Calorie restriction activates a gastric Notch-FOXO1 pathway to expand ghrelin cells

Restricting calories can extend lifespan (McCay et al., 1989) in humans, as well as other species (Weindruch et al., 1986). Calorie restriction also improves health by delaying the onset of age-related diseases including degenerative brain disease, heart disease, type 2 diabetes, and certain types of cancer (Colman et al., 2009; Madeo et al., 2019). The mechanism of the alleged protective effect of calorie restriction is unknown. Interestingly, energy expenditure and oxidative stress markers decrease in humans who restricted their diet by 15% over a 2-year period (Il’yasova et al., 2018; Redman et al., 2018). Cardiometabolic risk factors (Most et al., 2018) and circulating tumor necrosis factor-α (Ravussin et al., 2015), a marker of inflammation, also show a similar pattern.

Epithelial cell types have specialized functions. Distinct processes of the gastrointestinal (GI) tract, such as the absorption of lipids and nutrients, are driven, in part, by distinct subtypes of epithelial cells, which are repopulated by stem and progenitor cells, including Lgr5 multipotent stem cells in the intestine (Barker et al., 2007) and antral stomach (Barker et al., 2010).

Enteroendocrine cells located throughout the gut sense nutrients and release hormones that regulate food intake and satiety (Gribble and Reimann, 2016). These cells are distributed along the GI tract in a region-specific manner (Thompson et al., 2018). For example, intestinal K− and L− cells, which produce the gut incretins GIP and GLP-1 respectively, are enriched in the upper and lower intestines, respectively (Baggio and Drucker, 2007). On the other hand, the hormone ghrelin, released during food restriction to increase food intake (Kojima et al., 1999), is most prominently produced by gastric epithelial cells (Drucker, 2007). Mechanisms that drive this regional specificity in the distribution of enteroendocrine subtypes are not clear but likely involve differentiation of stem cells, such as Lgr5+ cells, and endocrine progenitor cells, marked by the transcription factor Neurogenin3 (Neurog3).

The gastrointestinal tract is affected by calorie restriction. Interestingly, in the small intestine, calorie restriction expands crypt size, while decreasing villi length (Yilmaz et al., 2012). In addition, it decreases the number of different cell types, including chromogranin A+ enteroendocrine cells, while increasing the proliferation of Lgr5+ stem cells (Igarashi and Guarente, 2016). Fasting also promotes fatty acid oxidation within intestinal stem cells (Mihaylova et al., 2018). The role of Lgr5+ stem cells in antral gastric tissue (Barker et al., 2010) upon calorie restriction has not been studied.

Distinct programs are activated by calorie restriction in each organ. For example, high-throughput transcriptomic, proteomic, and metabolomic analyses of the liver found that calorie restriction induces RNA processing (Rhoads et al., 2018). In contrast, it triggers remodeling of skin and fur (Forni et al., 2017). Less is known about how calorie restriction impacts stomach cells, although ghrelin levels reportedly increase (Reimer et al., 2010). It has also recently come to light that cells in each tissue have a unique transcription network to regulate the reversal of age-related gene expression changes by calorie restriction (Ma et al., 2020).

Herein, we report that calorie restriction initiates a distinct gastric program affecting stem and endocrine progenitors in the stomach by modulating Notch activity. Further, during calorie restriction, the transcription factor FOXO1 preferentially undergoes nuclear translocation in gastric endocrine progenitor cells to trigger an increase in ghrelin+ cells, thereby stimulating food intake.

Plasma ghrelin increases after calorie restriction (Reimer et al., 2010). To investigate whether this is due to increased ghrelin cell number, we calorie-restricted female mice by 30% of their daily food intake (CR) for 4 wk and assessed the abundance of gastric cell populations after food restriction. Controls had unlimited access to food (ad libitum) during this time. After calorie restriction, wild-type female CR mice showed decreased weight (Fig. 1 A), lean mass (Fig. 1 B), and fed glucose (Fig. 1 C) compared with controls. The latter gained weight, as expected (Fig. 1 A). A similar decrease in body weight (Fig. S1, A, C, D, and E; and Fig. 1 E), lean mass (Fig. S1, B, F, G, and H; and Fig. 1 H), and fasting glucose (Fig. S1 I) was seen in a second cohort of female FOXO1-Venus (FoxV) knock-in reporter mice (Kuo et al., 2019).

We analyzed the stomach of these mice in the corpus, composed of gastric oxyntic glands that contain acid-secreting parietal cells to aid in food digestion, and the antrum, which funnels digested food into the intestine for further processing. The glandular epithelium of the antral stomach contains gastrin-secreting endocrine cells, but not (in mouse tissue) parietal cells. Interestingly, when we assessed ghrelin+ cell number by immunostaining, we found a doubling of cell abundance in both antrum and corpus (Fig. 1 D and Fig. S2 A).

In contrast, other endocrine cell populations positive for 5HT and gastrin were unaltered (Fig. 1, E and F; and Fig. S2, B and C). We saw a slight but significant increase in glucagon cells (Fig. 1 G and Fig. S2 D). We next examined alterations in differentiated endocrine cells after calorie restriction using general endocrine cell markers. While the number of synaptophysin+ cells did not change, we found a significant increase of chromogranin A+ cells in the antrum (AL: 4.2% versus CR: 7.8%) and corpus (AL: 5.8% versus CR: 9.1%) (Fig. 1 H and Fig. S2 E).

To determine whether these changes were dependent on biological sex, we calorie-restricted male mice by 30% over 4 wk. At the end, animals had lower body weight (Fig. 1 I), lean mass (Fig. 1 J), and fed glucose (Fig. 1 K) compared to male mice fed ad libitum. Similar to female mice, males also had more ghrelin+ cells in both antrum (AL: 0.8% versus CR: 1.8%) and corpus (AL: 1.1% versus CR: 2.2%) (Fig. 1 L and Fig. S3 A). This corresponded with an increase in chromogranin A+ endocrine cells from 1.8 to 4.2% in the antrum and 3.4–5.7% in the corpus (Fig. 1 M and Fig. S3 B).

We recapitulated these changes ex vivo by culturing primary stomach cultures (PSC) in a serum-deprived medium. The abundance of ghrelin+ cells in PSC increased from 1.6% in normal medium to 3.8% when grown in serum-deprived medium (Fig. 2 A). Similarly, chromogranin A+ cells also increased from 3.8 to 9.0% with serum deprivation (Fig. 2 A).

To determine the signaling mechanisms regulating these changes in endocrine cell abundance, we surveyed molecular pathways involved in gastric cell differentiation. We found a substantial increase in cells positive for the Notch transcriptional target Hes1 after serum deprivation, consistent with increased Notch activity (Fig. 2 B). Importantly, we also saw increased Notch ligand levels under the same conditions (Fig. 2, B and D).

To test whether the increase in Notch-active cells with serum deprivation was linked to the increased ghrelin cells, we blocked Notch signaling in PSC with the γ-secretase inhibitor DAPT (Fig. 2 C). DAPT treatment diminished the increased Jag1 mRNA in serum-deprived PSC (Fig. 2 D). Interestingly, the increased number of ghrelin+ cells in PSC caused by serum deprivation was reversed by overnight treatment with DAPT (Fig. 2 E). Conversely, cre recombination–mediated expression of the Notch intracellular domain (NICD) triggered an increase of chromogranin A-immunoreactive cells under control (7.5 versus 3.2%) and serum-deprived (17.7 versus 10.4%) conditions (Fig. 2 F). Cells overexpressing NICD, identified by cre activation of a fluorescence Tomato reporter, also contained increased chromogranin A (7.93 ± 2.03 versus 1.08 ± 0.28) and ghrelin (3.21 ± 0.96 versus 0.55 ± 0.18) mRNA (Fig. 2 G). Adenoviral-mediated expression of NICD-GFP increased the propensity of cells to differentiate into ghrelin+ cells fourfold (Fig. 2 H). These data suggest that activation of Notch signaling is necessary and sufficient to increase chromogranin A+ and ghrelin+ cells during calorie restriction.

Notch signaling regulates gastric Lgr5 stem cell differentiation (Demitrack et al., 2015). To determine if changes in differentiated endocrine cells were due to changes in Lgr5+ stem and Neurog3+ progenitor populations, we calorie-restricted Lgr5-GFP and Neurog3-GFP knock-in reporter mice. In the stomach, the former reporter labels a population of self-renewing, multipotent antral stem cells that maintain gastric epithelial cell homeostasis (Barker et al., 2010), whereas the latter marks Neurog3+ endocrine progenitor cells that differentiate into endocrine cells (Schonhoff et al., 2004). These reporter mice had lower glucose levels (Fig. 3 A) and decreased lean mass (Fig. 3 B) upon calorie restriction. In the stomach, calorie restriction triggered a decrease in antral Lgr5-GFP+ cells (Fig. 3, C and D). In contrast, cells that express transcription factor Sox2, another marker of multipotent stomach stem cells (Arnold et al., 2011), increased in the stomach and corpus (Fig. 3 E).

To test the involvement of Notch signaling in stem cell maintenance after calorie restriction, we used PSC. Serum-deprived PSC isolated from Lgr5-GFP mice had 40% fewer Lgr5-GFP+ cells compared with the serum-containing medium (Fig. 3 F). This decrease was dependent upon Notch activation, as treatment with DAPT restored Lgr5-GFP+ cell number (Fig. 3 F). To determine which cells activated Notch in response to food restriction, we immunostained Lgr5-GFP CR stomach tissue. Interestingly, there was a broad increase in Hes1 and Jag1 abundance throughout the entire antral gastric epithelial with calorie restriction, indicating that Notch is also activated outside of the Lgr5-GFP stem cells themselves (Fig. 3, G and H).

Using the calorie-restricted reporter mice described above (Fig. 3, A and B), we next analyzed endocrine progenitor cell populations. Neurog3-GFP+ endocrine progenitor cells increased, particularly in the antral stomach, after calorie restriction (Fig. 4, A and B). Similar to in vivo, PSC grown in a serum-deprived medium increased Neurog3-GFP+ cells by 180% as detected by immunohistochemistry (Fig. 4, C and D). This increased abundance was reversed when cultures were treated with the Notch inhibitor, DAPT (Fig. 4, C and D). Serum deprivation also increased the abundance of Neurog3-GFP+ cells (0.58% ± 0.031 versus 0.47% ± 0.01) as detected by endogenous fluorescence using FACS (Fig. 4, E and F). Neurog3-GFP+ cells were enriched for Neurog3 mRNA, confirming the identity of the cell population (Fig. 4 G). Upon further analysis, Hes1 mRNA was also increased in serum-deprived Neurog3-GFP+ cells, indicating Notch activation in these cells (Fig. 4 H). Similar to previous reports (McKimpson et al., 2022; Talchai et al., 2012), Neurog3-GFP+ cells were also enriched for FOXO1 (Fig. 4 I). Interestingly, FOXO1 mRNA was further increased in Neurog3-GFP cells grown under serum-deprived conditions (Fig. 4 I).

Given the change in FOXO1 levels, we next analyzed the effect of calorie restriction on FOXO1-expressing cells in the stomach. To detect FOXO1, we used GFP immunohistochemistry in FoxV reporter mice (McKimpson et al., 2022) (denoted as FOXO1-GFP), with pancreatic islets acting as a positive control (Fig. S4 A). No reactivity was detected in wild-type mice (Figure S4 B). Immunohistochemistry revealed that stomachs of calorie-restricted mice had ∼135% more FOXO1-GFP+ cells in the antrum (Fig. 5, A and B), with some regions of FoxV CR mice having significantly more FOXO1-GFP+ cells (Fig. 5 C). There were also ∼170% more FOXO1-GFP+ cells in the corpus of FoxV CR mice (Fig. 5, D and E).

Changes in cell number can be due to proliferation. Interestingly, under basal conditions, FOXO1-positive stomach cells are minimally proliferative, as indicated by the paucity of FOXO1-GFP+ cells colocalizing with proliferation markers phospho-histone H3 (pHH3) (Fig. 5 F) and Ki67 (Fig. 5 G). However, there was a significant 195% increase in FOXO1-expressing GFP+, pHH3+ cells after calorie restriction (Fig. 5 H). FOXO1-expressing Neurog3+ cells also increased after calorie restriction (Fig. 5, I and J).

In ad libitum-fed conditions, FOXO1 is mainly detected in the cytoplasm of gastric cells by immunohistochemistry (McKimpson et al., 2022). However, FOXO1 actively translocates between the cytoplasm and the nucleus, the latter where it drives the expression of key transcriptional targets. To test the role of nuclear FOXO1 in gastric endocrine cell remodeling following calorie restriction, we studied PSC from knock-in mice homozygous for a constitutively nuclear acetylation-defective FOXO1 (FoxKR) (Banks et al., 2011). On C57BL/6 background, FoxKR mice display glucose and insulin levels similar to control mice and have lower plasma and hepatic triglyceride levels (Qiang et al., 2012). As expected, Foxo1 mRNA levels were similar in PSC from wild-type and FoxKR mice (Fig. 6 A). While expression of different stomach cell markers, including Lgr5 and Neurog3, were minimally altered under ad libitum-fed conditions, ghrelin mRNA levels (1.97 ± 0.18 in FoxKR versus 0.85 ± 0.13 in wild type) were significantly increased in FoxKR PSC (Fig. 6 A), as were Ghrelin+ cells (3.30% in FoxKR versus 0.71% in wild type) (Fig. 6 B).

We confirmed these results by immunohistochemistry of stomach sections. FOXO1 was nuclear in FoxKR mice (Fig. 6 C). While the number of ghrelin+ cells in the antrum of these mice was slightly changed from controls, there was a striking increase in ghrelin+ cells in the corpus (6.20% in FoxKR versus 2.58% in wild type) (Fig. 6, D and E). In line with ghrelin being an orexigenic hormone, FoxKR mice also had increased food intake (Fig. 6 F). Furthermore, treatment of serum-deprived PSC overnight with a FOXO1 inhibitor (compound 10 [Langlet et al., 2017]) reversed the increased abundance of ghrelin+ (Fig. 6 G) and chromogranin A+ (Fig. 6 H) cells under these culture conditions. Taken together, our results suggest a role for nuclear FOXO1 in promoting the differentiation of ghrelin+ cells with calorie restriction.

To test whether the role of FOXO1 was dependent on Notch signaling, we treated PSC from wild-type and FoxKR mice with the Notch inhibitor DAPT and measured chromogranin A+ and ghrelin+ cells by immunohistochemistry (Fig. 6 I). PSC generated from FoxKR stomachs had an increased number of ghrelin+ cells (3.3%) compared with ghrelin+ cells in wild-type PSC (1.4%) (Fig. 6 J). In PSC from wild-type mice, ghrelin+ cells increased to 2.8% upon serum deprivation and this was reversed to 1.2% by DAPT treatment. Neither treatment affected the number of ghrelin+ (Fig. 6 J) or chromogranin A+ (Fig. 6 K) cells in FoxKR PSC. We also assessed Notch signaling in wild-type and FoxKR stomach. Expression of Notch components Hes1 and Jag1 was similar in the antrum and corpus from both genotypes (Fig. 6, L and M). These data link nuclear FOXO1 with ghrelin cell differentiation downstream of Notch activation.

Similar to another report (Reimer et al., 2010), we found increased plasma ghrelin levels in calorie-restricted mice (Fig. 7, A and B). This increase occurred when either female (Fig. 7 A) or male (Fig. 7 B) mice were food-restricted. To determine whether increased ghrelin contributed to the beneficial effects of calorie restriction, we blocked ghrelin signaling using the ghrelin receptor antagonist GHRP-6 (Asakawa et al., 2003) for 5 days. Importantly, we continued to calorie-restrict mice during this treatment, and both vehicle- and GHRP-6-treated mice consumed their food in its entirety throughout the duration of the experiment. While body weight was minimally increased in mice treated with GHRP-6 (Fig. 7 C), these mice had elevated fed glucose levels (GHRP-6: 144.2 ± 5.8 mg/dl versus Vehicle: 113.2 ± 5.3 mg/dl) (Fig. 7 D). In addition, the lean mass of GHRP-6-treated mice was similar to their lean mass prior to treatment, whereas vehicle-treated mice continued to lose lean mass due to continued food restriction (Fig. 7 E).

To test the involvement of Notch signaling, we next treated food-restricted mice with the Notch inhibitor PF-03084014 (PF), which blocks γ-secretase. While the 5-day treatment with PF did not alter body weight (Fig. 7 F), mice that received the Notch inhibitor had higher fed glucose levels compared with vehicle-treated controls (Fig. 7 G). Importantly, both groups consumed all the food provided, which continued to be restricted for the duration of the experiment. Mice treated with PF also had minimal change in lean body mass, whereas animals receiving the vehicle had a significant decrease in lean mass after 5 days, consistent with calorie restriction (Fig. 7 H).

GIP and GLP-1 receptor agonists are being used to treat obesity and diabetes (Tan et al., 2022) and result in decreased eating. Daily treatment of wild-type mice with the dual- GIP/GlP1 receptor agonist tirzepatide (TZP) for 5 days significantly decreased weight (Fig. 7 I). TZP-treated mice had decreased food intake at the start of treatment, followed by normalization by treatment day 4 (Fig. 7 J). While mice had similar fed glucose levels at the start of the experiment, those receiving TZP had >50 mg/dl lower values (control: 142.6 ± 5.9 mg/dl versus TZP: 89.0 ± 2.6 mg/dl) by treatment day 5 (Fig. 7 K).

At the end of the experiment, TZP-treated mice also had decreased 5-h-fasting glucose (control: 138.8 ± 6.9 mg/dl versus TZP: 59.8 ± 5.9 mg/dl) (Fig. 7 L). Surprisingly, similar to calorie restriction, mice injected with TZP had an ∼4.4-fold increase in plasma ghrelin compared with vehicle (Fig. 7 M). This corresponded with an increased abundance of ghrelin+ cells in the antrum (control: 0.89% ± 0.20% versus TZP: 2.26% ± 0.24%) and corpus (control: 1.44% ± 0.08% versus TZP: 4.01% ± 0.38%) (Fig. 7 N and Fig. S5). Chromogranin A+ cells also increased by ∼400% in the antrum and by ∼290% in the corpus with TZP treatment (Fig. 7 O and Fig. S5). Taken together, these data suggest that ghrelin plays a role in the effects of TZP.

We found that, unlike in the intestine, calorie restriction increased differentiated endocrine and Neurog3+ endocrine progenitor cells through increased Notch signaling in gastric Neurog3+ cells. We further show, using an active FOXO1 mutant, that FOXO1 localizes to the nucleus of Neurog3+ cells with calorie restriction, associated with an increased number of ghrelin cells.

Mechanistically, our data suggest that the increase in ghrelin+ cell number is dependent upon active FOXO1. FOXO1 regulates cellular differentiation and growth (Tsuchiya and Ogawa, 2017) and has been found in Neurog3+ and 5HT+ cells in the large and small intestines of mice and humans (Bouchi et al., 2014; Talchai et al., 2012). We have recently characterized a population of FOXO1+ cells in the stomach (McKimpson et al., 2022) that colocalize with markers of acid-secreting parietal as well as Neurog3+ cells. In these cells, FOXO1 regulates transcription of the cyclin E1 gene (CCNE1). Our data suggest that, under calorie restriction, FOXO1 marks an expanding subpopulation of Neurog3+ cells. Interestingly, FOXO1 is found in the nucleus of these cells and is associated with increased numbers of ghrelin cells. Evidence suggests that FOXO1 mediates changes occurring with calorie restriction in other tissues as well. For example, FOXO1 haploinsufficiency is linked to reduced incidence of tumors with aging, although there was no effect on lifespan (Yamaza et al., 2010). FOXO1 also has been shown to have a role in maintaining telomere size during calorie restriction in the heart (Makino et al., 2016).

Here, we demonstrate that calorie restriction alters the abundance of stomach epithelial Neurog3+ cells by increasing Notch signaling. In support of Notch being activated in Neurog3+ cells, we found increased Hes1 mRNA after serum deprivation. It remains to be determined what triggers Notch activation. Stem and progenitor cells in different tissues are sensitive to diet. For example, fasting promotes the function of Lgr5+ stem cells in the intestine by inducing fatty acid oxidation (Mihaylova et al., 2018). Prolonged fasting also promotes self-renewal of hematopoietic stem cells (Cheng et al., 2014). Conversely, mice fed a high-fat diet have increased numbers of intestinal stem cells through peroxisome proliferator-activated receptor delta (Beyaz et al., 2016). Interestingly, a low-lipid diet activates Notch signaling to influence cell differentiation in the Drosophila intestine (Obniski et al., 2018).

Notch regulates stomach Lgr5+ cell proliferation (Demitrack et al., 2015) through a DLL1-expressing niche cell located in the antral gland base to blunt their differentiation (Horita et al., 2022). Notch inhibition generally promotes differentiation. It is unclear why the opposite result is seen in the stomach after calorie restriction, but it may indicate a differential effect of Notch activation on proliferation and differentiation within specific stem and progenitor cell subpopulations, including Neurog3+ progenitor and Lgr5+ stem cells. In support of this, while Lgr5+ stem cells were decreased with calorie restriction, we saw increased Sox2+ cells, a distinct multipotent stem cell population in the stomach (Arnold et al., 2011). A different program may also be activated in gastric stem cells under calorie restriction compared with a normal diet. Notch-regulated Lgr5 stem cell maintenance is associated with increased mTOR signaling (Demitrack et al., 2015). While we did not assess mTOR signaling, calorie restriction modulates the function of intestinal stem cells through mTORC1 in Paneth cells, a component of the intestinal stem cell niche (Yilmaz et al., 2012). This expansion is dependent on mTORC1 and SIRT1 (Igarashi and Guarente, 2016). In mouse intestine, calorie restriction also preserves the function of a stem cells reserve pool by inhibiting mTORC1 (Yousefi et al., 2018).

Finally, our data demonstrate that ghrelin cell numbers increase upon calorie restriction. While we do not know whether ghrelin is secreted from these cells, this is likely in view of the increase in plasma ghrelin during fasting (Asakawa et al., 2001; Tschöp et al., 2000), confirmed in our experiments. A link between increased ghrelin and calorie restriction has been suggested (Reimer et al., 2010). Interestingly, ghrelin may also have a protective effect against aging (Cheng et al., 2010; Kaplan et al., 2017). It is not clear whether the increase in ghrelin+ cells is an adaptive change to decreased food consumption or a protective effect of calorie restriction itself. Our in vivo data suggested the contribution of the latter, as blocking this pathway in food-restricted mice with either a Notch inhibitor or ghrelin receptor antagonist elevates glucose levels without affecting food intake. This pathway may also be broadly activated, as treatment of mice with the weight loss agent tirzepatide (Tan et al., 2022) also triggers the expansion of stomach ghrelin-producing cells, although a contributing factor may be the minimal food intake in TZP-treated mice, particularly within the first 2 days of treatment.

Taken together, our data suggest that the expansion of Neurog3+ endocrine cells into lineage-committed ghrelin-producing cells during calorie restriction contributes to the beneficial protective effects of calorie restriction, including improved metabolism.

FoxV mice were generated as previously described (Kuo et al., 2019). Neurog3-GFP (Lee et al., 2002) and FoxKR (Banks et al., 2011) mice were previously described, and Lgr5-GFP (strain #008875; RRID:IMSR_JAX:008875), NICD (strain #008159; RRID:IMSR_JAX:008159), and Tomato (strain #007909; RRID:IMSR_JAX:007909) mice were purchased from Jackson Laboratories. All mice were maintained on C57BL/6J background. Unless it is otherwise indicated, tissue and primary stomach cells were from mice 10–12 wk of age. Calorie restriction was performed in mice 12–16 wk old at the start of food restriction. Thus, tissue for these experiments, collected upon the completion of food restriction, was collected from mice that were 16–20 wk age. The specific age and biological sex of each cohort are indicated in each figure legend. Male and female mice were used in equal numbers and as indicated in the figure legends. All studies were approved by the Columbia University Institutional Animal Care and Utilization Committee.

Mice were fed a standard chow diet (SCD) (13.1, 62.4, and 24.5% calories from fat, carbohydrate, and protein respectively; PicoLab Rodent Diet 20, 5053; Purina Mills). All mice were acclimated to being singly housed for a week with individual food intake determined over the last 3 days. Thereafter, mice were assigned to groups with an equal starting body mass and either had access to food and water ad libitum (AL) or were calorie-restricted (CR) by 30% with food placed into the cage twice daily for 4 wk. CR mice received 2.6–2.8 g/day of SCD per mouse. Plasma ghrelin was measured using a Rat/Mouse Ghrelin (Total) ELISA kit (Millipore Sigma) where samples were assessed in blood collected from AL and CR mice after 5 h fast (food removed at 9 am and blood collected at 2 pm). Blood glucose was measured under fed conditions or after 5 h fast using a OneTouch glucose monitoring system (LifeScan). Lean mass was measured by EchoMRI. For tissue collection, to mimic ad libitum conditions, CR mice were given their last feeding at 9 am and tissue collected ∼3 h later at noon.

Notch activity in mice was blocked using PF-03084014 (PF) (Nirogacestat from SelleckChem) (Kitamoto et al., 2022). Specifically, PF was resuspended in N,N-Dimethylacetamide:Kolliphor HS 15: water = 5:10:85 (vol/vol/vol) solution, pH 4∼5, to a working concentration of 15 mg/ml using sonication. Mice were injected IP daily (at ∼4 pm) with vehicle or PF at the final dose of 150 mg/kg for 5 days. Ghrelin activity was blocked using the ghrelin receptor antagonist [D-Lys-3]-GHRP-6 (GHRP-6) (Bachem) (Asakawa et al., 2003). Specifically, GHRP-6 was resuspended in saline and administered by intraperitoneal (IP) injection for 5 days at the concentration 200 nmol/mouse. Prior to PF and GHRP-6 treatments, mice were calorie-restricted by 30% for 4 wk. Mice were 14 wk old at the start of caloric restriction and food restriction was continued throughout the duration of the treatments. Male mice were used for PF treatments and female mice were used for GHRP-6 treatments. Tirzepatide (MedChemExpress) was resuspended in dimethyl sulfoxide at a stock concentration of 1 mM. The drug was further diluted 1:100 with 1xPBS to create a working solution. 13-wk-old male mice were injected IP with either vehicle (1xPBS) or TZP at a concentration of 30 nM/kg daily (at ∼4 pm) for 5 days. The following day, mice were fasted at 9 am for 5 h, blood glucose was measured and collected (to measure ghrelin levels), and the tissue was collected for analysis. Throughout the duration of the experiment, mice were measured for food intake, body weight, and fed glucose levels.

After dissection, the stomach tissue was fixed in 4% paraformaldehyde (PFA, cat no. 15710; Electron Microscopy Sciences) for 1.5 h on ice. Tissue was subsequently placed in 30% sucrose (in 1xPBS) overnight, embedded in OCT, flash frozen, and sectioned at 5 μm for staining. For primary cultures, cells were grown on gelatin-coated chamber slides (cat. no. 354114; Corning), fixed in 4% paraformaldehyde for 10 m at room temperature, and permeabilized in PBS with 0.1% Triton X-100 (Cat. no. 42235-5000; Acros Organics) for 10 min.

For immunohistochemistry, tissue was blocked in 10% goat (Cat. no. S-1000; Vector) or donkey (cat. no. 017000121; Jackson Immuno Research Labs) serum for 1 h at room temperature and then incubated with primary antibodies diluted in blocking solution overnight. The primary antibodies used were as follows: GFP (cat no. A-6455, 1:200; RRID:AB_221570; Molecular Probes), GFP (cat no. ab6662, 1:100; RRID:AB_305635; Abcam), Ghrelin (cat no. MAB8200, 1:100; RRID:AB_2637039; Novus Biologicals), Ki67 (cat no. ab16667, 1:50; RRID:AB_302459; Abcam), Neurogenin3 (cat. no. ab38548, 1:50; RRID:AB_776709; Abcam), Sox2 (cat no. ab97959, 1:200; RRID:AB_2341193; Abcam), Chromogranin A (cat no. ab45179, 1:200; RRID:AB_726879; Abcam), 5HT (cat. no. NB120-16007, 1:50; RRID:AB_792022; Novus Biologicals), Gastrin (cat no. ab232775, 1:200, RRID:AB_3097695; Abcam), Glucagon (cat no. M182, 1:500; RRID:AB_2619627; Takara), CCK (cat no. PA5-32348, 1:400; RRID:AB_2549819; Thermo Fisher Scientific), Synaptophysin (cat no. MAB5258, 1:200; RRID:AB_2313839; Millipore), Jag1 (cat no. AF599, 1:100; RRID:AB_2128257; R and D), Hes1 (cat no. sc25392, 1:100; RRID:AB_647996; Santa Cruz), and pHH3 (cat no. ab5176, 1:100; RRID:AB_304763; Abcam). Alexa Fluor 488, 568, and 647 (dil. 1:1,000; Molecular Probes) secondary antibodies of corresponding species were used to detect primary antibodies. All secondaries were diluted 1:1,000 in blocking serum and incubated for 1 h at room temperature. Coverslips were applied using VECTASHIELD Vibrance Antifade Mounting Medium with DAPI (cat no. H-1800-10; Vector Laboratories) to counterstain nuclei. Images were obtained at room temperature using a Zeiss LSM 710 confocal microscope (Zeiss) and a Plan-Apochromat objective (20 ×/0.8) using ZEN (black edition) software. Images were taken in black and white and pseudocolored as indicated in the figures. When needed to optimize visualization, minor adjustments to brightness and contrast were performed to the entire image equally using ZEN (black edition) software. No additional imaging processing was performed.

Primary stomach cells were isolated from the indicated mice (McKimpson et al., 2022). Specifically, after serosa was removed, the antral stomach was incubated with digestion buffer (5 mM EDTA [cat. no. 15575-038; Invitrogen] in 1xPBS) at 4°C for 1.5 h. Following second digestion at 37°C for 30 m in TrypLE (cat. no. 12604013; Invitrogen) supplemented with 2 U/μl DNAse-1 (cat. no. 10104159001; Sigma-Aldrich) and 10 μM Y-27632 (cat. no. Y0503-1 MG; Rock Inhibitor, Sigma-Aldrich), cells were washed in ice-cold basic media (Advanced DMEM/F12 [cat. no. 12634010; Invitrogen] supplemented with 10 mM HEPES [cat. no. 15630080; Invitrogen], 1 x Glutamax [cat. no. 35050061; Invitrogen], 1% pen/strep [cat. no. 15140-122; Gibco], and 1% bovine serum albumin, fraction V [cat. no. BP1600-100; Thermo Fisher Scientific]) and passed through a 40-μM cell strainer (cat. no. 22363547; Thermo Fisher Scientific).

For 2-D culture, primary stomach cells were plated on mouse embryonic fibroblasts (cat. no. A34180; Gibco) at density.5 × 10⁶ cells/six-well plate coated with gelatin (cat. no. ES-006-B; Millipore). Cells were cultured in basic culture media (described above) supplemented with 10 μM rock inhibitor, 1 x N2 supplement (cat. no. 17502048; Invitrogen), 1 x B-27 supplement (cat. no. 17504044; Invitrogen), 1 mM N-Acetyl-L-cysteine (cat. no. A9165-5 G; Sigma-Aldrich), and 1 x Primocin (cat. no. NC9141851; Invitrogen). Rock inhibitor was removed from media after overnight recovery. Media for serum-deprivation conditions was modified to contain 0.1% bovine serum albumin, fraction V. Cells were cultured overnight at serum-deprived conditions. Where indicated, cells were treated overnight with 10 μM DAPT (N-[N-(3, 5-difluorophenacetyl)-l-alanyl]-s-phenylglycinet-butyl ester), 4 μM TAT-cre (cat. no. SCR508; Millipore), or 10 μM FOXO1 inhibitor (AstroZeneca; Compound 10 as published in Langlet et al., [2017]).

After culture, primary stomach cells were collected using trypsin and resuspended in basic culture media. Prior to sorting, cells were stained with the impermeant DNA dye DAPI. Cells were sorted using a BD Influx with a blue (488-nm) laser and 530/40 bandpass filter to detect endogenous GFP fluorescence and a green (561-nm) laser and 610/20 bandpass filter to detect tomato fluorescence and analyzed using FlowJo software (BD Life Sciences). To determine single-cell populations, cells were gated based on forward and side scatter and trigger pulse width. In addition, only viable cells (negative for DAPI) were considered for downstream analysis of GFP fluorescence.

RNA was isolated from primary stomach cultures using an Arcturus PicoPure RNA Isolation Kit (cat no. KIT0204; Applied Biosystems) as recommended by the manufacturer with the addition of the optional 15-minute DNaseI treatment. cDNA was generated from RNA using qScript cDNA SuperMix (cat no. 101414-106; Quanta Biosciences) by diluting RNA with 5x master mix, without adding any additional water, and incubating samples at 25°C for 5 m, 42°C for 30 m, 85°C for 5 m, and holding at 4°C (McKimpson and Accili, 2019). qPCR reactions were completed using GoTaq qPCR Master Mix (Promega) with the following primer pairs: Hes1: forward: 5′-CAC​TGA​TTT​TGG​ATG​CAC​TTA​AGA​AG-3′, reverse: 5′-CCG​GGG​TAG​GTC​ATG​GCG​TTG​ATC​T-3′; Jag1: forward: 5′-CCT​CGG​GTC​AGT​TTG​AGC​TG-3′, reverse: 5′-CCT​TGA​GGC​ACA​CTT​TGA​AGT​A-3′; Foxo1: forward: 5′-TCC​AGT​TCC​TTC​ATT​CTG​CAC​T-3′, reverse: 5′-GCG​TGC​CCT​ACT​TCA​AGG​ATA​A-3′; E-cadherin: forward: 5′-AGC​TTT​TCC​GCG​CTC​CTG-3′, reverse: 5′-CTT​CCG​AAA​AGA​AGG​CTG​TCC-3′; Ghrelin: forward: 5′-GCC​CAG​CAG​AGA​AAG​GAA​TCC​A-3′, reverse: 5′-GCG​CCT​CTT​TGA​CCT​CTT​CC-3′; Glucagon: forward: 5′-ATG​AAG​ACC​ATT​TAC​TTT​GTG​GCT​G-3′, reverse: 5′-CGG​CCT​TTC​ACC​AGC​CAC​GC-3′; Chromogranin A: forward: 5′-CAG​CTC​GTC​CAC​TCT​TTC​CG-3′, reverse: 5′-CCT​CTC​GTC​TCC​TTG​GAG​GG-3′; GAPDH: forward: 5′-CGT​ATT​GGG​CGC​CTG​GTCAC-3′, reverse: 5′-ATG​ATG​ACC​CTT​TTG​GCT​CC-3′; Lgr5: forward: 5′-ACC​CGC​CAG​TCT​CCT​ACA​TC-3′, reverse: 5′-GCA​TCT​AGG​CGC​AGG​GAT​TG-3′; and Ngn3: forward: 5′-TGG​CCC​ATA​GAT​GAT​GTT​CG-3′, reverse: 5′-AGA​AGG​CAG​ATC​ACC​TTC​GTG-3′. Using a Bio-Rad CFX96 real-time PCR system for measurements, relative gene expression was calculated using the ΔΔCt method with E-cadherin (as an epithelial marker of stomach cells) as the reference gene for primary stomach cultures and with GAPDH as the reference gene for sorted cells.

For the quantification of tissue, at least at least five independent fields/samples for a minimum of three mice in each experimental group were quantified. No samples were excluded from the analysis. Unless otherwise indicated, each dot in graphs of quantification of microscopic images represents an independent field. For each field, the percentage of positive cells was calculated for each marker by dividing the total immunoreactive+ cell by the total epithelial DAPI+ nuclei in the same field. For each sample, quantification was performed separately for antrum and corpus stomach. Note that Lgr5−GFP+ cells were only assessed in antrum to avoid any confounding effects from Lgr5-expressing chief cells in the corpus (Leushacke et al., 2017).To assess the proliferation of GFP+ cells in FoxV reporter mice, each FOXO1−GFP+ cell was determined to be positive or negative for respective proliferation markers, and the prevalence of each was assessed as a ratio over total GFP+ cell number for each microscopic field. The number of specific mice for each experiment is indicated in figure legends. For primary cultures, stomach cells in six to eight microscopic fields for at least three independent cultures were quantified.

A two-tailed student’s t test was used to compare the two groups. Analysis of variance followed by a Tukey post-hoc test was used to compare multiple groups. Data distribution was assumed to be normal, but this was not formally tested. GraphPad Prism 6 software was used for calculations and P < 0.05 was considered significant. Graphs are depicted as average ± standard deviation with individual samples being indicated using individual points. The specific measurements of each sample (i.e., mouse measurements and quantification of microscopic field) are as indicated in each of the figure legends.

This article contains five supplemental figures. Fig. S1 shows body weight, lean mass, and fasting glucose measurements from female mice before and after 4 wk of calorie restriction. Fig. S2 provides representative low-magnification fluorescent images of gastric endocrine cells in tissue from female mice fed ad libitum or calorie-restricted. Fig. S3 shows representative low-magnification fluorescent images showing the abundance of ghrelin+ and chromogranin A+ cells in the stomach tissue of male ad libitum-fed and calorie-restricted mice. Fig. S4 presents control staining for FOXO1-GFP for interpretation of results from FOXO1-Venus reporter mice. Fig. S5 shows representative fluorescent images of ghrelin+ and chromogranin A+ cells in stomach tissue from male mice treated with tirzepatide (TZP).

Data are available in the article itself and its supplementary materials. The data are available from the corresponding author upon reasonable request.

We thank members of the Accili laboratory for insightful discussions and Tommy Kolar, Ana M. Flete-Castro, Xi Sun, and Lu Caisheng for technical support.

This research was supported by a Naomi Berrie Diabetes Center (Columbia University) Berrie Fellow in Diabetes Research Award and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grant nos. K01DK121873 to W.M. McKimpson., DK103818 to U. Pajvani, and DK57539 to D. Accili. These studies used the CCTI Flow Cytometry Core, supported by NIH Office of the Director grant no. S10OD020056, and the resources of the Diabetes Research Center Flow Core, supported by NIDDK grant no. P30DK063608. D. Accili was a founder, director, stockholder, and chair of the board of Forkhead Biotherapeutics Corp.

Author contributions: W.M. McKimpson: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing - review & editing, S. Spiegel: Formal analysis, Investigation, Visualization, M. Mukhanova: Investigation, Writing—review & editing, M. Kraakman: Investigation, W. Du: Investigation, Resources, T. Kitamoto: Investigation, Resources, Writing—review & editing, J. Yu: Resources, Z. Deng: Data curation, Methodology, Resources, U. Pajvani: Conceptualization, Funding acquisition, Supervision, Writing—review & editing, D. Accili: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing.