Nature | Vol 585 | 24 September 2020 | 593colonic APCs obtained from wild-type mice and human intestine (Fig. 1k,
l), whereas co-culture of APCs deficient in Chrm1, Chrm2 and Chrm4
(mAChR TKO) with neurospheroids did not induce Aldh1a1 and Aldh1a2
expression (Extended Data Fig. 1i). Accordingly, generation of FOXP3+
pTreg cells was enhanced by colonic APCs from wild-type mice—but not
by those from mAChR TKO mice—precultured with either muscarine or
neurospheroids (Extended Data Fig. 1j, k). Collectively, these results
suggest that acetylecholine–mAChR signalling in APCs contributes to
maintenance of the pTreg cell population in the gut.
To investigate this idea, we assessed the requirement for signals via
the vagus nerve in preventing intestinal inflammation. VGx resulted in
increased susceptibility to dextran sulfate sodium (DSS)-induced colitis
(Extended Data Fig. 2a–c). Given that VGx reduced the number of pTreg
cells and induced the local inflammatory environment, we next sought
to identify the afferent neurons of the vagus nerve that are involved in
the regulation and maintenance of the gut pTreg pool. The vagus nerve
innervates a large part of the gastrointestinal tract, and its afferent neu-
rons relay sensory inputs to nodose ganglions (NGs) bilaterally^24. During
colitis, these sensory inputs are further projected to the nucleus tractus
solitarius (NTS) of the brainstem (Extended Data Fig. 2d–f ). Notably,
development of acute colitis led to the activation of hepatic sensory
afferents to the left NG and NTS in vivo, which was abrogated by the selec-
tive surgical division of the common hepatic branch of the vagus nerve
(HVx)^25 (Fig. 2a, b, Extended Data Fig. 3a–c). As the liver is continuously
exposed to nutrients, bacterial products, toxins and metabolites from
the intestine, this gut–liver axis, connected by the portal circulation, has
been demonstrated to contribute to liver diseases^26 ,^27. In addition, nutri-
ents and bacterial products can activate the vagus nerve through mecha-
nistic target of rapamycin complex 1 (mTORC1) signalling^28 (Extended
Data Fig. 2g–i), suggesting that the hepatic sensory afferents of the vagus
nerve are activated during colitis. Indeed, hepatic retrograde tracing
supported the idea that the liver senses the gut microenvironment, acti-
vates the hepatic sensory afferents of the vagus nerve and transmits
the signals to the brain via the left NG (Fig. 2c, d). Notably, the common
hepatic branch of the vagus nerve that is divided in HVx mice predomi-
nantly consists of capsaicin-sensitive TRPV1+ sensory afferents without
sympathetic TH+ neurons, as determined by electrophysiological and
immunohistological assessment (Extended Data Fig. 3d, e). Capsaicin
deafferentation of the common hepatic branch of the vagus nerve sig-
nificantly reduced the number of TRPV positive cells in the left, but not
the right, NG (Extended Data Figs. 3e, 5a). In addition, the number of
retrogradely labelled cells was significantly diminished in the left NG,
but not in the dorsal root ganglion (DRG) in HVx mice, indicating that
the common hepatic branch of the vagus nerve sends signals through
the left NG, but not through the right NG or the DRG (Fig. 2c, Extended
Data Fig. 3f–h). These results suggest that sensory information about
the intestinal environment is transmitted to the brain via lateralized
ascending pathways of the left vagus nerve from the liver to the brain.
The anatomical lateralization of the vagus nerve led us to explore
how the hepatic vagal sensory afferents influence gut pTreg cells, and weSham
HVx0510152025pERK1/2 expressionin NG per sectionP = 0.710P = 0.918P = 0.001
P < 0.001
P = 0.032
P = 0.400P = 0.012P = 0.019L RL RNG: DSS (day 7)
Sham HVxa b
Medulla oblongata : DSS (day 7)
Sham HVx
NTS AP NTSDMV DMV
LRNTS AP NTSDMV DMVLR0102030HVxL RShamLVxRVxf020406080P = 0.018P = 0.354L
Sham HVx ShamP = 0.592RR10
0203040c-Fos expressionin NTS per section c-Fos expressionin DMV per sectionP = 0.009Lc WGANG
(L)NG
(R)WGA / DAPId0246810No. of WGA+ neurons per eldeShamVGxHVx0204060FOXP3+ in CD4 cells (%)FOXP3+ in CD4 cells (%)Fig. 2 | The hepatic vagal sensory afferent pathway is essential for NTS
activation during colitis. a, b, Wild-type mice were subjected to HVx or sham
surgery and were then given DSS for 7 days, starting at day 2 after surgery (n = 4
per group). DMV, dorsal motor nucleus of the vagus; AP, area postrema. a, Top,
representative immunostaining for c-Fos in medulla oblongata. Scale bars,
200 μm. Bottom, relative c-Fos expression in NTS and DMV per section. L, left;
R, right. b, Top, representative immunostaining for phosphorylated ERK1/2
(pERK1/2) in NG. Scale bars, 100 μm. Bottom, relative pERK1/2 level in
NG per section. c, d, Wheat germ agglutinnin (WGA) retrograde tracing.
c, Representative f luorescence images of WGA–Alexa Fluor 488 (green) and
DAPI (blue) in NG at 1 week after injection of WGA in liver. Arrowheads indicate
WGA+ neurons in NG. Scale bars, 100 μm. d, Quantification of WGA+ neurons
from c. e, Wild-type mice were subjected to VGx, HVx or sham surgery (n = 9
per group). f, Wild-type mice were subjected to ventral subdiaphragmatic
vagotomy (LVx), dorsal subdiaphragmatic vagotomy (RVx) or sham surgery
(n = 4 per group). e, f, Frequency of FOXP3+ cells among CD4+ cells in colon at
day 2 after surgery. Representative of two (a–d, f) independent experiments or
pooled from three independent experiments (e). Data are mean ± s.e.m.
P values by unpaired two-tailed Student’s t-test (a, b, d) or one-way ANOVA
with Tukey’s post hoc test (e, f).