Anti-mouse phosphor-STAT3 (EP2147Y, Abcam), phosphor-NF-κB (pp65) (sc-52401, Santa), phosphor-JNK (pJNK)(9H8), phosphor-ERK (pERK)(197G2), STAT3 (124H6, Cell Signaling), NF-κBp65 (sc8008, Santa), JNK (SC), ERK(137F5), and β-actin (sc-47778, Santa) were purchased. FITC-, PE-, APC- APC/cy7-, PerCP/Cy5.5-, or PE/cy7-conjugated anti-mouse CD4 (RM4-5, Biolegend), CD8 (ZUT270.5, Biolegend), CD45.1 (30-F11, Biolegend), MHCII (I-A/I-E, M5/114.115.2, Biolegend), CD11c (MCA1441GA, Biolegend), CD11b (M1/70, eBioscience), F4/80 (BM8, Biolegend), CD206 (CO68C2), Gr-1(RB6-8C5, eBioscience), IL-10 (JES5-16E3, eBioscience), Foxp3 (MF23, eBioscience), IL-22 (140301, R&D system), IFNγ (XMG1.2, Biolegend), RORγt (AFKJS-9, eBioscience), IL-17A (eBio17B7, eBioscience), and CD3ζ (H146-968, Abcam) antibodies were purchased. Anti- REG3γ (PA517, Thermo) and anti-MUC2 (H300, Santa) were also purchased. FITC-, PE-, APC- APC/cy7-, PerCP/Cy5.5-, or PE/cy7-conjugated isotypic controls were from Biolegend. RPMI1640, DMEM, fetal bovine serum (FBS), and antibiotics were obtained from HyClone. Alexa fluor 488-conjugated goat anti-rabbit IgGH&L and Alexa fluor 594-conjugated goat anti-rabbit IgGH&L (Abcam) were also purchased. 7-AAD was from Abcam.

Lactobacillus reuteri (ATCC PTA 4659) was a gift of BioGaaia, Sweden.

Four- to six-week-old male or female C57BL/6, OT-I, OT-II, CD45.1, and TLR2^−/− mice were obtained from Nanjing Animal Center. The FVB-GFP/Fluc mice were kindly offered by Prof. Zongjin Li in Nankai University. C57BL/6 GF mice were generated by Shanghai SLAC Laboratory Animal Co. Ltd. All experiments for C57BL/6 GF mice were performed in Shanghai SLAC Laboratory Animal Co. Ltd.

For the preparation of huREG3γ^tgIEC mice, HD5 promoter, which may specifically promote the REG3γ expression in gut epithelial cells, was conjugated into Pinsulator–pHD5-promoter-CDS-poly plasmid. This conjugation was demonstrated by primers (REG3γ-HD5-tF: CATCCAACTCCAGGACGGAGTC and REG3γ-HD5-tR: CACCTGTAACATTGGCACTTTG) and sequencing using primers (Promoter-cx tF: GTTTGCTGGGTCAGAACTGA and Promoter-cx tR: GTAATTTAGGTGCGTACAGCC). Then the fragment of REG3γ-CDS and polyA were cloned into HD5 promoter-pinsulator. This conjugation was demonstrated using primers (REG3γ-PA-tF: CTTCCTGTGCAAGTACCGACC and REG3γ-PA-tR: GGTGTCTGCAGGCTCAAAGAG) and sequencing using the primers (HD5-Test-Tf: CTGTCTCAGGTCTTCTCCCAG and REG3γ -PA-F: GATCTTTTTCCCTCTGCCAAA). REG3γ transgenic mice were identified using the following primers (42: CTAGGCCACAGAATTGAAAGATCT and 43: GTAGGTGGAAATTCTAGCATCATCC, which produced 324 bp band). For experimental mice, WT mice first bred to huREG3γ^tgIEC mice, and then huREG3γ^tgIEC mice and their littermate WT mice were co-housed or separated into different cages after 4 weeks based on the experimental purpose.

All experimental litters were bred and maintained under specific pathogen-free (SPF) conditions in Nankai University. Experiments were carried out using age- and gender-matched mice. All the procedures were conducted according to the Institutional Animal Care and Use Committee of the Model Animal Research Center. Animal experiments were approved by the Institute’s Animal Ethics Committee of Nankai University. All experimental variables such as husbandry, parental genotypes, and environmental influences were carefully controlled. For high-fat diet-mediated obesity model, the experiments were strictly performed according to guidelines suggested by Stappenbeck and Virgin (20). Human samples were approved by the Institute’s Human Ethics Committee of Nankai University and agreed by individuals that offer fecal samples.

For microbiota transplantation, 6- to 8-week-old mice were treated with pan-antibiotics [ampicillin (A, 1 g/l, Sigma), vancomycin (V, 0.5 g/l), neomycin sulfate (N, 1 g/l), and metronidazole (M, 1 g/l)] via the drinking water. Water containing antibiotics was exchanged every 3 days. To confirm the elimination of bacteria, stool was collected from antibiotic-treated and untreated mice and cultured in anaerobic and aerobic conditions. The bacteria were counted under microscope. Then cecal contents from detergent treated mice or bacteria (1 × 10⁹) were suspended in 1 ml PBS with 30% glycerol. Mice (4 weeks old) were removed from the isolator and were orally administered 200 ml of fecal suspension or bacteria made using glycerol stocks.

For high-fat diet model, 6- to 8-week-old male mice and their control littermates were fed using high-fat diet (D12492, protein 26.2%, carbohydrate 26.3%, and fat 34.9%) and control diets (D12450B, 60% of calories may be derived from fat), which was from Research Diets, Inc. (NJ, USA).

For STAT3 inhibitor cucurbitacin I (JSI-124; Sigma Aldrich) treatment, pan-antibiotics-treated mice were infused with Lactobacillus NK318.1 (once/3 days, twice). Meanwhile, mice were treated ip with 1 mg/kg/of JSI-124 for 2 weeks (once/day).

For infusion of macrophages, mice before and after feeding high-fat diet were injected in tail vein using freshly isolated macrophages (1 × 10⁶/mice at day 0 and then once per week).

Gut microbiota was analyzed by Majorbio Biotechnology Company (Shanghai, China) using primers that target to the V3–V4 regions of the 16S rRNA. Once the PCR for each sample was completed, the amplicons were purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA, USA), quantified, normalized, and then pooled in preparation for emulsion PCR followed by sequencing using Titanium chemistry (Roche, Basel Switzerland) according to the manufacturer’s protocol. In the first step of data processing, the generated sequence data were de-convolved using the sample barcodes to identify sequences from each of the samples. Barcode, primer, and adaptor sequences were also trimmed as part of this step. PCR artifacts “chimeras” were identified using the ChimeraSlayer program (http://microbiomeutil.sourceforge.net; reference http://genome.cshlp.org/content/21/3/494.long) and removed prior to downstream analysis. The resulting de-convoluted and filtered sequence data were assigned taxonomy (to the genus level) using the Ribosomal Database Project classifier, and the genera classifications were used to generate a sample genus count matrix. Operational Taxonomic Unit (OTU) analysis of these sequences was performed as follows: sequences were processed (trimmed) using the Mothur software and subsequently clustered at 97% sequence identity using cd-hit to generate OTUs. The OTU memberships of the sequences were used to construct a sample-OTU count matrix. The samples were clustered at genus and OTU levels using the sample genus and sample-OTU count matrices, respectively. For each clustering, Morisita-Horn dissimilarity was used to compute a sample distance matrix from the initial count matrix, and the distance matrix was subsequently used to generate a hierarchical clustering using Ward’s minimum variance method. The Wilcoxon Rank Sum test was used to identify OTUs that had differential abundance in the different sample groups. Bacterial DNA sequencing data have been uploaded to the Sequence Read Archive database in NCBI with accession number SAMN04569211 and SAMN04569229.

For absolute numbers of gut microbiota, 16S rRNAs were extracted and then amplified using phylum such as Firmicutes, Bacteroidetes, or genus-specific primers (Table S1 in Supplementary Material). The concentration of each product was detected and then exchanged into copy numbers. The purified products were diluted into 10²–10⁸ copies/ml and then analyzed by quantitative real-time PCR (qRT-PCR), and then standard curve was prepared based on the CT value and concentration. Other sample DNAs were amplified by qRT-PCR, and the exact copy number was counted based on the molecules. 100 mg of stool was suspended in 900 µl of cysteine–peptone–water solution and homogenized. For quantification of mucosa-associated bacteria, total DNA was isolated from PBS-washed and weighted gut tissue using DNeasy Blood & Tissue Kit (Qiagen). DNA was then subjected to quantitative PCR as described above, and results are expressed as bacteria number per milligram of gut tissue, using a standard curve. Primers used were listed in Table S1 in Supplementary Material.

For lactobacillus isolation, 100 mg fresh fecal samples were collected and diluted in 2 ml BPS solution, cultured on Rogosa SL selective medium (Sigma Aldrich) for Lactobacillus enumeration, and then colonies were identified and purified for 16S ribosomal DNA sequence analyses for the speciation of colonial genotype. The lactobacilli were cultured in deMan, Rogosa, Sharpe (MRS; 3M Health Care, St. Paul, MN, USA) media and also grown on MRS agar containing either 10% sucrose. Anaerobic conditions were generated with sachets of AnaeroPack-Anaero (Mitsubishi Gas Chemical, Japan) in an air-tight jar. At 24 h of cultivation in liquid media, Lactobacillus NK318.1 could reach 1 × 10⁹ CFU/ml.

Mucus immunostaining was paired with fluorescent in situ hybridization (FISH) to analyze bacteria localization at the surface of the intestinal mucosa. In brief, ileum and colonic tissues (proximal colon, second centimeters from the cecum) containing fecal material were placed in methanol-Carnoy’s fixative solution (60% methanol, 30% chloroform, and 10% glacial acetic acid) for a minimum of 3 h at room temperature (RT). Tissue were then washed in methanol, ethanol, ethanol/xylene (1:1), and xylene, followed by embedding in paraffin with a vertical orientation. 5-mm sections were cut and dewaxed by preheating at 60°C for 10 min, followed by bathing in xylene at 60°C for 10 min, xylene at RT for 10 min, and 99.5% ethanol for 10 min. The hybridization step was performed at 50°C overnight with an EUB338 probe (59-GCTGCCTCCCGTAGGAGT-39 with a 5′ Cy3 label) diluted to a final concentration of 0.01 µg/ml in hybridization buffer (20 mM Tris–HCl, pH7.4, 0.9 M NaCl, 0.1% SDS, 20% formamide).

For Lactobacillus, the hybridization step was performed at 60°C overnight with an RNA probes (Lac663: FAM-O-ACATGGAGTTCCACT) diluted to a final concentration of 0.01 µg/ml in hybridization buffer [10% dextran sulfate (wt/vol), 10 mM NaCL, 30% (vol/vol) formamide, 0.1% (wt/vol) Ficoll, 5 mM Na₂EDTA, 0.1% (wt/vol) SDS, 0.1% (vol/vol)triton X-100, 50 mM Tris–HCl, pH = 7.4]. After washing for 10 min in wash buffer (wash solution: 2× SSC, 0.1% Tween-20) and 10 min in PBS, block solution (5% FBS in PBS) was added for 30 min at 50°C. Mucin 2 primary antibody (rabbit H300, Santa) was diluted to 1:1,500 in block solution and applied overnight at 4°C. After washing in PBS, block solution containing anti-rabbit secondary antibody diluted to 1:1,500 was applied to the section for 2 h. After washing in PBS, slides were mounted using Prolong anti-fade mounting media (Life Technologies). Nuclei were stained using Hoechst 33342. Observations were performed with a Zeiss LSM 700 confocal microscope with software Zen 2011 version 7.1. This software was used to determine the distance between bacteria and the epithelial cell monolayer, as well as the mucus thickness.

Student’s t-test, one-way analysis of variance, and the Mann–Whitney U test were used to determine significance. A 95% confidence interval was considered significant and was defined as p < 0.05. *p < 0.05, **p < 0.01, and ***p < 0.001.

Gut epithelial cells that produce bacteria-killing substance may affect the composition of gut microbiota (23, 24). REG3γ, a kind of bacteria-killing substance, may target Gram-positive bacteria (18). However, Reg3γ^−/− mice do not exhibit significant alterations in gut luminal bacterial communities (25). Interestingly, in Reg3γ^tgIEC mice, there exist significant differences in the composition of gut microbiota not only in gut mucus layer but also in luminal microbiota (24). Thus, REG3γ overexpressing mice offer a model to found some bacteria, which may potentially affect gut immune cell populations. We generated gut epithelial cell-restricted human REG3γ transgenic (huREG3γ^tgIEC) mice. Human REG3γ (Gene ID: 130120) is selectively expressed in mouse gut epithelial Paneth cells under the HD5 promoter (26). Consistent with other data (25), REG3γ could increase spatial space to segregate bacteria from gut epithelial cells. We compared the composition of gut microbiota of huREG3γ^tgIEC mice and their WT littermate mice by sequencing 16S rRNAs (V3–V4 regions) in pooled small intestine contents in male WT and huREG3γ^tgIEC mice and observed dramatic changes of the gut microbiota in mice fed with standard chow (Figure 1A). Although the antibacterial activity of REG3γ is restricted to Gram-positive bacteria (18), the proportion of Lactobacillus genus (Gram positive) was higher in huREG3γ^tgIEC mice than those in WT mice, consistent with the results from mouse Reg3γ^tgIEC mice (24). The high proportion of Lactobacillus genus in REG3γ overexpressing mice may be derived from the resistance of healthy microbiota (27). Notably, increased Lactobacilli were also detected not only in mucus layers of small intestine but also in colonic tissues by using q-PCR and fluorescence in situ hybridizations (Figures 1B,C). These results suggest that REG3γ overexpression in gut epithelial cells may affect the composition of gut microbiota.

Alterations of gut microbiota may affect the composition of gut immune cells and immune cell subsets (1, 14–16). Since Lactobacilli are rich in small intestine mucus layers of huREG3γ^tgIEC mice, we next investigated the composition of gut immune cells in the LP of ileum tissues using flow cytometry based on the gating strategies (Figure S1A). There was remarkably increased proportion of MHCII(+)F4/80(+) macrophages in the ileum LP in huREG3γ^tgIEC mice compared to control WT littermates (Figure 1D). Similar to previous data in WT mice (6), these MHCII(+)F4/80(+) macrophages in the huREG3γ^tgIECmice could also express CD11c and CD11b (Figure 1D). The absolute number of MHCII(+)F4/80(+) macrophages was remarkably higher in the small intestine of huREG3γ^tgIEC mice compared to control WT (Figure 1E). However, these MHCII(+)F4/80(+) cells also included CD206(+) and CD206(−), Gr-1(+) and Gr-1(−), and IL-10(+) and IL-10(−) macrophage subpopulations (Figures 1F,G). Furthermore, the proportion of MHCII(+)F4/80(+)CD206(+), MHCII(+)F4/80(+)Gr-1(+), and MHCII(+)F4/80(+)IL-10(+) macrophage subsets also increased in the LP of huREG3γ^tgIEC mice (Figures 1F,G). Notably, CD11b(+)Gr-1(+) myeloid-derived suppressive cells (MDSCs) and MHCII(+)CD11C(+) cells were much more in the LP of huREG3γ^tgIEC mice in different degrees. The proportion of Treg cell subsets also increased in huREG3γ^tgIECmice (Figure S2A in Supplementary Material), which may be related to enlarged anti-inflammatory macrophage pools (28). IL-10 levels in the ileum tissues of huREG3γ^tgIEC mice, which is related to anti-inflammatory macrophages and Treg, were higher than WT mice (Figure S2B in Supplementary Material). The proportion of other immune cell populations and subpopulations such as B cells, CD4, CD8, and other T cell subpopulations such as Th1, Th17, and IL-22 cells did not exhibit significant changes. These data indicate that REG3γ may specifically affect the differentiation of myeloid-derived cells such as macrophages.

Gut immune cells have been shown to possess broad trafficking abilities to lymphoid organs at distant sites, such as the MLN and spleen (29). Increased F4/80(+)MHCII(+) macrophages and macrophage subpopulations were also detected in the PP, MLN, and spleen of huREG3γ^tgIEC mice (Figures S2C,D in Supplementary Material). Notably, those increased subsets MHCII(+)F4/80(+)CD206(+), MHCII(+)F4/80(+)Gr-1(+), and MHCII(+)F4/80(+)IL-10(+) in the LP were also found in the PP and spleen of huREG3γ^tgIEC mice (Figures S2E,F in Supplementary Material). Taken together, our data demonstrate that REG3γ overexpression enlarges macrophage pools not only in the LP but also in the PP, MLN, and spleen.

Since MHCII(+) and F4/80(+) are makers of mature, anti-inflammatory macrophages in gut tissues (9), we also assessed immune tolerance situation at basic line. Beyond increased Treg cells and IL-10 in the LP of ileum tissues, myeloid cell-derived immune suppressor substance arginase-1, NO, and H₂O₂ were also higher in the LP of huREG3γ^tgIEC mice compared to control WT (Figure S3A in Supplementary Material). The expression levels of CD3 zeta chain of CD4 and CD8 cells in the LP of huREG3γ^tgIEC mice were significantly lower in huREG3γ^tgIEC mice (Figure S3B in Supplementary Material). The levels of inflammation cytokines including TNFα, IL-6, IL-1β, and IFNγ were lower in huREG3γ^tgIEC mice compared to control WT littermate mice (Figure S3C in Supplementary Material). Compared to control WT, freshly isolated MHCII(+)F4/80(+) macrophages (Figure S1B in Supplementary Material) from huREG3γ^tgIEC mice could express higher levels of IL-10 and higher levels of arginase-1, NO, and H₂O₂ (Figure S3D in Supplementary Material). These IL-10, NO, and H₂O₂ expression was further augmented by exposure to Lactobacillus NK318.1 (Figure S3E in Supplementary Material). Freshly isolated macrophages from the LP of huREG3γ^tgIEC mice also exhibited stronger suppression on CD4 and CD8 T cell proliferation (Figure S3F in Supplementary Material). Thus, our data suggest that increased macrophages in huREG3γ^tgIEC mice promote the homeostasis of gut tissues.

Next, we investigated how REG3γ cause the accumulation of anti-inflammatory macrophages. Previous studies showed that the composition of gut microbiota may affect the differentiation of gut immune cell populations or their subpopulations (1, 14–16). The proportion of macrophages in pan-antibiotics (ampicillin, vancomycin, neomycin sulfate, and metronidazole)-treated huREG3γ^tgIEC mice were remarkably deceased (not shown), implying that gut microbes are involved in the increased macrophages in huREG3γ^tgIEC mice. Since the proportion of Lactobacillus was remarkably increased among the composition of gut microbiota in huREG3γ^tgIEC mice, we hypothesized that these increased Lactobacilli might be related to the increased macrophages. One Lactobacillus strain named as Lactobacillus NK318.1, which is close to Lactobacillus Taiwanesis isolate (Figures S4A,B in Supplementary Material), was very often detected in the isolated Lactobacillus clones using Lactobacillus selective medium (Rogosa SL selective medium) from the fresh feces of huREG3γ^tgIEC mice (35% in huREG3γ^tgIEC mice, whereas 5% in WT mice). 16S rRNA copies of Lactobacillus NK318.1 were also much more in the fresh feces of huREG3γ^tgIEC mice compared to control WT littermates (Figure S4C in Supplementary Material). Other Lactobacillus such as L. reuteri, Lactobacillus gasseri, Lactobacillus animalis, Lactobacillus rhamnosus, and Lactobacillus plantarum did not significantly change and were even decreased in huREG3γ^tgIEC mice (Figure S4C in Supplementary Material). While equal amounts of Lactobacillus NK318.1, Lactobacillus NK318.2 with high homology toL. animalis (Figures S4D,E in Supplementary Material), L. reuteri and positive control huREG3γ^tgIEC mouse feces were transplanted into antibiotics-treated mice by orally gavage (Figure S5A,B in Supplementary Material), increased MHCII(+)F4/80(+) and F4/80(+)IL-10(+) macrophages were only detected in the Lactobacillus NK318.1-transplanted mice, whereas Lactobacillus NK318.2 and L. reuteri did not cause similar changes (Figures S5C–E in Supplementary Material). IL-10 levels in ileum tissues and sera of in the Lactobacillus NK318.1-transplanted mice were higher compared to control untransplanted mice (Figure S5F in Supplementary Material). All of these suggest that Lactobacillus NK318.1 may be a bacterium, which may enlarge gut anti-inflammatory macrophage pools.

To determine whether anti-inflammatory macrophages are really related to REG3γ-associated Lactobacillus NK318.1, we next colonized GF mice with Lactobacillus NK318.1. After 2 weeks, Lactobacillus colonization were demonstrated through FISH analyses with a specific probe. Consistent with other data (9), non-transplanted control GF mice housed under separate but similar conditions had very few macrophages (Figures 2A,B). In contrast, Lactobacillus-transplant induced robust accumulation of F4/80 macrophages not only in the LP of ileum tissues but also in the PP and spleen (Figure 2C). Moreover, the effect of Lactobacillus is bacterial species specific. Lactobacillus NK318.1 but not Lactobacillus NK318.2 could cause accumulation of MHCII(+)F4/80(+) macrophages in the gut LP, PP, and also spleen of GF mice (Figures 2A,C). The proportion of IL-10(+) macrophages was remarkably increased in the LP, PP, and spleen of huREG3γ^tgIEC mice (Figures 2D–F; not shown). The levels of IL-10 were also higher in the gut tissues of the LP in Lactobacillus NK318.1-transplanted GF mice (Figure 2D). Increased IL-10 (+) macrophages in the ileum LP were also further confirmed by immunohistochemistry staining (Figure 2E). In addition, transplantation of Lactobacillus NK318.1 could reverse many intestinal immune abnormalities, including prolonging the length of small intestine and colon and promoting the establishment and mature of gut immune system such as remarkably larger PP and expanded immune cells in the Lactobacillus-transplanted mice, which are similar to SPF mice. Taken together, our data show that REG3γ-associated Lactobacillus may enlarge MHCII (+)F4/80(+) macrophage pools not only in the LP but also in the PP and spleen.

To found a mechanism for how Lactobacillus causes increased macrophages, we first compared the gene expression profiles in the Payer’s patch (aggregated lymphoid nodules in gut LP) in GF mice with or without Lactobacillus colonization. Lactobacillus NK318.1 colonization in GF mice might induce at least a twofold change in expression of multiple gene expression (Figure 3A). KEGG analyses on those higher expression genes, which are induced by Lactobacillus, identified several critical signaling pathways, including Janus-activated kinase (JAK)-STAT signaling pathway, Toll-like receptor signaling pathway, NF-κB signaling pathway, TNF signaling pathway, intestinal immune network for IgA production, and cytokine–cytokine receptor interaction (Figure 3B). Notably, this strain of Lactobacillus did not affect immune-associated signaling pathways in gut epithelial cells (Figure 3C), indicating that Lactobacillus NK318.1-mediated accumulation of macrophages were not through gut epithelial cells. Activated STAT3 were further detected in the freshly gut tissues of Lactobacillus colonized GF mice through immunoblotting but not in GF mice without Lactobacillus colonization (Figure 3D). These data suggest that STAT3 may be involved in the Lactobacillus-mediated accumulation of macrophages in gut tissues.

Since Lactobacillus belongs to Gram-positive bacteria, we also examined whether TLR2 was involved in the accumulation of Lactobacillus NK318.1-mediated macrophages. Heated-dead Lactobacillus NK318.1 may induce IL-10 expression in the freshly isolated WT macrophages but not in TLR2^−/− macrophages (Figure 4A). These dead bacteria might not only activate multiple signal pathways such as NF-κB, JAK, and ERK but also activated STAT3, whereas these signal molecules in TLR2^−/− macrophages were not or weakly activated (Figure 4B). In vivo, Lactobacillus NK318.1 gavage caused the accumulation of macrophages in WT mice but not in TLR2^−/− mice (Figure 4C). The levels of IL-10 in gut tissues of WT but not in TLR2^−/− mice were significantly higher after Lactobacillus gavage (Figure 4D). All of these suggest that TLR2 is involved in Lactobacillus-induced macrophages in huREG3γ^tgIEC mice and meanwhile also indicate that Lactobacillus NK318.1 may directly activate STAT3 signal pathway.

We next further determine STAT3 contribution in gut anti-inflammatory macrophages. STAT3 is regarded as one of the primary signaling molecules for macrophage polarization (30, 31). In vitro experiment revealed that STAT3 inhibitor impeded the Lactobacillus-mediated IL-10 expression (not shown), indicating that Lactobacillus NK318.1 may affect the differentiation of gut macrophages. We next employed STAT3 inhibitor (JSI-124) to detect STAT3 effects on the macrophages in vivo. Injection of STAT3 inhibitor impeded Lactobacillus NK318.1-mediated increased macrophages. The proportion and total number of macrophages in the LP significantly decreased after injecting STAT3 inhibitor (Figures 5A,B). The proportion of the subsets such as MHCII(+)F4/80(+)CD11C(+)CD206(+), MHCII(+)F4/80(+)CD11C(+)Gr-1(+), and MHCII(+)F4/80(+)CD11C(+)IL-10(+) macrophages not only decreased in the LP but also in the PP and spleen after injecting STAT3 inhibitor (Figures 5C–E). Meanwhile, STAT3 activation was markedly reduced in the gut tissues of STAT3 inhibitor injected mice (Figure 5F). IL-10 levels in the ileum tissues of STAT3 inhibitor injected mice were also lower (Figure 5G). In addition, STAT3 inhibitor affects not only Lactobacillus-mediated macrophages but also the macrophages in WT and huREG3γ^tgIEC mice. Taken together, we demonstrate that STAT3 plays a critical role in REG3γ-associated Lactobacillus-mediated accumulation of anti-inflammatory macrophages.

The accumulated macrophages in huREG3γ^tgIEC mice may cause reduced inflammation, which is often related to HFD-mediated obesity (2, 32, 33). We next employed HFD-mediated obesity model to detect the effects of accumulated macrophages in huREG3γ^tgIEC mice on the obesity. We found that huREG3γ^tgIEC mice had a remarkable resistance to HFD-mediated obesity and reduced sensitivity to insulin and glucoses compared to WT littermate mice (Figures 6A–E). Lactobacillus NK318.1 gavages also impeded HFD-mediated obesity and reduced sensitivity to insulin and glucoses (Figures 6F–I). Interestingly, IL-10(+)F4/80(+) macrophages were much more in the adipose tissues of huREG3γ^tgIEC mice than in the those of control WT littermate mice fed HFD (Figure 6J). Similar phenomenon was also found in Lactobacillus infused mice (Figure 6K). The macrophages in the adipose tissues of huREG3γ^tgIEC mice or Lactobacillus infused mice also exhibited the anti-inflammatory phenotype (Figure 6L), which evenly distribute in adipose tissues (34). Because of critical role of anti-inflammatory macrophages in impeding obesity (34, 35), the resistance of huREG3γ^tgIEC mice and Lactobacillus infused mice to HFD-mediated obesity may be related to accumulation of these anti-inflammatory macrophages.

Next question is whether the accumulation of anti-inflammatory macrophages in adipose tissues is related to REG3γ-associated Lactobacillus. Previous data showed that gut immune cells possess broad trafficking abilities to lymphoid organs such as the inguinal lymph nodes and spleen (29). We next tested whether Lactobacillus-mediated macrophages might migrate into adipose tissues. To demonstrate this, we employed the β-Actin-luc LPTA^® animal model to examine the movement of Lactobacillus NK318.1-mediated macrophages. The isolated macrophages from the LP tissues of β-Actin-luc LPTA mice with Lactobacillus NK318.1 orally gavage were transplanted into syngeneic recipients. Bioluminescent foci were detected as early as 12 h after transplantation and the most frequently observed locations were at anatomic sites corresponding to the locations of the spleen and gut tissues (Figure 7A). However, these gut immune tissue-derived macrophages may migrate into pat-pad adipose tissues (Figure 7A). While isolated CD45.1(+) macrophages from gut LP tissues were injected into congenic CD45.2(+) WT mice, CD45.1(+) macrophages could move into the pat-pad adipose tissues of mice (Figures 7B,C). Infusion of isolated macrophages from Lactobacillus-infused mice could impede the HFD-mediated obesity and reduced sensitivity to insulin and glucoses (Figures 7D–G). Finally, we also investigated the proportion of Lactobacillus NK318.1 in obese and thin individuals. We found that the copies of Lactobacillus NK318.1 in thin individuals were remarkably increased, whereas the copies of Lactobacillus NK318.1 in obese individuals were less compared to normal individuals (Figure S6 in Supplementary Material). Taken together, our data suggest that the Lactobacillus NK318.1-associated macrophages are involved in the occurrence and development of obesity.