Mice of either sex (12 to 15 weeks old, 23 to 30 grams) were used in all experiments. The HSD11b2-Cre mice were purchased from Jackson Laboratories (Strain #030545, Bar Harbor, ME, United States). This line of mice was originally generated by Jarvie and Palmiter [13], who targeted Cre recombinase to the gene encoding HSD2 (HSD11b2). Mice were maintained on a 12-hour light/dark cycle at 22 $\pm$ 2 °C and 50% humidity, with free access to normal food (0.283% sodium) and water in a pathogen-free animal care facility. For experiments, the low-sodium diet (LSD) group was switched to low-sodium food (0.05% sodium) as needed. Both the normal-sodium food (XTI01WC-004) and low-sodium food (XTM09-008) were obtained from Jiangsu Synergic Bioengineering Company (Nanjing, Jiangsu, China). For euthanasia, mice were administered a lethal dose of urethane (4 g/kg, i.p., HY-B1207, MCE, Shanghai, China), followed either by decapitation or by transcardial perfusion with saline and 4% paraformaldehyde (PFA, A500684, Shenggong Biotech, Shanghai, China). All animal procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals (Chinese Association for Laboratory Animal Sciences) and approved by the Animal Care and Ethics Committee of Hebei Medical University.

Mice were anesthetized with pentobarbital sodium (60 µg/g, i.p.,P3761; Sigma-Aldrich, San Francisco, CA, USA) and secured in a stereotaxic frame. The NTS stereotaxic surgery procedure referenced our previously published paper [20]. Stereotaxic surgery was performed to inject viral vectors into the NTS of HSD11b2-Cre mice. rAAV9-hSyn-DIO-EGFP (Lot No. BC-0244, BrainVTA Co., Ltd, Wuhan, Hubei, China) labeled HSD2 neurons, while rAAV9-EF1$\alpha$-DIO-ChR2-mCherry (Lot No. BC-0108, BrainVTA Co., Ltd, Wuhan, China) enabled Cre-dependent ChR2 expression. Virus titers exceeded 10¹² GC/mL and were stored at –80 °C. Viral vectors were bilaterally injected into the NTS (60 nL per injection, 6 injections; calamus scriptorius: anteroposterior: +0.1 mm, mediolateral: $\pm$0.2 mm, dorsoventral: –0.1 mm; anteroposterior: 0.3 mm, mediolateral: $\pm$0.2 mm, dorsoventral: –0.1 mm; anteroposterior: 0.5 mm, mediolateral: $\pm$0.2 mm, dorsoventral: –0.1 mm) at a rate of 25 nL/min, using a pulled glass pipette adapted to a 2 µL microsyringe (Microliter #7102, Hamilton, Reno, NV, USA) connected to a pressure-driving syringe pump (Harvard Apparatus, Holliston, MA, USA). The pipette was left in place for 5 minutes post-injection to ensure diffusion before slow withdrawal. Ampicillin (125 mg/kg, i.p., A6920, Solarbio, Beijing, China) was given pre-surgery, and ketorolac (4 mg/kg, i.p., IK0560, Solarbio) was administered for analgesia. Mice were allowed to recover for 3 weeks before any behavioural, physiological, or tracing experiments were performed (see also Fig. 1), ensuring sufficient time for transgene expression.

Fig. 1.

Experimental protocol. NSD, normal sodium diet and water provided ad libitum; LSD, low sodium diet and water provided ad libitum; LSD-H, a low sodium diet and water are provided ad libitum, but during the final 2 hours, the water is replaced with a 3% NaCl solution. Two separate groups of mice were used in conducting the “two bottle choice test” and the “immunofluorescent staining” experiment.

The detail method refers to our previous published paper [20]. Mice were anesthetized (urethane, 4 g/kg, i.p.), perfused with cold saline followed by 4% PFA (PBS, pH 7.4). Brains were post-fixed (4% PFA, 4 °C, 12 h), cryoprotected in 30% sucrose, and sectioned coronally. Sections were blocked (5% Bovine Serum Albumin with 0.25% Triton X-100, PBS, 30 min, room tempreture), incubated with primary antibodies (2% Bovine Serum Albumin-PBS, 4 °C, overnight), washed (PBS $\times$6), and incubated with fluorescent secondary antibodies (2 h, room tempreture). After mounting (Vect ashield), NTS-Enhanced Green Fluorescent Protein (EGFP) neurons were imaged and counted in 6 sections/mouse. Antibody sources are in Supplementary Table 1. Fluorescence images were collected with a Leica DM6000B microscope (Leica Microsystems, Wetzlar, Germany).

Viral vectors encoding ChR2 were bilaterally injected into the NTS (rAAV9-EF1$\alpha$-DIO-ChR2-mCherry) of HSD11b2-Cre mice. To activate HSD2 neurons in freely moving mice, an optical fiber cannula was fixed above the skull over the nucleus of the NTS. To prevent damage to the NTS tissue and interference with HSD2 neurons due to mice movement or neck flexing, the fiber was positioned on the midline surface (Bregma, anteroposterior: –7.5 mm, mediolateral: 0, dorsoventral: 4 mm) of the area postrema (AP). The output power of light at the end of the optical fiber was 10 mW in all experiments, as measured with an optical power meter (PM20; Thorlabs, Newton, NJ, USA). For blue light stimulation of ChR2-expressing neurons, an optical fiber (Model: FOC-C-W-200-1.25-0.37-6.0, 200 µm in diameter, 0.37 NA, 6 mm length, Hangzhou Sanshi Biotechnology Co., Ltd., Hangzhou, Zhejiang, China) was connected to a 473 nm LED source (Newdoon Inc., Hangzhou, Zhejiang, China). In awake mice, we optogenetically activated HSD2 neurons with 60 s of blue-light pulses (473 nm, 10 Hz, 20 ms pulse width, 10 mW) to ensure adequate photostimulation. Under anesthesia, stimulation frequencies of 1, 5, 10, 15 and 20 Hz (10 ms pulse width, 10 mW, 60 s) were used for BP and PNDphrenic nerve discharge (PND) recordings. To illuminate the HSD2 neurons in anesthetized mice, an occipital craniotomy was made to expose the dorsal surface of the medulla oblongata over the NTS. ChR2-transduced HSD2 neurons were photostimulated by placing an optical fiber over the NTS on one side.

To assess the impact of sodium deprivation on salt appetite in mice, animals were individually housed and provided with two bottles of purified water for a one-week acclimation period. On the test day, one of the water bottles was replaced with a 1.5% NaCl solution (A610476, BBI life sciences, Shanghai, China). Fluid intake was measured by manually weighing the bottles before and after consumption to calculate the liquid ingested over a 24-hour period. For more precise measurement of licking behavior during optogenetic activation of HSD2 neurons, we used a two-bottle choice lickometer system (ZS-PH; Beijing Zhongshidichuang Science and Technology Development Co., Ltd., Beijing, China). This system records the number of licks with a temporal resolution of 0.01 seconds, allowing accurate assessment of sodium preference. Mice were housed individually in cages equipped with two drinking tubes containing either distilled water or 1.5% NaCl for at least one week prior to testing. The operating principle of this lickometer system is similar to that of the Stoelting Lickometer System (https://stoeltingco.com/Neuroscience/Stoelting-Lickometer-System~10447). An electrical signal is generated each time the mouse’s tongue contacts the sipper tube, and each lick is recorded as an event within the software (Trigger Master V4.0, Shanghai VanBi Intelligent Technology Co., Ltd., Shanghai, China).

The detail method refers to our previous published paper [21]. Mice were anesthetized (pentobarbital sodium, 60 µg/g, i.p.) and a PE50 tube was inserted into the left common carotid artery, connected to a Millar Mikro-Tip® catheter pressure transducer (MIT0699, ADInstruments, Sydney, Australia). Data was recorded at 0.1 kHz and analyzed with LabChart software (PowerLab26T, LabChart 7.0, ADInstruments, Sydney, Australia).

The protocol was used as previously depicted [22]. Mice were anesthetized with urethane (1.3 g/kg, i.p.) and secured in a stereotaxic frame, underwent tracheostomy and bilateral vagotomy, then paralyzed (pancuronium, 5 mg/kg, i.p., P1918, Sigma-Aldrich, Saint Louis, MO, USA) and ventilated (100% O₂). The left phrenic nerve was recorded using a silver electrode that was paraffin-immersed, while end-tidal CO₂ (ETCO₂) was maintained at approximately 4% as the basal level and was continuously monitored with a MicroCapStar capnograph (model 1000, CWE Inc., Ardmore, PA, USA). Signals were digitized (Micro1401), processed (rectified/smoothed: 0.1 s time constant, 2 kHz sampling, 30–3000 Hz bandpass), and analyzed offline (Spike2). Integrated PND frequency and peak amplitude were quantified.

Mice were anesthetized with pentobarbital sodium (60 µg/g, i.p.). A 27-G guide cannula (RWD Life Science, Shenzhen, China) was stereotaxically implanted into the fourth ventricle (i4V) at coordinates bregma, –6.23 mm; mediolateral, 0.0 mm; dorsoventral, –4 mm, and secured to the skull. Cannula placement was verified both intra-operatively and post-hoc. During surgery, the cannula (targeted to bregma –6.23 mm, referenced from literature [23]) was advanced while the attached 20 cm PE-50 line was filled with aCSF and held vertically. A sudden drop of the meniscus indicated penetration of the cerebellum and entry into the 4th ventricle. Following ventricle cannulation surgery, the mice were divided into three groups: aCSF (n = 10), KH7 (n = 14), and U0126 (n = 14). After a 5-day recovery period, i4V infusions were performed in conscious, freely moving mice using 33-G internal cannulas connected via 60 cm PE50 tubing to a 10 µL Hamilton 700 syringe. The syringe was mounted on a syringe pump (Pump 11 Elite, Harvard Apparatus) and infusions were delivered at a rate of 2 µL/min for one minute. KH7 (100 µM, HY-103194, MedChemExpress, Shanghai, China), a specific inhibitor of soluble adenylyl cyclase [24], and U0126 (10 µM, HY-12031A, MedChemExpress), a potent non-ATP competitive and selective MEK1/2 inhibitor [25, 26], were infused into the i4V. Control animals received an equivalent volume of vehicle (10% DMSO in corn oil). Because the internal cannula was connected to the pump via a 60 cm length of PE-50 tubing, hydraulic compliance and surface tension introduced a variable delay between pump advancement and fluid outflow. Although the pump was programmed to deliver 2 µL at 2 µL/min for 1 min, the actual volume that entered the 4th ventricle was smaller and could not be precisely quantified. Pilot tests showed that visible meniscus movement at the cannula tip lagged 15–30 s behind the pump command, estimating an actual delivered volume of approximately 1 µL. After infusion, the internal cannulas were removed and the guide cannulas were sealed. One hour later, 4 mice from each group were anesthetized and transcardially perfused for c-Fos and 11$\beta$-HSD2 immunohistochemistry. The remaining mice in each group were allowed unrestricted movement for the subsequent 24-h two-bottle choice test. Subsequently, 2 µL of 2% Pontamine Sky Blue (C8679, Sigma-Aldrich) was infused through each cannula; only animals in which the dye flowed freely into the 4th ventricle were retained. Applying these criteria, the final group sizes were: aCSF (n = 6/6 patent), KH7 (n = 8/10, two occluded), and U0126 (n = 9/10, one occluded).

We re-analyzed the published single-cell RNA-Seq data from Resch et al. [11] using the same preparation method as described previously. Publicly available RNA-seq data were obtained from the GEO database (accession GSE102332). Differential expression analysis was conducted using edgeR [27], with genes defined as significantly up- or down-regulated when they met the following thresholds: $|$log2 fold change$|$ $>$1 and p 0.05, without correction for multiple comparisons. Visualization of results was performed via the ggplot2 package in R (v4.2.0, R Foundation for Statistical Computing, Vienna, Austria).

Statistical analysis of data was performed using GraphPad Prism v9 (GraphPad Software Inc., SanDiego, CA, USA) and the results were expressed as the means $\pm$ standard deviation. Difference between the two groups was analyzed by were calculated by t test (2-tailed) including unpaired t test, or paired t test. The differences among multiple groups were analyzed using one-way ANOVA followed by Tukey’s post hoc test. p values less than 0.05 were considered statistically significant. Figures were prepared using Illustrator software (Adobe Illustrator CS6, San Jose, CA, USA).

To test whether HSD2 neurons are activated by sodium deprivation or silenced by high sodium and to verify their role in sodium appetite, HSD11b2-Cre mice received AAV-Syn-DIO-EGFP for specific labeling of these neurons. After three weeks of viral expression, animals were maintained on standard chow and water ad libitum and then randomly assigned to three experimental groups: normal-sodium diet (NSD, n = 15), low-sodium diet (LSD, n = 11) and a low-sodium group with a terminal 2-h exposure to 3% NaCl (LSD-H, n = 6). During the fourth week (days 22–28) all mice were single-housed with two bottles of purified water to habituate. On day 29, the NSD cohort was split: nine mice continued on normal chow but one water bottle was replaced with 1.5% NaCl for 24 h and intake was determined gravimetrically, while the remaining six mice were transcardially perfused with 4% PFA for c-Fos immunostaining to assess neuronal activation. The LSD group was handled similarly: five mice were given 1.5% NaCl for 24 h with intake measured, and six were perfused for c-Fos analysis. The LSD-H group remained on LSD until two hours before perfusion, when one water bottle was exchanged for 3% NaCl; all mice were then perfused for c-Fos labeling. The timeline is summarized in Fig. 1.

Immunofluorescence results confirmed HSD2 neurons to be located near the obex of the 4V, extended caudally through the commissural NTS, and approached the spinomedullary transition (Fig. 2A). Mouse brains were sectioned into 0.02 mm-thick slices and EGFP marked HSD2 neurons bilaterally EGFP-expressing neuron soma were manually counted across six coronal sections (bregma: –8.0 to –7.0 mm; thickness, 20 µm; each separated by 120 µm) from each mouse (n = 6). Only neurons with a clearly visible soma within the tissue section were counted as part of the denominator. Each mouse had approximately 121 $\pm$ 33 HSD2 neurons (Fig. 2B). In the NSD group, c-Fos expression was minimal in EGFP-positive NTS neurons, reflecting baseline neuronal activity (Fig. 2C, upper panel). Conversely, the LSD group showed a marked increase in c-Fos-positive cells among EGFP-labeled HSD2 neurons (Fig. 2C, middle panel), indicating strong neuronal activation under dietary sodium restriction conditions. The LSD-H group had the lowest c-Fos expression when given 3% NaCl for the last 2 hours (Fig. 2C, lower panel), suggesting that the activation of HSD2 neurons, induced by seven days of LSD, was suppressed by a brief period of high-sodium intake (Fig. 2D,E).

Fig. 2.

Dietary sodium deprivation increases c-Fos expression in hydroxysteroid dehydrogenase type 2 (HSD2) neurons. (A) Distribution of Enhanced Green fluorescent protein (EGFP)-labeled HSD2 neurons. Yellow areas indicating nucleus tractus solitarii (NTS) regions at different Bregma levels. 10N, vagus nerve nucleus; AP, area postrema; CC, central canal; 4V, fourth ventricle. n = 6, Scale bar = 20 µm. (B) Count of EGFP-labeled neurons across various Bregma levels. (C) Representative images showing c-Fos (red) expression in HSD2 neurons (green). Scale bar = 50 µm. (D) Distribution of c-Fos-positive neurons among EGFP-labeled neurons at varying Bregma levels. n = 6 mice per group. (E) Percentage of c-Fos-activated HSD2 neurons (EGFP⁺c-Fos⁺ neurons/total GFP⁺ neurons). **p 0.001, ****p 0.0001. One-way ANOVA with post hoc Tukey test.

Salt preference dynamically shifts depending on physiological state. To verify that sodium deprivation can enhance sodium appetite in mice, we subjected them to one week of sodium deprivation (Fig. 1). 24-hour total fluid intake was determined by weighing the bottles. We found that NSD mice drank ~7 mL of water daily versus only ~1 mL of 1.5% NaCl, indicating a preference for water (Fig. 3A). Control mice maintained moderate salt avoidance (13.5% saline intake). Strikingly, the LSD group mice reversed this pattern, with saline constituting over 70% of total fluid intake. This nearly complete behavioral reversal resulted in significantly elevated sodium intake in LSD mice, demonstrating state-dependent fluid preference (Fig. 3A). To observe the rapid sodium preference behavior in mice with activated HSD2 neurons, the optogenetic virus AAV-EF1a-DIO-ChR2-mCherry was injected into the NTS of HSD11b2-Cre mice. Fluorescent imaging shows that Cre-mediated recombination activates ChR2-mCherry expression (Fig. 3B). In vivo optogenetic activation of HSD2 neurons in awake, freely moving mice was performed concurrently with a two-bottle choice test conducted in a lickometer-equipped cage (Fig. 3C, left). Upon optogenetic activation, it was observed that stimulated animals robustly consumed up to 1.5% NaCl solution (Fig. 3C, right).

Fig. 3.

LSD or optogenetic activation of HSD2 neurons increased sodium appetite. (A) 24-hour water intake (g) was calculated by directly weighing the water bottles before and after consumption. n = 9 in NSD (normal diet group) and n = 5 in LSD group mice. Data were analyzed using two-way ANOVA with post hoc Tukey test, ****p 0.0001. (B) Schematic representation of viral injection (left panel) and verification of ChR2-mCherry expression in the NTS (right panel). Fiber optic cannula implanted above the AP. Scale bar = 100 µm. (C) Schematic illustration of in vivo optogenetic activation of HSD2 neurons for the two-bottle choice test (left panel). The parameters used were blue light (473 nm) at 10 Hz with a 20 ms pulse width for 60 s at 10 mW. The number of licks for water or 1.5% NaCl during a one-minute session with blue light on or off (right panel, n = 6 mice). **p 0.001, unpaired t-test.

A comprehensive brain-wide projection of HSD2 neurons was reported by Gasparini et al. [12]. The brain-wide projections of HSD2 neurons were reaffirmed through immunohistochemical labeling by mCherry, ensuring sensitive detection. Given that ChR2 integrates into the cell membrane along the entire length of the axon, brain tissues were utilized from mice subjected to the optogenetic experiments described above. To enhance the visibility of axonal terminals, immunostaining was conducted with an anti-mCherry antibody in brain sections. The whole brain projection pattern is depicted in Fig. 4A. Our observations confirmed that HSD2 neurons extend axons into three primary target regions: the lateral parabrachial nucleus (PBcL), the pre-locus coeruleus (pre-LC), and the ventral lateral bed nucleus of the stria terminalis (vlBNST) (Fig. 4B,C). The pre-LC receives a dense innervation of HSD2 axon terminals, which flank and streak through the LC laterally and extend medially between the LC and Barrington’s nucleus. Forebrain projections traverse the medial forebrain bundle and preoptic area to converge at the anterior commissure, where they form a dense terminal field within a specific subregion of the vlBNST. Outside the pre-LC, PBcL and vlBNST HSD2 projections are sparse, with minimal extension to the ventral lateral periaqueductal gray (vlPAG) (Fig. 4D) and scattered terminals near the ventral tegmental area (VTA) (Fig. 4E) in the ventral midbrain. Sparse boutons are also observed in the parasubthalamic nucleus (PSTh) (Fig. 4F) and lateral hypothalamus (LH) (Fig. 4G), with minimal HSD2 axonal presence in the central amygdala (CeA) (Fig. 4H) and paraventricular nucleus of the hypothalamus (PVH) (Fig. 4I) rostral to these areas. This mapping confirms the specific neural pathways of HSD2 neurons.

Fig. 4.

Validation the efferent projections of HSD2 neurons. (A) Sagittal diagram summarizing the axonal projections of HSD2 neurons. (B) Schematic of the lateral parabrachial nucleus (PBcL) and pre-locus coeruleus (pre-LC) locations in the mouse brain atlas, and fluorescent images of HSD2 axons (mCherry⁺, red) in the PBcL (left) and pre-LC (right). Tyrosine hydroxylase (TH)-positive neurons (blue) indicating the LC region. Sections were immunostained with an anti-mCherry antibody to enhance visibility. (C–I) Fluorescent images of HSD2 axons (mCherry⁺, red) in the ventral lateral part of the bed nucleus of the stria terminalis (vlBNST) (C), the ventrolateral periaqueductal gray (vlPAG) (D), the ventral tegmental area (VTA) (E), the parasubthalamic nucleus (PSTh) (F), lateral hypothalamus (LH) (G), the central amygdala (CeA) (H), and paraventricular nucleus of the hypothalamus (PVH) (I). Schematic drawing of coronal brain slices is shown in the left side of each fluorescence image (the specific nucleus area is indicated in yellow). Arginine vasopressin (AVP, green) denotes the location of the PVH.

To probe LSD-driven mechanisms in HSD2 neurons, we re-analyzed scRNA-seq data. This analysis expanded upon the foundational work of Resch et al. [11], by leveraging RNA-Seq data retrieved from the Gene Expression Omnibus database under the accession code GEO: GSE102332. In their experiments, mice were fed either a standard chow (0.3% Na⁺) or LSD (0.02% Na⁺) for 8–12 days. Subsequently, HSD2 neurons were manually picked from both the control (n = 33) and Na⁺ deficient (n = 32) groups for RNA profiling. Using the region-specific NTS marker Phox2b along with specific neuronal markers HSD2 (Hsd11b2) and MR (Nr3c2), it was confirmed that HSD2 neurons were successfully sequenced. Post the implementation of quality control protocols, we identified a cohort of 1193 genes that exhibited differential expression patterns. Specifically, 1079 genes were upregulated and 114 were downregulated in the LSD group when compared to the control group ($|$log2 fold change$|$ $>$1, p 0.05, Fig. 5A).

Fig. 5.

Re-analysis of RNA-seq data reveals enrichment of mitogen-activated protein kinase (MAPK) and cyclic adenosine monophosphate (cAMP) signaling pathways in HSD2 neurons under LSD conditions. (A) Volcano plot illustrating genes affected by sodium deprivation. Dots (red) outside the gray-shaded area, indicate genes that are significantly upregulated (1079), while dots (green) indicate genes that are downregulated (114) in response to sodium deprivation ($|$log2 fold change$|$ $>$1 and p 0.05). (B) Bubble plots of the top 10 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways enriched by the differential genes. (C) Heatmap showing single-neuron expression patterns for MAPK signaling pathway genes in HSD2 neurons from Chow- and LSD-fed mice. (D) Heatmap showing single-neuron expression patterns for cAMP signaling pathway genes in HSD2 neurons from chow- and LSD-fed mice. (E) Log2 fold change values for each gene enriched in the MAPK signaling pathway. (F) log2 fold change values for each gene enriched in the cAMP signaling pathway. See also Supplementary Table 2.

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of differentially expressed genes identified the axon guidance pathway as the most significantly enriched, followed by mitogen-activated protein kinase (MAPK) and cyclic adenosine monophosphate (cAMP) signaling pathways (Fig. 5A). A volcano plot distinctly marked all differentially expressed genes associated with these pathways (Fig. 5B). Detailed examination revealed complete upregulation of all 19 genes in the MAPK pathway (Fig. 5C) and predominant upregulation in the cAMP pathway (15 of 17 genes upregulated, 2 downregulated; Fig. 5D). Quantitative expression changes for these genes are presented in Fig. 5E (MAPK pathway) and Fig. 5F (cAMP pathway), demonstrating significant pathway-specific alterations. These findings highlight the potential involvement of axon guidance and intracellular signaling cascades in regulating the observed biological processes.

Gene Ontology (GO) analysis of differentially expressed genes in HSD2 neurons revealed key functional implications (Fig. 6). Enriched terms included “inorganic anion transport” (GO:0015698), critical for maintaining neuronal ion homeostasis and regulating membrane excitability; “postsynaptic membrane organization” (GO:0001941), indicating potential modifications in synaptic structure and receptor distribution; “glutamate receptor binding” (GO:0035254), suggesting altered excitatory input strength; and “channel activity” (GO:0015267), reflecting potential changes in ion channel function. These findings collectively indicate that the identified genes may significantly modulate HSD2 neuron excitability through coordinated effects on ion transport, synaptic architecture, and receptor signaling, with likely consequences for neural circuit function and behavior.

Fig. 6.

Single-cell transcriptomic re-analysis of HSD2 neurons during sodium deficiency. (A) Statistically significant gene ontology (GO) terms (David database) of genes upregulated by sodium-deficient conditions. BP, biological process; CC, cellular compartment; MF, molecular function. (B) Bubble Chart of GO Terms.

In conclusion, the LSD-associated excitation of HSD2 neurons may involve transcriptional changes that increase cAMP and activate cAMP dependent protein kinase A (PKA), potentially activating the MAPK pathway and modulating neuronal excitability and sodium appetite. Although aldosterone/MR signaling is a plausible mechanism, further studies are required to confirm this.

To determine whether MAPK and cAMP pathways mediate LSD-induced sodium appetite, cannulas were implanted into the fourth ventricle of LSD-fed mice and either the MAPK inhibitor (U0126) or cAMP inhibitor (KH7) were administered, with aCSF as control. Sodium preference was then assessed using two-bottle choice tests. Cannula placement accuracy was verified post-hoc by blue dye injection (Fig. 7A). Pharmacological inhibition of key signaling pathways significantly altered sodium preference in LSD-fed mice (Fig. 7B). Control (aCSF-treated) animals showed strong saline preference (6.87 $\pm$ 0.55 mL 1.5% NaCl vs. 2.43 $\pm$ 0.53 mL water). Both KH7 and U0126 reduced this preference, with 1.5% NaCl intake reduced to 2.81 $\pm$ 1.1 mL and 3.56 $\pm$ 0.79 mL respectively, with corresponding water intake being 3.56 $\pm$ 0.79 mL and 3.81 $\pm$ 0.64 mL. Consistent with behavioral results, c-Fos expression in HSD2 neurons was markedly reduced following KH7 or U0126 treatment compared to aCSF controls (Fig. 7C,D). These findings demonstrate that LSD-induced sodium intake may involve intact cAMP and MAPK signaling in HSD2 neurons, revealing a molecular mechanism through which these neurons detect hormonal signals to regulate sodium intake.

Fig. 7.

Disruption of sodium appetite by inhibiting MAPK or cAMP pathways in mice on LSD. (A) A diagram illustrates the cannula placement in the i4V for drug delivery. Blue areas in brain sections show precise Pontamine Sky Blue injection into the i4V. (B) Bar graphs showing the volume of 1.5% NaCl solution and water consumed by mice in the aCSF (n = 6), KH7 (n = 8), and U0126 (n = 9) groups over 24 hours. Data are presented as mean $\pm$ SEM. Unpaired t test, ****p 0.0001, ns, no significance difference. (C) Bar graph showing the percentage of c-Fos-positive cells in the HSD2 neurons of mice in the aCSF, KH7, and U0126 groups. n = 4 in each group. Data are presented as mean $\pm$ SEM. One-way ANOVA with Tukey’s multiple comparisons test, ****p 0.0001. (D) Representative immunofluorescence images showing the expression of HSD2 (green), c-Fos (red), and their overlay (yellow) in the NTS of mice. Scale bars = 50 µm. (E) Schematic of the experimental timeline and summary of results for the effects of MAPK and cAMP pathway blockade on sodium appetite in mice. ns, no statistically significant difference, p $>$ 0.05.

In summary, the MAPK and cAMP pathways are integral to the molecular mechanisms that underlie the activation of HSD2 neurons in response to low-sodium conditions. These pathways likely work in concert to regulate the neuronal activity and subsequent behavioral response that drives sodium intake in mice (Fig. 7E).

The knowledge regarding genes selectively targeted by aldosterone/MR in the brain is still incomplete. To address this, the currently reported canonical targets of aldosterone/MR signaling were analyzed (Fig. 8A,B). It was confirmed that known target genes of theoretical interest, such as the aldosterone-inducible subunits of the ENaC, including Scnn1a, Scnn1b and Scnn1g, were absent in both control and sodium-deprived HSD2 neurons. Interestingly, HSD2 neurons did express Sgk1, which is known to mediate aldosterone-induced trafficking of ion channels in the kidney. However, its expression was not upregulated in neurons from sodium-deficient mice. Among the known aldosterone/MR target genes expressed in HSD2 neurons, it was confirmed that two were found to be significantly upregulated under sodium-deficient conditions: WNK lysine-deficient protein kinase 1 (Wnk1) and the $\alpha$1 subunit of the Na⁺/K⁺-ATPase pump (Atp1a1). Other mediators of rapid aldosterone effects, including GPR30, epidermal growth factor receptor (EGFR), the insulin-like growth factor 1 receptor (IGF1R) and platelet-derived growth factor receptor (PDGFR), were also investigated. Their expression was either undetectable in HSD2 neurons or remained largely unchanged by sodium deficiency. This suggests that the molecular mechanisms underlying aldosterone activation of HSD2 neurons under low sodium conditions are distinct from peripheral mechanisms and cannot be fully elucidated by the currently known molecular mechanisms.

Fig. 8.

Re-analysis of RNA-seq data confirms altered expression of mineralocorticoid receptor (MR)-associated genes in HSD2 neurons under LSD conditions. (A) Volcano plot illustrating the differential expression of MR-regulated genes upon sodium deprivation. Genes represented by color dots exhibit significant expression changes (log2 fold change $>$1 or –1, p 0.05). Colored dots denote downregulated genes (blue dots), upregulated genes (red dots) and no significant change (grey dots) in expression levels. (B) Heatmap depicting single neuron expression patterns for MR-regulated genes. The color gradient from yellow to purple corresponds to low to high expression levels. Values are expressed in log2 counts per million (Log2 CPM).

HSD2 neurons location in the NTS, which plays a crucial role in regulating cardiorespiratory functions [28], would make it reasonable to hypothesize that HSD2 neurons regulate BP and breathing. To test this, an optogenetic virus was injected into the NTS of HSD11b2-Cre mice (Fig. 9A). Under anesthesia, these neurons were blue light -activated these neurons and monitored arterial BP and PND. This investigation will aim to reveal whether HSD2 neuron activation influenced these physiological parameters, offering insights into their broader regulatory functions beyond sodium appetite. Fig. 9B illustrate the immunofluorescence validation of ChR2-mCherry expression in NTS. Results revealed that blue light activation of HSD2 neurons at various frequencies had no significant impact on the arterial BP and HR of the mice (Fig. 9C,D). Similarly, the activation of HSD2 neurons did not significantly alter the frequency and amplitude of PNDs (Fig. 9E–G). To confirm that optogenetic stimulation indeed activated HSD2 neurons, mice were perfused with 4% PFA one hour after blue light (10 Hz) activation, followed by immunofluorescent c-Fos staining. Analysis of results from three mice revealed that 68.1% of mCherry-positive neurons were also c-Fos positive (Fig. 9H), demonstrating that optogenetic stimulation effectively activated a majority of HSD2 neurons. Taken together these results indicate that HSD2 neurons, as activated in the current experimental paradigm, do not appear to coordinate autonomic regulation of cardiovascular and respiratory functions.

Fig. 9.

Effect of activation HSD2 neurons on cardiovascular parameters and phrenic nerve discharge (PND). (A) Schematic of AAV-DIO-ChR2-mCherry injections. (B) Validation of ChR2-mCherry expression in NTS. Scale bars = 100 µm. (C) Typical traces of blood pressure (BP) and heart rate (HR) in response to 1, 5, 10, 15 and 20 Hz light frequencies stimulation (10 ms pulse width, 10 mW, 60 s). Scale bar = 30 seconds. (D) Changes in systolic BP, diastolic BP, and HR from BP recordings following optogenetic stimulation at various frequencies. The sample sizes (n) for 1 Hz, 5 Hz, 10 Hz, 15 Hz, and 20 Hz are 8, 6, 6, 5, and 9 respectively. (E) Optogenetic activation and schematic of PND recordings in anesthetized mice. (F) Representative traces showing the effect of 1 Hz, 5 Hz, 10 Hz, 15 Hz and 20 Hz photostimulation (10 ms pulse width, 10 mW, 60 s) on PND. (top) End-tidal CO₂ (ETCO₂); (middle) raw waveforms of PND; (bottom) PND integration ($\int$PND) derived from rectification and smoothing (time constant, 0.05 s). Scale bar = 30 seconds. (G) Changes in frequency (f) and amplitude of $\int$PND, and the product f*$\int$PND following optogenetic stimulation. The sample sizes (n) for 1 Hz, 5 Hz, 10 Hz, 15 Hz, and 20 Hz are 5, 4, 3, 6, and 7, respectively. (H) Immunofluorescence validation of optogenetic activation of ChR2-mCherry neurons and quantification of activation ratio. (Left) Representative immunofluorescence images of ChR2-mCherry (red) and c-Fos (green) in the NTS. (Right) quantification of activation ratio (n = 3 mice). Scale bars = 50 µm. Data were analyzed using paired t-tests. ns, no statistically significant difference, p $>$ 0.05.