Parallel neural pathways control sodium consumption and taste valence

Abstract

The hedonic value of salt fundamentally changes depending on the internal state. High concentrations of salt induce innate aversion under sated states, whereas such aversive stimuli transform into appetitive ones under sodium depletion. Neural mechanisms underlying this state-dependent salt valence switch are poorly understood. Using transcriptomics state-to-cell-type mapping and neural manipulations, we show that positive and negative valences of salt are controlled by anatomically distinct neural circuits in the mammalian brain. The hindbrain interoceptive circuit regulates sodium-specific appetitive drive , whereas behavioral tolerance of aversive salts is encoded by a dedicated class of neurons in the forebrain lamina terminalis (LT) expressing prostaglandin E2 (PGE2) receptor, Ptger3. We show that these LT neurons regulate salt tolerance by selectively modulating aversive taste sensitivity, partly through a PGE2-Ptger3 axis. These results reveal the bimodal regulation of appetitive and tolerance signals toward salt, which together dictate the amount of sodium consumption under different internal states.

Keywords: appetite; homeostatic neural circuits; internal state; prostaglandin; salt attraction; salt aversion; sensory modulation; sodium homeostasis; taste.

Figure 1.. Independent regulation of appetitive drive and tolerance toward salt.

(A) Consumption assay under sodium depleted (-Sodium, left), osmotic thirst (Thirst, middle), or sated (Sated, right) conditions. Animals were tested with pure water, KCl (500 mM), low salt (60 mM NaCl), high salt (500 mM NaCl), low salt + 440 mM KCl, low salt + 40 mM CaCl₂, and low salt + 40 mM MgCl₂, and low salt + 0.25 mM quinine (n = 6–21 mice). (B) A diagram of state-dependent salt tolerance. Sodium-depleted animals accept aversive high salt or low salt with additional aversive stimuli. Conversely, sated or thirsty animals reject the same solutions. (C) A bimodal regulation model of salt consumption under sodium depletion. (D) Comparison of pre-LC^Pdyn-stimulated and sodium-depleted conditions. Left, a scheme of acute photostimulation of pre-LC^Pdyn neurons. Middle and right, Cumulative consumption curves of pre-LC^Pdyn-stimulated (blue) and sodium depleted (red) conditions during a 30-min session. Low and high salt (60 and 500 mM NaCl) was accepted by both pre-LC^Pdyn-stimulated and sodium-depleted animals. Only sodium-depleted animals tolerated low salt with KCl or quinine but not photostimulated animals (n = 9–15 mice). Data are expressed as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001. See also Figure S1 and Table S1.

Figure 2.. Sodium depletion activates selective excitatory neuron types in the LT.

(A) Representative images of Fos (red) immunofluorescence signals in the forebrain SFO, and hindbrain pre-LC under sodium depletion and sated states. The SFO was counterstained by an excitatory marker nNos (blue) while the pre-LC was counterstained by Foxp2 (blue). The locations of the SFO and pre-LC are −5.52 and −0.7 mm relative to Bregma, respectively. (B) Stimulus-to-cell-type scRNA-seq mapping of neurons activated under sodium depletion. Left, experimental diagram of scRNA-seq. Right, UMAP embedding of SFO neurons. Data from sodium-depleted and sated mice were integrated using CCA alignment (sated n = 3380 neurons, sodium depleted n = 4439 neurons). (C) UMAP embedding for Fos log-normalized expression (red) in SFO excitatory neurons under sated (left) and sodium-depleted conditions (right). (D) Identification of Glut1-enriched genes. Log-normalized average gene expression in Glut1 cluster was compared to other excitatory neural clusters. Glut1- and Glut2–5-enriched genes were shown outside the dotted lines (Table S2). (E) UMAP embedding of Ptger3 log-normalized expression demonstrates faithful expression of Ptger3 in the Glut1 cluster. (F) Violin plots of Fos (red) and Ptger3 (blue) under sated (left) and sodium-depleted (right) conditions. Ptger3 is selectively expressed in Glut1 (bottom) and Glut1 was specifically activated during sodium depletion in the SFO (top). (G) SeqFish analysis of the SFO. Left, spatial distribution of all major cell types in representative anterior (top) and posterior (bottom) sections. After segmentation, cells were color-coded as indicated. Right, A total of 1,567 cells from five anterior sections and 1,614 cells from five posterior sections are quantified. Percentage of individual cell type is presented in the pie chart. Exc. Neurons, excitatory neurons; Inh. Neurons, inhibitory neurons; LT Astro, LT astrocytes; Astro, astrocytes; LT endo, LT endothelial cells; Endo, Endothelial cells; VSMCs, vascular smooth muscle cells; Ependy, Ependymal cells; Oligo, oligodendrocytes. (H) Spatial distribution of all cells is plotted according to the original cell center coordinates. Excitatory neurons are highlighted in colors (red and blue). The Glut1/Ptger3 cluster (red) is separate from other excitatory neuron clusters (blue). Neurons in gray represent excitatory neurons showing gene expression of multiple cluster markers. (I) Violin plots showing the log-normalized expression of Fos (red) and Ptger3 (blue) in SFO excitatory neurons. Fos expression was found specifically in the Glut1 cluster under sodium depletion. Scale bar, 100 μm (A), 50 μm (G-H). See also Figure S2 and Table S2.

Figure 3.. SFO^Ptger3 neurons mediate salt tolerance.

(A) The genetic structure of the Ptger3^Cre mouse line (top). Light and dark blue shades indicate UTRs and exons of Ptger3. Cre is inserted into the first exon, resulting in Ptger3 disruption. In situ hybridization showing co-expression of Ptger3 (green) and Cre:GFP (red) expression (bottom); 86% of Ptger3+ cells expressed Cre:GFP and 83% of Cre:GFP+ neurons expressed Ptger3 (n = 9 sections from 7 mice). (B) Representative Fos immunofluorescence signals (red) under distinct internal states and noxious stimulus. SFO^Ptger3 neurons are labeled green. Activation was highly selective under sodium depletion with a minor activation level induced by inflammation. Pie charts display the percentage of Fos+/Ptger3+ cells in all Fos+ cells. (C) Appetitive-drive test. Photostimulation of SFO^Ptger3 neurons did not induce salt consumption. Left, histological validation of Fos immunofluorescence signals (red) after photostimulation of ChR2-expressing SFO^Ptger3 neurons (green). Middle, the averaged lick numbers of water, low (60 mM), medium (250 mM) and high (500 mM) concentrations of sodium at 20 Hz stimulation. Right, the averaged lick numbers of high salt with different stimulation frequency (n = 4–11 mice). (D) High-salt tolerance test. Left, representative raster plots of licking behavior toward high salt during a 5-sec session from a sated + photostimulated (black), thirsty (blue), or thirsty + photostimulated (red) animal. Right, quantified lick numbers without (blue) or with (red) photostimulation across different concentrations of sodium: water, 60 mM, 250 mM, 500 mM (n = 4–12 mice). (E) Aversive-taste-tolerance test. The number of licks without (blue) or with (red) photostimulation for low salt supplemented with KCl or quinine (n = 6 mice). (F) LT^Ptger3 neurons are required for normal salt tolerance. Left, a diagram of chemogenetic loss-of-function experiments. Middle, the number of licks for water or high salt during a 30-min session with vehicle (grey) or CNO (red) injection under osmotic thirst (n = 5–7 mice). Right, the same behavioral analyses under sodium depletion for low salt, low salt with KCl, or quinine (n = 7 mice). Data are expressed as mean ± SEM, *p < 0.05, ***p < 0.001. Scale bar, 25 μm (magnified images), 50 μm (A), 100 μm (B-C). See also Figure S3, Videos S1–2 and Table S1.

Figure 4.. Parallel and independent activation of fore- and hindbrain circuits under sodium depletion

(A) Two distinct circuit models between fore- and hindbrain neurons related to sodium ingestion. Left, a parallel model where sodium depletion independently activates SFO^Ptger3 and pre-LC^Pdyn neurons. Right, a serial-activation model depicting the hierarchical relationship between the two neural circuits. (B) Testing pre-LC^Pdyn → SFO^Ptger3 projections. AAV-DIO-ChR2-EYFP was transduced in pre-LC^Pdyn neurons. Top, representative images showing axonal projections to the ventral bed nucleus of stria terminalis (vBNST) and SFO. Note that the vBNST is a known downstream area of pre-LC^Pdyn neurons. Bottom, similarly, photostimulation-induced Fos immunofluorescence signals were observed in the BNST but not SFO. Right, the number of Fos+ cells was compared between SFO (red) and pre-LC (grey, n = 4 sections from 4 mice). (C) Testing SFO^Ptger3 → pre-LC^Pdyn projections. Top, SFO^Ptger3 neurons showed no projection to the pre-LC when compared to the MnPO, a known downstream target of the SFO. Bottom, no activation was found in pre-LC^Pdyn neurons after photostimulating SFO^Ptger3 neurons. Right, the number of Fos+ cells by photostimulation (n = 4–6 sections from 3 mice). (D) Activation of SFO^Ptger3 neurons under sodium depletion did not rely on functioning pre-LC^Pdyn neurons. AAV-FLEX-taCasp3-TEVP was injected into the pre-LC bilaterally in Pdyn^Cre or wild-type mice. Middle, representative images of sodium-depletion-induced Fos in the SFO and pre-LC. Right, the numbers of Fos+ cells in the SFO and pre-LC were quantified in transgenic (Cre, red, n = 4–6 sections from 3 mice) and wild-type mice (WT, grey, n = 5–8 sections from 4 mice). (E) Pre-LC activation with ablated LT^Ptger3 neurons. AAV-FLEX-taCasp3-TEVP was injected to the SFO and OVLT in Ptger3^Cre/wt or wild-type mice. Middle, representative images of sodium-depletion-induced Fos. Right, quantified Fos+ cells in the SFO and pre-LC. Ablation of LT^Ptger3 neurons had no effect on the pre-LC activity (red, n = 4 sections from 2 mice; grey, n = 4 sections from 3 mice for SFO, and red, 4 sections from 2 mice and 5 sections from 3 mice for pre-LC). Data are shown as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar, 100 μm. See also Table S1.

Figure 5.. Selective modulation of aversive taste saliency by SFO^Ptger neurons

(A) Dose-dependent behavioral aversion toward bitter tastes. Left, the number of licks was quantified for water, 0.125, and 0.25 mM quinine under osmotic thirst conditions in the absence (black) or presence (red) of photostimulation to SFO^Ptger neurons. Right, preference changes are shown as a ratio by calculating the lick numbers with photostimulation divided by those without photostimulation (n = 5–8 mice). Lick number of water was reanalyzed from Figure 3D. (B) Similar analyses as (A) using sour taste. Water, 10, and 20 mM citric acid were used to quantify taste preference (n = 8–11 mice). Lick number of water was reanalyzed from Figure 3D. (C) Acute stimulation of SFO^Ptger3 neurons did not change preference toward attractive tastant under food deprivation. Average lick numbers for water, 0.5, 1, and 2 mM AceK are shown without (black) or with (red) photostimulation (n= 4–7 mice). (D) Dose-dependent behavioral aversion toward capsaicin, a non-taste oral aversive compound. 0, 0.3, and 1 μM capsaicin were used to calculate avoidance curve (n = 4–5 mice). (E) Saliency modulation by SFO^Ptger3 neurons was specific toward aversive stimuli through the taste system. The effect was not generalized to non-oral aversive stimuli. (F) Stimulation of Ptger3 neurons alleviate aversive taste response under hunger. Shown are raster plots from representative food-deprived mice to sweet (2 mM AceK), sweet with bitter (2 mM AceK with 0.125 mM quinine), and the same mixture with photostimulation. With the photostimulation of SFO^Ptger3 neurons, the bitter taste was tolerated. (G) The effect of photostimulation of SFO^Ptger3 neurons on glucose (500 mM), AceK (2 mM) and monopotassium glutamate + inosine monophosphate (30 mM + 0.6 mM) supplemented with bitter. The average lick numbers toward pure attractive tastants (blue), mixture with 0.125 mM quinine (grey), and together with photostimulation (red) are shown (n = 7–10 mice). Data are shown as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001. See also Figure S4 and Table S1.

Figure 6.. Ptger3 in the SFO is required for salt tolerance.

(A) Fluorescence in situ hybridization validates the lack of Ptger3 transcripts in the SFO in homozygous Ptger3^Cre/Cre (Ptger3^−/−) animals compared to heterozygous Ptger3^Cre/wt (Ptger3^+/−) animals. 89% of Ptger3 signals were abolished in Ptger3^Cre/Cre mice (n = 4 sections from 3 mice) compared to Ptger3^Cre/wt mice (n = 5 sections from 3 mice). (B) Left, Representative images of the SFO and pre-LC under sodium depletion in Ptger3^Cre/wt (grey) and Ptger3^Cre/Cre (red) animals. Sodium depletion activated the SFO in a Ptger3-dependent manner but Fos immunofluorescence signals in pre-LC were unaffected. Right, quantification of the cell activation (grey, n = 4 sections from 4 mice; red, n = 5 sections from 5 mice for SFO, and grey, n = 7 sections from 4 mice; red, n = 6 sections from 3 mice for pre-LC). (C) Aversion tolerance is abolished in Ptger3^Cre/Cre mice. Left, the number of licks toward water, low salt (60 mM NaCl), high salt (500 mM NaCl), and low salt with KCl or quinine in Ptger3^Cre/wt (grey) and Ptger3^Cre/Cre (red) mice under sodium depletion (n = 4–10 mice). (D) Fluorescence in situ hybridization validates the lack of Ptger3 transcripts in the Ptger3-shRNA-injected animals. Compared to scramble-shRNA-injected animals, 87% of Ptger3 signals were abolished in Ptger3-shRNA-injected mice (n = 4 sections from 2 mice) compared to scramble-shRNA-injected mice (n = 4 sections from 2 mice). (E) Left, representative images of the SFO and pre-LC under sodium depletion in scramble- (grey) and Ptger3- (red) shRNA-injected mice. Right, quantification of the cell activation (grey, n = 5 sections from 5 mice; red, n = 5 sections from 5 mice). (F) Aversion tolerance is abolished in Ptger3 KD mice. Left, a diagram of gene knockdown and behavioral paradigm. Right, consumption of the same solutions as Figure 6C was tested in scramble- (grey) and Ptger3- (red) shRNA-injected mice (n = 5–13 mice). Data are expressed as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar, 25 μm (A right, and D), 100 μm (A left, B, and E). See also Figure S5 and Table S1.

Figure 7.. Functional roles of PGE2-Ptger3 signaling for salt taste modulation

(A) ELISA measurement of the circulating PGE2 level. PGE2 was measured from sated (grey), and sodium depleted/repleted (red), and formalin-injected (blue) animals (n = 6–9 mice). (B) Rapid PGE2 access to the SFO. Ex vivo imaging of the SFO was performed after intravenous injection of PGE2-AMCA. Top, experimental diagram and the chemical structure of a synthesized PGE2-AMCA. Bottle left, representative images showing fluorescence of PGE2-AMCA in acutely dissected SFO. Fluorescence levels were quantified across indicated white lines. Bottom middle, representative traces of relative fluorescence levels in vehicle- or PGE2-AMCA-injected animals. Bottom right, quantification of the relative fluorescence level (n = 7 mice for PBS, and 9 mice for PEG2-AMCA). (C) Photometry recording from SFO^Ptger3 neurons. Shown are calcium dynamics and quantified responses (normalized ΔF/F) from SFO^Ptger3 neurons after subcutaneous injection of vehicle (grey) and PGE2 (red, n = 6 recordings from 5 mice for PGE2, and 7 recordings from 5 mice for vehicle). (D) Peripheral injection of PGE2 induces salt tolerance in a Ptger3-dependent manner. High-salt consumption after subcutaneous vehicle (grey) or PGE2 (red) injection in Ptger3^Cre/wt and Ptger3^Cre/Cre animals was analyzed (n = 4–5 mice). PGE2 did not induce a drive to consume salt under sated conditions. Under thirst condition, Ptger3^Cre/wt, but not Ptger3^Cre/Cre, animals showed enhanced salt tolerance with PGE2 injection. (E) Ptger3 KD mice did not tolerate high salt upon PGE2 administration. Scramble- (grey) and Ptger3- (red) shRNA- injected mice were tested in high salt consumption assay (n = 5–10 mice). (F) Anti-inflammatory drugs (NSAIDs) partially decreased salt tolerance in sodium depletion. Sodium-depleted mice supplemented with ibuprofen (drinking water) and aspirin (i.p. injection) showed decreased tolerance toward high salt (n = 4–5 mice). (G) Enhanced dopamine signals toward high salt with PGE2 injection. Left, dopamine release in the NAc was monitored with dLight1.3b during high-salt consumption was measured. Middle and right, averaged dLight1.3b dynamics and quantified data are shown under sodium depletion, thirst, and thirst with PGE2 subcutaneous injection. Lick frequency is shown under dLight traces (n = 8–9 mice). (H) Activation of SFO^Ptger3 neurons induced dopamine release during high-salt consumption. Left, an experimental diagram for virus injection and salt consumption assay. Middle and right, averaged dLight1.3b dynamics and quantified data are shown after i.p. vehicle (grey) or CNO (red) injection (n = 8 mice). Data are expressed as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar, 100 μm. See also Figure S6 and Table S1.