Science - USA (2022-05-27)

SCIENCE science.org 27 MAY 2022 • VOL 376 ISSUE 6596 939

founding factors such as diet being taken
into account ( 8 ). Accordingly, exploratory
clinical findings have shown that ingestion
of fermented dairy products alters brain
activity in regions associated with anxiety
( 9 ). Psychobiotics can improve mood and
affect (emotional states) as well as anxiety
that is associated with irritable bowel syn-
drome. Modulation of gut-derived metabo-
lites by an oral drug, antibiotic treatments,
or FMT appear to improve ASD-related
behaviors and GI symptoms in various in-
vestigational and open-label trials ( 10 , 11 ).
Although early microbiota-based interven-
tions appear promising, replication and
well-controlled studies are needed to move
from correlative research to validation of
disease modification.
Genome-wide association studies
(GWASs) have used large, collaborative team
approaches to increase sample size and un-
cover genetic risks for complex diseases.
However, the microbiome is more diverse
and dynamic than the human genome, fea-
tures that present distinct challenges. The
complexity of human microbiotas in people
living in different locations and environ-
ments, its temporal variability shaped by
lifestyle, and a still incomplete inventory of
human-associated bacteria pose barriers to
understanding its causal effects on human
outcomes. Bacterial and viral community
profiling of fecal DNA has generated evi-
dence of microbiome changes in a variety of
neurological disorders, but critical informa-
tion that is required to fully interpret these
results is still missing. For example, several
psychiatric and neurodegenerative disorders
are accompanied by intestinal problems; dis-
entangling whether the microbiota contrib-
utes to gut symptoms or was reshaped by the
pathology and/or lifestyle associated with GI
manifestations will affect interpretation of
observed microbiome differences. The rela-
tive lack of longitudinal studies does not in-
form whether microbiome changes precede
a clinical diagnosis.
Most cohorts compare community con-
trols to patients and do not capture the ef-
fects of shared environments and diets on
the microbiota that can be assessed through
analysis of household controls and/or sib-
lings. Gathering genetic information, such
as known risk alleles associated with neu-
ropsychiatric or neurodegenerative disor-
ders, would allow more refined analyses
on the level of contribution by the gut mi-
crobiota to impacts of gene-environment
interactions. Therefore, efforts to predict
causality by the microbiota to outcomes in
humans (even in limited contexts) requires
approaches that capture multiple variables
that can influence the microbiota’s plastic-
ity over time, confounds not encountered

in simplified animal models.

A key aspect of microbiome-brain re-

search is its cross-disciplinary nature, and

varied expertise is needed to unravel com-

plex effects on human neurologic or be-

havioral outcomes (see the figure). Large

natural history studies that follow human

populations in their native settings over

time are ideal platforms for exploring the

contributions of the microbiota to effects

of interest because accessible fecal samples

can be collected with accompanying life-

style data, supplying not just longitudinal

information about microbiome profiles

but also real-time tracking of lived experi-

ences that may influence its composition

and function. Notably, these experiences

include environmental changes as well

as emotional stressors, physical stressors,

circadian rhythms, injury, infection, and

neurologic disease diagnosis (e.g., potential

brain-to-body feedback). Similar to natural

history studies, birth cohorts are powerful

for studying certain conditions that mani-

fest early in life, such as ASD. This approach

enables microbiome profiling in high-risk

populations and can start during pregnancy,

with disease expected to be diagnosed 3 to

5 years after birth. Microbiome assessments

in birth cohorts can be coupled with ge-

netic testing, metabolic profiling, and diet

and lifestyle information. Longitudinally

correlating blood or urinary biomarkers,

immune profiles, and microbial products

with behavioral and neurobiological end

points will yield insights into cellular and

molecular mechanisms that may mediate

gut-brain interactions.

Replication across cohorts will be criti-

cal but poses logistical and material chal-

lenges. Defining the directionality of cause

and effect in large population studies would

provide unprecedented opportunities for

generating hypotheses that can be tested

in humans and experimental systems. Also,

the relative contribution of the microbiota

to normal behavioral responses in the ab-

sence of a disease state requires valida-

tion—for example, exploring whether the

microbiota plays a role in homeostatic in-

teroceptive (internal) responses, determin-

ing how microbiota effects on immunity or

metabolism shape everyday behaviors, and

unraveling whether interindividual vari-

ance in cognitive, social, or stress responses

is mediated by microbial influences.

Notably, it is conceivable that the influ-

ences of gut bacteria on human conditions

are not causal, outside of certain types of

infections. By setting homeostatic thresh-

olds for the activity and function of the im-

mune, metabolic, and endocrine systems,

an “unhealthy” microbiota may not itself

cause illness but rather make its host less

resilient to the effects of stressors that in-

clude genetic risks, unhealthy diets or life-

styles, or encountered emotional or physical

stresses. This idea of combinatorial etiolo-

gies, wherein the microbiota represents one

component of a complex disease, is sup-

ported by seminal observations that mice

that harbor specific genetic or environ-

mental risks for ALS, ASD, and PD display

varying symptoms depending on their mi-

crobiome composition ( 12 – 15 ). Accordingly,

interventions to establish a “healthy” mi-

crobiota through diet or FMT, or correcting

subtle microbial deviations with probiotics

or prebiotics, may confer biological resil-

ience that mitigates underlying genetic or

other contributions of illnesses that affect

the brain. Perhaps the view that animals

have intimately coevolved with their mi-

crobiotas, which fundamentally interact

widely and dynamically with the entire

body, explains why so many outcomes and

diseases have been attributed to gut bacte-

rial function (or dysfunction). Robustness

is achieved by overlapping and complemen-

tary functions of a complex microbiota, and

depletion or changes in microbial diversity

may erode resilience to a stressor, thereby

increasing disease risk.

The prospect of microbiota-based in-

terventional studies for proof-of-concept

or treatment objectives is exciting to con-

template. There is already evidence that

FMTs in adult humans are feasible, scal-

able, and safe ( 12 ). Successful interventions

that target the microbiota, not the human,

and improve altered behaviors or brain pa-

thology would help validate that gut bac-

teria contribute to a neurologic condition.

Additionally, the potential of delivering

drugs with targets in the gut, rather than

current approaches that require traversing

the blood-brain barrier, may offer greater

therapeutic tractability and safety. j

REFERENCES AND NOTES

1. J. F. Cryan et al., Physiol. Rev. 99 , 1877 (2019).


2. B. D. Needham et al., Nature 602 , 647 (2022).


3. H. E. Vuong, J. M. Yano, T. C. Fung, E. Y. Hsiao, Annu. Rev.
   Neurosci. 40 , 21 (2017).


4. M. P. O’Donnell, B. W. Fox, P.-H. Chao, F. C. Schroeder, P.
   Sengupta, Nature 583 , 415 (2020).


5. P. Fang, S. A. Kazmi, K. G. Jameson, E. Y. Hsiao, Cell Host
   Microbe 28 , 201 (2020).


6. E. Blacher et al., Nature 572 , 474 (2019).


7. E. Y. Hsiao et al., Cell 155 , 1451 (2013).


8. R. Gacesa et al., Nature 604 , 732 (2022).


9. K. Tillisch et al., Gastroenterology 144 , 1394 (2013).


10. A. Stewart Campbell et al., Nat. Med. 28 , 528 (2022).


11. D.-W. Kang et al., Sci. Rep. 9 , 5821 (2019).


12. C. R. Kelly et al., Gastroenterology 160 , 183 (2021).


13. A. Burberry et al., Nature 582 , 89 (2020).


14. T. R. Sampson et al., Cell 167 , 1469 (2016).


15. D. Matheoud et al., Nature 571 , 565 (2019).

ACKNOWLEDGMENTS

We thank E. Hsiao, R. Knight, and J. Ousey for helpful com-

ments and K. O’Riordan for assistance with the figure.

10.1126/science.abo