Main

Accumulating evidence suggest that most colorectal adenomas are initiated by the inactivation of APC gene, and progress into adenocarcinomas through additional genetic alterations in KRAS (for KRAS), TP53 (P53), SMAD4, TGFBR2 (TGF-β type-II receptor), etc.1 In colorectal carcinomas, marked tumor invasion and expansion of the stroma are well-known features associated with metastasis and poor prognosis.2, 3, 4

As a mouse model for colorectal adenocarcinomas, we previously constructed cis-Apc+/Δ716 Smad4+/− compound mutant (hereafter cis-Apc/Smad4 mice) carrying a knockout allele of the Smad4 gene on the same chromatid as that of Apc.5 Using these mice, we demonstrated that loss of SMAD4-dependent TGF-β family signaling converts intestinal adenomas to adenocarcinomas. Namely, the cis-Apc/Smad4 mice develop tumors that show marked invasion and stromal expansion, although the simple Apc mutant (ApcΔ716) mice form only benign adenomas.5, 6 To examine the mechanisms underlying such malignant changes in the cis-Apc/Smad4 tumors, we recently performed DNA microarray analyses and found that the adenocarcinomas expressed much higher levels of CC-chemokine ligand 9 (CCL9) than the ApcΔ716 adenomas. Further studies revealed that CCL9 released from the cis-Apc/Smad4 tumor epithelia recruits immature myeloid cells (iMCs) that carry its CC-chemokine receptor (CCR1) and express matrix metalloproteinase (MMP) 2 and 9 to help tumor invasion.7

Although it is the tumor stromal cells that produce most MMPs whose levels are increased in the colon cancer (eg MMP2 and MMP9), the tumor epithelial cells can express another MMP, MMP7.8 Interestingly, our microarray data7 showed increased expression of MMP7 in the cis-Apc/Smad4 adenocarcinomas compared with the ApcΔ716 adenomas. In human colorectal cancer, it has been reported that the level of MMP7 correlates with advanced Dukes stages, poor outcome, invasion and metastasis to the liver.9, 10, 11 Several studies have shown that overexpression of MMP7 increases the invasive ability and metastasis to the liver in some colon cancer cell lines,9, 12, 13 although another study suggests that MMP7 has little effect on the invasive potential or metastasis capacity of colon cancer cells.14 Interestingly, a recent report suggests that MMP7 can contribute to expansion of the tumor stroma, because recombinant MMP7 stimulates proliferation and migration of human colonic fibroblasts.15 As these studies relied on cell lines, however, it remains to be determined what actual roles MMP7 plays in colon cancer development in vivo.

To address this issue, we crossed the cis-Apc/Smad4 mice with the Mmp7−/− and generated triple mutant mice, cis-Apc/Smad4 Mmp7−/−. Here, we report that lack of MMP7 in the cis-Apc/Smad4 mice reduces the number and size of tumors, but does not affect the depth of invasion, number of stromal fibroblasts or expression of extracellular matrix components.

MATERIALS AND METHODS

Knockout Mice

The knockout alleles of the Apc, Smad4 and Mmp7 genes have been described previously.5, 16, 17 The cis-Apc/Smad4 male mice were crossed twice with the Mmp7−/− female mice (from Jackson Laboratory, Bar Harbor, ME, USA) to generate triple mutant cis-Apc/Smad4 Mmp7−/− mice. All animal experiments have been conducted according to the protocol approved by the Animal Care and Use Committee of Kyoto University.

RT-PCR

The levels of mRNAs were determined by RT-PCR as described previously,7 using the following primers: Mmp2 (F: 5′-GCAAGTTCCCGTTCCGCTTCC-3′ and R: 5′-CAGTACCAGTGTCAGTATCAGC-3′), Mmp3 (F: 5′-TGGAGATGCTCACTTTGAC G-3′ and R: 5′-GATGGAAGAGATGGCCAAAA-3′), Mmp7 (F: 5′-GCGGAGATGCTC ACTTTGAC-3′ and R: 5′-GCATCTATCACAGCGTGTTC-3′), Mmp9 (F: 5′-TACAGGGCCCCTTCCTTACT-3′ and R: 5′-ACTCCTTATCCACGCGAATG-3′), Mmp10 (F: 5′- CAGGTGTGGTGTTCCTGATG-3′ and R: 5′- TGGAAGTTAGCTGGGCTTGT-3′), Mmp14 (F: 5′-GCTCCGAGGAGAGATGTTTG-3′ and R: 5′-CATCACTGCCCATGAA TGAC-3′), Col1a1 (F: 5′-CACCCTCAAGAGCCTGAGTC -3′ and R: 5′-GGTGGAGG GAGTTTACACGA-3′), Col1a2 (F: 5′-TGGATTGACCCTAACCAAGG-3′ and R: 5′-CTTGCCCCATTCATTTGTCT-3′) and Gapdh (F: 5′-CCTTCATTGACCTCAACTAC-3′ and R: 5′-TGGGC CCTCAGATGCCTGCT-3′).

Western Blotting

Dissected tumors were homogenized in lysis buffer containing 50 mM Hepes, 150 mM NaCl, 20 mM NaF, 10 mM sodium pyrophosphate, 2 mM sodium vanadate, 5 mM EDTA, 1% NP-40 and a protease inhibitor cocktail (Roche, Indianapolis, IN, USA). The tissue lysates were separated in a sodium dodecyl sulfate–polyacrylamide gel (15%) and transferred onto polyvinylidene difluoride membrane, using standard techniques. Specific signals were detected by rabbit polyclonal anti-MMP7 antibody (Calbiochem, San Diego, CA, USA) and mouse monoclonal antiβ-actin antibody (Sigma, St Louis, MO, USA).

Polyp Number Scoring

The intestinal tissues were fixed with 4% paraformaldehyde and then transferred to PBS. The tumor diameters were measured under a dissection microscope, and the tumor numbers were scored for the respective size classes. Under a dissection microscope, we identified and scored the nascent polyps (0.2–0.5 mm in diameter) consisted of microadenomas.6

Immunohistochemistry

The intestinal tissues were fixed with 4% paraformaldehyde and embedded in paraffin wax for immunohistochemistry, or directly embedded in OCT compound (Sakura Finetek, Tokyo, Japan) for immunofluorescent staining. Sections were immunostained as described previously,7 using the following primary antibodies: rabbit polyclonal antibodies for MMP7 (Calbiochem), collagen type-I (LSL, Tokyo, Japan), fibronectin (Sigma), tenascin-C (Sigma) and pan-cytokeratin (Dako, Carpinteria, CA, USA), and rat monoclonal antibodies for CD31, CD34 and CD11b (BD Biosciences, San Jose, CA, USA). In the immunofluorescent staining, nuclei were detected by staining with DAPI (Calbiochem).

Counting Fibroblasts

Frozen sections of the intestinal polyps were stained with rat monoclonal antifibroblast antibody (AbD Serotec, Raleigh, NC, USA) as described above. The nuclei in the stained cells were counted under a fluorescent microscope ( × 400). Nine randomly selected fields were scored per mouse. Results were presented as the mean number of cells per field obtained from three mice for each strain.

Statistical Analysis

The values were shown as mean±s.d., and analyzed by Student's t-test. The ‘P’-values smaller than 0.05 were considered significant.

RESULTS

Expression of MMP7 is Increased in cis-Apc/Smad4 Intestinal Tumors

To investigate the mechanism of tumor invasion and stromal expansion in the cis-Apc+/Δ716 Smad4+/− (cis-Apc/Smad4) adenocarcinomas, we recently performed DNA microarray analyses7 and found higher expression of Mmp7 mRNA in the carcinomas than in the ApcΔ716 adenomas. To confirm the microarray data, we performed RT-PCR and western blotting. As shown in Figure 1a, the Mmp7 mRNA level in the cis-Apc/Smad4 adenocarcinomas was three times higher than that in the ApcΔ716 adenomas. Expression of the Mmp7 mRNA in the ApcΔ716 adenomas was twice higher than in the normal intestinal tissues of the wild-type C57BL/6 mice, although to a less extent than in the cis-Apc/Smad4 adenocarcinomas (sixfold). We also found that expression of the MMP7 precursor protein was increased moderately in the cis-Apc/Smad4 adenocarcinomas compared with the ApcΔ716 adenomas (1.3-fold) or normal tissues (1.9-fold; Figure 1b). Furthermore, the levels of the cleaved (active) form MMP7 were markedly elevated in the cis-Apc/Smad4 adenocarcinomas (27 times higher than in the ApcΔ716 adenomas). Immunostaining data showed that the MMP7 protein localized to the tumor epithelial cells in both ApcΔ716 and cis-Apc/Smad4 polyps (Figure 1c). Consistent with the western blotting data, the cis-Apc/Smad4 adenocarcinoma cells showed a stronger immunoreactivity to MMP7 than the ApcΔ716 adenoma cells. As described previously,18 Paneth cells in the normal intestinal crypt expressed MMP7 in both ApcΔ716 and cis-Apc/Smad4 mice at similar levels (Figure 1c). These results indicate that expression of MMP7 is increased during the progression of intestinal adenomas to adenocarcinomas, caused by impaired SMAD4 signaling in the cis-Apc/Smad4 mice.

Figure 1
figure 1

The levels of MMP7 in the cis-Apc/Smad4 polyps. (a) Expression of MMP7 mRNA determined by RT-PCR. Total RNA was prepared from the intestinal tissues of normal C57BL/6 mice (wild type) or polyps of ApcΔ716 and cis-Apc/Smad4 mice. (b) Expression of MMP7 protein determined by western blotting. Open and closed arrowheads indicate pro- and active-MMP7, respectively. (c) Intestinal polyps (top) and normal tissues (bottom) of ApcΔ716 (left) and cis-Apc/Smad4 (right) mice stained for MMP7. Insets show higher magnifications of the boxed areas. Scale bars show 200 and 100 μm (inset) in top panels, and 200 and 10 μm (inset) in bottom panels.

Full size image

Loss of MMP7 Reduces Number and Size of Tumors in cis-Apc/Smad4 Mice

To examine the role of the increased MMP7 expression in the development of SMAD4-deficient adenocarcinomas, we crossed the cis-Apc/Smad4 mice with the Mmp7−/− mice twice, and generated triple mutant mice. As shown in Figure 2a, we found no expression of MMP7 mRNA or protein in the intestinal polyps of the cis-Apc/Smad4 Mmp7−/− mice. Because some of these mice became moribund at the age of 14 weeks, we killed all animals at this age in this series of experiments.

Figure 2
figure 2

Effects of MMP7-deficiency on tumor formation in the cis-Apc/Smad4 mice. (a) Expression of MMP7 mRNA (left) and protein (right) in the intestinal polyps (P) and normal tissues (N) from MMP7-intact (+/+) and knockout (−/−) cis-Apc/Smad4 mice. (b) Total number of the intestinal polyps in the MMP7-intact (+/+) and knockout (−/−) cis-Apc/Smad4 mice. Results show a mean number of polyps/mouse±s.d. (n=9). P<0.001. (c) Size distribution of the intestinal polyps. Polyps were classified according to their diameters in millimeters. Results are shown as percentages for the respective size classes (n=9). Φ indicates polyp diameter (mm). (d) Intestinal tumors in the cis-Apc/Smad4 and cis-Apc/Smad4 Mmp7−/− mice stained with H&E. Scale bars show 200 μm.

Full size image

As loss of MMP7 suppressed the tumorigenesis of the intestinal adenomas in ApcMin mice,17 we first investigated whether the MMP7-deficiency affected the number or size of the adenocarcinomas in the cis-Apc/Smad4 mice. Consistent with our earlier report,5 the cis-Apc/Smad4 mice developed 93±23 intestinal polyps at the age of 14 weeks (Figure 2b). Compared with the cis-Apc/Smad4 mice, the cis-Apc/Smad4 Mmp7−/− mice had a 50% less number of tumors at the same age (43±19 polyps, Figure 2b), suggesting that the lack of MMP7 suppressed tumor formation. Furthermore, we found smaller tumor size in the cis-Apc/Smad4 Mmp7−/− mice (Figure 2c). Namely, the ratio of large polyps (1.0–3.0 mm in diameter) in the cis-Apc/Smad4 Mmp7−/− mice (35%) was significantly lower than that in the cis-Apc/Smad4 mice (57%). On the other hand, the ratio of small polyps (<1.0 mm) was higher in the cis-Apc/Smad4 Mmp7−/− (49%) than in the cis-Apc/Smad4 (32%). It is worth noting that the ratio for very large polyps (>3.0 mm) was similar between the two strains (Figure 2c). As shown in Figure 2d, there was no obvious difference in the polyp morphology between the two types of mutants. These results suggest that MMP7 facilitates the expansion of most adenocarcinomas in the cis-Apc/Smad4 mice, although some tumors can expand without MMP7. Consistent with the data of ApcMin mice,17 the absence of MMP7 reduced the number and size of tumors in ApcΔ716 mice, as well (data not shown).

Loss of Mmp7 does not Affect Invasion Depth of cis-Apc/Smad4 Tumors

We have demonstrated previously that the cis-Apc/Smad4 polyps show marked invasion, and that the invasion is helped by the iMCs that are recruited by CCL9 and express its receptor CCR1 as well as MMP2/9.7 Because increased expression of MMP7 can contribute to colorectal tumor invasion,9, 12, 13 we next studied the effect of MMP7-deficiency on the depth of tumor invasion in the cis-Apc/Smad4 tumors. In the cis-Apc/Smad4 mice, all tumors showed invasion into the submucosa, smooth muscle layer and serosa at the proportions of 35, 35 and 30%, respectively (Figure 3a). Contrary to our anticipation, polyps in the cis-Apc/Smad4 Mmp7−/− mice showed marked invasions like those in the cis-Apc/Smad4 mice. Namely, 54 and 34% of the tumors invaded into the submucosa and smooth muscle layer, respectively (Figure 3a). Although the proportion of tumors that invaded to the serosa (12%) was lower in the cis-Apc/Smad4 Mmp7−/− mutant, the difference was not significant statistically. Because it has been reported that the absence of MMP7 results in a compensatory increase in the level of MMP2,17 we examined the mRNA levels for MMPs involved in tumor invasion. In the cis-Apc/Smad4 Mmp7−/− polyps, mRNA levels for Mmp2, Mmp3, Mmp9 and Mmp10 were much higher than those in the normal intestinal tissues or ApcΔ716 polyps (Figure 3b). However, we could not find any significant differences in the mRNA levels for these MMPs between the cis-Apc/Smad4 and cis-Apc/Smad4 Mmp7−/− polyps. Although the mRNA level for Mmp14 was slightly higher in the cis-Apc/Smad4 Mmp7−/− polyps than in the cis-Apc/Smad4 polyps (1.5-fold), the difference was not significant statistically. We next examined the recruitment of iMCs in the cis-Apc/Smad4 Mmp7−/− tumors. To this end, we performed immunostaining for CD34 and CD11b, markers for this class of cells.7 Consistent with our previous findings, we found accumulation of CD34- and CD11b-expressing iMCs at the invasion front of the cis-Apc/Smad4 polyps (Figure 3b). Similarly, the invading glands in the cis-Apc/Smad4 Mmp7−/− adenocarcinomas were associated with iMCs (Figure 3b), suggesting that the cis-Apc/Smad4 tumor epithelial cells can attract the iMCs even in the absence of MMP7. These results indicate that increased expression of MMP7 in the tumor epithelium does not involve in the invasion of the adenocarcinomas in the cis-Apc/Smad4 mice.

Figure 3
figure 3

Intestinal tumor invasion in the cis-Apc/Smad4 Mmp7−/− tumors. (a) The invasion depths of the polyps (Φ ≥2 mm) in the cis-Apc/Smad4 and cis-Apc/Smad4 Mmp7−/− mice (n=50 tumors from three mice per group). The polyps stained with H and E were classified according to their invasion depth, ie, mucosa (m), submucosa (sm), muscularis propria (mp) and serosa (se). Results are presented as proportions among the polyps examined (mean±s.d. from 50 polyps in each strain). (b) Expression of MMP mRNAs determined by RT-PCR. Total RNA was prepared from the intestinal tissues of wild-type mice or ApcΔ716, cis-Apc/Smad4 and cis-Apc/Smad4 Mmp7−/− polyps. (c) Invasion front of the cis-Apc/Smad4 (left) and cis-Apc/Smad4 Mmp7−/− (right) polyps. Serial sections were double stained for pan-cytokeratin (green) and CD11b (red) or CD34 (green) and CD31 (red), respectively. Scale bars show 100 μm.

Full size image

Lack of Mmp7 does not Affect Number of Fibroblasts in cis-Apc/Smad4 Tumor Stroma

Expansion of the tumor stroma with fibroblasts and collagen fibers (fibrosis) is one of characteristic features in human colorectal carcinoma.2 Interestingly, MMP7 stimulates the proliferation of human colonic fibroblasts by cleaving insulin-like growth factor-binding protein 5 (IGFBP5) and subsequent release of insulin-like growth factor (IGF).15 We therefore speculated that loss of MMP7 might prevent the cleavage of IGFBP5 and subsequent proliferation of fibroblasts in the cis-Apc/Smad4 adenocarcinomas. As shown in Figure 4a, the cleaved form IGFBP5 was increased in the cis-Apc/Smad4 adenocarcinomas compared with that in the ApcΔ716 adenomas. Furthermore, immunostaining showed a higher number of fibroblasts in the cis-Apc/Smad4 tumor stroma (75±18 cells per field) than in the ApcΔ716 (44±16; Figure 4c and d). However, polyps in the triple mutant did not show much difference either in the extent of IGFBP5 cleavage or in the fibroblast numbers (77±15) compared with those in the cis-Apc/Smad4 (Figure 4b–d).

Figure 4
figure 4

Effects of MMP7-deficiency on the number of stromal fibroblasts in the cis-Apc/Smad4 tumor. (a) Expression of IGFBP5 protein determined by western blotting. Protein samples were prepared from the intestinal tissues of wild-type mice or ApcΔ716 and cis-Apc/Smad4 polyps. Open and closed arrowheads indicate intact- and cleaved-IGFBP5, respectively. (b) Expression of IGFBP5 protein in the MMP7-intact (+/+) and knockout (−/−) cis-Apc/Smad4 polyps. (c) The intestinal polyps of the ApcΔ716 (left), cis-Apc/Smad4 (middle) and cis-Apc/Smad4 Mmp7−/− (right) stained for fibroblast specific antigen. Insets show higher magnifications of the boxed areas. Scale bars show 200 and 100 μm (inset). (d) The number of fibroblasts in the intestinal polyps. Genotypes of Mmp7 are shown as +/+ (wild type) or −/− (knockout). Data are given as the mean numbers of fibroblasts per field±s.d. from nine fields per mouse (n=3). P<0.01.

Full size image

Absence of Mmp7 does not Affect Accumulation of Collagen Type-I in cis-Apc/Smad4 Tumors

In human colorectal carcinomas, fibronectin, tenascin and collagen type-I are major components of fibrotic foci, and are increased compared with normal tissues or adenomas.19, 20, 21 As shown in Figure 5a, we found more collagen type-I in the cis-Apc/Smad4 adenocarcinomas than in the ApcΔ716 adenomas. Consistently, the mRNA levels for the α1 and α2 chains of collagen type-I (Col1a1 and Col1a2, respectively) were markedly higher in the cis-Apc/Smad4 adenocarcinomas (Figure 5b). Because it was reported that expression of MMP7 was increased in fibrosis of the liver and lungs,22, 23 and because loss of MMP7 reduced the severity of the bleomycin-induced pulmonary fibrosis,23 we next investigated the effect of MMP7-deficiency on the amount of collagen type-I in the cis-Apc/Smad4 tumors. Contrary to our anticipation, the cis-Apc/Smad4 Mmp7−/− tumors showed essentially the same levels of collagen type-I protein and mRNA as the cis-Apc/Smad4 tumors (Figure 5a and c). Therefore, it appears unlikely that MMP7 contributes to the increased synthesis of collagen type-I in the cis-Apc/Smad4 adenocarcinomas.

Figure 5
figure 5

Expression of collagen type-I in the cis-Apc/Smad4 Mmp7−/− tumors. (a) The intestinal polyps of the ApcΔ716 (left), cis-Apc/Smad4 (middle) and cis-Apc/Smad4 Mmp7−/− (right) stained for collagen type-I. Insets show higher magnifications of the boxed areas. Scale bars show 200 and 100 μm (inset). (b) Expression of collagen type-I mRNA determined by RT-PCR. Col1a1 and Col1a2 show α1 and α2 chains of collagen type-I, respectively. Total RNA was prepared from the intestinal tissues of wild-type mice or ApcΔ716 and cis-Apc/Smad4 polyps. (c) Expression of collagen type-I mRNA in the intestinal polyps (P) and normal tissues (N) from cis-Apc/Smad4 (+/+) and cis-Apc/Smad4 Mmp7−/− (−/−) mice.

Full size image

Consistent with the clinical data,20 expression of fibronectin and tenascin was also increased in the cis-Apc/Smad4 adenocarcinomas compared with the ApcΔ716 adenomas (Figure 6). However, we found no distinct differences in their expression between the cis-Apc/Smad4 and cis-Apc/Smad4 Mmp7−/− tumors. Although fibronectin is a known substrate of MMP7,24, 25 these immunostaining data suggest that MMP7 is dispensable for its degradation in the intestinal tumors.

Figure 6
figure 6

Expression of fibronectin and tenascin in the cis-Apc/Smad4 Mmp7−/− tumors. The intestinal polyps of the ApcΔ716 (left), cis-Apc/Smad4 (middle) and cis-Apc/Smad4 Mmp7−/− (right) stained for fibronectin (top) and tenascin (bottom). Insets show higher magnifications of the boxed areas. Scale bars show 200 and 100 μm (inset).

Full size image

Taken together, the present results demonstrate that MMP7 promotes tumor formation, but it is not necessarily required for tumor invasion or fibrosis in the SMAD4-deficient intestinal adenocarcinomas.

DISCUSSION

We have demonstrated here that the tumor epithelial cells of the cis-Apc/Smad4 adenocarcinomas express higher levels of MMP7 than those of the ApcΔ716 adenomas. These cells lack the wild-type Smad4 alleles, and are devoid of the SMAD4-dependent TGF-β family signaling.5, 7 Thus, our results with the mouse models suggest that inactivation of the TGF-β family signaling stimulates expression of MMP7 during the course of progression from colon adenomas to carcinomas. Consistently, it has been reported that expression of MMP7 is increased in human colorectal carcinomas compared with adenomas.26, 27

MMP7 has been implicated in benign colon adenomas. For example, lack of MMP7 reduces the number and size of the adenomas in ApcMin mice.17 However, it remains to be investigated whether MMP7 inactivation also suppresses development of tumors carrying mutations in Apc and an additional tumor suppressor (eg p53, Smad4 or Tgfbr2). In this study, we have demonstrated that loss of MMP7 reduces both number and size of tumors in the cis-Apc/Smad4 mice, suggesting that inhibition of MMP7 can prevent tumor formation initiated by the Apc inactivation in the presence of Smad4 mutation.

In human colon carcinomas, it has been reported that expression of MMP7 correlates with poor prognosis, advanced Dukes stage, liver metastasis and depth of invasion.9, 10, 11 Furthermore, cell line studies have shown that overexpression of MMP7 increases the invasiveness and metastatic potential of colon cancer cells,9, 13 and that an antisense oligonucleotide against MMP7 suppresses liver metastasis.28 These results suggest a significant role for MMP7 in tumor invasion and metastasis at late stages of colon cancer progression. On the other hand, we have shown here that lack of MMP7 does not suppress tumor invasion in the cis-Apc/Smad4 adenocarcinomas. It is worth noting that the cis-Apc/Smad4 adenocarcinomas do not carry mutations in the Ras (K-ras)5 or p53 (Trp53; data not shown) gene, suggesting that the tumors in the cis-Apc/Smad4 mice represent human colon carcinomas at relatively early stages. Thus, our results indicate that MMP7 is not required for invasion of such mouse tumors, and are consistent with the clinical data showing no significant correlation between MMP7 expression and depth of invasion in the early colorectal carcinomas.29 It is still conceivable, however, that MMP7 be involved in colon cancer progression such as metastasis beyond the stage examined here. Inhibitors against MMPs are in clinical development for various tumors and negative findings such as those reported in this study may provide an important framework of properly testing these drugs in the right group of patients.

In this study, we have shown that the cis-Apc/Smad4 adenocarcinomas show some fibrosis, ie, increased number of fibroblasts and higher levels of collagen type-I compared with the ApcΔ716 adenomas. To investigate the mechanisms underlying the tumor fibrosis, we analyzed mRNA samples from ApcΔ716 and cis-Apc/Smad4 polyps using DNA microarray.7 The results showed that almost all secreted factors, other than CCL9 7 and MMP7, were expressed at essentially the same levels between the two types of polyps. Consistent with such data, we found no differences between the ApcΔ716 and cis-Apc/Smad4 polyps in the levels of mRNA for profibrotic growth factors such as TGF-β, IGF, connective tissue growth factor, or platelet derived growth factor (Supplementary Figure 1). It is unlikely that the increased levels of CCL9 in the cis-Apc/Smad4 polyps stimulate the tumor fibrosis, because loss of its cognate receptor CCR1 does not affect the stromal expansion (data not shown). Accordingly, we have examined the possibility whether MMP7 can contribute to the tumor fibrosis through a posttranscriptional mechanism, eg, release or activation of profibrotic growth factors. Although lack of MMP7 reduces the experimental pulmonary fibrosis in a mouse model,23 the loss of MMP7 alone has not suppressed the fibrosis in the cis-Apc/Smad4 tumors. It is conceivable that other MMPs or proteinases compensate the defect in tumor fibrosis caused by the lack of MMP7. Further investigation will be required to elucidate the mechanism(s) underlying the stromal expansion in the cis-Apc/Smad4 polyps.

Using the cis-Apc/Smad4 mice, we have scrutinized here the role of MMP7 in the intestinal tumor progression in vivo. We have found that MMP7 promotes tumor expansion throughout development of the SMAD4-deficient intestinal adenocarcinomas, but it is not required for the invasion or fibrosis in these tumors. Although MMP7 is expressed essentially in the tumor epithelium, other invasion-associated MMPs such as MMP2 and MMP9 are primarily localized in the stromal cells in human colon cancer.8, 30, 31 In the cis-Apc/Smad4 mice, we have recently demonstrated that the MMP2/9-expressing stromal iMCs promote tumor invasion, without affecting the tumor size or multiplicity.7 Our findings, taken together, suggest separate roles for MMP7 and MMP2/9 in the development of some colon carcinomas. Namely, MMP7 supports the tumor formation, whereas MMP2/9 help tumor invasion of colon cancer whose SMAD4-dependent TGF-β family signaling is inactivated. Furthermore, the present results support our novel therapeutic strategy against tumor invasion, which targets MMP-producing stromal iMCs with CCR1 inhibitors, rather than direct inhibition of MMPs produced by the tumor epithelial cells.32