function (6).
A low-protein diet is another position is not the only explanation for a
dietary intervention used in chronic kid-
ney disease patients, and a benefit of this
approach appears to be a decrease in ure-
mic toxins. Uremic toxins are compounds
that are retained in the blood when kidney
function is compromised, and increases in
uremic toxins promote a variety of pathol-
ogies, including endothelial dysfunction.
Because gut microbes produce uremic
toxins (including p-cresyl sulfate, indole
and indole derivatives, and trimethylamine
N-oxide), the production or modulation of
these substances by gut microbes has also
been a focus of research (7). Although the
role of the gut microbiota to produce ure-
mic toxins is well-known, Lobel et al. in-
troduce an entirely new twist on this para-
digm: the idea that dietary changes can
trigger posttranslational modifications of
microbial proteins that alter uremic toxin
formation, and thereby influence chronic
kidney disease progression. Specifically,
Lobel et al. report a dietary intervention
that induced a posttranslational modifica-
tion (S-sulfhydration) of a microbial en-
zyme, leading to a decrease in uremic toxin
production that had a protective effect in
a mouse model of kidney disease. This is a
notable finding because posttranslational
changes drove this outcome, and the gut
microbial community composition itself
was not found to be altered (see the figure).
In recent years, a number of studies
have shown that changes in the gut mi-
crobiota (typically, changes in the relative
proportion of various bacteria) are associ-
ated with a wide diversity of diseases and
conditions, including obesity, diabetes, bi-
polar disorder, and depression. However,
many of these studies have been correla-
tive—they have reported shifts in bacterial
abundances in affected individuals, but
it is often unclear why these shifts occur,
or whether they are causal. Are the shifts
somehow promoting the associated pheno-
type? Or are they reacting to it? In at least
one case, a change in bacterial abundance
in type 2 diabetes was tied to a pharma-
ceutical intervention common in the af-
fected group, rather than a consequence
of the disease itself (8), highlighting the
complexity inherent in these studies. To
move research forward, it is imperative
that studies go beyond reporting associa-
tions, and begin to unravel mechanisms
underlying these shifts, and to understand
the consequences.
The study of Lobel et al. goes a step be-
yond the taxonomic approach by highlight-
ing the idea that a “shift” in bacterial com-
Department of Physiology, Johns Hopkins School of
Medicine, Baltimore, MD, USA. Email: jpluznick@jhmi.edu
SCIENCE sciencemag.org
IMMUNOTHERAPY
change in function of the gut microbiota. If
posttranslational modifications of the bac-
terial proteome modulate host physiology
or pathophysiology, then simply determin-
ing bacterial abundance will not reveal all
of the key information. In retrospect, this
should not be surprising: In mammalian
research, it has been known for decades
that the relative abundance of messenger
RNA (mRNA) is not informative about
everything that is happening at the pro-
tein, posttranslational, or functional level.
Similarly, it is not surprising that, for some
situations, species-level abundances of gut
microbes fail to provide the full picture of
what is happening in health and disease.
The findings of Lobel et al. emphasize
the importance of not simply measuring
bacterial abundances, but instead truly
understanding functional processes that
underlie host–microbiota interactions.
Although such studies are difficult, Lobel
et al. provide a blueprint for how they can
be accomplished. Furthermore, these stud-
ies imply that the progression of chronic
kidney disease may be modified by new
strategies that alter the gut microbiota
and/or the enzymatic activities of the gut
microbial proteome.
Looking ahead, these types of ap-
proaches can yield dividends for not only
chronic kidney disease, but for a large array
of other diseases and conditions for which
the microbiota have been implicated. By
definition, these types of studies require
a multidisciplinary approach as one must
be cognizant of not only microbial biol-
ogy and posttranslational modification
modalities but must also understand the
whole-animal physiology and pathophysi-
ology of disease. Thus, this represents a
great opportunity for scientists from di-
verse fields to come together. By pooling
the diverse knowledge and approaches of
varied fields, advances in understanding of
host–microbiome interactions and disease
progression can be made, and perhaps new
approaches to disease treatment and pre-
vention can be uncovered. j
REFERENCES AND NOTES
1. G. B. D. C. K. D. Collaboration, Lancet 395, 709 (2020).
2. S. Noel et al., Nephron Clin. Pract. 127, 139 (2014).
3. S. Mills, C. Stanton, J. A. Lane, G. J. Smith, R. P. Ross,
Nutrients 11, 923 (2019).
4. J. L. Sonnenburg, F. Bäckhed, Nature 535, 56 (2016).
5. L. Lobel et al., Science 369, 1518 (2020).
6. C. M. Rebholz et al., Am. J. Kidney Dis. 68, 853 (2016).
7. W. L. Lau, J. Savoj, M. B. Nakata, N. D. Vaziri, Clin. Sci.
(London) 132, 509 (2018).
8. K. Forslund et al. , Nature 528, 262 (2015).
ACKNOWLEDGMENTS
This work was supported by NIH DK107726 and NIH HL128512
(to J.L.P.).
10.1126/science.abd8344
Published by AAAS
Messengers
from the
microbiota
Microbiota produce
inosine, a metabolite that
modulates effector T cells
and tumor immunity
By Fyza Y. Shaikh1 and Cynthia L. Sears1,2
Downloaded from http://science.sciencemag.org/ on September 18, 2020
I
mmunotherapy sometimes leads to
stunning therapeutic success in a wide
range of cancers. However—except in
melanoma, microsatellite instable (MSI)
colorectal cancer, Hodgkin’s lymphoma,
and Merkel cell carcinoma—only ~15 to
30% of cancer patients respond to treatment.
Thus, identification of the determinants of
response are under intense investigation,
and the gut microbiota is proposed to be a
critical determinant for immunotherapy re-
sponses. On page 1481 of this issue, Mager
et al. (1) show that inosine produced by
specific gut bacteria, Bifidobacterium pseu-
dolongum or Akkermansia muciniphila,
enhances immune checkpoint inhibitor
(ICI; a type of immunotherapy) efficacy in
numerous mouse models. They propose
that impaired gut barrier function from ICI
treatment facilitates inosine systemic trans-
location, resulting in adenosine 2A receptor
(A2AR)–dependent activation of T helper 1
(TH1) antitumor effector cells that yield tu-
mor shrinkage. These intriguing data pro-
vide a potential framework to directly link
microbe-induced metabolite synthesis and
ICI therapeutic efficacy.
Step-by-step examples of a micro-
biota, metabolite, immune effector, tu-
mor response (or nonresponse) cascade
in humans do not yet exist. However, the
complexity presented by the hundreds to
thousands of human gut bacteria may be
reduced by enhancing the understanding
of the biosynthetic metabolite redundancy
among bacterial species (2). Accruing ex-
perimental and human data linking bac-
terial metabolites to human physiology
and disease are encouraging. For example,
studies of the interactions of bacterial me-
1
Department of Oncology, Johns Hopkins University School
of Medicine, Baltimore, MD, USA. 2Department of Medicine,
Johns Hopkins University School of Medicine, Baltimore,
MD, USA. Email: csears@jhmi.edu
18 SEP TEMBER 2020 • VOL 369 ISSUE 6510 1427
INS IGHTS | P E R S P E C T I V E S
tabolites and human G protein–coupled
receptors (GPCRs), a critical receptor class
that mediates drug action, suggest that
multiple simple bacterial metabolites, pro-
duced by more than one bacterial species,
are GPCR agonists and affect immune and
neurologic function (3, 4).
Furthermore, molecular approaches
to modify bacterial genomes, such as
CRISPR-Cas9–based gene deletion, en-
able characterization of how single mol-
ecules produced by specific bacteria affect
host function. For example, using this ap-
proach, analysis of a Clostridium sporo-
genes mutant deficient in the production
of branched short chain fatty acids (SCFAs)
revealed a previously unknown capacity
for branched SCFAs to regulate the num-
bers of immunoglobulin A (IgA)–produc-
ing plasma cells in the small intestine (5).
Fermentation of dietary fiber by the gut
microbiota produces SCFAs (acetate, pro-
pionate, and butyrate). SCFAs are the most
abundant microbial metabolites in the
gut that act through gut mucosal GPCRs
(GPCRs 41, 43, and 109A) and are known
for their critical role in colon homeosta-
sis. SCFAs regulate the phenotype and/
or function of macrophages, neutrophils,
dendritic cells, and CD4+ T cells [especially
regulatory T cells (Treg cells)] and have re-
cently been found to promote CD8+ T effec-
tor function and memory potential (6, 7).
Early translational studies in cancer
patients suggest that SCFAs limit anti–
CTLA-4 (cytotoxic T lymphocyte–associ-
ated antigen 4) ICI responses but promote
anti–PD-1 (programmed cell death protein
1) ICI responses (8, 9). ICIs block inhibi-
tory immune cell and tumor molecules to
unleash an antitumor immune response.
ICIs, particularly anti–CTLA-4, likely al-
ter intestinal barrier function in humans.
Altered barrier function is associated with
microbiota disruption (dysbiosis) that
contributes to acute colitis, inflammatory
bowel diseases, as well as colorectal cancer.
Thus, better understanding of the effect of
ICIs on the gut microbiota, its metabolic
capacity, and metabolite-mediated down-
stream effects on colon mucosal immunity
and tumor immune responses is needed.
It is clear that untangling which gut mi-
crobiota species and metabolites, singly or
together, modify human disease and thera-
peutic outcomes is a difficult challenge.
Inosine is another key metabolite pro-
duced by the gut bacteria. This endog-
enous purine nucleoside is formed by
deamination of adenosine. Prior data
support an immunosuppressive role for
inosine, which activates A2AR signaling to
limit inflammation, block TH1 cell differen-
tiation, promote Treg cell activity, and thus
1428 18 SEP TEMBER 2020 • VOL 369 ISSUE 6510
Microbial signaling
Diet and medications yield microbiota that produce
metabolites such as inosine or short chain fatty
acids (SCFAs), which regulate mucosal and
systemic immune cells. Inosine acts in the tumor
microenvironment, through activation of adenosine
2A receptor (A2AR) signaling in T helper 1 (TH1) cells
and regulates antitumor immunity in a context-
dependent manner.
Immune checkpoint inhibitor (ICI) therapy
ICIs disrupt the gut barrier, which allows bacterial
metabolites, such as inosine, to promote antitumor
immunity in an interferon-g (IFNg)–dependent manner.
Akkermansia Bifdobacterium
muciniphila pseudolongum
Gut bacteria
Inosinee S
SCFAs
Mucus
Colon epithelial cells
Immune cells
Inosine, SCFAs, and
Inosinee mucosal immune cells travel
to tumors within the patient
A2AR
Cytotoxic
IFNg
T cells
TH1 cell
Tumor bed
Healthy tissue
inhibit tumor immunity in vivo. A2AR is
also a GPCR and a target for drug develop-
ment. Pharmacologic antagonists of A2AR
enhance antitumor immunity in multiple
mouse models and are currently in on-
cology clinical trials. Conversely, inosine
supplementation was recently reported
to enhance the antitumor efficacy of ICI
therapy and adoptive T cell transfer in pre-
clinical mouse models, but only in models
that used tumor cells unable to catabolize
inosine to support cell growth (10). These
data suggest that, at least in some circum-
stances, inosine can relieve tumor-imposed
metabolic restrictions on T cells.
Mager et al. build on this theme by show-
ing with in vitro assays that inosine dis-
plays context-dependent actions (see the
figure). Inosine boosted or inhibited TH1
differentiation of naïve T cells in the pres-
Published by AAAS
ence or absence, respectively, of exogenous
interferon-g. Prior work similarly identi-
fied highly context-dependent, oncogenic
Treg cell action (11). Collectively, these data
highlight that “company matters” in the
highly pleomorphic and ever-expanding
characterization of immune cell function,
especially within the tumor microenviron-
ment. Whether context-dependent inosine
action is relevant to human TH1 cell dif-
ferentiation or effector cell biology and/or
affects current human cancer therapy tri-
als of A2AR antagonists deserves scrutiny.
Further analysis of microbial, serum, and
immune cell samples from patients under-
going ICI treatment will clarify the impact
of the observations by Mager et al.
Despite the exciting explosion of micro-
biota studies over the past 15 years, pin-
ning down which microbial members or
Downloaded from http://science.sciencemag.org/ on September 18, 2020
mediators are validated biomarkers for
human disease prognosis or therapeutic
response, has been elusive. Encouragingly,
microbiota-based therapeutics are emerg-
ing for the treatment of Clostridioides dif-
ficile infection (12). Although the highly
positive causal findings in microbiota
studies in mice are thrilling, perhaps im-
plausibly so, these results contrast sharply
with diverse and inconsistent findings in
microbiota human studies to date. Diet
and drug exposure (antibiotics and non-
antibiotics) (13), both recent and distant
(14, 15), seem to modify microbiota-linked
human disease outcomes, yet these factors
are rarely included as prospectively col-
lected variables in human study analyses.
These gaps in outcomes emphasize that to
harness the putative power of the microbi-
ota for therapeutic success, well-designed,
well-powered human studies tied to trans-
lational laboratory studies are needed to
define circumstances that are addressable
with microbiota-based treatments. j
RE FERENCES AND NOTES
1. L. F. Mager et al., Science 369, 1481 (2020).
2. P. C. Dorrestein, S. K. Mazmanian, R. Knight, Immunity
40, 824 (2014).
3. H. Chen et al., Cell 177, 1217 (2019).
4. D. A. Colosimo et al., Cell Host Microbe 26, 273 (2019).
5. C.-J. Guo et al., Science 366, eaav1282 (2019).
6. M. Luu et al., Sci. Rep. 8, 14430 (2018).
7. A. Bachem et al., Immunity 51, 285 (2019).
8. C. Coutzac et al., Nat. Commun. 11, 2168 (2020).
9. M. Nomura et al., JAMA Netw. Open 3, e202895 (2020).
10. T. Wang et al., Nat. Metab. 2, 635 (2020).
11. A. L. Geis et al., Cancer Discov. 5, 1098 (2015).
12. B. H. McGovern et al., Clin. Infect. Dis. ciaa387 10.1093/
cid/ciaa387 (2020).
13. L. Maier et al., Nature 555, 623 (2018).
GRAPHIC: C. BICKEL/SCIENCE
14. L. Derosa et al., Ann. Oncol. 29, 1437 (2018).
15. J. Zhang et al., Gut 68, 1971 (2019).
ACKNOWL EDGMENT
The authors thank J. Powell and D. Pardoll for helpful
discussions.
10.1126/science.abe0709
sciencemag.org SCIENCE
Messengers from the microbiota
Fyza Y. Shaikh and Cynthia L. Sears

Science 369 (6510), 1427-1428.

DOI: 10.1126/science.abe0709

Downloaded from http://science.sciencemag.org/ on September 18, 2020

ARTICLE TOOLS http://science.sciencemag.org/content/369/6510/1427

RELATED http://science.sciencemag.org/content/sci/369/6510/1481.full
CONTENT
http://stm.sciencemag.org/content/scitransmed/12/530/eaax0876.full
http://stm.sciencemag.org/content/scitransmed/10/443/eaan4116.full
http://stm.sciencemag.org/content/scitransmed/8/328/328rv4.full
http://stm.sciencemag.org/content/scitransmed/9/379/eaah3560.full
REFERENCES This article cites 14 articles, 3 of which you can access for free
http://science.sciencemag.org/content/369/6510/1427#BIBL

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. The title Science is a registered trademark of AAAS.
Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of
Science. No claim to original U.S. Government Works