Time-restricted feeding may increase gut hormone GLP-1 and control appetite through tryptophan use by gut Lactobacillus in pigs

The animal procedure was carried out according to the guidance of the Animal Care and Use Committee of Nanjing Agricultural University (SYXK2019–0066) and implemented based on the standard of Experimental Animal Care and Use Guidelines of China (EACUGC2018–01). To investigate the potential associations between time-restricted feeding and appetite regulation, 28 healthy male Duroc × Landrace × Large White pigs (12.66 ± 0.11 kg) were randomly assigned to four groups, including ALF, TRF, eTRF and mTRF groups. Each group consisted of 7 replicates (pens) with 1 pig per pen. All pigs were subjected to a 12-h light/12-h dark cycle lighting protocol (Figure 1a), with the light period commencing at Zeitgeber time 0 (ZT0, 7:00) and ending at ZT12 (19:00). In the ALF group, pigs were allowed sufficient time for eating without any restrictions. The TRF group had a restricted eating window of 30 min at ZT0 (7:00), ZT5 (12:00), and ZT10 (17:00). The eTRF group had an eating window of 60 min at ZT0 (7:00) and 30 min at ZT5 (12:00). The mTRF group had a half-hour eating window at ZT5 (12:00) and a one-hour window at ZT10 (17:00). The total feeding duration for the three time-restricted feedings was 90 min per day. This duration was based on our previous studies, which found that the total eating time of pigs did not exceed 75 min when consuming 90% of ad libitum intake.²¹ During their respective permitted time windows, adequate food was provided to all groups. The nutrient composition and levels of feed were provided in Table S1. The feed intake and body weight were monitored on day 0, 14, and 29. The dynamic anterior vena cava blood was collected at 6-h intervals over a 24-h period, starting on the 26th day of the experiment. On day 30, the pigs were euthanized after an overnight fast at ZT0 (7:00). Fasting blood was collected and immediately stored at 4°C, then centrifuged within 30 min to minimize the proteolytic variability. The samples of serum were stored at −80°C for determination of hormones. The proximal colonic digesta was collected and stored at −80°C for metagenomics and metabolomics analyses. The tissues of hypothalamus and colon were collected and stored at −80°C for determination of gene and protein expressions.

The behavior of pigs was determined by video camera for three days on day 21. Feeding behavior was defined as the head of animal being within feeder for least 5 s, or when the animal was chewing food or grazing with its mouth in contact with the food.

The hypothalamus total RNA was extracted using Trizol reagent following the manufacturer’s protocol. Subsequently, high-quality RNA samples were utilized for mRNA purification through Dynabeads Oligo (dT) (Thermo Fisher, CA, USA). The resulting mRNA was fragmented and reverse-transcribed into cDNA, which served as a template for synthesizing U-labeled second-stranded DNAs. After size selection and amplification, a final cDNA library with an average length of 300 ± 50 bp was constructed. Paired-end sequencing (PE150) was conducted on an Illumina Novaseq™ 6000 (LC-Bio Technology CO., Ltd., Hangzhou, China), adhering to the recommended protocol provided by the vendor. The clean data was obtained and aligned to the reference genome Ensembl_v107. The gene expression abundance was estimated by calculating the fragment per kilobase of transcript per million mapped reads (FPKM). The differentially expressed genes (DEGs) were identified using the criteria of p < 0.05 and |log₂ (FC)| > 1, and enrichment analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway was performed using the OmicStudio platform (https://www.omicstudio.cn/tool↗).

The hypothalamic grinding powder was dissolved in sterile saline and centrifuged to obtain the supernatant. The blood samples were centrifuged to obtain serum. Gastrointestinal hormones, including ghrelin, leptin, glucagon, insulin (INS), 5-hydroxytryptamine (5-HT), GLP-1, PYY, cholecystokinin (CCK), glucose-dependent insulinotropic polypeptide (GIP), gastrin and dopamine (DA), were quantified using ELISA kits following the manufacturer’s instructions (Nanjing MALLBIO Biological Technology Co.,Ltd, Nanjing China).

The total RNA of hypothalamic and colonic tissues was extracted using the EASYspin Plus RNA extraction kit (Aidlab, China). Following dilution, the RNA was reverse transcribed into cDNA using the Evo M-MLV Mix Kit (Accurate Biology, AG11728, China). Target gene expressions were determined using the SYBR Green Premix Pro Taq HS qPCR Kit (Rox Plus) (Accurate Biology, AG11718, China) and quantified by normalizing with the reference gene β-actin according to the formula 2^−ΔΔCt. The primer sequences were provided in Table S2.

The colonic tissues were lysed in RIPA buffer (NCM Biotech, Suzhou, China) supplemented with protease inhibitors (Beyotime Biotech, Shanghai, China). The lysate was collected by centrifugation for gel electrophoresis. Following adjustment to the appropriate concentration, protein samples were separated using a 4–20% precast polyacrylamide gel electrophoresis (PAGE) gel and transferred onto polyvinylidene difluoride (PVDF) membranes. PVDF membranes were blocked using NcmBlot blocking buffer (NCM Biotech, Suzhou, China), followed by overnight incubation at 4°C with antibodies against ChgA (1:100 diluted; ABclonal, A9576), Notch1 (1:1000 dilution; ABclonal, A19090) and β-actin (1:2000 dilution; proteintech 20,536–1-AP). After washing steps, the membranes were further incubated with a secondary antibody (1:3000 dilution; ABclonal AS014) at room temperature. Finally, the blots were visualized using ChemiDoc Imager system (Bio-Rad) after immersion in enhanced chemiluminescence (ECL) HRP substrate solution (Yeasen Biotech, Shanghai, China). The ratio of target protein to β-actin was calculated using imageJ software (v 1.54 g).

The colonic tissues were fixed with paraformaldehyde and subsequently embedded in wax blocks. Following sectioning and de-waxing, permeation was performed at room temperature using 0.5% Triton X-100 (Solarbio, Beijing, China). Tissue sections were then subjected to blocking and incubated with primary antibodies targeting Olfm4 (1:100 diluted; Cell Signaling Technology, #14369), Muc2 (1:100 diluted; ABclonal, A14659), and ChgA (1:100 diluted; ABclonal, A9576). After three washes, the tissue sections were incubated with secondary antibodies for 60 min at room temperature. DAPI was employed for nuclear staining, and fluorescence micrographs were acquired using the panoramic scan. The mean fluorescence intensity of Olfm4 and Muc2 protein staining were quantified using the “Threshold” feature in ImageJ software (v 1.54 g). The quantity of positive cells was determined using the “Analyzed Particles” feature in ImageJ software (v 1.54 g).

The total DNA of colonic digesta was extracted using the E.Z.N.A.® Viral DNA Kit (Omega Bio-tek, Norcross, GA, U.S.) following the manufacturer’s protocols. The TruSeq DNA Library Preparation kit of Illumina was employed to construct sequencing libraries using high-quality DNA samples. Clean reads were obtained after quality trimming using Trimmomatic²² and removal of host-genome contaminations using the Burrows−Wheeler Alignment Tool (http://bio-bwa.sourceforge.net/bwa.shtml↗). The contigs of each sample were generated from clean sequence reads using MegaHit.²³ Prediction of the open reading frames (ORFs) of assembled contigs was performed using Prodigal v2.6.3, and clustering analysis was conducted using CD-HIT.²⁴ The values of transcripts per kilobase of exon model per million mapped reads (TPM) were calculated to represent gene abundance. The species abundance and gene function were annotated using the non-redundant protein (NR) database and KEGG database. Linear discriminant analysis effect size (LEfSe) analysis was conducted to identify the characteristic species using the R microeco package (v 1.6.0).

For non-targeted metabolomics analysis, the colonic digesta sample was mixed with 80% methanol and subjected to ultrasound treatment. After incubation at −40°C for 1 h, the supernatant was obtained through high-speed centrifugation and supplemented with 2-Chlorophenylalanine for subsequent analysis. The determination of metabolites was performed using Liquid Chromatograph Mass Spectrometer (LC-MS) analysis with mobile phases consisting of water containing 0.05% ammonium and acetonitrile. Raw data was normalized by probability quotient normalization (PQN) with quality control (QC), log transformation and auto scaling. The SIMCA-p software (v 14.0, Umetrics, Umea, Sweden) was used to calculate variable importance projection (VIP) values. The identification of different metabolites was based on the criteria of VIP > 1 and p < 0.05 between the ALF pattern and three time-restricted feeding patterns.

For targeted metabolomics analysis, the colonic digesta was mixed with 80% methanol followed by grinding. The supernatant was collected after centrifugation and diluted with 10% methanol. The diluent was mixed with the 20 μg/L L-Tryptophan-d5 solution for subsequent LC-MS analysis. Standards including tryptophan, indole-3-lactic acid (ILA), indole-3-propionic acid (IPA), indole, tryptamine, IAA, indole-3-aldehyde (IAld), and 5-HT were obtained to quantify the concentrations of these metabolites.

Ligilactobacillus salivarius and Ligilactobacillus animalis stains isolated from pig gut were cultured in MRS liquid medium contained 1 mM tryptophan for 12 h. After filtration, the supernatant was mixed with a fourfold volume of methanol and placed in the −80°C refrigerator for 0.5 h. The supernatant was collected after centrifugation and then evaporated. Methanol was subsequently added for redissolution. The supernatant was taken for High performance liquid chromatography (HPLC) measurement after high-speed centrifugation. A total of 10 μL supernatant was injected and separated by gradient elution. The concentration of acetonitrile containing 0.1% formic acid (v/v) was varied from 5% to 95% over a period of 20 min at a flow rate of 0.8 mL/min, with detection performed at a wavelength of 280 nm.

To investigate the impact of ILA on GLP-1 secretion in vivo, a 7-day short-term experiment using weaned piglets was conducted. A total of 12 Duroc × Landrace × Large White pigs (7.41 ± 0.16 kg) were randomly assigned to two groups: the control group and ILA group (Figure 8a). Each group consisted of 6 replicates (pens) with 1 pig per pen. Pigs in the control group were fed a basal diet, while those in the ILA group received the same basal diet supplemented with 100 mg/kg of ILA. All pigs were provided with feed and water ad libitum. After an overnight fast, blood was collected from the anterior vena cava and stored at 4°C. Serum samples were obtained by centrifugation within 30 min for subsequent measurement of GLP-1 levels.

The jejunal crypts were isolated from a 7-day-old piglet and cultured as previously described.²⁵ A suitable segment of middle jejunum was longitudinally dissected, and the intestinal mucosa was removed using sterilized slides. Then, the jejunum was cut into 1 cm × 1 cm pieces. Throughout this process, cold PBS containing antibiotics was continuously rinsed until the supernatant remained clear. The tissue pieces were digested with 25 mL of Gentle Cell Dissociation Reagent (Stemcell) for 30 min at room temperature. The crypts were isolated by filtration through a 70 μm filter followed by centrifugation, and subsequently resuspended in PBS supplemented with 1% fetal bovine serum. After counting, the intestinal crypts were resuspended in IntestiCult organoid growth medium (Stemcell) and Matrigel (BD Biosciences), and planted onto a 24-well plate. Following gelation, the crypts were supplemented with 500 μL of organoid growth medium and incubated at 37°C with 5% CO₂. After passaging, the intestinal organoids were treated with different concentrations of ILA for 60 h. The total RNA of intestinal organoids was extracted using the EASYspin Plus RNA extraction kit (Aidlab, China), and the medium supernatant was collected for GLP-1 determination using an ELISA kit (Nanjing MALLBIO Biological Technology Co., Ltd, Nanjing China). The expression levels of ChgA at the protein level were determined using immunofluorescence staining and western blot analysis. Briefly, the organoids were fixed using paraformaldehyde and permeabilized using 0.2% Triton X-100 (Solarbio, Beijing, China). After blocking, the organoids were incubated with primary antibodies targeting ChgA (1:100 diluted; ABclonal, A9576) overnight at 4°C. After washing, the organoids were incubated with the secondary antibody (1:250 diluted; Bioworld, BS10017) for 1 h at room temperature. After DAPI staining, the fluorescence of the organoids was identified using the panoramic scan. Additionally, the protein of organoids was extracted using RIPA buffer supplemented with protease inhibitors for subsequent western blot analysis.

The intestinal secretin tumor cell line (STC-1) was cultured in Dulbecco’s modified eagle medium (DMEM, high glucose). ILA at concentrations of 10, 100, and 1000 μM was added to the medium for 6 h and 12 h. Diprotin A was added to the cell medium in order to prevent degradation of GLP-1. After incubation, the cell supernatant was collected for GLP-1 determination using an ELISA kit.

The growth performance, hormone levels, metabolite levels, as well as gene and protein expressions from both in vivo and in vitro experiments were analyzed using one-way ANOVA followed by post-hoc Duncan’s test and Student’s t-test. The behavioral data following a normal distribution were subjected to one-way ANOVA analysis, while the behavioral data with a non-normal distribution were analyzed using the Kruskal-Wallis test. The microbiological data were assessed using the Kruskal-Wallis test.

The average daily feed intake of ALF at 0–14 d exceeded that of the three time-restricted feeding groups, and surpassed both eTRF and mTRF at 14–29 d. During the final 15 days of the experiment, the average daily feed intake of TRF pattern was higher than eTRF and mTRF patterns (Figure 1b). Correspondingly, ALF resulted in a significant increase in body weight compared to the other groups on day 14, and further increased body weight compared to eTRF and mTRF on day 29. TRF increased body weight compared to eTRF and mTRF on day 29, but there was no significant difference between TRF and ALF (Figure 1c). Additionally, the feed/gain ratio of ALF was significantly higher than that of other groups at 14–29 d (Figure 1d). The analysis of eating behaviors revealed that ALF group increased the overall duration of eating compared to other groups (Figure 1e). The frequency of eating was significantly higher in the ALF pattern compared to the TRF and mTRF patterns (Figure 1f). No significant impact was observed on eating rate (Figure 1g). It is noteworthy that, based on the hourly statistics, the eating time of pigs with three time-restricted feeding patterns fell within the permissible time range for each restricted window (Figure 1h).

The levels of neurotransmitter 5-HT and gastrointestinal hormones were quantified using dynamic serum samples collected every 6 h during the normal feeding periods, taking into account the influence of food intake timing and circadian clocks (Figure 2a). As shown in Figure 2b, the levels of appetite-related neurotransmitter 5-HT in the eTRF pattern were significantly elevated compared to those in the ALF pattern at ZT23 (1), ZT5 and ZT23(2). The levels of glucagon in the mTRF pattern were consistently elevated across four time points, exhibiting significant increases compared to eTRF and TRF at ZT5, significant increases compared to TRF at ZT11, and significant elevations compared to TRF, eTRF, and mTRF at both ZT17 and ZT23(2) (Figure 2c). However, INS levels were higher in the mTRF group compared to the other three groups at ZT11 (Figure 2d). Higher levels of ghrelin and the ghrelin-to-leptin ratio in the mTRF group were observed, except for ghrelin at ZT23(2) and the ghrelin-to-leptin ratio at ZT23(1), despite there being no significant differences in leptin levels among the four groups across 24 h (Figure 2e-g). The levels of dynamic gastrin, CCK, and PYY showed no significant differences over a 24-h period (Figure 2h,i and l). However, at ZT23 (2), the GIP levels of ALF were found to be higher compared to mTRF and eTRF (Figure 2j). The levels of GLP-1 in ALF group were consistently low, except at ZT17, indicating a noticeable trend at ZT23 (1) and a significant difference at ZT11 (Figure 2k). To investigate the potential associations between neurotransmitters and hormones with central regulation of appetite, the levels of GLP-1, DA, 5-HT, ghrelin, and leptin in hypothalamic tissue were determined. The results revealed a significant decrease in GLP-1 levels in the ALF pattern compared to those on time-restricted feeding patterns (Figure 2m). However, the hypothalamic levels of DA, 5-HT, ghrelin and leptin exhibited no significant differences (Figure 3n–q). Furthermore, fasting serum GLP-1 levels exhibited a significant decrease in the ALF pattern compared to eTRF and mTRF patterns (Figure 3r,s).

Transcriptomic analysis of the hypothalamus, a pivotal feeding center, revealed that ALF significantly upregulated 481, 856, and 1011 gene expressions and downregulated 674, 856, and 1279 gene expressions compared to TRF, eTRF and mTRF, respectively (Figure S1A, S2A and S3A). Interestingly, the DEGs between ALF pattern and the three time-restricted feeding patterns were found to be significantly enriched in neurotransmitter pathways, including GABAergic synapse, glutamatergic synapse, cholinergic synapse, serotonergic synapse and dopaminergic synapse (Figure S1B, S2B and S3B). ALF resulted in the upregulation of a majority of gene expressions that were enriched in neurotransmitter pathways (Figure 3a and Figure S1C, S2C and S3C). Notably, ALF upregulated the expressions of DRD1, DRD5, HTR4 and HTR1B. Furthermore, real-time PCR analysis demonstrated higher gene expressions for DRD1 and HTR1B in the ALF pattern compared to the three time-restricted patterns (Figure 3b). However, no statistically significant differences were observed in the expressions of NPY, AGRP, POMC and CART (Figure 3c,d).

To elucidate the underlying factors contributing to elevated GLP-1 levels under time-restricted feeding, the expression levels of genes involved in stem cell proliferation and differentiation in the ileum and colon were assessed. The results demonstrated that feeding patterns had no significant impact on the expression levels of markers related to stem cell proliferation and differentiation in the ileum (Figure 4a). Additionally, the expressions of active intestinal stem cell marker (Lgr5) and tuft cell marker (Dclk1) showed no significant difference in the colon, as shown in Figure 4b. And no significant difference was observed in the mean fluorescence intensity of the active intestinal stem cell marker Olfm4 in the colon under different feeding patterns (Figure 4c). However, compared to ALF and mTRF, TRF significantly upregulated the expression of goblet cell marker gene (Muc2) in the colon. Additionally, eTRF exhibited an upregulation in the expression of EEC marker gene (ChgA) compared to ALF, TRF, and mTRF in the colon (Figure 4b). Meanwhile, the immunofluorescent staining of colonic tissues revealed that the mean fluorescence intensity of Muc2 was higher in the TRF pattern compared to the ALF pattern, and the number of EECs increased in the eTRF pattern compared to the ALF, TRF, and mTRF patterns. (Figure 4c). The results of the western blot analysis also demonstrated that eTRF significantly increased ChgA protein expression compared to the other groups (Figure 4d). Further investigations into EEC marker genes revealed that eTRF resulted in the upregulation of the L cell marker gene Gcg compared to ALF (Figure 4e). Consistently, among the three groups of time-restricted feeding, eTRF exhibited a significant downregulation of Notch1 gene and protein expressions (Figure 4f,g), as well as upregulation of gene expressions of Math1 and Ngn3 compared to ALF (Figure 4f).

In order to investigate the potential microbiological mechanisms underlying the induction of EEC differentiation by feeding patterns, metagenomic sequencing was conducted to determine microbial composition and metabolic functions. The results shown in Figure 5a revealed that 17 genera exhibited statistically significant differences among the genera with relative abundance exceeding 1%. It is noteworthy that three time-restricted feeding patterns facilitated the colonization of Lactobacillus and Ligilactobacillus. Furthermore, TRF significantly facilitated the colonization of Faecalibacterium and Megasphaera, while both TRF and mTRF promoted the colonization of Roseburia. At the species level, LEfse analysis revealed that eTRF was predominantly characterized by Lactobacillus, including Ligilactobacillus animalis, Lactobacillus absiana, Lactobacillus delbrueckii, and Limosilactobacillus mucosae. TRF was primarily associated with bacteria involved in carbohydrate metabolism, such as Megasphaera elsdenii (Figure 5b).

To investigate the potential mechanisms underlying the difference in GLP-1 levels between the ALF pattern and the three time-restricted feeding patterns, an untargeted metabolomic analysis was conducted to identify possible intermediates involved in host-gut microbial interactions. As shown in Figure 6a, the correlation analysis between different genera and metabolites showed that Lactobacillus and Ligilactobacillus were positively correlated with indole metabolites, while negatively correlated with amino acids such as levodopa and cysteine. Thereinto, ILA, as a metabolite from tryptophan by intestinal microbes, was positively related to Lactobacillus and Ligilactobacillus. Targeted metabolomics analysis was conducted to quantify the concentrations of tryptophan metabolites in colonic digesta. The results demonstrated that eTRF and mTRF significantly elevated ILA levels compared to ALF. Additionally, TRF and mTRF showed higher concentrations of IPA relative to ALF and eTRF. Conversely, ALF led to a significant increase in 5-HT levels (Figure 6b). Based on metagenomic information, the metabolic pathway of tryptophan metabolism was elucidated. As shown in Figure 6c, eTRF exhibited a higher abundance of genes involved in the conversion of tryptophan into ILA and IAA. Notably, the key gene ldhA responsible for the conversion of tryptophan into ILA primarily contributed by Lactobacillus and Ligilactobacillus.²⁶ And Limosilactobacillus primarily contributed to the key gene amiE responsible for the conversion of tryptophan into IAA.

The species assigned by ldhA were shown in Figure 7a. The relative abundance of ldhA contributed by Limosilactobacillus sp012843675, Limosilactobacillus mucosae, and Ligilactobacillus salivarius in the eTRF pattern exhibited a significantly higher levels compared to that observed in the ALF pattern. And the eTRF showed a significantly higher abundance of ldhA contributed by Lactobacillus nasalidis and Ligilactobacillus animalis compared to the TRF and ALF patterns. Correlation analysis showed that these species were positively correlated with ILA levels in colonic digesta. Among them, the correlation coefficients of Lactobacillus nasalidis, Ligilactobacillus salivarius, and Ligilactobacillus animalis were higher (Figure 7b). To validate the tryptophan conversion capability, Ligilactobacillus salivarius and Ligilactobacillus animalis were cultured in MRS medium supplemented with 1 mM tryptophan. As shown in Figure 7c, ILA and IPA were identified in the supernatant of Ligilactobacillus salivarius, while IPA was also detected in the supernatant of Ligilactobacillus animalis.

To further investigate the impact of ILA on GLP-1 secretion in vivo, a short-term 7-day experiment was conducted (Figure 8a). Although no significant difference was observed in the average daily feed intake, body weight, and feed-to-gain ratio (Figure 8b–d), supplementation with ILA tended to increase serum GLP-1 levels (Figure 8e). To investigate whether microbial metabolite ILA play a pivotal role in EEC differentiation, jejunal crypts were isolated from a 7-day-old pig and cultured with different concentrations of ILA for a duration of 60 h (Figure 9a). The results indicated that exposure to 5 μM, 20 μM and 80 μM ILA led to an increase in the surface area of the intestinal organoids, whereas the budding efficiency remained unchanged compared to the mock-treated group (Figure 9b,c). As shown in Figure 9d, 80 μM ILA significantly increased the gene expressions of the goblet cell marker (Muc2) and 20 μM ILA significantly increased the expressions of the EEC marker gene (ChgA) compared to the mock-treated group. Additionally, treatment with 20 μM ILA resulted in the upregulation of Gcg gene expressions and an increase in GLP-1 levels within the supernatant of the organoid culture medium (Figure 9e,f). Moreover, 20 μM ILA significantly upregulated the expressions of Math1, Ngn3, Neurod1 and Isl1 (Figure 9g). The results of the immunofluorescence staining and western blot analysis demonstrated that 20 μM ILA significantly increased the number of ChgA-positive cells (Figure 9h) and the protein expression levels of ChgA (Figure 9i). In order to investigate the potential of ILA in promoting GLP-1 secretion in intestinal EECs, STC-1 cells were treated with ILA for 6 h and 12 h, and the levels of GLP-1 in the medium were determined. The results showed that only treatment with 1000 μM ILA significantly enhanced the secretion of GLP-1 by STC-1 cells (Figure 9J).

The RNA-Seq transcriptome datasets are available at NCBI project PRJNA1151847. The metagenome sequencing data are available at NCBI project PRJNA1177368.