L. plantarum OLL2712, which had been heat-treated by incubation at 75°C for 60 min and lyophilized, after being cultured in de Man, Rogosa, and Sharpe (MRS) broth (Becton Dickinson, USA), and stored at -20°C.

C57BL/6 male mice were purchased from Charles River Laboratories (Japan, RRID : IMSR_CRL:027). Mice were fed sterilized (121°C, 20 min) water and maintained at an appropriate temperature (23 ± 2°C) and humidity (50 ± 5%) with a 12-hour light-dark cycle. All experiments were conducted with the approval of the Experimental Animal Ethics Committee of the Graduate School of Agriculture and Life Sciences of the University of Tokyo.

To create obese C57BL/6NCrl mice, mice were fed a HFD (60% kcal from fat; Oriental Yeast, Japan) from 8-week-old, and mice in the control group were fed a normal chow diet (AIN-93M; Oriental Yeast, Japan) ( Figure S1A ). 6 individuals were used for each group to investigate the proinflammatory changes induced by a 4-week HFD. The nutrient composition of HFD-60 and AIN-93M is shown in Tables S1 , S2 .

To investigate the effects of oral administration of L. plantarum OLL2712 in obese mice, C57BL/6NCrl mice were fed a HFD from 8-week-old for 4 weeks. In the last 3 weeks, L. plantarum OLL2712, suspended in the sterilized water to the concentration of 20 mg/mL, was orally administered every day, 4 mg to each mouse ( Figure 1A ). 8 – 12 individuals were used for each group.

To investigate the effects of a short-term oral administration of L. plantarum OLL2712 in mice, L. plantarum OLL2712, suspended in the sterilized water to the concentration of 20 mg/mL, was orally administered every day, 4 mg to each C57BL/6NCrl mouse from 9-week-old ( Figure S1B ). 5 individuals were used for each group.

Epididymal adipose tissue (EAT) and mesenteric adipose tissue (MAT) were shredded into 2-3 mm pieces and dissociated with collagenase type II (1 mg/mL; Sigma-Aldrich, USA, Cat#C6885) at 37°C for 45-60 min until the adipose tissue was almost dissolved, and then the reaction was stopped with EDTA (10 mM) for 5 min. After being filtered with a 115 µm nylon mesh (Tokyo Screen, Japan), stromal vascular fraction (SVF) derived from adipose tissue was treated with red blood cell lysis buffer, made from ammonium chloride, potassium carbonate, and EDTA, for 5 min at room temperature. After centrifugation, the EAT SVF and MAT SVF were suspended in 10% FCS-RPMI.

Peyer’s patches (PPs) were treated with collagenase I (1 mg/mL; FUJIFILM Wako Pure Chemical, Japan, Cat#032-22364) and 10 μg/mL DNase I (Roche Diagnostics, Germany, Cat#10104159001) at 37°C for 60–90 min before being filtered with an 86 µm nylon mesh (Tokyo Screen, Japan). The PP cells were centrifuged twice and suspended in 10% FCS-RPMI. 10% FCS-RPMI was prepared using RPMI 1640 (Nissui Pharmaceutical, Japan, Cat#05918), containing 100 U/ml penicillin G potassium (Meiji Seika Pharma, Japan), 100 μg/ml streptomycin sulfate (Meiji Seika Pharma, Japan), 50 μM 2-mercaptoethanol (FUJIFILM Wako Pure Chemical, Japan, Cat#137-06862), 0.03% L-glutamine (FUJIFILM Wako Pure Chemical, Japan, Cat#074-00522), and 0.2% sodium hydrogen carbonate (FUJIFILM Wako Pure Chemical, Japan, Cat#191-01305), and 10% heat-inactivated fetal calf serum (Thermo Fisher Scientific, Germany, Cat#173012).

After F4/80 MicroBeads Ultrapure (Miltenyi Biotec, Germany, Cat#130-110-443) were added to PP whole cells, F4/80⁺ cells were isolated using a magnetic-activated cell sorting (MACS) system (Miltenyi Biotec, Germany). The obtained F4/80⁺ cells were used as macrophages.

The intestinal contents were removed, and the intestinal tract was washed with PBS (-), added to TRIzol (Invitrogen, USA, 15596026), and homogenized using TissueRuptor (QIAGEN, Germany) until the tissue was barely visible. The intestinal tissue was immediately frozen in liquid nitrogen and stored at -80°C.

The intestinal tissue samples were thawed at 4°C. Then, 0.2 mL of chloroform (FUJIFILM Wako Pure Chemical, Japan, Cat#038-02606) was added to 1 mL of the sample in TRIzol reagent, and the mixture was stirred manually and kept at room temperature for 3 minutes. After centrifugation, the upper layer was transferred to a new tube, and 0.5 mL of isopropanol (FUJIFILM Wako Pure Chemical, Japan, Cat#166-04836) was added. After being kept at room temperature for 10 min, the sample was centrifuged. The sample was washed with 1 mL of 75% ethanol and dried at room temperature until the precipitate turned translucent. The RNA solution derived from the intestinal tissue was dissolved in sterilized water, and any DNA was removed using an RNase-Free DNase (QIAGEN, Cat#79254).

Total RNA from cells was isolated using QIAshredder (QIAGEN, Germany, Cat#79656), 2-mercaptoethanol (FUJIFILM Wako Pure Chemical, Japan, Cat#137-06862), and an RNeasy mini kit (QIAGEN, Cat#74106) according to the provided protocol. Complementary DNA (cDNA) was synthesized using SuperScript VILO MasterMix (ThermoFisher Scientific, USA, Cat#11755-050) and the GeneAmp PCR System 9700 (Applied Biosystems).

Synthesized cDNA samples were added to a LightCycler 480 Multiwell Plate 96 (Roche Diagnostics), and quantitative PCR was performed with a QuantiTect SYBR Green PCR Kit (QIAGEN, Cat#204143) using a CFX Connect Real-Time PCR Detection System (Bio-Rad, USA). The relative expression levels of each gene were standardized against the gene expression levels of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The primer sequences for quantitative PCR (qPCR) are shown in Table S3 .

Intestinal permeability in vivo, was measured using 4 kDa Fluorescein isothiocyanate-dextran (FITC-Dextran) (Sigma, Cat#46944). 4 hours after FITC-dextran was orally administered to mouse (12 mg per mouse), the serum was collected and diluted in a microplate reader (Greiner bio one, Austria), and the fluorescence was measured at 485(ex)/528(em) nm using a SpectraMax iD5 (Molecular Devices, Japan). The concentration of FITC-dextran was then calculated.

Intestinal permeability ex vivo was investigated according to a previous report (22). The whole digestive tract from the stomach to the final part of colon was collected. After MAT was removed, specific intestinal sections were collected. A 4 cm segment under the stomach was selected as the duodenum, a segment from the 5th to the 10th centimetre below the stomach as the jejunum, a 5 cm intestinal section proximal to the cecum as the ileum, and a 5 cm segment below the cecum as the colon ( Figure 2B ). The collected intestinal tracts were washed by PBS (-) and the contents were gently removed without breaking the intestinal tissues. A 1 mg/mL solution of 4 kDa FITC-dextran was injected into the selected intestinal sections tied with surgical sutures, and each segment was moved to DMEM (Sigma-Aldrich, USA, Cat#11965092) and placed at 37°C ( Figure S1C ). The concentration of FITC-dextran transported from the lumen to the DMEM was measured every 30 minutes, and the cumulative concentration (Q_t) of the DMEM, collected at each time point, was calculated using the following formula.

Q_t = (C_t*V_r) + (C_{t sum}*V_s), where:

Q_t = Cumulative concentration at time t

C_t = Concentration at time t (ng/mL)

V_r = Volume at the receiver side (mL)

C_tsum = Sum of all previous C_t

V_s = Volume sampled (mL)

Q_t versus time (t) was plotted and the slope (δQ_t/δt) was calculated. And the apparent permeability (P_app) of each individual intestinal sac was calculated using the following formula:

P_app = (δQ_t/δt)/(A*Co), where:

A = Area of tissue (cm²)

Co = Initial concentration (ng/mL)

The cecal contents were collected in 1.5 mL tubes and stored at -80°C. The gut microbiota was analysed with next-generation sequencing and amplicon sequencing by TechnoSuruga Laboratory (Japan). DNA was extracted according to the method previously reported (23), using an automated DNA isolation system (GENE PREP STAR PI-480 KURABO, Japan). The details of the analysis were provided by TechnoSuruga Laboratory (Japan). The 341f/R806 primers and dual-index method was used to amplify the V3-V4 regions of Bacterial 16S rRNA (23–26). And then barcoded amplicons were paired-end sequenced on 2×284-bp cycle using the MiSeq system with MiSeq Reagent Kit version 3 (600 Cycle) chemistry. Paired-end sequencing reads were merged by fastq-join ver 1.3.1 with default setting (27).

FASTX-Toolkit ver 0.0.13 was being used to extract joined-reads, which had quality value score of ≥ 20 for more than 99% of the sequence. After the chimeric sequences were deleted with usearch61 (28, 29), nonchimeric reads were submitted for 16S rDNA-based taxonomic analysis using the Ribosomal Database Project ver 2.11 (RDP, RRID : SCR_006633). Finally, Metagenome@KIN Ver 2.2.1 analysis software (World Fusion, Japan) was used to perform the identification with confidence ≥ 0.8.

The results are given as the mean ± SD, and Student’s t-test was used for statistical analyses. A difference was considered significant at p<0.05.

It has been frequently reported that diet-induced obesity is related to chronic inflammation. A long period of intake of HFD, usually more than 12 weeks, could cause abnormal cytokine production, a disorder of lipid metabolism, and elevated blood glucose, followed by disruption of the regulatory mechanism of adipocytokine production (30). To investigate the inflammation in early obesity induced by a short period of HFD ingestion, C57BL/6N mice were fed HFD (60% kcal from fat) for 4 weeks, and mice fed a normal chow diet (AIN-93 M) were used as a control group (ND) ( Figure S1A ). The mice were weighed every 7 days, and it was found that the HFD group showed increasing body weights ( Figure 3A ), followed by increasing weight of their EAT and MAT ( Figures 3B, C ).

We isolated SVF, which contained immune cells, from the EAT and MAT, and the gene expression of proinflammatory cytokines in the EAT SVF and MAT SVF of mice was measured by qPCR ( Figures S2A–C , 3D–F ). And the gene expression of F4/80 (F4/80), as a marker of macrophages, was measured simultaneously ( Figures S2D , 3G ). Cytokine chemokine (C-C motif) ligand 2 (CCL2; Ccl2), as a macrophage-specific chemokine, increased with a 4-week HFD diet in EAT SVF and MAT SVF ( Figures S2A , 3D ). Nevertheless, other major proinflammatory cytokine such as IL-1β (Il1b) and TNF (Tnf), and macrophage marker F4/80 (F4/80) did not show a remarkable change ( Figures S2B–D , 3E–G ). These data suggested that a short-term of HFD feeding could induce increases in body weight and fat mass, followed by slight increase in inflammation in the adipose-derived SVF by inducing CCL2 (Ccl2).

Many previous studies have reported that both obese patients and obese mice show an increase in Firmicutes and a decrease in Bacteroidetes in their gut microbiota (31, 32). We collected the contents of the cecum from mice fed a HFD for 4 weeks and analysed the gut microbiota composition using next-generation sequencing applications. The relative abundance of Firmicutes was found to increase in the HFD group compared to normal diet (ND) group ( Figure 4A ), which is a relevant marker of gut dysbiosis, and we detected a descending tendency in the relative abundance of Bacteroidetes ( Figure 4B ), although there was no change in the Firmicutes/Bacteroidetes ratio ( Figure 4C ). At the genus level, the mice in the HFD group showed an obvious distinction from the ND group ( Figure 4D ). The HFD group exhibited a decreasing tendency in the relative abundance of Lactobacillus (p = 0.06, Figure 4E ) and the relative abundance of Lactococcus, Lachnospiracea incertae sedis, and Peptococcus which was almost absent in the cecum of the control group, was found to increase with a HFD feeding ( Figures 4F, H, J ). Although Pseudoflavonifractor did not change ( Figure 4I ), an increasing tendency induced by HFD was detected in the relative abundance of Clostridium cluster XIVa (p = 0.08, Figure 4G ). These data indicated that a short-term HFD feeding had already resulted in a substantially different gut bacterial flora compared with the ND group, which could be involved in inflammation-associated diseases.

To explore the anti-inflammatory effects of L. plantarum OLL2712, we orally administered the heat-treated strain to mice on a HFD during the last 3 weeks ( Figure 1A ). We were unable to detect a significant difference in body weight or fat mass between mice treated with L. plantarum OLL2712 and those treated with sterilized water ( Figures 1B–D ). Nevertheless, in the EAT SVF, the gene expression of F4/80 (F4/80) decreased significantly in the LAB-treated mice ( Figure S3D ). Although not significant, the gene expression of CCL2 (Ccl2) showed a declining tendency with the LAB treatment (p = 0.05) ( Figure S3A ), while there was no change found in the gene expression of IL-1β (Il1b) and TNF (Tnf) ( Figures S3B, C ). Simultaneously, remarkable changes were found in the MAT SVF, as the gene expression of proinflammatory cytokines, CCL2 (Ccl2), IL-1β (Il1b) and TNF (Tnf), and macrophages marker, F4/80 (F4/80), decreased in the LAB group ( Figures 1E–H ).

Colonic macrophages play important roles in the induction of obesity-associated insulin resistance. Both macrophage-specific CCR2 knockout and intestinal epithelial cell-specific tamoxifen-inducible CCL2 knockout mice have been observed to be resistant to HFD, showing improved glucose and insulin tolerance (18). To investigate the colonic macrophage infiltration underlying dietary obesity, we evaluated the changes of CCL2 and F4/80 under HFD and LAB treatment. Unexpectedly, we did not find colonic inflammation in mice fed an HFD for 4 weeks ( Figures S4A–D ). Nevertheless, in macrophages derived from PP cells, the gene expression of CCL2 (Ccl2) and TNF (Tnf) was upregulated by dietary-induced obesity ( Figures S4E, G ), suggesting that intestinal inflammation was already elicited in the small intestinal compartment, although there was no detected change in the gene expression of IL-1β (Il1b) and IL-6 (Il6) in PP macrophages ( Figures S4F, H ).

On the other hand, a 3-week oral administration of L. plantarum OLL2712 elicited decreased expression of CCL2 (Ccl2), IL-1β (Il1b) and F4/80 (F4/80) ( Figures 1I, J, L ) in the colon, and TNF (Tnf) did not change in the gene expression levels ( Figure 1K ). Meanwhile, there was no change detected in the duodenum, jejunum, and ileum (data not shown). From these results, we supposed that the LAB strain had an anti-inflammatory effect on the large intestine of mice. Furthermore, with the oral administration of heat-treated OLL2712, the gene expression of the proinflammatory cytokines IL-1β (Il1b) and TNF (Tnf) were found to be downregulated in PP macrophages ( Figures 1N, O ), although CCL2 (Ccl2) and IL-6 (Il6) did not change ( Figures 1M, P ).

We analysed the gut microbiota in the cecum of mice treated daily with heat-treated L. plantarum OLL2712 for 3 weeks compared with the control mice treated with water. At the phylum level, Firmicutes and Bacteroidetes did not show a significant change under the OLL2712 treatment ( Figures 5A–C ). However, at the genus level, the gut bacterial flora displayed a remarkable difference between mice treated with LAB and the control group treated with water ( Figure 5D ). The relative abundance of Lactobacillus dramatically increased ( Figure 5E ). Furthermore, the relative abundance of Clostridium cluster XIVa, Lachnospiracea incertae sedis, and Pseudoflavonifractor, which had been increased by the HFD and are considered to be associated with host inflammation and diseases, showed a significant decrease under OLL2712 treatment ( Figures 5G–I ). And the relative abundance of Lactococcus showed a declining tendency (p = 0.05) ( Figure 5F ), while Peptococcus did not change with the LAB treatment ( Figure 5J ). These data suggested that the oral administration of OLL2712 could modulate the gut microbiota composition related to obesity.

According to the experimental results obtained thus far, we confirmed the anti-inflammatory effects of heat-treated OLL2712 on adipose tissue, PP macrophages, and the colon. In addition, the LAB treatment caused a substantial change in the gut microbiome.

Obesity has been reported to increase intestinal permeability, after which the products of obesity-induced gut dysbiosis are allowed to translocate into the bloodstream, adipose tissue, or other organs, as one of the causes of chronic systemic inflammation (15, 33). Since intestinal inflammation and gut dysbiosis both affect the intestinal barrier (34–36), we next assessed the effects of the LAB strain on intestinal permeability. Seven days orally administration of heat-treated OLL2712 ( Figure S1B ) induced the lower levels of orally administered FITC-dextran in the serum compared to the control group, supporting our hypothesis ( Figure 2A ).

Since the proinflammatory cytokines in the colon decreased after 3 weeks of LAB treatment, while those in the jejunum did not, there was a possibility that different parts of the gut played different roles and might respond to the LAB strain in different ways. Assessment of the intestinal permeability in the previous in vivo experiment measured overall gastrointestinal absorption without any site specificity.

To further investigate site specificity of the protective effects of LAB treatment on the gut barrier, we collected duodenum, jejunum, ileum, and colon from the LAB treated mice and assessed the permeability ex vivo ( Figures 2B , S1C ). After 3 weeks of LAB treatment, the permeability of the ileum derived from HFD mice showed a significant reduction ( Figure 2E ), while the permeability of other segments had barely changed compared to that of HFD mice treated with water ( Figures 2C, D, F ). To gain mechanistic insight, we investigated the relative expression of barrier-related genes in each intestinal section. In the ileum, the expression of Occludin (Ocln), as one of the proteins forming tight junctions, decreased with a 4-week HFD feeding ( Figure 2H ). In addition, the gene expression of a secreted mucin with a physical barrier function, MUC2 (Muc2), also decreased with obesity ( Figure 2I ). And there was no significant change detected in the gene expression of ZO-1 (ZO1) ( Figure 2G ). The gene expression of ZO-1 (ZO1) and Occludin (Ocln) increased with the LAB treatment ( Figures 2J, K ), although no significant change was found in MUC2 (Muc2) ( Figure 2L ). These data suggested that LAB treatment blocked intestinal barrier disruption in the ileum of HFD-fed mice.