Science - USA (2022-05-27)

andP. distasonisproduce high amounts of
b-lactamase, which allowsC. bolteaeand
B. productato colonize ampicillin-treated
mice. This approach can provide mechanistic
insights into interactions between the micro-
biota and host in different contexts.

Establishing experimental Koch’s postulates

An ecological version of Koch’spostulates,
which are the four criteria designed to assess
whether a microorganism causes a disease,
offers a useful framework for functional studies
of the microbiota ( 23 , 24 )thatcanbemodified
for experimentation with in vivo models. The
goal is to show that a microbe (or microbes)
isolated from an animal with a particular
phenotype can be isolated in pure culture and
then reintroduced into a susceptible host to
reproduce the phenotype. This procedure can
also be applied to microbes isolated from
humans with disease and transferred into
animal models to test for phenotypes that oc-
cur with the disease. For example,Bacteroides
spp. symbionts isolated from healthy litter-
mate mice induced intestinal inflammation
in a genetically susceptible mouse: a double-
knockout mouse that is deficient in both trans-
forming growth factorbreceptor II (TGFbRII)
and interleukin-10R2 (IL-10R2). Although a
singleBacteroidesspecies was not sufficient to
induce disease, these findings fulfilled Koch’s
postulates in a host genotype–specific fashion.
More importantly, this study established a
conceptual framework for assessing a microbe’s
disease-inducing potential ( 25 ). A recent
study found that dual-specificity phospha-
tase 6 (Dusp6)–deficient mice (Dusp6−/−mice)

can counteract colonic mucosal damage in an

experimental model because of enhanced col-

onic barrier integrity. In this case, a single bac-

terial strain belonging to the genusDuncaniella

was isolated fromDusp6−/−mice, and if germ-

free mice were colonized with the isolated

Duncaniellastrain, they showed less chem-

ically induced injury. Although it is likely that

other microbial strains inDusp6−/−mice could

have similar and redundant effects, this study

showed the potential of using a specific bac-

terial isolate to prevent intestinal epithelium

damage during colitis ( 26 ). A similar discovery

showed that colonization of murine norovirus

in mice that contain a mutation in the Crohn’s

disease susceptibility gene [Atg16L1 hypo-

morphic (Atg16L1HM) mice] recapitulates the

abnormalities observed in the Paneth cells

of conventionally raised norovirus-positive

Atg16L1HMmice ( 27 ). When colonizing con-

ventional or gnotobiotic mice, wound healing

triggered by the increased expression of chemo-

kine CCL5 was impaired by the presence of

strains of the fungusDebaryomyces hansenii

isolated from inflamed tissues of Crohn’sdis-

ease patients as well as by strains of this

fungus extracted from antibiotic-treated con-

ventional mice ( 28 ). Recovery ofD. hansenii

from affected mice verified that both the host

and causative microbes can be potential tar-

gets for therapeutics in Crohn’sdisease.

The next step from ecological tests is to

develop molecular tests of Koch’s postulates

( 29 ) by determining the underlying mecha-

nisms that might support further therapeutic

studies. The human opportunistic fungal path-

ogenC. albicanscolonizes many mucosal sites,

particularly in the gastrointestinal and geni-

tourinary tracts. Alterations in the indigenous

microbiota and host immunity lead to the

overgrowth ofC. albicansand the production

of invasive hyphae that secrete candidalysin,

a cytolytic peptide toxin. This toxin damages

epithelial cells, resulting in phenotypes such

as leaky gut syndrome, and causes life-

threatening disseminated infections.C. albicans

strains that lack this toxin are avirulent and

have not been seen to damage epithelial cells

in a mucosal-infection model ( 30 ). These are

predominately preclinical studies and need

translation to human disease to be useful for

developing targets for therapies.

Interkingdom interactions within microbiota

Mucosal surfaces are densely colonized with a

wide variety of microbes, some of which have

potential pathogenic capabilities (i.e., bacterial

and fungal pathobionts). Major perturbations

of microbiota composition are commonly ob-

served in patients with chronic inflammatory

disorders, such as IBD and asthma. Dysbiosis

triggered by antibiotic treatment can allow the

expansion of pathobionts because of the loss of

commensal microbes, especially those at mu-

cosal surfaces. Innate inflammatory responses

may cause damage to host mucosal barriers

and allow invasion and systemic dissemina-

tion of pathobionts ( 31 ). As a result of colo-

nization resistance by a mature microbiota,

C. albicanscannot colonize the gastrointestinal

track in adult mice. Two commensal bacteria,

B. productaandBacteroides thetaiothaomicron,

antagonizeC. albicansin the mouse intestine.

These bacteria induce expression of hypoxia

Lu et al., Science 376 , 950–955 (2022) 27 May 2022 3of6

Associate microbes/molecules

with a disease/phenotype

Isolation of putative effector

microbes/molecules

In vitro and in vivo

functional validation

Screen with microbes or

microbial products

Functional validation in mouse

Identification of effector microbes

Omics-initiated approach

Rational design

of microbial-based

therapies

Humans with a disease or

animals with a phenotype

Screen-initiated approach

Fig. 2. Omics- and screen-initiated approaches in microbiome research.
The omics-initiated approach (top) begins with multi-omics data generated from
humans with a disease or animals with a phenotype that has features of the
disease state. By computational analysis of the data, individual bacteria, minimal
consortia, or altered microbial products that are enriched or diminished are
further tested in animal models to identify potentially causal microorganisms or

their products. The screen-initiated approach (bottom) begins with a functional

screen of microbial products (metabolites and enzymes) or microbes themselves

on disease-relevant host pathways or cells. By using functional tests in both

in vitro and in vivo studies, the causal molecules and microbes can be identified.

Integration of these two approaches in microbiome studies is a very promising

method to find microbial targets for therapies in human diseases.

THE SYSTEMIC MICROBIOME

ILLUSTRATION: KELLIE HOLOSKI/

SCIENCE