Comparative transcriptomics of lateral hypothalamic cell types reveals conserved growth hormone-tachykinin dynamics in feeding

Abstract

Objectives: The lateral hypothalamus (LH) plays a central role in appetite control, however the functional and evolutionary conservation of its subcircuits remain unclear. This study aimed to define the molecular and cellular identities of zebrafish LH neurons, identify conserved LH cell types across vertebrates, and investigate their roles in appetite regulation.

Methods: We performed the Act-seq method of single-cell RNA sequencing in the larval zebrafish LH under food-deprived and voracious feeding states to capture activity-dependent transcriptional signatures. Using integrative comparative transcriptomics, we aligned zebrafish neuronal clusters with a published mouse LH dataset to identify conserved neuronal sub-populations, and performed functional and molecular characterisation of a highly-conserved cell type in both zebrafish and mice.

Results: We identified several LH neuronal subtypes in zebrafish that are differentially activated during voracious feeding. Cross-species mapping revealed overlapping cellular clusters, especially for GABAergic neurons. We report a conserved GABAergic cluster co-expressing growth hormone (GH) receptors and tachykinin. In both species, feeding activates these neurons and elevates GH receptor and tachykinin expression. In zebrafish, upstream GH signaling is similarly regulated by feeding state, and acute GH administration both activates this cluster and enhances food intake.

Conclusions: These findings uncover a conserved GH receptor-tachykinin LH population which may link metabolic hormone signaling to appetite control. Beyond its established long-term roles in growth and metabolism, we propose that GH exerts acute appetite-enhancing effects through activation of this neuronal pathway. Our comparative LH atlas highlights the evolutionary biology of hypothalamic appetite circuits.

Keywords: Appetite regulation; Comparative transcriptomics; Growth hormone; Lateral hypothalamus; Zebrafish.

Graphical abstract

Figure 1

Act-seq identifies conserved neuronal sub-clusters in the LH activated by voracious feeding. A) Schematic of the Act-seq workflow including the identification of LH cells by Gal4FF expression (Created with BioRender.com). B) UMAP (Uniform Manifold Approximation and Projection for Dimension Reduction) of LH cells sub-clustered based on their GABAergic and C) glutamatergic identities. Activated clusters are circled in red (all three Immediate Early Genes (IEGs) show increased expression in Voracious feeding relative to Food-deprived) or orange (two out of three IEGs show increased expression). D) Individual and average IEG expression of each GABAergic or E) glutamatergic cluster in Food-deprived (yellow) or Voracious feeding (purple) conditions. F) Expression of selected neuromodulators and receptors (cluster marker genes or appetite-related genes) in Food-deprived (yellow) or Voracious feeding (purple) conditions in GABAergic or G) glutamatergic populations. Activated clusters are indicated by red (three IEGs) or orange (two IEGs) boxes. Statistical analysis was performed using two-group unpaired Wilcoxon rank-sum test. Significant increases in gene expression across conditions are also indicated with asterisks. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Figure 2

Conserved receptor and neuromodulator expression in zebrafish larval hypothalamic regions. A) Representative images showing expression of important neuromodulators (ccka, penka, penkb, sst1.1, pdyn, trh, tac1, vip) in 7 dpf larval brains using Hybridisation Chain Reaction (HCR). A = Anterior, L = Lateral. B) Heatmaps showing dynamic expression of neuromodulators in medial lateral hypothalamus (mLH, top) and lateral lateral hypothalamus (lLH, bottom) respectively under voracious feeding and satiated conditions with respect to food-deprived conditions. The mLH and lLH were analysed separately as they may have different functional roles in sensory versus consummatory feeding behaviour [14]. C) Heatmaps showing dynamic activity reflected by elevated c-fos expression of neuromodulator expressing neurons under different feeding states with respect to food-deprived conditions (mLH, top; lLH, bottom). Expression and activity values were averaged from individual neurons from 8 to 10 fish per condition per gene. Genes with significant differential expression or activity as compared to the food-deprived state are labeled with an asterisk (FDR <0.05, statistical analysis performed using unpaired t-test with Welch correction). D) Schematic showing how activity patterns of different neuropeptidergic populations may reflect the composite LH activity patterns reported in Wee et al. (2019) [14].

Figure 3

Inter-species comparison of mouse and zebrafish LH transcriptomes. a) UMAP of mouse LH sub-clustered by A) GABAergic and C) glutamatergic identities. B) Common markers expressed in mouse and zebrafish GABAergic and D) glutamatergic clusters. Clusters with similar expression of marker genes are indicated by the same-coloured boxes. Clusters that are highly conserved in both species are indicated by dotted line boxes and clusters that are highly conserved and also activated in zebrafish are highlighted with solid line boxes. Note that the latter are all GABAergic. GABAergic: M2^GABA and Z0^GABA (yellow); M7^GABA and Z7^GABA (blue); M9^GABA and Z5^GABA (purple); M11^GABA and Z1^GABA (pink). Glutamatergic: M9^Glut and Z7^Glut (gray); M10^Glut and both Z0^Glut and Z5^Glut (brown).

Figure 4

Single-cell transcriptomic integration of mouse and zebrafish LH. Cross-species integration of zebrafish and mouse LH [18] using the Seurat CCA Integration workflow. UMAP visualisation of the integrated dataset sub-clustered by A) GABAergic and D) glutamatergic identities. Species specific distribution of mouse (cyan) and zebrafish (magenta) cells in B) GABAergic and E) glutamatergic populations. Proportions of mouse (cyan) versus zebrafish (magenta) cells in C) GABAergic and F) glutamatergic populations. Expression of selected neuromodulators and receptors in mouse (cyan) or zebrafish (magenta) cells in the G) integrated GABAergic or H) integrated glutamatergic populations. Color code of outlined boxes is shared with Figure 3. Clusters that are highly conserved in both species are indicated by dotted line boxes and clusters that are highly conserved and also activated in zebrafish are highlighted with solid line boxes.

Figure 5

tac1/ghraneuronal cluster is dynamically regulated by feeding state. A) HCR validation of a GABAergic cluster (tac1/ghra) which is activated in Voracious feeding conditions (Scale bar = 20 μm, magnified image scale bar = 10 μm). A = Anterior, L = Lateral. B) Validation of the GABAergic identity of tac1 neurons (Scale bar = 20 μm). C)c-fos elevation in tac1/ghra neurons is significantly higher in Voracious feeding in medial lateral (p-value <0.0001) and lateral lateral (p-value = 0.0002) hypothalamus. Statistical significance was calculated using Mann–Whitney test. Food-deprived: n (no. of tac1 neurons): mLH = 585, lLH = 105 (from 9 fish); Voracious feeding: n (mLH) = 844, n (lLH) = 69 (from 8 fish), Fed (Satiated): n (mLH) = 470, n (lLH) = 74 (from 7 fish). D) Number of tac1 neurons is significantly higher in Voracious feeding in mLH (p-value = 0.01). For sub-figure C, the different coloured larger dots in the SuperPlots [32] represent the mean of fluorescence intensity per individual fish brain, and the smaller dots of these colors indicate individual cells being quantified from each animal. For sub-figure D, the different coloured larger dots in the SuperPlots [32] represent the average number of tac1 positive neurons per individual fish brain. Statistical significance was calculated using Mann–Whitney test. No. of fish: Food-deprived = 9, Voracious feeding = 8, Fed (Satiated) = 7. ∗p<0.05, ∗∗p<0.01, ∗∗∗p<0.001, ∗∗∗∗p<0.0001.

Figure 6

tac1-positive hypothalamic neurons are responsive to food chemosensory cues. A) Photograph of the imaging setup showing the flow chamber in which the larva is mounted in agarose with the nose/mouth region exposed and facing the inflow of food cues (E3 conditioned with AP100 larval feed) from the inlet (white arrow). B) Schematic of food cue delivery protocol. The fish is exposed to a continuous flow of E3 (white) until the stimulus is switched to food cues (dark gray) via a valve system. This protocol is repeated twice. The green function depicts changes in food cue concentration as a function of time in the chamber, as estimated using a red dye (which was not present in the actual experiment). Hence, the light gray area indicates the period where the food cues are likely still present but decreasing in concentration. C) The dorsal and D) ventral plane of the ventral hypothalamus of an example fish and spatial map (C’& D′) of food-cue responsive neurons (solid circles) versus non-responsive neurons (transparent circles). A = Anterior, P = Posterior. Calcium traces of representative food-responding neurons are displayed and numbered in C' and D'. Scale bar = 20 μm. Sorted heat maps show the average food cue-induced calcium activity (deltaF/F) of food-responding neurons in the medial (E) and lateral (F) hypothalamus across both stimulus periods. (G) Mean functions from the food-responsive neurons in the MH (magenta) versus the LH (green) averaged across 19 fish. H) Boxplot showing number of food-responsive (left) vs non-responsive (right) neurons in the MH versus LH per fish (n = 19 fish). I) Ratio of food versus non-food neurons in the MH versus the LH.

Figure 7

Growth hormone modulates feeding and neuronal activity in zebrafish. A) Growth hormone (gh1) expression in larval zebrafish pituitary (Scale bar = 20 μm) A = Anterior, L = Lateral. B)gh1 transcript expression [Food-deprived: n (no. of endocrine cells) = 355 (from 9 fish); Voracious feeding: n = 412 (from 8 fish), Fed (Satiated):, n = 343 (from 9 fish); Food-deprived vs Voracious feeding: p-value <0.0001; Food-deprived vs Fed (Satiated): p-value <0.0001] and activity (measured by elevation in c-fos expression) of gh1 expressing endocrine cells is modulated by feeding state with highest expression during voracious feeding (Food-deprived vs Voracious feeding: p < 0.0001, Food-deprived vs fed: p-value = 0.001) C) Growth hormone receptor (ghra) expression in the mLH [Food-deprived N (no. of fish) = 8, Voracious feeding N = 8, Fed N = 9; mLH: Food-deprived vs Voracious feeding: p-value = 0.003; Fed (Satiated) vs Voracious feeding: p-value = 0.005] and lLH [Food-deprived N = 8, Voracious feeding N = 8, Fed N = 9; lLH: Food-deprived vs voracious feeding: p-value = 0.004; Fed (Satiated) vs Voracious feeding: p-value = 0.03] is highest during voracious feeding. D) Human Growth Hormone (hGH) treatment results in increased feeding in satiated fish, as measured by gut fluorescence normalised to mean gut fluorescence of controls for each biological replicate [Voracious feeding (control): N (No. of fish) = 230, Voracious feeding (GH treated): N = 246, Fed (control): N = 247, Fed (GH treated): N = 261; p-value = 0.03, from 6 replicates per condition] and E) increased activity in tac1 neurons [Food-deprived_control: n (no. of tac1 neurons) = 1508 (from 7 fish); Food-deprived_GH treated: n = 1471 (from 6 fish); Voracious feeding_control: n = 2336 (from 13 fish); Voracious feeding_GH treated: n = 2528 (from 19 fish); Fed_control: n = 1729 (from 13 fish); Fed_GH treated: n = 1941 (from 13 fish); p-value <0.0001 for all three conditions]. For sub-figures B, C and E, the different coloured larger dots in the SuperPlots [32] represent the mean of fluorescence intensity per individual fish brain, while the smaller dots of these colors indicate individual cells being quantified from each animal. For sub-figure D, the individual larger dots in the SuperPlots [32] represent average gut fluorescence per experimental replicate while the smaller dots of the same colors represent the normalised gut fluorescence per fish in each replicate. Statistical significance was calculated using Mann–Whitney test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001.

Figure 8

Tac1/Ghr neuronal cluster is conserved in mouse hypothalamus and is modulated by feeding state. A) RNAscope validation of Tac1/Ghr cluster in mouse brain sections shows co-expression of Ghr puncta (purple) in Tac1 neurons (green) (Overview stitched image: Scale bar = 100 μm; Magnified image: Scale bar = 20 μm) (Brain section cartoon created with BioRender.com). B) Representative maximum z-projection and stitched confocal images of mouse brain sections from “Fasted only”, “Fasted and refed” and “Ad-libitum” groups (Scale bar = 100 μm) with probes for growth hormone receptor (Ghr) (magenta), Tac1 (cyan), c-Fos as an activity marker (yellow) and a merged image with an overlay of three channels. M = Medial, L = Lateral. C) Scatter plot showing a significant increase in expression of Tac1 in Fasted and refed and Ad-libitum conditions [Fasting only: n (no. of Tac1 cells) = 7171 (from approximately 8–9 sections per brain, 4 brains in total); Fasted and refed: n = 13,881 (from approx 8–9 sections per brain, 4 brains in total); Ad-libitum: n = 12,448 (from approx 8–9 sections per brain, 4 brains in total); [Fasting only vs Fasted and refed: p-value <0.0001; Fasting only vs Ad-libitum: p-value <0.0001]. D) Scatter plot showing a significant overall increase of Tac1 neurons in “Fasted and refed” state (p = 0.01) and “Ad-libitum” state (p = 0.02). E) Scatter plot showing an increase in number of Tac1 neurons with elevated c-Fos expression in “Fasted and refed” condition vs Fasting only condition (p = 0.04). [Fasting only: n (no. of left and right halves of brain sections analysed) = 67 (from 4 brains); Fasted and refed: n = 73 (from 4 brains), Ad-libitum: n = 76 (from 4 brains). F) Scatter plot showing higher Ghr expression in activated Tac1 neurons in all conditions (p < 0.0001) [Fasting only: n (no. of Tac1 cells) = 7171 (from approx 8–9 sections per brain, 4 brains in total); Fasted and refed: n = 13,881 (from approx 8–9 sections per brain, 4 brains in total); Ad-libitum: n = 12,448 (from approx 8–9 sections per brain, 4 brains in total)]. The different coloured larger dots in the SuperPlots [32] C–F represent the mean fluorescence intensity or cell numbers per individual mouse brain. The smaller dots of these colors indicate individual cells being quantified from each animal for C, F. In D and E, the smaller colored dots indicate the individual right and left halves of each brain section analysed per mouse brain. Statistical significance was calculated using Mann–Whitney test. ∗p < 0.05, ∗∗∗∗p < 0.0001.