Tumor fibroblast-derived epiregulin promotes growth of colitis-associated neoplasms through ERK

Abstract

Molecular mechanisms specific to colitis-associated cancers have been poorly characterized. Using comparative whole-genome expression profiling, we observed differential expression of epiregulin (EREG) in mouse models of colitis-associated, but not sporadic, colorectal cancer. Similarly, EREG expression was significantly upregulated in cohorts of patients with colitis-associated cancer. Furthermore, tumor-associated fibroblasts were identified as a major source of EREG in colitis-associated neoplasms. Functional studies showed that Ereg-deficient mice, although more prone to colitis, were strongly protected from colitis-associated tumors. Serial endoscopic studies revealed that EREG promoted tumor growth rather than initiation. Additionally, we demonstrated that fibroblast-derived EREG requires ERK activation to induce proliferation of intestinal epithelial cells (IEC) and tumor development in vivo. To demonstrate the functional relevance of EREG-producing tumor-associated fibroblasts, we developed a novel system for adoptive transfer of these cells via mini-endoscopic local injection. It was found that transfer of EREG-producing, but not Ereg-deficient, fibroblasts from tumors significantly augmented growth of colitis-associated neoplasms in vivo. In conclusion, our data indicate that EREG and tumor-associated fibroblasts play a crucial role in controlling tumor growth in colitis-associated neoplasms.

Figure 1. Growth features of sporadic and colitis-associated tumors.

(A) Endoscopic images of patients without IBD. The left panel shows noninflamed colonic mucosal the right panel demonstrates a polypoid precursor lesion of a sporadic CRC with HGD, as confirmed by histopathological analysis. (B) Endoscopic pictures of patients suffering from UC. The right image demonstrates moderate intestinal inflammation (Mayo score 2); the left image displays a flat-appearing CAC, as confirmed by histopathological analysis. All images were acquired from the distal sigmoid colon, as documented by the colonoscopist. (C and D) Pictures from corresponding mouse models: noninflamed colon (C, left image), sporadic colorectal tumors in the Apc^min/+ model (C, right image), intestinal inflammation in the DSS model of colitis (D, right image), and colitis-associated neoplasms in the AOM/DSS model (D, left image). Scale bars: 2 mm.

Figure 2. Differential gene expression studies in mouse models of sporadic and colitis-associated tumors identify EREG as a potential target.

(A) Expression patterns from experimental models of CAC, sporadic CRC, and control epithelia were groupwise compared as described in Methods. (B) Distribution of transcripts from differential expression analysis (CAC vs. sporadic CRC) are displayed in a volcano plot (transcripts with P < 0.05 and FC > 2 in yellow). (C) The 25 significantly expressed genes with highest up- or downregulation are derived from differential expression analysis between CAC and sporadic CRC. (D) GO enrichment analysis was performed, as described in Methods. The displayed GO terms are derived from the genes that were differentially expressed with upregulation in the CAC model. Proliferat, proliferation; resp, response; oxid, oxidative; sign, signaling; rec, receptor; pathw, pathway. (E) Distribution of probe sets from differential expression analysis (CAC vs. control, sporadic CRC vs. control) are shown in volcano plots as indicated (transcripts with P < 0.05 and FC > 2 in yellow). (F) Gene lists from various differential expression analyses were compared with each other as visualized by Venn diagrams (blue, CAC vs. sporadic CRC; bright yellow, CAC vs. control; gray, sporadic CRC vs. control). The numbers of probe sets per group are displayed.

Figure 3. Human precursor lesions of colitis-associated tumors display significantly elevated EREG expression.

(A) H&E stainings of paraffin-embedded tissue sections from colorectal tumors and control tissue, as indicated. Scale bars: 50 μm. (B) Immunostaining for human EREG in the same samples as in A. The analysis included 47 samples (non-IBD control group, n = 7; sporadic HGD, n = 12; sporadic CRC, n = 11; colitis-associated HGD, n = 9; CAC, n = 8). Scale bars: 50 μm. (C) For quantitative analyses, 4–5 representative high power fields (HPF) per sample were scored for the number of positive cells. Original magnification, ×40. Data are shown as mean values per group ± SD. Significant differences are indicated. *P < 0.05; **P < 0.01.

Figure 4. Colitis-associated tumorigenesis in Ereg-deficient mice.

(A–C) Colitis-associated tumors were induced in WT controls and Ereg^–/– mice by AOM and 3 cycles of DSS. Tumor development in the rectosigmoid was assessed by multispectral fluorescence imaging of integrin α_Vβ₃, stereomicroscopic inspection, and histopathological analysis at day 80 as shown by representative pictures of each group. Scale bars: 400 μm. Both images in C are composites of multiple overlapping images to allow a broader field of view. (D and E) Total tumor numbers and average tumor load per mouse were quantified in WT controls and Ereg^–/– mice. Data are shown as mean values ± SD of 8–10 mice per group, and similar results were obtained in 3 independent experiments. *P < 0.05; **P < 0.01.

Figure 5. EREG promotes growth rather than initiation of colitis-associated tumors.

(A and B) Serial analyses of AOM/DSS tumors was performed in controls and Ereg^–/– mice by endoscopy, as indicated. Tumor load development over time is shown by representative endoscopic pictures and average tumor load progression analysis. (C) The relative contribution of different tumor sizes (s1 = small, s5 = very large) to the total tumor load in WT and Ereg^–/– mice was determined. The serial analysis was performed with 5 mice per group and was repeated twice with similar results. Data are shown as mean values ± SD. **P < 0.01.

Figure 6. EREG facilitates intestinal wound healing in vitro and in vivo.

(A) Experimental colitis was induced by 3 cycles of DSS. EREG expression was measured in the distal colon by qPCR. Data are mean values ± SD of 2–4 animals per data point. Similar results were obtained in 2 independent experiments. (B and C) Gaps were scratched into monolayers of IEC. PBS or EREG (100 ng/ml) was applied, and the wound gap closure was analyzed by microscopy after 18 hours. Representative images from 1 experiment out of 3 are shown. Scale bars: 100 μm. Data shown as mean values ± SD. *P < 0.05. (D and E) In vivo wounding was performed by targeted biopsies in control (n = 4) and Ereg^–/– (n = 4) mice. The mucosal healing was monitored by endoscopy every 1–2 days. Representative images are displayed from days 0, 3, and 5. The experiment was repeated twice with similar results. Arrows highlight the margins of the mucosal lesions. *P < 0.05. (F) Cross sections of the in vivo wounded distal colon (d3, left panel) from EREG-reporter mice (Ereg^+/–) were analyzed by X-gal staining. The dotted line indicates the wound gap. Control tissue (right panel) was obtained from the unwounded distal colon of the same mouse. Numerous EREG-expressing cells (arrows) could be identified adjacent to the wound but not in control tissue. Scale bars: 100 μm.

Figure 7. EREG directly increases proliferation in IECs and promotes colon tumor growth.

(A) Full-thickness rectosigmoidal pieces were incubated with PBS or EREG (500 ng/ml) and analyzed by immunohistochemistry for Ki67. Representative images out of 3 experiments are shown. Scale bars: 100 μm. (B) Staining was quantified by counting of positive cells per crypt. *P < 0.05. (C) IEC were grown in 12-well plates with varying concentrations of EREG. Cell numbers were determined after 24 and 48 hours. The experiment was performed in duplicate and repeated 3 times with similar results. *P < 0.05; **P < 0.01. (D–F) Rectosigmoidal tumors were induced in WT animals by AOM and 1 cycle of DSS. Starting at day 30, repeated injections of PBS (n = 5), recombinant EREG (n = 5), EGF (n = 3), or AREG (n = 3) were performed in the base of small tumors 4 times in 10 days. The change in tumor load was assessed by endoscopy at day 45. Data from 1 out of 2 experiments are shown and represent mean values ± SD. *P < 0.05.

Figure 8. Fibroblasts are the main producers of EREG in colitis-associated tumors.

(A) Consecutive rectosigmoidal cross sections (nos. 1–6) of 4-μm thickness from AOM/DSS-tumors of lacZ-Ereg-reporter mice were stained with X-gal plus eosin (left panels) or with specific antibodies against fibroblast markers as indicated (right panels). Representative brightfield and confocal immunohistochemical images are shown. Arrows indicate double-positive cells in consecutive sections. Scale bars: 25 μm. (B) Tumor cells with positive X-gal staining were quantified for colocalization with markers VIM, FSP1, and PDGFR-β of cancer-associated fibroblasts (CAF). Data represent mean values ± SD from 5 tumors. (C) Human cross sections from a patient with UC and HGD were stained for EREG and VIM (left panel) or with an antibody control (right panel) and studied by confocal laser microscopy. Scale bars: 25 μm. (D and E) Fibroblasts were purified from AOM/DSS tumors of Ereg^+/– mice as specified in Methods and grown in cell culture. At day 5, cells were analyzed for α-SMA, VIM, FSP1, PDGFR-β, and VCAN by immunocytochemistry as indicated. Additionally, EREG expression was determined by X-gal staining. Scale bars: 50 μm. (F and G) Fibroblasts from AOM/DSS tumors were isolated and cultured with PBS, EREG (100 ng/ml), LPS (1000 ng/ml), LTA (1000 ng/ml), TNF-α (100 ng/ml), or IL-6 (100 ng/ml) for 6 hours. EREG expression was measured by qPCR. Data represent mean values ± SD (n = 2–3 per group). Similar results were obtained in 2 independent experiments. *P < 0.05; **P < 0.01.

Figure 9. Adoptive transfer model: fibroblast-derived EREG promotes the growth of colitis-associated tumors.

(A) IEC were grown in 48-well plates with conditioned medium from Ereg^–/– or WT tumor fibroblasts, and cell numbers were determined after 24 and 48 hours. The experiment was performed in triplicate and repeated twice with similar results. *P < 0.05. Data represent mean values ± SD. (B and C) Fibroblasts from AOM/DSS tumors of heterozygous lacZ-Ereg-reporter mice were endoscopically injected into a tumor of a WT mouse. At day 2 after injection, the tumor was harvested, and cross sections were stained for X-gal. EREG-expressing fibroblasts were identified in the tumor stroma upon cell transfer (arrows). Scale bars: 100 μm. (D–F) Ereg^–/– and WT mice were treated with AOM and 1 cycle of DSS. Starting at day 30, repeated injections (4 times in 15 days) of tumor fibroblasts from Ereg^–/– or WT animals were performed into the base of small tumors of Ereg^–/– mice (n= 3 per group). The change in tumor load was endoscopically monitored until day 50. Data are from 1 out of 2 independent experiments with similar results and represent mean values ± SD. *P < 0.05; **P < 0.01.

Figure 10. EREG promotes the development of colitis-associated tumors via activation of ERK.

(A) CMT93 cells were cultured with EREG (100 ng/ml) or PBS. Cell lysates were quantified for phospho-ERK1/2, total ERK1/2, and β-actin by Western blots. (B) IEC were cultured in the presence of PBS or EREG (100 ng/ml) and ERK inhibitors PD98059 (1 μg/ml) or U0126 (40 ng/ml), as indicated. Cell numbers were determined after 24 and 48 hours. The experiment was performed twice with similar results. *P < 0.05. (C) Freshly harvested tumors of sizes 3–4 were cut into pieces and equally distributed into tissue culture wells. Next, tumor tissue was cultured with EREG and ERK inhibitors (U0126, PD98059). After 60 minutes, tumors were lysed and quantified for phospho-ERK1/2, total ERK1/2, and β-actin by Western blotting. The experiment was repeated twice with similar results. (D–F) Colitis-associated tumors were induced in WT mice by AOM and 1 cycle of DSS. Apc^min/+ mice were screened at the age of 6 weeks for the presence of sporadic rectosigmoidal tumors. Endoscopically guided injections of PBS, EREG, or ERK inhibitor (U0126) were made into growing tumors 4 times in 10 days starting at day 30. The change of tumor load was monitored by endoscopy until day 45. Data are from 1 out of 3 experiments and represent mean values ± SD (n = 3–5 per group). *P < 0.05.