Food Perception Primes Hepatic ER Homeostasis via Melanocortin-Dependent Control of mTOR Activation

SUMMARY

Adaptation of liver to the postprandial state requires coordinated regulation of protein synthesis and folding aligned with changes in lipid metabolism. Here we demonstrate that sensory food perception is sufficient to elicit early activation of hepatic mTOR signaling, Xbp1 splicing, increased expression of ER-stress genes, and phosphatidylcholine synthesis, which translate into a rapid morphological ER remodeling. These responses overlap with those activated during refeeding, where they are maintained and constantly increased upon nutrient supply. Sensory food perception activates POMC neurons in the hypothalamus, optogenetic activation of POMC neurons activates hepatic mTOR signaling and Xbp1 splicing, whereas lack of MC4R expression attenuates these responses to sensory food perception. Chemoge-netic POMC-neuron activation promotes sympathetic nerve activity (SNA) subserving the liver, and norepinephrine evokes the same responses in hepatocytes in vitro and in liver in vivo as observed upon sensory food perception. Collectively, our experiments unravel that sensory food perception coordinately primes postprandial liver ER adaption through a melanocortin-SNA-mTOR-Xbp1s axis.

In Brief

The sight and smell of food are sufficient to induce liver endoplasmic reticulum reprogramming through a hypothalamic circuit, thereby anticipating the metabolic changes required for nutrient intake

Graphical Abstract

INTRODUCTION

More than 150 years ago, the French physiologist Claude Bernard proposed the concept of milieu interieur–that survival of an organism depends on keeping a constant internal environment despite changes in the external environment. The concept was later elaborated by Walter Cannon, who introduced the concept of homeostasis–keeping physiological responses within a narrow limit through feedback regulation (Cooper, 2008). Although food is necessary for survival, food intake poses a threat on multiple processes in whole-body homeostasis (Woods, 1991), such as leading to elevated glucose levels and activated protein and lipid synthesis.

Although the majority of homeostatic mechanisms operate based on internal feedback regulatory mechanisms, the organism has evolved systems to anticipate the impact of food consumption on homeostasis (Moore-Ede, 1986; Smith, 2000). Seeing, smelling, and tasting food, often termed as the cephalic phase, elicits a myriad of physiological changes, including increased heart rate (LeBlanc and Cabanac, 1989), saliva production, and secretion of digestive enzymes (Feldman and Richardson, 1986). This ensures that nutrients are rapidly metabolized and more efficiently removed from the circulation (Power and Schulkin, 2008). The cephalic phase is considered to be fast and transient in nature and mediated by both the parasympathetic (PNS) and sympathetic (SNS) branches of the autonomic nervous system (Diamond and LeBlanc, 1987). Functionally, intragastric feeding, which bypasses the sensory phase of food consumption or pre-meal treatment of humans with a ganglionic blocker, leads to exaggerated glucose excursions (Ahren and Holst, 2001; LeBlanc et al., 1984). Thus, the mechanisms through which food perception primes body physiology limit the potential harm arising from incoming nutrients. Nevertheless, the neuronal pathways involved and the molecular mechanisms affected in peripheral organs to mediate this adaptive response are only partly defined.

Notably the liver is a metabolically highly flexible organ that undergoes rapid changes in the transition from the fasted to the fed state (Seitz et al., 1977). In the postprandial phase, elevated circulating levels of insulin and nutrients activate hepatic signaling through the mechanistic target of rapamycin (mTOR) pathway (Inoki et al., 2002). mTOR is a key subunit of both mTORCI and mTORC2 with distinct and overlapping functions (Saxton and Sabatini, 2017). Similar to the mTOR pathway, the endoplasmic reticulum (ER) stress response is activated in nutrient-repleted liver in response to accumulation of unfolded proteins (Deng et al., 2013). This ER-stress response requires the activity of a serine/threonine-protein kinase/endoribonu-clease, the inositol-requiring enzyme 1 (IRE1) (Cox et al., 1993; Mori et al., 1993), to increase protein-folding capacity and to adapt ER homeostasis. Once unfolded proteins accumulate in the ER, the chaperone BIP is released from IRE, resulting in activation of its endonuclease activity to execute unconventional splicing of the X-Box binding protein (Xbp1) mRNA and translation of the mature Xbp1 transcription factor (Cox and Walter, 1996; Liou et al., 1990). This in turn activates a complex transcriptional response to readapt protein folding through expression not only of chaperone genes (Lee et al., 2003) but also of key enzymes of lipid synthesis to promote ER expansion (Shaffer et al., 2004; Sriburi et al., 2004). Accumulating data indicate an overlapping regulation of mTOR and Xbp1 in the postprandial liver (Pfaffenbach et al., 2010).

In addition to direct nutrient and insulin signaling in hepato-cytes, also the central nervous system plays an important role in coordinating the physiological responses in metabolic tissues including the liver (Ruud et al., 2017). In particular, specialized energy-state-sensing neurons in the hypothalamus can adapt autonomic outflow to organs such as liver, adipose tissue, and pancreas in accordance to energy availability of the organism. An important neurocircuit in these processes comprises the functionally antagonistic orexigenic agouti-related peptide (AgRP)-expressing and the anorexigenic proopiomelanocortin (POMC)-expressing neurons in the arcuate nucleus (ARC) of the hypothalamus (Cone, 2006; Gautron et al., 2015), which, in addition to regulating food intake, also coordinate multiple metabolic processes in peripheral organs (Kőnner et al., 2007; Ruud et al., 2017). Both AgRP and POMC neurons express receptors for energy state-communicating hormones, leading to the predominant view that their activity is subject to feedback regulation via hormones that increase subsequent to nutrient intake (Bel-gardt and Bruning, 2010; Vogt and Bruning, 2013). However, it has recently been demonstrated that these neurons rapidly and transiently change their activity already upon food perception before nutrients are consumed (Betley et al., 2015; Chen et al., 2015; Mandelblat-Cerf et al., 2015).

We sought to test whether the sight and smell of food are sufficient to prime the liver for the incoming nutrients. Through a combination of phosphoproteomic, transcriptomic, and lipido-mic screens, we found that food cues transiently activate hepatic mTOR and ER-stress signaling to prime ER adaptation for incoming nutrients. Food perception rapidly activated POMC neurons, and chemogenetic activation of these neurons resulted in a rapid increase of sympathetic output to the liver. Acute op-togenetic stimulation of POMC neurons in the ARC was sufficient to activate mTOR signaling and Xbp1 splicing in the liver. In he-patocytes, norepinephrine stimulation lead to mTOR activation and induced Xbp1 splicing. Thus, we uncovered a principal pathway through which food perception primes liver adaptation by engaging melanocortin-dependent control of liver SNA to promote mTOR-and Xbpl-dependent coordination of ER homeostasis.

RESULTS

Food Perception Promotes Hepatic ER-Stress Gene Expression

Given the rapid effects of food perception on the activity of the melanocortin circuitry, which also regulates peripheral metabolism, we investigated whether food perception might evoke changes in the liver transcriptome. Wild-type mice were fasted for 16 hr and then either allowed to eat a standard rodent chow (refed) or presented with inaccessible standard rodent chow that could be seen and smelled but not consumed (caged food) (Figures 1A and S1A). Mice were sacrificed after fasting or after different periods of refeeding or caged food exposure. Refeeding rapidly elevated blood glucose and insulin levels and resulted in a delayed increase in leptin levels after 2 hr (Figure S1B). Caged food exposure resulted after 30 min in a slight increase in blood glucose levels, which returned to baseline after 1 hr (Figure S1B), whereas circulating insulin concentrations remained unchanged (Figure S1C). Neither refeeding nor caged food presentation increased circulating norepinephrine concentrations (Figure S1C).

Figure 1. Food Perception Activates an ER-Stress Gene-Expression Signature in Liver.

(A) Experimental set-up and workflow.

(B) PC-analysis sample plot of all the time points.

(C) Bar plot representing the number of significantly changed transcripts at a FDR < 5% and Log₂-fold > 1.

(D) Point plot showing the averaged Log₂ ratiowith 95% confidence interval (CI) compared tofasted micefortranscripts assigned tothe indicated GO terms. Line plots visualize the time profiles of individual genes.

(E) Verification ofgenes byquantitative real-time PCR; mean indicated by a horizontal line, and data represented as boxplots with individual values (n = 6 animals/ group) relative to hprt expression.

(F) Xbp1 mRNA expression following refeeding or caged food presentation (n = 10–11 animals/group) relative to hprt expression.

(G) Xbp1 splicing assay showing the PCR products of Xbp1(u) and Xbp1(s) (n = 3/group).

Statistical analysis: (D) mean ± 95% CI. (E and F); *p < 0.05, **p < 0.01, and ***p < 0.001; one-wayANOVAfollowed by Dunnet’s multiple comparisonstest. Data in (E) and (F): Box-whisker plots with upper and lower quartile, median, the minimum and maximum values, and all individual data points

A principal component (PC) analysis on the hepatic mRNA expression in the different groups of mice revealed a clear separation of liver samples from refed mice compared to fasted animals; this separation was evident as early as 30 min after food exposure, persisting at all later time points (Figure 1B; Table S1). Interestingly, caged food exposure also induced a different gene expression relative to the fasted mice within 30 to 60 min after sensory food perception (Figure 1B). The number of differentially expressed genes in livers of refed mice or those exposed to caged food (Log₂-fold change > 1, FDR < 0.05) exhibited a differential kinetic. Exposure to caged food resulted in differential expression of 23 genes after 30 min, 23 genes after 1 hr, and 21 genes after 2 hr, whereas only 1 gene was differentially expressed 4 hr after sensory food perception. In contrast, refeeding resulted in a constant increase in the number of differentially expressed genes from 180 after 30 min to 639 after 4 hr (Figure 1C; Table S1). Next, we postulated that gene-expression changes induced by caged food exposure and refeeding could overlap. We performed a 1D gene annotation analysis aiming for the identification of annotation terms (GO, KEGG, and GSEA) that are systematically larger/smaller than the global distribution of Log2 ratios comparing each group to the group of fasted mice (Table S1). This analysis revealed that “activation of chaperone genes by Xbp1” and “protein processing in the endoplasmic reticulum” annotations were significantly upregu-lated in response to both caged food and refeeding (Figure 1D; Table S1). Analysis of candidate gene expression in these GO terms by real-time PCR (qPCR) confirmed that expression of Dnajb9, Pdia6, Dnajc3, and Herpudl was indeed increased in livers of refed mice or those exposed to caged food (Figure 1E). In contrast, classical insulin- and nutrient-regulated pathways such as “fatty acid beta oxidation” were downregulated by refeeding and not upon exposure to caged food, whereas genes of the GO term “glucokinase by GK regulatory protein” were up-regulated in response to refeeding but remained unchanged upon sensory food perception (Figure S1D).

Given that splicing of Xbp1 mRNA constitutes a key step in the initiation of the ER-stress-induced transcriptional program, we assessed changes in expression of spliced and unspliced Xbp1. Exposure to caged food increased the expression of spliced Xbp1(s) mRNA after 30 and 60 min, whereas both Xbp1(s) and unspliced Xbp1(u) expression continued to increase until 120 min after refeeding (Figure 1F). These results were further validated by visualization of Xbp1(s) in a complementary approach (Figure 1G). Importantly, 30 min exposure of accessible or inaccessible faked food (wooden pellet) did not increase Xbp1(s) expression (Figure S1E).

Food Perception Rapidly Activates Hepatic mTOR Signaling

Next, we performed an unbiased phos-phoproteomic screen on liver tissue obtained from mice that were either fasted, refed, or exposed to caged food for 30 min employing an in vivo SILAC-based quantification strategy as outlined in Figure 2A and Table S2 for all identified phos-phosites. In total we identified aproxi-mately 19,000 phosphosites (Table S2), and the pattern of phosphorylation was highly reproducible across different liver samples within each experimental group (Figure S2). A PC analysis showed clear separation of liver samples from 30 min refed mice compared to livers of fasted mice, with livers from mice exposed to caged food in between (Figure 2B). Employing a restricted analysis of the whole dataset, GO terms related to insulin signaling were enriched only in livers from refed mice (Figure 2C). This was further confirmed when comparing phosphorylated sequence windows for proteins containing an Akt kinase motif (Figure 2D).

Figure 2. Sensory Food Perception Activates Hepatic mTOR Signaling.

(A) Workflow of SILAC-based phosphoproteomics.

(B) PC analysis plot.

(C) Gene Ontology-Restricted analysis of the whole data. Identified enriched Gene Ontology terms determined by Fisher’s exact test comparing significantly regulated phosphorylation sites to the complete dataset. p values were corrected by the Benjamini-Hochberg procedure.

(D) Boxplot showing the distribution of phosphorylation sites displaying an Akt substrate motif (RXRXX (S/T)).

(E) Scatterplot of the Log₂ ratios of fasted/refed versus caged food/fasted.

(F) Fisher’s exact test results represented in a bar chart showing significantly enriched Gene Ontology terms and kinase motifs in the group of phosphorylation sites that were either similarly or exclusively regulated.

(G) Heatmap of a 2D annotation enrichment analysis combining the 30 min time point of the RNA-seq data and the 30 min time point of the phosphoproteomics data (FDR < 0.02).

Next, we aimed to identify phosphosites commonly regulated in livers from refed mice and those exposed to caged food. This unrestricted analysis revealed that phosphoproteins annotated to pathways related to ribosomal S6, TOR, Rho, Ras, and AMPK were enriched in the group of phosphorylation sites that were similarly regulated in both groups compared to fasted mice (Figures 2E and 2F). Phosphorylation sites on the ribosomal protein S6 (S6) were found to have a Log₂-fold change of 2 to 5 both in refed mice and in those exposed to caged food and ranked among the strongest regulated sites (Figure 2E; Table S2). S6 phosphosites were found to include the most commonly regulated sites within the C-terminal region, including Ser235/ 236/240/244/247 (Figure 2E; Table S2). Moreover, mTOR signaling regulates initiation of mRNA translation by controlling the phosphorylation of a range of eukaryotic initiation factors (Jiang et al., 2016), and we found altered phosphorylation of eIF3a, eIF4b, eIF4G, and eIF5b, which all form a part of the initiation complex together with S6 (Table S2). We also compared our data to published liver phosphoproteomic data from liver samples of fasted and 1 hr refed rats treated with and without rapamycin (Demirkan et al., 2011). Out of the 28 rapamycin-sensitive phosphosites activated by refeeding, we found 12 of these sites to be commonly upregulated in livers of refed mice and those exposed to caged food in our dataset (Table S2).

To potentially link the observed changes in the phosphopro-teome to the transcriptional changes described above, we further performed a 2D analysis combining our phosphoproteo-mic data with the RNA-sequencing data at the 30 min time point. This 2D enrichment analysis corroborated the finding of enrichment of pathways related to ribosomal S6 protein and ribosome biogenesis and with downregulation of beta-oxidation, glycogen, lipid, and ketone catabolism upon refeeding (Figure 2G; Table S2). In addition, we identified commonly regulated transcripts and phosphoproteins in the GO terms related to the synthesis of phosphatidylcholines (PC) and phosphatidyletha-nolamines (PE) upon both refeeding and exposure to caged food (Figure 2G). By manual inspection of the phosphoproteomic data, we detected phosphorylation of proteins, which were also detected as differentially expressed transcripts in the 2D analysis. Thus, phosphorylation of CTP:phosphocholine cytidylyl-transferase (Pcytla), the rate-limiting enzyme in PC synthesis, and phosphorylation of ethanolamine-phosphate cytidylyltrans-ferase (Pcyt2), involved in the synthesis of phosphatidylethanol-amines, were detected in both liver of refed mice and those exposed to caged food (Table S2).

Because we had focused our initial phosphosproteomic screen on samples from animals that had been exposed to food or caged food for 30 min, we potentially detected changes in downstream components of rapidly activated phosphorylation cascades. For example, S6 phosphorylation is a distal signaling event in the mTORCI pathway. However, our phosphoproteomic screen revealed increased phosphorylation of mTOR (Ser2481) and the mTOR complex protein Raptor (Ser859/863) only in livers of refed mice (Figure 2E; Table S2). Therefore, we investigated rapid phosphorylation events upstream of S6 by western blot analysis in samples of mice that had been exposed to refeeding or caged food for shorter periods of time. Strikingly, after only 5 min of refeeding or food perception, we detected increased Akt phosphorylation (Ser473) and mTOR phosphorylation (Ser2448) (Figure 3A). Caged food exposure and refeeding both also increased phosphorylation of 70 kDa ribosomal S6 kinase (P70S6K) (Thr389) at this time point. (Figure 3A). In contrast, only S6 (Ser235/236) phosphorylation was increased under refed conditions with no effect on S6 (Ser240/244) phosphorylation (Figure 3A). In contrast, after 10 min, phosphorylation of Akt (Ser473) in liver from mice exposed to caged food had already returned to baseline, whereas refeeding induced a maintained elevated Akt (Ser473) phosphorylation (Figure 3B). In liver from animals exposed to caged food or refed for 10 min, mTOR (Ser2448), P70S6K (Thr389), S6 (Ser235/236), and (Ser240/ 244) phosphorylation were all elevated compared to fasted mice. In contrast to these early changes in mTOR signaling, livers from mice exposed to caged food or refed for 10 min did not exhibit elevated levels of Xbp1(s) (Figure S3A).

Figure 3. Food Perception-Induced Xbp1 Splicing Is Attenuated upon mTOR Inhibition.

(A) Representative western blots using liver extracts from fasted mice or those refed or caged-food exposed for 5 min, showing pmTOR (Ser2448), pAkt (Ser473), pP70S6K (Thr389), pS6 (Ser235/236), (Ser240/244), and the corresponding beta-actin loading control.

(B) Same as in (A) with the exception that mice were refed or exposed to caged food for 10 min (n = 6 mice/condition). See Figure S6 for western blots of total proteins.

(C) Representative western blots using liver extracts from Rapamycin-treated mice that were fasted or refed or exposed to caged food for 30 min. Full blots used for quantification are represented in Figure S6.

(D) Quantification of pP70S6K (Thr389) and pS6 (Ser235/236) as assessed by western blots and Xbp1(u) and Xbp1(s) expression as assessed by qPCR in liver of mice exposed to caged food or refed with or without rapamycin administration (n = 10 mice/condition). See Figure S6 for all western blots used for quantification. Statiscal analysis: *p < 0.05, **p < 0.01, and ***p <0.001; One-way ANOVA followed by Dunnet’s multiple comparisons test in (A)-(C); in (D), two-way ANOVA followed by Sidak’s multiple comparisons test. Data presented as box plots. See also Figures S3 and S6.

Western blots of liver lysates from mice refed for 30 min confirmed elevated mTOR/P70S6K/S6 phosphorylation with only pAkt (Ser473) having returned to fasted levels in livers from refed mice (Figure S3B). In contrast, in livers from mice exposed to caged food, some residual phosphorylation of S6 was observed with all other phosphosites having returned to fasted levels (Figure S3B). In livers from refed mice, P70S6K (Thr389) and S6 (Ser235/236) remained increased after 1 hr but returned to fasted levels after 2 hr (Figure S3C).

Next, we examined the ability of food perception and refeeding to activate S6 phosphorylation and Xbp1 splicing in rapa-mycin-treated mice. Rapamycin treatment did not change food intake or blood glucose levels compared to vehicle-treated mice (Figure S3D). Rapamycin treatment blocked phosphorylation of P70S6K (Thr389) and S6 (Ser235/236) induced by exposure to either caged food or refeeding (Figure 3C). In addition, rapamycin treatment also attenuated the ability of food perception and refeeding to induce Xbp1 splicing (Figure 3D).

Food Perception Promotes Hepatic PC and PE Synthesis

Given that hepatic mTOR signaling and Xbp1 splicing are linked to hepatic lipid metabolism (Lee et al., 2008; Li et al., 2010; Quinn et al., 2017), we also compared the effect of refeeding and food perception on hepatic lipid profiles. Untargeted lipidomic profiling on liver tissue of mice that were fasted, exposed to caged food, or refed for 30, 60, 120, or 240 min revealed that the dominant lipid classes detected were phosphatidylcholines, triacylglycerols, phosphatidylethanolamines, and diacylglycer-ols (Figures 4A–4C; Table S3). Further PC analysis revealed that lipidomic profiles of refed mice at all time points were separated from those of fasted mice (Figure 4D). Mice that were exposed to caged food separated from the fasted mice (30 and 60 min time points) but later overlapped with fasted animals (Figure 4D). Long-chain polyunsaturated phosphatidylcholines, diacylglycerols, and phosphatidylethanolamine species were identified as key lipid classes contributing to the observed clustering (Figures 4E–4H). These experiments were consistent with the notion that key enzymes of PC, PE, and VLDL synthesis are targets of both mTOR and Xbp1 regulation (Quinn et al., 2017; Sriburi et al., 2004; Wang et al., 2012) and had been detected as being posttranslationally modified in the phospho-proteomic screen described above.

Figure 4. Food Perception Promotes Hepatic Phosphatidylcholine Synthesis.

(A) Workflow for determination of lipid signatures by LC-MS.

(B) Histogram showing the descriptive power (DP) and number of annotated compounds.

(C) Lipid class distribution showing numbers within each lipid class.

(D) PC-analysis sample plot of all lipids.

(E) PC-analysis sample plot considering only the class of phosphatidylcholines.

(F) PCA drivers of phosphatidylcholines—chain length is color encoded, and the number of double bounds is represented by the scatter point size.

(G and H) PC-analysis sample plots of diacylglycerols and phosphatidylethanolamines.

Next, we aimed to address whether food perception and/or refeeding promotes structural changes in ER morphology. To this end, we used an electron microscopy-based morphological analysis to assess the morphology of the hepatic ER in mice that had been fasted, exposed to caged food, or refed for 30 min. This analysis revealed that both food perception and refeeding lead to rapid ER remodeling and elongation (Figures S4A-S4C).

MC4R Signaling Is Required for Food Perception-Induced Xbp1 Splicing and mTOR Activation

Having identified the ability of food perception to prime hepatic ER adaptations, we aimed to investigate whether this regulation might be linked to the transient changes in the activity of the energy-sensing melanocortin neurons. We employed fiber photometry to determine POMC neuron activity in freely behaving mice, which exhibited POMC neuron-restricted GCaMP6s expression (Figure 5A). Analysis of Ca²⁺ dynamics in POMC neurons showed that refeeding resulted in a robust increase in POMC neuron activity (Figure 5B). In contrast, food perception rapidly and transiently increased POMC neuron activity, whereas exposure to fake food failed to elicit activity changes in POMC neurons (Figure 5B). We observed no fluorescence intensity changes when we expressed GFP instead of GCaMP6s in POMC neurons (Figure 5B).

Figure 5. Food Perception RapidlyActivates POMC Neurons.

(A) Confocal imagesshowing Cre-dependent, virally mediated expression of GCaMP6s by immunohistochemistry(GFP; green), Pomc (magenta), and Agrp (blue) mRNA expression as assessed by FISH as well as fiber placement.

(B) Ca²⁺ responses to faked food, caged food, or accessible food presentation in POMC neurons in fasted mice expressing GCaMP6s or GFP. Areas underthe curve (AUC) of the individual mice (n = 3) are shown. Arrows indicate the time points of the respective food stimulus.

(C) Representative images of Pomc and Fos mRNA expression in the ARC of fasted, fake-food-exposed, caged-food-exposed, or refed mice (30 min) (n = 4/each group).

(D) Top left, quantification of Fos/Pomc mRNA-expressing cells. Top right, spliced Xbp1(s) in liverfrom the same mice (n = 4; forthe fasted group n = 3). Bottom, Spearman correlation between Fos-positive POMC neurons in the ARC and Xbp1(s) in the livers from the same animals.

Statistical analysis: (B and D) *p < 0.05, **p < 0.01, One-way ANOVA followed by Dunnet’s multiple comparisons test. Data in (B) are individual values and in (D) are Box-whisker plots.

Similarly, both refeeding and exposure to caged food, but not fake food exposure, robustly induced Fos expression in POMC neurons (Figures 5B and 5C). In the same mice, refeeding and caged food exposure, but not fake food exposure, induced Xbp1(s) expression (Figure 5D). Of note, analysis of Fos expression in POMC neurons and hepatic Xbp1(s) expression in these mice revealed a positive correlation (Figure 5D).

Next, we optogenetically activated POMC neurons in the ARC of mice expressing Channelrhodopsin-2 (ChR2) in POMC neurons. We crossed POMC-Cre mice with mice allowing Cre-dependent expression of ChR2 and implanted optical fibers over the ARC in the resulting ChR2^POMC animals and littermate controls (ChR2^WT)(Figures 6A). Following 30 min photoactivation of POMC neurons, animals were sacrificed, and livers were subjected to analysis. We found that POMC neuron stimulation induced a 5-fold increase in hepatic S6 phosphorlyation and a 2-fold increase in Xbp1(s) expression (Figures 6B–6C).

Figure 6. POMC Neuron Activation Promotes Hepatic mTOR Signaling and Xbp1 Splicing.

(A) Expression of ChR2 in POMC neurons as assessed by GFP and POMC immunohistochemistry as well as fiber placement.

(B) Representative western blots for pP70S6K(Thr389) and pS6 (Ser235/236) from liver lysates of ChR2^POMC and ChR2^WT mice after photostimulation for 30 min.

(C) Quantification of Xbp1(s) and Xbp1(u) mRNA in livers of ChR2^POMC and ChR2^WT mice after photostimulation. Western blot analysis for pS6(Ser235/236) and pP70S6K(Thr389) (n = 7 mice/group). See Figure S6 for all western blots used for quantification and additional loading controls.

(D) Representative western blots for pS6 (Ser235/236) in livers of fasted, exposed to caged food, or refed (30 min) control and MC4R-deficient mice. Lower panel shows the quantification of results obtained under the same conditions (n = 6–8 mice/group). See Figure S6 for all western blots used for quantification.

(E) Analysis of hepatic Xbp1(s) and Xbp1(u) expression in livers of the same mice as analyzed in (D).

Statistical analysis: (C) *p < 0.05 and **p < 0.01 indicate significantly different from ChR2^WT group by unpaired two-sided Student’s t test. (D) *p < 0.05 and **p < 0.01; two-way ANOVA followed by Sidak’s multiple comparisons test. (E) *p < 0.05 and **p < 0.01; one-way ANOVA followed by Dunnet’s multiple comparisons test. In (C)–(E), data are presented as box plots.

To investigate the necessity of POMC/αMSH-dependent signaling in food perception-evoked induction of mTOR signaling andXbp1 splicing in the liver, we compared these pathways in Melanocortin-4 receptor (MC4R)-deficient mice and littermate controls. We found that caged food exposure induced hepatic S6 phosphorylation (Ser235/236) in control but not in MC4R-deficient mice (Figure 6D). In contrast, refeeding induced hepatic S6 phosphorylation (Ser235/236) to a similar extent in both control and MC4R-deficient mice (Figure 6D). Moreover, whereas refeeding similarly induced Xbp1(s) expression in the livers of control and MC4R-deficient mice, food perception failed to significantly induce Xbp1(s) expression in the livers of MC4R-deficient mice (Figure 6E).

POMC Neuron Activation Increases Hepatic Sympathetic Nerve Activity to Induce mTOR Activation and Xbp1(s) Expression

We next determined whether POMC neuron stimulation promotes hepatic sympathetic nerve activity (SNA) subserving the liver as a potential effector mechanism to promote the observed ER adaptation. To this end, we employed mice that express the stimulatory chemogenetic receptor hM3Dq selectively in POMC neurons. Clozapine-N-oxide (CNO) administration in fasted hM3Dq^POMC but not in control hM3Dq^WT mice effectively activated POMC neurons but not AgRP neurons in the ARC as revealed by in situ analysis of Fos, Pomc, and Agrp expression (Figures S5A and S5B). We next assessed hepatic SNA activity changes in response to hM3Dq/CNO-mediated POMC neuron stimulation. Within only 1 min after CNO administration, hepatic SNA activity was significantly increased in hM3Dq^POMC mice, and this increase lasted for 60 min. CNO administration failed to induce changes in hepatic SNA activity in control hM3Dq^WT mice (Figures 7A).

Figure 7. Norepinephrine Activates mTOR Signaling and Xbp1 Splicing in Liver.

(A) Changes in hepatic SNA after intravenous (i.v.) administration of CNO (3 mg/kg) or vehicle in hM3Dq^POMC mice (n = 6) or hM3Dq^WT control mice (n = 7). Data are presented as mean with 95% CI.

(B) mRNA expression (qPCR) of Xbp1(s) and Xbp1(u) in liver of mice fasted for 16 hr followed by intraperitoneal (i.p.) administration of either saline or alpha₁- adrenergic receptor blocker (Alphal, Prazosin) followed by a second i.p. administration (30 min later) of saline or NE.

(C) Western blots of pmTOR(Ser2448), pP70S6K(Thr389), and pS6(Ser235/236) and loading control (Calnexin). mRNA expression of Xbp1(s) and Xbp1(u) in fasted hM3Dq^Alfp and hM3Dq^WT control mice sacrificed 30 min after CNO (3 mg/kg) i.p. administration (n = 5 for hM3Dq^WT and n = 8 for hM3Dq^Alfp). See Figure S6 for additional loading controls.

(D) Xbp1(s) and Xbp1(u) expression in Hepa1–6 cells treated with 10 μM NE for the indicated times.

(E) Representative western blots from Hepa1–6 cells treated with 10 μM NE for different periods.

(F) Representative western blots of Hepa1–6 cells pre-treated with 100 nM rapamycin for 30 min followed by 10 μM NE for 30 min.

(G) Xbp1(s) and Xbp1(u) expression from Hepa1–6 cells pre-treated with 20 nM rapamycin for 30 min followed by 10 μM NE administration for 30 min. Statistical analysis: (A) Left: **p < 0.01 and ***p < 0.001; two-way ANOVAfollowed by Turkey’s multiple comparisons test. Right (AUC): **p < 0.01 and ***p < 0.001; one-way ANOVA followed by Turkey’s multiple comparisons test. (B) *p < 0.05 from (Sal/Sal) and §p < 0.05 from (Sal/Alpha₁); one-way ANOVA followed by Turkey’s multiple comparisons test. (C) **p < 0.01 and ***p < 0.001; unpaired two-sided Student’s t test. (D and G) *p < 0.05 and **p < 0.01; one-way ANOVA followed by Dunnet’s multiple comparisons test. (A): Data on the left are presented as mean ± 95% confidence intervals; (A, right), (B), (C), (D), and (G): Data are presented as box plots. See also Figures S5 and S6.

In liver, NE is released from sympathetic nerve endings reaching the hepatocytes, which express both alpha₁- and beta₂-adrenergic receptors (Kawai et al., 1986) (Figure S5C). In control mice, 30 min following administration of NE (2 mg/kg), hepatic Xbp1(s) was increased (Figure 7B), whereas pre-treatment with an alpha₁-adrenergic receptor blocker (Prazosin) abrogated NE-induced hepatic Xbp1(s) expression (Figure 7B).

Binding of NE to alpha₁-adrenergic receptors activates the Gq subunit of the heterotrimeric G protein. Therefore, we generated mice that express the chemogenetic hM3Dq receptor selectively in hepatocytes (hM3Dq^Alfp). CNO administration increased Xbp1(s) expression in hM3Dq^Alfp mice but not in littermate controls (Figure 7C). Moreover, hM3Dq/CNO-mediated stimulation of hepatocytes leads to robust mTOR, P70S6 kinase, and S6 phosphorylation (Figure 7C).

Similarly, NE stimulation of Hepa1–6 cells increased Xbp1(s) expression after 30 min, which returned to baseline after 2 hr (Figure 7D). In addition, NE stimulation increased Akt/mTOR/ P70S6K/S6 phosphorylation within 5 min, returning to baseline after 60 min (Figure 7E). Pre-treatment of Hepa1–6 cells with rapamycin for 30 min blocked the NE-induced P70S6K (Thr389) and S6 (Ser235/236) phosphorylation (Figure 7F) as well as NE-stimlated Xbp1 splicing (Figure 7G). Taken together, our in vitro and in vivo data indicate that direct POMC neuron activation as observed upon food perception increases liver SNA, which in turn can activate mTOR/S6 phosphorylation and Xbp1 splicing.

DISCUSSION

Predictive physiological responses to food cues ensure that disturbance in internal homeostasis due to ingested nutrients are limited. These anticipatory responses are initiated by visual, olfactory, and cognitive inputs related to the nutritional value and the internal energy state of the animal (Power and Schulkin, 2008), leading to rapid activation of the sympathetic nervous system (Diamond and LeBlanc, 1987). Although numerous studies have contributed to the understanding of the so-called cephalic response, few studies have examined the role of specific neuronal populations and how this impacts on signaling in peripheral organs.

Here we provide a conceptual framework of how food perception is relayed from the hypothalamus to the liver via melanocor-tin-dependent regulation of liver SNA to induce hepatic mTOR and Xbp1 activation. This notion is consistent with the recent description that food perception rapidly activates POMC neurons (Betley et al., 2015; Chen et al., 2015; Mandelblat-Cerf et al., 2015). These experiments have revealed a novel layer of regulation of the melanocortin circuitry beyond the well-established long-term activity control via hormonal signals such as insulin, leptin, and GLP-1 (Cowley et al., 2001; Secher etal., 2014).

This regulatory principle is in line with the generalized concept that melanocortin neurons are ideally positioned to integrate numerous signals informing them about energy availability of the organisms as well as sensory signals instructing them about upcoming nutrient influx. Thereby, they can adapt numerous processes in the integrated regulation of whole-body metabolism, including control of hepatic glucose metabolism and VLDL secretion, BAT activation, and adipose tissue lipolysis, (Balthasar et al., 2004; Berglund et al., 2012; Kőnner et al., 2007; Parton et al., 2007; Ruud et al., 2017; Scherer et al., 2011, 2016; Shin et al., 2017; Steculorum et al., 2016). Importantly, POMC neurons have been demonstrated to promote sympathetic nerve activation in response to leptin and insulin (i.e., in control of SNA in various tissues) (Bell et al., 2018; Chhabra et al., 2017; Chitravanshi et al., 2016; Rahmouni et al., 2003). Although our experiments clearly reveal that chemogenetic POMC neuron activation is sufficient to rapidly induce hepatic SNA and that optogenetic POMC neuron activation promotes S6 phosphorylation and Xbp1 splicing, they do not rule out the possibility that the concomitant inhibition of AgRP neuron activity may also contribute to these effects upon sensory food perception. Our experiments are also consistent with studies revealing that MC4R expression in sympathetic pre-ganglionic neurons regulates thermogenesis as well as hepatic glucose production through their ability to reciprocally control sympathetic and parasympathetic pre-ganglionic neuron activity (Berglund et al., 2014; Sohn et al., 2013). In fact, human subjects with mutations in the MC4R, when compared to BMI-matched controls, have reduced sympathetic muscle tone (Sayk et al., 2010), have reduced urinary norepinephrine excretion, and are less prone to hypertension (Greenfield et al., 2009). Given the pleiotropic regulatory function of melanocortin neurons on SNA subserving multiple tissues, the anticipatory effects of food sensation on liver physiology may have more general implications for a broad-based organismal adaptation to incoming nutrients (i.e., through regulation of adipose tissue or pancreatic SNA).

Finally, it will be interesting to investigate whether the ability of food perception to adapt hepatic ER homeostasis might be compromised in obesity. This appears possible in light of the impairment in the regulation of melanocortin neurons in high-fat-diet-fed mice (Enriori et al., 2007; Kleinridders et al., 2009). Thereby, increased food intake in obesity may lead to increased nutrient load in liver, which is not primed to expand protein-folding capacity and secretion, thus further propagating ER stress in a cell-autonomous manner, which can contribute to the development of obesity-associated insulin resistance (Ozcan et al., 2004, 2006).

Using an unbiased transcriptomic approach, we identified the ER-stress response (and in particular the Xbp1 branch) as the primary transcriptional program activated by food cues in the murine liver. Thus, a subset of transcripts related to the ER-stress response is initially regulated by a mechanism that seems not to require food intake (i.e., being independent of glucose, insulin, and leptin) but that requires primarily the perception of food. Induction of the ER-stress response and more specifically Xbp1 splicing upon feeding in murine liver suppresses glucose production, enhances insulin sensitivity, controls lipogenesis, and is required for export of lipids in the form of VLDL (Deng et al., 2013; Lee et al., 2003; Ning et al., 2011; Ozcan et al., 2004; Wang et al., 2012). Thus, activation of Xbp1 splicing by food perception may also regulate hepatic metabolism.

The liver is a highly metabolic organ, the primary source for circulating plasma proteins and lipid-rich particles (Trefts et al., 2017). Fasting results in suppression of the secretory machinery (van Leeuwen et al., 2018), which needs to be re-established in order for the liver to coordinate incoming nutrients with protein synthesis and export of lipids to other tissues. Therefore, inducing ER-stress signaling represents an ideal mechanism to adapt according to this future requirement. Moreover, our data are in line with more recent experiments in nematodes, providing proof of principle for a non-cell-autonomous regulation of the ER-stress response through the CNS (Taylor and Dillin, 2013), thus pointing to an evolutionarily conserved, food-perception-dependent control of hepatic ER homeostasis and adaptation. Similarly, it had been demonstrated in mice with POMC neuron-specific overexpression of Xbp1(s) that they exhibit increased hepatic Xbp1(s) expression under HFD-feeding conditions, pointing to the possibility of non-cell-autonomous ER-stress regulation via POMC neurons (Williams et al., 2014). However, that study did not reveal the physiological stimulus, efferent pathways, and molecular mechanisms activated in liver to promote this response.

Collectively, the present study reveals a principle pathway through which sensory food anticipation coordinately primes peripheral tissue function. The regulation of food-sensing-dependent control of ER homeostasis may have far-reaching implications not only for metabolic homeostasis and potentially disease but also for the regulation of aging and development of a whole range of aging-associated diseases, in light of the critical role of proper proteostasis in these processes (Balch et al., 2008).

STAR★METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken anti-GFP antibody	Abcam	Cat# ab13970; RRID: AB_300798
Rabbit anti-POMC precursor antibody	Phoenix Pharmaceuticals	Cat# H-029–30; RRID: AB_2307442
Chicken anti-GFP antibody	Life Technologies	Cat# A10262; RRID: AB_2534023
Alexa-fluorophore secondary antibody	Molecular Probes	Cat# R37119; RRID: AB_2556547
Beta-Actin antibody	Sigma Aldrich	Cat# A5441; RRID: AB_476744
Calnexin antibody	Calbiochem	Cat# 208880; RRID: AB_2069031
p70 S6 kinase antibody	Santa Cruz	Cat# sc-230; RRID: AB_632156
p-p70 S6 kinase antibody	Cell Signaling	Cat# 9234; RRID: AB_2269803
mTOR antibody	Cell Signaling	Cat# 2972; RRID: AB_330978
p-mTOR antibody	Cell Signaling	Cat# 2971; RRID: AB_330970
S6 ribosomal protein antibody	Cell Signaling	Cat# 2217; RRID: AB_331355
p-S6 ribosomal protein antibody	Cell Signaling	Cat# 2215S; RRID: AB_331682
p-S6 ribosomal protein antibody	Cell Signaling	Cat# 4858; RRID: AB_916156
Akt (pan) antibody	Cell Signaling	Cat# 4685; RRID :AB_2225340
p-Akt antibody	Cell Signaling	Cat# 4060; RRID: AB_2315049
GFP (Immuno) Gcamp6Ss antibody	Abcam	Cat# ab13970; RRID: AB_300798
POMC (Immuno) antibody	Phoenix Pharmaceuticals	Cat# H-029–30; RRID: AB_2307442
GFP (Immuno) ChR2	Life Technologies	Cat# A10262; RRID: AB_2534023
Anti-mouse antibody	Sigma Aldrich	Cat# A4416; RRID: AB_258167
Anti-rabbit antibody	Sigma Aldrich	Cat# A0545; RRID: AB_257896
Goat anti-chicken FITC antibody	Jackson ImmunoResearch	Cat# 103–095-155; RRID: AB_2337384
Donkey anti-rabbit antibody	Molecular Probes	Cat# R37119; RRID: AB_2556547
Goat anti-chicken antibody	Molecular Probes	Cat# A-11039; RRID: AB_142924
Bacterial and Virus Strains
AAV1 .Syn.Flex.GCaMP6s.WPRE.SV40	Upenn Vector Core	N/A
AAV1 .pCAG.Flex.eGFP.WPRE	Upenn Vector Core	N/A
Chemicals, Peptides, and Recombinant Proteins
GlutaMAX	Gibco	Cat# 35050061
Prazosin	Sigma Aldrich	Cat# P7791
Clozapine-N-oxide	Sigma Aldrich	Cat# C0832
Buprenorphine	Bayer	PZN 01498870
Meloxicam	Boerhinger Ingelheim	PZN 07578423
Rapamycin	Calbiochem (Merck)	Cat# 553210; Lot 2848513
Qiazol	Qiagen	Cat# 79306
MOPS buffer	Invitrogen	Cat# NP001
Western blotting reagent	Roche	Cat# 11829200
Glutaraldehyde	Electron Microscopy Sciences	Cat# 16220
Protease III	MBL life science	Cat# 322340
Protease Plus	MBL life science	Cat# 322330
ProLong Gold Antifade Mountant	ThermoFisher	Cat# P36931
Tris- HCL	VWR chemicals	Cas# 103156X
Trisbase	AppliChem	Cas# 1025
cOmplete Mini Protein inhibitor cocktail	Roche	Cat# 1183617001
EDTA	Sigma Aldrich	Cas# 60–00-4; E6758
EGTA	Sigma Aldrich	Cas# 67–42-5; E3889
Na3VO4	Sigma Aldrich	Cas# 13721–39-6
NaF	Sigma Aldrich	Cas# 7681–49-4
Na4P2O7	Sigma Aldrich	Cas# 7722–88-5
Agarose	Sigma Aldrich	Cas# 9012–36-6
Sucrose	Sigma Aldrich	Cas# 57–50-1
PFA	Sigma Aldrich	Cas# 30525–89-4
L (−)− Norepinephrine bitartrat salt monophosphate	Sigma Aldrich	Cat# A9512; Lot SLB3018V
Rapamycin	Calbiochem (Merck)	Cat# 553210; Lot 2848513
L-norvaline	Sigma Aldrich	Cat# 176079–5G Lot MBBB0258V
4 x Laemmli sample buffer	Bio-Rad	Cat# 161–0747
SPLASH standard single-vial lipidomic analytic standard for human plasma lipids, deuterium labelled lipids	Avanti Polar Lipid Inc	Cat# 330707
NP-40	Sigma Aldrich	Cas# 9016–45-9
Ethanol	VWR chemicals	Cas# 64–17-5
10 × TGS Tris/glycine running buffer	Bio-Rad	Cat# 161–0732
PMSF	Sigma Aldrich	Cat# 93482
SYBRGreen mastermix	Applied Biosystems	Cat# 4472897
TaqMan mastermix	Thermo Scientific	Cat# 436906
Tween 20	VWR France	Cas# 9005–64-5
Methanol	VWR France	Cas# 67–56-1
DMEM (1x) + GlutaMAX™, Delbeccos Modified Eagle Medium 4.5g/L D-glucose	Gibco	Cat# 111960044
Dichloromethane	Sigma Aldrich	Cas# 75–09-2
b-Mercaptoethanol	Sigma Aldrich	Cat# M3148
DPBS (1x)	Gibco Life Technologies	Cat# R14190–094
Dimethyl sulfoxide (DMSO)	Merck Darmstadt	Cas# 25322–68-3
Critical Commercial Assays
Leptin, Mouse Leptin ELISA Kits	Crystal Chem	Cat# 90030 RRID: AB_2722664
Insulin, Ultra Sensitive Mouse Insulin ELISA	Crystal Chem	Cat# 90080
Norepinephrine 2-CAT (A-N) Research ELISA	Labor Diagnostika Nord	Cat# BA E-5400
RNeasy Mini kit	Qiagen	Cat# 74004
High-Capacity cDNA Reverse Transcription Kit	Applied Biosytems	Cat# 4368814
BCA assay	Pierce	Cat# 23225
RNAscope Multiplex Fluorescent Assay for Pomc+Fos	ACD Bio	Cat# 320851
Tyramide-based RNAscope Multiplex Fluorescent v2 Assay	ACD Bio	Cat# 323110
TruSeq RNA sample preparation Kit v2	Illumina	Cat# RS-122–2001 /2
2 -CAT (A-N) Research ELISA	Labor Diagnostika Nord	Cat# BA E-5400
Deposited Data
Phosphoproteome	this study	PRIDE: PXD005681
RNA-Seq Data	this study	NCBI GEO: GSE118973
Experimental Models: Cell Lines
Hepa1–6	ATCC	(ATCC CRL-1830™) passage number 30 RRID: CVCL_0327
Experimental Models: Organisms/Strains
C57BL/6 N	Charles River	Strain# 027
Melanocortin-4 receptor (Mc4r)-deficient (Mc4rtm1Lowl/JloxTB)	Jackson Laboratory	Strain# 006414 RRID: IMSR_JAX:006414
POMC-Cre	Jackson Laboratory	Strain# 005965 RRID: IMSR_JAX:005965)
Alfp-Cre	Kellendonk et al., 2000	N/A
hM3Dqfl/fl (ROSA26CAGsloxSTOPloxhM3DGq)	Steculorum et al., 2016	N/A
ChR2 fl/fl mice (ROSA26loxSTOPloxChR2(H134R)-EYFP- WPRE) with a conditional allele (Ai32)	Jackson Laboratory	Strain# 012569 RRID: IMSR_JAX:012569)
Oligonucleotides
See Table S5 for qPCR primers.
Software and Algorithms
Perseus	Tyanova et al., 2016	http://wwwcoxdocs.org/doku.php?id=perseus:start
InstantClue	Nolte et al., 2018	www.instantclue.uni-koeln.de
MaxQuant Andromeda	Cox et al., 2011	http://www.coxdocs.org/doku.php?id=maxquant:start
ImageJ	Schneider et al., 2012	www.imagej.nih.gov
R	The R Foundation for Statistical Computing c/o Institute for Statistics and Mathematics Wirtschaftsuniversi tät Wien	https://www.r-project.org
Matlab R2G14b	MATLAB and Statistics Toolbox Release 2012b, The MathWorks, Inc., Natick, MA, USA	https://de.mathworks.com
Other
ssniff Chow Diet	ssniff Spezialdiaten GmbH	Cat# V1554
Criterion TGX Precast Gels 4–15 %	Bio-Rad	Cat# 5671024
Criterion TGX Precast Gels 10 %	Bio-Rad	Cat# 5671035
Criterion TGX Precast Gels 7.5 %	Bio-Rad	Cat# 5671085
Trans-Blot (R)Turbo, Midi Format, 0.2 μm PVDF, single application	Bio-Rad	Cat# 1704157
C18 Cartridges	Waters GmbH	Cat# WAT 054955
XBridgeBEH Column	Waters GmbH	Cat# 186007207
TiO2 beads, 5 μm Titansphere	GL Sciences	Cat# 5020–75000

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Jens C. Brüning (bruening@sf.mpg.de).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse models

Husbandry.

All animal procedures were conducted in compliance with protocols approved by the local authorities (Bezirksregierung Koln) and were in accordance with NIH guidelines. Mice were housed at 22–24°C with 12 hr light/dark cycle. Animals had ad libitum access to water and a standard rodent chow (ssniff V1554,59494 Soest, Germany). Mice were housed in individually ventilated cages (IVCs). After weaning, the number of mice per cage was between 2 and 5. For Calcium imaging by fiberphotometry, mice were singlehoused after viral injection and stereotactic surgery in non-ventilated open cages due to the laser detection system. All studies, except the study using hM3Dq^POMC mice to assess sympathetic activation at the level of the liver, were conducted at the Max Planck Institute for Metabolism Research, Cologne, NRW, Germany. Measurement of SNA subserving the liver in hM3Dq^POMC mice was performed in Iowa, USA, in the laboratory of Dr Kamal Ramhouni and approved by the University of Iowa, Animals Research Committee.

Mouse lines

C57BL/6N. This mouse line was obtained from Charles River, France.

Melanocortin-4 receptor (MC4R)-deficient (MC4R^{tm1Lowl/JloxTB}) mice were obtained from Jackson laboratory (stock# 006414). Heterozygous breedings were used to generate wt/wt and fl/fl mice in the facility of the Max Planck Institute for Metabolism Research, Cologne, NRW, Germany. Both genders were used for experiments. At the time of study, the MC4R-deficient mice were obese (Body weight: wt/wt 17.9+/−2.9 g; fl/fl 28.1+/−2.9 g), while fasting blood glucose concentrations did not differ significantly between groups (blood glucose: wt/wt 59.7 mg/dl; fl/fl 64.3 mg/dl).

Alfp-Cre. This line has previously been described (Kellendonk et al., 2000). The line was maintained by crossing to C57BL/6 N mice from Charles River, France, in the facility of the Max Planck Institute for Metabolism Research Cologne, NRW, Germany

POMC-Cre. This line has previously been described (Balthasar et al., 2004) and was obtained from Jackson Laboratory (stock# 005965). Mice were bred to C57BL/6N mice from Charles River, France, to maintain the line in the facility of the Max Planck Institute for Metabolism Research, Cologne, NRW, Germany.

hM3Dq fl/fl (ROSA26CAGSloxSTOPloxhM3D_Gq). This line has previously been described (Steculorum et al., 2016). The mice were maintained as homozygous fl/fl animals.

ChR2 fl/fl mice (ROSA26loxSTOPloxChR2(H134R)-EYFP-WPRE) with a conditional allele (Ai32) (Madisen et al., 2012) were obtained from Jackson Laboratory (stock# 012569). The mice were maintained as homozygous fl/fl animals in the facility of the Max Planck Institute for Metabolism Research, Cologne, NRW, Germany.

hM3Dq^Alfp. The breeding scheme was mating heterozygous Alfp Cre+ mice to homozygous hM3Dq fl/fl mice. The heterozygous hM3Dq^Alfp and hM3Dq^WT mice were used as experimental animals. Littermates of both sexes were used for the experiment.

hM3Dq^POMC. POMC-Cre mice were first crossed to hM3Dq fl/fl mice to generate heterozygous hM3Dq^POMC (Cre/wt, fl/wt) mice, which were then bred to hM3Dq fl/fl mice to generate the experimental animals. Littermates of both sexes were used for the experiment.

ChR2^POMC. Heterozygous POMC-Cre mice were crossed to homozygous ChR2 fl/fl mice to generate male ChR2^POMC mice as experimental animals. Male ChR2^WT littermates were used as controls.

Cell lines

Hepa1–6 cells (gender: female) were obtained from ATCC (ATCC® CRL-1830) passage number 30 and were not further authenticated in our laboratory. They were tested and found negative for ectromelia virus by ATTC as well as for mycoplasma in our laboratory. Cells were maintained under standard conditions in humidified incubator (Binder International) at 37°C, 5% CO₂ in DMEM (1x) + GlutaMAX™ Dulbeccos Modified Eagle Medium; 4.5 g/l D-glucose (GIBCO) supplemented with 10% fetal calf serum and 5% penicillin/streptomycin. Every 3–4 days the cells were passaged at 70 – 90% confluency.

METHOD DETAILS

Mouse experiments

Study consideration and design

The circadian clock impacts on almost all metabolic pathways in the murine liver and ER-stress networks have a periodicity of 12 hr with lowest activity mid through the light cycle (Zeigtgeber, ZT6) (Cretenet et al., 2010). ER volume (Chedid and Nair, 1972), norepinephrine levels (Terazono et al., 2003), ribosomal biogenesis (Sinturel et al., 2017), all oscillate in a circadian manner. Therefore, to control for this effect, livers used for any analysis in this paper were removed halfway through the light cycle at ZT6 ± 30 min.

Fasting, refeeding, and presentation of caged and faked food

For this experiment, 8-week-old male C57BL/6N mice were obtained from Charles River, France, and acclimatized to the facility for 8 days priorto the experiment. As mice were always sacrificed at ZT6 ± 30 min, mice were fasted accordingly. If mice were to be refed for either 30 min or 2 hr and sacrificed at ZT6, the former group was fasted at ZT13.5 and the latter ZT12. The mice were then refed at ZT5.5 and ZT4, respectively. In order to minimize the time of sacrificing the first to the last mouse in the same cage, the number of mice in each cage varied among experiments. For the induction of Fos mRNA in POMC neurons, mice were housed in pairs, as only two mice were perfused at the same time (Figures 6A and 6B). For studies, in which the liver was used for RNA sequencing (Figure 1A) or lipidomics (Figure 2A), mice were housed in groups of 3. For protein analysis and phosphoproteomic experiments, mice were housed in groups of 5 for the 30 min, 1 hr, 2 hr, and 4 hr time point and groups of 2 for the 5 min and 10 min exposure (Figures 3 and 4). The estimated time from killing the first mouse to the next, including removing the liver and snap freezing in liquid nitrogen, was 45 s for all studies except for the lipidomics, which was closer to 1 min due to washing of the liver in ice cold PBS to remove blood. Overall, the approximate time to collect 5 livers was 4 min. We did not observe any correlation between liver Xbp1 -splicing from the first to the fifth mouse indicating that the 4 min between the first and the fifth mouse in a group did not affect the results.

Norepinephrine injection with α1 adrenergic blockade (Prazosin)

C57BL/6N male mice, 5 in each cage, from Charles River, France, were fasted for 14 hr at ZT13 toZT13.25 and injected from ZT4.50 to Z5.15 with prazosin (0.5 mg/kg) at ZT 5.20 to ZT5.45 and after 30 min injected with either saline or norepinephrine (2 mg/kg) as previously described (Im et al., 1998). The liver was removed 30 min following injection of saline or norepinephrine at ZT5.50 to ZT6.15.

CNO administration in hM3Dq^Alfp and control littermates

Mice were fasted for 15 hr, then intraperitoneally injected with CNO (3 mg/kg body weight) and sacrificed 30 min later.

Stereotaxic surgical procedures

For all stereotaxic surgeries, animals were anesthetized with isoflurane and placed into a stereotaxic apparatus. For pain relief and postoperative care, mice were injected with buprenorphine (0.1 mg/kg) and meloxicam (5 mg per kg). Post-surgery, animals received tramadol in the drinking water (1 mg/mL), were inspected twice daily and body weight was monitored to ensure regain of pre-surgery weight.

Viral injections.

300–500 nL of AAVs expressing GCaMP6s or GFP (AAV.Syn.Flex.GCaMP6s.WPRE.SV40 or AAV1.pCAG.Flex. eGFP.WPRE) were injected into the ARC (coordinates from bregma AP: −1.45 mm; ML: ± 0.25 mm; DV: −5.80 mm to −6 mm) at 50 nL per min and the pipette was withdrawn 5–10 min after injection.

Fiber implantation.

Optical fibers (optogenetics: fiber core = 200μm, NA = 0.48, flat tip; fiber photometry: fiber core = 400μm, NA = 0.48, flat tip; both Doric Lenses Inc) were implanted over the ARC (coordinates from bregma, optogenetics, AP: −1.45 mm, DV: −5.6 mm, ML: 0 mm, fiber photometry, AP: −1.45 mm, DV: −5.7 mm, ML: 0.35 mm) and affixed to the skull using dental acrylic.

In vivo optogenetic studies

Fiber optic cables were firmly attached to the implanted fiber optic cannulae. ChR2^POMC and control ChR2^WT mice were acclimated to this procedure in their experimental cages for one week prior to optogenetic stimulation. On the day of the photostimulation experiment, food was removed and animals were attached to fiber optic cannulae at ZT0. At ZT6, blue laser (473 nm) light stimulation consisting of pulse trains (5 ms pulses of 20 Hz; 3 s on, 1 s off) was delivered. Following 30 min of photostimulation, mice were deeply anesthetized and liver samples were rapidly removed. Then, animals were perfused transcardially, their brains were removed and the locations of fiber tips were identified post hoc.

Fiber photometry

Experiments were performed 4 weeks after viral injections to allow for expression of the transgene. Animals were single housed for 1.5 weeks prior to the experiments. Five days before the first experimental day, animals were attached to fiber cable and kept connected throughout the experimental period. Photometry was performed using a custom-built fiber photometer (Model 925, Michael Dubbert, Electronic Laboratory, Institute for Zoology, University of Cologne), which was built following the principal specification described in (Chen et al., 2015) and as previously described (Steculorum et al., 2016). For ensuring compatability throughout all studies, mice were fasted from ZT15.30 to ZT5.45. At ZT5.45 the laser was turned on. The recordings consisted of a baseline period of 15 min followed by exposure to either food, caged food or caged fake food at ZT6 and the calcium signal was recorded for 30 min. Each experimental day was separated by 8–10 days. The mice were allowed to sense the food as for the other experiments. Photometry data were subjected to minimal processing using InstantClue. Since the signal was recorded every second and remains 0 in between, we calculated the maximum value within in a rolling window of 4500 data points. Smoothing of data was performed by accessing the moving average (4500 data points). To correct for differences in the baseline, we calculated the mean of the first 10 min and divided the data by this value. This procedure was performed individually for each recording. The area under curve was calculated using trapezium rule between the smoothed signal data and a horizontal line at y = 0.

Hepatic sympathetic nerve activity recording

Mice for this experiment were bred at the Max Planck Institute for Metabolism Research, Cologne, Germany. At the age of 7 weeks they were shipped to Dr. Kamal Rahmouni, University of IOWA, where the study was conducted as previously described (Bell et al., 2018; Tanida et al., 2015). hM3Dq^POMC or hM3Dq^WTmice were anesthetized by i.p. injection of a 9.1 mg/kg ketamine and 9.1 mg/kg xylazine mixture and a tapered micro-renathane tubing (MRE-40 for mice) was inserted into the jugular vein fori.v. injection. For monitoring of blood pressure, an MRE-40 arterial catheter was inserted into the left carotid artery. The trachea was then cannulated with PE-50 tube allowing the mice to spontaneously breath oxygen-enriched room air. The level of anesthesia was sustained throughout the experimental protocol by a slow infusion of α-chloralose through the jugular vein (12 mg/kg bolus followed by a 6 mg/kg constant infusion). Body temperature was monitored with a thermometer inserted into the rectum and was maintained at 37.0 −37.5°C using a heating pad. To record liver sympathetic nerve activity (Liv-SNA), the celiac and liver branch of the ventral splanchnic nerve were identified and exposed at the level of liver artery using a dissecting microscope. The nerve was attached to a pair of 36-gauge stainless steel wire electrodes and quickly fixed with a silicon gel (Kwik-Sil; WPI) to prevent dehydration and for electrical insulation. After surgery, each animal was allowed to stabilize for 20–40 min. Electrical activity in each nerve was amplified 50,000–100,000 times with a band path of 100–1000 kHz and monitored by an oscilloscope. The amplified and filtered nerve activity was converted to standard pulses by a window discriminator, which separated discharge from electrical background noise that was determined postmortem. Both the discharge rates and the neurogram were sampled with a Power-Lab analog-to-digital converter for recording and data analysis on a computer. Background noise, which was determined 30–60 min after the animal was sacrificed, was subtracted. Nerve activity was rectified and integrated with baseline nerve activity normalized to 100%. Baseline measurements of liver SNA was obtained during 5–10 min before iv injection of vehicle or CNO (3 mg/kg) and recorded for the following hour.

Preparations of solutions for in vivo experiments

α₁ blocker.

The dose used was 0.5 mg/kg. Prazosin was dissolved in H₂O to a concentration of 0.05 mg/mL and injected as 10 μl/g body weight (Im et al., 1998).

Norepinephrine.

The dose was 2 mg/kg dissolved in saline at a concentration of 0.2 mg/mL. For keeping the potency, each injected group had received separately prepared solutions, just prior to injection of the first mouse (Im et al., 1998).

CNO.

Dose used was 3 mg/kg. CNO was first dissolved in pure DMSO to a concentration of 50 mg/mL, then further diluted in saline to 0.3 mg/mL for injection.

Rapamycin.

The dose used was 5 mg/kg body weight. Rapamycin was dissolved in 75% saline, 10% ethanol, 10% PEG300 and 5% Tween 80 and injected as 5 μL/g body weight as described before (Liu et al., 2016).

Cell experiments

Hepa1–6 cell

Experimental procedure: 24 hours before the experiment, the cells were trypsinized, counted by using an automated cell counter (EVE automatic Cell counter NanoEnTek) and plated at a density of 2*10⁵ cells/well in 12-well plates (Becton Dickinson labware, France). Six hours before the stimulation, the medium DMEM (1x) + GlutaMAX™ -I Dulbeccos Modified Eagle Medium 4.5g/L D-glucose was replaced with serum free medium. Cells were stimulated with norepinephrine (10 μM dissolved in PBS (GIBCO)). The experiment was stopped by rapid removal of the cell medium and by directly submerging the plate in liquid nitrogen and kept at −80°C until processed for either RNA or protein preparation. For technical replicates the mean of three to six wells were used. At least three independent experiments were performed for each study.

Preparations of solutions for in vitro experiments

Norepinephrine was dissolved in PBS to a final concentration of 1 mM and 10 μl of this was added to each well containing 1 mL of medium. Rapamycin was dissolved in pure DMSO at a concentration of 10 μM. This was further diluted in cell medium to a final concentration of 100 nM for inhibition of protein phosphorylation (Liu et al., 2016) and 20 nM for blocking transcription as described before (Jefferies et al., 1997)

Tissue, serum, and plasma processing

Blood measurement.

Blood glucose was measured by a hand-held glucometer ContourXT, with Contour Next strips (REF #84167879) (Bayer, Germany).

Serum and plasma analysis.

For the fasted, 30, 60 and 120 minutes study, leptin and insulin concentrations were determined in duplicates using 5 ml of serum according to the manufacturer’s ELISA instructions. To generate serum, whole blood was allowed to clot at RT for 30 minutes followed by a 30 minutes centrifugation at 3000 g and stored at −80°C until used. For the determination of insulin and norepinephrine concentrations in mice which had been fasted or exposed to caged food or refed for 5,10 and 30 minutes, plasma was used. Plasma was prepared in the following way: 4 μl of 1 M sodium metabisulfite was added to EDTA-coated tubes, mixed with 400 – 500 μl blood and put on ice for 30 min before centrifugation for 30 minutes at 3000 g.

Processing of murine livers for the different analysis

For all studies, the gall bladder was quickly removed before freezing the livers in liquid nitrogen. For lipidomics, the livers were briefly washed in ice cold PBS to remove blood. During the perfusion of the mice with 0.9% saline, a small piece of the liver was removed and quickly frozen in liquid nitrogen.

RNA isolation from murine liver

The same part of the outer part of the left lobe was always taken for analysis. The piece of liver (approx. 10 mg) was snap-frozen in liquid nitrogen in 1.5 mLtube containing beads (1.4mm Zirconium oxide beads, Cat-No. KT03961–1-103.BK, Bertin Technoloiges, France). One mL ice-cold Qiazol (QIAGEN) was added to a small piece of liver (approx. 10 mg) on ice, a maximum of 12 samples was processed at one time. The samples were homogenized at RT using the fast prep machine (MP Biomedicals, Ohio, USA) for 40 s speed 6 m/s, allowed to stand at RT for 5 min, then 300 μL of chloroform was added. The tubes were vigorously shaken, spun for 15 min at 13000 rpm at RT. 300 μl of supernatant was mixed with 400 μL of 70% EtOH. From this step on the RNeasy Mini kit (QIAGEN) was used according to the manufacturers instruction. No DNase treatment was used for regular cDNA synthesis, qPCR and Xbp1 -splicing assay, but only used for RNA sequencing. RNA was eluted with 50 μL RNase-free water and kept on ice. The concentration was measured by Nanodrop ND-100 (Peqlab Biotechnology, France), and diluted to a final concentration of 12001500 ng/ μl. This RNA was used for RNA sequencing and cDNA synthesis. Expression of spliced Xbp1 using cDNA synthesized from this part correlated well with the expression using cDNA generated from RNA from a piece of liver from the inner part of the right lobe (data not shown).

RNA isolation from Hepa1–6 cells

For RNA isolations 12 well plates were allowed to thaw from −80°C. Then 300 μL of the RTL buffer (RNeasy Mini Kit (QIAGEN)) with 1 μLper 1 mL β-mercaptoethanol was added to each well. The solution was transferred to 300 μLof 70% EtOH. From this step on, the procedure was done according to the manual provided by the manufacturer of the RNeasy Mini kit (QIAGEN).

cDNA synthesis

From total liver mRNA, 2 μg were reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosytems, Cat #4368814). 2 μL buffer, 4.2 μLof MilliQ-purified H₂o, 2 μL random hexamer primers, 1 μL RT enzym and 0.8 μLdNTP were added to 2 μg of total liver RNA in 10 μL H₂O. Reverse transcription was performed at 25°C for 10 min, at 37°C for 60 min, and at 85°C for 5 min. 2 μg cDNA/RNA were diluted to a total of 200 μL with MilliQ-purified H₂O. RNA from Hepa1–6 was reverse-transcribed as described above, except that cDNA was synthesized from only 1 μg of total Hepa1–6 RNA and diluted to 100 μl.

qPCR

qPCR was performed using the QuantStudio 7 Flex Real-Time PCR System (Applied Biosystems) for amplification. Quantification was done using the delta Ct method and normalized to the control group in the given experiment. Hypoxanthine guanine phosphor-ibosyl transferase (Hprt) was used to normalize within each sample. See Table S4 for list of primers, TaqMan probes and SYBRGreen primer pairs. The SYBRGreen primers were synthesized by Eurogentec (Germany) and TaqMan probes were obtained from Applied Biosystems. The primers for Dnajc3 (p58ipk) and Herpud1 have previously been described (Yang et al., 2015) and the Xbp1(s) and Xbp1(u) primers, which share a common antisense primer, have been described before (Shao et al., 2014).

Xbp1-splicing assay

2 μl of cDNA (concentration 10 μg/μL), same as for qPCR, were used for amplification of unspliced and spliced Xbp1 in a total volume of 25 μL containing 1 × PCR buffer, MgCl (2.5 mM) mXbp11.3S (0.4 μM), mXbp1.12AS (0.4 mM) dNTPs (2.5 mM) Taq (5 U/mL) and MilliQ-purified H₂O. Mxbp 1.3S (5’ A AAC AGA GTA GCA GCG CAG ACT GC 3’) and mxbp 1.12AS (5’ TC CTT CTG GGT AGA CCT CTG GGA A 3’). PCR protocol: 94°C 4 min, (94°C 10 s, 68°C 30 s 72°C 30 s) for 35 cycles 72°C 10 min. Expected products: 473 bp unspliced and 450 bp spliced. PCR products were separated on a 2% agarose gel with ethidium bromide for 2.5 hr at 150 V.

Protein isolation from murine liver

Frozen livers were broken into smaller pieces in liquid nitrogen and 60 – 80 mg were transferred to 1.5 mL-tubes containing beads as for total RNA isolation. Livers were homogenized for 40 sat RT in 1 mLof the following buffer: Tris-HCL25 mM, Na₃VO₄100 mM, NaF 100mM, Na₄P₂O₄ 50 mM, EGTA 10 mM, EDTA 10 mM, NP-40 1%, PMSF 1mM pH 7.4 and cOmplete mini (1 tablet/10 mL) were freshly added. The samples were left on ice for 5 min, visually inspected for complete homogenization. If samples were not completely homogenized then the procedure was repeated. Samples were then centrifuged for 30 min at 17.000 g, the supernatant was taken for determination of protein concentration using the BCA assay (Pierce) relative to a BSA standard. Protein was mixed with Laemmli buffer (Bio-Rad) (1:4) containing β-Mercaptoethanol (1:10) and H₂O was added to a final concentration of 2 μg protein per μL.

Protein isolation from Hepa1–6 cells

Cells were lysed by first thawing the cells on ice and then adding 150 μL cell lysis buffer (Tris-HCl 25 mM, NaCl 25 mM, Na₃VO₄ 10 mM, NaF 10 mM Na₄P₂O₇ 10 mM and EGTA 1 mM with freshly added NP-40 1% and cOmplete Mini 1 tablet /10 mL pH 7.4) to each well. After pipetting up and down, cell lysates were transferred to a cold 1.5 mL-tube and centrifuged for 15 min at 17.000 g. The supernatant was mixed with Laemmli buffer (Bio-Rad) (1:4) and β-Mercaptoethanol (1:10). 10 μL were separated by SDS-PAGE using the same system and antibodies as for the liver (see below).

Western blotting

Proteins were separated using precasted gels from BioRad 4%−15%, 10% or 7,5% with either 26 or 18 wells. The voltage was set to 200 Vand kept constant. Transfer of protein to PVDF membranes was done using the Trans-Blot Turbo system (Bio-Rad) with eitherthe 7 or 10 min protocol. Tris/glycine running buffer (1×TGS buffer, Bio-Rad) was used as running buffer. After the transfer, the membranes were blocked in (1:20) western blotting reagent (Roche 11829200) for 1–2 hr and incubated over night in primary antibody. The next day, membranes were washed three times for 5–7 min in TBS-T pH 7.4 and incubated for 1–2 hr in secondary antibody, and then washed 4 times for 10 min in TBS-T pH 7.4. Proteins were visualized using the Fusion Solo Vilber Lourmat system. Band intensities were quantified using ImageJ (National Institutes of Health, Bethesda, United States). As a loading control, all membranes were probed with either beta actin orcalnexin. The intensity of phospho protein bands wasdevided by the intensity of beta actin orcalnexin. No stripping was done. For individual primary and secondary antibodies, lot numbers, company, and dilutions see Table S5. Forthe amount of protein loaded and blotting conditions, please see Table S5 for details. All primary and secondary antibodies were diluted in TBS-T pH 7.4,5% 1:20 western blotting reagent. Membranes: Trans-Blot (R)Turbo Bio-Rad Midi Format, 0.2 μm PVDF, single application. Gels 7.5%, 26 wells, 15 μL, 1.0 mm Criterion TGX Precast gel. Gels 10%, 26 wells, 15 μL, 1.0 mm Criterion TGX Precast gel. Gels 4 −15%, 26 wells, 15 μL, 1.0 mm Criterion TGX Precast gel.

Ultrathin sections and electron microscopy

Tissues were post-fixed in 2% glutaraldehyde (Electron Microscopy Sciences) in 0.12 M phosphate buffer and treated with 1% osmium tetroxide (Electron Microscopy Sciences). After dehydration using ethanol and propylene oxide, tissues were embedded in Epon (Sigma). For electron microscopy, 70 nm ultrathin sections were cut from Epon-blocks and stained with uranyl acetate (Plano GMBH) and lead nitrate (Sigma). Images were acquired using a transmission electron microscope (JEM-2100 Plus) equipped with Gatan ONE View camera. ER surface area was measured on individual electron micrographs from three mice per treatment condition with 8 images from each mouse and presented as a mean per picture. All quantifications were done in a blinded fashion using ImageJ (National Institutes of Health, Bethesda, United States).

Perfusion for RNA scope and immunohistochemistry, confocal imaging, and quantification

Tissue processing.

Animals were perfused transcardially with 0.9% saline followed by ice cold 4% paraformaldehyde (PFA; pH 7.4). The brains were removed and post-fixed in 4% PFA at RT for 18 hr, followed by incubation with 25% sucrose in 0.1 M phosphate buffered saline (PBS, pH 7.4) at 4°C for 24 hr. The brains were cut at 14 μm for RNAscope or 30 μm for immunohistochemistry on a freezing microtome and collected in bins containing sterile anti-freeze solution (30% ethylene glycol and 20% glycerol in PBS) and subsequently stored at −20°C until further processing.

RNAscope

Fluorescent in situ hybridization for the simultaneous detection of Fos, Pomc and Agrp mRNA was performed using RNAscope^®. All reagents were purchased from Advanced Cell Diagnostics (ACD, Hayward, CA) if not stated otherwise. The Fos probe (Cat No. 316921-C3) contained 20 oligo pairs and targeted region 407–1427 (Acc. No. NM_010234.2) of the Fos transcript, and the POMC probe (Cat No. 314081-C2) contained 10 oligo pairs and targeted region 19–995, (Acc. No: NM_008895.3) of the Pomc transcript. The AgRP probe (Cat No. 400711-C2) constituted 16 oligo pairs and targeted region 11–764 of the AgRP transcript (Acc. No. NM_001271806.1). Three-plex negative and three-plex positive control probes recognizing bacterial dihydrodipicolinate reductase, DapB (Cat No. 320871) and PolR2A, cyclophilin and Ubiquitin (Cat No. 320881), were processed in parallel with the target probes to ensure tissue RNA integrity and optimal assay performance. All incubation steps were performed at 40°C using the ACD HybEz hybridization system (Cat No. 321462). On the day before the assay, every 12th section throughout the ARH was mounted on SuperFrost Plus Gold slides (Cat No. FT4981GLPLUS; ThermoFisher), dried at RT, briefly rinsed in autoclaved MilliQ-purified water, air-dried and baked at 60°C overnight. From each animal, one section from the same region of the brain was also mounted for use with the negative control probe to enable subsequent calculation of background. On the day of the assay, slides were first incubated for 7 min in ACD, submerged in Target Retrieval (Cat No. 322000) at a temperature of 98.5 – 99.5°C for 8 min, followed by two brief rinses in autoclaved MilliQ-purified water. The slides were quickly dehydrated in 100% ethanol and allowed to air dry for 5 min. A hydrophobic barrier was then created around the sections using an ImmEdge hydrophobic barrier pen (Cat No. 310018). The sections were incubated with Protease III (Cat No. 322340) for 40 min for the detection of Pomc and Fos, or Protease Plus (Cat. No. 322330) for 25 min for the detection of Pomc, AgRP and Fos. The subsequent steps, i.e., hybridization of the probes and the amplification and detection steps, were performed according to the manufacturer’s protocol for RNAscope^® Multiplex Fluorescent Assay (Cat No. 320851; for Pomc+Fos), which in the end labeled the POMC probe with Atto 550 and the Fos probe with Atto 647, or, for the simultaneous detection of Pomc+AgRP+Fos, the tyramide-based RNAscope^® Multiplex Fluorescent v2 Assay (Cat. No. 323110) that rendered probes labeled with OpaI520 (1:750), Cy3 (1:3000) and Cy5 (1:3000) respectively. Sections were counterstained with DAPI and coverslipped with ProLong Gold Antifade Mountant (Cat. No. P36931; ThermoFisher) and stored in the dark at 4°C until imaging.

Immunofluorescence

Following the RNAscope procedure for AgRP and POMC (see above, with the modification that incubation with Protease III was only 10 minutes to avoid degradation of the GFP epitope), slides were briefly washed in PBS and then blocked in 3% goat serum/1 x PBS (Triton-X) for 1 hr at RT. After a brief wash in 1 x PBS the slides were incubated over night in chicken anti-GFP antibody (Abcam no. 13970,1:500). The following morning, sections were washed intensively for 30 min in 1 x PBS and incubated with a goat anti-chicken antibody (1:500) for 1 hr at RT. After washing for 30 min, sections were counterstained with DAPI and coverslipped with ProLong Gold Antifade Mountant (Cat No. P36931; ThermoFisher) and stored in the dark at 4°C until imaging. Staining of POMC and GFP: Brain sections were washed in PBS with Tween-20, (pH 7.4 (PBST)) and blocked in 3% normal donkey serum in PBST for 1 hr at RT. Brain sections were then incubated overnight at RT in blocking solution containing primary antiserum (rabbit anti-POMC precursor, Phoenix Pharmaceuticals H-029–30, 1:1.000; chicken anti-GFP, Life Technologies A10262, 1:1.000). The next morning, sections were extensively washed in PBS for 30 min followed by incubation in Alexa-fluorophore secondary antibody (Molecular Probes, R37119, A-11039, both 1:1,000) for 1 hr at RT. After several washes in PBS, sections were mounted on slides after being counterstained with DAPI.

Imaging and quantification

Images were captured using a confocal Leica TCS SP-8-X microscope, equipped with a 40x/1.30 oil objective. Tile scans and Z stacks (optical section of 1.0 μm) of the ARC were captured unilaterally from rostral to caudal, rendering approx. 5 sections per animal. Laser intensities for the two probe channels were kept constant throughout the entire material. Images were imported into Fijii (NIH) where maximum intensity projections were made. The DAPI channel was enhanced regarding brightness and contrast, but the probe channels were left unmodified. The images were then imported and fused into the Halo software (Indica Labs) for quantification of double-labeled neurons. The software relies on the DAPI stain for cellular identification and automatically calculates the cell intensity for each cell and probe (an integrated number containing both the fluorescent intensity and the area covered by the probe within the designated cell). The threshold for probe recognition was calculated as the mean cell intensity present in the negative control sections + 3xSD. All labeling above this value was considered to be true signal.

Transcriptomics

After total RNA isolation (see above), 2 μg of total RNA was sent to Cologne Center for Genomics. The quality of the RNA integrity was assessed by Agilent 2200 TapeStation System and all samples had an RNA integrity number above 7. PolyA mRNA libraries were prepared using TruSeq^® RNA sample preparation Kit v2 (Illumina). Libraries were sequenced for 50 million reads 50 bp singleend on a Illumina HiSeq 2000 sequencer with a paired-end (101×7×101 cycles) protocol. After sequencing, raw reads were assessed for quality, adaptor content and duplication rates with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Reaper version 13–100 was employed to trim reads after a quality drop below a mean of Q20 in a window of 10 nucleotides. Only reads between 30 and 150 nucleotides were cleared for further analysis. Trimmed and filtered reads were aligned versus the Ensemble mouse genome version mm10 (GRCm38) using STAR 2.4.0a with the parameter -outFilterMismatchNoverLmax 0.1 to increase the maximum ratio of mismatches to mapped length to 10% (Dobin et al., 2013). The number of reads aligning to genes was counted with featureCounts 1.4.5-p1 tool from the Subread package. Only reads mapping at least partially inside exons were admitted and aggregated per gene. Reads overlapping multiple genes or aligning to multiple regions were excluded. Differentially expressed genes were identified using DESeq2 version 1.62 (Love et al., 2014). The Ensemble annotation was enriched with UniProt data (release 06.06.2014) based on Ensemble gene identifiers (Activities at the Universal Protein Resource (UniProt)). Only genes with a minimum fold change of ± 2 (Log₂ ± 1), a maximum Benjamini-Hochberg corrected p value of 0.05, and a minimum combined mean of 5 reads were deemed to be significantly differentially expressed

Phosphoproteomics

Protein lysis and digestion.

Five biological replicates of each experiment (refed, caged food and fasted) were performed. Livers were crushed in liquid nitrogen using a mortar. Six Lys6 labeled SILAC livers were used to generate an internal standard that was the same for all experiments. Liver powder was then dissolved in RIPA buffer (150 mM NaCl, 1% NP-40, 0.2% Na-Desoxycholat, 0.1% SDS in 10 mM Tris HCl pH 7.5) containing protease and phosphatase inhibitors (Biotol) using 1 mL buffer per 10 mg powder. Samples were sonicated on ice and centrifuged at 14.000 g at 4°C. Supernatant was then used for acetone precipitation overnight at −20°C, the pellet was washed with 90% acetone, dried and dissolved in 6 M Urea, 2 M Urea in 10 mM HEPES. Then, protein concentration was determined and 5 mg of protein were mixed with the heavy SILAC standard at a one to one ratio. Proteins were reduced by 10 mM dithiothreitol and alkylated using 55 mM Iodacetamide as described before (Nolteet al., 2014). Lys-C endopeptidase was added in an enzyme to protein ratio of 1 to 100 and incubated for 2 hr at RT. Urea was diluted to 2 M by adding 50 mM Ammonium bicarbonate and additional Lys-C was used in the same ratio. Digestion was stopped by adding 5% Acetonitrile and 0.2% TFA in a one to one ratio.

High pH fractionation and phosphopeptide enrichment.

Peptides were desalted by C18 Cartridges (Waters GmbH) including several washing steps of bound peptides by 0.1% TFA and three-step elution by 40% and twice 60% Acetonitrile (ACN) containing 0.1% TFA (600 μL each). Samples were dried in a speed-vac and dissolved in 10 mM Ammonium hydroxide to a final volume of 1.6 mL. High pH fractionation was done on an Ultimate 3000 HPLC (Thermo Scientific). Complete samples were loaded via 4-step injection onto a Waters X-bridge column (Waters XBridge BEH130 C18 3.5 μm 4.6 × 250 mm column). The column temperature was set to 35°C. For peptide separation a binary buffer system was used: Buffer A) 10 mM ammonium hydroxide and B) 90% ACN, 10 mM ammonium hydroxide. Due to the relative high sample volume and smaller sample-loop volume (400 μL) each sample was injected four times within a high pH run. Peptides were loaded onto the column using 5% B at 1 mL/min between each injection step for 1 min and after the fourth injection this was held for further 2 min. In a linear shape the buffer B content was increased from 10% to 25%, ramped to 40% in 5 min, further increased to 95% B within 5 min and held for 3 min. Followed by a re-equilibration step of 5 min after ramping the gradient back to loading conditions (5% B). Fractions were collected for a total volume of 1000 μL and prior collection 200 μL of 80% ACN and 5% TFA was added to each vial to immediately shift the pH to acidic conditions. Fractions were concentrated in a speed-vac for at least 12 h to almost complete dryness and pooled as follows. Every 13^th fraction was pooled and adjusted to binding conditions of TiO₂ beads 80% acetonitrile and 7% TFA in a 5 μL tube (final volume 2–3 mL). Prior to phosphopeptide enrichment TiO₂ beads were washed in 40% ACN and 10% ammonium hydroxide, followed by 2 washing steps of 70% acetonitrile and 80% acetonitrile/ 5% TFA. Then TiO₂ beads (5 μm Titansphere, GL Sciences) were dissolved in binding buffer (80% ACN, 7% TFA by 10 μL/mg of beads) as described previously (Krishnan et al., 2015). Then, 2.5 mg TiO₂ beads were added to each fraction and incubated on a rotating wheel for 30 min at RT. Beads were collected by centrifugation (1000 g, 30 s) and supernatant was transferred to a fresh 5 mL tube containing further 2.5 mg of TiO₂ beads. Beads were washed with 2 × 2 mL 80% ACN and 7% TFA and loaded onto C8-StageTips. Beads were washed further three times on-tip using 30% ACN, 1% TFA, followed by 60% ACN, 1% TFA and 80% ACN, 1% TFA. Beads were dried with a syringe and phosphopeptides were eluted with 5% ammonium hydroxide and twice with 10% ammonium hydroxide, 40% ACN. Eluate was collected fraction-wise in a 96 well plate and concentrated in a speed-vac for 2.5 hr for complete dryness and resuspended in 10 μL 2.5% ACN and 5% formic acid.

Liquid chromatography and mass spectrometry.

LC-MS/MS instrumentation consisted out of an Easy nLC 1000 (Thermo Fisher) coupled via a nono-spray ionization source to a QExactive Plus mass spectrometer. Peptides were separated on a 50 cm column (I.D. = 75 μm, packed with 1.7 μm C18 Beads, Dr. Maisch) using a two-buffer system: A) 0.1% formic acid and B) 0.1% formic acid in acetonitrile. Content of buffer B was increased over 90–120 min gradient time from 5% to 23% within 70% of the total time, followed by an increase to 55% and 85% and a re-equilibration step to 5% before the next sample was loaded onto the column. MS1 level spectra were acquired at a resolution of 70,000 at 200 m/z, using 3e6 as an AGC target within a maximum injection time of 20 ms. The instrument operated in atop 10 mode and sequentially isolated the top 10 most intense peaks using an isolation window of 1.9 m/z for HCD fragmentation at normalized collision energy of 27. The resolution was set to 35,000 at 200 m/z. We used an AGC target of 5e5 and a maximum injection time of 110–120 ms maximizing parallelization and high quality MS2 spectra for phosphosite localization and identification (Kelstrup et al., 2012).

LC-MS/MS data analysis

All raw data were processed using MaxQuant 1.5.3.8 and the implemented Andromeda search engine (Cox and Mann, 2008; Cox et al., 2011). Acquired MS/MS spectra were used for identification of modified and unmodified peptides using the Uniprot reference proteome of Mus musculus (2016). A Lys-C/P was set as the used protease and a maximum of 2 missed cleavages was tolerated. Heavy SILAC channel was defined by Lys6 as the label. Mass tolerances were kept as by default. The FDR was estimated by the implemented decoy algorithm and calculated to 0.01 at the protein, site and peptide-spectrum-match level by the revert method. Carbamidomethylation at cysteine residues was set as a fixed modification while phosphorylation at Serine, Threonine and Tyrosine as well as Methionineoxidation and protein N-term acetylation were defined as variable modifications. A minimal ratio count of 1 was required for quantification and peptide length of 7 amino acids was defined. The re-quantify and match-between runs algorithms were enabled using default settings. Log₂ SILAC ratios were normalized by median substrac-tion. In order to identify significantly different phosphorylation sites a t test was applied. The set of similar regulated phosphorylation sites were identified by the fold change compared to fasted (1.5 fold in the same direction e.g., up or downregulated compared to fasted). Protein annotations such as Gene Ontology terms, Pfam and KEGG pathways are based on Uniport while substrate motif information were annotated via Perseus (Tyanova et al., 2016). Perseus was also used to analyze data by Principal Component Analysis (for all omics-levels). To identify enriched gene ontologies we performed a fisher exact test and used a specific target group (for example the group of phosphorylation sites that were regulated similarly between refed and caged food compared to fasted against all identified phosphorylation sites. We aimed to compare phosphorylation and mRNA adaptions in response to refed and caged food. To achieve this, we conducted a 2D enrichment (Cox and Mann, 2012). We mapped mRNA expression data based on Uniprot IDs to the phosphorylation data and performed a 2D enrichment for refed compared to fasted and caged food compared to fasted on these two omics levels separately. Finally, we overlapped the two lists of systematically up/downregulated annotations. Phosphoproteomics data are available via ProteomeXchange with identifier PXD005681 (Vizcaino et al., 2016).

Lipids

The whole liver was quickly removed without the gallbladder and washed in ice cold PBS on ice and then snap frozen in liquid nitrogen and stored at - 80°C until processed further. Livers were then transferred to a ceramic mortar containing liquid nitrogen also submerged in liquid nitrogen. Using the mortar, the livers were crushed to fine powder and 50–80 mg was weighed off and transferred to a plastic tube containing ceramic beads. Extraction was done by adding 1.5 mL methanol/H₂0 (1:1) containing 2 mM deuterium labeled acetate, 40 mM norvaline and 2 μL of a deuterium-labeled lipid standard (Aventis). The samples were homogenized using the fast prep machine (MP Biomedicals, Ohio, USA) 4 × 40 s speed 6 m/s. After inspection of the sample for total tissue lysis, the samples were centrifuged 10 min 16.000 g at 4°C and the supernatant was transferred to a new 1.5 mL-tube, and dried using a SpeedVac (aqueous extract) used for GC-MS. 1.5 mL of ice cold dichlororomethane/methanol mixture (3:1) was added to the pellet and the samples were homogenized again using the fast prep machine 4 × 40 s speed 6 m/s. After centrifugation at 16.000 g for 10 min, the supernatant was collected and dried in a vacuum concentrator. This is the organic extract for GC-MS, which was sent to MS-Omics ApS (Copenhagen, Denmark) for lipidomics.

Sample analysis was carried out by MS-Omics as follows: The samples reconstituted in an isopropanol/methanol/water (2:1:1) mixture. The analysis was carried out using a UPLC system (UPLC Acpuity, Waters) coupled with a time of flight mass spectrometer (Xevo G2 Tof, Waters). An electrospray ionization interface was used as ionization source. Analysis was performed in negative and positive ionization mode. The UPLC was performed using a slightly modified version of the Waters application note described by Isaac et al. (Giorgis Isaac, Stephen McDonald, and Giuseppe Astarita, Lipid separation using UPLC with charged surface hybrid technology, Waters application notes, 2011). Data processing was performed in MZmine 2 (T. Pluskal, S. Castillo, A. Villar-Briones, M. Oresic, MZmine 2: Modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data (Pluskal et al., 2010)). Identification of compounds was performed using a both accurate mass (with an accepted deviation of 0.01 Da) and retention time in relation to chain length and double bonds.

Analysis of lipid data

Lipids were filtered for a Descriptive Power (relative standard deviation to control samples) above 2.5 and intensity values were Z-Score normalized (mean = 0, standard deviation = 1). Lipids were filtered for the main classes and unidentified features were deleted. Note that in the provided Table S3 neither a filtering nor a normalization was done. Then we used a PC analysis to determine effects within samples with the help of InstantClue.

QUANTIFICATION AND STATISTICAL ANALYSIS

Group sizes commonly applied in murine studies were used depending on the analysis in question. Sample numbers of data obtained for physiological parameters or on mouse tissue refer to the number of individual mice entering the analysis as specified in the figure legends (Figures 1,2,3,4,5,6, 7, S1, S2, S3, S4, S5, and S6). Experiments on cultured cells were performed in at least 3 independent experiments, each performed at least in triplicates as specified in the figure legend (Figure 7). For comparing two groups an unpaired two tailed t test was used to dertermine significance. For differences among more than 2 groups a one-way ANOVA was used for an overall difference. If significant, this was followed by multiple comparisons tests comparing all other groups to the control group. A one-way ANOVA was also used when a caged food and refed sample were compared to a common fasted group, i.e., comparing the effect of time of caged food and refeeding on gene expression in liver or serum parameters like insulin or leptin and only this is depicted in the Figures. For comparing the effect of two parameters eg Rapamycin or genotype under different conditions, a two-way ANOVA was applied. When significant, this was followed by multiple comparisons tests to find individual differences within each condition and only this is shown in the Figure. The Western blot data in the Rapamycin-treated mice was analyzed with a one-way ANOVA as Rapamycin completely prevented phosphorylation. Data are presented as Box-whisker plots with upper and lower quartile, median, the minimum and maximum values and all individual data points with the y axis being determined by InstantClue. Statistical Analysis was performed using GraphPad Prism 7 Software.

Highlights.

• Food perception activates hepatic mTOR and Xbp1 signaling

• Food perception promotes changes in hepatic phosphatidylcholine synthesis

• POMC neurons rapidly respond to food perception and control SNA to liver

• Norepinephrine signaling in liver stimulates mTOR and Xbp1 activation