To investigate the effects of SG on HFD-feeding induced colonic pro-inflammation, twenty 4-week-old male C57BL/6J mice were purchased from Charles River (Beijing, China) and allowed to acclimate in the laboratory for two weeks prior to the start of HFD intake. Before surgery, all mice were fed 60% HFD (Research Diets D12492, New Brunswick, New Jersey, USA.) for twelve weeks. Post that, the HFD-induced obese mice were randomized based on body weight to receive either SG (n=11) or sham surgery (SHAM, n=9). All mice were maintained on the same HFD for eight weeks following surgery until euthanasia. One mouse died and three mice suffered abdominal abscess after SG and were excluded.

To study the role of the gut microbiota in the improvements of HFD-feeding related colonic pro-inflammation following SG, another twenty 4-week-old male C57BL/6J mice (Charles River, Beijing, China) were used. All mice were fed 60% HFD (Research Diets D12492, New Brunswick, New Jersey, USA.) for twelve weeks starting at 6 weeks of age. Two days prior to surgery, all mice were provided with broad-spectrum antibiotics cocktails added in the drinking water to delete most of the intestinal bacteria (26). Then, these mice were randomized based on body weight to undergo either SG (SG-ABX, n=11) or sham surgery (SHAM-ABX, n=9). All mice were maintained on the same HFD and antibiotics cocktails for eight weeks following surgery until euthanasia. One mouse died and two mice suffered abdominal abscess after SG and were excluded.

All mice were housed under specific pathogen-free conditions at an ambient temperature with a 12-12 light-dark cycle and had ad libitum access to food and water. All animal experiments were conducted in accordance with National Research Council Guide for Care and Use of Laboratory Animals and approved by the Department of Laboratory Animal Science Fudan University.

The broad-spectrum antibiotics cocktails comprises four types of antibiotics purchased from Sigma Aldrich (Shanghai, China), namely neomycin trisulfate (#N6386), metronidazole (#M1547), vancomycin hydrochloride (#V2002) and ampicillin (#A9518). The antibiotics cocktails were freshly prepared by adding the antibiotics powders into drinking water, reaching a concentration of 1 g/L for neomycin trisulfate, 0.25 g/L for metronidazole, 0.5 g/L for vancomycin hydrochloride and 1 g/L for ampicillin (26). The antibiotics-containing drinking water was stored in the light-protected bottles and changed three times weekly.

SG and sham procedures were performed as previously described (21). Mice were anesthetized by intraperitoneal injection of pentobarbital sodium (50 mg/kg; Sigma, Shanghai, China). The stomach was gently exposed after dissection of the gastrosplenic ligaments. For SG, the glandular stomach was closed at 4 mm proximal of the pylorus toward the fundus using a 5-mm titanium clip (Ethicon, Somerville, NJ). After that, 80% glandular stomach and entire non-glandular stomach were excised along the outside of the clip, leaving a tubular gastric remnant. The gastric remnant with the clip was then enhanced using interrupted 8-0 Prolene sutures. For sham surgery, gentle pressure was applied on the stomach with nontoothed blunt forceps. Immediately after surgery, mice were placed on a heat mat and subcutaneously administered 1 ml warm 5% Glucose and Sodium Chloride Injection. Mice were fasted for food on the day of surgery and resumed HFD one day after surgery. Body weight following surgery was monitored daily for the first week and then weekly.

Mixed-meal tolerance test (MMTT) was performed 6 weeks following surgery. All mice were fasted for 4 hours before oral gavage of liquid meal (volume 200 ml Ensure Plus spiked with a 25 mg dextrose). After that, tail vein blood glucose levels were measured using glucometers (Contour TS, Shanghai, China) at 0, 15, 30, 45, 60, 90, and 120 minutes.

Blood samples were collected when mice were euthanized after 4-hour fasting. Serum levels of triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) were measured using automatic biochemical analyser (Siemens Healthcare Diagnostics Inc, ADVIA XPT, USA) according to the manufacturer’s instructions.

Total RNA was extracted from colon tissues using TRIzol (BioTNT, Shanghai, China) and then reversed into Complementary DNA by PrimeScript RT kit (Takara RR047, Beijing, China). PCR was performed using the TB Green Premix (Takara RR420, Beijing, China) on a QuantStudio 6 (ThermoFisher) system. Primers of target genes were purchased from Integrated DNA Technologies (Sangon Biotech, Shanghai, China) and verified by melting curve analysis. The expression levels of target genes were normalized to β-actin gene and calculated using the 2–ΔΔCT method.

The colon segments were fixed in 4% buffered formalin for 48 hours prior to paraffin embedding and hematoxylin and eosin (H&E) staining. Crypt depth of colon of each mouse was measured in three different fields under x 400 high power field (HPF) by the software (K-Viewer 1.5.5.2, China). For evaluation of the macrophage infiltration in the colon, CD68 staining was performed using a rabbit anti-mouse CD68 primary antibody (Abcam ab283654, Shanghai, China) and secondary antibody (Jackson, Philadelphia, USA), and further counted in three different sections under HPF by two blinded observers.

The macrophages from colonic lamina propria were isolated by lamina propria dissociation kit (Miltenyi Biotec mouse130-097-410, Shanghai, China) according to the manufacturer’s instructions. Briefly, intestinal fat was removed, and colon was cut open and washed slowly in PBS (Ca²⁺ and Mg²⁺ free). The colon segments were cut into 1 mm pieces and added into Enzyme D, Enzyme R, and Enzyme A for incubation at 37°C for 45 minutes. The supernatant was then filtered through 75 um nylon mesh and centrifuged at 500g for with 10 minutes. After centrifuge, the pellets were collected and re-suspended in 40% Percoll (Solarbio, Beijing, China). The 40% Percoll solution with suspended cells was transferred into 80% Percoll and re-centrifuged at 2000 rpm for 20 minutes at room temperature. The white interphase was collected after centrifuge and washed twice with PBS. Before flow cytometric analysis, single-cell suspensions isolated from the colon were further stained for 30 minutes on ice with fluorophore-conjugated commercial antibodies to F4/80 (Invitrogen 11-4801-82, USA), CD11b (Invitrogen 12-0112-82, USA) and CD11c (Invitrogen 17-0114-81, USA). After preparation, those cells were re-suspended in PBS with 0.5% FBS and analyzed using FACSAriaIII (BD Bioscience, USA). The data were analyzed by FlowJo software (Becton, Dickinson and Company, USA).

Fecal and cecal samples were collected when mice were euthanized, and immediately frozen at -80°C upon collection. Total genomic DNA was extracted from samples using the OMEGA Soil DNA Kit (M5635-02) (Omega Bio-Tek, Norcross, GA, USA), following manufacturer’s instructions. Then the DNA was stored at -20°C prior to further analysis. PCR amplicons were purified with Vazyme VAHTSTM DNA Clean Beads (Vazyme, Nanjing, China) and quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). After quantification, amplicons were pooled in equal amounts. Pair-end 2 x 250 bp sequencing was performed using the Illlumina MiSeq platform with MiSeq Reagent Kit v3 at Shanghai Personal Biotechnology Co., Ltd (Shanghai, China). Sequencing data analyses were performed using QIIME2. Briefly, raw sequences after trimming were analyzed by the cutadapt plugin and the dada2 plugin. After that, non-singleton amplicon sequence variants (ASVs, 100% operational taxonomic units (OTUs)) were generated. Microbial taxonomic classification was performed to ASVs according to the classify-sklearn alignment algorithm (27) against the Greengenes database (Release 13.8) of 99% OTUs reference sequences (28). Alpha diversity metrics including Chao1 and Shannon calculated by the diversity plugin were performed to estimate richness and diversity respectively. Beta diversity metrics including unweighted UniFrac distance matrix were scaled and visualized through principal coordinates analysis (PCoA), and significance of the clustering between groups was determined via permutational multivariate analysis of variance (PERMANOVA). Random Forest Classifier with 10-fold cross-validations, MetagenomeSeq analysis, and Linear discriminant analysis (LDA) effect size (LEfSe) with default parameters were computed to identify significantly different microbes in abundance between groups at different taxonomic levels.

Total RNA was isolated from colon tissues using Trizol Reagent (Invitrogen Life Technologies, USA) and sequenced on NovaSeq 6000 platform (Illumina, USA) by Shanghai Personal Biotechnology Cp. Ltd. R language Pheatmap (1.0.8, China) software package was used to perform bi-directional clustering analysis to identify all differentially expressed genes (DEGs) between two surgical groups. The top Gene Ontology (GO) was used to perform GO enrichment analysis based on the DEGs. P-value was calculated by hypergeometric distribution method (the standard of significant enrichment is P <0.05). Cluster Profiler (3.4.4) software was used to carry out the enrichment analysis of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway of DEGs, focusing on the significant enrichment pathway with P <0.05.

Data were presented as mean ± SEM. Differences in body weight and blood glucose levels during MMTT between two surgical groups were evaluated using two-way analysis of variance (ANOVA) with post-hoc Sidak test for multiple comparisons (29). Other simple comparisons between two surgical groups were assessed with Student’s t-test or non-parametric Mann–Whitney U tests. In addition, comparisons among four surgical groups were assessed using two-way analysis of variance (ANOVA) with post-hoc Tukey’s test for multiple comparisons. All statistical analyses were conducted using GraphPad Prism 8 software (La Jolla, CA). Data were considered statistically significant when P < 0.05 (2-sided significance testing).

Twenty 4-week-old male C57BL/6J mice were fed 60% HFD for 12 weeks and then randomized based on body weight to receiving either SG or SHAM operation. Mice were kept on the same 60% HFD following surgery until euthanasia ( Figure 1A ). SG led to significant weight loss as compared to SHAM operation ( Figure 1B ). SG mice also displayed significantly lower glucose levels during mixed meal tolerance test (MMTT) ( Figure 1C ), suggesting an improved glucose tolerance following SG. Besides, SG-treated mice had significantly decreased levels of TC, LDL-C and HDL-C ( Figure 1D ).

In terms of colonic morphology, there was no difference in the length of colon and weight of cecum between SG and SHAM groups ( Figures 1E, F ). However, H & E staining indicated a decreased depth of colonic crypts of SG mice as compared to SHAM mice ( Figure 1G ). To determine the effects of SG on HFD feeding associated colonic pro-inflammation, we measured mRNA expression levels of different cytokines in the colon using qPCR. The mRNA expression levels of pro-inflammatory cytokines such as IL-6, IL-1β, IL-18, and IL-23 were downregulated in the colon of SG mice relative to SHAM mice, while those anti-inflammatory factors such as IL-10, TNF-β, IL-33 were not impacted by SG ( Figure 1H ). Given that HFD feeding associated colonic pro-inflammation induces gut barrier defects (30), we measured mRNA expression levels of two common intestinal tight junction proteins zonula occludens 1 (ZO-1) and Occludin. SG increased relative expression levels of ZO-1 and Occludin in the colon as compared to SHAM operation ( Figure 1I ), implicating a possible improvement in the colonic barrier following SG. Collectively, these data demonstrate that SG results in improvements in the HFD feeding related colonic pro-inflammation.

Literature has recently proposed an important role of macrophage infiltration in the HFD feeding associated colonic pro-inflammation (31). We therefore measured mRNA expression levels of macrophage-related chemokine genes including CCL2, CCL7 and CCL12 and marker genes such as F4/80, CD68, CD11b and CD11c in the colon. All these genes except CD11c showed comparable expression levels between SG and SHAM groups ( Figures 2A, B ). Likewise, immunohistochemistry tests revealed no difference in the numbers of CD68⁺ macrophages infiltrated in the colon epithelium between SG and SHAM mice ( Figure 2C ). This was further confirmed by flow cytometry analysis of colonic lamina propria where SG mice had similar numbers of F4/80⁺CD11b⁺CD11c^- sub-population to SHAM mice ( Figure 2D ). These results indicate that SG has no effect on the HFD-feeding induced macrophage infiltration in the colon.

Given the association of colonic pro-inflammation upon HFD intake with alterations to the gut microbiota (32), we next characterized the intestinal bacterial changes following SG. The fecal microbiota of SG mice displayed higher richness than that of SHAM mice, as estimated by higher levels of Chao 1 index ( Figures 3A, B ). For the overall composition of the gut microbiota, unweighted UniFrac PCoA revealed a differential clustering of fecal samples between SG and SHAM mice ( Figure 3C ). This was confirmed by PERMANOVA which showed significance in unweighted UniFrac distances of samples between SG and SHAM groups (F value = 1.85, P = 0.003), suggesting an alteration in the overall composition of the gut microbiota following SG. In terms of the detailed compositions of the gut microbiota, Firmicutes was the major phylum in the fecal microbial communities of SHAM and SG mice, while Desulfovibrio subspecies showed a decreasing trend in the abundance following SG ( Supplementary Figure 1A ). Random Forest classifier was used to identify differentially enriched microbes between SHAM and SG groups. It indicated that bacterial subspecies belonging to Lactobacillus genus were the main discriminators for SG compared to SHAM gut microbiota, with substantial rises in the relative richness following SG ( Figure 3D ). This was consistent with results generated from MetagenomeSeq analysis showing that multiple bacteria under Lactobacillales order were abundant following SG relative to SHAM operation ( Figure 3E ). Additional LDA effect size (LEfSe) analysis revealed more taxonomic differences in the microbial composition between SHAM and SG groups ( Supplementary Figure 1D ). For instance, SG microbiota were enriched for Blautia subspecies, whereas SHAM microbiota were enriched for Desulfovibrio subspecies. Taken together, these data indicate that SG leads to significant alterations of the gut microbiota, featured by decreases of pro-inflammatory microbes such as Desulfovibrio and increases of anti-inflammatory microbes like Lactobacillus. This suggests a potential association of changed gut microbiota with decreased colonic pro-inflammation following SG.

We next sought to determine whether the improvements of colonic pro-inflammation following SG were dependent on the alterations of the gut microbiota. Another cohort of twenty 4-week-old male C57BL/6 mice were fed 60% HFD for 12 weeks and then randomized based on body weight to receiving either SG (SG-ABX) or SHAM (SHAM-ABX) operation. All mice were kept on the same 60% HFD following surgery until euthanasia. They were also provided with broad-spectrum antibiotics in the drinking water to delete most intestinal bacteria (26) starting from 2 days before surgery till the end of the study ( Figure 4A ).

To confirm the efficacy of antibiotics cocktails in the gut microbiota deletion, 16s rRNA sequencing was used to characterize the bacterial changes in the cecum contents of SG and SHAM mice that received antibiotics treatment. Chao1 and Shannon index indicated extremely low levels of richness and diversity of the gut microbiota in the antibiotics-treated groups ( Supplementary Figures 2A, B ). This was consistent with compositions of the microbial community where Proteobacteria was the main phylum left in the gut, accounting for 90% of the entire bacterial abundance following antibiotics administration ( Supplementary Figure 2D ). At species level, there were fewer microbes that had much lower richness in the cecum contents ( Supplementary Figure 2E ). These data suggest that the broad-spectrum antibiotics treatment successfully deleted most microbes in the cecum. In this situation, no differences were observed between SG and SHAM mice in the microbial richness and diversity, as estimated by Chao1 and Shannon index, respectively ( Supplementary Figures 2A, B ). Likewise, no bacterial taxa were identified as varying significantly in the relative abundance between SG-treated and sham-operated mice using Random Forest classifier, MetagenomeSeq analysis or LEfSe (data not shown given the extremely low richness of the microbes detected). For the overall composition of the gut microbiota, unweighted UniFrac PCoA did not show pronouncedly differential clustering of samples between SG-ABX and SHAM-ABX groups (PERMANOVA F value =0.84, P =0.614) ( Supplementary Figure 2C ). Together, these data reveal no marked alterations of the gut microbiota following SG relative to SHAM operation in the context of antibiotics treatment.

Although antibiotics cocktails eliminated most bacteria in the gut, SG still led to significant weight loss, improved glucose tolerance and decreased serum levels of TC and LDL-C ( Figures 4B–D ). Interestingly, weight loss, glucose tolerance and serum lipid levels were all comparable between SG and SG-ABX groups ( Supplementary Figures 3A–F ). For the histological features of HFD feeding associated colonic pro-inflammation, no difference was observed in the crypt depth between the two surgical groups under antibiotics treatment ( Figure 4E ). However, unlike what was shown in the Figure 1 , SG had no effect on the expressions of various inflammation cytokines in the colon in the absence of the gut microbiota changes, as the pro-inflammatory cytokines including IL-6, IL-1β, IL-18, and IL-23 displayed comparable mRNA expression levels between antibiotics-treated SG and SHAM groups ( Figure 4F ; Supplementary Figure 4A ). Likewise, there were no differences in the relative expression levels of ZO-1 and Occludin genes in the colon between antibiotics-treated SG and SHAM mice ( Figure 4I ; Supplementary Figure 4B ), implicating no improvement in the gut barrier integrity following SG when there were no intestinal bacterial changes resulting from oral administration of antibiotics. On the other hand, whereas antibiotics-treated animals had reduced colonic macrophage infiltration, SG-ABX and SHAM-ABX groups showed comparable levels of macrophage infiltration and related genes expressions in the colon ( Figures 4G, H , Supplementary Figures 4C, D ). Altogether, these results indicate an important role of the gut microbiota in the improvements of HFD related colonic pro-inflammation following SG.

To further probe the relationship between improvements of colonic pro-inflammation and the gut microbiota changes following SG at transcriptional level, we conducted RNA sequencing analysis of colon tissues obtained from both SG and SHAM mice with and without antibiotics treatment. For mice that did not receive antibiotics cocktails, volcano plots illustrated 179 differentially expressed genes (DEGs) in the colon between SG and SHAM groups ( Figure 5A ). Both Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses of these DEGs showed a significant enrichment of multiple pathways that were related to inflammation regulation in the colon including inflammatory response, leukocyte migration and chemotaxis as well as PPAR signaling pathway ( Figure 5B ; Supplementary Figure 5A ). On the other hand, for mice that received antibiotics cocktails, volcano plots identified only 73 DEGs between SG and SHAM groups. Neither GO nor KEGG pathway analyses of these DEGs showed an enrichment of inflammation-associated pathways ( Figures 5C, D ; Supplementary Figure 5B ). Taken together, these findings demonstrate that SG exerts transcriptional modifications of inflammatory pathways in the colon in a manner that was gut microbiota relevant.