An essential mesenchymal function for miR-143/145 in intestinal epithelial regeneration

Abstract

Downregulation of the miR-143/145 microRNA (miRNA) cluster has been repeatedly reported in colon cancer and other epithelial tumors. In addition, overexpression of these miRNAs inhibits tumorigenesis, leading to broad consensus that they function as cell-autonomous epithelial tumor suppressors. We generated mice with deletion of miR-143/145 to investigate the functions of these miRNAs in intestinal physiology and disease in vivo. Although intestinal development proceeded normally in the absence of these miRNAs, epithelial regeneration after injury was dramatically impaired. Surprisingly, we found that miR-143/145 are expressed and function exclusively within the mesenchymal compartment of intestine. Defective epithelial regeneration in miR-143/145-deficient mice resulted from the dysfunction of smooth muscle and myofibroblasts and was associated with derepression of the miR-143 target Igfbp5, which impaired IGF signaling after epithelial injury. These results provide important insights into the regulation of epithelial wound healing and argue against a cell-autonomous tumor suppressor role for miR-143/145 in colon cancer.

Figure 1. Lethal failure of intestinal regeneration in miR-143/145^−/− mice

(A) Northern blot analysis of bladder miRNA expression in mice of indicated genotypes. (B) H&E-stained colon sections of adult miR^+/+ and miR^−/− mice. Representative images from at least 5 mice of each genotype are depicted. (C) Proliferation of ulcer-adjacent crypts assessed by H&E (upper panels), Ki67, and pH3 staining. Histology (H&E) was examined in over 20 mice of each genotype while a subset of sections (n ≥ 3) were further stained for Ki67 and pH3. (D) Kaplan-Meier survival curves of miR^+/+, miR^+/−, and miR^−/− mice, administered 3.5% DSS in drinking water for 5 days. *, p<0.0001, log rank test. (E) miR^+/+ and miR^−/− animals dissected 2 days after completion of DSS treatment. Images are representative of over 20 mice examined per genotype. See also Figures S1–S3.

Figure 2. miR-143/145 are exclusively expressed in the intestinal mesenchyme

(A) In situ hybridization analysis of miR-143 and miR-145 expression in mouse colon. Representative images from 3 animals per genotype shown. Blue, miRNA; red, nuclear fast red counterstain. (B) In situ hybridization (left panels) showing miR-143/145-expressing pericryptal cells located in a position consistent with SMA+ lamina propria myofibroblasts (right panel). (C) Quantitative RT-PCR analysis of mRNA and miRNA expression in full-thickness mouse colon specimens and purified epithelium. The purity of epithelial preparations was demonstrated by enrichment of the epithelial markers Villin (Vil) and E-cadherin (Cdh1) and near complete depletion of the mesenchymal transcripts smooth muscle actin (Sma) and vimentin (Vim). n ≥ 3 samples per condition. For this and all subsequent figures, error bars represent standard deviations. (D) mRNA and miRNA expression in human colorectal tumor and paired normal colon specimens and in corresponding purified epithelium. n=10 samples per condition. *, p<0.05 (Student’s t-test). (E) Quantitative RT-PCR analysis of miR-143/145 expression in purified intestinal epithelial cells (IEC), isolated epithelial stem cells (Lgr5+), and ISEMFs, compared to full-thickness colon specimens. See also Figure S4.

Figure 3. Mesenchymal, but not epithelial, miR-143/145 deletion phenocopies germline deletion

(A) X-gal-stained colon sections demonstrating domains of Cre-mediated recombination in mice of the indicated genotypes. Blue, LacZ expression; red, nuclear fast red counterstain. (B) Colons from mice of the indicated genotypes, dissected 2 days after completion of DSS treatment. Representative images from 5–10 examined mice per genotype. (C) Kaplan-Meier survival curves of mice of indicated genotypes, administered 3.5% DSS in drinking water for 5 days. *, p=0.001, log rank test. (D) Proliferation of ulcer-adjacent crypts assessed by H&E (upper panels), Ki67, and pH3 staining. Histology (H&E) was examined in 5–10 mice of each genotype while a subset of sections (n ≥ 3) were further stained for Ki67 and pH3. (E) Quantification of the average number of pH3+ cells within 500 µm of deepthelialized zones. Multiple ulcer-adjacent areas in at least 3 mice per genotype were quantified. **, p<0.01 (Student’s t-test). See also Figure S5.

Figure 4. Defective epithelial regeneration of Myh11-Cre/eGFP; miR^−/flox mice

(A) Proliferation of ulcer-adjacent crypts in mice of the indicated genotypes assessed by H&E (upper panels), Ki67, and pH3 staining. Histology (H&E) was examined in 8–24 mice of each genotype while a subset of sections (n ≥ 3) were further stained for Ki67 and pH3. (B) Quantification of the average number of pH3+ cells within 500 µm of deepthelialized zones. Multiple ulcer-adjacent areas in at least 3 mice per genotype were quantified. **, p<0.01 (Student’s t-test). See also Figure S6.

Figure 5. Disorganized myofibroblasts are associated with defective epithelial regeneration

(A) SMA immunofluorescence showing muscularis mucosa and pericryptal myofibroblasts in uninjured miR^+/+ and miR^−/− colons. Images representative of ≥10 animals of each genotype. (B–C) SMA immunofluorescence showing muscularis mucosa and pericryptal myofibroblasts in ulcerated regions of colon from mice of the indicated genotypes 2 days after completion of DSS treatment. Images representative of ≥10 miR^+/+ and miR^−/− animals and ≥3 animals of the other genotypes. (D) GFP fluorescence showing labeled smooth muscle cells and their progeny in Myh11-Cre/eGFP; mTmG mice (right panels) before and after DSS injury, compared to Myh11-Cre/eGFP alone (left panels). White arrowheads show representative smooth muscle-derived cells that have migrated into the lamina propria. Images representative of 3 animals per genotype. See also Figure S6.

Figure 6. De-repression of Igfbp5 in miR-143/145^−/− myofibroblasts in vitro and in vivo

(A) Quantitative RT-PCR analysis of Igfbp5 expression in cultured ISEMFs with or without LPS treatment and in whole mouse colons with or without DSS administration. n= 3 independently-derived WT or KO ISEMF cell lines or animals per condition. *, p<0.05 (Student’s t-test). (B) Sequence and evolutionary conservation of the miR-143 binding site in the 3’ UTR of Igfbp5. Mutations introduced into luciferase reporters (panel C) are indicated in red. (C) Relative firefly luciferase activity of reporter constructs containing the miR-143 binding site or its mutated version following transfection into 293T cells alone or with control or miR-143 or miR-145 synthetic miRNA mimics. n=3 replicates per condition. *, p<0.05; **, p<0.01 (Student’s t-test). (D) In situ hybridization analysis of Igfbp5 expression in DSS-treated miR^+/+ and miR^−/− colon sections. Upper panels, uninjured regions; lower panels, ulcerated regions with residual crypts on the left. n=7–8 mice per genotype examined. Blue, Igfbp5; red, nuclear fast red counterstain. (E) Phospho-Igf1r (pIGF1R) immunofluorescence of ulcer-adjacent crypts in DSS-treated miR^+/+ and miR^−/− colons. Quantification of membrane fluorescence from n=4 mice of each genotype is shown in graph on right. Arrowheads, baso-lateral membranes of ulcer-adjacent crypts with (solid) or without (open) activated Igf1r. **, p<0.01 (Student’s t-test). See also Figure S7.

Figure 7. Model of miR-143/145 action in intestinal epithelial regeneration

Upon intestinal injury, lamina propria myofibroblasts derived from multiple sources including smooth muscle participate in the wound healing response. In the absence of miR-143/145 (right side of figure), excessive secretion of IGFBP5 by myofibroblasts results in sequestration of IGF ligand and diminished IGF pathway activation within epithelial cells, likely contributing to the failure of epithelial cells to activate the regenerative proliferation program.