Epimorphin deletion inhibits polyposis in the Apc^min/+ mouse model of colon carcinogenesis via decreased myofibroblast HGF secretion

Abstract

Interactions between the epithelium and surrounding mesenchyme/stroma play an important role in normal gut morphogenesis, the epithelial response to injury, and epithelial carcinogenesis. The tumor microenvironment, composed of stromal cells including myofibroblasts and immune cells, regulates tumor growth and the cancer stem cell niche. Deletion of epimorphin (Epim), a syntaxin family member expressed in myofibroblasts and macrophages, results in partial protection from colitis and from inflammation-induced colon cancer in mice. We sought to determine whether epimorphin deletion protects from polyposis in the Apc^min/+ mouse model of intestinal carcinogenesis. Epim^−/− mice were crossed to Apc^min/+ mice; Apc^min/+ and Apc^min/+/Epim^−/− mice were killed at 3 mo of age. Polyp numbers and sizes were quantified in small intestine and colon, and gene expression analyses for pathways relevant to epithelial carcinogenesis were performed. Primary myofibroblast cultures were isolated, and expression and secretion of selected growth factors from Apc^min/+ and Apc^min/+/Epim^−/− myofibroblasts were examined by ELISA. Small bowel polyposis was significantly inhibited in Apc^min/+/Epim^−/− compared with Apc^min/+ mice. Apc^min/+/Epim^−/− compared with Apc^min/+ polyps and adjacent uninvolved intestinal mucosa had increased transforming growth factor-β (TGF-β) expression and signaling with increased P-Smad2/3 expression. Myofibroblasts isolated from Apc^min/+/Epim^−/− vs. Apc^min/+ mice had markedly decreased hepatocyte growth factor (HGF) expression and secretion. We concluded that Epim deletion inhibits polyposis in Apc^min/+ mice, associated with increased mucosal TGF-β signaling and decreased myofibroblast HGF expression and secretion. Our data suggest that Epim deletion reduces tumorigenicity of the stromal microenvironment.

the mammalian intestine contains a highly proliferative epithelium that is completely renewed and replaced every 3–5 days. The crypts of Lieberkuhn contain stem cells that give rise to proliferating daughter/progenitor cells, which differentiate into the four major cell types of the intestine. Surrounding stromal cells, including myofibroblasts, play a critical role in modulating epithelial cell proliferation (19, 22) and carcinogenesis (34). Stromal myofibroblasts that encircle the intestinal crypts contribute to maintaining the normal stem cell niche (19, 26, 32, 37), and tumor-associated myofibroblasts play a critical role in promoting epithelial carcinogenesis, in part by regulating cancer stem cells (27, 31, 34) and enhancing tumor growth (2, 16). Myofibroblasts enhance Wnt activity and “stemness” of adjacent cells in colon cancers; colonic myofibroblasts from colon cancers modulate Wnt activity in adjacent tumor cells, inducing a tumorigenic stem cell phenotype (34). Bone marrow-derived myofibroblasts promote tumor growth in gastric cancer models (27).

To better understand the role of myofibroblasts in regulating proliferation and carcinogenesis of the gut epithelium, we have studied epimorphin (Epim), a member of the syntaxin family of vesicle-docking proteins expressed in gut myofibroblasts (10, 33, 35). Syntaxins are SNARE complex structural proteins that are required for membrane fusion events. One of the best studied, syntaxin 1, functions as part of a target or t-SNARE complex to mediate neurotransmitter release in synapses (11). We have shown that Epim, or syntaxin 2, plays a role in regulating myofibroblast secretion of growth factors and cytokines (33). Epim is also unique among the syntaxins; in addition to its intracellular function as a vesicle-docking protein that regulates cellular secretion, it also has an extracellular location (6, 12, 29). Epim has been shown to bind to αV-integrins and regulate transcription of genes involved in branching morphogenesis and fibrosis (6, 12). For example, exposure of mammary epithelial cells to Epim resulted in increased expression of genes associated with branching morphogenesis in the mammary gland, such as matrix metalloproteinase (MMP)-3, and downregulation of luminal epithelial markers, such as keratin-16 (6). Also, administration of an anti-Epim-blocking antibody to mice in a kidney injury model resulted in increased transforming growth factor-β (TGF-β) and decreased MMP-2 and MMP-9 expression (36).

Stromal Epim has also been postulated to play a role in epithelial carcinogenesis. We previously generated Epim^−/− mice and showed that Epim deletion results in partial protection from colitis (35) and from inflammation-induced carcinogenesis in an azoxymethane/dextran sodium sulfate model (33). Epim plays a role in carcinogenesis in other epithelial organs; for example, ectopic overexpression of Epim in hepatocellular carcinoma cells results in increased invasion and metastasis in an orthotopic liver implantation model, associated with increased MMP-9 expression (14).

Herein, we further address the hypothesis that myofibroblast Epim affects tumor growth and epithelial-stromal microenvironment interactions in the gut, by examining the effect of Epim deletion on tumorigenesis in the Apc^min/+ mouse model of colon carcinogenesis. These mice, which contain a mutation in the adenomatous polyposis coli gene, develop multiple adenomatous polyps in the small bowel; the Apc^min/+ strain of mice utilized in the present studies has few colonic polyps. We show that Epim deletion reduces the number of small bowel polyps in specific regions of the gut and acts by regulating expression of two key stromal modulators of epithelial carcinogenesis, including increasing TGF-β expression and downstream signaling pathways, and inhibiting hepatocyte growth factor (HGF) expression and secretion from myofibroblasts.

MATERIALS AND METHODS

Animals

All mice were housed in a barrier mouse facility with a 12-h:12-h light/dark cycle and free access to water and to standard rodent chow diet (PicoLab 20; Purina, St. Louis, MO). All animal experimentation was approved by the Animal Studies Committee of Washington University School of Medicine.

Apc^min/+ and Apc^min/+/Epim^−/− mice.

All mice in this study are in the C57BL/6J background. Two-month-old male Apc^min/+ (C57BL/6J/Apc^min/+) mice were purchased from the Jackson Laboratory (Bar Harbor, ME) and were housed at the Washington University School of Medicine animal facility for breeding; all progeny used to generate Apc^min/+/Epim^−/− mice and for breeding were housed in the same mouse room. Apc^min/+ mice were mated to C57BL/6J wild-type Epim^+/+ (WT) females raised in the Washington University School of Medicine mouse facility or to C57BL/6J/Epim^+/− female mice, as detailed below. All mice used for these experiments were born in the Washington University animal facility and were housed in the same room from birth.

Epim^−/− mice (with global deletion of Epim) were generated as previously described (35) and backcrossed to C57BL/6J mice to congenicity by speed congenics performed in the Murine Models Core of the Digestive Diseases Research Core Center of Washington University School of Medicine. C57BL/6J Apc^min/+ males were crossed with C57BL/6J Epim^+/+ or C57BL/6J Epim^+/− female mice, generating WT, Apc^min/+, and Apc^min/+/Epim^+/− mice; male Apc^min/+/Epim^+/− mice were then bred to female Epim^+/− or Epim^−/− mice to generate Apc^min/+/Epim^−/− mice, all in a C57BL/6J background.

All littermates were screened for the presence of the Min mutation (9) and for epimorphin genotype by PCR assay. PCR was performed using the Eppendorf Mastercycler Gradient thermocycler. Total tail DNA was isolated with EZHMW mouse tail DNA isolation kit (EZBioResearch, St. Louis, MO).

Intestinal Polyp Quantification

Polyp numbers and polyp sizes.

Apc^min/+ and Apc^min/+/Epim^−/− female mice were killed at 3 mo of age (n = 7 mice per group). The entire small intestine and colon were removed. Duodenum was identified and separated distal to the pylorus up to the ligament of Treitz. The remaining small intestine was divided into three equal segments that were designated proximal jejunum, distal jejunum, and ileum. All small intestine fragments and colon were cut longitudinally, washed with PBS, and pinned. Tissue was fixed overnight with 10% buffered formalin, followed by 70% ethanol. Polyps in each segment were counted, and polyp size was quantified. Polyp numbers were counted by visualizing intestines using a Nikon SMZ 800 inverted stereo dissecting microscope at ×10 magnification. Polyp size was quantified by visualizing intestinal segments using a Nikon SMZ 800 inverted microscope, and polyps were photographed using a Photometrics CoolSnap CCD camera using Metavue 6.3 (Molecular Devices, Sunnyvale, CA). Polyp size was quantified using Image J1.42q in Java (National Institutes of Health, Bethesda, MD).

Quantitative Real-Time RT PCR

Total RNA was isolated from polyps and adjacent, grossly uninvolved proximal and distal intestinal segments of 4-mo-old female Apc^min/+ and Apc^min/+/Epim^−/− mice, using Trizol Reagent (Invitrogen, Carlsbad, CA) followed by treatment with RNase-free DNase I (Ambion, Austin, TX).

First-strand CDNA was synthesized from 1 μg of total RNA using SuperScript II Reverse Transcriptase (Invitrogen), with random primers, dNTP, and RNase OUT Ribonuclease Inhibitor (Invitrogen). The absence of DNA contamination in final RNA samples was proven by 18S amplification of cDNA without adding reverse transcriptase enzyme to the reaction.

Quantitative real-time RT-PCR with SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) was performed on ABI Prism 7000 Sequence Detection apparatus (Applied Biosystems). Oligonucleotide primer sequences are listed in Table 1.

Table 1.

Primers for qRT-PCR

Gene	Forward Primer	Reverse Primer
18S	5′-TCG AGG CCC TGT AAT TGG AA-3	5′-CCC TCC AAT GGA TCC TCG TT-3′
mTGF-β1	5′-GCA GTG GCT GAA CCA AGG A-3′	5′-AGC AGT GAG CGC TGA ATC G-3′
mTGF-β2	5′-CCA CCT CCC CTC CGA AAA-3	5′GAG ACA TCA AAG CGG ACG ATT C–3′
mTGF-β3	5′-CAA TTA CTG CTT CCG CAA CCT–3′	5′-GCC TAG ATC CTG CCG GAA GT-3′
mHGF	5′-CCT GAC ACC CCT TGG GAG TA-3′	5′-TCC ATA GGG ACA TCA GTC TCA TTC-3′

TGF, transforming growth factor; HGF, hepatocyte growth factor.

Apc^min/+ genotyping.

For each reaction, three primer sets were used: 4 μM MApc MT (5′-TTCTGAGAA AGACAGAAGTTA-3′), 1 μM MApc 15 (5′-TTCCACTTTGGCATAAGG-3′), and 0.02 μM MApc 9 (5′-GCCATCCCTTCACGTTAG-3′); 1× Tris·HCl reaction buffer (Sigma, St. Louis, MO), 3.5 μM MgCl₂, 0.5 mM d NTP mix, 0.1% Triton X-100, 0.01 U Sigma Red Taq DNA Polymerase, and 1 μl total tail DNA were used. PCR amplification was as follows: denaturation for 3 min at 94°C, followed by 34 cycles (94°C for 0.5 min, 60°C for 2 min, 72°C for 3 min), and final extension at 72°C for 10 min. Electrophoresis in a 2% agarose gel with ethidium bromide in 1× Tris-acetate-EDTA buffer exhibited 618-bp WT and 327-bp Min mutant product sizes.

Epimorphin genotyping.

For each reaction, three primer sets were used: 0.8 μM WT (5′-TTAGATTCCCTCCTAGAGACGC-3′), 0.4 μM COM (5′-GACAGTGTGTTAACTAGTCAAGC-3′), and 0.2 μM Epim^−/− (5′-GGACACGCTGAACTTGTGGC-3′); 1× Tris·HCl reaction buffer (Sigma), 1.2 mM betaine (Sigma), 4 mM dNTP mix (Invitrogen), 0.04 U Red Taq DNA Polymerase (Sigma), and 1 μl total tail DNA were used. Amplification (94°C denaturation for 3 min, followed by 34 cycles of 94°C for 1 min, 60°C for 0.5 min, 72°C for 3.5 min) yields 582-bp WT and 398-bp Epim^−/− products.

Myofibroblast isolation and culture.

Myofibroblasts were isolated from ileum of 8-day-old Apc^min/+ and Apc^min/+/Epim^−/− mice as described (25, 33). Briefly, intestines were removed and placed into Hanks balanced salt solution (HBSS, Invitrogen), cut longitudinally to remove stool, and minced into ∼10 2–3-mm fragments. The fragments were washed eight times in HBSS and then treated with collagenase XI (300 U/ml, Sigma) and dispase (0.1 mg/ml, Invitrogen) for 25 min at room temperature. After the pieces were further minced smaller, fragments were washed with HBSS 1× and centrifuged at 200 g for 10 min. Pellet debris was washed five times with DMEM (Invitrogen) with 2% sorbitol. Myofibroblast culture medium was composed of DMEM with 10% FBS (Invitrogen), epidermal growth factor (2 ng/ml; Sigma), insulin (10 μg/ml; Sigma), transferrin (10 μg/ml; Roche, Indianapolis, IN), and gentamicin (10 μg/ml; Sigma). Fragments were plated in the culture medium and incubated at 37°C. Culture medium was changed on day 3 after isolation, and cells were passaged after washing in PBS (2×) with trypsin (0.05%)/EDTA (0.53 mM). Trypsin was inactivated by adding 1.0 ml of culture medium with serum. Cells were used for analysis between passages 3 and 4. At least three to four independent primary cultures from individual mice were established for each genotype. Absence of epimorphin expression was confirmed in Apc^min/+/Epim^−/− myofibroblasts by qRT-PCR and immunoblot as done previously (33).

Myofibroblast characterization.

Myofibroblasts from each primary culture from each genotype (n = 3–4 per genotype) were grown on four-well plastic slides and fixed in formalin. To confirm the myofibroblast phenotype [defined as cells that are positive for α-smooth muscle actin (SMA) and vimentin, negative or weakly positive for desmin, and negative for cytokeratin], slides were incubated with either rabbit polyclonal anti-α-SMA (1:200; Abcam, Cambridge, MA), anti-desmin (D8281, 1:20 dilution; Sigma), anti-cytokeratin 18 (1:100 dilution; Millipore, Billerica, MA), or antivimentin (1:20 dilution; Sigma) antibodies and detected with anti-rabbit fluorescein (FITC)- or rhodamine (TRITC)-conjugated antibodies. Cells were mounted in Vectashield Mounting Medium with DAPI and examined under a fluorescence microscope (Axioskop 2; Zeiss Microscopes, Jena, Germany).

HGF expression and secretion from myofibroblasts.

Myofibroblasts from three to four independent primary myofibroblast cultures isolated from intestines of Apc^min/+ or Apc^min/+/Epim^−/− mice were plated (3 × 10⁵ cells per well in 6-well plates) and grown for 48 h in myofibroblast culture media containing FBS. Media was removed and replaced with serum-free (SF) myofibroblast media, SF myofibroblast media containing LPS (5 μg/ml; List Biological Laboratories, Campbell, CA), or myofibroblast media with 10% FBS. After 24 h, media was removed and frozen, and cells were harvested for total RNA and protein isolation. HGF mRNA expression was measured by qRT-PCR. Myofibroblast cellular and secreted HGF protein was measured by ELISA of cell lysate and media (HGF mouse ELISA kit, Abcam), per the manufacturer's instructions. Epim^−/− or WT myofibroblasts were also plated as above and incubated with SF myofibroblast media or SF myofibroblast media containing TGF-β (10 ng/ml) or IL-1β (10 ng/ml), and media was collected after 24 h to quantify secreted HGF by ELISA. Each sample for ELISA was repeated in duplicate, and each experiment was performed at least twice.

Immunohistochemical Analyses

Intestines were fixed in 10% formaldehyde overnight, dehydrated, embedded in paraffin, and sectioned (8 μm). Sections were deparaffinized, rehydrated, and washed with PBS. Antigen retrieval was performed in heat retrieval solution (Diva Cloaker- Biocare Medical, Concord, CA) for 3 min. Sections were cooled for 20 min, washed in water, and rinsed with PBS. Endogenous peroxidase activity was eliminated by incubation with 1.5% hydrogen peroxide for 30 min and nonspecific binding blocked by incubating in 2% BSA for 30 min. Tissue sections were incubated overnight at 4°C with rabbit monoclonal anti-P-SMAD2/3 (1:200; Cell Signaling, Beverly, MA). Antibodies were mixed with 1% BSA/0.1% Triton. Sections were washed with PBS, incubated with avidin-biotin (Biocare Medical) and biotin-conjugated secondary antibodies, goat anti rabbit- (Perkin Elmer) and rat-adsorbed horse anti-mouse (Vector Laboratories, Burlingame, CA) for 1 h at room temperature. Slides were washed with PBS and incubated for 1 h at room temperature with streptavidin-biotinylated horseradish peroxidase (Jackson ImmunoResearch Laboratories) and developed in 3,3,-diaminobenzidine solution (Betazoid DAB Chromogen Kit, Biocare Medical). After being washed with distilled water, sections were counter stained with hematoxylin, dehydrated, and cover slipped. Sections were examined under light microscopy. Myofibroblasts were grown on four-well plastic slides and fixed in formalin.

Statistical Analyses

Data are expressed as means ± SE and were compared between control and experimental groups with two-tailed Student's t-test (Microsoft Excel 2007) or by two-way ANOVA with Bonferroni's post hoc analysis.

RESULTS

Apc^min/+/Epim^−/− Mice Have Decreased Polyposis

Polyp numbers and sizes were examined in 3-mo-old Apc^min/+ and Apc^min/+/Epim^−/− female mice. Small intestines were removed and divided into four segments including duodenum, proximal and distal jejunum, and ileum. Intestines were examined under the dissecting microscope for polyp quantification. Apc^min/+/Epim^−/− mice had significantly fewer polyps compared with Apc^min/+ mice, in proximal jejunum and ileum (Fig. 1A). The decrease in polyp numbers was most remarkable in ileum, the intestinal segment in which polyps were most abundant (64% decrease, mean number of polyps = 156 in Apc^min/+, n = 7 mice, vs. 72 polyps in Apc^min/+/Epim^−/−, n = 7 mice, P < 0.0001, Fig. 1A). In proximal jejunum, polyp numbers decreased by 45% (P < 0.05; mean number of polyps = 84 in Apc^min/+, n = 7 mice, vs. 47 in Apc^min/+/Epim^−/−; n = 7 mice, Fig. 1A). The duodenum contained few polyps, and there were no significant differences comparing Apc^min/+/Epim^−/− to Apc^min/+ mice. There were no differences in polyp number in Apc^min/+/Epim^−/− vs. Apc^min/+ distal jejunum. Few polyps were found in the colon in either Apc^min/+ or Apc^min/+/Epim^−/− mice (average 1.83 in Apc^min/+/Epim^−/− colon, 1.16 in Apc^min/+/Epim^−/− cecum vs. 2.64 in Apc^min/+ colon and 0.9 in Apc^min/+ cecum). There were no significant differences in the number of colonic polyps in either genotype, except for a small but significant decrease in the number of polyps in Apc^min/+ mice that were heterozygous for Epim (Apc^min/+/Epim^+/−) compared with Apc^min/+ colon (mean number of colonic polyps 2.6 vs. 1.14, P < 0.04).

Fig. 1.

Reduced polyposis in Apc^min/+/Epim^−/− compared with Apc^min/+ mice. Apc^min/+/Epim^−/− and Apc^min/+ mice were killed at 3 mo after birth. A: polyps in proximal jejunum and ileum were counted by visualizing intestinal segments under the Nikon SMZ 800 inverted stereo dissecting microscope at ×10 magnification. The total number of polyps in Apc^min/+/Epim^−/− ileum was decreased by 64%, and the total number of Apc^min/+/Epim^−/− jejunal polyps was decreased by 45% (*P < 0.05). Epim, epimorphin. B: number of polyps of all sizes is decreased in Apc^min/+/Epim^−/− vs. Apc^min/+ ileum (*P < 0.05). Polyps were visualized under a Nikon SMZ 800 inverted microscope, and photographs were taken. Polyp size was quantified using Image J1.42q in Java (National Institutes of Health). C: in jejunum, the number of smallest polyps ≤0.2 mm² is decreased in Apc^min/+/Epim^−/− vs. Apc^min/+ mice (*P < 0.05). Data are expressed as mean number of polyps in each group ± SE.

Polyps were divided into four groups based on size. The number of polyps of all sizes was decreased in ileum from Apc^min/+/Epim^−/− compared with Apc^min/+ mice (Fig. 1B). When stratified by size, there were significantly reduced numbers of only the smallest polyps in proximal jejunum (Fig. 1C, ≤0.2 mm², P < 0.05).

Isolation of Primary Myofibroblast Cultures

Dysregulated tumor-associated myofibroblast secretion of HGF has been postulated to play a role in colon cancer pathogenesis (34). Epim is expressed in myofibroblasts, which produce and secrete HGF. To further explore the mechanisms regulating the inhibition of tumorigenesis mediated by Epim, primary myofibroblast cell cultures were established from ileum from Apc^min/+ and Apc^min/+/Epim^−/− mice, using myofibroblast isolation and culture techniques as previously described (25, 33). At least three independent cell isolates were established and examined for each genotype.

To verify that these cells were myofibroblasts, each was examined for expression of myofibroblast markers, including α-SMA, vimentin, and desmin, as well as absence of expression of the epithelial cell marker cytokeratin (Fig. 2). Abundant expression of α-SMA and vimentin was demonstrated by qRT-PCR, with low levels of desmin expression and absent cytokeratin in all cultured cells (Fig. 2) regardless of genotype. A small decrease in vimentin expression was noted in Apc^min/+/Epim^−/− compared with Apc^min/+ ileal myofibroblasts (*P < 0.05). The morphology of each primary culture appeared similar with typical spindle-shaped cells (Fig. 3, A and B). Selected cell cultures from Apc^min/+ mice (Fig. 3, C and E) or from Apc^min/+/Epim^−/− mice (Fig. 3, D and F) were immunostained for α-SMA (Fig. 3, C and D) and vimentin (Fig. 3, E and F), confirming the myofibroblast phenotype in cells from both genotypes.

Fig. 2.

Apc^min/+/Epim^−/− and Apc^min/+ myofibroblasts have similar patterns of α-smooth muscle actin (SMA), vimentin, and desmin mRNA expression. MRNA levels for myofibroblast marker genes were similar in both genotypes except for a decrease in vimentin expression in Apc^min/+/Epim^−/− myofibroblasts (*P < 0.05). α-SMA was most abundantly expressed, desmin mRNA levels were almost undetectable, and cytokeratin, an epithelial marker, was absent.

Fig. 3.

Apc^min/+ and Apc^min/+/Epim^−/− ileal myofibroblasts have similar microscopic morphology and express α-SMA and vimentin. A and B: morphology is similar in primary cultures of Apc^min/+ (A) and Apc^min/+/Epim^−/− (B) myofibroblasts, visualized by inverted microscopy (magnification; ×100). C and D: Apc^min/+ (C) and Apc^min/+/Epim^−/− ileal myofibroblasts (D) express α-SMA (red, TRITC labeled). E and F: Apc^min/+ (E) and Apc^min/+/Epim^−/− (F) ileal myofibroblasts express vimentin (green, FITC labeled, E and F). C–F, visualized under fluorescence microscopy; magnification is ×400.

To rule out the possibility that Epim deletion resulted in decreased numbers of myofibroblasts, small intestines and polyps from Apc^min/+ and Apc^min/+/Epim^−/− mice (n = 4 per group) were immunostained to detect myofibroblasts, using an anti-SMA-α antibody. Myofibroblast numbers were quantified by light microscopy, counted per ×40 high-power field. There were no significant differences in small bowel or tumor-associated myofibroblasts comparing Apc^min/+ (average n = 83 SMA+ cells per polyp) and Apc^min/+/Epim^−/− mice (average n = 82 SMA+ cells per polyp).

Myofibroblast HGF Secretion is Reduced by Epim Deletion in Apc^min/+ Mouse Myofibroblasts

To examine the effect of Epim deletion on HGF secretion, ELISAs were performed on conditioned media from Apc^min/+/Epim^−/− and Apc^min/+ myofibroblast cultures. Cells from three independent Apc^min/+ myofibroblast isolates and from four Epim^−/−/Apc^min/+ myofibroblast isolates were plated in individual wells and incubated in serum-replete (RM) media for 48 h. Media was removed and replaced with either SF or RM myofibroblast media for 24 h, and then media was removed to quantify HGF secretion. Myofibroblasts cultured in SF media were also stimulated with LPS and media collected for analysis of HGF secretion. As a control, myofibroblast media supplemented with FCS was analyzed by ELISA for mouse HGF, and none was detected.

HGF secretion from Apc^min/+/Epim^−/− ileal myofibroblasts cultured in RM media was significantly reduced compared with Apc^min/+ ileal myofibroblasts (Fig. 4A, RM P < 0.03, n = 4 independent primary cultures for each genotype). There was a trend toward a decrease in HGF secretion from LPS- treated Apc^min/+/Epim^−/− compared with Apc^min/+ ileal myofibroblasts (LPS, P = 0.07). HGF secretion was almost completely inhibited from both Apc^min/+/Epim^−/− and Apc^min/+ myofibroblasts when cultured in SF media only. HGF mRNA expression was significantly reduced in Apc^min/+/Epim^−/− compared with Apc^min/+ ileal myofibroblasts (14-fold reduction, P < 0.02; Fig. 4B). Cellular HGF protein abundance, as measured by ELISA, was not significantly different in Apc^min/+/Epim^−/− compared with Apc^min/+ ileal myofibroblasts (Fig. 4C, P = 0.16).

Fig. 4.

A: hepatocyte growth factor (HGF) secretion from Apc^min/+/Epim^−/− myofibroblasts is inhibited compared with Apc^min/+ ileal myofibroblasts. HGF protein concentration was measured in conditioned media harvested from myofibroblasts grown either in serum-free media (SF), in SF media containing lipopolysaccharide (LPS), or in serum-replete media (RM). A marked decrease in HGF protein secretion was observed from Apc^min/+/Epim^−/− myofibroblasts grown in RM (*P < 0.03). A trend toward a decrease in HGF secretion from LPS-treated Apc^min/+/Epim^−/− vs. Apc^min/+ myofibroblasts was noted (*P = 0.07). HGF secretion was inhibited from both Apc^min/+/Epim^−/− and Apc^min/+ myofibroblasts incubated with SF media. B: HGF mRNA expression is decreased in Apc^min/+/Epim^−/− vs. Apc^min/+ ileal myofibroblasts. HGF mRNA levels were quantified by qRT-PCR. HGF mRNA levels are reduced 14-fold Apc^min/+/Epim^−/− compared with Apc^min/+ ileal myofibroblasts (*P = 0.015). C: HGF protein concentrations measured by ELISA are unchanged in Apc^min/+ vs. Apc^min/+/Epim^−/− ileal myofibroblast cell lysates (P = 0.16). Results were normalized to total protein.

TGF-β Signaling is Increased in Apc^min/+/Epim^−/− Compared with Apc^min/+ Polyps and Adjacent Intestinal Mucosa

Deletion of TGF-β receptor II in fibroblasts (which blocks TGF-β signaling) results in increased fibroblast HGF expression and activation of the HGF receptor cMet in adjacent epithelia, with increased proliferation and induction of epithelial carcinogenesis (5). To determine whether the observed reduction in HGF expression and secretion from myofibroblasts might be associated with increased TGF-β signaling in Apc^min/+/Epim^−/− mucosa, we examined TGF-β expression in polyps from Apc^min/+/Epim^−/− compared with Apc^min/+ mice by quantitative RT-PCR. Although the three TGF-β isoforms have similar biological activities and act via the TGF-β receptor II, each isoform was analyzed with qRT-PCR to more accurately assess total TGF-β expression. TGF-β2 mRNA expression was increased in polyps from Apc^min/+/Epim^−/− mice compared with Apc^min/+ mice (P < 0.01; 1.8-fold increase, Table 2). A trend toward increased TGF-β3 expression in polyps from Apc^min/+/Epim^−/− mice compared with Apc^min/+ mice was also observed (Table 2, 1.38-fold, P = 0.086). To determine whether the increase in TGF-β2 mRNA levels in Apc^min/+/Epim^−/− vs. Apc^min/+ polyps reflected increased mucosal TGF-β signaling, we performed immunohistochemical analysis to detect expression of the immediate downstream TGF-β target, phosphorylated (P)-Smad2/3. Expression was compared in polyps and in uninvolved adjacent mucosa (Fig. 5). Multiple small and large polyps (n = 21 polyps from 6 Apc^min/+ mice and n = 19 polyps from 7 Apc^min/+/Epim^−/− mice) were examined, and location and intensity of staining were observed. All slides were processed for immunohistochemical analysis at the same time, under identical conditions to permit staining intensity comparisons. Consistent with the increase in TGF-β2 mRNA expression, p-SMAD 2/3 expression was increased in nuclei of Apc^min/+/Epim^−/− (Fig. 5, D–F) compared with Apc^min/+ polyps (Fig. 5, A–C). Brown nuclear staining (arrows) was seen uniformly in Apc^min/+/Epim^−/− polyp epithelium from the surface to the crypts, whereas Apc^min/+ polyps had little staining in the crypts (arrowheads, Fig. 5, A and B) and patchy nuclear staining in the polyp epithelium above the crypts. Increased p-Smad2/3 expression was consistently observed in both small and large Apc^min/+/Epim^−/− polyps compared with Apc^min/+ polyps. Similarly, p-Smad2/3 expression was increased in Apc^min/+/Epim^−/− (Fig. 5H) compared with Apc^min/+ uninvolved mucosa (Fig. 5G). Patchy brown nuclear staining was noted in Apc^min/+ villi (Fig. 5G, arrows); villus nuclear staining was more intense and more uniform from villus tip to the base in Apc^min/+/Epim^−/− mice (Fig. 5H, arrows). p-Smad2/3 expression was also noted in nuclei of the Apc^min/+/Epim^−/− crypts but was largely absent in Apc^min/+ crypts (arrowheads, Fig. 5H). Consistent with our observations of increased TGF-β signaling in Apc^min/+/Epim^−/− polyps, we observed a trend toward increased TGF-β1 mRNA expression in Apc^min/+/Epim^−/− polyps compared with uninvolved mucosa (1.39-fold, P = 0.09, Table 2); this trend was not observed in Apc^min/+ mice (Table 2).

Table 2.

TGF-β mRNA expression in Apc^min/+/Epim^−/− compared with Apc^min/+ polyps and uninvolved ileum

Genotype	Gene	Location	Fold Change	P Value
Apcmin/+/Epim−/− compared with Apcmin/+	TGF-β2	Ileum - polyps	+1.8↑ Apcmin/+/Epim−/− vs. Apcmin/+	P = 0.004
Apcmin/+/Epim−/− compared with Apcmin/+	TGF-β3	Ileum - polyps	1.38↑ Apcmin/+/Epim−/− vs. Apcmin/+	NS (P = 0.086)
Apcmin/+/Epim−/−	TGF-β1	Ileal polyp vs. uninvolved ileum	1.39↑	NS (P = 0.09)
Apcmin/+	TGF-β1	Ileal polyp vs. uninvolved ileum	1.11	NS (P = 0.55)

Expression of TGF-β1, 2, and 3 mRNA was quantified by qRT-PCR in ileal polyps and uninvolved ileum from Apc^min/+/Epim^−/− and Apc^min/+ mice. Epim, epimorphin.

Fig. 5.

Increased phospho (p)-Smad 2/3 expression in Apc^min/+/Epim^−/− compared with Apc^min/+ ileal polyps and uninvolved ileum. Sections of ileal polyps from 3 Apc^min/+ mice (A–C) or from 3 Apc^min/+/Epim^−/− mice (D–F) were incubated with a rabbit monoclonal anti-p-SMAD2/3 antibody, incubated with avidin-biotin/biotin-conjugated secondary antibodies and visualized with streptavidin-biotinylated horseradish peroxidase. All slides were processed at the same time under identical conditions. Arrows depict mucosa with brown-stained nuclei positive for p-Smad2/3 expression; arrowheads depict mucosa with blue nuclei that lack p-Smad2/3 expression. G and H: p-Smad 2/3 expression in uninvolved ileum from Apc^min/+ (G) vs. Apc^min/+/Epim^−/− (H) mice. (Magnifications: all at ×200). Arrows depict patchy brown p-Smad3-positive nuclei in G and uniformly brown-stained nuclei in H; arrowheads in G indicate blue nuclei that lack p-Smad2/3.

To determine the effect of TGF-β on HGF secretion from myofibroblasts with Epim deletion, Epim^−/− and WT myofibroblasts (n = 3 primary myofibroblast cultures from each genotype) were incubated with TGF-β3 for 24 h (20). Each primary culture was also treated with IL-1β, which stimulates HGF secretion, as a normalizing control. HGF secretion from Epim^−/− myofibroblasts was inhibited by 60% in response to TGF-β treatment, compared with only 34% inhibition of secretion from WT myofibroblasts (P = 0.02).

DISCUSSION

In the present studies, we have shown that Epim deletion significantly inhibited small intestinal polyposis in Apc^min/+ mice. The inhibitory effect of Epim deletion is associated with a marked reduction in HGF secretion from Apc^min/+/Epim^−/− myofibroblasts compared with Apc^min/+ myofibroblasts and by an increase in TGF-β signaling in Apc^min/+/Epim^−/− polyps and uninvolved mucosa compared with Apc^min/+ mice. HGF plays a role in stromal regulation of colon carcinogenesis (18, 34), and TGF-β has well-described antiproliferative, antitumorigenic effects, which contribute to the inhibition in polyp multiplicity (1, 20, 21). Our data support a novel role for Epim in regulating HGF secretion from ileal myofibroblasts and suggest that the effects of Epim deletion can be exploited to alter myofibroblast function and thereby reduce the protumorigenic effects of the stromal microenvironment.

In the present study, we have shown that HGF secretion from Apc^min/+/Epim^−/− myofibroblasts was profoundly reduced compared with Apc^min/+ ileal myofibroblasts, providing a possible mechanism for the effects of Epim on polyposis. HGF is a protumorigenic growth factor that interacts with its receptor, cMET, to increase epithelial proliferation. It also enhances colon cancer growth, invasiveness, and metastasis (15, 30). Myofibroblasts or conditioned media from myofibroblast cultures, when added to colon cancer stem cell cultures, increased Wnt signaling in stem cells and enhance “stemness,” as assessed by in vivo tumorigenicity studies (34). HGF was identified as a key soluble mediator of these effects (34). Thus the profound inhibition of HGF secretion from myofibroblasts would be predicted to decrease polyp number as well as polyp growth. In the present study, we showed that, in ileum, polyps of all sizes were reduced in number, implicating effects on polyp initiation. Polyp growth was also inhibited, as indicated by the observation that the largest percent decrease in number of polyps was observed for the larger polyp groups (61–67% for polyps ranging from >0.4 to >0.9 mm² vs. 47–49% for the smallest groups ranging from <0.2 to 0.4 mm²; Fig. 1B). In jejunum, the number of only the smallest polyps was decreased, suggesting that the effects of Epim deletion in Apc^min/+ jejunum vs. ileum are temporally regulated; whether this difference reflects intrinsic regional differences in myofibroblasts and their interactions with jejunal vs. ileal epithelium or is a result of differences in the luminal environment (i.e., with a more “permissive” environment in the ileum) is unclear. Epim protein and mRNA are expressed at equal levels in jejunum and ileum (data not shown), thus eliminating differential expression as an explanation; however, this does not equate to similar effects of deletion.

We have also shown that polyps and adjacent uninvolved mucosa from Apc^min/+/Epim^−/− mice have increased TGF-β2 mRNA expression compared with Apc^min/+ mice and increased TGF-β signaling as measured by p-SMAD2/3 immunohistochemical staining. TGF-β signaling plays an important and complex role in colorectal cancer (1). TGF-β is expressed in cancer cells and in cancer-associated stromal cells (21). In early-stage colon cancers and other epithelial tumors, TGF-β signaling inhibits epithelial cell proliferation and is tumor suppressive (3, 20, 21). Increased TGF-β signaling would thus also be expected to suppress adenoma growth in Apc^min/+/Epim^−/− mice.

The mechanisms regulating the profound reduction in HGF mRNA expression and secretion from Apc^min/+/Epim^−/− myofibroblasts are complex. Loss of the function of Epim as a syntaxin and t-SNARE vesicle-docking protein involved in plasma membrane fusion events and exocytosis (4, 11, 28) would be expected to have direct effects on HGF secretion. Syntaxins are part of the SNARE complex that regulates protein secretion in multiple cell types. Inhibition of HGF secretion resulting from Epim deletion is consistent with our previous observations showing that Epim deletion also affects the secretion of protumorigenic proinflammatory cytokines and growth factors in LPS-stimulated colonic myofibroblasts (33). The effects of Epim deletion are, however, specific for a subset of secreted proteins, as we have shown that secretion of constitutively expressed proteins such as Bmp4 increases; however, LPS-stimulated secretion of IL-6 is markedly inhibited (10, 33, 35).

In addition, Epim deletion resulted in increased TGF-β expression and signaling; multiple studies have shown that the TGF-β and HGF signaling pathways are closely linked, as TGF-β regulates HGF expression and secretion (5, 7, 18). Loss of TGF-β responsiveness in fibroblasts, induced by fibroblast-specific deletion of the TGF-β receptor TGFβRII, resulted in prostate and forestomach squamous cancer in mice, with increased stromal cell abundance (5). Loss of TGF-β responsiveness resulted in increased fibroblast HGF expression and activation of the HGF receptor cMet in adjacent epithelium, with increased proliferation and induction of epithelial cancer. Further evidence for linkage between TGF-β and HGF signaling and its importance in epithelial carcinogenesis comes from a recent study showing that intestinal myofibroblast deletion of tumor progression locus 2 (Tpl2), a MAP kinase that regulates inflammatory and oncogenic pathways, is associated with increased HGF secretion, reduced sensitivity to the negative regulation of HGF by TGF-β3, and increased colitis-associated tumorigenesis (18). These data are consistent with our observations in the present studies and our previous work in colitis-associated cancer (33), both showing reduced tumorigenicity in the setting of Epim deletion. Herein, we showed that, in addition to inhibition of HGF secretion, Apc^min/+/Epim^−/− compared with Apc^min/+ ileal myofibroblasts had markedly reduced HGF mRNA levels (Fig. 4B). Our data suggest that increased TGF-β expression resulting from Epim deletion also inhibits HGF mRNA expression in myofibroblasts. We have shown herein that TGF-β also inhibits HGF secretion from myofibroblasts. It is also likely that steady-state HGF protein levels are not significantly reduced in Apc^min/+/Epim^−/− compared with Apc^min/+ myofibroblasts (Fig. 4C) due to the profound decrease in secretion, which in turn may lead to intracellular protein accumulation.

The mechanisms by which Epim deletion in Apc^min/+ mice results in increased polyp and mucosal TGF-β mRNA expression and signaling (Fig. 5 and Table 2) are likely related to its unique structure and context-specific cellular localization. Depending upon the cell-specific context, Epim has an extracellular (6, 12, 29) as well as intracellular function (4, 28). Epim translocates to the cell surface and is secreted via nonclassical mechanisms (6, 13). Epim interacts with αV-integrins on target epithelial cells (13), affecting downstream signaling pathways that induce branching morphogenesis and regulating transcription of multiple target genes, including MMPs (6). Relevant to the present observations are studies in kidney that showed that Epim is expressed in interstitial myofibroblasts and Epim protein abundance increases in response to urinary obstructive injury; blocking Epim signaling with an anti-Epim antibody increases TGF-β mRNA expression in kidney during the injury recovery phase (36). Herein, we have shown that lack of Epim signaling in the gut in Apc^min/+ mice also results in increased mucosal TGF-β expression. TGF-β is expressed in stromal cells and epithelium in Apc^min/+ mice (8) and in tumor stroma as well as cancer cells in colon cancers (21). Thus the observed increase in TGF-β signaling in Apc^min/+/Epim^−/−normal mucosa and polyps could have resulted from effects of Epim deletion on myofibroblast expression and secretion of TGF-β, or from loss of Epim signaling directly from myofibroblasts to the epithelium resulting in increased epithelial expression of TGF-β. Our data suggest the latter possibility because TGF-β2 mRNA expression was increased in Apc^min/+/Epim^−/− vs. Apc^min/+ polyps, yet TGF-β2 mRNA expression was unchanged in myofibroblasts from Apc^min/+/Epim^−/− vs. Apc^min/+ mice (data not shown). Also, our attempts to quantify TGF-β2 secretion from myofibroblasts have been limited by lack of an adequate ELISA. Although not definitive, it is likely based on these results that increased epithelial TGF-β signaling results directly from the loss of myofibroblast Epim signaling to the epithelium.

In summary, our data suggest that Epim deletion reduces small bowel polyposis and affects both myofibroblast HGF secretion and epithelial TGF-β signaling. Reducing the tumorigenicity of the stromal environment is an attractive and emerging target for therapy of other epithelial cancers (17, 23, 24) in addition to colon cancer, and our results suggest a novel means for achieving this in vivo. An examination of the complete secretory and expression profile of Apc^min/+/Epim^−/− compared with Apc^min/+ myofibroblasts may lead to the identification of new stromal targets for colon cancer therapy.

GRANTS

This work was supported by NIH NIDDK R01DK61216, R01DK46122, R01 DK50466, and the Digestive Diseases Research Core Center P30 DK52574.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: E.A.S., S.B., J.L., A.S., and G.K. performed experiments; E.A.S., S.B., J.L., G.K., and D.C.R. analyzed data; E.A.S., J.L., and D.C.R. interpreted results of experiments; E.A.S., S.B., and D.C.R. prepared figures; E.A.S., S.B., A.S., M.S.L., and D.C.R. approved final version of manuscript; A.S., M.S.L., and D.C.R. edited and revised manuscript; M.S.L. and D.C.R. conception and design of research; D.C.R. drafted manuscript.