All animal protocols were reviewed and approved by the Animal Care and Use Committee of Sichuan Agricultural University (SICAU-2020-280, 12 May 2020). All animal studies were performed in accordance with the Use and Care of Laboratory Animals. Pla2g10 ^loxp+/− mice were constructed with C57BL/6 background Embryonic stem cells ( Figure 1A ). C57BL/6J-background tomoxifen-dependent Lgr5 ^cre+ mice (T003768) were purchased from GemPharmatech LLC., China. Pla2g10 knock-in (KI, Pla2g10 ^loxp+/− Lgr5 ^cre+/−, n = 6) mice were generated by mating Pla2g10 ^loxp+/− mice with Lgr5 ^cre+ mice, and Pla2g10 ^loxp+/− mice (n = 6) were used as the control. All the mice were housed under the individual ventilated caging system, maintained in a 12-h light/12-h dark environment, with a room temperature maintained of 22°C and humidity of 60%, with free access to food and water.

Mice with the same genotype were housed together, with 2 mice per cage. At an age of 4 months, the mice were weighted and injected with 100 mg/kg tamoxifen (#S1238, Selleck, Shanghai, China) once daily for a week to induce Cre expression. After injection, their body weight were recorded weekly for 4 weeks, and fed blood glucose levels were measured twice weekly, and food intake were recorded daily for 4 days. 4 weeks after tamoxifen injection, mice were euthanized with CO₂. The colons were collected, and the chyme in the colons was collected for microbial DNA extraction.

The initial weight of food in each cage on the first day was recorded at 5 pm, while the weight of the remaining food was measured at 5 pm on the following days. The food intake was calculated as (initial food weight - remaining food weight)/2.

Total RNA was extracted from tissues using RNAiso Plus (#9109, TaKaRa Bio, Dalian, China) following the manufacturer’s instructions. 1 μg of total RNA was used to synthesize cDNA using the PrimeScript FAST RT reagent Kit with gDNA Eraser (#RR092A, TaKaRa Bio). The cDNA was then subjected to qPCR using the iTaq Universal SYBR Green Supermix (#1725124, Bio-Rad, Shanghai, China) with a 384-well real-time PCR machine (QuantStudio 5, Thermo Fisher, Shanghai, China). The following primers were used: Gapdh F: 5’-AGGTCGGTGTGAACGGATTTG-3′, R: 5′-TGTAGACCATGTAGTTGAGGTCA-3′; Pla2g10 F: 5’-GTGCAGGTGTGACGAGGAG-3′, R:5′-CACTTGGGAGAGTCCTTCTCA-3′.

After homogenized in Western ReProbe (TM) PLUS (#C500035, Sangon Biotech, Shanghai, China) supplemented with PMSF (#abs812852, Absin, Shanghai, China), the protein samples were centrifuged at 12,000 rpm for 30 min. The protein lysates were quantified using a BCA protein assay kit (#23227, Thermo Fisher, Shanghai, China). Protein samples with equivalent concentrations were separated by SDS-PAGE. The following antibodies were used: rabbit anti-GAPDH Ab (#60004-1-lg, Proteintech, Wuhan, China) and mouse anti-PLA2G10 Ab (#sc-514324, Santa cruz, Shanghai, China).

Genomic DNA (gDNA) was extracted from the colonic chyme using a soil genomic DNA extraction kit (#DP336, Tiangen biotech, Beijing, China). The gDNA was used as a template to amplify the highly-variable V3–V4 region of the 16S RNA gene by PCR, using the forward primer 341F (5’-ACTCCTACGGGAGGCAGCAG-3’) and the reverse primer 806R (5’-GGACTACHVGGGTWTCT AAT-3’). All PCRs were performed in 15μL of Phusion High-Fidelity PCR Master Mix (New England Biolabs, Manchester, UK), with 0.2 μM of both the forward and reverse primers, and approximately 10 ng of template DNA. Thermal cycling consisted of an initial denaturation at 98°C for 1 minute, followed by 30 cycles of denaturation at 98°C for 10s, annealing at 50°C for 30s, and elongation at 72°C for 30s, and finally an extension at 72°C for 5 min.

The PCR products were purified using magnetic beads. The samples were mixed in equal-density ratios based on the concentrations of the PCR products. After the PCR products were thoroughly mixed, they were detected and purified using a universal DNA purification and recovery kit (#DP214-03, Tiangen Biotech, Beijing, China) to isolate the nucleic acids. The nucleic acids were subjected to construct libraries using a library construction kit (#E7645S, New England BioLabs, Manchester, UK). The libraries were sequenced using the Illumina NovaSeq 6000 (PE250) system.

Paired-end reads were merged using the FLASH (v1.2.11) to obtain raw tags (Magoč and Salzberg, 2011). The Cutadapt software (v3.3) was then used to match the reverse primer sequence and trim the remaining sequences to prevent their interference in subsequent analyses. Quality filtering of the raw tags was performed using the fastp software (v0.23.1) to obtain high-quality clean tags (Bokulich et al., 2013). To detect chimeric sequences, the tags were compared with the reference database (Silva database (16S/18S), and the effective tags were obtained by removing the chimeric sequences using the vsearch package (v2.16.0) (Edgar et al., 2011). To obtain the initial amplicon sequence variants (ASVs), the DATA2 program in the QIIME2 software (v202202) was performed to denoise the previous effective tags (Y. Wang et al., 2021). The species annotation was performed with QIIME2 software.

The absolute abundance of ASVs was normalized using a standard of sequence number corresponding to the sample with the fewest sequences. The subsequent analyses of α-diversity and β-diversity were all performed based on the output normalized data. The top 10 taxa at the phylum level and top 22 taxa at the genus level of each sample were selected to plot the distribution histogram of relative abundance using Perl software (v5.26.2). The Venn diagram was produced in R software (v4.0.3) using the Venn Diagram function, and the α-diversity was calculated using Chao1 and Shannon indices in QIIME2 software to analyze the diversity, richness and uniformity. The β-diversity was calculated by the unweighted Unifrac distance using QIIME2 software, and was visualized by principal coordinate analysis (PCoA) and non-metric multidimensional scaling (NMDS) analysis using R software (v4.0.3). The Linear Discriminant Analysis Effect Size (LEfSe) was applied to analyze the differences at each level using R software (v4.0.3). The functional genes and Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologs were predicted by the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) (v2.3.0). The network plot was produced using Chiplot (https://www.chiplot.online/).

Data are presented as mean ± standard error of the mean (SEM). Growth data were preliminarily analyzed using Microsoft Excel 2019 and then analyzed statistically using the SPSS 21.0 software. Normality of the data was evaluated by Shapiro-Wilk test using the SPSS 21.0 software. Microbial data were organized and analyzed statistically using the R software (v4.0.3). Data were visualized using the GraphPad Prism (v8.0) and R software (v4.0.3) The P - values used to compare two groups were calculated with Student’s t-test. A trend was deemed present when 0.05 < P < 0.1; the significance of differences was defined as *P < 0.05.

To explore the effects of intestinal Pla2g10 overexpression on the homeostasis of gut microbiota, Pla2g10 ^loxp+/− mice were mated with tamoxifen-dependent Lgr5 ^cre+ mice to generate intestine-specific Pla2g10 knock-in (KI) mice. Both the gene expression and protein levels of PLA2G10 were significantly higher in the colons of the Pla2g10 KI mice than in those of the control mice ( Figures 1B, C ). In the subsequent 4 weeks after CRE induced, the body weights and glucose levels did not differ significantly between the two groups ( Figures 2A, B ). Although there was no significant difference in the food intake of the two groups, the Pla2g10 KI mice consistently consumed 1 g more than the control mice ( Figure 2C ). A PCoA with an unweighted UniFrac distance algorithm revealed differences in the microbial compositions of the two groups ( Figure 3A ). The first principal axis (PCoA1) could explain 30.44% of the sample difference, and the second principal axis (PCoA2) could explained 18.45% of the sample difference. The NMDS stress value was 0.039, indicating that the subsequent analytical results accurately reflected the differences between the samples with high confidence ( Figure 3B ). We assessed the α-diversity of the gut microbiota by calculating the Chao1 and Shannon indices, both of which were significantly lower in the Pla2g10 KI mice than in the control group ( Figures 3C, D ), indicating that the total number of gut bacteria and their richness and evenness were reduced after the overexpression of Pla2g10. A Venn diagram confirmed this, showing that the number of operational taxonomic units (OTUs) was lower in the Pla2g10 KI mice than in the control group ( Supplementary Figure 1A ).

Analysis of the community composition and species differences between the Pla2g10 KI and control mice showed that the overexpression of Pla2g10 altered the composition of the gut microbiota at the phylum ( Figure 4A ) and genus levels ( Figure 4B ). The detailed distributions of the gut microbiotal compositions of the two groups are shown in Supplementary Table 1 . The abundance of Actinobacteria in the colonic chyme of the Pla2g10 KI group had an upward trend compared with the control group, whereas the abundance of Bacteroidetes was significantly lower ( Figure 5A ), and the Firmicutes/Bacteroidetes (F/B) ratio also showed an increasing trend in the Pla2g10 KI mice ( Supplementary Figure 1B ). At the genus level, the relative abundances of Dubosiella (P = 0.020), Enterorhabdus (P = 0.030), Lactococcus (P = 0.033) and Lactobacillus (P = 0.045) were significantly elevated in the Pla2g10 KI mice compared to control group ( Figure 5B ), while Faecalibaculum only showed an increasing trend (P = 0.066). One genus, Allobaculum, showed a significant decline in abundance in the gut microbiota of the Pla2g10 KI mice compared with control mice. Differences in bacteria at other levels between the Pla2g10 KI and control groups are shown in Supplementary Figures 1C–E .

To determine the effect of PLA2G10 on the gut microbial community, a linear discriminant analysis (LDA) and its effect size (LEfSe) analysis were performed to identify the biomarkers in the microbiomes of the two groups ( Figure 6A and Supplementary Figure 2A ). A logarithmic LDA cutoff score of 3.0 was set to identify important taxonomic differences between the two groups. We then calculated the scores that reflected the effect size of each differentially enriched taxon from these results. The LDA value histogram for the samples at the genus level ( Figure 6B ) showed that the gut microbiota of the control group contained more Allobaculum and Desulfovibrio than that of the Pla2g10 KI group, whereas Dubosiella and Enterorhabdus were dominant in the Pla2g10 KI group. Taken together, these results suggest that Pla2g10 overexpression altered the abundances of the specific gut bacteria.

To further investigate the changes in the metabolic functions of the gut microbiota associated with Pla2g10 overexpression, we performed a functional prediction analysis based on the KEGG orthology (KO) database, using the PICRUSt2 software. Combination of the PICRUSt2 functional clustering heat map ( Figure 7A ) with a t test analysis ( Supplementary Figure 2B ) clearly showed that Pla2g10 overexpression significantly altered various functions of the gut microbiota in mice, and most of the differentially expressed functions were more abundant in the Pla2g10 KI mice than in the control mice, as shown in Supplementary Table 2 . We then constructed a network plot of the 30 most differentially abundant functions identified in the KEGG analysis and the 10 most abundant genera in both groups of mice. The results revealed that the main genera whose functions were affected by Pla2g10 overexpression in mice were Dubosiella, Lactobacillus, X Ruminococcus_torques_group, and Allobaculum ( Figure 7B ).