In this study, all animals were housed at 21°C with 50% humidity, on a 12-h light:12-h dark schedule in the standard animal facility, five per cage, in accordance with the guidelines of Tongji University, and all animal work was conducted under ethical permission from the Tongji University ethical review panel.

MiR-146a^+/- mice were generated by Cyagen Biosciences Inc. (Guangzhou, China) using CRISPR/Cas9 technology. Briefly, in vitro transcribed Cas9 mRNA and gRNA (gRNA1: TGCACCACCCATGATAGGGCAGG, gRNA2: GCTTCGCTTTTCCCAACCGGAGG) were injected into fertilized eggs of C57BL/6N mice. PCR product sequencing using two pairs of primers (miR-146a F1, 5′ AAGGGAAGGATTGAACATGACACA 3′, miR-146a R1 5′ TTATTGCCTCTCTACAAGGACCTG 3′, and miR-146a R2 5′ ACCATCAATAGCAGAGATGACTGG 3′) was performed to identify F0-generation mice carrying a miR-146a knock-out allele. The positive F0-generation mice were then mated with wild-type C57BL/6N mice to obtain F1-generation mice, and the F1-generation miR-146a^+/- mice were identified by PCR, and Sanger sequencing confirmed a deletion of 1,467 bp, which did not affect the gene sequence of the gene Gm12148 which is close to miR-146a. Male and female experimental WT C57BL/6N mice were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Zhejiang, China).

During colonoscopy, biopsy samples were collected from 22 adult patients with ongoing gut symptoms by the Department of Gastroenterology, The Third Xiangya Hospital, Central South University according to the ethics permit (R19057) approved by the Ethics Committee of The Third Xiangya Hospital. All fresh samples from patients who requested biopsy examination and agreed to donate biopsy tissue for our research were included in this study, with no exclusion criteria. Among them, six subjects were without IBD (non-IBD), and seven and nine patients were with IBD in the active stage (active IBD) or inactive stage (inactive IBD). The information of patients with IBD is provided in Table S1 . Biopsy samples were immediately frozen in dry ice and stored in a −80°C freezer until further analysis.

Experiment 1: 2% DSS (molecular weight of 36–50 kDa, MP Biomedicals, Irvine, CA, USA) in water was used to induce C57BL/6N mice and miR-146a^+/- mice to become IBD models. WT and miR-146a^+/- mice were randomly divided into four groups: WT control group, miR-146a^+/- control group, DSS-treated WT group, and DSS-treated miR-146a^+/- group. Control group mice had free access to tap water, and DSS group mice and three additional miR-146a^-/- mice were given 2% DSS in drinking water for 6 days to induce experimental colitis and fed with water on day 7. Mice were sacrificed on day 7 and samples were collected.

Experiment 2: To investigate the efficacy of miR-146a mimics in DSS-induced miR-146a^-/- IBD mice, miR-146a^-/- mice were randomly divided into three groups: saline control group, saline + DSS group, and miR-146a mimics + DSS group. Saline control group mice had free access to tap water and treated with vehicle saline. Saline + DSS group and miR-146a mimics + DSS group mice were given 2.5% DSS (a new batch from the same supplier) in drinking water for 7 days to induce experimental colitis; meanwhile, saline and miR-146a mimics [the modified miR-146a-5p mimics (40 pg per gram of body weight) and miR-146a-3p mimics (80 pg per gram of body weight)] were purchased from Suzhou Biosyntech Co. Suzhou, China ( Supplementary Table S2 ) and administrated via gavage and intraperitoneal injection every other day from day 1 for five times, respectively. Mice were sacrificed on day 9 and samples were collected.

Experiment 3: To examine the efficacy of miR-146a mimics in DSS-induced WT mice, C57BL/6N mice were randomly divided into three groups: saline control group, saline + DSS group, and miR-146a mimics + DSS group. Saline control group mice had free access to tap water and treated with vehicle saline. Saline + DSS group and miR-146a mimics + DSS group mice were fed with 2.5% DSS containing water on day 1 and followed with 2% DSS containing water from day 2 to day 6, and treated with vehicle saline and miR-146a mimics [the modified miR-146a-5p mimics (6.5 pg per g body weight) and miR-146a-3p mimics (3.25 pg per g body weight)], respectively, via intraperitoneal injection every other day from day 1 for five times. Mice were sacrificed on day 9 and samples were collected.

The disease activity index (DAI) was determined by weight loss, stool consistency, and fecal occult blood.

The weight loss score is as follows: 0 points, <1%; 1 point, 1%–5%; 2 points, 5%–10%; 3 points, 10%–20%; and 4 points, >20%. The stool consistency is divided into three conditions: 0 points, normal stool; 2 points, loose stool; and 4 points, water-like stool. The feces was stained with the fecal occult blood test kit (Zhuhai BaSO Biotechnology Co., Zhuhai, China), and the fecal occult blood was scored according to the following scale: 0 points, negative; 2 points, positive; and 4 points, obvious blood on feces. Note that spouted rectal bleeding, which was indicated by bloodstain on mouse tail, was not counted in the score of fecal occult blood.

Colons were quickly removed from mice, and their length was measured and photographed. After removing the content, the colon was washed twice with cold saline. Dissected colon tissues were snap-frozen and stored at −80°C for RT-qPCR detection. For immunofluorescence and histochemical staining detection, colons were immersed in 4% PFA overnight and embedded in paraffin after progressive dehydration.

Total RNA was isolated from colons using Trizol Reagent (Thermo Fisher Scientific, Massachusetts, USA) and 1 µg of total RNA was used for library preparation according to Illumina standard instruction (VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina^®). An Agilent 4200 bioanalyzer was employed to evaluate the concentration and size distribution of cDNA library before sequencing with an Illumina novaseq6000. The protocol of high-throughput sequencing was followed according to the manufacturer’s instructions (Illumina). The raw reads were filtered by Seqtk before mapping to genome using Hisat2 (version: 2.0.4) (28). The gene fragments were counted using stringtie (v1.3.3b) followed by TMM (trimmed mean of M values) normalization (29–31). Significant differentially expressed genes (DEGs) were identified as those with a p value < 0.01 and fold change >2 using the edgeR software (32). The RNA-seq raw data as well as gene read counts for individual samples are accessible at Gene Expression Omnibus (GEO) under accession number GSE247433.

Total RNA was extracted from mouse colons and human samples using Trizol reagent (Thermo Fisher Science, Massachusetts, USA) and 2 μg of total RNA was reverse-transcribed with the Revert Aid First Strand Gene Synthesis Kit (Thermo Fisher Science, Massachusetts, USA). Using SYB-Green and the ABI QuantStudio 3 machine (Thermo Fisher Scientific, Massachusetts, USA), the expression level of Mmp3, Mmp8, Mmp10, Il6, Il1a, Il1b, Cxcl2, Cxcl3, S100a8, S100a9, Lcn2, Serpine1, Serpine2, Ccl3, Saa3, and Csf3 was quantified by real-time PCR with specific primers listed in Supplementary Table S3 and normalized to Gapdh RNA. For miRNA detection, 1 μg of total RNA was polyadenylated and reverse-transcribed using the Catch-All™ miRNA & mRNA RT-PCR Kit (Pengekiphen, Suzhou, China) according to the manufacturer’s instructions. Subsequently, qRT-PCR assays were conducted using the Catch-All™ miRNA & mRNA universal PCR primer as the reverse primer, and the specific miRNA forward primers listed in Supplementary Table S3 or as previously described (18). The amplification conditions were as follows: an initial step at 95°C for 10 min, and 40 cycles of 15 s at 95°C and 1 min at 60°C. All assays were performed in triplicate and negative controls were included by omitting the template. The Ct value was recorded for each reaction, and the expression level of miRNAs was calculated relative to U6, a ubiquitously expressed snRNA.

Paraffin sections with a thickness of 4 µm were stained using a hematoxylin and eosin (HE) staining kit (Solarbio Life Sciences, Beijing, China) according to the user manual. The slides were evaluated blindly by experienced pathologists, and histological scores were evaluated according to previous studies (33, 34) based on the following parameters: the area affected by inflammatory infiltration and tissue damage.

Three random areas of each colon sample were photographed, with the grades of inflammatory cell infiltration and tissue damage scored according to the following criteria: (1) Inflammatory cell infiltration: 0 point, no inflammatory infiltration; 1 point, less inflammatory infiltration in the lamina propria; 2 points, the muscular layer of the mucosa infiltrates, thickening of the mucosa; 3 points, inflammatory cells penetrate the submucosa; (2) tissue damage: 0 point, no epithelial cell changes; 1 point, goblet cell deletion; 2 points, a large area of goblet cells is missing, crypt is missing; 3 points, extensive destruction of the structure of the mucosa and extension to the muscular layer of the intestine. The mean of scores of the three areas represented the grade of colitis in each animal.

Immunostainings on 4-μm paraffin sections were performed as described previously (35). The following primary antibodies were used: MMP3 antibody (Abcam, #AB52915, 1:100), MMP8 antibody (Abcam, #AB53017, 1:1,000), MMP10 antibody (Abcam, #AB261733, 1:100), IL6 antibody (CST, #12912s, 1:200), IL1A antibody (Abcam, #AB300499, 1:50), IL1B antibody (Abcam, #ab234437, 1:50), CXCL2 antibody (Thermo, #MA5-23737, 1:100), CXCL3 antibody (Abcam, #ab220431, 1:1,000), S100A8/S100A9 antibody (Abcam, #ab288715, 1:500), NF-κB p65 antibody (Abcam, #ab16502, 1:1,000), Phospho-NF-κB p65 antibody (CST, #3033, 1:1,000), TRAF6 antibody (Abcam, #ab33915, 1:1,000), and IRAK-1 antibody (Abcam, #ab218130, 1:200). Secondary antibodies were conjugated with AlexaFluor 488 (Thermo Fisher Scientific, Massachusetts, USA, 1:1,000). Nuclei were counterstained with DAPI (Sigma, #MBD0015, 1:10,000). Fluorescent images were captured with a confocal laser microscope (Spin SR10, Olympus, Japan) and processed with Adobe Photoshop software.

The fluorescence signal intensities of MMP3, MMP8, MMP10, IL6, IL1A, IL1B, CXCL2, CXCL3, S100A8+S100A9, NF-κB (p65), Phospho-NF-κB (p65), TRAF6 and IRAK1 were measured using ImageJ on three sections, which came from every ninth serial sections from distal colon of three WT mice and three miR-146a+/- mice; three DSS-treated WT mice and three DSS-treated miR-146a+/- mice in experiment 1, and every ninth serial sections from distal colon of three saline treated miR-146a-/- mice, three saline + DSS treated miR-146a-/- mice and four mimics + DSS miR-146a-/- mice in experiment 2, and three every ninth serial sections from distal colon of three saline treated WT mice, three saline + DSS treated WT mice and four mimics + DSS WT mice.

Sixteen 8-week-old C57BL/6N mice were randomly divided into four groups and fasted for 6 h, and then mice were intraperitoneally injected or administrated via gavage with miR-146a-5p mimics (6.5 pg per gram of body weight) and miR-146a-3p mimics (3.25 pg per gram of body weight), or treated in combination with miR-146a-5p mimics (6.5 pg per gram of body weight) and miR-146a-3p mimics (3.25 pg per gram of body weight), or served as untreated controls. Colon tissues were collected 4 h after treatment of miR-146a mimics and fixed in 4% PFA overnight. Colon tissues were embedded in OCT after immersion in 15% sucrose and 30% sucrose overnight, and sectioned in a cryostar (NX50, Thermo Fisher Scientific) at a thickness of 14 μm. The colon cryosections were stained with DAPI and photographed using a confocal microscope (Zeiss 710, Germany).

To investigate whether miR-146a is involved in IBD, we employed RT-qPCR to quantify the expression level of miR-146a in the colon of patients, and the results demonstrated that miR-146a-3p was significantly expressed higher in the colon of patients with IBD in both the active stage and the inactive stage than that in non-IBD patients who had other types of bowel disease, and that there was an increasing trend in expression level of miR-146a-5p in the colon of patients with IBD compared to patients without IBD ( Figure 1A ). Since the expression of miR-146a could be increased in patients with other types of bowel diseases (36) and it is difficult to obtain colon tissues from healthy people, we generated an IBD mouse model by induction with 2.5% DSS and found that both miR-146a-5p and miR-146a-3p were upregulated by more than two fold in the colon of DSS-treated wild type (WT) mice when compared to untreated control WT mice ( Figure 1B ). These findings suggest that miR-146a-5p and miR-146a-3p were involved in the development of IBD.

To further decipher the function of miR-146a in IBD, we generated a global miR-146a knock-out (KO) mouse line using CRISPR/Cas9 technology ( Supplementary Figures S1A–C ) and used DSS to induce mice for the development of colitis symptoms ( Figure 1C ). During the treatment of 2% DSS in drinking water from day 1 to day 6, the body weight of DSS-treated female ( Figure 1D ) and male ( Figure 1E ) miR-146a ^+/- mice dropped significantly from day 5 to day 7 compared with DSS-treated WT mice, untreated control miR-146a ^+/-, and WT mice, while DSS-treated female and male WT mice did not show significant body weight loss when compared to untreated WT control mice ( Figures 1D, E ). The stool scores of both DSS-treated WT and miR-146a^+/- mice were dramatically increased from day 3 when compared to untreated controls, but there was no significant difference in stool score between DSS-treated WT and miR-146a^+/- mice ( Figures 1F, G ). The DAI was significantly higher in DSS-treated female and male miR-146a^+/- mice than in DSS-treated female and male WT mice ( Figures 1H, I ) from day 5 to day 7. Moreover, all 2% DSS-treated female and male miR-146a^+/- mice had bloodstain on their tails, which indicated massive rectal bleeding, while 2% DSS-treated WT female and male mice did not ( Figures 1J, K ). Colon length and histological score are two key parameters that indicate the severity of colitis. The results showed that there was no difference in colon length between untreated WT and miR-146a^+/- mice, and that 2% DSS treatment significantly shortened colon length (19%) in WT female, but not male, mice and the colon lengths of DSS-treated female and male miR-146a^+/- mice were shortened by 39% and 38% when compared to untreated female and male miR-146a^+/- control mice, and were also significantly shorter than DSS-treated female and male WT mice, respectively ( Figures 1L, M ). Colon sections of different groups were histologically stained and examined. We found that DSS treatment caused significant destruction of epithelial layer and induced inflammation in the colon of WT mice when compared to untreated WT control, and the colon derived from DSS-treated miR-146a^+/- mice showed a higher histological score, with more inflammatory cell infiltration and tissue damage, than untreated miR-146a^+/- and DSS-treated WT mice. Interestingly, untreated miR-146a^+/- mice exhibited a slightly higher histological score, which can be attributed to inflammation when compared to untreated WT mice, indicating that miR-146a represses inflammation responses under physiological conditions ( Figures 1N, O ). These findings indicate that the insufficiency of miR-146a promotes the development of IBD from three aspects: colitis, bleeding, and body weight loss, and suggest that miR-146a plays a vital role in preventing the development of IBD.

We next investigated the genetic regulatory network regulated by miR-146a in IBD. Total RNA was extracted from colons of three untreated WT mice, three 2% DSS-treated WT mice, and three DSS-treated miR-146a^-/- mice. RNA sequencing data demonstrated that 479 and 512 genes that met the criteria of p < 0.01 and |log2(Fold change)| > 1 were up- and downregulated in WT mice after DSS treatment, respectively ( Figure 2A ; Supplementary Table S4 ). However, much more genes (1,465 versus 479) were upregulated and 791 genes were downregulated in DSS-treated miR-146a^-/- mice with the same criteria compared to untreated WT mice ( Figure 2B ; Supplementary Table S4 ). Maintaining p < 0.01 but lowering fold-change criteria from sixfold to fourfold or to twofold led to a disproportionate increase in upregulated genes shared between DSS-treated WT and miR-146a^-/- mice and maintaining close to 67% shared gene regardless of fold-change threshold ( Figures 2C, D ). A disproportionate increase in IBD-related genes was also observed to be shared between DSS-treated WT and miR-146a^-/- mice when lowering the fold-change criteria and maintaining more than 77% shared IBD-related gene from change thresholds of twofold (77%) to sixfold (98%) ( Figures 2E, F ). When increasing fold-change threshold to 10, 28 out of 71 upregulated genes in DSS-treated WT mice were IBD-related genes, while 128 out of 381 upregulated genes in DSS-treated miR-146a^-/- mice were IBD-related genes, and 29 (such as Il1a, Il11, Il5ra, Il1r2, Cxcl13, Cxcl5, Osm, Arg1, Igfbp5, and Vsig4) out of 128 genes were predicted to be direct target genes of miR-146a by TargetScan8.0 ( Supplementary Figures S2A, B ; Supplementary Table S5 ). When further elevating fold-change threshold to 45, the number of upregulated IBD-related genes in DSS-treated miR-146a^-/- mice was 10 times more (55 versus 5) than that in DSS-treated WT mice ( Supplementary Figures S2C, D ). However, the number of genes downregulated more than 10-fold in DSS-treated miR-146a^-/- mice was not remarkably increased when compared to DSS-treated WT mice (64 versus 45, Supplementary Figures S2E, F ; Supplementary Table S5 ). These findings suggest that miR-146a mainly inhibits expression of DSS-induced genes, especially IBD-related genes.

Further analyses found that 963 and 529 genes that met the criteria of p < 0.01 and |log2(fold change)| > 1 were up- and downregulated in DSS-treated miR-146a^-/- when compared to DSS-treated WT mice, respectively ( Figure 2G ; Supplementary Figure S3A ). Principal component analysis also showed that the mRNA expression profile in the colon of DSS-treated miR-146a^-/- mice was separated from DSS-treated WT mice and naïve mice ( Supplementary Figure S3B ).When increasing fold-change threshold to 10, 225 and 41 genes were up- and downregulated in DSS-treated miR-146a^-/- mice when compared to DSS-treated WT mice, respectively, and 64 out of 225 were IBD-related genes ( Supplementary Figures S3C, D ). Hundreds of IBD-related genes were further induced by DSS in miR-146a^-/- mice more than twofold and 141 IBD-related genes were upregulated by more than fourfold when compared to DSS-treated WT mice ( Figure 2H ; Supplementary Figure S3E ). These upregulated IBD-related genes belong to genetic regulatory networks of innate immune responses, intestinal epithelial and vascular endothelial barriers, and coagulation, corresponding to all symptoms of IBD (colitis, bleeding, bowel problems, and body weight loss) ( Supplementary Figures S3C, E ). It is known that cytokines, chemokines, and other pro-inflammatory factors are upregulated in patients with IBD (39–41). Matrix metalloproteinases (MMPs) destroy the extracellular matrix, damage the intestinal epithelial barrier and vascular endothelial barrier, and subsequently affect digestion and absorption, leading to bleeding (6). Another reason for massive bleeding is dysfunction in coagulation (13). Thus, 16 out of 141 further upregulated more than fourfold genes, which shows that multiple genetic regulatory networks involved in IBD were selected for validation: Il1a, Il1b, and Il6 from the cytokine family; Ccl3, Cxcl2, and Cxcl3 from the chemokine family; Serum Amyloid A3 (Saa3), colony stimulating factor 3 (Csf3), Lipocalin-2 (Lcn2), S100a8, and S100a9 from pro-inflammatory factors; Mmp3, Mmp8, and Mmp10 from the MMP family; and Serpine1 and Serpine2 from the Serpin family, which inhibits serine proteases including thrombin, urokinase, and plasmin and is related to coagulation ( Figure 2I ; Table 1 ).

Quantitative RT-PCR confirmed the dramatic upregulation of these 16 genes in the colon of DSS-treated miR-146a^-/- mice compared to DSS-treated WT mice, and the expression levels of Mmp3, Mmp8, Mmp10, Il1a, Il1b, Cxcl2, Cxcl3, S100a8, S100a9, Lcn2, Ccl3, and Saa3 increased in the colon of DSS-treated WT mice compared to WT control mice ( Figure 2J ). It is also interesting to find that the expression levels of Mmp8, Mmp10, Cxcl2, Cxcl3, S100a8, S100a9, Ccl3, and Csf3 increased in the colon of DSS-treated WT mice comparing to WT control mice ( Figure 2J ). Further immunostaining demonstrated that the protein expression levels of MMP3, MMP8, MMP10, IL1A, IL1B, IL6, CXCL2, CXCL3, S100A8, and S100A9 significantly increased in the colon of WT mice after DSS treatment, and further increased in the colon of DSS-treated miR-146a^+/- mice. Since it was reported that miR-146a negatively regulated TRAF6 and IRAK1, and that p65 (RelA) is a predicted target of miR-146a-3p, we examined and found that the expression of TRAF6, IRAK1, p65, and p-p65 (phosphorylated p65) was also further upregulated in the colon of DSS-treated miR-146a^+/- mice when compared to DSS-treated WT mice ( Figures 3A, B ). Note that the expression levels of MMP3, MMP8, MMP10, IL1A, IL1B, IL6, CXCL2, CXCL3, S100A8, S100A9, TRAF6, P65, pP65, and IRAK1 increased in the colon of untreated miR-146a^+/- mice when compared to untreated WT mice, indicating that miR-146a prevents the allergic response of the intestine to a normal environment ( Figures 3A, B ). These findings suggest that de-repression of genetic regulatory networks of immunity, intestinal epithelial and vascular endothelial barriers, and coagulation is the molecular mechanism for massive bleeding and body weight loss in DSS-treated miR-146a KO mice.

Next, we examined whether supplementation with miR-146a mimics could relieve DSS-induced IBD in miR-146a^-/- mice. MiR-146a^-/- mice were divided into three groups, treatment with saline, treatment with 2% DSS and saline, treatment with 2% DSS and miR-146a mimics ( Figure 4A ). The results showed that 2% DSS induced severe IBD in miR-146a^-/- mice with obvious body weight loss and an increase in stool score and DAI, and that the combination of gavage administration and intraperitoneal injection of the modified miR-146a-5p and miR-146a-3p mimics for four times every other day significantly reduced body weight loss, stool score, and DAI in DSS-treated miR-146a^-/- mice when compared to DSS + saline-treated miR-146a^-/- mice ( Figures 4B–D ). DSS-induced spouted bleeding of miR-146a^-/- mice was eliminated by supplementation with miR-146a mimics ( Figure 4E ). MiR-146a mimics prevented 32% DSS-induced reduction in the colon length and attenuated the destruction of the intestine as well as inflammatory cell filtration ( Figures 4F–H ). These results demonstrated that supplementation with miR-146a mimics can relieve DSS-induced IBD in miR-146a^-/- mice.

To identify the downregulated IBD-related genes by miR-146a mimics, RNA sequencing was performed and the results illustrated that 135 and 69 genes that met the criteria of p < 0.01 and |log2(Fold change)| > 1 were down- and upregulated in the colon of miR-146a mimics + DSS-treated miR-146a^-/- mice, when compared to DSS + saline-treated miR-146a^-/- mice, respectively, and 50 out of the 135 genes were downregulated IBD-related genes ( Figures 5A, B ; Supplementary Table S6 ). Principal component analysis showed that the mimics treatment pulled the expression profile back towards health status ( Supplementary Figure S4 ). Quantitative RT-PCR showed that the expression levels of Mmp3, Mmp8, Mmp10, Il1a, Il6, Il1b, Cxcl2, Cxcl3, S100a8, S100a9, Lcn2, Serpine1, Serpine2, Ccl3, Saa3, and Csf3 were induced in the colon of DSS + saline-treated miR-146a^-/- mice, but reversed by miR-146a mimics ( Figure 5C ). Further immunostaining demonstrated that the expression levels of MMP3, MMP8, MMP10, IL1A, IL1B, IL6, CXCL2, CXCL3, S100A8, S100A9, TRAF6, p65, p-p65, and IRAK1 were induced by DSS in the colon of miR-146a^-/- mice, but repressed by miR-146a mimics ( Figures 6A, B ).

We next examined whether miR-146a mimics could attenuate DSS-induced IBD in WT mice. WT mice were divided into three groups. Two groups of mice were fed with 2.5% DSS in water on day 1, followed by 2% DSS from day 2 to day 6, and water on day 7; the other group of mice were fed with water ( Figure 7A ). The results showed that DSS induced severe IBD in WT mice with obvious body weight loss and an increase in stool score and DAI, and that intraperitoneal injection of miR-146a-5p and miR-146a-3p mimics for four times every other day did not reduce body weight loss, but decreased stool score and DAI in DSS-treated WT mice when compared to DSS + saline-treated WT mice ( Figures 7B–D ). Moreover, miR-146a mimics prevented DSS-induced reduction in colon length of WT mice ( Figure 7E ), and decreased damage to the intestine as well as inflammatory cell filtration ( Figure 7F ). Quantitative RT-PCR showed that the expression levels of Mmp3, Mmp8, Mmp10, Il6, Il1a, Cxcl2, Cxcl3, S100a8, S100a9, and Lcn2 were induced in the colon of DSS + saline-treated WT mice, but significantly downregulated by miR-146a mimics ( Figure 7G ). The upregulation of Saa3 was not significantly repressed by miR-146a mimics, and downregulation of Serpine1, Serpine2, and Csf3 was detected in DSS + saline-treated WT mice when compared to saline-treated WT mice. MiR-146a mimics restored the expression levels of Serpine1 and Csf3 ( Figure 7G ). Surprisingly, the expression of Il1b and Ccl3 was further increased in the colon of miR-146a mimics + DSS-treated WT mice compared to DSS + saline-treated WT mice. Further immunostaining illustrated that the expression levels of MMP3, MMP8, MMP10, IL1A, IL1B, IL6, CXCL2, CXCL3, S100A8, S100A9, TRAF6, p65, p-p65, and IRAK1 were induced by DSS in the colon of WT mice, but reversed by miR-146a mimics ( Figure 8 ). To investigate whether the modified miR-146a mimics were uptaken by intestinal cells, Cy3-conjugated miR-146a-5p mimics and Cy3-conjugated miR-146a-3p mimics were administrated into WT mice via gavage or intraperitoneal injection or combinatory injection, and the fluorescence images captured by confocal microscopy showed that miR-146a mimics were uptaken by intestinal cells 4 h after administration, and more absorption of miR-146a mimics was found in the gavage and intraperitoneal injection combination group ( Supplementary Figure S5 ).

We next determined whether targets regulated by miR-146a increased in the colon of patients with IBD. Quantitative RT-PCR showed that the expression levels of MMP3, MMP10, IL6, IL1B, S100A8, S100A9, SERPINE1, and CSF3 were higher in the colon of patients with active IBD than non-IBD subjects and patients with IBD in the inactive stage, and the expression level of IL1A was downregulated in patients with IBD in the inactive stage when compared to patients with active IBD ( Figure 9 ). However, MMP8, CXCL2, CXCL3, LCN2, SERPINE2, and CCL3 were not upregulated in the colon of patients with active IBD when compared to non-IBD subjects and patients with IBD in the inactive stage ( Figure 9 ), but we should bear in mind that non-IBD subjects and patients with IBD in the inactive stage do not have health conditions, and these genes could be upregulated in these two groups of patients. Nevertheless, these findings suggest that MMP3, MMP10, IL6, IL1B, S100A8, S100A9, SERPINE1, CSF3, and IL1A are involved in active IBD.