Transcriptional Basis for Rhythmic Control of Hunger and Metabolism within the AgRP Neuron

Abstract

The alignment of fasting and feeding with the sleep/wake cycle is coordinated by hypothalamic neurons, though the underlying molecular programs remain incompletely understood. Here, we demonstrate that the clock transcription pathway maximizes eating during wakefulness and glucose production during sleep through autonomous circadian regulation of NPY/AgRP neurons. Tandem profiling of whole-cell and ribosome-bound mRNAs in morning and evening under dynamic fasting and fed conditions identified temporal control of activity-dependent gene repertoires in AgRP neurons central to synaptogenesis, bioenergetics, and neurotransmitter and peptidergic signaling. Synaptic and circadian pathways were specific to whole-cell RNA analyses, while bioenergetic pathways were selectively enriched in the ribosome-bound transcriptome. Finally, we demonstrate that the AgRP clock mediates the transcriptional response to leptin. Our results reveal that time-of-day restriction in transcriptional control of energy-sensing neurons underlies the alignment of hunger and food acquisition with the sleep/wake state.

Keywords: AgRP; Agouti-related protein; RNA sequencing; RNA-seq; RiboTag; SCN; circadian; metabolism; suprachiasmatic nucleus; time-restricted feeding.

Figure 1.. Neural Molecular Clock Regulates Feeding Time and Glucose Homeostasis

(A) Food intake (% of total daily intake) in Bmal1^fx/fx (Fx/Fx) (white) and CamK2𝛼-Cre;;Bmal1^fx/fx (BKO) (green) mice in normal light:dark (LD; 12:12 h) (top) and constant darkness (DD) (bottom) conditions (left panels). Total daily food intake (g) over the corresponding 24-h period in LD and DD (right panels) (n = 4–7). (B) Endogenous glucose production, glucose infusion rate, glucose disposal rate, and tissue-specific glucose uptake (in white adipose tissue [WAT] and skeletal muscle) during hyperinsulinemic-euglycemic clamping (n = 6). (C) Glucose and insulin levels measured every 4 h in ad-lib-fed Bmal1^fx/fx and BKO mice (n = 3–5). (D) Time course of plasma glucose levels following an intraperitoneal injection of pyruvate (2 mg/kg) in fasted Bmal1^fx/fx and BKO mice at ZT2, following either ad libitum feeding (n = 10–19) (left panel) or following two weeks of dark-only feeding (n = 7) (right panel). (E) Respiratory exchange ratio (RER) values (VCO2/VO2) over 24 h in either ad-lib-fed (n = 4/genotype) or dark-only fed (n = 4) BKO mice. (F) Time course of plasma glucose levels following pyruvate injection (2 mg/kg) in sham and vagotomized Bmal1^fx/fx and BKO mice at ZT2 (n = 3–5). Data are represented as mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001). See also Figures S1 and S2.

Figure 2.. RNA Profiling Reveals Distinct Morning versus Evening Networks in Energy-Sensing AgRP Neurons by Whole-Cell and RiboTag Sequencing

(A) Endogenous glucose production during hyperinsulinemic-euglycemic clamping in Bmal1^fx/fx (Fx/Fx) (white) and AgRP-Cre;;Bmal1^fx/fx (ABKO) (blue) mice (n = 5–9). (**p< 0.01). (B) Experimental protocol outlining fluorescence-activated-cell-sorting (FACS)-based methods for whole-cell transcriptomics analyses (top row) and RiboTag-based methods for ribosome-bound transcriptomics analyses (bottom row) in AgRP neurons from fasted NPY-hrGFP and AgRP-Cre;;RiboTag mice, respectively, in morning (ZT2) and evening (ZT14) (see STAR Methods for specific details). (C) Representation of significant (FDR-adjusted p value < 0.05) gene networks identified from KEGG pathway analyses that are enriched in either the morning (ZT2) (i.e., JAK-STAT signaling, focal adhesion, and fatty acid biosynthesis) or evening (ZT14) (i.e., axonal guidance, mitophagy, and cholinergic signaling) in AgRP neurons by FACS (n = 7–8). Volcano plot comparing FDR-adjusted p values (y axis) and fold change (ZT2/ZT14) (x axis) displays the transcripts that are differentially expressed between morning and evening as orange dots (FDR-adjusted p < 0.05). (D) Log₂-fold change in gene expression in AgRP neurons from ABKO mice compared with Bmal1^fx/fx controls (both lines crossed with NPY-hrGFP for FACS) for indicated genes, color-coded for the associated KEGG (or literature-defined) pathways that each gene is associated with (n = 2–3). (E) Volcano plot comparing FDR-adjusted p values (y axis) and fold change (ZT2/ZT14) (x axis) displays the transcripts that are differentially expressed between morning and evening in AgRP-Cre;;RiboTag mice as red dots (FDR-adjusted p < 0.05) (n = 5–7). Representative genes changed in the mitochondrial oxidative phosphorylation pathway are shown. See also Figure S3.

Figure 3.. Dynamic Whole-Cell and RiboTag Energy State Transcriptional Response in Energy-Sensing AgRP Neurons across the Sleep/ Wake Cycle

(A) Overview of sample collection paradigm in the morning (ZT2) and evening (ZT14). Mice were either ad libitum fed or fasted at ZT12 for subsequent collection, using both FACS (NPY-hrGFP mice) and RiboTag (“Ribo” AgRP-Cre;;RiboTag mice) approaches. (B) Quadrant plot of transcriptional responses at ZT2 and ZT14, comparing log₂-fold changes induced by fasting (versus the fed state) in samples collected by FACS (x axis) (n = 3–8) and RiboTag (y axis) (n = 3–7) (showing significant genes; FDR-adjusted p values < 0.05). (C) Hierarchical clustering of normalized log₂-fold changes in gene expression across all identified transcripts in AgRP neurons across morning and evening, comparing changes induced by fasting (versus the fed state), in animals collected by both FACS and RiboTag. (D) Significant pathway enrichment (FDR-adjusted p < 0.05) in fasted versus adlibitum-fed animals across time points and comparing whole-cell (by FACS) and ribosome-bound (RiboTag) mRNAs, showing pathways that are either specific in terms of significance to FACS (top panel) or RiboTag (middle panel) or that are shared by both methods. (E) Heatmaps of significantly altered (FDR-adjusted p < 0.05) genes in significantly enriched pathways, across time points (ZT2 and ZT14) and utilized methods (FACS and RiboTag). Shown as log₂-fold changes (cutoff at ± 3) in response to fasting versus ad libitum feeding. See also Figures S3 and S4.

Figure 4.. Clock Timing of AgRP Neuron Transcriptional Response to Restricted Feeding and Leptin

(A) Selected KEGG pathways that were significantly enriched (FDR-adjusted p < 0.05) based on genes that were up- or down-regulated in morning (ZT2) and evening (ZT14) in AgRP neurons of mice (NPY-hrGFP) that were refed 1 h prior to collection, compared with fasted mice (all collected by FACS) (n = 4–8). (B) Significantlydown-regulated components (highlighted in boxes) ofcomplexesalong thespliceosome KEGG pathway in response to restricted feedingversus fasting at ZT14. (C) Hierarchical clustering of normalized log2-fold changes in gene expression inAgRP neurons at ZT2 and ZT14in Bmal1^fx/fx (Fx/Fx) versus AgRP-Cre;;Bmal1^fx/fx (ABKO) mice (lines crossed with NPY-hrGFP for FACS) (n = 3–4), showing all transcripts that were significantly different in response to leptin versus vehicle administration across the time points and genotypes. (D) Log₂-fold change in expression ofgenesin significantlyenriched KEGG pathwaysin ABKO mice at ZT2 andZT14in responseto leptinversusvehicle, showing log₂-fold changes (± 3) across time points and genotypes (significant genes with FDR-adjusted p < 0.05 highlighted in the right column for each pathway). (E) Motif analysis with significantly enriched (q value < 0.05) consensus sequences and the associated transcription factors, based on the transcriptional response (for up-regulated genes) to acute leptin versus vehicle administration in Fx/Fx and ABKO mice at the specified time points (no significant motifs found for other time points). See also Figures S3 and S4.