Nature | Vol 585 | 24 September 2020 | 595wild-type HVx mice, but not mAChR TKO mice, treated with bethane-
chol exhibited an increased frequency of pTreg cells, suggesting that
the liver–brain–gut neural arc stimulates colonic APCs and creates
the pTreg niche (Fig. 3e, Extended Data Fig. 8k, l). Collectively, these
results support the essential nature of the neural input from the hepatic
sensory afferents for initiating this vago–vagal liver–brain–gut neural
arc, and that this reflex arc is independent of the sympathetic system
and axon reflex.
As the generation and maintenance of pTreg cells are highly depend-
ent on the microbiome and metabolites, we investigated the role of the
microbiome in these processes. The gut microbiome derived from HVx
mice and sham-operated control mice showed no significant differ-
ences in composition and diversity, and transferring faecal bacteria
from these mice induced similar numbers of gut pTreg cells in germ-free
mice (Extended Data Fig. 9a–e). Consistently, HVx mice co-housed or
parabiosed with sham-operated mice maintained their original phe-
notypes and exhibited reductions in gut pTreg cells (Extended Data
Fig. 9f–k), suggesting that the liver–brain–gut neural arc functions
to harbour a pTreg cell pool independently of HVx-induced alterations
in gut microbiome and metabolites. In addition, we observed no fur-
ther reduction in pTreg cells, particularly microbiome-independent
pTreg cells^13 , in gut-sterilized HVx mice (Extended Data Fig. 9l, m). Col-
lectively, these results indicate that the liver–brain–gut neural arc
maintains basal levels of gut pTreg cells, which are dependent on tonic
microbial input.
This action of the liver–brain–gut neural arc as a modulator of intes-
tinal pTreg cells was previously unanticipated; we therefore sought to
determine whether it is relevant to the development of colitis. Mice
treated with surgical or chemical division of the hepatic vagal branch
exhibited reduced pTreg frequencies (Fig. 3c–e, Extended Data Figs. 4a, b,
5c, d), which resulted in the increased susceptibility to colitis induced by
DSS and 2,4,6-trinitrobenzene sulfonic acid (TNBS) (Fig. 4a–c, Extended
Data Fig. 10a–c). Consistently, HVx did not worsen colitis in Rag2−/−
mice, unlike T-cell-sufficient mice (Extended Data Fig. 10d–f ). In addi-
tion, splenectomy had little effect on the severity of colitis in HVx mice,
unlike in endotoxaemia models^37 –^39 (data not shown). As HVx did not
lead to substantial changes in gut microbiome composition (Extended
Data Fig. 9a–c), HVx mice exhibited increased severity of colitis com-
pared with co-housed sham-operated mice (Extended Data Fig. 10g, h).
Moreover, neither antibiotic-treated mice nor MyD88-deficient mice
exhibited increased susceptibility to DSS-induced colitis after HVx
(Extended Data Fig. 10i–l), suggesting that tonic microbial input is
required for the liver–brain–gut neural arc to maintain the gut pTreg
pool. By contrast, exacerbation of DSS-induced colitis in HVx mice was
inhibited by a cholinergic agonist (Fig. 4d–f, Extended Data Fig. 10m–o).
Collectively, these results indicate that the liver–brain–gut neural arc
serves as a feedback loop to protect the intestine from excessive inflam-
mation (Extended Data Fig. 10p).
In summary, our work reveals an activity of extrinsic vago-vagal
reflexes that connects hepatic vagal sensory afferents, the brainstem
and vagal efferents, and enteric neurons to stimulate mAChR+ APCs
and maintain a reservoir of peripheral regulatory T cells. As demon-
strated by a retrospective cohort study that reported an increased
risk of patients with new-onset depression developing inflammatory
bowel diseases^44 , autonomic imbalance is likely to contribute to the
pathogenesis of inflammatory bowel diseases. Adding to the direct and
reciprocal gut–brain neural reflexes that control appetite, food reward,
cancer, fatty liver, Parkinson’s disease and other neural diseases^45 –^48 ,
our findings provide a distinct view of tissue-specific immune cell adap-
tation mediated by both the liver and central nervous system, which
tunes the levels of gut pTreg cells and prevents potential gut inflamma-
tion. Dysfunction of this liver–brain–gut neural arc predisposes the
gut to inflammation, raising the possibility that denervation-induced
suppression of tumorigenesis could be attributable to the decreased
number of colonic pTreg cells. Our work highlights the essential roles
of the liver–brain–gut neural arc, which specifies the immunoregula-
tory niche and fine-tunes immune responses in the intestine. Interven-
tions that target this liver–brain–gut neural arc could provide broad
applications to promote the treatment of IBD^49 , infectious diseases
and cancer of the gut.Online content
Any methods, additional references, Nature Research reporting sum-
maries, source data, extended data, supplementary information,
acknowledgements, peer review information; details of author con-
tributions and competing interests; and statements of data and code
availability are available at https://doi.org/10.1038/s41586-020-2425-3.ab
6KDP +9[c '66de
6KDP +9[f '66
P$FK57.2 '66 :7 '66 6KDP+9[6FRUH3 6KDP+9[6FRUH3 P$FK57.2 7LPH G 7LPH G 3HUFHQWLQLWLDOERG\ZHLJKW(^)
6KDP
+9[
3HUFHQWLQLWLDOERG\ZHLJKW
6KDP
+9[
6KDP +9[
'$,
3
6KDP +9[
'$,
3
P$FK57.2
Fig. 4 | Perturbation of hepatic vagal afferents exacerbates mouse colitis in
a muscarinic signalling-dependent manner. a–c, Wild-type mice were
subjected to HVx or sham surgery and were given DSS for 7 days, starting at day
2 after surgery. Graphs show pooled data from three independent experiments
(n = 15 per group). d–f, mAchR TKO mice were subjected to HVx or sham surgery
and were given DSS for 7 days, starting at day 2 after surgery. Graphs show
pooled data from three independent experiments (n = 12 per group). a, d, Per
cent change in body weight during acute colitis. b, e, Disease activity index
(DAI) score, on a scale of 0 (mild) to 12 (severe). c, f, Representative
haematoxylin and eosin staining of colon sections (left; scale bars, 200 μm)
and histological scores (right). Data are mean ± s.e.m. P values by unpaired
two-tailed Student’s t-test.