Nature | Vol 581 | 14 May 2020 | 207verified by patch-clamp recording with brain slices (Extended Data
Fig. 7d). Administration of CNO via the drinking water throughout
the course of immunization significantly reduced the abundance of
SPPCs in CRH-IRES-Cre mice as compared to control B6 mice (Fig. 3h, i).
Together, these data show that basal CRH neuronal activity in the CeA
and PVN are required for optimal SPPC formation.
To determine whether enhanced CRH neuronal activity would
increase SPPC production during an immune response, we used a
recombinant AAV that conditionally expresses the hM3D(Gq) recep-
tor; as control, we used an AAV expressing a fluorescent protein (Fig. 3b,
activation and control setup). Upon CNO activation, the hM3D(Gq)
receptor triggers action potentials^12 , as verified by recordings of single
CRH neurons in brain slices (Extended Data Fig. 7e). As shown in Fig. 3j, k,
when CNO was given from day 8 to day 12 after immunization, mice
that harboured hM3D(Gq)-expressing CRH neurons produced more
SPPCs than did control mice. Ablation, inhibition, or activation of CRH
neurons did not significantly change the GC response (Fig. 3g, i, k).
To confirm that the connection between CeA and PVN CRH neurons
and the splenic nerve is essential for optimal SPPC production, we
denervated the spleen and then pharmacogenetically activated CeA
and PVN CRH neurons during immunization. The two sets of surgery
required—splenic denervation and transcranial injections (Extended
Data Fig. 8a)—tended to cause variably high background levels of
GCs and plasma cells (X.Z. et al., unpublished observations). We thus
assessed NP-binding, isotype-switched IgG+ plasma cells, which were
exclusively generated from the NP-KLH immunization. Pharmacoge-
netic activation of CRH neurons in the CeA and PVN led to an increase
in NP-specific IgG+ plasma cells only in mice with intact splenic nerves,
but not in denervated mice (Extended Data Fig. 8b, c).
Having identified a functionally important connection between
CeA and PVN CRH neurons and the splenic nerve, we next investigated
whether bodily behaviours could activate this pathway and thereby
enhance the outcome of immunization. PVN CRH neurons respond
to stress and drive the production of immunosuppressive glucocor-
ticoids from the adrenal gland. For this reason, we thought that an
immunostimulatory behavioural paradigm should provoke PVN CRH
neurons but not too strongly, perhaps in the form of mild stress. We
developed a protocol in which mice are made to stand on a round,
transparent platform that is 10 cm in diameter and elevated to 1.5 m
above the ground (Extended Data Fig. 9a). We call this behavioural
regimen elevated-platform standing (EPS), during which mice exhibit
signs of acrophobic stress (Supplementary Video 1). In CRH-IRES-Cre
mice that had been stereotactically injected with a recombinant AAV
that conditionally expresses the DIO-GCaMP6m calcium indicator,
fibre photometry revealed that EPS induced calcium fluxes in CRH
neurons of the CeA and PVN (Extended Data Fig. 9b, Fig. 4a, d). B6 mice
were subjected to 3-min EPS twice daily between days 8 and 12 after
NP-KLH immunization; whereas the GC magnitude was not changed
compared to mice that had not undergone EPS, SPPC abundance was
significantly increased (Fig. 4c). Notably, when spleen-denervated
mice were immunized and subjected to EPS, the SPPC-enhancing effect
if EPS was absent (Fig. 4d), indicating that the splenic nerve-dependent
neural pathway is responsible for the EPS-mediated immunostimula-
tory effect.
To test whether all stressors would stimulate SPPC formation, we
conducted similar analyses of mice subjected to prolonged physi-
cal restraint (PPR), a strong stress-inducing paradigm. As shown in
Extended Data Fig. 9c, d, during a typical 90-min PPR session, PVN
CRH neurons were strongly activated while CeA CRH neurons were
suppressed. PPR led to an increase in circulating corticosterone levels
as compared to EPS (Extended Data Fig. 9e, f ). Notably, PPR during
the course of immunization markedly suppressed GC formation and
did not enhance SPPC formation (Extended Data Fig. 9g). Thus, only
behavioural stress of appropriate form and strength, as represented
by EPS, could be immunostimulatory.
To evaluate whether EPS-promoted SPPC formation is translated
into an enhanced antigen-specific antibody response, we also meas-
ured serum NP-specific IgG titres 2 weeks after immunization. In
EPS-experienced mice, titres increased by about 70% when compared
with control mice (Fig. 5a). This EPS-induced increase required intact
CRH neurons in the CeA and PVN, because no increase was detected
when CRH neurons were ablated (Fig. 5b, c). The enhancing effect
also depended on the splenic nerve, because it was not observed in
spleen-denervated mice (Fig. 5d, e). The EPS effect also required expres-
sion of α9 AChRs on B cells (Fig. 5f, g). Finally, in a separate replication
study, we followed the course of immunization for 4 weeks and found
that, in addition to increased antibody titres (Extended Data Fig. 10a, b),
EPS induced an increase in hypermutation-laden cells among SPPCs
(Extended Data Fig. 10c), and led to a significant increase in the abun-
dance of antigen-specific bone-marrow plasma cells (Extended Data
Fig. 10d, e), which represent the long-lived compartment of humoral
memory^13. Together, these results indicate that, by increasing neural
activity in the CeA/PVN–splenic nerve axis, bodily behaviours can
enhance the outcome of immunization.
Activation of dopaminergic neurons in the ventral tegmental area
(VTA) has been reported to enhance immunity^14. Although chemo-
genetic activation of neurons in the VTA did not increase GC or SPPC
formation (data not shown), it is interesting to consider whether VTA
dopaminergic neurons communicate with the CeA/PVN–splenic nerve
pathway. A reflexive form of neuronal regulation of innate inflamma-
tion, the anti-inflammatory reflex, is well established^15. ChAT-expressing
T cells are essential for this anti-inflammatory reflex^1 and are probably
involved in neural enhancement of the adaptive response describedSPPC (%)012340.5 0.7 GC (%)1.1 1.100.51.01.5Control EPSaΔF/F(%)ΔF/F(%)b–20020400100 200300
Time (s)CeA CRH–500501001500100 200 300PVN CRHΔF/F(%) (PVN)Baseline During AfterΔF/F(%) (CeA)
–1001020050100cGL7FASB220CD138
ControlEPS ControlEPSNerve+SPPC (%)EPS00.51.01.5+
––++––0.6 0.90.4 0.5ShamDenervated
B220CD138Control EPSdP = 0.7 P = 0.008P < 0.0001
P < 0.0001P = 0.1P = 0.01P = 0.01P = 0.047P = 0.01Fig. 4 | EPS stimulates CRH neurons and enhances plasma cell formation.
a, Representative traces of integrated calcium signals from PVN (top) or CeA
(bottom) CRH neurons, read by photometry in CRH-IRES-Cre mice during EPS,
presented as normalized changes in the GCaMP6m f luorescence intensity
(mean in red, s.e.m. in grey). Dashed lines mark the beginning and the end of a
5-min EPS session. b, Average GCaMP6m signals, collected as in a, before,
during and after a 5-min EPS session; grey lines, individual mice; black line with
error bars, mean ± s.e.m. of six mice. One-way ANOVA with Bonferroni’s
correction. c, Representative contour plots (left) and summary statistics of
percentage GC and SPPC (right), 13 days after immunization. Data pooled from
five independent experiments; each symbol indicates one mouse, lines denote
means. d, Representative contour plots (top) and summary statistics of
percentage SPPC (bottom), 13 days after immunization of B6 mice that were
sham-operated or denervated in postnatal week 3 and allowed to recover for
6 weeks before immunization and EPS. Data pooled from three independent
experiments; each symbol indicates one mouse, lines denote means.
Two-tailed unpaired t-test (c, d).