JMJD1A expression is upregulated in the colon and colonic epithelial cells in response to C. rodentium infection

To investigate the patterns of JMJD1A expression in the colons of WT mice after C. rodentium infection, we collected the colons of WT mice on day 7 after enteric infection with C. rodentium and then performed western blot assays and quantitative RT-PCR assays. As shown in Fig 1A and 1B, the protein and mRNA levels of JMJD1A were significantly increased in the colons of mice on day 7 after C. rodentium infection. JMJD1A immunohistochemistry (IHC) staining showed that JMJD1A expression was mainly detected in the colonic epithelial cells on day 7 after C. rodentium infection, suggesting that C. rodentium infection can induce JMJD1A expression in the colonic epithelial cells (Fig 1C). Consistently, western blot and quantitative RT-PCR analysis showed that the protein and mRNA expression of JMJD1A was significantly upregulated in the colonic epithelial cells of mice on days 7 and 10 after C. rodentium infection (Fig 1D and 1E). Furthermore, heat-killed C. rodentium (HKCR) induced JMJD1A protein and mRNA expression in CT26 cells (Fig 1F and 1G). To determine the mechanism of JMJD1A upregulation in the colonic epithelial cells, we analyzed the binding sites of the transcription factors on the JMJD1A promoter as well as the RNA-sequencing data of the colons of mice on day 7 after C. rodentium infection to predict potential transcription factors regulating JMJD1A expression. To determine which transcription factor can regulate the expression of JMJD1A, we screened the sequence of mouse JMJD1A promoter in Jaspar databases to search for the potential transcription factor binding sites. 114 potential transcription factors were obtained. Then, we searched the expression model of these 114 potential transcription factors in RNA-sequencing data of the colons of mice on days 7 and 0. Hence, we found some significantly upregulated transcription factors such as Rel and IRF1, and other significantly downregulated transcription factors such as Jun and Fosl2 in the colons of mice on day 7 after C. rodentium infection (Fig 1H). Nineteen transcription factors that showed significant changes between the day 7 and day 0 was provided in supplementary information and listed in excel (S1 Data, Excel file). Consistently, mRNA and western blot analysis showed that the expression of IRF1 was significantly increased in the colons after C. rodentium infection (Fig 1I). Knockdown of IRF1 in CT26 cells reduced the protein and mRNA levels of JMJD1A after C. rodentium infection (Fig 1J), suggesting that IRF1 can regulate JMJD1A expression. An IRF1 binding site was found between -1391 bp and -992 bp on the JMJD1A promoter, implicating that IRF1 may regulate JMJD1A expression at the transcriptional level. To test it, we performed JMJD1A promoter-reporter (p (-1984/-88)) assays and found that IRF1 could increase JMJD1A promoter reporter activity in a dose dependent manner (Fig 1K and 1L). However, IRF1 could not increase the reporter activity of a truncated JMJD1A promoter-reporter (p (-992/-88)) with IRF1 binding site absence (Fig 1L). These results supported the notion that IRF1 regulates JMJD1A expression at the transcriptional level. In addition, we found c-Rel binding site (TAGAATTTCC, -557 ~ -566) on the JMJD1A promoter, implicating that c-Rel may bind to this site to induce the expression of JMJD1A. However, the results of JMJD1A promoter-reporter (p (-1984/-88)) assays showed that c-Rel could not enhance JMJD1A promoter activity in CT26 cells (S1 Fig), excluding the c-Rel-mediated regulation of JMJD1A. To determine whether IRF1 could be recruited to the IRF1 binding site on the JMJD1A promoter after HKCR treatment, ChIP assays were performed and the results showed that IRF1 could be recruited to the IRF1 binding site on the JMJD1A promoter (Fig 1M). Taken together, these results demonstrated that JMJD1A expression was significantly increased in the colon, at least in part mediated by IRF1 after C. rodentium infection, implicating that JMJD1A may play a role in host defense against C. rodentium infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. JMJD1A expression is upregulated in the colon and colonic epithelial cells in response to C. rodentium infection.

(A) The protein level of JMJD1A was significantly increased in the colons of mice on day 7. The protein level was measured in the colons of mice on day 7 and day 0 after C. rodentium infection. Densitometric analysis of JMJD1A were performed. (B) The mRNA level of JMJD1A was significantly increased in the colons of mice. The mRNA level was measured in the colons of mice on day 7 and day 0 after C. rodentium infection. (C) Anti-JMJD1A immunohistochemistry staining showed that JMJD1A expression was mainly detected in the colons of WT and JMJD1A^-/- mice on day 7 after C. rodentium infection. Anti-JMJD1A immunohistochemistry staining in the colons of mice on day 7 and day 0. Scale bars represent 50 μm. The black arrow represents JMJD1A-positive cells. (D) The protein level of JMJD1A was significantly upregulated in the colonic epithelial cells of WT mice on days 7 and 10 compared with day 0 after C. rodentium infection. (E) The mRNA level of JMJD1A was significantly upregulated in the colonic epithelial cells of WT mice on days 7 and 10 after C. rodentium infection. (F) The mRNA level of JMJD1A was significantly upregulated in CT26 cells treated with HKCR (MOI = 200) for 30 min and 60 min. (G) The protein level of JMJD1A was significantly upregulated in CT26 cells treated with HKCR (MOI = 200) for 30 min and 60 min. Densitometric analysis of JMJD1A were performed. (H) Prediction of JMJD1A promoter-binding transcription factors by JASPAR and heatmap of DEG in the colons of WT mice infected with C. rodentium for 7 days relative to control group (for 0 day) from the RNA-seq database. (I) The levels of mRNA and protein of IRF1 were significantly increased in the colons of mice after C. rodentium infection. (J) Knockdown of IRF1 in CT26 cells reduced the mRNA and protein levels of JMJD1A after C. rodentium infection. n = 4. Densitometric analysis of JMJD1A were performed. (K) IRF1 could increase JMJD1A promoter reporter activity in a dose dependent manner in CT26 cells. CT26 cells were co-transfected with Renilla plasmids, murine JMJD1A promoter, and IRF1 expression plasmids using Lipofectamine 2000. The luciferase and Renilla values were measured at 24 h after co-transfection. (L) IRF1 could not increase the reporter activity of a truncated JMJD1A promoter-reporter (p (-992/-88)) with IRF1 binding site absence. (M) IRF1 could be recruited to the IRF1 binding site on the JMJD1A promoter. Data are mean ± SEM. Results are representative of three independent experiments. P values were assessed with the two-tailed Student’s t-test, or one-way ANOVA test. *P < 0.05; **P < 0.01; ***P < 0.001.

https://doi.org/10.1371/journal.ppat.1014009.g001

JMJD1A^-/- mice exhibit increased mortality, worsened colonic histopathology, and enhanced proinflammatory cytokine production after C. rodentium infection

To determine the role of JMJD1A in host defense against C. rodentium infection, we infected WT and JMJD1A^-/- mice with C. rodentium orally and monitored the body weight change and the survival of mice. C. rodentium infection led to more body weight loss in JMJD1A^-/- mice compared with WT mice at day 10 post-infection (Fig 2A). Strikingly, all JMJD1A^-/- mice succumbed to C. rodentium infection whereas all WT mice survived (Fig 2B). Increased mortality of JMJD1A^-/- mice infected with C. rodentium may be associated with more severe colonic injury and colonic inflammation. Indeed, a shorter colon length, which is one of the inflammatory features, was observed in infected JMJD1A^-/- mice compared with WT mice on day 12 after C. rodentium infection (Fig 2C). C. rodentium infection-induced colonic injury and colonic inflammation are characterized by epithelial integrity impairment, inflammatory cell infiltration, and submucosal oedema [25]. Therefore, we performed H&E staining to assess C. rodentium-induced colonic injury. H&E staining showed that worsened histopathology was observed in the colons of JMJD1A^-/- mice (Fig 2D). To determine whether the colonic inflammation is aggravated in JMJD1A^-/- mice during C. rodentium infection, we measured the levels of proinflammatory cytokines in the colon culture supernatant of WT mice and JMJD1A^-/- mice by ELISA. Although the concentrations of TNFα and IL-1β were comparable in the colons of WT mice and JMJD1A^-/- mice, the concentration of IL-6 in the colons of JMJD1A^-/- mice was lower compared with WT mice on day 7 after infection (Fig 2E). The concentrations of TNFα, IL-1β, and IL-6 in the colons of JMJD1A^-/- mice were significantly higher compared with WT mice on day 12 after infection (Fig 2E). Collectively, these results suggest that JMJD1A can protect mice from C. rodentium-induced colonic injury and colonic inflammation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. JMJD1A^-/- mice exhibit increased mortality, worsened colonic histopathology, and enhanced proinflammatory cytokine production after C. rodentium infection.

(A) Body weight loss was greater in JMJD1A^-/- mice (n = 11) compared with WT mice (n = 12) on day 14 after C. rodentium infection. (B) Survival kinetics of C. rodentium-infected mice revealed significantly higher mortality in JMJD1A^-/- mice (n = 8) compared to WT (n = 8). (C) Colon length of JMJD1A^-/- mice was shorter than WT mice on day 12 after C. rodentium infection. n = 4-8. (D) H&E staining of colon sections exhibited more severe histopathological damage in JMJD1A^-/- mice compared with WT mice on days 7 and 12 after C. rodentium infection. n = 4-10. Representative pictures of H&E staining of the colon sections (Left panel); Quantification of histopathology scores (Right panel). Scale bars represent 100 μm. (E) The concentrations of TNFα, IL-1β, and IL-6 were significantly higher in the colons of JMJD1A^-/- mice compared with WT mice on day 12 after C. rodentium infection as measured by ELISA. n = 6-10. The data were pooled from three independent experiments. Data are mean ± SEM. The survival curve was assessed with the log-rank test; the other data were assessed with the two-tailed Student’s t-test or nonparametric tests. *P < 0.05; **P < 0.01; ***P < 0.001.

https://doi.org/10.1371/journal.ppat.1014009.g002

JMJD1A^-/- mice exhibit impaired clearance of C. rodentium

Increased mortality and worsened colonic histopathology of JMJD1A^-/- mice infected with C. rodentium may be caused by impaired clearance of C. rodentium. As shown in Fig 3A-3C, JMJD1A^-/- mice exhibited more C. rodentium load compared with WT mice in the feces, colons, and caecum with feces on days 7 and 12 after C. rodentium infection. Furthermore, JMJD1A^-/- mice exhibited more severe systemic spreading of C. rodentium to spleens, livers, and lungs compared with WT mice on days 7 and 12 after C. rodentium infection (Fig 3D-3F). Immunohistochemical staining of C. rodentium in the colons of WT mice and JMJD1A^-/- mice showed that JMJD1A^-/- mice had more invaded crypts and C. rodentium loads compared with WT mice on days 7 and 12 after C. rodentium infection (Fig 3G). Taken together, these results suggest that JMJD1A protects mouse from C. rodentium infection by enhancing clearance of C. rodentium and limiting the systemic spreading of C. rodentium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. JMJD1A^-/- mice exhibit impaired clearance of C. rodentium.

(A-F) JMJD1A^-/- mice exhibited more C. rodentium load compared with WT mice in feces (A), colons (B), caecum + feces (C), spleen (D), liver (E), and lung (F) after oral administration of C. rodentium for 7 days and 12 days. (G) Anti-C. rodentium immunohistochemistry staining showed that the colon sections of JMJD1A^-/- mice had more invaded crypts and C. rodentium loads compared with WT mice on days 7 and 12 after C. rodentium infection. n = 4-7. The black arrow indicates the crypt tissue invaded by C. rodentium with brown signals. Representative pictures of anti-C. rodentium immunohistochemistry staining of the colon sections (Left panel); Quantification of crypt invasion by C. rodentium in JMJD1A^-/- mice and WT mice (Right panel). Scale bars represent 100 μm. (H) The protein levels of JMJD1A in the colons and colonic epithelial cells of WT mice infected with AAV9-shJMJD1A were significantly decreased compared with AAV9-shCtrl for 4 weeks. Densitometric analysis of JMJD1A were performed. (I) H&E staining of colon sections from WT mice infected with AAV9-shJMJD1A showed a severe colonic histopathology than in AAV9-shCtrl group after C. rodentium infection. Representative pictures of H&E staining of the colon sections (Left panel); Quantification of histopathology scores (Right panel). Scale bars represent 100 μm. n = 6. (J) Mice infected with AAV9-shJMJD1A exhibited more C. rodentium load compared with mice infected with AAV9-shCtrl in the colons on day 14 after C. rodentium infection. n = 11. (K) Anti-C. rodentium immunohistochemistry staining of the colon sections of WT mice infected with AAV9-shJMJD1A or AAV9-shCtrl after C. rodentium infection revealed that JMJD1A-knockdown mice had more invaded crypts and C. rodentium loads compared with control mice. Brown signals represent C. rodentium. Representative pictures of anti-C. rodentium immunohistochemistry staining of the colon sections (Left panel); Quantification of crypt invasion by C. rodentium in WT mice infected with AAV9-shJMJD1A or AAV9-shCtrl (Right panel). Scale bars represent 100 μm. n = 6. The data were pooled from three independent experiments. Data are mean ± SEM. P values were assessed with the two-tailed Student’s t-test or nonparametric tests. *P < 0.05; **P < 0.01; ***P < 0.001.

https://doi.org/10.1371/journal.ppat.1014009.g003

Previous study has demonstrated that adeno-associated virus (AAV) pseudotype 9 could be used for transduction of the epithelium in the colon, which is ideal for studying mucosal inflammation in the experimental colitis [26]. Since JMJD1A was mainly expressed in the colonic epithelial cell after C. rodentium infection (Fig 1C), we wondered whether JMJD1A in colonic epithelial cell is responsible for the clearance of C. rodentium. We knocked down JMJD1A in the colonic epithelial cells in vivo by infecting WT mice with AAV9 carrying JMJD1A short hairpin RNA (AAV9-shJMJD1A) or negative control short hairpin RNA (AAV9-shCtrl) via portal vein injection, and then orally administered C. rodentium to these mice four weeks later. To address the specificity of our AAV-mediated JMJD1A knockdown in vivo, we performed a series of cellular and molecular analyses. First, immunohistochemical staining confirmed that AAV9-shJMJD1A treatment significantly reduced JMJD1A protein levels within the colonic epithelial compartment, while expression in the lamina propria stromal cells remained comparable to the control group (S2 Fig). To further validate this specificity and examine potential off-target effects on some crucial immune cells, we isolated colonic epithelial cells, macrophages (isolated using anti-mouse CD11b MicroBeads), and CD4⁺ T cells (anti-mouse CD4 MicroBeads) from the colons of the infected mice treated with AAV9-shCtrl and AAV9-shJMJD1A. Consistent with the IHC results, qPCR results revealed a significant reduction of JMJD1A mRNA exclusively in the epithelial cells of AAV9-shJMJD1A-treated mice. In contrast, JMJD1A expression in sorted macrophages and CD4⁺ T cells, as well as in the unlabeled cells, was not significantly changed (S3A Fig). We further assessed the cytokines produced by CD4⁺ T cells isolated from AAV9-shCtrl mice and AAV9-shJMJD1A mice on day 14 after infection. qPCR analysis in these sorted CD⁺ T cells showed comparable levels of IFNγ, IL-10, and IL-17A between AAV9-shCtrl group and AAV9-shJMJD1A group (S3B Fig). As shown in Fig 3H, western blot results showed that JMJD1A expression was significantly decreased in the colons and colonic epithelial cells of mice infected with AAV9-shJMJD1A. C. rodentium infection led to more body weight loss in AAV9-shJMJD1A mice compared with AAV9-shCtrl mice at day 14 post-infection (S4A Fig). However, all AAV9-shCtrl and AAV9-shJMJD1A mice survived after C. rodentium infection (S4B Fig), which differed from the lethal phenotype observed in JMJD1A^-/- mice (Fig 2B). The different causes for this result might be that AAV9-shJMJD1A could not fully get rid of JMJD1A in IECs of AAV9-shJMJD1A mice, or the AAV9-shJMJD1A mice mainly affected IECs, while JMJD1A^-/- mice had a much broader range of affected cells. H&E staining showed a severe colonic histopathology in AAV9-shJMJD1A group compared with AAV9-shCtrl group after C. rodentium infection (Fig 3I). Furthermore, mice infected with AAV9-shJMJD1A exhibited more C. rodentium load compared with mice infected with AAV9-shCtrl in the colons on day 14 after C. rodentium infection (Fig 3J). Consistent with these results, immunohistochemical staining of C. rodentium in the colons of mice infected with AAV9-shJMJD1A or AAV9-shCtrl showed that JMJD1A-knockdown mice had more invaded crypts and C. rodentium loads compared with control mice on day 14 after C. rodentium infection (Fig 3K). These results suggest that JMJD1A expressed in the colonic epithelial cell promotes the clearance of C. rodentium.

The expression of antimicrobial peptide (AMPs) in the colons of WT and JMJD1A^-/- mice is comparable with or without C. rodentium infection

One of the important functions of colonic epithelial cell is to produce AMPs to control bacterial infection. Since JMJD1A expressed in the colonic epithelial cell promoted the clearance of C. rodentium, we wondered whether JMJD1A regulates the expression of colonic AMPs to clear C. rodentium. However, there was no difference observed in the mRNA levels of AMPs in the colons between WT and JMJD1A^-/- mice with or without C. rodentium infection (S5 Fig), suggesting that JMJD1A-mediated protection against C. rodentium is not through regulation of AMPs.

JMJD1A^-/- mice exhibit impaired recruitment of macrophages and CD4⁺ T cells and reduced production of C. rodentium-specific antibodies after C. rodentium infection

The other important function of colonic epithelial cells is to produce chemoattractants to recruit immune cells. Many immune cell types, including macrophages, CD4⁺ T cells, and neutrophils, play critical roles in C. rodentium clearance [11,27,28]. We therefore performed F4/80 IHC staining for macrophages, CD4 IHC staining for CD4⁺ T cells, and MPO IHC staining for neutrophils in the colons of WT mice and JMJD1A^-/- mice after C. rodentium infection. Without C. rodentium infection, the number of macrophages and CD4⁺ T cells was comparable in the colons of WT mice and JMJD1A^-/- mice (Fig 4A and 4B). However, on days 7 and 12 after C. rodentium infection, more macrophages and CD4⁺ T cells were observed in the colons of WT mice compared with JMJD1A^-/- mice (Fig 4A and 4B). The number of neutrophils was comparable in the colons of WT and JMJD1A^-/- mice after C. rodentium infection (S6 Fig). Additionally, the numbers of CD20-positive cells were significantly decreased in the colons of JMJD1A^-/-mice compared to WT mice after C. rodentium infection on day 12 (S7 Fig). Previous studies have shown that CD4⁺ T cell-dependent IgG response is required for the clearance of C. rodentium [11], but IgA and IgM are dispensable [13]. Especially, luminal IgG is essential for eradicating C. rodentium and promoting the host survival [10]. Therefore, we assessed C. rodentium-specific IgG and IgA in the luminal content of WT mice and JMJD1A^-/- mice after C. rodentium infection. As shown in Fig 4C, the titers of IgG and IgA in the luminal contents of JMJD1A^-/- mice were reduced compared to WT mice on day 12 after C. rodentium infection. These results demonstrated that the recruitment of macrophages and CD4⁺ T cells, as well as CD4⁺ T cell-dependent antibody responses were impaired in the colons of JMJD1A^-/- mice after C. rodentium infection, leading to impaired clearance of C. rodentium. Similar to JMJD1A^-/- mice, less macrophages and CD4⁺ T cells were recruited into the colons of the AAV9-shJMJD1A-infected mice compared with the AAV9-shCtrl-infected mice after C. rodentium infection (Fig 4D and 4E), and the C. rodentium-specific IgG and IgA in the luminal content of mice infected with AAV-shJMJD1A were less than those of mice infected with AAV-shCtrl on day 14 after C. rodentium infection (Fig 4F). Taken together, these results suggest that JMJD1A in the colonic epithelial cell plays a critical role in host defense against C. rodentium infection by promoting the recruitment of macrophages and CD4⁺ T cells to clear C. rodentium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. JMJD1A in the colonic epithelial cell promotes the clearance of C. rodentium and the recruitment of macrophages and CD4⁺ T cells to the colon.

(A) Anti-F4/80 immunohistochemistry staining showed that more macrophages were observed in the colons of WT mice compared with JMJD1A^-/- mice on days 7 and 12 after C. rodentium infection. n = 3-5. Representative pictures of anti-F4/80 immunohistochemistry staining of the colon sections (Left panel); Quantification of F4/80⁺ cells (Right panel). Arrows represent F4/80-positive cells. Scale bars represent 100 μm. (B) Anti-CD4 immunohistochemistry staining showed that more CD4⁺ T cells were detected in the colons of WT mice compared with JMJD1A^-/- mice on days 7 and 12 after C. rodentium infection. n = 3-5. Representative pictures of anti-CD4 immunohistochemistry staining of the colon sections (Left panel); Quantification of CD4⁺ T cells (Right panel). Arrows represent CD4-positive cells. Scale bars represent 50 μm. (C) The contents of IgG and IgA in the luminal contents of JMJD1A^-/- mice were reduced compared to WT mice on day 12 after C. rodentium infection. n = 3-8. (D) Anti-F4/80 immunohistochemistry staining showed that less macrophages were recruited into the colons of the AAV9-shJMJD1A-infected mice compared with the AAV9-shCtrl-infected mice after C. rodentium infection. Representative pictures of anti-F4/80 immunohistochemistry staining of the colon sections (Left panel); Quantification of F4/80⁺ cells (Right panel). Arrows represent positive cells. Scale bars represent 100 μm. n = 3-5. (E) Anti-CD4 immunohistochemistry staining showed that less CD4⁺ T cells were recruited into the colons of the AAV9-shJMJD1A-infected mice compared with the AAV9-shCtrl-infected mice after C. rodentium infection. Representative pictures of anti-CD4 immunohistochemistry staining of the colon sections (Left panel); Quantification of CD4⁺ T cells (Right panel). Scale bars represent 50 μm. n = 3-5. (F) Production of total IgG and IgA against C. rodentium were decreased in the luminal content of AAV9-shJMJD1A infected mice compared with AAV9- shCtrl infected mice on day 14. n = 6-8. Data are mean ± SEM. The data were pooled from three independent experiments. P values were assessed with the two-tailed Student’s t-test or nonparametric tests, or two-way ANOVA test. *P < 0.05; **P < 0.01; ***P < 0.001.

https://doi.org/10.1371/journal.ppat.1014009.g004

JMJD1A^-/- mice exhibit reduced CCL8 induction, which is responsible for reduced recruitment of macrophages and CD4⁺ T cells to the colons after C. rodentium infection

Since chemokines play important roles in the recruitment of many immune cell types, including macrophages and CD4⁺ T cells, we hypothesized that the difference in the recruitment of macrophages and CD4⁺ T cells in the colons of infected WT and JMJD1A^-/- mice may be due to the difference in the expression of chemoattractants. Therefore, we detected the expression of chemoattractants in the colons of WT mice and JMJD1A^-/- mice by RNA-sequencing on day 7 after C. rodentium infection. RNA-sequencing analysis showed that the macrophage chemoattractants CCL3 and CCL4, the neutrophil chemoattractants CXCL5, and the macrophage and CD4⁺ T cell chemoattractants CXCL9 and CCL8 were significantly decreased in the colons of JMJD1A^-/- mice on day 7 after C. rodentium infection (Fig 5A). However, the quantitative RT-PCR results showed that only CCL8 was significantly reduced in the colons of JMJD1A^-/- mice after C. rodentium infection (Fig 5B). The expression pattern of colonic CCL8 protein was in accord with the expression pattern of CCL8 mRNA (Fig 5C). Furthermore, the expression of CCL8 in the colonic epithelial cells of JMJD1A^-/- mice was markedly decreased compared to WT mice on day 7 after C. rodentium infection (Fig 5D). These results suggest that reduced induction of CCL8 may lead to reduced recruitment of macrophages and CD4⁺ T cells in the colons of JMJD1A^-/- mice.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. JMJD1A^-/- mice exhibit reduced CCL8 induction, which is responsible for reduced recruitment of macrophages and CD4⁺ T cells to the colons after C. rodentium infection.

(A) RNA-sequencing was performed in the colons of JMJD1A^-/- and WT mice infected with C. rodentium for 0 and 7 days. Selected genes involved in chemokines and proinflammatory cytokines have shown as a heat map from the RNA-seq database: differentially expressed genes (DEGs) were highlighted with “*”; for visualization, value of the Log₂(FPKM+0.1) are represented as row Z-scores. Red indicated high expression, and the blue indicated low expression relative to the gene mean. Each sample is pooled from 3 mice. (B) CCL8 was significantly reduced in the colons of JMJD1A^-/- mice compared with WT mice on days 7 and 12 after C. rodentium infection. n = 3-7. (C) The concentration of CCL8 was decreased in the colon cultured supernatants of JMJD1A^-/- mice compared with WT mice as measured by ELISA on days 7 and 12 after C. rodentium infection. n = 3-5. (D) The expression of CCL8 in the colonic epithelial cells was reduced in the JMJD1A^-/- mice compared to WT mice on day 7 after C. rodentium infection. n = 3. (E) The concentration of CCL8 in the colon cultured supernatants was increased in AAV9-mCCL8 infected mice compared to AAV9-flag infected mice. n = 3. (F) H&E staining of colon sections showed that AAV9-mCCL8 infected mice suffered from less severe pathology compared to AAV9-flag infected mice after C. rodentium infection for 14 days. Representative pictures of H&E staining of the colon sections (Left panel); Quantification of histopathology scores (Right panel). Scale bars represent 100 μm. n = 3-5. (G) C. rodentium loads in the colons of AAV9-mCCL8 infected mice (n = 10) was significantly lower than in those AAV9-flag (n = 8) after oral administration of C. rodentium for 14 days. (H) Anti-C. rodentium immunohistochemistry staining showed that CCL8-overexpressed mice had less invaded crypts and C. rodentium loads compared with control mice on day 14 after C. rodentium infection. Brown signals represent C. rodentium. Representative pictures of anti-C. rodentium immunohistochemistry staining of the colon sections (Upper panel); Quantification of crypt invasion by C. rodentium in WT mice infected with AAV9-mCCL8 or AAV9-flag (Lower panel). Scale bars represent 100 μm. n = 5. (I) Anti-F4/80 immunohistochemistry staining revealed more macrophages infiltration in the colons of AAV9-mCCL8 infected mice compared to AAV9-flag group with C. rodentium infection for 14 days. Representative pictures of anti-F4/80 immunohistochemistry staining of the colon sections (Left panel); Quantification of F4/80⁺ cells (right panel). Scale bars represent 100 μm. n = 3-5. (J) Anti-CD4 immunohistochemistry staining showed the enhanced CD4⁺ T cells accumulation in the colons of WT mice infected with AAV9-mCCL8 compared to AAV9-flag group with C. rodentium infection for 14 days. Representative pictures of anti-CD4 immunohistochemistry staining of the colon sections (Left panel); Quantification of CD4⁺ T cells (Right panel). Scale bars represent 50 μm. n = 3-5. (K) Production of IgG and IgA against C. rodentium were increased in the luminal content of AAV-mCCL8-infected mice compared to AAV-Flag-infected mice on day 14 after C. rodentium infection. Data are mean ± SEM. The data were pooled from three independent experiments. P values were assessed with the two-tailed Student’s t-test or nonparametric tests, or two-way ANOVA test. *P < 0.05; **P < 0.01.

https://doi.org/10.1371/journal.ppat.1014009.g005

Additionally, we also attempt to explore other pathways that may operate downstream of JMJD1A. As shown in Fig 5A, JMJD1A^-/- mice exhibited lower colonic IFNγ mRNA levels. We quantified IFNγ protein levels by ELISA (S8C Fig) and assessed STAT1 phosphorylation by western blot in colonic tissues from WT and JMJD1A^-/- mice uninfected and infected with C. rodentium for 7 days. Correspondingly, the protein levels of STAT1 phosphorylation in colons of JMJD1A^-/- mice were lower than that of the WT mice (S8B Fig). These results suggest that IFNγ-STAT1-IRF1 axis could be a second crucial arm of the JMJD1A-mediated host defense against enteric infection.

To verify the role of CCL8 in host defense against C. rodentium in vivo, WT mice were infected with AAV9-mCCL8 or control AAV9 (AAV9-flag) via portal vein injection before C. rodentium infection, and then C. rodentium was orally administered to these mice infected with AAV9-mCCL8 or AAV9-flag four weeks later. As shown in Figs 5E and S9, mouse CCL8 expression was significantly increased in the colons of mice infected with AAV9-mCCL8. H&E staining exhibited an alleviative colonic histopathology in the AAV9-mCCL8 group compared with the AAV9-flag group after C. rodentium infection (Fig 5F). Furthermore, the number of C. rodentium in the colons of mice infected with AAV9-mCCL8 was less than those of mice infected with AAV9-flag on day 14 after C. rodentium infection (Fig 5G). Consistent with these results, immunohistochemical staining of C. rodentium in the colons of mice infected with AAV9-mCCL8 or AAV9-flag showed that CCL8-overexpressing mice had fewer invaded crypts and C. rodentium loads compared with control mice on day 14 after C. rodentium infection (Fig 5H). Additionally, more macrophages and CD4⁺ T cells were recruited to the colons of the AAV9-mCCL8 group compared with the AAV9-flag group after C. rodentium infection (Fig 5I and 5J). The C. rodentium-specific IgG and IgA in the luminal content of mice infected with AAV9-mCCL8 were more than those of mice infected with AAV-flag on day 14 after C. rodentium infection (Fig 5K). Taken together, these results suggest that CCL8 in the colonic epithelial cell is responsible for the host defense against C. rodentium infection by recruiting CD4⁺ T cells and macrophages to the colons.

CCL8 restore rescues JMJD1A-knockdown-caused susceptibility to C. rodentium infection

To determine whether CCL8 is a key downstream effector mediating JMJD1A’s protective role, we performed a functional rescue experiment by restoring CCL8 expression in JMJD1A-knockdown (AAV9-shJMJD1A) mice via co-infection of AAV9-mCCL8. The protein level of JMJD1A was significantly decreased in the colonic epithelial cells of both AAV9-shJMJD1A-flag mice and AAV9-shJMJD1A-mCCL8 mice on day 14 after C. rodentium infection (S10A Fig), indicating that co-infection of AAV9-mCCL8 did not affect the knockdown efficiency of JMJD1A by AAV9-shJMJD1A. Then, we examined the efficacy and specificity of CCL8 overexpression after co-infection of AAV9-mCCL8. JMJD1A knockdown in AAV9-shJMJD1A-flag mice markedly reduced CCL8 mRNA expression in the colonic epithelial cells on day 14 after infection, which was rescued by co-infection of AAV9-mCCL8 (S10B Fig), but CCL8 mRNA expression was comparable in the macrophages, CD4⁺ T cells, and unlabeled cells from the colons of AAV9-shJMJD1A-CCL8 mice and AAV9-shJMJD1A-flag mice on day 14 after infection (S10C Fig), indicating that CCL8 expression was specifically rescued in the epithelial compartment of the rescue group. Consistently, the protein level of CCL8 was significantly decreased in the colon cultured supernatants of JMJD1A-knockdown mice on day 14 after C. rodentium infection, but rescued by co-infection of AAV9-mCCL8 (S10D Fig).

Critically, CCL8 rescue by co-infection of AAV9-mCCL8 markedly reversed the susceptibility to C. rodentium infection caused by JMJD1A knockdown. Compared to AAV9-shJMJD1A-flag mice, AAV9-shJMJD1A-CCL8 mice exhibited a marked decrease in colonic C. rodentium burden (S11A Fig), significantly attenuated pathology (S11B Fig), and diminished bacterial crypt invasion (S11C Fig). Additionally, compared to AAV9-shJMJD1A-flag mice, IHC analysis showed increased infiltration of macrophages (S11D Fig) and CD4⁺ T cells (S11E Fig) in the colons of AAV9-shJMJD1A-CCL8 mice on day 14 after C. rodentium infection. Compared to AAV9-shJMJD1A-flag mice, AAV9-shJMJD1A-CCL8 displayed restored levels of C. rodentium-specific fecal IgG and IgA on day 14 after C. rodentium infection (S11F Fig). These results demonstrate that CCL8 restore is sufficient to rescue the host defense defect caused by JMJD1A knockdown.

JMJD1A cooperates with STAT1 to promote IRF1 expression which further enhances CCL8 expression

As C. rodentium-induced CCL8 expression was reduced in colonic epithelial cells of JMJD1A^-/- mice, we hypothesized that CCL8 may be regulated by JMJD1A directly in colonic epithelial cells after C. rodentium infection. To test it, we transfected JMJD1A-specific small interfering RNA to knock down JMJD1A in CT26 cells and then assessed the effects of JMJD1A knockdown on CCL8 expression after HKCR treatment. As shown in Fig 6A, HKCR treatment increased CCL8 expression in control CT26 cells, whereas the induction of CCL8 expression was markedly impaired in JMJD1A-knockdown CT26 cells, supporting the hypothesis that JMJD1A promotes CCL8 expression in the colonic epithelial cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. JMJD1A cooperates with STAT1 to promote the expression of IRF1.

(A) The mRNA level of CCL8 was decreased in JMJD1A knock-down CT26 cells with heat-killed C. rodentium treatment. (B) The mRNA level of CCL8 was decreased in the IRF1 knock-down CT26 cells with heat-killed C. rodentium treatment. (C) The mRNA level of IRF1 was decreased in the JMJD1A knock-down CT26 cells with C. rodentium treatment. (D) The protein levels of JMJD1A and IRF1 were decreased in JMJD1A knock-down CT26 cells by western blot with heat-killed C. rodentium (MOI = 200) treatment. (E) The protein level of JMJD1A and IRF1 were decreased in colons of WT and JMJD1A^-/- mice by western blot after C. rodentium infection. (F) KEGG pathway analysis showed that Jak-STAT signaling pathway was enriched. (G) The protein levels of p-STAT1 and STAT1 were increased in the colons of mice after C. rodentium infection. (H) JMJD1A could cooperate with STAT1 to enhance IRF1 promoter activity in CT26 cells. CT26 cells were co-transfected with Renilla plasmids, murine IRF1 promoter, JMJD1A expression plasmids and STAT1 expression plasmids using Lipofectamine 2000. The luciferase and Renilla values were measured at 24 h after co-transfection. n = 3. (I) The endogenous JMJD1A and STAT1 in CT26 cells could be detected by co-immunoprecipitation using STAT1 and JMJD1A antibodies, respectively. (J) STAT1 bound to the IRF1 promoter without HKCR treatment, and the binding was increased after HKCR treatment. The affinity of STAT1 on the IRF1 promoter in CT26 cells via ChIP assay using STAT1 antibodies or IgG antibodies. (K) JMJD1A bound to the IRF1 promoter without HKCR treatment, and the binding was increased after HKCR treatment. The affinity of JMJD1A on the IRF1 promoter in CT26 cells was measured via ChIP assay using JMJD1A antibodies or IgG antibodies. (L) The binding of H3K9me2 was decreased after HKCR treatment. The affinity of H3K9me2 on the IRF1 promoter in CT26 cells was measured via ChIP assay using H3K9me2 antibodies or IgG antibodies. Data are mean ± SEM. Results are representative of three independent experiments. P values were assessed with the two-tailed Student’s t-test or nonparametric tests, or one-way ANOVA test. *P < 0.05; **P < 0.01.

https://doi.org/10.1371/journal.ppat.1014009.g006

To determine which transcription factor can regulate the expression of CCL8, we screened the sequence of mouse CCL8 promoter in Jaspar databases to search for the potential transcription factor binding sites. We found IRF1 binding site (AGAAAAAGGAA, -519 ~ -509) on the CCL8 promoter, implicating that IRF1 may bind to this site to induce the expression of CCL8. Indeed, knockdown of IRF1 significantly reduced CCL8 expression (Fig 6B). To determine whether JMJD1A knockdown reduces CCL8 expression by decreasing IRF1 expression, we examined the effect of JMJD1A knockdown on IRF1 expression. As shown in Fig 6C and 6D, knockdown of JMJD1A markedly decreased HKCR-induced IRF1 mRNA and protein expression in CT26 cells. Consistently, western blotting results showed that IRF1 expression was significantly reduced in the colons (Fig 6E) and the colonic epithelial cells (S8A Fig) of JMJD1A^-/- mice compared with WT mice on day 7 after C. rodentium infection.

Since KEGG analysis showed that C. rodentium infection was associated with the activation Jak-STAT signaling pathway (Fig 6F), the protein levels of p-STAT1 and STAT1 were increased in colons and colonic epithelial cells after C. rodentium infection (Figs 6G and S8A), and it has been reported that the transcription factor STAT1 can regulate IRF1 expression at the transcriptional level [29], we wondered whether JMJD1A could cooperate with STAT1 to induce IRF1 expression. We therefore performed IRF1 promoter reporter assays to test this hypothesis. The results of IRF1 promoter reporter assays showed that JMJD1A could cooperate with STAT1 to enhance IRF1 promoter activity in CT26 cells (Fig 6H). Furthermore, to determine whether JMJD1A could interact with STAT1, we treated CT26 cells with HKCR and then performed Co-IP assays. As shown in Fig 6I, JMJD1A antibodies could immunoprecipitate STAT1 and STAT1 antibodies could immunoprecipitate JMJD1A, suggesting that JMJD1A can interact with STAT1. ChIP assays also revealed that JMJD1A and STAT1 bound to the IRF1 promoter without HKCR treatment, and the binding was increased after HKCR treatment (Fig 6J and 6K)，whereas the levels of H3K9me2 were reduced (Fig 6L). These results suggest that JMJD1A cooperates with STAT1 to promote IRF1 expression, which further enhances CCL8 expression.

JMJD1A cooperates with IRF1 to promote CCL8 expression

In addition to enhancing IRF1 expression to promote CCL8 expression, we wondered whether JMJD1A could cooperate with IRF1 to promote CCL8 expression. We therefore performed CCL8 promoter reporter assays. As shown in Fig 7A, JMJD1A could cooperate with IRF1 to enhance CCL8 promoter activity, whereas the JMJD1A (H1120Y) mutant, which loses the histone demethylation function, failed to cooperate with IRF1 to enhance CCL8 promoter activity. These results suggest that JMJD1A can cooperate with IRF1 to promote CCL8 expression and the demethylation activity of JMJD1A is required for promoting CCL8 expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. JMJD1A cooperates with IRF1 to promote CCL8 expression.

(A) JMJD1A could cooperate with IRF1 to enhance CCL8 promoter activity. CT26 cells were co-transfected with Renilla plasmids, murine CCL8 promoter reporter, JMJD1A expression plasmids or JMJD1A mutant expression plasmids, and IRF1 expression plasmids using Lipofectamine 2000. The luciferase and Renilla values were measured at 24 h after co-transfection. n = 3. (B) JMJD1A antibodies could immunoprecipitate IRF1 and IRF1 antibodies could immunoprecipitate JMJD1A. HEK293 cells were co-transfected with Flag-JMJD1A expression plasmids and Myc-IRF1 expression plasmids using Lipofectamine 2000. Co-IP analysis of the interaction of JMJD1A and IRF1 with indicated tag antibodies. n = 3. (C) JMJD1A antibodies could immunoprecipitate IRF1 and IRF1 antibodies could immunoprecipitate JMJD1A in CT26 cells. CT26 cells were treated with HKCR (MOI = 200), and cell lysis was collected. Co-IP analysis of the interaction of JMJD1A and IRF1 with indicated antibodies. n = 3. (D) JMJD1A and IRF1 bound to the CCL8 promoter without HKCR treatment, and the binding was enhanced after HKCR treatment, but the level of H3K9me2 was reduced. CT26 cells were treated with or without HKCR (MOI = 200), and ChIP assays were performed. The affinity of IRF1 on the CCL8 promoter in CT26 cells was measured by ChIP assay using IRF1 antibodies or IgG antibodies. The affinity of JMJD1A on the CCL8 promoter in CT26 cells was measured by using JMJD1A antibodies or IgG antibodies. The affinity of H3K9Me2 on the CCL8 promoter in CT26 cells was measured by using H3K9me2 antibodies or IgG antibodies. (E-H) HKCR treatment caused an increase in the chemoattraction of Raw264.7 and CTLL-2 cells, while knockdown of JMJD1A or IRF1 in CT26 cells decreased the chemoattraction of Raw264.7 and CTLL-2 cells. Graph showing number migrating toward in a chemotaxis assay. The data were normalized by the cell number of negative control (siNC) group. n = 4. (I) Schematic model for the mechanism by which JMJD1A protect mice from C.rodentium infection by upregulating CCL8 to recruit macrophages and CD4⁺ T cells. The image was created by “ScienceSlides 2005” plugin and Microsoft PowerPoint software. Data shown in graphs are Mean ± SEM or SD. The data were pooled from three independent experiments. P values were assessed with the two-tailed Student’s t-test or nonparametric tests, or one-way ANOVA test. *P < 0.05; **P < 0.01; ***P < 0.001.

https://doi.org/10.1371/journal.ppat.1014009.g007

To determine whether JMJD1A could interact with IRF1, we transfected 293T cells with JMJD1A and IRF1 expression plasmids for 24 h and then performed Co-IP assays. As shown in Fig 7B, JMJD1A antibodies could immunoprecipitate IRF1 and IRF1 antibodies could immunoprecipitate JMJD1A. In addition, the interaction of endogenous JMJD1A and IRF1 in CT26 cells could be detected by co-immunoprecipitation using JMJD1A antibodies and IRF1 antibodies, respectively (Fig 7C). Because JMJD1A can cooperate with IRF1 to enhance the CCL8 promoter activity, we determined whether JMJD1A could bind to the IRF1 binding site on the CCL8 promoter. CT26 cells were treated with HKCR and then the ChIP assays were performed. The results showed that JMJD1A and IRF1 bound to the CCL8 promoter without HKCR treatment, and the binding was enhanced after HKCR treatment (Fig 7D, left and middle panels). Furthermore, H3K9me2 could be detected at the IRF1 binding site on the CCL8 promoter without HKCR treatment, but the level was reduced after HKCR treatment (Fig 7D, right panel). Taken together, these results suggest that JMJD1A can be recruited to IRF1 binding site on the CCL8 promoter to demethylate H3K9me2 for enhancement of CCL8 expression.

To verify that JMJD1A and IRF1-induced CCL8 could chemoattract macrophages and T cells, chemotaxis assays in a transwell system were performed utilizing mouse macrophage cell line RAW264.7 and T cell line CTLL-2, respectively. As shown in Fig 7E-7H, HKCR treatment caused an increase in the chemoattraction of Raw264.7 and CTLL-2 cells, while knockdown of JMJD1A or IRF1 in CT26 cells decreased the chemoattraction of Raw264.7 and CTLL-2 cells after HKCR treatment. Furthermore, CCL8 overexpression in JMJD1A-knockdown or IRF1-knockdown CT26 cells could partly rescue the chemoattraction of Raw264.7 and CTLL-2 cells.

JMJD1A impacts on the health status of gut microbiota

Because the gut microbiota plays an important role in regulating intestinal inflammation and enteropathogens colonization, we assessed the fecal microbiota composition from naïve WT and JMJD1A^-/- mice by 16S rRNA amplicon sequencing. The principal co-ordinated analysis (PCoA) of beta diversity based on a Bray-Curtis comparison showed that the fecal samples from naïve WT and JMJD1A^-/- mice were separated (S12A Fig). Beta diversity of the fecal microbiomes from the WT mice was significant difference compared with JMJD1A^-/- mice at genus level (S12B Fig). Gut Microbiome Health index (GMHI) showed that the health conditions of gut microbiota from WT mice were better than that of gut microbiota from JMJD1A^-/- mice (S12C Fig). Furthermore, microbial dysbiosis index (MDI) showed that the microbial dysbiosis of JMJD1A^-/- mice was more severe than that of WT mice (S12D Fig). Linear discriminant analysis effect size (LEfSe) analysis and linear discriminant analysis (LDA) scores revealed that there was significantly different Oscillospirales Order between  WT and JMJD1A^-/- mice, which belongs to Firmicutes phylum (S12E and S12F Fig). There were significantly different bacterial species, including norank_f_Eubacterium_coprostanoligenes_group and Colidextribacter, identified by intergroup significance analysis at genus level between WT and JMJD1A^-/- mice (S12G and S12H Fig). These results suggest that JMJD1A promotes the growth of some gut microbiota and impacts on the health status of gut microbiota.

In summary, our study demonstrates that JMJD1A plays an essential role in protecting colon from C. rodentium infection. Mechanistically, JMJD1A cooperates with STAT1 to enhance IRF1 expression, and then cooperates with IRF1 to promote CCL8 expression, thereby enhancing the recruitment of macrophages and CD4⁺ T cells to the colons. Recruited macrophages kill C. rodentium through phagocytosis and induction of inflammatory response, and recruited CD4⁺ T cells activate B cells to produce antibodies against C. rodentium, both actions limit C. rodentium infection (Fig 7I).

Ethics statement

All experimental procedures were carried out in line with animal protocols, which are approved by the Laboratory Animal Center of Xiamen University. The animal experimental design was approved by the Ethics Committee of Xiamen University (approval number: XMULAC20190099).

Mice

JMJD1A^-/- mice were generated on a C57BL/6 background as previously described [79]. WT mice from the same litter served as a control group. Ten-week-old female mice were employed in all experiments.

C. rodentium infection

After fasting for 8 h, WT and JMJD1A^-/- mice were infected with C. rodentium strain ATCC51459 (1.0 × 10⁹ CFU mouse^-1 in 0.2 mL PBS) by oral gavage. For quantification of C. rodentium, feces and tissues samples were collected and weighed. Then, the samples were homogenized in a Tissuelyser-24 at 30 Hz for 6 min. Homogenates were serially diluted with PBS, and seeded onto MacConkey agar plates. C. rodentium colonies were counted after incubating overnight at 37 °C.

Cell culture

The human embryonic kidney 293 cell line (HEK293T) and the murine cell lines CT26 were all purchased from Cell Bank of Type Culture Collection of Chinese Academy of Sciences (Shanghai, China) and the cell lines were stored in the lab of Professor Yu. HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM), and the CT26 cells were cultured in RPMI-1640, which were supplemented with streptomycin (100 mg mL^-1), penicillin (100 IU) and fetal bovine serum (FBS, 10%). The cells were all cultured in a humidified incubator containing 5% CO₂ at 37 ℃.

RNA interference

JMJD1A siRNA [22] and nonspecific siRNA were purchased from Genepharma. CT26 cells were transfected with JMJD1A siRNA (20 nM) or negative siRNA using Lipofectamine 2000 (ThermoFisher Scientific, Waltham, US) according to the manufacturer’s instructions.

Histological analysis, immunofluorescence and immunohistochemistry

10% neutral Formalin-fixed distal colons were embedded in paraffin and 5 μm sections were cut. The sections were then stained with hematoxylin and eosin (H&E) by standard techniques or prepared for immunohistochemistry. For H&E staining, histological scoring was determined as described previously [80]. For immunohistochemistry, the method was described in our previous study [76]. The antibodies were incubated as follows: rabbit anti-C. rodentium antibody, rabbit anti-JMJD1A (NB100–77282, Novus), rabbit anti-CD4 antibody (Cat#ab183685, Abcam), and rabbit anti-F4/80 antibody (Cat#70076, CST) for overnight at 4 ℃.

For immunofluorescence staining (IF), the antibodies Rabbit anti-Mouse-CCL8 antibody (Cat#AF790, R&D Systems) was incubated for overnight at 4 ℃. The Fluorescent Secondary Antibody (Cat#A23330, Abbkine) for immunofluorescence was incubated for 30 min. Slides were mounted with the Antifade Mounting Medium with DAPI (Cat# H-1200, Vector Laboratories, USA).

Isolation of colonic epithelial cells

The colons were longitudinally opened and cleaned with cold PBS containing penicillin and streptomycin. The colons were cut into about 2 mm pieces and subsequently transferred them into chelating buffer for 30 min at 4 ℃ [81]. And then the colon segments were vigorously shaken for 1 min to dissociate the epithelial cells. The colon segments were removed by a 100 μm cell-strainer and the colon epithelial cells were collected by centrifuging at 400 × g for 10 min at 4 ℃.

Isolation of lamina propria cells

The methods of lamina propria cells isolated from the colonic lamina propria were described in our previous study [80]. Colonic macrophages and CD4⁺ T cells were purified by positive selection with a magnetic cell separation system by using anti-mouse CD11b Microbeads (Cat#130-049-601, Miltenyi Biotec) and anti-mouse CD4⁺ microbeads (Cat#K1302-10, RWD life science Co.).

Colon culture and cytokine analysis by ELISA

For colon culture and cytokine analysis, the methods were described in our previous study [76].

Measurement of C. rodentium-reactive immunoglobulins

The feces were immersed with 1 mL cold PBS containing protease inhibitor cocktail (Roche) and homogenized in a Tissuelyser-24 for a total of 6 min at 30 Hz. The homogenate was centrifuged and the supernatants were collected. The 96-well ELISA plates were coated with heat-killed C. rodentium for overnight at 4 ℃. After washing 3 times with 1 × PBST, blocking solution (1% BSA in 1 × PBST) was added to the plate and incubated for 1 h at room temperature. After washing 3 times with 1 × PBST, the diluted feces supernatants (1:1000) were then added to the coated plate for 2 h at room temperature. After washing 5 times with 1 × PBST, HRP-conjugated polyclonal goat anti-mouse IgG, IgA, or IgM antibodies (Southern Biotechnology Associates, Birminghan, AL) were added to the plate and incubated for 1 h at room temperature. After washing 5 times with 1 × PBST, 100 μL TMB solution was added to the plate and incubated for 15 min at room temperature. And then 2 N sulfuric acid was added to the plate to stop reaction, and OD₄₅₀ was detected.

Western blot

The proteins from colons, isolated epithelial cells and CT26 cells were extracted in RIPA buffer containing Protease Inhibitor (MedChemExpress, US). Then, the western blot was performed according to previous study [76]. The antibodies were incubated as follows: anti-JMJD1A (Cat#12835–1-AP, Proteintech), anti-STAT1 (Cat#14994, CST), anti-phosphorylation STAT1 (Cat#9167, CST), anti-IRF-1 (Cat#8478S, CST), and anti-β-actin (Cat #A5441, Sigma-Aldrich). Partly protein band intensities were quantified using ImageJ software (National Institutes of Health). The signal intensity of interest protein was normalized to that of its corresponding loading control (β-actin).

Co-IP assays

HEK-293T and CT26 cells were lysed in cell lysis buffer containing protease inhibitor cocktail (Roch) and whole cell lysates were collected in a new 1.5 mL tube. The supernatants were immunoprecipitated with 3 μg of the following antibodies or control immunoglobulin G (IgG) with rotation overnight at 4 °C: anti-JMJD1A (Cat#12835–1-AP, Proteintech), anti-STAT1 (Cat#14994, CST), anti-IRF-1 (Cat#8478S, CST) and Myc-Tag antibody (Cat#AE010, Abcolonal). And then the supernatants were incubated with protein A/G magnetic beads (Cat#HY-K0202, MCE) and rotated for 2 h at 4 °C. After washing three times with lysis buffer, the immunoprecipitants were eluted and boiled for 5 min with 1 × loading buffer. For exogenous Co-IP, the supernatants were incubated with 20 μL anti-flag magnetic beads (Cat#B26102, Biomake). The immunoprecipitants were then performed to western blot.

Gene expression analysis

Total RNA from colons and CT26 cells was extracted using TRIzol (ThermoFisher Scientific, US). Real-time RT-PCR was performed according to our previous study [76]. The primers were as follows: mJMJD1A forward: 5’-TCCCCAGGCAGCCAATT

CTCCA-3’, mJMJD1A reverse: 5’-TGGCTGTGGAGCAGACTCCAGT-3’; mCCL3 forward: 5’- TTCTCTGTACCATGACACTCTGC-3’, mCCL3 reverse: 5’- CGTGGAATCTTCCGGCT

GTAG-3’; mCCL4 forward: 5’- TTCCTGCTGTTTCTCTTACACCT-3’, mCCL4 reverse:

5’-CTGTCTGCCTCTTTTGGTCAG-3’; mCxcl5 forward: 5’-GGTCCACAGTGCCCTACG

-3’, mCxcl5 reverse: 5’-GCGAGTGCATTCCGCTTA-3’; mCxcl9 forward: 5’-AATGC

ACGATGCTCCTGCA-3’, mCxcl9 reverse: 5’-AGGTCTTTGAGGGATTTGTAGTGG-3’;

mCCL8 forward: 5’- CCAGATAAGGCTCCAGTCACC-3’, mCCL8 reverse: 5’- AGAGAGACATACCCTGCTTGGTC-3’; mIL-10 forward: 5’-GGTTGCCAAGCCTTATCG

GA-3’, mIL-10 reverse: 5’-ACCTGCTCCACTGCCTTGCT-3’; mIL-17A forward: 5’-GCTCCAGAAGGCCCTCAGA-3’, and mIL-17A reverse: 5’-CTTTCCCTCCGCATTGA

CA-3’; mIFNγ forward: 5’-TCAAGTGGCATAGATGTGGAAGAA-3’, and mIFNγ reverse: 5’-TGGCTCTGCAGGATTTTCATG-3’; GAPDH forward: 5’-AACTTTGGCATTGTGGAAG

G-3’, GAPDH reverse:5’-GGATGCAGGGATGATGTTCT-3’; L32 forward: 5’-AAGCGAA

ACTGGCGGAAAC-3’, L32 reverse: 5’- TAACCGATGTTGGGCATCAG-3’.

AAV9 packaging

The shuttle plasmids were cotransfected into HEK-293T cells with AAV9-cap plasmids and pHelper. The AAV viral particles were isolated and purified according to a previously reported protocol [82]. For AAV9-mediated CCL8 overexpression and JMJD1A knockdown, 5 × 10¹⁰ viral genomes were injected into female 8-week-old WT mice via portal vein injection before C. rodentium infection. Four weeks after operation, these mice were infected with C. rodentium by oral gavage. JMJD1A knockdown plasmid was constructed with the primers: shJMJD1A forward, 5’-GATCCGAGTATGTGTAGATTGCTATTCAAGAGATAGCAATCT

ACACATACTCTTTTTA-3’, shJMJD1A reverse, 5’-AGCTTAAAAAGAGTATGTGTAGATT

GCTATCTCTTGAATAGCAATCTACACATACTCG-3’, and then inserted into the pAAV-ZsGreen1-shRNA vector at the BamHI and HindIII sites to construct the JMJD1A knockdown plasmid. Mouse CCL8 cDNA was amplified by PCR from colonic epithelial cells with the following primers: forward, 5’-CGCGGATCCAAT

GAAGATCTACGCAGTGCT-3’, and reverse, 5’- CGACGCGTTCAAGGCTGCAGAATTTG

AG-3’ and then inserted into the pAAV-MCS vector at the BamHI and HindIII sites to produce the mouse CCL8 expression plasmid.

16S rRNA sequencing

The fecal samples from naïve WT and JMJD1A^-/- mice were collected, frozen, and sent to Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The Genomic DNA extraction, PCR amplification and Illumina sequencing were performed with extracted samples by Majorbio. Principal coordinate analysis (PCoA) was used to determine the similarity among the microbiota communities in different fecal samples through Vegan v2.5-3 package. To identify the bacteria taxa, which was significantly abundant among different samples, the linear discriminant analysis (LDA) effect size (LEfSe) [83] (http://huttenhower.sph.harvard.edu/LEfSe) was administered (LDA score > 2 and P < 0.05). The raw sequencing reads were deposited into the NCBI Sequence Read Archive (SRA) database (BioProject ID: PRJNA1093258).

RNA-sequencing analysis

RNA-sequencing experiment was performed referring to our previous study [84]. In short, total RNA samples were extracted from the colon tissues of mice, which included WT and JMJD1A KO (JMJD1A^-/-) mice, using TRIzol reagent (Invitrogen). Then, the RNA-sequencing analysis was performed with extracted RNA samples at GENEWIZ Company. Three representative mice were pooled in a sample. Significantly differential genes were selected, and the thresholds were defined as Fold change>1.5 and P < 0.05. The raw sequencing reads were deposited into the NCBI SRA database (BioProject ID: PRJNA1094454).

Construction of plasmids and luciferase reporter assays

The mouse JMJD1A promoter fragments were amplified by PCR using mouse genomic DNA as a template with the following primers: forward, JMJD1A (-1984 bp) 5’-TCCCCCGGGACAAGATGCCAATGTAGCTTTCC-3’, JMJD1A (-1716 bp) 5’-TCC

CCCGGGTCCTTCATTCCATTTCCCACCT-3’, JMJD1A (-1391 bp) 5’-TCCCCCGGGCCT

CTGGTCCAAACCATAAAACC-3’, JMJD1A (-992 bp) 5’-T CCCCCGGGGTCTTCGCTCGA

CAGCCTAC-3’; reverse, JMJD1A (-88 bp) 5’-GGAAGATCTTTGCCTAAGTGTTCGTCC

CC-3’. The fragments were ligated into the SmaI and BglII sites of the PGL3-basic to yield the JMJD1A promoter plasmids. The mouse CCL8 promoter fragment was amplified by PCR using mouse genomic DNA as a template with the following primers: forward, 5’-GGGGTACCGGTGGATGTCTAAAAGTG-3’, and reverse, 5’-TCCC

CCGGGTTTGGAGTGAAGGCT-3’. The fragment was ligated into the KpnI and SmaI sites of the PGL3-basic to yield the JMJD1A promoter plasmids. The mouse STAT1 promoter fragments were also amplified by PCR using mouse genomic DNA as a template with the following primers: forward, STAT1, 5’- TCTATCGATAGGTACAC

TTGGACACTTCAAAAATATGAA-3’, and reverse STAT1,5’-CTTAGATCGCAGATCGT

CCCAAGTGGGTCTGAGGGGCG-3’. CT26 cells were seeded in a 24-well plate and transfected with Renilla plasmisds and indicated plasmids using Lipofectamine 2000 (ThermoFisher Scientific, Waltham, US) according to the manufacturer’s instructions. The luciferase and Renilla values were measured at 24 h after transfection by a Dual luciferase reporter assay kit (Promega, Madison, US) according to the manufacturer’s instructions. The Renilla served as an internal control.

Chromatin immunoprecipitation (ChIP) assay

ChIP assays were performed based on the protocol described by Abcam (Cambridge, MA). The sheared chromatin was immunoprecipitated with anti-JMJD1A (Cat#12835–1-AP, Proteintech), anti-IRF-1 (Cat#8478S, CST), anti-STAT1 (Cat#14994, CST), anti-H3K9me2 (Cat#ab1220, Abcam) or nonspecific IgG (Santa Cruz, Dallas, US) antibodies. ChIP DNA fragments were analyzed by qRT-PCR with specific primers for the JMJD1A or CCL8 promoter. The following primers were used: JMJD1A promoter IRF1 binding site, forward, 5’-CTCTGGTC CAAACCATAAAACCA-3’, and reverse, 5’-TGAGAAAGTAGATGGAGTTG

ATGA-3’; IRF1 promoter STAT1 binding site, forward, 5’-AGGTACCCATCACTAACA

AGGC-3’, and reverse, 5’- CCTACTGAAAGCAACCAGAGACTA-3’; CCL8 promoter IRF1 binding site, forward, 5’-TCACTTACGACGACAGTTGGG-3’, and reverse, 5’- TCAAATGCCTAAGACAGTTTTCAA-3’.

Transwell migration assays

4 × 10⁵ RAW264.7 cells in 100 μL of DMEM or 2 × 10⁵ CTLL-2 cells in 100 μL of RIMP-1640 supplemented with penicillin (100 IU) and streptomycin (100 mg mL^-1) were placed in upper chamber of transwell containing permeable polyester membrane (8 μm pores) pretreated with calcein-AM (Yeasen, Shanghai, China). In the lower chamber, 500 μL of RPMI-1640 medium without FBS containing different treatments (siNC, siIRF1, siIRF1 plus PLV-mCCL8, siJMJD1A, siJMJD1A plus PLV-mCCL8 and PLV-mCCL8), CT26 cells were placed and incubated at 37 ℃ for 4h. The number of cells transmigrated into the bottom chamber was counted using the Fluorescence Microscope (Nikon TE2000) for each treatment condition.

Statistical analysis

Bioinformatic analysis of the gut microbiota was carried out using the online platform of Majorbio Cloud platform (www.majorbio.com). Data analysis was performed by using GraphPad Prism 8.0 software. Data shown in graphs are Mean ± SEM or SD. Statistical analyses were calculated with GraphPad Prism 8.0. P < 0.05 was considered statistically significant. P values were assessed with the two-tailed Student’s t-test or nonparametric tests, one-way ANOVA or two-way ANOVA test. The survival curve was assessed with the log-rank test.