Expression of matrix metalloproteinases during rat skin wound healing: evidence that membrane type-1 matrix metalloproteinase is a stromal activator of pro-gelatinase A.

Skin wound healing depends on cell migration and extracellular matrix remodeling. Both processes, which are necessary for reepithelization and restoration of the underlying connective tissue, are believed to involve the action of extracellular proteinases. We screened cDNA libraries and we found that six matrix metalloproteinase genes were highly expressed during rat skin wound healing. They were namely those of stromelysin 1, stromelysin 3, collagenase 3, gelatinase A (GelA), gelatinase B, and membrane type-1 matrix metalloproteinase (MT1-MMP). The expression kinetics of these MMP genes, the tissue distribution of their transcripts, the results of cotransfection experiments in COS-1 cells, and zymographic analyses performed using microdissected rat wound tissues support the possibility that during cutaneous wound healing pro-GelA and pro-gelatinase B are activated by MT1-MMP and stromelysin 1, respectively. Since MT1-MMP has been demonstrated to be a membrane-associated protein (Sato, H., T. Takino, Y. Okada, J. Cao, A. Shinagawa, E. Yamamoto, and M. Seiki. 1994. Nature (Lond.). 370: 61-65), our finding that GelA and MT1-MMP transcripts were expressed in stromal cells exhibiting a similar tissue distribution suggests that MT1-MMP activates pro-GelA at the stromal cell surface. This possibility is further supported by our observation that the processing of pro-GelA to its mature form correlated to the detection of MT1-MMP in cell membranes of rat fibroblasts expressing the MT1-MMP and GelA genes. These observations, together with the detection of high levels of the mature GelA form in the granulation tissue but not in the regenerating epidermis, suggest that MT1-MMP and GelA contribute to the restoration of connective tissue during rat skin wound healing.

The mouse one P-domain (pS2) and two P-domain (mSP) genes exhibit distinct patterns of expression.

We have previously shown that the human pS2 gene, which codes for a secreted peptide of 60 amino acids, is expressed in a number of human carcinomas, including carcinomas of the breast, the pancreas, and the large bowel. Strong pS2 gene expression was also observed in the normal gastric mucosa and in the regenerative tissues surrounding ulcerous lesions of the gastrointestinal tract. A number of pS2 similar peptides, designated as P-domain peptides, have been described, notably the porcine (PSP), murine (mSP), and human (hSP) spasmolytic polypeptides, which correspond to duplicated pS2 proteins. We have now cloned a mouse homolog of the human pS2 cDNA to dispose of an animal model to study the pS2 protein function, which remains unknown at the present time. We show that the mouse putative pS2 protein sequence and the physiological pattern of expression of the mouse pS2 gene are well conserved. The mouse pS2 gene is highly expressed in the stomach mucosa cells, whereas no pS2 gene expression could be detected in the mouse mammary gland, even during postnatal development processes dependent on growth factors or hormones. Using in situ hybridization, we show that although coexpressed in the fundus, the antrum and the antrum-pyloric regions of the stomach, the mouse pS2 and mSP genes exhibit distinct and complementary cellular patterns of expression.

The breast cancer-associated stromelysin-3 gene is expressed during mouse mammary gland apoptosis.

We have cloned from a mouse placenta cDNA library a mouse homologue of the human stromelysin-3 (ST3) cDNA, which codes for a putative matrix metalloproteinase expressed in breast carcinomas. The ST3 protein is well conserved between humans and mice, and the pattern of ST3 gene expression is similar in both species, and shows expression in the placenta, in the uterus, and during limb bud morphogenesis. We show that the ST3 gene can also be expressed in the normal mouse mammary gland. ST3 gene expression was not detected during mammary growth, neither in virgin nor in pregnant mice, but was specifically observed during postlactating involution of the gland, an apoptotic process associated with intense extracellular matrix remodeling. ST3 transcripts were found in fibroblasts immediately surrounding degenerative ducts, suggesting that ST3 gene expression may be associated with the basement membrane dissolution, which occurs during mammary gland involution. Since the ST3 gene is also specifically expressed in fibroblastic cells surrounding invasive neoplastic cells of breast carcinomas, we suggest that ST3 is implicated in extracellular matrix remodeling processes common to mammary apoptosis and breast cancer progression.

In vivo evidence that the stromelysin-3 metalloproteinase contributes in a paracrine manner to epithelial cell malignancy.

Stromelysin-3 (ST3; Basset, P., J.P. Bellocq, C. Wolf, I. Stoll, P. Hutin, J.M. Limacher, O.L. Podhajcer, M.P. Chenard, M.C. Rio, P. Chambon. 1990. Nature. 348:699-704) is a matrix metalloproteinase (MMP) expressed in mesenchymal cells located close to epithelial cells, during physiological and pathological tissue remodeling processes. In human carcinomas, high ST3 levels are associated with a poor clinical outcome, suggesting that ST3 plays a role during malignant processes. In this study we report the ST3 gene inactivation by homologous recombination. Although ST3 null mice (ST3-/-) were fertile and did not exhibit obvious alterations in appearance and behavior, the lack of ST3 altered malignant processes. Thus, the suppression of ST3 results in a decreased 7, 12-dimethylbenzanthracene-induced tumorigenesis in ST3-/- mice. Moreover, ST3-/- fibroblasts have lost the capacity to promote implantation of MCF7 human malignant epithelial cells in nude mice (P < 0.008). Finally, we show that this ST3 paracrine function requires extracellular matrix (ECM)-associated growth factors. Altogether, these findings give evidence that ST3 promotes, in a paracrine manner, homing of malignant epithelial cells, a key process for both primary tumors and metastases. Therefore, ST3 represents an appropriate target for specific MMP inhibitor(s) in future therapeutical approaches directed against the stromal compartment of human carcinomas.

Breast cancer-associated pS2 protein: synthesis and secretion by normal stomach mucosa.

The human pS2 gene is specifically expressed under estrogen transcriptional control in a subclass of estrogen receptor-containing human breast cancer cells. The pS2 gene encodes an 84-amino acid protein that is secreted after signal peptide cleavage. The distribution of pS2 protein in normal human tissues was studied with antibodies to pS2; pS2 was specifically expressed and secreted by mucosa cells of the normal stomach antrum and body of both female and male individuals. Moreover, no estrogen receptor could be detected in these cells, indicating that pS2 gene expression is estrogen-independent in the stomach. The function of the pS2 protein in the gastrointestinal tract is unknown. However, the pS2 protein is similar in sequence to a porcine pancreatic protein that has been shown to inhibit gastrointestinal motility and gastric secretion.

Gastric mucosa abnormalities and tumorigenesis in mice lacking the pS2 trefoil protein.

To determine the function of the pS2 trefoil protein, which is normally expressed in the gastric mucosa, the mouse pS2 (mpS2) gene was inactivated. The antral and pyloric gastric mucosa of mpS2-null mice was dysfunctional and exhibited severe hyperplasia and dysplasia. All homozygous mutant mice developed antropyloric adenoma, and 30 percent developed multifocal intraepithelial or intramucosal carcinomas. The small intestine was characterized by enlarged villi and an abnormal infiltrate of lymphoid cells. These results indicate that mpS2 is essential for normal differentiation of the antral and pyloric gastric mucosa and may function as a gastric-specific tumor suppressor gene.

TRAF4 overexpression is a common characteristic of human carcinomas.

Tumor necrosis factor receptor (TNFR) associated factor 4 (TRAF4) was initially identified as a gene amplified and overexpressed in breast carcinomas. Our aim was to evaluate whether TRAF4 protein overexpression exists in other cancer types. Immunohistochemistry analysis of tumor samples from 623 patients with 20 different tumor types showed that TRAF4 was overexpressed in 268 tumors (43%), including 82 of 137 lung adenocarcinomas (60%). Interestingly, 32 primary tumors and their matching metastases exhibited mostly similar TRAF4 expression pattern. TRAF4 protein overexpression was limited to cancer cells and the subcellular localization was consistently cytoplasmic in a large majority of cases. To investigate changes in TRAF4 gene copy number, 125 cases from six different types of carcinomas were also analysed by fluorescence in situ hybridization. Out of the 28 cases (22%) showing an increased TRAF4 gene copy number, 23 (82%) were overexpressing the protein. Thus, TRAF4 gene amplification is one of the mechanisms responsible for TRAF4 protein overexpression in human cancers. Considering that TRAF4 is located at 17q11.2 in a region of amplification devoid of known oncogenes and is commonly overexpressed in cancer, our data support an oncogenic role for TRAF4.

Stromelysin-3 expression promotes tumor take in nude mice.

Stromelysin-3 (ST3) is a matrix metalloproteinase expressed in human carcinomas in ways suggesting that it may play a role in tumor progression. To test this possibility, we have performed gene transfer experiments using both anti-sense and sense ST3 expression vectors, and malignant cells either expressing (NIH 3T3 fibroblasts) or not (MCF7 epithelial cells) endogenous ST3. We have compared the ability of parental and transfected cells to cause subcutaneous tumor development in nude mice. 3T3 cells expressing anti-sense ST3 RNA showed reduced tumorigenicity, and MCF7 cells expressing mouse or human ST3 were associated with reduced tumor-free period leading to a significant increased tumor incidence(P<10(-4)). However, once established, the ST3 expressing tumors did not grow faster than those obtained with the parental MCF7 cell line. In addition, tumors obtained after sub-cutaneous injection of ST3-expressing or nonexpressing cells did not exhibit obvious histological differences, and careful examination did not reveal any local invasive tissue areas nor systemic metastases. These in vivo observations were in agreement with those obtained in vitro showing that ST3 expression did not modify proliferative nor invasive properties of transfected cells. Altogether, these results indicate that ST3 expression promotes tumor take in nude mice, presumably by favoring cancer cell survival in a tissue environment initially not permissive for tumor growth. These findings represent the first experimental evidence showing that ST3 can modulate cancer progression.

[Paradoxical interactions between pregnancy and breast cancer]. / Effets paradoxaux de la grossesse sur le cancer du sein.

Interactions between pregnancy and breast cancer are complex and paradoxical. Epidemiological data show that nulliparity and late full-term pregnancy increase breast cancer risk. By contrast, early full-term pregnancy and multiparity are thought to be the most effective means of decreasing lifetime breast cancer risk. Paradoxically, young women diagnosed with breast cancer during pregnancy have a higher risk of dying from their disease. Moreover, there is a transient increase in risk of breast cancer in the first three to four years after pregnancy. After breast cancer treatment, there is no evidence that pregnancy increases the risk of breast cancer recurrence. Thus, it is not contraindicated in women previously treated for breast cancer and free of recurrence. Various physio-pathological mechanisms are involved in the protective effect of pregnancy, like cellular differentiation of mammary cells, mammary gland involution, circulating anti-mucin antibody and excretion in the milk of breast carcinogens. In the past, unfavorable effects of pregnancy were mainly attributed to precancerous cell proliferation induced by pregnancy-associated hormonal changes. However, recent studies suggest that the remodeling of cellular microenvironment and extracellular matrix during pregnancy and involution may contribute to enhanced invasive and metastatic potential of breast carcinomas.

[Proteomics and breast cancer]. / Protéome et cancer du sein.

Breast cancer is the first cause of death between 35 and 55 years. Genetic alterations and modifications in gene expression are found during different steps of tumor progression. These changes are translated at the protein level where quantitative and qualitative modifications are found in tumor compared to normal samples. Similarly to studies aimed at deciphering transcriptional changes important in cancer, proteomic approaches allow the global and comparative study of proteins in normal and pathological samples. The objective of this article is to present common proteomic methods and to review the first published results concerning proteomics studies applied to breast cancer with an emphasis on reports obtained using the SELDI-TOF MS (Surface Enhanced Laser Desorption Ionization Time-Of-Flight Mass Spectrometry). In breast cancer, it is possible to explore the tumoral proteome and/or the blood derived proteome. The first studies are aimed at globally understanding the disease while the latter are aimed at discovering serum proteins or biomarkers useful for the early detection, diagnosis, prognosis and management of cancer. Promising results are obtained using these emerging methods and these novel biomarkers should be validated in the future and will have an important impact for the management of breast cancer patients.

[Circulating proteinic biomarkers and breast cancer]. / Marqueurs protéiques circulants et cancer du sein.

Circulating proteinic biomarkers are secreted by tumor cells or by their environmental cells and they have a variable specificity. In case of breast cancer, carcino-embryonic antigen (CEA) was for a long time the only circulating biomarker used. Nowadays, the most useful biomarkers measure circulating levels of fragments of MUC1-polymorphic epithelial mucin (MUC1-PEM): cancer antigen (CA) 15.3, mucin-like carcinoma-associated antigen (MCA), CA 27-29, CA 549... They are useful for general disease follow-up. Other circulating markers belonging to keratins (tissue polypeptide antigen, TPA, TPS or Cyfra 21.1) are correlated with proliferative activity of breast tumors. More recently, the measure of the c-erb B2 circulating part (extra cellular domain, ECD) was proposed as a prognostic biomarker for breast tumors with c-erb B2 overexpression. Moreover, the determination of urinary level of trefoil factor1 (PS2-TFF1) might be useful for the follow-up of hormonodependent breast cancers. The present review describes the clinical interest of these different circulating biomarkers in case of breast cancer, emphasizing their biological characteristics.

Two distinct amplified regions at 17q11-q21 involved in human primary breast cancer.

Chromosomal segment 17q11-q21 is a commonly amplified region in human breast carcinomas. Several lines of evidence suggest that ERBB2 is the gene responsible for the emergence of this amplicon, but four novel genes (called MLN 50, MLN 51, MLN 62, and MLN 64) in 17q11-q21 have recently been found to be amplified and overexpressed in breast cancer cell lines. We investigated 98 primary breast tumors for amplification of these five loci. Twenty-five tumors (25.5%) showed amplification of at least one of these markers, but most amplifications did not encompass all of the tested loci. The genes most frequently amplified were ERBB2 and MLN 64 (22 of 25 amplified cases). MLN 64 was always coamplified with ERBB2, and to a similar level. Amplification of these five genes always leads to overexpression of their mRNA; we observed no cases of overexpression without amplification in any of these genes. Our results suggest that: (a) an independent, amplified region defined by MLN 62 (also called CART1 or TRAF4) is located in 17q11-q12; (b) in addition to ERBB2, MLN 64 is a major target for the 17q12-q21 amplicon; and (c) these MLN genes could be of pathogenetic significance in breast cancer.

Identification of CGA as a novel estrogen receptor-responsive gene in breast cancer: an outstanding candidate marker to predict the response to endocrine therapy.

The estrogen receptor (ER) status of breast tumors is used to identify patients who may respond to endocrine agents such as tamoxifen. However, ER status alone is not perfectly predictive, and there is a pressing need for more reliable markers of endocrine responsiveness. Here, we identified the well-known CGA gene (coding for the alpha subunit of glycoprotein hormones) as a new ERalpha-responsive gene in human breast cancer cells. We used a real-time quantitative reverse transcription-PCR assay to quantify CGA mRNA copy numbers in a large series of breast tumors. CGA overexpression (> 10 SD above the mean for normal breast tissues) was observed in 44 of 131 (33.6%) breast tumor RNAs, ranging from 20 to 16,500 times the level in normal breast tissues; the highest levels of CGA gene expression were close to those observed in placenta. Significant links were observed between CGA gene overexpression and Scarff-Bloom-Richardson histopathological grade I+II (P = 0.015), and progesterone (P = 0.0009) and estrogen (P < 10(-7)) receptor positivity, which suggested that CGA is a marker of low tumor aggressiveness. We observed CGA mRNA overexpression in 44 of 90 (48.9%) ERalpha-positive tumors and in none of the 41 ERalpha-negative tumors. Immunohistochemical studies demonstrated that human chorionic gonadotropin alpha protein was strictly limited to ERalpha-positive tumor cells. Overexpression of the CGA gene was not accompanied by overexpression of the CGB gene. Our results also suggest that CGA could be a more reliable marker than PS2 and PR for ERalpha functionality and, thus, for endocrine responsiveness. Moreover, the CGA marker has the added value of dichotomizing ERalpha-positive patients into two subgroups of similar size. Specific antibodies directed to secreted human chorionic gonadotropin alpha protein are commercially available, thus facilitating the future application of this marker to the clinical management of breast cancer.

A screening method to identify genes commonly overexpressed in carcinomas and the identification of a novel complementary DNA sequence.

We describe a differential screening method for cDNA libraries which used a combination of subtracted and PCR-amplified cDNA probes, and which can be applied to the selection of genes expressed in multiple tissues. This technique was used to identify genes commonly overexpressed in breast and basal cell carcinomas. These represent stromally dependent, invasive tumors with and without metastatic capacity. Thus, this screening sought to identify genes involved in the early stages of tumor progression. We identified a total of 16 genes, including c-erbB-2 and tissue inhibitor of metalloproteinases 3 whose products have been implicated in tumorigenesis or invasion. We also identified a novel sequence (D52) showing little homology with others described in any species, which maps to the human chromosomal band 8q21. In situ RNA hybridizations of breast carcinoma sections indicated that the D52 gene was expressed in cancer cells, whereas other genes identified in the differential screening were expressed in fibroblastic or inflammatory cells within the tumor stroma. Thus, the procedure developed in this study selected genes expressed in a diversity of cell types, indicating its potential usefulness in other systems.

pS2 expression and response to hormonal therapy in patients with advanced breast cancer.

Seventy-two patients with advanced breast carcinoma (42% bone, 25% visceral, 5.5% soft tissue, and 27.5% multiple site metastases) were evaluated to determine the relationship between tumor expression of the estrogen-regulated protein pS2, estrogen receptor (ER) or progesterone receptor (PgR) content, and response to hormonal therapy. Twenty-nine % of tumors were pS2 positive, 64% were ER positive, and 29% were PgR positive. Of the ER-positive patients (n = 43), 15 (35%) had greater than 10% of the invasive carcinoma which immunostained for pS2 (these were considered pS2 positive). Only 3 of 24 ER-negative tumors were pS2 positive. A weak association between pS2 expression and ER content (P = 0.08) but not PgR content was observed. Of pS2-positive patients, 52% had a partial or complete response to hormonal therapy. In 24% of pS2-positive patients the disease stabilized with treatment. In contrast, 27% of pS2-negative patients had a partial or complete response. In 10% of these patients the disease stabilized. Similar associations between therapeutic response and ER or PgR were not observed. The odds of having a clinical response to hormonal therapy was greater for pS2-positive than for ER- or PgR-positive tumors. pS2 expression may define a subset of ER-positive tumors that are more likely to respond to hormonal treatment.

Identification of a new interferon-alpha-inducible gene (p27) on human chromosome 14q32 and its expression in breast carcinoma.

A new complementary DNA, p27, has been cloned and sequenced from estradiol-treated MCF7 human breast carcinoma cells. It encodes a putative highly hydrophobic protein of 122 amino acids which has a 33% overall sequence similarity to the product of the 6-16 gene (R. L. Friedman, S. P. Manly, M. McMahon, I. M. Kerr, and G. R. Stark, Cell, 38: 745-755, 1984), which is transcriptionally induced by interferons of the alpha/beta type. We demonstrate here that the p27 gene, which is located in band q32 of human chromosome 14, is also induced by interferon-alpha in human cell lines of different origin and that expression is independent of the presence of estradiol receptor in the cells. High levels of p27 RNA were found in vivo in approximately 50% of primary human breast carcinomas (21 were tested by Northern blotting). In situ hybridization to some of the p27-overexpressing tumors showed that the p27 RNA is localized in cancer cells and sometimes also in fibroblastic cells of tumor stroma. p27 RNA levels in the tumors did not correlate with the presence of estrogen receptor or with the expression of the estrogen-induced pS2 gene. Further studies are now necessary to elucidate the cause of p27 gene overexpression in breast carcinoma and in particular to determine whether it corresponds to chromosomal rearrangements in the 14q32 region and/or to induction by interferons of the alpha/beta type.

Increased stromelysin 3 gene expression is associated with increased local invasiveness in head and neck squamous cell carcinomas.

Matrix metalloproteinases are believed to play an important role in tumor invasion and metastasis. To examine the expression of the stromelysin 3 (ST3) gene, a new member of the matrix metalloproteinase gene family, 111 head and neck squamous cell carcinomas and 21 metastatic lymph nodes were analyzed by Northern blot. ST3 gene expression was observed in 106 carcinomas and 19 metastatic nodes, but in only 2 of 60 samples of corresponding normal tissue tested in parallel. ST3 RNA, by in situ hybridization, and ST3 protein, by immunohistochemical analysis, were specifically detected in fibroblastic cells immediately surrounding invasive cancer cells. This fibroblastic expression of the ST3 gene is characteristic among the matrix metalloproteinase genes known to be overexpressed in head and neck carcinomas, since stromelysin 2 transcripts were specifically detected in neoplastic cells, and type I collagenase transcripts in both neoplastic cells and stromal fibroblasts. Furthermore, there was a highly significant positive correlation (P < 0.0001) between ST3 RNA levels and local invasiveness by the cancer cells, suggesting that enhanced expression of the ST3 gene may contribute to the neoplastic phenotype in head and neck carcinomas.

Prediction of relapse and survival in breast cancer patients by pS2 protein status.

Application of systemic adjuvant therapy for primary breast cancer patients requires a more accurate identification of patients at high risk for recurrence. We have quantitatively assessed the cytosolic levels of estrogen-regulated pS2 protein in tumors of 205 breast cancer patients (median follow-up, 47 mo). There were no significant associations between the level of pS2 protein and tumor size, lymph node status, and differentiation grade. Using length of relapse-free survival (RFS) and overall survival (OS) as end points, 11 ng of pS2 protein/mg of cytosol protein were found as the best cutoff level to discriminate between positive (pS2+) and negative (pS2-). Patients with pS2- tumors showed significantly shorter RFS and OS (P less than 0.0001) than patients with pS2+ tumors. Also after adjustment for tumor size, lymph node status, and estrogen receptor (ER) status, pS2 negativity was associated with earlier recurrence and death. Tumors positive for pS2 (55 of 205, 27%) were almost exclusively confined to the subclass of ER+ tumors (53 of 55, 96%). The death rate for patients with pS2+ tumors was one-tenth of the death rate for patients with pS2-/ER- tumors. In the patients with ER+ tumors, the prognostic power of the pS2 status was especially present in patients whose tumors were also positive for the progesterone receptor (5-yr RFS and OS, 85% and 97% for ER+/PgR+/pS2+ tumors compared with 50% and 54% for the patients with ER+/PgR+/pS2- tumors). In patients with axillary lymph node involvement (N+), pS2 status could discriminate strongly between a good and bad prognosis group (5-yr RFS and OS, 65% and 88% for N+/pS2+ compared with 32% and 34% for N+/pS2-). A similar phenomenon was observed in patients without axillary lymph node involvement (5-yr RFS and OS, 89% and 95% for N0/pS2+ compared with 58% and 82% for N0/pS2-). We conclude that the pS2 status of human primary breast tumors is an important variable for the identification of patients at high risk for recurrence and death. Knowledge of the cytosolic pS2 status appeared of particular importance to identify patients at high risk in the ER+/PgR+ subclass of tumors, and in both the N0 and N+ subclasses of patients.

High cancer cell death in syngeneic tumors developed in host mice deficient for the stromelysin-3 matrix metalloproteinase.

Matrix metalloproteinases (MMPs) are extracellular enzymes. Some of them are known to be involved in tumor development and/or progression. Several cellular functions have been proposed for MMPs during malignant processes. Notably, they may be involved in tissue-remodeling processes through their ability to digest matrix components or to participate in tumor neoangiogenesis and, subsequently, in cancer cell proliferation. One of these MMPs, stromelysin-3 (ST3/MMP11), although devoid of enzymatic activity against the matrix components, is associated with human tumor progression and poor patient clinical outcome. Using several in vivo experimental models, it has been demonstrated that ST3 expression by the fibroblastic cells surrounding malignant epithelial cells promotes tumorigenesis in a paracrine manner. The present study was devoted to the identification of the cellular function underlying this ST3-induced tumor promotion using a syngeneic tumorigenesis model in mice. Our results show that ST3 exhibits a new and unexpected role for a MMP, because ST3-increased tumorigenesis does not result from increased neoangiogenesis or cancer cell proliferation but from decreased cancer cell death through apoptosis and necrosis. Thus, during malignancy, the cellular function of ST3 is to favor cancer cell survival in the stromal environment.

Demonstration in vivo that stromelysin-3 functions through its proteolytic activity.

Stromelysin-3 (ST3), a matrix metalloproteinase (MMP) expressed in aggressive carcinomas, has been shown to promote tumor development in different in vivo experimental models. However, the inability of its mature form to degrade extracellular matrix components casts doubt on whether ST3 functions in vivo as a protease. In this study, we evaluated whether the ST3 tumor-promoting effect could be ascribed to its proteolytic activity and whether this putative protease could be targeted with MMP inhibitors. Catalytically inactive mutant cDNA of human (h) ST3 or mouse (m) ST3 were generated and transfected into MCF7 cells. When injected into nude mice in the presence of matrigel, the mutant-bearing cells did not exhibit the enhanced tumorigenicity elicited by MCF7 cells transfected with wild-type ST3 cDNA. In a second approach, TIMP2 overproduction in MCF7 cells expressing hST3 was induced by retroviral infection. The co-expression of ST3 and TIMP2 failed to enhance the tumorigenicity of MCF7 cells. Notably, matrigel depleted of low-molecular-weight proteins and growth factors failed to promote the tumorigenicity of ST3-expressing MCF7 cells. These findings provide the first in vivo evidence that ST3 is indeed a protease that can modulate cancer progression by remodeling extracellular matrix and probably by inducing it to release the necessary microenvironmental factors. Thus, ST3 represents an interesting target for specific MMP inhibition.