intestinal and blood-brain barrier permeability,
as well as microbiota composition—are also
likely to play a role ( 32 , 37 – 39 ).
From an evolutionary perspective, female-
specific appetite control systems may have
been positively selected because of the greater
importance of energy balance in females for
sexual maturity and pregnancy ( 40 ). In our
study, the differences in weight became sig-
nificant around 6 months of age and increased
with time. This time period correlates with
hormonal changes in female mice associated
with pre-menopausal states in humans ( 41 ),
such as the substantial decrease in estradiol
production. This hormone plays an important
role not only in sexual maturity but also in
energy balance. Estradiol (E2) controls food
intake via NPY and POMC neurons. Specific
deletion of estrogen receptora(ERa)inPOMC
neurons leads to body weight increase in fe-
male mice, and E2-induced anorexia is blunted
in mice lacking NPY neurons ( 42 ). Our findingsshow that MDP’s effects on food intake are
mediated via the GABAergic neurons in the
ARC, which includes NPY neurons. Therefore,
estradiol and MDP may exert their effects
through similar pathways. To explain the ap-
pearance of this phenotype in older females
alone, we hypothesize that MDP’s effect on
food intake is masked by the presence of
higher levels of estradiol in younger females,
as estradiol may be a more potent inducer of
anorexia. As the levels of estradiol begin toGabanyiet al.,Science 376 , eabj3986 (2022) 15 April 2022 6 of 12
GC Fiber trackEF# of spontaneous events per minBaselineFood02468Vgat creOptical connector &
implanted optic fiberOptic fiber
(NA:0.50)AAV9-FLEX-GCaMP
inVgat Arcuate nucleusDBaseline3rd VPost-foodMDPctr MDP# of spontaneous events
(per min - normalized by baseline)# of spontaneous events
(per min - normalizedby baseline)2h post
MDP2h post
MDPctrBaselineBaselineWild TypeTime (min)Maximum nb of spikes (%)AIKJ500ms20mvMDPMDPctr75pA30min0 10 20 30
0204060801001200 10 20 30
02004006008001000Time (min)Rm(mOhm)20%
F/F
20 secHSlc32a1 Nod2 Slc32a1 Nod2Nod2KO
Slc32a1 Nod2 Slc32a1 Nod20 10 20 30
01020304050Rheobase (pA)Time (min)BMDP Control MDP0.00.40.81.21.6VgatcreVgatNod2MDPctr MDP0.00.20.40.60.81.01.21.4
*Fig. 4. MDP decreases VgatARCneuronal activity.(A) Representative images of
RNAscope withNod2(green) andSlc32a1(Vgat, magenta) probes and nuclear staining
(DAPI, blue) in the ARC. Colocalization ofNod2andSlc32a1can be seen, with
Nod2KOmice used as a negative control. Scale bars, 25mm (except first image,
100 mm). (BtoG) In vivo fiber photometry (females >6 months). (B) Scheme of
calcium activity recording scheme for VgatARCneurons. (C) Representative
immunofluorescence image of the brain of a mouse subjected to fiber photometry,
showing expression of GCaMP (GFP, green); cell nucleus is stained with DAPI
(blue). Scale bars, 200mm. Injection site and canula track (dashed lines) are
highlighted; 3rd V, third ventricle. (D) Representative traces of GCaMP6f fluorescence
signals from VgatARCneurons of fasted mice. (E) Number of spontaneous events
before (baseline) and after food presentation (food) for fasted mice. Lines connect
each individual (n= 4). (F) Number of spontaneous events after MDPctr or MDP
gavage for fastedVgatcremice (n= 6). (G) Number of spontaneous events without
fasting forVgatcre(n= 7) andVgatDNod2(n= 5) mice. Lines connect each individual.
Data were normalized by baseline obtained before gavage. (HtoK)Exvivopatch-
clamp recordings in Nod2+VgatARCneurons (females >6 months;n= 4 mice;
nine MDPctr and seven MDP cells). (H) Representative traces of triggered action
potentials from cells treated with MDPctr or MDP at 0 and 30 min. (I) Maximum
number of spikes, (J) membrane resistance, and (K) rheobase measurements along
the recording. Data are averages ± SEM. *P≤0.05 [pairedttest in (E) and (F);
two-way ANOVA in (G), (I), (J), and (K)].RESEARCH | RESEARCH ARTICLE