C57BL/6J mice were procured from Chengdu Dossy Experimental Animals Co., Ltd. The mice were housed in individual ventilated cages with unlimited access to sterilized food and water. The animal studies were approved by the Animal Research Ethics Committee of the West China Hospital in Sichuan University (2017105A).

Eight-week-old male mice received a single intraperitoneal injection of glyoxylate (80 mg/kg body weight; Sigma-Aldrich, Shanghai, China) for 7 consecutive days (CaOx group) (25, 26). The acetate (C2), propionate (C3), and butyrate (C4) groups received intraperitoneal injections of glyoxylate (80 mg/kg body weight) with 150 mM sodium acetate, sodium propionate, or sodium butyrate, respectively. To deplete the gut microbiota, antibiotic cocktails (Abx), combined with ampicillin (0.5 g/L), vancomycin (0.25 g/L), neomycin (0.5 g/L), and metronidazole (0.5 g/L), were added to the drinking water two weeks before glyoxylate and SCFA administration (Abx, Abx+C2, Abx+C3, and Abx+C4 groups, respectively; n = 8 mice/group). The blood, kidney, and cecal contents of the mice in each group were determined on day 8.

To knockdown GPR43 expression in vivo, mice were transduced with an AAV serotype 9 vector encoding a green fluorescent protein reporter together with either short hairpin RNAs (shRNAs) targeting GPR43 in the kidney (AAV-shGPR43) or an empty vector (AAV-null). The shRNA sequences targeting the mouse GPR43 gene were cloned into the AAV by Shanghai GeneChem Co., Ltd. (Shanghai, China). Mice were injected with either AAV-shGPR43 or AAV-null via the tail vein (1 × 10¹¹ viral particles/mouse), followed by an additional 4 weeks of glyoxylate injection. The mice were randomly divided into five groups and treated with the AAV-null vector (AAV-null group), the AAV-shGPR43 vector and glyoxylate injection (AAV-shGPR43 group), the AAV-shGPR43 vector with C2 in the drinking water (AAV-shGPR43+C2 group), the AAV-shGPR43 vector with C3 in the drinking water (AAV-shGPR43+C3 group), or AAV-shGPR43 vector with C4 in the drinking water (AAV-shGPR43+C4 group).

A panel of CyTOF antibodies combined with 42 metal isotope-tagged antibodies ( Supplemental Table 1 , Fluidigm, San Francisco, USA) was used to detect immune cell populations in mouse kidneys. The kidneys were placed in Dulbecco’s modified Eagles’ medium and cut into pieces with collagenase IV (1 mg/mL; Merck, Shanghai, China) and DNase I (100 μg/mL, Merck) at 37°C for 30 min (27). The digested kidney cells were resuspended in phosphate-buffered saline (Thermo Scientific, Shanghai, China) containing 0.02% NaN₃ and 0.5% bovine serum albumin and stained for viability with 194Pt (Fluidigm) for 5 min on ice. The cells were then blocked using Fc receptors, followed by cell-surface staining with metal-labeled monoclonal antibodies (mAbs) for 30 min. The cells were incubated overnight at 4°C with DNA Intercalator-Ir (Fluidigm) to discriminate singly nucleated cells from doublets. After fixation and permeabilization, the cells were intracellularly stained with metal-labeled mAbs for another 30 min. The cells were washed and prepared for data acquisition. The Helios system (Fluidigm Sciences) was used to acquire CyTOF data at ≤500 events/s.

All CyTOF files were analyzed using FlowJo software, version 10.5.3. (TreeStar, Ashland, OR, USA). The CD45-positive populations were clustered using the X-shift algorithm (28), which automatically pools cells to identify underlying immune cell subpopulations. The normalized expression levels of markers were visualized with a heatmap. Cell frequencies were calculated by dividing the total number of CD45⁺ cells events by the number of assigned cell events. The R software package was used to perform dimension reduction, based on t-distributed stochastic neighbor embedding (t-SNE) (29).

Kidney digestion is the same as described above. Renal cells were surface-stained with fluorophore-conjugated antibodies against the mouse CD45 antigen (FITC, BioLegend, San Diego, CA, USA), then renal CD45⁺ cells were sorted by Flow cytometry. The RNeasy Mini Kit (Qiagen, Germany) was used to extract total RNA from renal CD45⁺ cells following the manufacturer’s instructions. One microgram total RNA was used to prepare an RNA-seq transcriptome library using the TruSeq™ RNA Sample Preparation Kit (Illumina, San Diego, CA). Then, complementary DNA target fragments (200–300 base pairs) were selected, followed by polymerase chain reaction-based amplification with Phusion DNA polymerase (Thermo Scientific, Shanghai, China). An Illumina HiSeq X instrument was used for sequencing.

The raw data were trimmed and qualified using the SeqPrep and Sickle tools. Then, the clean data were aligned to the reference genome using TopHat software (version 2.0.0) (30), and the mapped reads were assembled using StringTie.

Transcript expression levels were calculated using the transcripts per million reads method to identify differentially expressed genes (DEGs) between groups. The RSEM (31) and EdgeR (32) software programs were utilized to quantify gene abundances and for differential expression analysis, respectively. Furthermore, Gene Ontology (GO) functional-enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed on DEGs with a Bonferroni-corrected P-value of ≤ 0.05.

Isolated kidney CD45⁺ cells were used to analyze immune subtypes. The cells were surface-stained with fluorophore-conjugated antibodies against the following mouse antigens: CD45 (FITC, APC), CD11b (PE), CD11c (FITC), F4/80 (PE/CY7), CX3CR1 (BV510), CD24 (APC), GR-1 (APC), I-A/I-E (APC/CY7), and CD103 (BV421) (BioLegend, San Diego, CA, USA). A FACSAria SORP flow cytometer (BD Biosciences, CA, USA) and FlowJo software were used to acquire and analyze the flow cytometric data, respectively.

Kidneys were fixed in paraformaldehyde, embedded in paraffin, and cut into 4 µm sections. Hematoxylin and eosin staining and von Kossa staining were performed using staining kits (Solarbio, China), according to the manufacturer’s protocols. For immunofluorescence, the sections or cells were incubated with an anti-F4/80/Alexa Fluor^® 647-conjugated antibody (1:100, BioLegend), an anti-GPR43/FITC-conjugated antibody (1:100, Bioss, Beijing, China), an anti-CX3CR1/PE-conjugated antibody (1:100, Novus, CO, USA), and an anti-GR-1/Alexa Fluor^® 594-conjugated antibody (1:100, BioLegend), and with DAPI (Merck) or Hoechst (Thermo Scientific, Shanghai, China) dye.

Mouse sera were collected, and IL-1β, IL-6, and TNF-α levels were measured using enzyme-linked immunosorbent assays (Invitrogen, Shanghai, China), according to the manufacturer’s recommendations.

An E.Z.N.A.^® Soil DNA Kit (Omega Bio-tek, Inc., GA, USA) was used to extract microbial DNA from cecal samples. According to our previous study, primers 338F (5’-ACTCCTACGGGAGGCAGCAG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’) were used to amplify the bacterial 16S rRNA genes (V3–V4 hypervariable regions) (18). Sequencing was performed by Majorbio Technology Co. Ltd. (Shanghai, China) using the Illumina MiSeq PE300 platform (Illumina).

The raw 16S rRNA sequence data were demultiplexed, quality-filtered, and merged using fastp software (version 0.20.0) (33) and FLASH software (version 1.2.7) (34). All sequences were clustered into different operational taxonomic units (OTUs) at 97% similarity using UPARSE software (version 7.1) (35) by removing the chimeric sequences. The taxonomy of the 16S rRNA amplicon sequences was analyzed using the RDP Classifier software (version 2.2) (36) against the Silva v138 database with a confidence threshold of 0.7.

Alpha diversity and coverage indices were calculated. Principal coordinates analysis (PCoA) and the Adonis test were used to evaluate beta diversity among the groups. Linear discriminant analysis effect size (LEfSe) analysis was used to identify differentially abundant bacteria among the three groups with a cutoff threshold of 3.0. KEGG metabolic pathways were predicted using the PICRUSt software package.

Mouse TCMK-1 kidney epithelial cells (3 × 10⁴) were treated with lipopolysaccharide (LPS, 1 µg/ml) in a 24-well plate and cocultured with or without acetate (C2), propionate (C3), butyrate (C4), or immune cells for 24 h. Then, CaOx crystals (100 µg/ml) were added to the coculture system for 10 min, followed by fixation. Crystal adhesion to the cell surface was observed using an inverted microscope.

CD45⁺F4/80⁺CD11b⁺CX3CR1⁺CD24^- macrophages and CD45⁺CD11b⁺GR1⁺ neutrophils were sorted by flow cytometry. Macrophages (with or without GPR43-antagonist treatment for 24 h) were cocultured with neutrophils for another 24 h. GPR43 expression was detected by immunofluorescence. Using a Transwell system, macrophages were cultured in the upper chamber with or without a GPR43 antagonist for 24 h. Treated macrophages were added to another upper chamber of a 24-well plate with neutrophils, and TCMK-1 cells were treated with LPS and CaOx crystals seeded in the lower chamber for 24 h. Neutrophil migration and GPR43 expression were also observed.

Statistical analyses were performed using SPSS. Data are shown as mean ± standard deviation (SD). All data were analyzed by one-way analysis of variance for multiple groups, and Student’s t-test was applied for comparisons between two groups. Statistically significant differences are shown as *p < 0.05 or **p < 0.01.

After 7 consecutive days of intraperitoneal glyoxylate injection, CaOx crystal deposits were found in mouse kidneys with elevated serum creatinine and urea levels ( Supplemental Figure 1 ). To explore the immunomodulatory role of SCFAs in CaOx stones, we first studied the phenotypic diversity of immune cell populations in kidneys with CaOx crystals by analyzing CD45⁺ cell populations clustered with the X-shift algorithm.

Unsupervised clustering of 383,985 single CD45⁺ kidney cells from six samples created a detailed t-SNE map of distinct cell populations ( Figure 1A ). The clusters were classified into 29 distinct immune cell types ( Figures 1A, B ), based on the expression levels of lineage markers. The normalized expression of clusters is shown in a heatmap ( Figure 1C ). We identified the following major cell subsets: CD4 T cells, CD8 T cells, B cells, macrophages, DCs, natural killer cells, monocytes, ILCs, and neutrophils ( Supplemental Table 2 ).

The frequencies of each cluster are shown in Fig. 2. The frequencies of clusters 12, 15, 17, 22, and 23 in CaOx crystals were significantly lower than those in control mice ( Figure 2A ). Based on lineage and functional markers, the cells of cluster 17 were defined as DCs with high CD103 expression that also expressed CD24, MHC-II, BST2, CXCR3, CD69, CCR2, and CD68 ( Figure 2B ). Clusters 22 and 23 were both defined as CX3CR1⁺CD24^- macrophage populations, where CD11c expression differentiated cluster 22 from cluster 23 ( Figure 2B ). Clusters 12 and 15 expressed low levels of CD45, and their cell types could not be determined ( Figure 2B ).

When analyzing the above results, we did not find a significant difference in the frequencies of T cells and B cells. We then depleted CD3⁺ and CD19⁺ cells to an unsupervised cluster using the X-shift algorithm to identify more myeloid cell subsets. We observed 27 clustered immune cell populations, composed mainly of macrophages and DCs ( Figures 3A, B ). Cluster 18 (defined as neutrophils) increased significantly in mice with CaOx crystals, and cluster 20 (defined as CX3CR1⁺CD24^- macrophages) was significantly lower in mice with CaOx crystals than in control mice ( Figures 3C, D ), indicating that macrophages and neutrophils were involved in CaOx crystal formation in the kidney.

Bulk RNA sequencing was applied to identify the functional and transcriptional characteristics of immune kidney cells in mice with CaOx crystals. We identified 172 DEGs in mice with CaOx crystals (adjusted P < 0.05; fold-change ≥ 2 or ≤ 0.5), including 89 upregulated and 83 downregulated mRNAs ( Figure 4A ). GO analyses revealed DEGs associated with responses to external stimuli, extracellular regions, and other parameters ( Figure 4B ). KEGG analyses showed that the DEGs were involved in circadian rhythms, the PPAR signaling pathway, complement and coagulation cascades, retinol metabolism, porphyrin and chlorophyll metabolism, and steroid hormone biosynthesis ( Figure 4C ). To further identify the functional and transcriptional features of the more significant DEGs, we analyzed DEGs with 10-fold changes in expression. Interestingly, DEGs with 10-fold expression differences were enriched for GO terms related to acute inflammatory responses, neutrophil and granulocyte migration, and interleukin-6 and interleukin-8 secretion, among others ( Figure 4D ). In addition, KEGG analysis showed that DEGs with 10-fold expression differences were linked to the IL-17 and TNF signaling pathway, and cytokine–cytokine receptor interactions ( Figure 4E ), where most genes were associated with CXCL2, CXCL1, and IFN-γ ( Figure 4E ). Previous findings showed that these genes were all expressed at sites of inflammation and were related to monocytes and neutrophils (37, 38). The results are in accordance with our CyTOF results, suggesting that the formation of kidney stones was associated with inflammation promoted by neutrophil migration.

Previously, we found that SCFAs were associated with kidney stones (18), which play an important role in immunomodulation. Thus, we next administered SCFAs to mice with renal CaOx crystals to observe the effect of SCFAs on crystal formation through immunomodulation. The results showed that C2, C3, and C4 reduced CaOx crystal deposition ( Figure 5A ), with decreased secretion of serum IL-6, TNF-α, and IL-1β ( Figure 5B ) and decreased expression of NLRP3 ( Supplemental Figure 2 ).

The frequencies of CD45⁺F4/80⁺CD11b⁺CX3CR1⁺CD24^- macrophages (CX3CR1⁺CD24^- macrophages) and CD45⁺GR1⁺CD11b⁺ neutrophils (GR1⁺ neutrophils) in mice with CaOx crystals were detected by flow cytometry ( Supplemental Figure 3 ). In mice with renal CaOx crystals, CX3CR1⁺CD24^- macrophages decreased, and GR1⁺ neutrophils increased significantly ( Figures 6A, B ). Interestingly, we found that C2 and C3 administration significantly increased the frequencies of CX3CR1⁺CD24^- macrophages ( Figure 6C ). Administration of C2, C3, and C4 decreased the frequency of GR1⁺ neutrophils ( Figure 6C ). These results suggest that SCFAs prevent renal CaOx crystal formation through immunomodulation.

To determine whether SCFAs affected the composition of the gut microbiota, caecum microbiota were detected after SCFA treatment in mice. All sequences were clustered into 2199 OTUs belonging to 767 genera and 47 phyla. The coverage indices were greater than 98%. The Wilcoxon rank-sum test showed that alpha diversity indices (Sobs, Simpson, Shannon, Ace, Chao, and Coverage) were not significantly different among the groups ( Figure 7A ). PCoA at the OTU level showed that the microbiota composition of the control mice, CaOx crystal mice, and CaOx crystal mice administered with C2, C3, and C4 were different, which was confirmed by the Adonis test (p < 0.01) ( Figure 7B ).

At the phylum level, Firmicutes was the most common bacteria in the gut microbiota, followed by Bacteroidetes and Actinobacteria ( Figure 7C ). At the genus level, we found that the abundance of Allobaculum was highest in mice with CaOx crystals, although the abundance decreased sharply in mice with CaOx crystals that were administered C2 and C3 (p < 0.01, Figure 7D ). The abundances of Dubosiella, Candidatus, Saccharimonas, and Bifidobacterium significantly increased in mice with CaOx crystals administered C3 (p < 0.05, Figure 7D ). Faecalibaculum was significantly increased in mice with CaOx crystals administered C4 ( Figure 7D ). LEfSe analysis significantly different abundances (p < 0.05) for six, ten, three, six, and 18 bacterial genera in the gut microbiota of mice with CaOx crystals administered C2, C3, and C4; control mice; and CaOx crystal mice, respectively ( Figure 7E and Supplemental Table 3 ). The differentially abundant metabolic pathways are shown in Supplemental Table 4 . PICRUSt analysis revealed that the abundance of 24 KEGG pathways differed significantly among the five groups, including those related to oxidative phosphorylation; fructose and mannose metabolism; glycerophospholipid metabolism; and valine, leucine, and isoleucine degradation (p < 0.05).

To determine whether SCFAs regulated the formation of CaOx crystals in a manner dependent on the gut microbiota, antibiotics were administered to deplete intestinal bacterial loads before SCFAs were used. Administering C2, C3, and C4 with antibiotics reduced CaOx crystal deposition ( Figure 8A ), without decreasing secretion of IL-6, TNF-α, and IL-1β ( Figure 8B ). Interestingly, we found that administration of C2, C3, and C4 did not modulate the frequencies of CX3CR1⁺CD24^- macrophages and GR1⁺ neutrophils ( Figure 9 ) after depletion of the intestinal bacterial load. These results suggest that SCFAs prevent renal CaOx crystal formation, although the immunomodulatory effects depend on the gut microbiota.

GPR43, known as SCFA receptors, is activated by SCFAs (39) and has been reported to be expressed on extensive types of immune cells, particularly neutrophils and monocytes (40, 41). To identify the role of GPR43 in SCFAs in regulating crystal formation through kidney immune cells, GPR43 was detected in the kidneys of mice with or without SCFA administration. GPR43 expression on CX3CR1⁺CD24^- macrophages in the kidneys of mice with CaOx crystals was lower than that in control mice ( Figure 10 ). C2, C3, and C4 administration increased the infiltration of CX3CR1⁺CD24^-GPR43⁺ macrophages with GPR43 ( Figure 10 ). In contrast, GPR43 expression on GR1⁺ neutrophils in the kidney decreased after SCFA administration, compared to that in mice with CaOx crystals ( Figure 11 ).

Then, AAV carrying GPR43 shRNA was administered to mice with CaOx crystals, along with SCFAs. Renal GPR43 expression was knocked down by transduction with the AAV vector after four weeks ( Figure 12 ). GPR43 knockdown inhibited the alleviation of crystal formation in the kidney by SCFAs ( Figure 12 ). We also found that GPR43 knockdown blocked the regulation of CX3CR1⁺CD24^- macrophages and GR1⁺ neutrophils by SCFAs ( Supplemental Figure 4 ).

To evaluate the effect of SCFAs on immune cells via GPR43, SCFAs and GPR43 antagonists (with or without immune cells) were added to mouse renal tubular epithelial cells (TCMK-1 cells) and a CaOx crystal coculture system in vitro. C2, C3, and C4 treatment reduced CaOx crystal adhesion to TCMK-1 cells. CX3CR1⁺CD24^- macrophages showed decreased adhesion to CaOx crystals on TCMK-1 cells, but GR1⁺ neutrophils showed increased adhesion to CaOx crystals, when compared to the CaOx group ( Figure 13 ). After coculturing with SCFAs, CX3CR1⁺CD24^- macrophages showed enhanced adhesion to CaOx crystals on TCMK-1 cells, which was reversed by the GPR43 antagonist ( Figure 13 ). GR1⁺ neutrophils showed decreased adhesion CaOx crystals on TCMK-1 cells after coculture with SCFAs; however, the GPR43 antagonists enhanced CaOx crystal adhesion to TCMK-1 cells when cocultured with GR1⁺ neutrophils and SCFAs ( Figure 13 ). These results suggest that SCFAs not only alleviate crystal adhesion to renal cells, but also play a role in regulating immune cells via GPR43.

In addition, we found that CX3CR1⁺CD24^- macrophages inhibited CaOx crystal adhesion to TCMK-1 cells induced by GR1⁺ neutrophils through GPR43 ( Figure 13 ). Next, we cocultured CX3CR1⁺CD24^- macrophages and GR1⁺ neutrophils in a non-Transwell or a Transwell system, with or without TCMK-1 cells or CaOx crystals. CX3CR1⁺CD24^- macrophages suppressed GPR43 expression on GR1⁺ neutrophils ( Figure 14A ) and decreased the migration of GR1⁺ neutrophils ( Figure 14B ). However, treating CX3CR1⁺CD24^- macrophages with GPR43 antagonists restored GPR43 expression on neutrophils and the migration of GR1⁺ neutrophils ( Figure 14 ). The results indicate that the interaction between CX3CR1⁺CD24^- macrophages and GR1⁺ neutrophils was mediated through GPR43.

In summary, we found multiple immune cell subpopulations in the kidney. We and also found that SCFAs reduced CaOx crystal formation in the kidney by decreasing the secretion of IL-1β, IL-6, and TNF-α by modulating CX3CR1⁺CD24^- macrophages and GR1⁺ neutrophils in a manner that depended on the gut microbiota. Furthermore, SCFAs alleviated crystal formation and crystal adhesion in the kidney by regulating macrophages and neutrophils with GPR43. Additionally, our results revealed that CX3CR1⁺CD24^- macrophages regulate GR1⁺ neutrophils through GPR43 ( Figure 15 ).