Male C57BL/6J mice were purchased from Japan SLC (Shizuoka, Japan), housed in a conventional animal room at 24°C, and maintained under a 12 h light/dark cycle. Gpr84^−/− mice with a C57BL/6J background were generated using the CRISPR/Cas9 system. Mice were acclimated to the CLEA Rodent Diet (CE-2, CLEA Japan, Inc., Tokyo, Japan) for 1 week prior to treatment. Seven-week-old mice were placed on an MCT diet or modified D12492 diet (60% kcal fat, Research Diets Inc., New Brunswick, NJ, USA) for 5 weeks. These diets were formulated as either lard or MCT (C8:0 triglyceride, C10:0 triglyceride, or C12:0 triglyceride, Nisshin OilliO Group, Ltd., Tokyo, Japan), or based on the D12492 diet (Research Diets Inc.) supplemented with or without 5% decanoate (C10:0, FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan). The composition of the diets is provided in Supplementary Tables 1, 2. Body weight was measured once per week for the duration of the experiment. All mice were sacrificed under deep isoflurane induced anesthesia. All experimental procedures involving mice were performed according to protocols approved by the Committee on the Ethics of Animal Experiments of the Kyoto University Animal Experimentation Committee (Lif-K21020) and the Tokyo University of Agriculture and Technology (permit number: 28–87). All efforts were made to minimize suffering.

Blood glucose levels were assessed using a portable glucometer (OneTouch® Ultra®, LifeScan, Milpitas, CA, USA). The concentrations of plasma triglycerides (TG) (LabAssay™ Triglyceride, FUJIFILM Wako Pure Chemical Corporation), non-esterified fatty acids (NEFAs) (LabAssay™ NEFA, FUJIFILM Wako Pure Chemical Corporation), and cholesterol (LabAssay™ Cholesterol, FUJIFILM Wako Pure Chemical Corporation), were measured according to the manufacturer's instructions. Plasma insulin [(Insulin enzyme-linked immunosorbent assay (ELISA) KIT (RTU), Shibayagi, Gunma, Japan] and GLP-1 [glucagon-like peptide-1 (Active) ELISA, Merck Millipore, Darmstadt, Germany] levels were determined using ELISA as described previously (16). For plasma GLP-1 measurement, the samples were treated with a dipeptidyl peptidase IV (DPP-IV) inhibitor (Merck Millipore) to prevent the degradation of active GLP-1.

The MCFA levels in the plasma and intestinal samples were determined following a previously described protocol with modifications (17). Briefly, the samples containing an internal control (C19:0) were homogenized in methanol and mixed with chloroform and water to extract the lipids. The samples were centrifuged at 2,000 × g at 17 °C for 10 min. The supernatant containing MCFAs was collected and dried. The samples were resuspended in chloroform:methanol (1:3, v/v) and subjected to liquid chromatography with tandem mass spectrometry (LC-MS/MS) analysis using an ultra-performance LC system (UPLC, Waters, Milford, MA, USA) equipped with an Acquity UPLC system coupled to a Waters Xevo TQD mass spectrometer (Waters). The samples were separated on an ACQUITY UPLC BEH C18 column (2.1 × 150 mm, 1.7 μm; Waters) using a methanol gradient in 10 mM ammonium formate aqueous solution.

For the oral glucose tolerance test (OGTT), mice were allowed to fast for 16 h, and administered MCT (C8:0 triglyceride, C10:0 triglyceride, or C12:0 triglyceride; 2.4 g/kg body weight) or decanoate (C10:0; 1 g/kg body weight) in 0.5% carboxymethylcellulose through oral gavage. At 1 h post-administration (MCT administration) or 2 h post-administration (MCFA administration), mice were orally administered glucose (2.5 mg/g body weight). The plasma glucose concentration was monitored before injection and at 15, 30, 60, 90, and 120 min post-injection.

Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and RNAiso Plus (Takara, Shiga, Japan). cDNA was transcribed using total RNA as the template with Moloney murine leukemia virus reverse transcriptase (Invitrogen, CA, USA). qRT-PCR analysis was performed using SYBR Premix Ex Taq II (Takara) and the StepOne™ real-time PCR system (Applied Biosystems, CA, USA), as described previously (18, 19). The sequences of the primers were as follows: Gpr84, 5′-AGGTGACCCGTATGTGCTTC-3′ (forward) and 5′-AGATGATGGGATTGATCACAGGAG-3′ (reverse); 18S, 5′-ACGCTGAGCCAGTCAGTGTA-3′ (forward) and 5′-CTTAGAGGGACAAGTGGCG-3′ (reverse).

STC-1 cells (mouse enteroendocrine cell line; ATCC) were cultured in Dulbecco's modified eagle medium (DMEM; Sigma, St. Louis, MO, USA) containing 1% penicillin and streptomycin (Invitrogen), 5% fetal bovine serum (FBS; Funakoshi, Tokyo, Japan), and 15% horse serum (Gibco, NY, USA). To measure GLP-1 secretion, the cells were plated in 24-well-plates (approximately 1 × 10⁵ cells/well) and cultured for 48 h (17). The cells were treated with octanoate (C8:0), decanoate (C10:0), dodecanoate (C12:0), and embelin (Cayman Chemical, MI, USA) for 1 h. The culture supernatant was collected in the presence of a DPP-IV inhibitor. For siRNA-mediated knockdown, STC-1 cells were transfected with 30 nM siRNA (universal control: 5′-UGGUUUACAUGUCGACUAA-3′, Gpr84: 5′-GUUGGGCUAUCGAUACUUU-3′, Dharmacon, Lafayette, CO, USA) using Lipofectamine 2000 transfection reagent (Invitrogen), following the manufacturer's instructions. For inhibitor treatment, STC-1 cells were pretreated with the Gα(i/o) blocker NF023 (10 μM; Calbiochem, CA, USA), Gi/o-type G protein inactivator pertussis toxin (PTX; 1 μg/mL; Wako), Gβγ blocker Gallein (30 μM; Merck Millipore), the phosphoinositide-specific phospholipase C (PLC) inhibitor U73122 (1 μM; Wako), or GPR40 antagonist GW1100 (10 μM; Merck Millipore) for 20 min prior to the addition of decanoate (C10:0).

Flp-In T-REx HEK293 cells were transfected with a mixture containing mouse HA–Gpr84 or human FLAG-GPR40 cDNA in pcDNA5/FRT/TO and pOG44 vectors using Lipofectamine (Invitrogen) as described previously (20). Flp-In GPR84 or GPR40 T-REx HEK293 cells were cultured in DMEM containing 10 μg/mL blasticidin S (Funakoshi, Tokyo, Japan), 100 μg/mL hygromycin B (Invitrogen), and 10% FBS. The cells were incubated at 37 °C in an atmosphere of 5% CO₂. To determine cAMP levels, cells were plated on 24-well-plates (~1 × 10⁵ cells/well) and incubated for 24 h. Each well was then treated with or without doxycycline (10 μg/mL) for 24 h. The cells were treated with 3-isobutyl 1-methylxanthine (IBMX; 500 μM; Sigma; phosphodiesterase inhibitor) for 30 min and then stimulated with forskolin (2 μM; Sigma; adenylate cyclase activator) for 10 min. cAMP concentration wa determined using a cAMP enzyme immunoassay (EIA) kit (Cayman Chemical) according to the manufacturer's protocol (20). For intracellular calcium ([Ca²⁺]i) response analysis, cells were plated on 96-well black plates (~3 × 10⁴ cells/well) and incubated for 24 h. Each well was treated with or without doxycycline (10 μg/mL) for 24 h. After treatment, the cells were further incubated in Hank's balanced salt solution (pH 7.4) containing a calcium assay kit component A (Molecular Devices, CA, USA) for 1 h at 37 °C. These plates were placed on a FlexStation 3 multi-mode microplate reader, and the mobilization of [Ca²⁺]i was monitored (17).

All values are presented as mean ± standard error of mean (SEM). Differences between groups were examined for statistical significance using two-tailed unpaired Student's t-test (two groups) and two-tailed one-way analysis of variance, followed by Dunnett's or Tukey-Kramer's post-hoc test (≥ three groups). ROUT and Smirnov-Grubbs' tests were used for evaluating outliers.

We first investigated the MCT-induced changes in metabolic parameters after a 20% fat diet, which replaced lard with each type of MCT (octanoate; TriC8, decanoate; TriC10, and dodecanoate; TriC12). After 5 week-feeding of MCT diet (Supplementary Table 1), the body weights of TriC10 and TriC12 diet-fed mice were significantly lower than those of Lard diet-fed mice during growth (Lard > TriC8 > TriC10 > TriC12; Figure 1A). In addition, the weight of white adipose tissue (WAT) was also lower in TriC10 and TriC12 diet-fed mice than in Lard diet-fed mice at 12 weeks of age (Lard > TriC8 > TriC10 > TriC12), whereas the liver weight of TriC8 diet-fed mice was significantly higher than those of Lard diet-fed mice (Figure 1B). The levels of plasma MCFAs C10:0 and C12:0 but not C8:0 were significantly increased by each type of MCT diet (Supplementary Figure 1A), whereas plasma ketone body levels were comparable between Lard and each type of MCT diet-fed mice (Supplementary Figure 1B). Moreover, plasma glucose levels in TriC12 diet-fed mice were significantly lower (Lard > TriC8 > TriC10 > TriC12), and plasma total cholesterol levels in all MCTs diet-fed mice were lower than those in Lard diet-fed mice (Figure 1C). In addition, plasma TG levels in TriC12 diet-fed mice were significantly higher than those in Lard-fed mice (Lard < TriC8 < TriC10 < TriC12). Thus, although all MCT intake prevented obesity, different types of MCTs showed different metabolic functions.

We next performed OGTTs by oral MCT administration to examine the influence of MCT intake on glucose homeostasis. Following administration of MCTs (2.4 g/kg) and glucose, we found that the administration of all MCT types individually significantly suppressed the increase in blood glucose level compared to that in control mice (Figure 2A). Moreover, plasma insulin levels following glucose administration at 30 min in all MCT-administered mice were higher than those in control mice, whereas plasma GLP-1 levels in TriC10- but not TriC8- and TriC12-administered mice were significantly higher than those in control mice (Figure 2B). GLP-1 is secreted from enteroendocrine L cells, which mainly exist in the ileum and colon. At 1 h after the oral administration of each type of MCTs, C10:0 was detected in the ileal content and tissue at the highest levels among all MCFAs (Figures 2C,D). Thus, oral administration of MCTs enhanced glucose tolerance and increased plasma GLP-1 levels, especially with TriC10 administration.

MCFAs are ligands for FFAR GPR84. Hence, we examined the influence of GPR84 on GLP-1 secretion by MCTs. Gpr84 was expressed in the intestine, especially the ileum, and STC-1 cells as enteroendocrine cells, but not in the pancreas (Figure 3A). C10:0 and especially C12:0 strongly induced GLP-1 secretion in a dose-dependent manner in STC-1 cells, whereas C8:0 hardly induced these effects (Figure 3B). GPR40 is also a receptor for MCFAs; GPR84 couples to Gi/o, whereas GPR40 couples to Gq (1). Using a heterologous expression system, we found that C10:0 and C12:0 but not C8:0 significantly suppressed cAMP levels induced by forskolin in GPR84-overexpressing HEK293 cells (C10:0 > C12:0), whereas C10:0 and C12:0 but not C8:0 significantly increased [Ca²⁺]i levels in GPR40-overexpressing HEK293 cells (C10:0 < C12:0), and these effects were not observed in doxycycline (–) control HEK293 cells (Figures 3C,D). Thus, C10:0 and C12:0 but not C8:0 promotes GLP-1 secretion in STC-1 cells.

Moreover, we investigated whether C10:0-stimulated GLP-1 secretion is mediated by GPR84. The GPR84 agonist embelin (21), as well as C10:0, directly promoted GLP-1 secretion in a dose-dependent manner in STC-1 cells (Figure 4A). The GLP-1 secretion by either C10:0 or embelin stimulation was significantly inhibited by Gpr84 RNA interference in STC-1 cells (Figures 4B,C). Additionally, C10:0-stimulated GLP-1 secretion was effectively inhibited by treatment with pertussis toxin (PTX) (Gi/o-type G protein inactivator) (18, 22), Gallein (Gβγ blocker) (18, 23), and U73122 (PLC inhibitor), but not NF023 (Gα(i/o) blocker) (18, 24) (Figures 4D,E) in STC-1 cells. These results showed that C10:0-stimulated GLP-1 secretion in STC-1 cells was mediated by G(i/o)βγ-PLC signaling, but not Gq signaling. Moreover, C10:0-stimulated GLP-1 secretion in STC-1 cells was not abolished by treatment with the GPR40 antagonist GW1100 (25) (Figure 4F). Thus, C10:0-stimulated GLP-1 secretion is promoted via GPR84 rather than via GPR40.

We next investigated the effects of C10:0 via GPR84 on glucose homeostasis and GLP-1 secretion in vivo. In OGTT, either TriC10 or C10:0 administration significantly suppressed the increase in blood glucose levels in wild-type mice, whereas this effect was attenuated in Gpr84^−/− mice (Figures 5A,B). Additionally, oral C10:0 administration markedly increased plasma GLP-1 and insulin levels in wild-type mice, whereas these effects were attenuated in Gpr84^−/− mice (Figure 5C). Thus, C10:0 administration enhances glucose tolerance via GPR84-mediated GLP-1 secretion.

Finally, we investigated the beneficial effects of C10:0 supplementation on the metabolic functions of HFD-induced obese mice. After C10:0-supplemented HFD-feeding (Supplementary Table 2), we observed that the body weights of the mice were markedly lower than those of HFD-fed control mice during growth (Figure 6A). In addition, the weight of WAT was also lower in C10:0-supplemented HFD-fed mice than in HFD-fed control mice at 12 weeks of age (Figure 6B). Plasma C10:0 levels were significantly increased by C10:0-supplemented HFD feeding (Supplementary Figure 2). Moreover, plasma glucose and total cholesterol levels in C10:0-supplemented HFD-fed mice were significantly lower than those in HFD-fed control mice (Figure 6C). Plasma GLP-1 levels in C10:0-supplemented HFD-fed mice were significantly higher than those in HFD-fed control mice (Figure 6D). Additionally, food intake in C10:0-supplemented HFD-fed mice tended to be lower than that in the HFD-fed control mice (Figure 6E). Thus, C10:0 supplementation efficiently prevented HFD-induced obesity.