Sex as a biological variable. Sex was not considered as a biological variable in these studies. Due to increased variability of sodium appetite in female rodents (56), we performed all experiments in male mice.

Human, porcine, and rat brainstem tissue. We obtained the caudal brainstem from 12 human brains removed after death from a variety of nonneurologic causes (Table 1). The postmortem interval between death and autopsy was 33.5 ± 18 hours (mean ± SD; range 15–50 hours). Average age at death was 35 ± 29.6 years (range: 33 gestational weeks to 77 years). Sex was female in 2 cases and male in 9. After fixing for 2–3 days at 4°C in 10% formalin-PBS (SF100-20, Thermo Fisher Scientific), we cryoprotected the brainstem in 30% sucrose-PBS for an additional 1–2 days until it sank. Next, we cut 40 μm–thick tissue sections in the transaxial plane using a freezing-stage microtome. We collected tissue in a separate 1-in-12 series with approximately 0.5 mm between successive sections in each series, which were stored in cryoprotectant at –30°C. We used the same protocol to process brainstem tissue, gifted to us from Michael Welsh at the University of Iowa, from 4 pigs that were euthanized and autopsied after previous research studies. For rat brainstem tissue, we obtained unpublished images from slides prepared and described in a previous publication (84).

Human postmortem lower brainstem tissue used for HSD2 immunolabeling and (in a subset of cases) HSD11B2 in situ hybridization

Mice. All mice were group-housed in a temperature- and humidity-controlled room with a 12/12-hour light/dark cycle and, initially, ad libitum access to water and standard rodent chow (Envigo 7013). Some were provided low-sodium chow (TD-130591, Teklad/Envigo) for 1 or more days in protocols described below. In addition to C57BL/6J mice, we used hemizygous Hsd11b2-Cre or Th-IRES-Cre mice maintained on a C57BL/6J background (Table 2).

Stereotaxic injections. To selectively ablate Cre-expressing neurons in the NTS, we used a head-down stereotaxic approach (82, 85) to inject AAV-DIO-dtA-mCherry [AAV1-Ef1a-Lox-Cherry-lox(dtA)-lox2.ape, 5.2 × 1012 vg/mL; UNC Gene Therapy Center Vector Core]. After post hoc histology (described below), we included all experimental-ablation mice for analysis if they had fewer than 50% of the average number of HSD2 neurons counted control mice. To chemogenetically stimulate HSD2 neurons, we instead injected AAV-DIO-hM3Dq-mCherry (AAV1-Ef1a-DIO-hM3Dq-mCherry, 4 × 1012 vg/mL; Duke Viral Vector Core). In additional mice with CNO continuous infusion pumps, we used AAV8-DIO-hM3Dq-mCherry (44361-AAV8, Addgene; 2.1 × 1013 gc/mL), and control mice received AAV8-DIO-mCherry (50459-AAV8, Addgene; 2.2 × 1013 gc/mL).

Cannula implantation and s.c. osmotic minipumps. Under anesthesia, the scalp was shaved, and the mouse was placed gently in the stereotaxic ear bars. A scalp incision was made wide enough to expose the skull at both bregma and lambda. After leveling the skull (both AP and ML) and zeroing at bregma, a 1 mm burr hole was made using a high-speed rotary tool with a 0.8 mm bit. An L-shaped 28-gauge cannula (3280PM-SPC, 5 mm; Plastics One) was attached by tubing (0.69 mm ID × 1.14 mm OD; Scientific Commodities) to an osmotic minipump (ALZET 1007D) then inserted through the burr hole to the target coordinate (i4V: 0 mm lateral, –6.5 mm caudal, and 5.0 mm deep to bregma for i4V; lateral ventricle [LV]: 0.8 mm right, 0.35 mm caudal, and 2.5 mm deep to bregma). Cyanoacrylate glue with dental cement was used to fix the cannula. After the cannula was firmly cemented in place, the minipump was inserted into a s.c. pocket in the midscapular area. To infuse aldosterone peripherally, we placed the osmotic minipump in the same s.c. position without connecting any cannula or tubing. The skin was closed using Vetbond (3M). In every mouse with an i4V or LV cannula, under anesthesia and immediately before transcardial perfusion, we disconnected the catheter from the minipump and used a 28 gauge syringe to inject Evans blue dye (1% in sterile water) to check cannula patency. We excluded from analysis all mice without blue dye in the i4V or LV.

BioDAQ fluid intake measurement. Mice were housed individually in BioDAQ cages (Research Diets) as described previously (39). Each cage had fresh ALPHA-dri+ Plus bedding (Shepherd Specialty Papers), an overhead hopper with standard rodent chow (Teklad 7013), and 2 bottles (distilled water and 3% NaCl) that could be blocked by a gate. Fluid intake was recorded continuously, but we analyzed data from prespecified intervals, as detailed below.

Dietary sodium deprivation and potassium supplementation. We habituated mice in individual BioDAQ cages with low-sodium chow and ad libitum access to water and 3% NaCl for at least 3 days. We then closed the access gate for 3% NaCl within 30 minutes before lights-off. To prevent consumption of excreted sodium, the mouse was provided with a fresh, clean cage every day. Six days later, we reopened the saline access gate 10 minutes before lights-off and measured 3% NaCl and water intakes across the following 6 hours.

For dietary sodium deprivation plus potassium supplementation, we modeled our approach after a similar dietary protocol followed by acute infusion of potassium chloride (KCl) in patients (40). However, instead of i.v. infusion at the end of our protocol, we gave a bolus of potassium by gavage (3M KCl, 0.1 mL per 10 g), similar to previous rodent experiments (86). Also, after pilot experiments testing a range of concentrations, we found that mice show no aversion to 1.5% or less KCl in drinking water (they consumed this solution at the same daily volumes as dH2O). Thus, our protocol for dietary sodium deprivation plus potassium supplementation was ad libitum access to 1.5% KCl in the drinking water, plus the same low-sodium chow as above, and fresh cage changes. After 6 days, these mice (n = 8) were perfused immediately following CSF and blood collection.

Aldosterone dose-response testing. We infused aldosterone using ALZET 1007D osmotic minipumps, which deliver 0.5 μL/h. The day before implanting, we preloaded each minipump with 100 μL of vehicle (1% ethanol in sterile 0.9% saline) or aldosterone solution and then placed it in sterile saline overnight. We tested separate doses in separate groups of mice, with each mouse receiving 1 dose.

For i4V infusions, we monitored saline and water intake for up to 10 days in separate groups receiving either vehicle (n = 4) or aldosterone 1 (n = 3), 2.5 (n = 2), 5 (n = 6), 10 (n = 5), 25 (n = 5), 50 (n = 4), or 100 (n = 5) ng/h. After identifying a peak effect at 10 ng/h aldosterone (i4V), we used an additional group of mice to infuse that dose into the lateral ventricle (n = 7). We discarded 3 days of data (2 presurgery, 1 day-of surgery) from 1 mouse (1 ng/h i4V aldosterone group) with aberrant readings caused by a loose fluid hopper.

For peripheral infusions, we monitored saline and water intake for up to 10 days in separate groups receiving either vehicle (n = 4) or aldosterone 10 (n = 2), 100 (n = 3), 250 (n = 3), 500 (n = 5), 750 (n = 3), or 1,000 (n = 6) ng/h.

Cell-type–specific neuronal stimulation and ablation. Four weeks after injecting AAV-DIO-hM3Dq-mCherry into the NTS of Hsd11b2-Cre mice, and after at least 3 days of habituation to individual BioDAQ cages, we injected mice 30 minutes prior to lights-off with either CNO (1 mg/kg in sterile 0.9% NaCl with 1% DMSO) or sterile 0.9 % NaCl (0.1 mL per 10 g mouse). After at least 3 additional days, we repeated the injection with the other solution (vehicle or CNO). We analyzed water and 3% NaCl intake across the 6 hours after lights-off. In additional cohorts, we habituated mice to individual BioDAQ cages for 5 days and then implanted an ALZET 1007D minipump (delivering CNO 0.5 mg/kg/h s.c.) and continued monitoring water and 3% NaCl intake for 7d. At the end of each experiment, we perfused each mouse for histological assessment of HSD2 neuron transduction.

In separate cohorts, we stereotaxically microinjected AAV-DIO-dtA-mCherry in Hsd11b2-Cre experimental mice (n = 19 i4V aldosterone; n = 12 s.c. aldosterone) and in Cre– littermates as controls (n = 10 i4V aldosterone; n = 6 s.c. aldosterone). Four weeks later, mice were habituated in BioDAQ cages for at least 3 days. We then implanted osmotic minipumps, returned each mouse to its BioDAQ cage, and monitored fluid intake for 9 more days. We repeated this protocol with s.c. aldosterone infusion in Th-IRES-Cre (n = 8) and Cre– littermate control mice (n = 11).

Metabolic cage studies. To test whether peripheral aldosterone infusion increased urinary fluid loss, we used 4 groups of mice. Two received aldosterone (1,000 ng/h s.c.) and 2 received vehicle. Mice were habituated for 3d in individual metabolic cages with ad libitum food and water access (Ugo Basile, 41700). Urine was collected, and water and food intake were measured daily. After s.c. minipump implantation, mice were returned to the metabolic cages for 10d. Two groups (aldosterone and vehicle, n = 4 each) had ad libitum water access, and the other 2 (aldosterone and vehicle, n = 4 each) received 5 mL water — which is slightly more than vehicle-infused mice drank but less than s.c. aldosterone-infused mice drank in our initial BioDAQ recordings — every day for 7d; they then gained ad libitum water access for the final 3d. In addition to expected evaporative losses, 2–3 urine samples from 2 mice in the 5 mL water–restricted experiment had fecal contamination, requiring removal of 2 data points from animal 7863 (days 6 and 10) and 3 data points from animal 7864 (days 1, 5, and 6).

Measurements in blood plasma and CSF. After 6 days of aldosterone infusion, or dietary sodium deprivation with or without potassium supplementation, we collected blood and CSF. Prior to collection, mice receiving s.c. infusion (vehicle and 250, 500, and 1,000 ng/h aldosterone) were housed in standard cages with ad libitum access to regular chow and water. Mice receiving i4V infusion (vehicle and 5 and 10 ng/h aldosterone) were monitored in BioDAQ cages with ad libitum access to regular chow, dH2O, and 3% NaCl. Because CSF collection precluded our blue-dye assay for i4V cannula patency, we used a behavioral assay for cannula patency and only collected blood and CSF from mice that were consuming at least 1 mL of 3% NaCl per day by d6, which was at the low end of our dose-response results with 5 and 10 ng/h i4V aldosterone infusion (above). Mice consuming little to no 3% NaCl were presumed to have a nonpatent cannula and euthanized without blood or CSF collection.

To collect CSF, we used a pulled glass capillary pipette (80–100 μm tip inner diameter). We anesthetized each mouse and, using the NTS surgical exposure described above, inserted the tip slowly through the atlantooccipital membrane until CSF flowed spontaneously from the cisterna magna into the pipette, which was left in place for 10–30 minutes. CSF volumes collected from most mice ranged from 3 to 12 μL. After slowly removing the pipette, CSF was ejected from the pipette into a 200 μL tube by inserting a 22 gauge blunt needle and using an attached 10 mL syringe to push air into the pipette. CSF was stored at –80°C.

Immediately after collecting CSF, we injected ketamine/xylazine for terminal blood collection and euthanasia. After 5 minutes, we opened the peritoneum and collected blood from the inferior vena cava with an insulin syringe/needle. Blood was ejected from the insulin syringe into a microcentrifuge tube before being spun for 5 minutes at 6,000g at 4°C. Serum was extracted using a pipettor and stored at –80°C.

We used an ELISA to measure the aldosterone concentration (IB79134, IBL) in mouse blood plasma and pooled CSF. After thawing and diluting samples (1:10 for aldosterone), we followed the standard instructions for each commercially available sandwich ELISA kit. In a subset of mice receiving s.c. infusion of aldosterone (1,000 ng/h) or vehicle, we used an ion-sensitive electrode to measure serum sodium (Prolyte; Diamond Diagnostics). To measure plasma osmolality, we also used an osmometer. In an additional cohort of mice receiving s.c. infusion of aldosterone (1,000 ng/h) or vehicle, we provided ad libitum access to food and water (no saline), for 7d; then, we collected blood from the mandibular vein and centrifuged as above, extracted, and stored serum at –80°C. In these samples, we measured copeptin with an ELISA kit (MBS160428, MyBioSource) used in previous work in mice (87). We also measured blood glucose using a glucometer (FreeStyle Lite, Abbott).

Transcardial perfusion and tissue sectioning. All mice with stereotaxic AAV injection or implanted cannulas and a subset of mice with s.c. infusion pumps were transcardially perfused as follows. First, each mouse was anesthetized with ketamine (150 mg/kg) and xylazine (15 mg/kg), dissolved in sterile 0.9% saline, and injected i.p. It was then perfused transcardially with phosphate-buffered saline (PBS) prepared from 10× stock (P7059, MilliporeSigma), followed by 10% formalin-PBS (SF100-20, Thermo Fisher Scientific). After perfusion, we removed each brain and fixed it overnight in 10% formalin-PBS. We sectioned each brain into 40 μm–thick coronal slices using a freezing microtome and collected tissue sections into separate, 1-in-3 series. We stored all tissue sections in cryoprotectant solution at –30°C.

Immunolabeling and in situ hybridization. To label HSD2 in human and porcine brain tissue sections, we used a rabbit polyclonal antiserum (Table 3) and nickel-diaminobenzene (NiDAB) IHC as described in previous work (28). For human cases, we selected 16–20 total sections from one 1-in-12 series spanning approximately 1 cm of the caudal medulla oblongata. This tissue series contained intermediate through caudal levels of the human NTS, as well as the cuneate, gracile, and spinal trigeminal nuclei, plus the caudal inferior olivary complex and pyramidal tracts back through the spinomedullary transition. For immunofluorescence labeling, we used previously described protocols (28, 39, 88) and primary antisera in Table 3. To label mRNA for HSD11B2, we selected adjacent tissue sections from 5 human cases with immunolabeling for HSD2. We then used the RNAscope 2.5 HD Detection Reagent-BROWN (322310; Advanced Cell Diagnostics [ACD]) and a probe for human HSD11B2 mRNA (432351; ACD) with a previously described protocol (39).

For bright-field analysis in human tissue, we identified HSD2-immunoreactive and neuromelanin-containing neurons throughout each tissue section in VS-ASW. We selected 3 adult brainstems with the highest tissue quality (MH001, MH004, and MH005), we counted every HSD2-immunoreactive neuron that contained a nuclear void and measured its short-axis diameter through the center of its nucleus (in μm). We used the average diameter to perform Abercrombie correction (49) and multiplied this number by the section interval (×12 for 1-in-12 series) to estimate the total number of HSD2 neurons.

For epifluorescence analysis in mouse tissue, we used QuPath (89) to count all HSD2-immunoreactive neurons and catecholaminergic NTS neurons, which are immunoreactive for the enzyme TH. We excluded 5 Th-IRES-Cre– littermate control mice due to tears in the dorsal hindbrain, which occurred during brainstem removal and caused histological artifacts preventing analysis of the full caudal NTS.

Statistics. We exported continuous fluid intake records from the BioDAQ DataViewer (Research Diets) and then used Microsoft Excel to organize data and to calculate total intake volumes. After acute chemogenetic stimulation with CNO and after week-long dietary sodium deprivation, we analyzed the first 6 hours of access to 3% NaCl. For all multiday infusion protocols, we analyzed 3% NaCl and water intake in 24-hour bins. We then used GraphPad Prism to plot data and run statistical tests, which are described in figure legends. To compare dose-response effects of aldosterone on 3% NaCl and water intake, we calculated AUC across 9d following osmotic minipump implantation and then used 1-way ANOVA followed by Tukey’s multiple-comparison correction. To assess the effect of Cre-conditional neuronal ablation on aldosterone-induced intake of 3% NaCl and water, CNO effects on sodium and water intake, and sodium and water intake induced by sodium deprivation, we used unpaired, 2-tailed t tests. We also used unpaired, 2-tailed t tests to compare immunolabeled cell counts between individual groups of experimental and control mice and to compare sodium, copeptin, and glucose measurements between aldosterone-infused and control mice. To compare average daily water intake and average daily urine output between aldosterone- and vehicle-infused mice with restricted and unrestricted water access in metabolic cages, we used repeated-measures 2-way ANOVA. All results are expressed as mean ± SD. We considered P < 0.05 statically significant.

Study approval. All procedures involving animals were conducted in accordance with the guidelines of the IACUC at the University of Iowa (protocol nos. 00720011, 3072011, and 3102343). Human tissue procurement protocols were reviewed by the University of Iowa’s IRB and determined not to represent patient research under the Revised Common Rule. Consent for research tissue donation was obtained from the next of kin by the University of Iowa Department of Pathology in accordance with federal and Iowa law. In accordance with Iowa law, no tissue from elective terminations was used for this project.

Data availability. The data that support the findings of this study are in the Supporting Data Values file and are also available from the corresponding author upon reasonable request.