaeal
16S rRNA sequences were derived from biopsies taken
from six regions of the colon and one stool sample from
each individual. In the mouse study, 16S rRNA bacterial
sequences were obtained from the cecal contents. Neither study targeted Eukarya, although recent work in
mice suggests an unappreciated diversity of fungi (Scupham et al., 2006).
The human intestinal samples contained members of
seven divisions of Bacteria (Firmicutes, Bacteroidetes,
Actinobacteria, Fusobacteria, Proteobacteria, Verrucomicrobia, Cyanobacteria), which, together with divisions described in previous studies (Spirochaeates, VadinBE97
[Bäckhed et al., 2005]), brought the total number to nine.
The mouse survey showed a very similar number and
set of divisions, although Fusobacteria was not detected
and TM7 was found. The Firmicutes and the Bacteroidetes dominated, together accounting for >98% of all
16S rRNA sequences in each mammalian host.
Radiation of a Few Lineages
The human and mouse intestine contains relatively few
bacterial and archaeal divisions compared with other microbial habitats (Bäckhed et al., 2005; Figure 2A) even
though the number of deposited sequences is much
higher for the former. For example, in soil, where plant
polysaccharides are also degraded to simple carbon
compounds, there are at least 20 bacterial divisions (Dunbar et al., 2002). The dearth of divisions in the gut contrasts with the multiplicity of phylogenetically shallow
Figure 2. Comparison of Microbial Diversity in the Human Colon, Mouse Cecum, Ocean, and Soil
(A) Percent representation of divisions in each environment. For the human gut, 16S rRNA sequence data (n = 11,831) were obtained from the
Arb alignment supplied by Eckburg et al. (2005). We used our available
Arb alignment for mouse data (n = 5088; Ley et al., 2005); the ocean
data set (n = 1082) alignment was supplied by Acinas et al. (2004);
the zebrafish gut data set (n = 1387) was obtained from adult conventionally raised adults (Rawls et al., 2004; J.F. Rawls, R.E.L., and J.I.G.,
unpublished data). Soil data sets from two independent studies were
combined (n = 1379; Axelrood et al., 2002; Tringe et al., 2005). Stromatolite data (n = 321; Papineau et al., 2005) were obtained from GenBank. Stromatolite, soil, ocean, and fish sequences were aligned using
the nast online Arb alignment tool (http://greengenes.lbl.gov/cgi-bin/
nph-NAST_align.cgi).
(B) Phylogenetic architecture of the microbial communities shown in
(A). For each habitat, the number of phylotypes per 100 16S rRNA
gene sequences is shown for differing thresholds of 16S rRNA gene
pairwise sequence identity (%ID). The blue bar highlights the phylotypes with R97% ID, the cutoff used to designate species and subspecies-level taxa. Distance matrices were calculated using the entire
16S rRNA sequence (including hypervariable regions) with Olsen correction in Arb, followed by OTU determination using a furthest-neighbor method in DOTUR (Schloss and Handelsman, 2005). For further
details about methods, see online Supplemental Data. Note that compared to the soil and ocean, the gut shows the steepest decline in phylotype abundance at %IDs %97%. The shape of the curve reflects the
structure of diversity. Sample size affects the relative position of the
curves (phylotype abundance estimates based on the resampling of
equal numbers of sequences for all environments show the same
trends; data not shown).
lineages within its divisions. Over half (7,555) of the 13,335
sequences in the human colonic data set were sampled
only once and thus can be considered to represent strains
(subspecies). The high number of strains sort into a dramatically lower number of ‘‘species-level’’ phylotypes:
395 when the threshold for defining membership in a given
species is set at R99% 16S rRNA sequence identity (Eckburg et al., 2005). (Note that the estimate of 395 species at
99% ID for the human colon is equivalent to the number of
phylotypes at 97% ID [generally used to delimit species]
when whole 16S rRNA sequences are used [see Supplemental Data available with this article online]. The estimate
of 395 species is conservative since it was obtained after
excluding hypervariable regions that comprise 10% of
the length of 16S rRNA genes.)
Mucosal communities sampled at different points along
the colon were similar within each of the three human
hosts (Eckburg et al., 2005; Zoetendal et al., 2002), and
their composition overlapped with the fecal microbiota
(the colon has incomplete peristalsis that permits backflow and mixing). Interpersonal differences in microbial
community composition are illustrated in Figure 3: in several places in the 16S rRNA tree, regions representing
phylogenetic fans were unique to each person.
This pattern of high levels of strain variation but far fewer
intermediate and deep lineages was also observed in the
mouse gut microbiota (Ley et al., 2005). This type of extreme fan-like phylogenetic architecture may be a signature feature of the gut ecosystem. Shallow phylotype
fans have been described in other natural habitats (e.g.,
the ocean [Acinas et al., 2004]), but these habitats appear
to harbor a greater number of intermediate-level phylogenetic groups than the gut (Figure 2B). The pattern of phylotype fans in the gut recalls the patterns in classic adaptive radiations (Schuter, 2000), such as the radiation of
terrestrial snail species of French Polynesia (Garrett,
1884) where a few successful early colonists gave rise to
a variety of descendants. Shallow, wide radiations result
from extreme selection pressure followed by détente. In
the case of the snails, the selection pressure came in the
form of a wide ocean to cross, but once there, the early
snails filled an island’s worth of empty niches (professions)
across a range of habitats (addresses). Similarly, the phylogenetic architecture of the gut could have resulted from
the diversification of a discrete limited initial community (a
population bottleneck) into strains. In addition, the phylogenetic ‘‘shallowness’’ of the intestinal community reflects
the relatively short time that the mammalian gut has existed as a habitat (100 million years for mammals with
placentas [Murphy et al., 2001], versus >3.85 billion years
for the ocean).
How Do They Get There? 100 Trillion Family
Heirlooms
Pathogens that use an oral-fecal route for their transmission (e.g., members of the Proteobacteria such as Vibrio
cholera) can exploit environmental reservoirs outside of
their hosts to proliferate. However, members of the
Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc. 839
Figure 3. Variation in Bacterial Diversity within the Colonic
Microbiotas of Three Healthy Humans
These phylogenetic trees are based on the 16S rRNA bacterial sequence data set (n = 11,831) and alignment of Eckburg et al. (2005)
(see Supplemental Data). (A), (B), and (C) show the portion of the whole
tree that is contributed by individuals A, B, and C from the study. Each
tree represents the whole data set. Within a tree, colored portions (red,
blue, yellow) represent diversity unique to the individual, white
branches indicate portions of the tree that are shared with another individual, and black branches indicate diversity that was not encountered in a given individual. Individual B (blue) had the highest levels
of bacterial diversity, and archaeal (M. smithii) sequences were recovered from all biopsy samples and stool. Individual C (yellow) had intermediate bacterial diversity and archaeal sequences in stool only. Individual A (red) the lowest bacterial diversity and no archaeal sequences
840 Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc.
Firmicutes and Bacteroidetes detected in the human gut
do not appear to grow outside of their host and likely
rely on the close contact of parents and offspring for transmission. One testable prediction of this parent-to-offspring transmission hypothesis is that microbial communities will be similar in members of a given family. In
experiments with C57Bl/6 mice, we showed that animals
inherited their microbial communities from their mothers
(Ley et al., 2005), using the recently developed UniFrac
metric (Lozupone and Knight, 2005). Related mice shared
similar bacterial communities: the effect was evident
across multiple generations: the mothers that were sisters
shared microbiotas with each other as well as with their
offspring. These findings demonstrate that a microbiota
can be inherited vertically from mothers and is stable
enough over time that kinship relationships are reflected
in community composition.
Culture-based studies in humans indicate that babies
acquire their initial microbiota from the vagina and feces
of their mothers (Mandar and Mikelsaar, 1996). Babies delivered by caesarian section have an altered colonization
pattern relative to their vaginally-delivered counterparts
(Gronlund et al., 1999). Host genotype represents a confounding factor when attempting to show vertical transmission of microbes from parents to offspring. In a study
of adult monozygotic twins, their siblings, and marital partners, Zoetendal et al. (2001), used DNA fingerprinting
methods to show greater similarities between the gut microbial communities of monozygotic twins than between
monozygotic twins and their unrelated marital partners.
The similarities between the monozygotic twins’ gut communities were interpreted as an effect of genotype on microbial diversity. An alternative explanation is that the observed similarities between twins’ communities were due
to their colonization from a shared mother. Indeed, the
study also included dizygotic twins and siblings: their microbiota were as similar to each other’s as the microbiota
of monozygotic twins, despite a lower level of host genetic
relatedness (Zoetendal et al., 2001). Similarly, in the only
small subunit (SSU) rRNA sequence-based study of kinship effects on the normal human microbiota, Frank
et al. (2003) showed that marital partners had different
communities in their ear canals but that the same bacterial species was dominant within members of a given
family.
In our study of mice, genotype was shown to act on the
relative abundance of groups whose specific composition
was determined by kinship. The mothers were heterozygotes for a mutation (ob) in the leptin gene. The offspring
were produced from matings of ob/+ males and females,
(Eckburg et al., 2005). Division names are indicated. The scale bar indicates degree of sequence divergence. Note that that 16S rRNA sequenced-based surveys do not always recover identical communities,
even from subsamples of the same material; rather, statistically related
communities may be recovered. Therefore, while the exact sequences
may not be resampled, related sequences from the same phylogenetic
fan may be retrieved.
so a subset was obese (ob/ob), while their wild-type (+/+)
and heterozygous ob/+ siblings were lean. The composition of the cecal microbiota (i.e., phylotypes present irrespective of their abundance) was dictated by the mother
regardless of their offspring’s genotype. However, there
was a division-wide 50% reduction in the abundance of
Bacteroidetes in ob/ob animals and a proportional division-wide increase in the Firmicutes compared to their
lean +/+ or ob/+ littermates (Ley et al., 2005).
This mouse study allowed the effects of genotype to be
decoupled from effects of kinship because the experimental conditions could be controlled and because analytical
methods are now available that take into account composition with or without abundance data (Lozupone and
Knight, 2005). These approaches can now be used, in
concert, to address some basic unanswered questions
about human microbial ecology (see Table 1). For example, we do not understand how resilient (resistant) an individual’s microbiota is to potential colonists/pathogens
from contaminated water, fecal material aerosolized by
modern toilets (Barker and Jones, 2005), or cohabitation
with an unrelated partner (spouse), or to disturbances
such as antibiotic treatment, infection with parasites, or
drastic shifts in diet. Is mother/caregiver-to-infant transmission a lifelong shaper of an individual’s microbiota?
Is there is a window of opportunity during infancy when
the gut is particularly ‘‘open’’ or susceptible to colonization or, for that matter, to an enduring perturbation of its
microbial ecology following antibiotic therapy? Does the
early-acquired microbiota persist into adulthood, having
gained a foothold by initially ‘‘educating’’ the immune system to help retain a particular suite of organisms? What
are the effects of genotype or health status? Controlled
studies in mice could tease apart the effects of age, disturbance, exposure, and successional order on the composition and resilience of microbial communities. Such studies should inform and direct sampling designs aimed at
answering similar questions in humans. These basic studies will provide more than interesting ‘‘natural history’’: if
people acquire their microbiome and its gene content
from their families, it represents another form of genetic
heritability—one that has the potential to be reprogrammed.
The Evolutionary Gauntlet: Selection Pressures
for Form and Function
The method of gut colonization is one factor that influences diversity. Other factors include the physical and
chemical environment and the selection pressures on
the host. These selection forces (Figure 1), both direct
and indirect, are what mutualistic and pathogenic microbes alike must contend with to be successful.
Chemistry
The gut offers fewer chemical niches (professions) to its
microbiota than a habitat such as soil. Gut microbes can
derive energy from transferring electrons from organic
carbon either to organic carbon (fermentation), to inorganic carbon (methanogenesis), or to sulfate (sulfate re-
duction), all of which are available in the gut at levels
that can sustain populations. In addition, H2 is undoubtedly an important intermicrobe electron transfer shuttle.
Other microbial niches, such as nitrification or, photosynthesis, are not viable options in the gut for lack of substrates and light, and therefore entire functional guilds
common in other microbial habitats are excluded. Further
reduction in the number of niches may result from the peristalsis and churning of the gut. Carbon resource heterogeneity or spatial partitioning of different organic substrates
correlates with diversity patterns in soil (Zhou et al., 2002).
However, the mixing of gut contents is a homogenizing
force that would reduce niche breadth.
Some of the metabolic niches in the gut can be loosely
mapped onto its major lineages. Only Archaea are known
produce methane. Sulfate reduction is carried out by Delta
Proteobacteria (e.g., Desulfovibrio spp.) and one clade
within the Firmicutes (Desulfotomata spp.). Fermentation,
however, is a phylogenetically widely held skill, and the
principal energy pathway for members of the Bacteroidetes and Firmicutes (and thus the dominant energy-producing pathway for the microbiota).
In principle, microbes from any phylogenetic division
could acquire the necessary genes to survive as a fermenter in the gut: for example, key glycoside hydrolases that
liberate sugars from dietary polysaccharides could be acquired by lateral gene transfer (LGT). Yet the dominance of
the Firmicutes and Bacteroidetes shows that acquisition
of a gene suite required to colonize the gut has been limited to members of a few phylogenetic divisions. Microbes
are more likely to swap genes with microbes from the
same environment (Beiko et al., 2005): therefore, dominant groups would monopolize the gene swapping market
and newcomers would be rare. However, LGT is a homogenizing evolutionary force (Woese, 2002). If members
of the Bacteroidetes and Firmicutes have continually
swapped genes over their evolution, their genomic contents would be drawn from a common pool of gut-adapted
genes. Whole genome data are needed to make such interdivision comparisons, but the known differences in
GC content between members of the two divisions is evidence that LGT has not homogenized their genomes.
The codominance of the Bacteroidetes and Firmicutes
likely stems from their distinct and complementary metabolic roles within the community. Moreover, these coevolved, cooperative roles were likely traits of early colonists in mammalian gut evolution and are encoded by
a genomic repertoire that is too large and eclectic to be
acquired and assembled solely by stochastic processes
such as LGT.
Cooperation
It remains unclear how much traditional ecological theory
can be strictly applied to the microbial world. However,
concepts borrowed from ecology can frame our view of
microbial interactions in the gut. Ecological theory teaches
that head-to-head competition for limiting resources results competitive exclusion—the loser goes extinct and
the community becomes a monoculture of the winner
Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc. 841
Table 1. Key Questions and Future Studies
Question
Approaches
1. How is the gut microbiota transmitted
in humans?
If the transmission is primarily parent-child, we should detect kinship
patterns in microbial communities of the hosts. One caveat is that modern
standards of hygiene may have altered transmission mechanisms. Parallel
studies of families in developed areas where exposure to water and food
contaminated with fecal microbes is minimal and families in
underdeveloped areas where such exposure is common are required
elucidate patterns of transmission.
2. How distinctly human is our gut microbiota?
If the diversity of the microbiota coevolved with mammalian host species,
the relationships of gut communities from different mammals should mirror
the evolutionary relationships of the mammals. Furthermore, the microbial
communities of distantly related host species should be less functionally
interchangeable than the communities of closely related hosts.
3. How functionally redundant are the members
of the human gut microbiota?
Natural selection at the host level will favor functionally stable communities.
This ‘‘top-down’’ pressure should produce a community with functional
redundancy encoded in its genomes. An opposing ‘‘bottom-up,’’ cell-level
selection pressure would favor niche specialization to avoid competition.
The sequencing of a large collection of microbial genomes from the gut,
from closely and distantly related bacteria and the building-up of
‘‘intentional communities’’ within gnotobiotic mouse models should help
answer fundamental questions about how the gut community is assembled
and operates.
4. Are members of our gut microbiota more likely
to have exchanged genes with each other than
with microbes from other habitats?
The lateral exchange of genes is a fundamental process of microbial
genome evolution, one that can confer new characteristics to organisms,
allowing them to jump from peak to peak within fitness landscapes.
However, microbes are restricted in the gene space they can sample by the
suite of potential donors that share their habitat. This would have the effect
of homogenizing the genome pool over time within a habitat. The gut is
a natural laboratory for studying genome evolution within spatially distinct
habitat islands. Moreover, these islands are distributed at a host scale
across the globe. The gut is also an alluring model for studies of the
biogeographic effects of habitat scale and fractionation on genome
evolution in both mutualists and pathogens.
5. What controls the rate of lateral gene transfer
among microbes in the gut?
If lateral gene transfer (LGT) homogenizes genomes over time, genes in any
given genome will be drawn randomly from a habitat’s gene pool. The
degree to which microbiomes differ between habitats will be a reflection of
the degree of mixing between habitats and the rates of LGT within a habitat.
Experiments in gnotobiotic models may help quantify the rate of LGT among
prominent members of the gut microbiota.
6. What is the impact and role of Archaea?
The Archaea within the gut appear to follow a different set of rules than the
Bacteria. Archaeal diversity appears to be low, although this needs more
verification (Rieu-Lesme et al., 2005). By providing the final step in energy
extraction from degradation of organic compounds, Archaea can alter the
thermodynamics of the whole system, with profound consequences for the
host. Are Archaea the power brokers of the gut? Are they pillars of the
community and a target for therapeutic manipulation?
7. Does the immune system respond to gene
suites/gene expression rather than to species?
The role of the immune system in shaping the microbiota remains largely
enigmatic. The traditional view of bacterial species as entities with discrete
phenotypes that the immune system responds to oversimplifies the
problem. We need to rethink how an immune system can aid in the retention
of a functional suite of microbes whose composition may change over the
lifetime of the host. The immune system and the microbiota should be
viewed as a coevolved system: a driving force for evolution of the immune
system is the need to accommodate (ongoing) diversity in a host microbiota;
this in turn allows the host to accommodate environmental (including food)
antigens and possibly self-antigens.
842 Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc.
Table 1. Continued
Question
Approaches
8. How does genotype, diet, health status
influence the microbiota?
How does the microbiota influence our health, and can we reshape our
microbial communities microbiota as a therapeutic strategy? Are we dealt
an unalterable microbial hand during infancy, or is the community continually
turning over as new colonists arrive and establish? Controlled studies using
mice with different genetic backgrounds and germ-free mice inoculated with
a defined microbiota will identify dominant factors that shape the microbiota,
the elements within the microbiota that impact host health, and the homeostatic
controls on diversity. Results from mouse-based studies should provide
a framework for conducting case-controlled studies of humans.
(Putman, 1994). Competition is avoided if organisms partition resources and/or cooperate or, alternatively, if another factor, such as the host, limits their growth. Resource partitioning is initiated when an organism finds
an empty niche and uses a resource that is not already
tapped. E. coli populations have been shown to differentiate into diverse niches when introduced into germ-free
mouse guts (Giraud et al., 2001a). Genetic diversification
can enable exploitation of new niches by increased mutation rates within a clonal population (Bjedov et al., 2003)
and in a mixed community by LGT (Gogarten et al.,
2002). Diversification can create positive feedback loops:
diversity begets diversity by creating new niches. This
process has been invoked to explain high diversity among
plants and animals on islands (Emerson and Kolm, 2005).
At the microbial level, diversity may expand niche space to
a degree that is not possible in plant and animal communities. Microbes produce chemical food webs where the
product of one microbe becomes the substrate for another, and the removal of waste products (e.g., H2) can enhance the thermodynamic yield of the first (Thauer et al.,
1977). Microbial chemical webs have the potential to be
inconceivably complex and changeable as cells adjust
their transcriptomes and metabolomes to variations in
substrate availability and thermodynamic gradients. The
sum of activities of members of the community is an emergent property of the community with consequences for
host fitness.
Selection for Emergent Functionality: Pressure on
Hosts
Each microbial cell is under selection pressure in the gut:
this pressure acts on the cell’s phenotype and results in
fixation of genes in the genomes of its daughters. Each
cell is also part of a community that provides critical services (e.g., nutrient extraction, invasion resistance) for
the higher level in the hierarchy—namely the host. Hierarchy theory states that the higher levels impose constraints
on the lower levels (Koestler, 1967). Emergent properties
of the gut microbial community impact host fitness and
consequently the future availability of a gut habitat. Therefore, irrespective of the success of particular suites of bacteria in the gut, if the host does not benefit from their collective behavior, the entire group can be selected against
when the host dies without leaving many offspring. Ecological theory suggests that for group selection to take
place, the community must occasionally be transferred
to a new habitat (Wilson, 1975): this is what happens
when new hosts are born and colonized. When the community is transferred, cheaters that benefit from the community without contributing to it are purged (they are present at lower abundance, and the probability that they will
be transferred is low; Travisano and Velicer, 2004). Furthermore, transmission to new hosts changes the scale
at which competition between members of the microbiota
occurs from local (within host) to global (between hosts):
this type of change would favor cooperation and altruism
over cheating (Griffin et al., 2004).
The diversity of the gut microbiota today may reflect
‘‘the ghost of group selection past.’’ However, the host
and his/her gut microbial community constitute a rare biological system: one where group selection could be demonstrated empirically. To test empirically if an emergent
trait drives a phenotype that can be selected against at
the host level, experiments could be designed in which
an intentional community of microbes is seeded into a previously germ-free population of mice and community-level
traits (such as the ability to degrade a specific compound
requiring a multispecies consortium) used as the basis for
artificial selection. Group selection based on emergent
properties of the community is standard in engineered
systems where a specific product is desired. Functional
stability is a property that is always selected for in
waste-water treatment systems (by the engineer) and
one that may play a critical role in promoting host fitness:
after all, the gut must digest food and liberate nutrients
and energy in a reliable fashion.
Functional Redundancy and the Reduced Need for
‘‘Keystone Species’’
Our gut microbiota is an efficient and stable natural bioreactor (Sonnenburg et al., 2004): it is resistant and resilient
to chaotic blooms of subpopulations (or pathogens) that
could be disruptive and reduce host fitness (Bäckhed
et al., 2005). Functional redundancy in a microbiota confers
stability (also known as the insurance hypothesis [Yachi
and Loreau, 1999]) that will be selected for at the host
level. Functional redundancy can also obviate the need
for ‘‘keystone species’’ (defined as a species with a central
role in the system whose loss causes a dramatic change in
processes and diversity). If selection pressure at the host
level for a reliable stable community is a dominant shaper
Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc. 843
of diversity in the gut, then the resulting functional redundancy should be written into the microbiome as well as
scripted in host-microbial interactions.
An extreme test of the functional redundancy of the gut
microbiota is replacement of an entire gut microbiota with
just one species. Monoassociation of germ-free mice with
a single species of Bacteroidetes can recapitulate a number, but certainly not all, host phenotypes observed in
‘‘conventionally raised’’ animals (mice that acquire a complete microbiota beginning at birth). Analysis of the finished genome sequences of three members of Bacteroidetes prominently represented in the human colonic
microbiota has revealed a considerable degree of redundancy in their ability to process plant polysaccharides
(see below). It is likely that the phylotype fans observed
in the gut correspond to ‘‘ecotypes’’ (habitat specialists;
Acinas et al., 2004; Palys et al., 1997) with differing levels
of functional redundancy. Different species within the family may be functionally redundant, but to coexist they partition the resource base by expressing species-specific
substrate preferences and/or ‘‘use efficiencies’’ (Sonnenburg et al., 2005). The alternative strategy, exclusive niche
specialization, can result in decreased ability to diversify
further (Buckling et al., 2003) and risk to the host that a keystone species is irreplaceably lost in a selective sweep
(Cohan, 2002) from, for example, attack by bacteriophages that are abundant in the gut (Breitbart et al., 2003).
Although functional redundancy may be a dominant
theme within lineages that occupy the gut, one exceptional organism that appears to be irreversibly adapted
to a niche for which there appears to be few if any alternative players is the archaeon Methanobrevibacter smithii.
M. smithii couples H2 oxidation to CO2 reduction to produce methane (CH4) in one of the least energy-yielding reactions in biology. It competes directly for H2 with sulfatereducing bacteria but is thought to hold a competitive
advantage in vivo (Strocchi et al., 1994). In the comprehensive 16S rRNA sequence-based enumeration study
of the colonic microbiota of three healthy individuals, the
person with the highest levels of bacterial diversity was
the one with the most archaeal (M. smithii) sequences
(Eckburg et al., 2005) (Figure 3). This observation could
be anecdotal, or it could be a hint that archaea have a pronounced effect on the thermodynamics and diversity of
the gut microbiota. Interestingly, in engineered systems
where functionality is constant, populations of bacteria
can fluctuate chaotically, but archaeal populations remain
constant in composition and abundance (Fernandez et al.,
1999; Leclerc et al., 2004). Gnotobiotic mouse experiments are needed in which ‘‘intentional’’ communities of
sequenced bacterial and archaeal species, representing
prominent representatives of the microbiota, are created
in previously germ-free animals in an attempt to circumscribe functional redundancy and thereby examine if
‘‘true’’ keystone species exist for specific biological processes (the term ‘‘gnotobiotic’’ comes from the Greek
‘‘gnosis’’ and ‘‘bios’’ meaning known life, and refers to animals whose living components are defined).
844 Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc.
Direct Selection by the Host: Effects of the Immune
System
Since emergent properties of the microbiota can potentially have dire consequences for the host, the host has
a strong incentive to control the composition of the microbiota directly. The immune system is the host’s first line of
contact with the microbiota in the gut and can be expected to play a role in shaping the microbiota. However,
the microbiota’s functional redundancy and strain variation could render a species-specific immune response ineffective at suppressing a specific microbial activity. Several observations support the view that the immune
system responds to whole suites of bacteria. First, although the microbiota provokes inflammatory bowel disease (IBD; Asseman et al., 2003; Dianda et al., 1997; Mizoguchi et al., 2000; Sadlack et al., 1993; Sellon et al.,
1998), no single species has been consistently linked to
all cases (Duchmann et al., 1999; Sydora et al., 2005). Indeed, T cells cloned from IBD patients (i.e., T cells that
should be monospecific) respond to multiple and divergent bacterial species (Duchmann et al., 1999). Furthermore, toll-like receptors recognize ligands produced by
a variety of mutualists in an interaction that maintains intestinal epithelial homeostasis (Rakoff-Nahoum et al.,
2004). Second, a series of papers from the Kasper laboratory demonstrate that capsular polysaccharide A (PSA)
from Bacteroides fragilis, an organism that comprised
5% of all Bacteroidetes and 2.5% of all bacteria in the enumeration study of the human colonic microbiota described
above, can stimulate a surprisingly large fraction of the
available population of T cells in mice and humans
(Cobb et al., 2004; Mazmanian et al., 2005). Thus, capsula
polysaccharide generated from multiple species could be
responsible for broad-spectrum T cell stimulation. Finally,
16S rRNA sequence-based analysis of the gut microbial
communities of immunodeficient mice has failed to demonstrate marked changes in species composition compared to normal controls (Suzuki et al., 2004), suggesting
that the adaptive immune system is not effective in eliminating a single species but rather modulates the relative
abundance of strains representing groups with shared
functional and/or structural properties.
Pathogenic Species and ‘‘Pathogenic Communities’’
In the classical view of an enteric pathogenic species, the
pathogen is a cheater that benefits from gut microbial
community dynamics while imposing a fitness cost to
the host and to the community. In this view, pathogenic
species are like mutualistic species in that they have
a shared interest in their hosts species’ survival (Lederberg, 2000). Selection on the host results in selection for
pathogens that are not too virulent, and/or a microbial
community that can prevent pathogens from building up
the necessary population density to cause injury (Czuprynski and Balish, 1981; Zachar and Savage, 1979). Theoretical studies suggest that mutualists typically have
a competitive advantage over cheaters (Ferriere et al.,
2002). Rather than competing head-to-head with
mutualists, a successful pathogenic species can use
a high infective dose, specialized organelles for attachment/invasion, and/or enterotoxins to induce secretion
of water so that it can enter or create habitats that are
not occupied by members of the normal microbiota (Nataro and Kaper, 1998). While this behavior results in efficient amplification and return to the environment, it is not
compatible with long-term colonization, due to the demise
of the host or a robust elicited immune response. As a result, virtually all enteropathogenic bacteria have an environmental reservoir and often a definitive host (e.g., poultry for Salmonella).
In addition to the classic pathogenic species, we propose that another kind of pathogenicity exists in the gut:
one in which the whole community is ‘‘pathogenic’’
when its emergent properties contribute to disease. In
a ‘‘pathogenic community,’’ no single microbe is pathogenic alone. Instead, the community assemblage is an environmental risk factor that contributes to a disease state.
A microbial community will be pathogenic within the context of other risk factors, such as host genotype, diet, and
behavior. For instance, the amount of calories available to
the host from food is a value modulated to a significant degree by the gut microbiota (Bäckhed et al., 2005). A microbial community whose energy extraction is very efficient
could constitute a risk factor for obesity in a person with
ready access to food, whereas it might promote health
in a individual with more limited access. We have argued
that a community whose emergent properties decrease
host fitness will be selected against. Diseases such as
obesity do not necessarily reduce host fitness today if fitness is only measured as the number of offspring produced. However, they do impose a cost to both the individual and to society. It is up to us as a human society to
recognize and select against the microbial communities
that may be risk factors for these types of maladies within
individuals as part of preventive medicine.
The Human Gut Microbiome Initiative
Comparing multiple genomes, representing 16S rRNA
phylotypes with different degrees of relatedness would
help answer a number of key questions about the microbiome. Which gene families are widespread among lineages and therefore essential for survival in the gut ecosystem? How much lateral gene transfer occurs between
distant versus close relatives in the densely populated distal gut, and how does this relate to the evolution and functional stability of the microbiota’s metabolome? How have
mutualists evolved, and how are they continuing to evolve
to interact with the immune system? Can features of microbial genome structure and microevolution be used as
biomarkers of health or of susceptibility to specific diseases? To what extent are shared elements in mutualists
represented in pathogenic microbes that face similar ecological pressures?
To begin to answer these questions, we have turned initially to members of the genus Bacteroides. B. thetaiotaomicron comprises 12% of all Bacteroidetes and 6% of all
Bacteria in the 11,831 member human colonic bacterial
16S rRNA data set (Eckburg et al., 2005). B. thetaiotaomicron’s 6.3 Mb genome reflects some of the selective pressures that have defined its habitat and its niche (Xu et al.,
2003). It contains an unusually large ensemble of genes involved in acquiring and metabolizing carbohydrates: this
arsenal includes 163 outer membrane proteins with homology to two proteins, SusC, SusD, that bind and import
starch, 226 predicted glycoside hydrolases, and 15 polysaccharide lyases (Xu et al., 2003). (By contrast, our 2.85
Gb genome contains a relatively ‘‘paltry’’ 99 known or putative glycoside hydrolases and no polysaccharide lyases:
it is deficient in enzymes required for degradation of xylan-,
pectin-, and arabinose-containing polysaccharides that
are common components of dietary fiber [http://afmb.
cnrs-mrs.fr/CAZY/]). We have also produced finished genome sequences for B. vulgatus (31% of Bacteroidetes
and 15% of Bacteria), and B. distasonis (0.8% and
0.4%) (J. Xu and J.I.G., unpublished data). These Bacteroides species also have highly evolved ‘‘glycobiomes’’
(genes involved the acquisition, breakdown, or synthesis
of carbohydrates), that may result from ‘‘top-down’’ selection for functional redundancy.
These large and similar suites of genes within bacterial
genomes could be paralogs resulting from duplication,
but they may well be genes acquired by LGT from bacteria
that are entrenched members of the microbiota or microorganisms acquired from the environment that are simply
passing through our gut (allochthonous members). Indeed, the genomes of the sequenced Bacteroides are
peppered with mobile elements that can facilitate LGT.
Intriguingly, the capsular polysaccharide synthesis loci of
B. thetaiotaomicron contain a number of predicted glycosyltransferases that appear to have been acquired by LGT
(Xu et al., 2003), providing a intriguing perspective about
the dynamic nature of the interface that exists between
our environment, our microbiome, and the surface properties of the microbiota that are seen by our immune system.
Lederberg (2000) has suggested that a good pathogen is
one whose epitopes stimulate an immune response to
competitors: could this (also) be the hallmark of a good
mutualist?
For these Bacteroides species to codominate in the gut,
they likely have distinct hierarchical substrate preferences
would allow resource partitioning and/or metabolic cooperation between congeners (i.e., there would be niche
specialization). Our in vivo functional genomic and mass
spectrometry-based metabolomic studies of the adaptive
foraging behavior of B. thetaiotaomicron in gnotobiotic
mice support this view (Sonnenburg et al., 2005). Groups
of bacteria assemble on undigested or partially digested
food particles and shed elements of the mucus gel layer
and/or exfoliated epithelial cells. Attachment to these nutrient reservoirs is directed by nutrient-regulated glycanspecific outer-membrane binding proteins. Attachment
helps oppose washout from the gut bioreactor and facilitates harvest of carbohydrates by adaptively expressed
glycoside hydrolases. Moreover, when polysaccharide
Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc. 845
availability from the diet diminishes, the organism turns to
polysaccharides in host mucus (Sonnenburg et al., 2005).
Coevolution of glycan structural diversity in the host, and
an elaborate collection of nutrient-regulated glycoside hydrolases in members of the microbiota such B. thetaiotaomicron likely ensures that the host and the microbiota can
adapt to dietary change and maximize energy harvest,
without having to undergo dramatic changes in the representation of this species, or those microbes that use products produced by B. thetaiotaomicron’s ‘‘digestive tract’’
(Sonnenburg et al., 2005).
To obtain a more comprehensive view of the microbiome, we have proposed a human gut microbiome initiative (Gordon et al., 2005; HGMI) that will deliver deep draft
whole genome sequences for 100 species representing
the bacterial divisions known to comprise our distal gut
microbiota. We have identified 86 cultured representatives (22%) of the 395 phylotypes identified in the human
colon. The deposited curated genome sequences would
herald another phase of the ‘‘human’’ genome sequencing
project, provide a key point of reference for interpreting
‘‘metagenome’’ sequencing projects that use total microbial community DNA as starting material for shotgun sequencing, serve as a model for future initiatives that
seek to characterize our other extraintestinal microbial
communities, and facilitate analysis of how selective pressures and community dynamics have shaped the microbiome in diseased humans and in gut pathogens.
Finally, global warming, loss and homogenization of biological diversity due to species invasions and extinctions
(Clavero and Garcia-Berthou, 2005; Mooney and Cleland,
2001), plus human deposition of biologically essential elements such as phosphorus and nitrogen around the planet
(Falkowski et al., 2000), impact our biosphere at the microbial level. Defining the gut microbiota and microbiome in
people who live in various geographic regions, under various levels of economic development, should provide an
opportunity to monitor human ‘‘microevolution’’ during
this period of profound social, economic, and ecological
change, and, hopefully, help us forecast changes in our
disease susceptibility.
Supplemental Data
Supplemental Data include Supplemental Experimental
Procedures and can be found with this article online at
http://www.cell.com/cgi/content/full/124/4/837/DC1/.
ACKNOWLEDGMENTS
We thank Lars Angenent, Justin Sonnenburg, Buck Samuel, Jian Xu, John F. Rawls, and Fredrik Bäckhed for helpful comments. Work from our lab cited in this review has
been supported by grants from the NIH (DK30292,
DK58529, T32-DK07130 [R.E.L.], T32-CA09547 [D.A.P.]),
the NSF (033284), the Ellison Medical Foundation, and
the W.M. Keck Foundation.
846 Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc.
REFERENCES
Acinas, S.G., Klepac-Ceraj, V., Hunt, D.E., Pharino, C., Ceraj, I., Distel,
D.L., and Polz, M.F. (2004). Fine-scale phylogenetic architecture of
a complex bacterial community. Nature 430, 551–554.
Asseman, C., Read, S., and Powrie, F. (2003). Colitogenic Th1 cells are
present in the antigen-experienced T cell pool in normal mice: control
by CD4+ regulatory T cells and IL-10. J. Immunol. 171, 971–978.
Axelrood, P.E., Chow, M.L., Radomski, C.C., McDermott, J.M., and
Davies, J. (2002). Molecular characterization of bacterial diversity
from British Columbia rain forest soils subjected to disturbance. Can.
J. Microbiol. 48, 655–674.
Bäckhed, F., Ley, R.E., Sonnenburg, J.L., Peterson, D.A., and Gordon,
J.I. (2005). Host-bacterial mutualism in the human intestine. Science
307, 1915–1920.
Barker, J., and Jones, M.V. (2005). The potential spread of infection
caused by aerosol contamination of surfaces after flushing a domestic
toilet. J. Appl. Microbiol. 99, 339–347.
Beiko, R.G., Harlow, T.J., and Ragan, M.A. (2005). Highways of gene
sharing in prokaryotes. Proc. Natl. Acad. Sci. USA 10, 2230–2239.
Bjedov, I., Tenaillon, O., Gerard, B., Souza, V., Denamur, E., Radman,
M., Taddei, F., and Matic, I. (2003). Stress-induced mutagenesis in
bacteria. Science 300, 1404–1409.
Breitbart, M., Hewson, I., Felts, B., Mahaffy, J.M., Nulton, J., Salamon,
P., and Rohwer, F. (2003). Metagenomic analyses of an uncultured viral community from human feces. J. Bacteriol. 185, 6220–6223.
Buckling, A., Wills, M.A., and Colegrave, N. (2003). Adaptation limits diversification of experimental bacterial populations. Science 302,
2107–2109.
Clavero, M., and Garcia-Berthou, E. (2005). Invasive species are a leading cause of animal extinctions. Trends Ecol. Evol. 20, 110.
Cobb, B.A., Wang, Q., Tzianabos, A.O., and Kasper, D.L. (2004). Polysaccharide processing and presentation by the MHCII pathway. Cell
117, 677–687.
Cohan, F.M. (2002). Sexual isolation and speciation in bacteria. Genetica 116, 359–370.
Czuprynski, C.J., and Balish, E. (1981). Pathogenesis of Listeria monocytogenes for gnotobiotic rats. Infect. Immun. 32, 323–331.
DeLong, E.F., and Pace, N.R. (2001). Environmental diversity of bacteria and archaea. Syst. Biol. 50, 470–478.
Dianda, L., Hanby, A.M., Wright, N.A., Sebesteny, A., Hayday, A.C.,
and Owen, M.J. (1997). T cell receptor-alpha beta-deficient mice fail
to develop colitis in the absence of a microbial environment. Am. J.
Pathol. 150, 91–97.
Duchmann, R., May, E., Heike, M., Knolle, P., Neurath, M., and Meyer
zum Buschenfelde, K.H. (1999). T cell specificity and cross reactivity
towards enterobacteria, bacteroides, bifidobacterium, and antigens
from resident intestinal flora in humans. Gut 44, 812–818.
Dunbar, J., Barns, S.M., Ticknor, L.O., and Kuske, C.R. (2002). Empirical and theoretical bacterial diversity in four Arizona soils. Appl. Environ. Microbiol. 68, 3035–3045.
Eckburg, P.B., Bik, E.M., Bernstein, C.N., Purdom, E., Dethlefsen, L.,
Sargent, M., Gill, S.R., Nelson, K.E., and Relman, D.A. (2005). Diversity
of the human intestinal microbial flora. Science 308, 1635–1638.
Emerson, B.C., and Kolm, N. (2005). Species diversity can drive speciation. Nature 434, 1015–1017.
Falkowski, P., Scholes, R.J., Boyle, E., Canadell, J., Canfield, D., Elser,
J., Gruber, N., Hibbard, K., Hogberg, P., Linder, S., et al. (2000). The
global carbon cycle: a test of our knowledge of earth as a system. Science 290, 291–296.
Fernandez, A., Huang, S., Seston, S., Xing, J., Hickey, R., Criddle, C.,
and Tiedje, J. (1999). How stable is stable? Function versus community
composition. Appl. Environ. Microbiol. 65, 3697–3704.
Ferriere, R., Bronstein, J.L., Rinaldi, S., Law, R., and Gauduchon, M.
(2002). Cheating and the evolutionary stability of mutualisms. Proc.
Biol. Sci. 269, 773–780.
Frank, D.N., Spiegelman, G.B., Davis, W., Wagner, E., Lyons, E., and
Pace, N.R. (2003). Culture-independent molecular analysis of microbial constituents of the healthy human outer ear. J. Clin. Microbiol.
41, 295–303.
Garrett, A. (1884). The terrestrial mollusca inhabiting the Society Islands. J. Acad. Nat. Sci. 9, 17–114.
Nataro, J.P., and Kaper, J.B. (1998). Diarrheagenic Escherichia coli.
Clin. Microbiol. Rev. 11, 142–201.
Palys, T., Nakamura, L.K., and Cohan, F.M. (1997). Discovery and classification of ecological diversity in the bacterial world: the role of DNA
sequence data. Int. J. Syst. Bacteriol. 47, 1145–1156.
Papineau, D., Walker, J.J., Mojzsis, S.J., and Pace, N.R. (2005). Composition and structure of microbial communities from stromatolites of
Hamelin Pool in Shark Bay, Western Australia. Appl. Environ. Microbiol. 71, 4822–4832.
Putman, R. (1994). Community Ecology (New York: Chapman and
Hall).
Giraud, A., Matic, I., Tenaillon, O., Clara, A., Radman, M., Fons, M., and
Taddei, F. (2001a). Costs and benefits of high mutation rates: adaptive
evolution of bacteria in the mouse gut. Science 291, 2606–2608.
Rakoff-Nahoum, S., Paglino, J., Eslami-Varzaneh, F., Edberg, S., and
Medzhitov, R. (2004). Recognition of commensal microflora by tolllike receptors is required for intestinal homeostasis. Cell 118, 229–
241.
Giraud, A., Radman, M., Matic, I., and Taddei, F. (2001b). The rise and
fall of mutator bacteria. Curr. Opin. Microbiol. 4, 582–585.
Rappe, M.S., and Giovannoni, S.J. (2003). The uncultured microbial
majority. Annu. Rev. Microbiol. 57, 369–394.
Gogarten, J.P., Doolittle, W.F., and Lawrence, J.G. (2002). Prokaryotic
evolution in light of gene transfer. Mol. Biol. Evol. 19, 2226–2238.
Rawls, J.F., Samuel, B.S., and Gordon, J.I. (2004). Gnotobiotic zebrafish reveal evolutionarily conserved responses to the gut microbiota.
Proc. Natl. Acad. Sci. USA 101, 4596–4601.
Gordon, J.I., Ley, R.E., Wilson, R., Mardis, E., Xu, J., Fraser, C.M., and
Relman, D.A. (2005). Extending our view of self: the human gut microbiome initiative (HGMI). (http://www.genome.gov/10002154).
Griffin, A.S., West, S.A., and Buckling, A. (2004). Cooperation and
competition in pathogenic bacteria. Nature 430, 1024–1027.
Gronlund, M.M., Lehtonen, O.P., Eerola, E., and Kero, P. (1999). Fecal
microflora in healthy infants born by different methods of delivery: permanent changes in intestinal flora after cesarean delivery. J. Pediatr.
Gastroenterol. Nutr. 28, 19–25.
Koestler, A. (1967). The Ghost in the Machine (London: Hutchinson).
Leclerc, M., Delgenes, J.P., and Godon, J.J. (2004). Diversity of the archaeal community in 44 anaerobic digesters as determined by single
strand conformation polymorphism analysis and 16S rDNA sequencing. Environ. Microbiol. 6, 809–819.
Lederberg, J. (2000). Infectious history. Science 288, 287–293.
Ley, R.E., Backhed, F., Turnbaugh, P., Lozupone, C., Knight, R., and
Gordon, J.I. (2005). Obesity alters gut microbial ecology. Proc. Natl.
Acad. Sci. USA 102, 11070–11075.
Lozupone, C., and Knight, R. (2005). A new phylogenetic model for
comparing environments clarifies the roles of biogeography and habitat in structuring microbial communities. Appl. Environ. Microbiol. 71,
8228–8235.
Majewski, J., Zawadzki, P., Pickerill, P., Cohan, F.M., and Dowson,
C.G. (2000). Barriers to genetic exchange between bacterial species:
Streptococcus pneumoniae transformation. J. Bacteriol. 182, 1016–
1023.
Mandar, R., and Mikelsaar, M. (1996). Transmission of mother’s microflora to the newborn at birth. Biol. Neonate 69, 30–35.
Mazmanian, S.K., Liu, C.H., Tzianabos, A.O., and Kasper, D.L. (2005).
An immunomodulatory molecule of symbiotic bacteria directs maturation of the host immune system. Cell 122, 107–118.
Mizoguchi, A., Mizoguchi, E., Saubermann, L.J., Higaki, K., Blumberg,
R.S., and Bhan, A.K. (2000). Limited CD4 T-cell diversity associated
with colitis in T-cell receptor alpha mutant mice requires a T helper 2
environment. Gastroenterology 119, 983–995.
Mooney, H.A., and Cleland, E.E. (2001). The evolutionary impact of invasive species. Proc. Natl. Acad. Sci. USA 98, 5446–5451.
Murphy, W.J., Eizirik, E., O’Brien, S.J., Madsen, O., Scally, M.,
Douady, C.J., Teeling, E., Ryder, O.A., Stanhope, M.J., de Jong,
W.W., and Springer, M.S. (2001). Resolution of the early placental
mammal radiation using Bayesian phylogenetics. Science 294,
2348–2351.
Rieu-Lesme, F., Delbes, C., and Sollelis, L. (2005). Recovery of partial
16S rDNA sequences suggests the presence of Crenarchaeota in the
human digestive ecosystem. Curr. Microbiol. 51, 317–321.
Rondon, M.R., Goodman, R.M., and Handelsman, J. (1999). The
Earth’s bounty: assessing and accessing soil microbial diversity.
Trends Biotechnol. 17, 403–409.
Sadlack, B., Merz, H., Schorle, H., Schimpl, A., Feller, A.C., and Horak,
I. (1993). Ulcerative colitis-like disease in mice with a disrupted interleukin-2 gene. Cell 75, 253–261.
Schloss, P.D., and Handelsman, J. (2005). Introducing DOTUR, a computer program for defining operational taxonomic units and estimating
species richness. Appl. Environ. Microbiol. 71, 1501–1506.
Schuter, D. (2000). The Ecology of Adaptive Radiation (New York: Oxford University Press).
Scupham, A.J., Presky, L.L., Wei, B., Bent, E., Grifith, N., McPherson,
M., Zhu, F., Oluwandana, O., Rao, N., Brawn, J., and Boreneman, J.
(2006). Abundant and diverse fungal microbiota in the murine intestine.
Appl. Environ. Microbiol. 72, 793–801.
Sellon, R.K., Tonkonogy, S., Schultz, M., Dieleman, L.A., Grenther, W.,
Balish, E., Rennick, D.M., and Sartor, R.B. (1998). Resident enteric
bacteria are necessary for development of spontaneous colitis and immune system activation in interleukin-10-deficient mice. Infect. Immun. 66, 5224–5231.
Sonnenburg, J.L., Angenent, L.T., and Gordon, J.I. (2004). Getting
a grip on things: how do communities of bacterial symbionts become
established in our intestine? Nat. Immunol. 5, 569–573.
Sonnenburg, J.L., Xu, J., Leip, D.D., Chen, C.H., Westover, B.P.,
Weatherford, J., Buhler, J.D., and Gordon, J.I. (2005). Glycan foraging
in vivo by an intestine-adapted bacterial symbiont. Science 307, 1955–
1959.
Strocchi, A., Furne, J., Ellis, C., and Levitt, M. (1994). Methanogens
outcompete sulphate reducing bacteria for H2 in the human colon.
Gut 35, 1098–1101.
Suzuki, K., Meek, B., Doi, Y., Muramatsu, M., Chiba, T., Honjo, T., and
Fagarasan, S. (2004). Aberrant expansion of segmented filamentous
bacteria in IgA-deficient gut. Proc. Natl. Acad. Sci. USA 101, 1981–
1986.
Sydora, B.C., Tavernini, M.M., Doyle, J.S., and Fedorak, R.N. (2005).
Association with selected bacteria does not cause enterocolitis in IL10 gene-deficient mice despite a systemic immune response. Dig.
Dis. Sci. 50, 905–913.
Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc. 847
Thauer, R.K., Jungermann, K., and Decker, K. (1977). Energy conservation in chemotrophic anaerobic bacteria. Bacteriol. Rev. 41, 100–
180.
Yachi, S., and Loreau, M. (1999). Biodiversity and ecosystem productivity in a fluctuating environment: the insurance hypothesis. Proc. Natl.
Acad. Sci. USA 96, 1463–1468.
Travisano, M., and Velicer, G.J. (2004). Strategies of microbial cheater
control. Trends Microbiol. 12, 72–78.
Zachar, Z., and Savage, D.C. (1979). Microbial interference and colonization of the murine gastrointestinal tract by Listeria monocytogenes. Infect. Immun. 23, 168–174.
Tringe, S.G., von Mering, C., Kobayashi, A., Salamov, A.A., Chen, K.,
Chang, H.W., Podar, M., Short, J.M., Mathur, E.J., Detter, J.C., et al.
(2005). Comparative metagenomics of microbial communities. Science 308, 554–557.
Whitman, W.B., Coleman, D.C., and Wiebe, W.J. (1998). Prokaryotes:
the unseen majority. Proc. Natl. Acad. Sci. USA 95, 6578–6583.
Wilson, D.S. (1975). A theory of group selection. Proc. Natl. Acad. Sci.
USA 72, 143–146.
Woese, C.R. (2002). On the evolution of cells. Proc. Natl. Acad. Sci.
USA 99, 8742–8747.
Xu, J., Bjursell, M.K., Himrod, J., Deng, S., Carmichael, L.K., Chiang,
H.C., Hooper, L.V., and Gordon, J.I. (2003). A genomic view of the human-Bacteroides thetaiotaomicron symbiosis. Science 299, 2074–
2076.
848 Cell 124, 837–848, February 24, 2006 ª2006 Elsevier Inc.
Zhou, J., Xia, B., Treves, D.S., Wu, L.Y., Marsh, T.L., O’Neill, R.V., Palumbo, A.V., and Tiedje, J.M. (2002). Spatial and resource factors influencing high microbial diversity in soil. Appl. Environ. Microbiol. 68,
326–334.
Zoetendal, E.G., Akkermans, A.D.L., Akkermans-van Vliet, W.M., de
Visser, J.A.G.M., and de Vos, W.M. (2001). The host genotype affects
the bacterial community in the human gastronintestinal tract. Micro.
Ecol. Health Dis. 13, 129–134.
Zoetendal, E.G., von Wright, A., Vilpponen-Salmela, T., Ben-Amor, K.,
Akkermans, A.D., and de Vos, W.M. (2002). Mucosa-associated bacteria in the human gastrointestinal tract are uniformly distributed along
the colon and differ from the community recovered from feces. Appl.
Environ. Microbiol. 68, 3401–3407.