Intestinal GPR119 activation by microbiota-derived metabolites impacts feeding behavior and energy metabolism

Objective The gastrointestinal tract affects physiological activities and behavior by secreting hormones and generating signals through the activation of nutrient sensors. GPR119, a lipid sensor, is indirectly involved in the secretion of incretins, such as glucagon-like peptide-1 and glucose-dependent insulinotropic peptide, by enteroendocrine cells, while it directly stimulates insulin secretion by pancreatic beta cells. Since GPR119 has the potential to modulate metabolic homeostasis in obesity and diabetes, it has attracted interest as a therapeutic target. However, previous studies have shown that the deletion of Gpr119 in mice does not affect glucose homeostasis and appetite in either basal or high-fat diet-fed conditions. Therefore, the present study aimed to explore the role of GPR119 signaling system in energy metabolism and feeding behavior in mice. Methods Gpr119 knockout (KO) mice were generated using CRISPR-Cas9 gene-editing technology, and their feeding behavior and energy metabolism were evaluated and compared with those of wild type (WT) mice. Results Upon inducing metabolic stress via food deprivation, Gpr119 KO mice exhibited lower blood glucose levels and a higher body weight reduction compared to WT mice. Although food intake in WT and KO mice were similar under free-feeding conditions, Gpr119 KO mice exhibited increased food intake when they were refed after 24 h of food deprivation. Further, food-deprived Gpr119 KO mice presented shorter post-meal intervals and lower satiety for second and later meals during refeeding, resulting in increased food intake. Associated with this meal pattern, levels of oleoylethanolamide (OEA), an endogenous agonist of GPR119, in the luminal contents of the distal gastrointestinal tract were elevated within 2 h after refeeding. The large-intestinal infusion of OEA prolonged post-meal intervals and increased satiety in the first meal, but not the second meal. On the other hand, infusion of oleic acid increased cecal OEA levels at 2 h from the beginning of infusion, while prolonging post-meal intervals and increasing satiety on the meals that occurred approximately 2 h after the infusion. Cecal OEA levels were low in antibiotic-treated mice, suggesting that the gut microbiota partially synthesizes OEA from oleic acid. Conclusions Collectively, our results indicate that the activation of gastrointestinal GPR119 by microbiota-produced OEA derived from oleic acid is associated with satiety control and energy homeostasis under energy shortage conditions.


INTRODUCTION
The gastrointestinal tract is involved in physiological regulation, including regulation of metabolism and feeding behavior, through the secretion of gut hormones and generation of signals via receptors in response to nutrients. Several G protein-coupled receptors (GPCRs) have been identified as sensors of lipids, such as fatty acids, monoacylglycerols (MAGs), and their metabolites, the levels of which are increased in the intestine after meals [1,2]. GPR40 and 120 are wellknown receptors for dietary long-chain fatty acids and their 1 metabolites produced by gut microbiota [1,3,4]. In addition, GPR119 is a receptor for MAGs [i.e. 2-oleoylglycerol (2-OG)], lysophosphatidylcholine (LPC), and fatty acid ethanolamides (FAEs) [i.e. oleoylethanolamide (OEA)] [5]. Although enterocytes, enteroendocrine cells, and neural fibers have been postulated to sense lipids via GPCRs in the gut, most studies imply that enteroendocrine cells are the primary cells that sense lipids, which results in the production of hormones like cholecystokinin (CCK) and glucagon-like peptide-1 (GLP-1) after a meal [6,7]. As indicated by previous in vitro and in vivo studies, GPR119 is highly expressed in the pancreas (pancreatic beta cells) and the gut, suggesting that GPR119 directly controls insulin secretion by the pancreas and indirectly controls incretin secretion by the intestine [5,8]. However, Gpr119 knockout (KO) mice show no changes in islet morphology and size, glucose homeostasis, insulin levels, and appetite under either lab-chow-or high-fat diet-fed conditions [8,9]. In addition, endogenous ligands, such as LPC and OEA, have failed to enhance glucosestimulated insulin secretion (GSIS) by islets isolated from either wild type (WT) or Gpr119 KO mice [8]. Interestingly, EX-4, a GLP-1 receptor agonist, enhanced GSIS in both WT and Gpr119 KO mice, while Gpr119 KO mice exhibited reduced postprandial plasma GLP-1 levels [8]. Therefore, the physiological function of intestinal GPR119 as a pharmacological target for regulating metabolism and feeding behavior should be further elucidated. Notably, a recent study suggested that the activation of distal-intestinal GPR119 slowed gastric emptying and reduced food intake in Cyp8b1 KO mice that exhibited impaired lipid absorption in the upper gut and absorbed dietary fat from the distal intestine [10]. OEA, an endogenous ligand of GPR119, is a well-known lipid mediator that modulates lipid metabolism and feeding behavior [11]. Detailed investigations have shown that OEA is directly derived from the diet or synthesized from dietary oleic acid (OA) in gastrointestinal enterocytes, especially in the proximal intestine [12,13]. OEA produced in the proximal intestine plays a vital role in inducing meal satiety by activating sensory fibers of the vagus nerve via peroxisome proliferatoractivated receptor a (PPARa) activation [13,14]. In addition, a study suggests that OEA can modulate the secretion of the gastrointestinal peptide hormone GLP-1, which induces satiation, by enteroendocrine cells [15]. Therefore, GPR119 may be involved in modulating feeding behavior through hormone secretion when ligands are formed in the distal intestine; however, OEA levels in the distal-intestinal tissue are unaffected after meals [12]. Studies indicate that gut microbiota produce various types of small molecules, including N-oleoylserinol, which was recently identified as a GPR119 ligand (EC 50 , 1.6 mM), and FAE [16e18]. FAEs naturally exist as signaling molecules in many organisms, from simple life forms to humans. Therefore, intestinal GPR119 may be activated by biomolecules produced by the gut microbiota, such as N-oleoylserinol and FAEs. However, the effects of these molecules on the host via GPR119 activation have not been elucidated yet. The present study investigated the role of intestinal GPR119 in feeding behavior and energy homeostasis using a new Gpr119 KO mouse line.
Our results indicate that GPR119 is essential for modulating energy homeostasis, particularly during energy deficit, and that the distalintestinal GPR119 induces satiety in modulating meal patterns by sensing microbiota-produced metabolites such as OEA.

MATERIAL AND METHODS
2.1. Animals C57BL/6 J mice were obtained from Japan SLC Inc. (Tokyo, Japan). Gpr119 KO mice were generated using the CRISPR-Cas9 genome editing system by Unitech Co., Ltd. (Kashiwa, Japan) (see details below). Mice were maintained with CLEA Rodent Diet CE-2 (CLEA Japan, Inc., Tokyo, Japan) and housed under the following conditions: room temperature (20e25 C), 40e60% humidity, and a 12/12 h lightedark cycle (light period: 8:00e20:00). All animal experiments were performed in accordance with the guidelines of the Committee on the Ethics of Animal Experiments of Tokyo University of Agriculture and Technology. The Animal Research Ethics Subcommittee approved the experiments conducted by the Tokyo University of Agriculture and Technology (permit: 28e87).

Gpr119 knockout mouse model
Gpr119 KO mice were generated using the CRISPR-Cas9 genome editing system by consignment production (Unitech Co., Ltd.). gRNA sequences (target sequence 1, TCAGAGTCACAGCACCGTTC; target sequence 2, GTACAGGTATACTCGCTTCA) were designed with the following PAM sequence outside the protein coding sequence (CDS) of GPR119 to delete the whole CDS and gRNAs were microinjected into fertilized embryos of C57BL/6 J mice with Cas9 mRNA (Supplemental Fig. 1A). Four male mice (KO1e4) were obtained from Unitech Inc. after screening for the deletion of target sequences using a genotyping primer set (WT allele, 2199 bp; target allele, approximately 800 bp) ( Table 1 and Supplemental Fig. 1C). One of the four mice (KO4) exhibited breeding issues. Hence, sequences of target alleles of rest of the mice (KO1e3) were analyzed by Sanger sequencing (Eurofins Genomics, K.K., Japan; Supplemental Fig. 1B). Finally, based on the sequence of the target allele and breeding ability of mice, KO1 was used as a Gpr119 KO mice line in this study. Herein, F3eF6 litters were used. The confirmed sequence of the Gpr119 KO mice (KO1) shows a 1372-bp deletion around the target and full deletion of the CDS of GPR119. Moreover, Gpr119 was not detected in various tissues of Gpr119 KO mice (Supplemental Fig. 1D; data for cecum and liver tissues are shown).

Feeding restriction
For the experiments related to feeding restriction, mice were housed individually in wired-bottom cages to avoid coprophagia and were acclimated to the housing conditions for at least 3 days. Five different feeding conditions were set up for the respective groups: free-feeding (FF), food-deprived (FD), refeeding (RF)-1 h, RF-2 h, and RF-5 h. Mice in the FD group were deprived of food for 24 h (from 9:00 to 9:00). Mice in the RF-1 h, RF-2 h and RF-5 h groups were provided access to food for 1, 2, and 5 h, respectively, after 24 h food deprivation. , and an incision (2.0 cm) was made through the skin and abdominal muscle along the midline of the lower abdomen. The lower ileum and cecum were gently exposed, and a small puncture in the ventral wall at approximately 1.0 cm proximal to the end of the ileum was made using a needle. One side of the tube (approximately 0.5 cm) was inserted through the puncture, facing the cecum. A tissue adhesive (Aron Alpha A Sankyo; Sankyo, Tokyo, Japan) was applied to the mesh for adherence to the gut wall. The outer portion of the tube with the attached square piece of mesh was subcutaneously routed to the back of the neck, exteriorized, and fixed to the underlying tissue by applying the tissue adhesive. The abdominal muscle wall was sutured using absorbent sutures [ELmelt; Natsume Seisakusho Co., Ltd., (Nastume), Tokyo, Japan], and the skin was closed using surgical sutures (silk; Natsume) or stainless-steel wound clips (AUTOCLIPÒ; Becton, Dickinson and Company, Franklin Lakes, NJ, USA). The outer end of the catheter was plugged with a stainless-steel pin to avoid leakage of intestinal fluid. After the operation, the mice were intraperitoneally injected with AntisedanÒ (3.0 mg/kg, atipamezole hydrochloride, Nippon Zenyaku Kogyo Co., Ltd.) for recovery against the action of medetomidine hydrochloride. The catheter was flushed with saline (0.5 mL) once every 2e3 days to avoid blockage. Following surgery, the mice were individually housed with free access to food and water and were allowed 7e10 days of surgical recovery before the initiation of experiments.

Lipid infusion from the large-intestinal catheter
The mice implanted with large-intestinal catheters were deprived of food for 24 h (from 9:00 to 9:00) before lipid infusion: OA (Sigma Aldrich, Burlington, MA, USA) or OEA (Cayman Chemical, Ann Arbor, MI, USA). OA was dissolved in phosphate-buffered saline (PBS) containing 0.3% xanthan gum (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan) to obtain a final concentration of 10 mg/mL. OEA was dissolved in PBS containing 0.3% xanthan gum and 10% dimethyl sulfoxide (DMSO) to obtain a final concentration of 4 mg/mL. The infusion volume was 0.1 mL/mice, and infusion was manually conducted for 30 s. Mice were placed in a feeding measurement system right after lipid infusion. For lipid measurement in the cecum, mice were sacrificed to collect cecum samples 2 h after OA administration or 30 min after OEA administration.

Feeding behavior
Feeding activities of mice were recorded using a feeding measurement system MFI-01 equipped with an FIC-001 sensor (Muromachi Kikai Co., Ltd), which can monitor both access to the food container and weight of food in the food container. Raw data were collected every 5 min using the system with the CompACT AMS software. In addition, each test session was videotaped to verify the exact time of accessing the food container. Mice were acclimatized to the system and food containers for at least 3 days before the experiment. Meal pattern analysis was performed as described in a previous study by Gaetani et al. [19] with some modifications: meal interval (min), interval time between meals; minimal meal interval, 5 min; meal, feeding periods included bouts; meal latency (min), the time interval from the onset to the first eating episode; meal length (>1 min), time from start to finish of the meal; frequency (meals/session), meal number in the session period; total food intake (g/kg body weight/session), the amount of food consumed during the session; meal size (g/kg body weight), the amount of food consumed during the meal; satiety ratio [min/(g/kg)], the meal interval to the next meal per meal as an indicator of satisfaction.
2.9. Food intake measurement Food intake was measured by manual weighing of food before and after a feeding period in several experiments that did not require the analysis of meal patterns. To test the effects of OEA on food intake under FF conditions, we intraperitoneally injected mice aged 11e16 weeks with OEA (20 mg/kg/mL) or a control vehicle (DMSO; Sigma Aldrich) at 19:30. The mice were then exposed to food at the onset of the dark period (20:00), and the remaining food was measured at the beginning of the light period (8:00). To test the effects of OEA on FD conditions, we intraperitoneally injected mice deprived of food for 24 h (10:00e10:00) with OEA (20 mg/kg/mL) or control vehicle (20% PBS in DMSO) 30 min before the start of the measurement (9:30). The mice were exposed to food for 5 h (10:00e15:00) to measure food intake.

Sample harvesting
At the end of the experiment, the mice were anesthetized using isoflurane (Pfizer Inc., New York, NY, USA), and tissue samples were collected. When needed, we measured blood glucose levels in the tail vein using a portable glucometer (OneTouchÒ UltraÒ; LifeScan Inc., Milpitas, CA, USA) with an LFS quick sensor (LifeScan Inc.) before anesthesia was administered. Blood was collected from the portal vein and inferior vena cava using heparin-treated tubes (sodium salt; Yoshindo Inc., Toyama, Japan). The plasma was separated by centrifugation at 8,000Âg for 5 min at 4 C. The plasma samples were prepared for GLP-1 analysis by mixing a portion of blood obtained from the portal vein with dipeptidyl peptidase IV inhibitor (Merck Millipore, Burlington, MA, USA) at 1% (v/v). The gut (stomach, duodenum, jejunum, ileum, cecum, and colon) was opened, gut contents were collected if needed, and the tissue was rinsed with ice-cold PBS. Gut tissue samples and plasma were immediately frozen in chilled 2methyl butane (Wako) on dry ice, and all samples were stored at À80 C until further use. Fresh fecal droppings were collected during fasting for 5 h to analyze gut microbiota composition.

Lipid extraction
Lipid extraction was performed according to the method described in a previous study with minor modifications [20]. The electrospray ionization interface remained in the positive ionization mode, and the ion spray voltage was set to À3000 V. For those analyses, nitrogen gas was used as the drying gas at a flow rate of 15 L/ min and as the nebulizer gas at a flow rate of 3 L/min. The desolvation and heat block temperatures were set to 250 C and 400 C, respectively. Argon at a pressure of 230 kPa was used as the collisioninduced dissociation gas. The temperature in the autosampler was maintained at 15 C. Original Article (bacteria only) and excluded based on the average abundance of each bacterium (<0.0001 for phylum and <0.00001 for family), which were then re-normalized. Raw data were deposited into the DNA Data Bank of Japan database (Accession No. DRA014241). In addition, total bacterial number in the cecum of mice was determined using qPCR (see details above) [21]. Primer sequences are shown in Table 1, and DNA obtained from Bifidobacterium breve (JCM1192T; Japan Collection of Microorganisms, RIKEN BioResource Research Center, Tsukuba, Japan) was used as standard for quantification.

Histological analysis
Fresh cecum was washed with PBS, embedded in Tissue-TekÒ OCT compound (Sakura Finetek Japan Co. Ltd., Tokyo, Japan) and frozen at À80 C until freeze sectioning. We prepared 10 mm-sections using a cryostat (Leica CM1860, Leica Biosystems, Wetzlar, Germany) at À20 C, and the sections were fixed on glass slides with 4% paraformaldehyde for 10 min for hematoxylin-eosin (HE) staining. Briefly, the sections were immersed in hematoxylin solution and placed in 70% ethanol for color fixation. Next, the sections were immersed in eosin solution for 4 min. After washing with running water, we immersed the sections in 70% ethanol, twice in 100% ethanol for dehydration, and twice in xylene to make the sections transparent.
After air-drying, the sections were sealed with Mount-Quick (Daido Sangyo, Toda, Japan) and cover glass. After 1 h, the sections were observed under a microscope (BZ-X710, Keyence Co., Osaka, Japan). At least two sections of cecum tissue obtained from each mouse were selected for determining villi length (mm) and number of villi (number/ mm of serosa length in transverse section) using ImageJ software (NIH, Bethesda, MD, USA).

Other analysis
Plasma GLP-1 concentrations were measured using the GLP-1 (Active) ELISA kit (Merck Millipore) according to the manufacturer's protocol. Plasma FFA concentrations were measured using the LabAssayÔ NEFA laboratory assay kit (Wako). Total plasma cholesterol concentrations were measured using the LabAssayÔ Cholesterol kit (Wako). Plasma triglyceride concentrations were measured using the Lab-AssayÔ triglycerides kit (Wako).

Statistical analysis
The measured values are expressed as mean AE standard error of mean (SEM). Statistically significant differences between groups were analyzed using the following tests: two-group comparison, Student's ttest; three-group comparison, one-way analysis of variance (ANOVA) and Tukey's multiple comparisons test; and two-group correspondence test, two-way ANOVA and Bonferroni's multiple comparisons test. * Represents p < 0.05, ** represents p < 0.01, and *** represents p < 0.001.

3.1.
Gpr119 KO mice exhibit reduced body weight and increased energy expenditure Gpr119 KO mice were generated using the CRISPR-Cas9 genomeediting system to evaluate the physiological roles of GPR119. As shown in Supplementary Fig. 1, a 1372 bp-deletion around the target gene was confirmed by Sanger sequencing analysis. Gpr119 KO mice maintained under FF conditions exhibited lower body weight ( Figure 1A, p ¼ 0.012) than that exhibited by littermate WT mice. However, no differences in growth, food intake ( Figure 1C), and blood glucose levels ( Figure 1E) were observed between WT and Gpr119 KO mice maintained under FF conditions. Furthermore, no differences in plasma triacylglycerol (TG; Figure 1F), non-esterified fatty acid (NEFA; Figure 1G), and total cholesterol ( Figure 1H) levels were observed between WT and Gpr119 KO mice maintained under FF conditions. Therefore, energy metabolism and locomotor activity were evaluated in these mice. Energy expenditure ( Figure 2C, p ¼ 0.0146 for the dark cycle; Figure 2D, p ¼ 0.003 for 22:00, p ¼ 0.0423 for 1:00), but not respiratory quotient (Figure 2A,B) and locomotor activity ( Figure 2E,F), was higher in the Gpr119 KO mice than that in the littermate WT. Thus, Gpr119 KO mice expended more energy than that expended by WT mice even when consuming similar amounts of food. Further, we assessed the MMPC Energy Expenditure analysis using raw data of bodyweight and energy expenditure to examine if increased energy expenditure mainly impacts body weight. The ANCOVA genotype effect considering bodyweight as covariate was not significantly different ( Figure 2G, p ¼ 0.126). Adjusted energy expenditure (kcal/day/23.1 g mouse) was significantly increased in Gpr119 KO mice compared with WT mice ( Figure 2H, p ¼ 0.0276), which explains the relatively lean phenotype of Gpr119 KO mice compared with WT mice.

Gpr119 KO mice display hyperphagic responses during RF after food deprivation
To investigate the role of GPR119 in metabolic stress, we subjected the seven-week-old Gpr119 KO mice and WT littermates to fooddeprivation for 24 h. Gpr119 KO mice showed lower body weight than WT mice under FD conditions ( Figure 1A, p ¼ 0.0024). The reduction in body weight in Gpr119 KO mice was significantly greater than that in littermate WT mice ( Figure 1B, p ¼ 0.0078). Interestingly, Gpr119 KO mice consumed more food than that consumed by WT mice 5 h after 24 h of food deprivation ( Figure 1D, p ¼ 0.049). In addition, blood glucose levels tended to be lower in Gpr119 KO mice than in WT mice ( Figure 1E, p ¼ 0.0501). No differences in plasma TG ( Figure 1F), NEFA ( Figure 1G), and total cholesterol ( Figure 1H) levels were observed between Gpr119 KO and WT mice deprived of food for 24 h.
These results indicate that Gpr119 KO mice exhibited hyperphagic responses under the stress of food deprivation.

3.3.
Gpr119 KO mice exhibit shorter post-meal intervals during RF after food deprivation Following the results in Figure 1 (Figure 3). Consistent with the results of the manual measurement of food intake shown in Figure 1D (RF-5 h), food intake measured using the system was significantly increased in Gpr119 KO mice compared with that in WT mice ( Figure 3A, p ¼ 0.0201). Furthermore, meal frequency ( Figure 3C) was significantly higher in Gpr119 KO mice than in WT mice, while no differences in latency ( Figure 3B) or total meal duration ( Figure 3D) in the session were observed between the Gpr119 KO and WT mice. No differences in the first meal parameters, such as meal size ( Figure 3E), meal duration ( Figure 3F), post-meal interval ( Figure 3G), and satiety ratio ( Figure 3H) were observed between the Gpr119 KO and WT mice. Interestingly, the average post-meal interval ( Figure 3K) and average satiety ratio ( Figure 3L), but not average meal size ( Figure 3I) and average duration ( Figure 3J), were significantly decreased in Gpr119 KO mice compared with those in WT mice. Therefore, we analyzed second meal parameters, results of which indicated that post-meal interval (Supplemental Fig. 2C, p ¼ 0.065) and satiety ratio (Supplemental Fig. 2D, p ¼ 0.0077) of the second meal were lower in Gpr119 KO mice than in  WT mice. No differences in the second meal size (Supplemental Fig. 2A) and meal duration (Supplemental Fig. 2B) were observed between the Gpr119 KO and WT mice. These results suggested that Gpr119 deletion induced a reduction in meal satiety during RF after food deprivation, and its effect was specifically delayed from the onset of feeding.
Further, we measured plasma GLP-1 levels in WT and Gpr119 KO mice under FD and RF conditions ( Figure 1I). GLP-1 levels showed an increasing tendency under RF in WT and Gpr119 KO mice; however, there was no statistical difference between feeding conditions. In addition, deletion of Gpr119 did not impact plasma GLP-1 levels. In addition, we verified the effects of food deprivation on Gcg and Gpr119 Figure 3: Gpr119 KO mice exhibited increased food intake and reduced satiety for subsequent meals under re-feeding conditions. After 24 h of food deprivation, WT or Gpr119 KO mice were individually placed in a cage with food in the feeding behavior analysis system, and their behavior was recorded for up to 5 h. The following feeding parameters were analyzed in the first 3 h: (A) food intake, (B) latency, (C) meal frequency, (D) total meal duration, (E) first meal size, (F) first meal duration, (G) first post-meal interval, (H) first satiety ratio, (I) average meal size, (J) average meal duration, (K) average post-meal interval, and (L) average satiety ratio. Data are expressed as means AE SEM (n ¼ 6 for WT and n ¼ 7 for KO; n ¼ 5 for latency of KO). Statistical significance is indicated by asterisks (*p < 0.05). WT, wild type; KO, knockout.
Original Article 8 expressions in the cecum. Interestingly, Gpr119 expression was increased by food deprivation in the cecum of WT mice ( Figure 1K, p ¼ 0.0128), while Gcg expression was decreased ( Figure 1J, p ¼ 0.0119). In Gpr119 KO mice, Gcg expression was not affected by the feeding conditions; however, the expression levels were relatively low compared to WT-FF ( Figure 1J, p ¼ 0.0062, WT-FF v.s. KO-FF; p ¼ 0.0054, WT-FF vs KO-FD).

Gpr119 is not involved in satiety induced by systemic OEA
To investigate if GPR119 is involved in satiety induction by systemic treatment with OEA, food intake was measured in Gpr119 KO mice that were intraperitoneally administered with OEA under FF or FD conditions. Under FF conditions, food intake in the dark cycle decreased as a result of OEA treatment in both WT (p ¼ 0.016; Figure 4A) and Gpr119 KO mice (p ¼ 0.0203; Figure 4A) compared with that in mice that were treated with the vehicle. Consistent with Figure 1D, higher food intake was observed during 5 h of RF after food deprivation in Gpr119 KO mice compared with that in WT mice under vehicle treatment; however, OEA treatment reduced food intake in both WT (p ¼ 0.036; Figure 4B) and Gpr119 KO mice (p < 0.001; Figure 4B). Overall, these results indicate that the induction of satiety by systemic OEA does not require GPR119 activation under either FF or FD conditions.

Feeding mobilizes OEA in intestinal contents
Here, we confirmed that Gpr119 was highly expressed in the lower part of the gastrointestinal tract, including the ileum, cecum, and colon ( Figure 5A). This expression pattern was associated with the expression of Gcg (Supplemental Fig. 3B) and Pyy (Supplemental Fig. 3C), but not Cck (Supplemental Fig. 3A) [23]. Analysis of the intestinal tissue of WT mice housed under FF conditions revealed that OEA levels were approximately 60e300 pmol/g tissue (Supplemental Fig. 3D). Therefore, we quantified OEA in the cecal contents of the mice where Gpr119 was highly expressed; OEA levels in the cecal contents were 2.4 AE 0.4 nmol/g and 2.3 AE 0.0 nmol/g in WT and Gpr119 KO mice, respectively, housed under FF conditions, and the levels decreased after 24-h food deprivation in WT (p ¼ 0.0061; Figure 5B) and Gpr119 KO mice (p < 0.001; Figure 5B). 2-OG levels in the cecum contents were similar between WT and Gpr119 KO mice under FF conditions, and food deprivation did not affect cecal 2-OG levels in either WT and Gpr119 KO mice ( Figure 5C). Plasma OEA levels were similar between WT and Gpr119 KO mice under FF conditions, and the levels were approximately 4 pmol/mL plasma (Supplemental Fig. 3E). In addition, Gpr119 KO mice did not exhibit abnormal changes in the cecum tissues compared with WT mice, as determined by hematoxylin and eosin staining, and no changes in villi number and length (Supplemental Fig. 3F). Furthermore, we analyzed the microbiota composition of the fecal contents of WT and Gpr119 KO mice. There was no significant difference at the phylum and family levels (Supplemental Fig. 3G) between WT and Gpr119 KO mice. Then, we tested ligand activity for GPR119 of OEA and 2-OG using HEK293 cells expressing DOX-inducible human GPR119. A synthetic agonist of GPR119, Ar231453 (0.3 mM and 3 mM), increased cellular cAMP levels by 7 times compared with vehicle control (Figure 5D), and Forskolin treatment increased cAMP levels by 27 times (data not shown). In the cells, OEA (30 mM) increased cAMP levels, whereas such increase was not observed with 2-OG treatment (30 mM). In DOXuninduced cells, the activation by Ar231453 and OEA was not observed (data not shown). Therefore, we focused on OEA in the subsequent experiments.
The diet used in this study contained OEA at a concentration of 1.3 AE 0.2 nmol/g, which is much higher than the OEA levels in the gut tissue and plasma of the mice. Therefore, we expected that RF would alter the luminal levels of OEA because the diet contained OEA. Our analysis revealed that luminal OEA levels (nmol/g luminal contents) were higher in the cecum and colon than those in the duodenum, jejunum, and ileum in mice refed for 1 or 2 h ( Figure 5E). However, cecal OEA levels in the mice refed for 2 h after food deprivation were not as high as those in the free-fed mice ( Figure 5B). In addition, luminal OEA levels were slightly higher in the stomach than in the proximal intestine, suggesting dilution and absorption of OEA in the proximal intestine. These results suggest that luminal OEA levels in the large intestine were elevated as a result of feeding, which might cause the accumulation of dietary OEA or the production of OEA by gut microbiota [16,17]. 3.6. Large-intestinal OEA infusion modifies meal patterns Next, we examined whether OEA infusion into the large intestine modified the meal pattern exhibited by mice because OEA was increased in the large intestine, where GPR119 was highly expressed ( Figure 5). OEA was infused through a catheter inserted at the end of ileum after 24 h of food deprivation, and the meal behavior was monitored. In the OEA group, the first post-meal interval was significantly longer than that in the vehicle group (p ¼ 0.0062; Figure 6D). In contrast, there were no differences in latency ( Figure 6A), first meal size ( Figure 6B), or first meal duration ( Figure 6C) between the groups, resulting in an increased satiety ratio for the first meal in the OEAinfused mice (p ¼ 0.0317; Figure 6E). Furthermore, we analyzed the meal pattern in 2 h of the mice, which showed that the total food intake (p ¼ 0.0436; Figure 6F) and meal frequency (p ¼ 0.100; Figure 6G) were lower in the OEA-infused mice. In addition, there were no differences in the average meal size, meal duration, post-meal interval, and satiety ratio between the groups (Supplemental Fig. 4), indicating that large-intestinal OEA infusion immediately affected the meal pattern; however, it was not long-lasting. These results collectively suggest that satiety can be induced by the sensing of OEA at the large intestine.
3.7. OEA generated from OA by gut microbiota modulates meal behavior Our previous study showed that microorganisms could convert unabsorbed fatty acids reaching in the distal intestine to unique fatty acid metabolites, such as conjugated fatty acids and hydroxy fatty acids, which have physiological functions [3]. We hypothesized that luminal OEA levels in the large intestine were increased ( Figure 5E) because of the synthesis of OA by microorganisms, although it was possible that dietary OEA was accumulated in the luminal contents of the large intestine. To verify this hypothesis, we first quantified OEA levels in the cecal contents of mice administered antibiotics for one week. Cecum weight dramatically increased in antibiotic-treated mice compared with those in mice which were not administered antibiotic (p < 0.001; Supplementary Fig. 5A); however, there were no differences in the water level in the cecal content between the groups ( Supplementary Fig. 5B). Further, the total number of bacteria was reduced by 99.9% in the cecum of antibiotic-treated mice (p < 0.001; Supplementary Fig. 5C). Thus, we analyzed the amounts of OEA and OA (as 18:1 FFA) in the cecum of mice administered antibiotics. Cecal OEA levels were significantly decreased after antibiotic treatment (p ¼ 0.0046; Figure 7B), although there was no significant difference in the 18:1 levels between the groups ( Figure 7A). These results suggest that the gut microbiota synthesizes OEA from OA that reaches the large intestine. However, antibiotic treatment did not affect food intake during either 24 h under FF conditions ( Figure 7C), or RF for 3 h after 24 h of food deprivation ( Figure 7D). Antibiotic treatment can lead to many changes in the host, gut microbiota, and their metabolites, including OEA production; therefore, this model might not exhibit OEA-specific effects on food intake in mice. Although food intake did not change, antibiotic treatment affected meal pattern in mice: increased meal frequency (p ¼ 0.005; Supplementary Fig. 5E), and reduced average meal size (p ¼ 0.0414; Supplementary Fig. 5G), and average meal duration (p ¼ 0.0096; Supplementary Fig. 5H), but not latency (Supplemental Fig. 5D), total meal duration ( Supplementary Fig. 5F), average meal interval ( Supplementary Fig. 5I) and average satiety ratio ( Supplementary Fig. 5J), suggesting that antibiotic treatment disrupted the mechanisms underlying satiation. Next, we examined whether the gut microbiota in the cecum biosynthesizes OEA using OA as a substrate. OA was directly administered into the large intestine of mice through an implanted catheter at the end of ileum after 24 h of food deprivation, and OEA levels were determined in the cecal contents collected 2 h after OA infusion. As a result, cecal OEA levels increased after OA infusion (p ¼ 0.0204; Figure 7E), indicating that the microbiota in the large intestine biosynthesizes OEA from OA. Therefore, we next investigated whether OEA biosynthesis from OA induces a similar change in meal patterns as that observed after OEA infusion. The OA infusion did not affect latency ( Figure 7F), first meal size ( Figure 7H), first meal duration ( Figure 7I), first post-meal interval ( Figure 7J), or first satiety ratio ( Figure 7K). We then analyzed the meal pattern for the second meal, which occurred approximately 90e120 min after OA infusion. OA infusion significantly increased the second post-meal interval (p ¼ 0.010; Figure 7N). In contrast, no change was observed in the second meal size ( Figure 7L) and second meal duration ( Figure 7M), resulting increase of second satiety ratio as a trend (p ¼ 0.0505; Figure 7O) and reduction of meal frequency (p ¼ 0.0077; Figure 7G). Collectively, these results indicate that microbiota generated OEA using OA as substrate and the OEA induces meal satiety. The mice were placed into the feeding measurement system, and test sessions were recorded. Meal parameters were calculated using raw data, including (A) latency of feeding onset at the beginning of the trial, (B) first meal size, (C) first meal duration, (D) first post-meal interval, (E) first satiety ratio, (F) food intake, and (G) meal frequency (n ¼ 5). Data are expressed as means AE SEM. Statistical significance is indicated by asterisks (*p < 0.05, and **p < 0.01). GPR119 has already been pharmacologically targeted for obesity and diabetes because it stimulates insulin secretion by directly acting on pancreatic beta cells or through incretin secretion by gut enteroendocrine cells. Therefore, strategies aimed at enhancing the activation of GPR119, such as modulation of microbiota composition to increase OEA production or inhibition of OEA degradation, may help to induce meal satiety in obesity and other eating disorders.

DATA AVAILABILITY
Data will be made available on request.

ACKNOWLEDGMENTS:
This work was supported by JSPS KAKENHI, Japan Grant Numbers 20K05920 (MI) and AMED, Japan Grant Numbers JP21gm1010007 (IK).