Thirst-associated preoptic neurons encode an aversive motivational drive

Abstract

Water deprivation produces a drive to seek and consume water. How neural activity creates this motivation remains poorly understood. We used activity-dependent genetic labeling to characterize neurons activated by water deprivation in the hypothalamic median preoptic nucleus (MnPO). Single-cell transcriptional profiling revealed that dehydration-activated MnPO neurons consist of a single excitatory cell type. After optogenetic activation of these neurons, mice drank water and performed an operant lever-pressing task for water reward with rates that scaled with stimulation frequency. This stimulation was aversive, and instrumentally pausing stimulation could reinforce lever-pressing. Activity of these neurons gradually decreased over the course of an operant session. Thus, the activity of dehydration-activated MnPO neurons establishes a scalable, persistent, and aversive internal state that dynamically controls thirst-motivated behavior.

Fig. 1. TRAP2 efficiently and specifically labels dehydration-activated neurons in the median preoptic nucleus (MnPO)

(A and B) TRAP2 design and principle. (C) Experimental timelines to determine the efficiency and specificity of Thirst-TRAP, by comparing Thirst- or Homecage-TRAP tdTomato expression with Fos immunolabeling in response to 48 hours of water deprivation (Thirst-Fos) or 4 hours at 37°C (Warm-Fos). (D) tdTomato expression in MnPO after recombination of TRAP2;Ai14 mice, under water-satiated (Homecage-TRAP) and 48-hours water-deprived (Thirst-TRAP) conditions. a.c., anterior commissure; D, dorsal;V, ventral; M, medial; L, lateral. (E) Representative confocal images of tdTomato/Fos overlap for three conditions, as indicated above. (F) Quantification of total tdTomato induction in MnPO after recombination in Homecage- or Thirst-TRAP mice; t test. (G) Efficiency and specificity of MnPO Thirst-TRAP. (H) TRAP/Fos overlap (Double⁺/Fos⁺) for the three experimental groups, one-way ANOVA, Holm-Šidák correction. Numbers of mice quantified for each experiment are in parentheses. **P < 0.01, ****P < 1 × 10⁻⁴. Data are presented as mean ± SEM.

Fig. 2. Molecular identity of MnPO Thirst-TRAPed neurons

(A) Single-cell RNA-sequencing of MnPO Thirst-TRAPed neurons. Tissue containing MnPO of Thirst-TRAP mice is microdissected from live brain slices. Neurons are dissociated then FAC-sorted into 96-well plates. Single-cell cDNA libraries are prepared and sequenced together. (B) t-distributed stochastic neighbor embedding (t-SNE) representation of 348 transcriptomes showing two clusters. (C) Top 30 differentially expressed genes between the two clusters, per cluster, sorted by average difference in expression per cluster and Z-scored per gene. (D and E) (Top) t-SNE representation of cells colored by Gad1 (D) or Slc17a6 (E) expression. (Bottom) Expression of Gad1 or Slc17a6 in cells within Cluster 1 (purple) and Cluster 2 (green). Expression, log-normalized TPM (transcripts per million) values (see methods). (F) Three-color smFISH of tdTomato (red), Gad1 (blue), and Slc17a6 (green) in the MnPO (dotted oval) of a Thirst-TRAP mouse. (Left) Low-magnification view; two fields of view highlighted. (Right) High-magnification view of tdTomato⁺ (red) cells outside MnPO (1) or within MnPO (2), along with Gad1 and Slc17a6 expression in the same cells, indicated with arrows. (G) Fraction of tdTomato⁺Slc17a6⁺ and tdTomato⁺Gad1⁺ neurons within MnPO, out of total tdTomato⁺ cells. N = 2 mice. tdT, tdTomato. (H to J) Double smFISH of tdTomato and Cluster 2 markers, Adcyap1, Agtr1a, or Nxph4, in Thirst-TRAP mice. (Top left) Low-magnification view of marker and tdTomato with MnPO circled. (Top right) High-magnification view of tdTomato (red) and marker gene (green) expression. Arrows, double-labeled cells. (Bottom left) t-SNE representation of cells colored by expression of marker gene. (Bottom right) Fractional combinations of tdTomato⁺ with each marker gene. Adcyap1: N = 276 total cells; Agtr1a: 396 total cells; Nxph4: 402 total cells, across N = 3 mice per marker gene.

Fig. 3. Optogenetic activation of MnPO Thirst-TRAPed neurons induces a scalable, aversive thirst motivational state

(A) Experimental timeline for Thirst- and Homecage-TRAP to express a virally delivered effector. (B) (Left) Fiber-optic implant for illuminating MnPO. (Right) Viral constructs for Cre-inducible iC++ or ChR2 expression. (C) Total licks in water-restricted mice expressing iC++ in MnPO during 7.5-min laser-ON or laser-OFF session; paired t test. (D) Licking for water after 20-Hz stimulation (blue) of Thirst-TRAPed MnPO neurons expressing ChR2 in 5 water-satiated mice. (E and F) Total licks over 15-min prestimulation or stimulation period of N = 5 Thirst- (E) or Homecage- (F) TRAPed MnPO neurons at 2.5, 5, 10, and 20 Hz. (G) Cumulative reinforcements (water rewards) over 20 min, before, during (blue), and after 20-Hz stimulation, after training. N = 5 animals per group. (H and I) Total reinforcements within 15 min prestimulation or stimulation period of N = 5 water-satiated Thirst- (H) or Homecage- (I) TRAPed MnPO neurons at 2.5, 5, 10, and 20 Hz. (J) Real-time place preference (RTPP) to 20-Hz stimulation. (K) Aggregate RTPP data of difference in time on nonstimulated and stimulated sides; unpaired t test. (L) Reinforcements per 30-min session of mice learning to lever-press to receive a 20-s break in otherwise constant 20-Hz stimulation. Two-way ANOVA, Holm-Šidák correction. (M) Cumulative reinforcements over 30-min session. N = 5 Homecage-(gray) and N = 5 Thirst- (green) TRAP mice. (N) Total reinforcements within a 30-min session for Thirst-TRAP mice lever-pressing to turn off 0-, 5-, 10-, and 20-Hz MnPO stimulation; Kruskal-Wallis test, false discovery rate correction. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant. Data presented as mean ± SEM.

Fig. 4. MnPO thirst-associated neurons integrate water intake to control motivational drive

(A) Diagram of possible models relating MnPO activity dynamics and behavior. Behavioral trace represents idealized cumulative reinforcements as a mouse lever-presses for water. (B) Diagram of MnPO fiber photometry recordings with multiplexed Ca²⁺-independent control (405 nm) and Ca²⁺-dependent (490 nm) illumination. (C) Coexpression of GCaMP6s and tdTomato in Thirst-TRAP MnPO. (D) Single-trial MnPO activity dynamics upon drinking water in a water-restricted Thirst-TRAP mouse. (E) (Top) Average MnPO activity upon licking full (green) or empty (black) water bottle. (Bottom) Lick rate over time. N = 5 mice. (F) Single-trial MnPO activity dynamics during FR1 lever-pressing in water-restricted Thirst-TRAP mouse. (G) (Top) Average MnPO activity during task performance on a FR1 (green) or Extinction (black) schedule. (Extinction, no water delivered.) (Middle) Average licking behavior. (Bottom) Average cumulative reinforcements. N = 5 mice, averaged across N = 3 sessions per mouse. (H) Same as (G) but for mouse on FR6 schedule. (I) Quantification of average fluorescence during 5-min interval from T = 10 min to T = 15 min, after presentation of empty bottle or water bottle (E), or after onset of Extinction and FR1 schedule lever-pressing for water (G); paired t test. (J and K) Lick rate (J) or Z-scored MnPO activity (K) in 2-min bins for free drinking from water bottle (E), FR1 lever-pressing (G), and FR6 lever-pressing (H). (L) Time for activity to decrease to half minimal value for each mouse, based on sigmoid fit; two-way ANOVA, Benjamini-Hochberg correction. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant. Data presented as mean ± SEM.