Transcriptional control of the RECK metastasis/angiogenesis suppressor gene

Abstract

The RECK gene is widely expressed in normal human tissues but is downregulated in tumor cell lines and oncogenically transformed fibroblasts. RECK encodes a membrane-anchored glycoprotein that suppresses tumor invasion and angiogenesis by regulating matrix-metalloproteinases (MMP-2, MMP-9 and MT1-MMP). Understanding of the transcriptional regulation of tumor/metastasis suppressor genes constitutes a potent approach to the molecular basis of malignant transformation. In order to uncover the mechanisms of control of RECK gene expression, the RECK promoter has been cloned and characterized. One of the elements responsible for the Ras-mediated downregulation of mouse RECK gene is the Sp1 site, to which Sp1 and Sp3 factors bind. Other regulatory events, such as DNA methylation of the RECK promoter and histone acetylation/deacetylation have been studied to understand the underlying mechanisms of RECK expression. Understanding of the mechanisms which control RECK gene transcription may lead to the development of new strategies for cancer prevention and treatment.

Introduction

Mutations in the ras proto-oncogenes are found in a wide variety of human tumors [1]. Ras proteins are essential components in several intracellular signaling pathways involved in regulating gene expression and various aspects of cellular behavior [2], [3], [4], [5]. Therefore, in order to understand the mechanism of cell transformation and propose novel strategies for cancer intervention, it is important to identify Ras targets that are relevant to manifestation of the malignant phenotype.

We have been isolating and characterizing genes which induce flat, normal fibroblast-like morphology (or ‘flat reversion’) when expressed in DT cells, a v-K-ras-transformed NIH-3T3 cell line [6]. Products of the reversion-inducing genes characterized so far, include a Ras-related small G-protein (Krev-1/Rap1A) [7], [8], a truncated form of the MSX-2 homeobox protein [9], a Kunitz-type serine protease inhibitor [10], previously termed TFPI-2 [11], [12] or PP5 [13] and RECK [14] ($\text{re}$version-inducing-$\text{c}$ysteine-rich protein with $\text{K}$azal motifs). This reversion-inducing gene encodes a glycoprotein (molecular weight 110,000 Da) with Kazal-type serine protease inhibitor-like domains that is associated with the cell membrane through a carboxy-terminal glycosilphosphatidylinositol (GPI)-modification [14].

Interestingly, the RECK gene is ubiquitously expressed in a number of normal tissues and non-neoplastic cell lines, while its expression is low or undetectable in many tumor-derived cell lines; also, RECK expression can be downregulated by several oncogenes, including ras [14], [15]. Restoration of RECK expression in malignant cells resulted in suppression of their invasive and metastatic activity, possibly by suppressing the secretion of matrix metalloproteinase (MMP)-9 and directly inhibiting its proteolytic activity [14]. RECK was recently found to negatively regulate two other MMPs, MMP-2, and MT1-MMP [16]. MMP-2, MMP-9, and MT1-MMP are all potentially involved in carcinogenesis [17], [18], [19], [20]. Taken together, these lines of evidence suggest that the RECK gene is an invasion and metastasis suppressor gene, whose downregulation by oncogenic signals contribute to manifestation of the malignant phenotype.

The biological functions of RECK have been recently explored in vivo through gene targeting experiments as well as tumor transplantation experiments in mice [16]. Mice lacking a functional RECK gene die around E10.5 with defects in collagen fibrils, the basal lamina, and vascular development; this phenotype was partially suppressed by MMP-2 null mutation. Vascular sprouting was dramatically suppressed in tumor derived from RECK-expressing fibrosarcoma cells grown in nude mice. These results support a role for RECK in the regulation of MMP-2 in vivo and implicate RECK downregulation in tumor angiogenesis [16].

Therefore, control of the MMP-inhibitor RECK gene expression has been analysed in order to gain insights into the mechanisms of tumor progression and to help with designing new strategies for cancer prevention and treatment.

RECK inhibits three MMP (matrix metalloproteinase) family members, namely, MMP-2, MMP-9, and MT1-MMP [14], [16]. The MMP family is essential during animal growth and development for proper ECM (extracellular matrices) remodeling [21], [22], which is important in certain pathological conditions, such as wound healing, osteoporosis, rheumatoid arthritis, and cancer [23].

Restoration of RECK expression of in malignant cells, such as B16-BL6 mouse melanoma and HT1080 fibrosarcoma, resulted in suppression of the invasive and metastatic activity of these tumor cells [14]. RECK negatively regulates MMP-9 through two different mechanisms, namely, suppression of pro-MMP-9 secretion and direct inhibition of MMP-9 enzymatic activity [14].

Processing of pro-MMP-2 takes place by two consecutive proteolytic cleavages which occur, preferentially, at the plasma membrane (reviewed in [24]). It has been reported that restored expression of RECK in the HT1080 fibrosarcoma cell line resulted in reduced amounts of the secreted intermediate form of MMP-2 and of active MMP-2 [16]. Biochemical analysis indicates that RECK regulates activation of pro-MMP-2 by inhibiting two proteolytic enzymes required for its processing, namely MT1-MMP and active MMP-2. [16].

These findings suggest that: (a) the function of the RECK protein is to suppress pro-MMP-9 secretion and to inhibit pro-MMP-2 processing at or near the cell surface; (b) the oncogene-induced downregulation of RECK contributes to the morphological transformation by deregulating these MMPs, thereby promoting excessive degradation of ECM around the cells (Fig. 1).

Cancer cells feeding depends on recruiting of blood vessels into the tumor mass (tumor angiogenesis). MMPs are known to play an important role during this step [25], [26]. To evaluate the patho-physiological implications of RECK expression, HT1080 cells were transfected with either a control vector or a RECK expressing vector [16]. Upon drug selection, pools of transfectants were inoculated sub-cutaneously into nude mice. Expression of high levels of RECK in tumor tissue was shown to limit the angiogenic sprouting and lead to massive tumor cell death. Interestingly, animals bearing the RECK-expressing tumors tended to live longer than those bearing the control tumors [16]. Moreover, a good correlation has been found between the abundance of RECK expression in hepatoma samples and patient survival [27]. Therefore, RECK expression can be a good prognostic indicator for cancer patients. Moreover, since RECK overexpression in tumors was found to affect tumor invasion, metastasis and angiogenesis, RECK may be a promising target for prevention and treatment of cancer invasion and metastasis.

Therefore, regulation of RECK metastasis/angiogenesis suppressor gene expression may help designing new strategies for cancer prevention and treatment.

Mutations in ras proto-oncogenes resulting in constitutive activation of the Ras signaling pathways are frequently found in a large variety of human tumors [1]. Ras is known to affect cell–CM interaction in several ways: activated Ras downregulates several ECM proteins such as fibronectin and collagen I as well as ECM-receptor (integrins) [28], [29], [30], while it upregulates several MMPs [31], [32]. These changes contribute, in concert, to manisfestation of the transformed phenotype, such as altered morphology, reduced cell adhesion, and invasive behavior. The RECK gene may be a critical target for Ras signaling since its expression is strongly suppressed in NIH-3T3 cells transformed by activated ras oncogenes and induces “flat reversion” when expressed in such transformed cells [14], [15]. Ras-mediated downregulation of RECK allows MMPs to display their full activity, adding another essential component for the Ras-induced phenotype. By exploring how the RECK gene is regulated, one may gain important insights into how oncogenic signals mediate malignant transformation. To this end, the mouse RECK gene promoter was isolated, characterized and one of the cis-regulatory elements responsible for RECK-downregulation in response to the activated ras oncogene was mapped [15].

The endogenous mouse RECK mRNA was clearly detected in untransformed mouse NIH-3T3 cells, but was downregulated in Ha-ras-transformed cells. Using the dual luciferase reporter system and co-transfection of activated ras oncogene, we demonstrated that this downregulation occurs at the transcriptional level [15]. The upstream 52-base promoter region was found to contain promoter activity, which is suppressed, to some extent, by the ras oncogene. This region contains two Sp1-binding motifs, one cEBP-binding motif and one CAAT box. Although both Sp1 sites were found to associate with Sp1 as well as with Sp3 transcription factors, Ras responsiveness seems to be mediated by only one of the Sp1 sites [15], indicated in Fig. 2 as Sp1(B).

The Sp1 transcription factor activates transcription by associating with one of the TATA-binding protein (TBP) co-activators in the TFIID complex. Interaction between glutamine-rich activation domains of Sp1 and the TBP-associated factor dTAF is an important component of the Sp1 transactivating activity [33]. Sp1 appears to provide a basal level of transcription but, in conjunction with other transcriptional activators or regulatory proteins, Sp1 can also participate in the dynamic regulation of gene expression. Sp1 activates transcription by cooperative interaction either with itself [34] or with other transcriptional factors, such as the papillomavirus E2 protein [35], Tat protein of the human immunodeficiency virus type 1 (HIV-1) [36], [37], NF-kB p65 [38], [39], Rb protein [40], GATA-1 [41], Egr-1 [42], Ap1 [43], [44], cEBPb [45], E2F [46], p53 [47], etc. These observations suggest that modulation of Sp1 activity may play a critical role in the regulation of cell proliferation and differentiation.

The existence of a family of Sp transcription factors [48], [49] suggests that gene regulation by Sp1 is more complex than had previously been assumed. Sp3, another member of this family that shares with Sp1 the same consensus binding sequence, was reported to be a dual-function regulator that can either induce or inhibit transcription, depending upon both the promoter and the cellular context [38], [50], [51], [52], [53]. Therefore, in the case of downregulation of the mouse RECK gene expression through the Sp1 site, an obvious possibility would be that Sp3 is induced or somehow activated in oncogene-transformed cells, thus occupying and blocking the Sp1 site. However, this hypothesis has not been supported by experimental data [15].

Sp1 has been shown to be involved in the Ras/Raf pathway. The Raf-mediated signaling pathway leads to alterations in Sp1 activity, resulting in higher levels of transcription of two growth-responsive genes, namely, rep-3b and mdr1 [54]. Sp1 binding sequences were found to be critical for the Ha-ras effect of activating transcription of the human 12-lipoxygenase gene [55]. The Sp1 binding site present in the VEGF promoter exerts an essential role in the Ras-mediated induction of this gene, which is mediated by the p42/p44 MAP kinase cascade [56]. Activation of the EGF receptor increases binding of the Sp1 transcription factor to the gastrin promoter region and stimulates its expression via the Ras-ERK signaling pathway [57]. Interestingly, it was reported that Sp1 plays a critical role in Erb-B2 and v-ras-mediated downregulation of the α-2-integrin gene in human mammary epithelial cells [58]. However, in contrast to the Ras-mediated downregulation of the mouse RECK gene [15], a slight reduction in binding of Sp1 to the critical Sp1 site was observed [58]. Thus, multiple mechanisms may exist in oncogene-mediated transcriptional suppression through Sp1 sites.

Sp1 can be regulated via post-translational modifications, such as phosphorylation [59], [60] and O-linked glycosylation [61]. Several studies suggest that the Sp1/Sp3 transcriptional activity can either increase or decrease, depending on the protein domain which is post-translationally modified. A decrease in the Sp1/Sp3 DNA binding activity by phosphorylation [62], [63], [64] or an increase in the Sp1 binding activity by dephosphorylation [64], [65], [66] was observed in some cases. On the other hand, studies correlating the Sp1/Sp3 protein phosphorylation with an increase in the DNA binding activity are available [57], [67], [68], [69], [70], [71]. Phosphorylation of the Sp1 transcription factor also seems to modulate protein-protein interactions, thus influencing its transactivating function [37], [72], [73], [74]. It has been suggested that Sp1 glycosylation competes with Sp1 phosphorylation at some specific residues of the protein, thus modulating its transcriptional activity [75]. It has also been shown that Sp1 glycosylation inhibits its interaction with itself or with the basal transcription machinery [76]. Another important role of Sp1 protein glycosylation seems to be the control of its stability [77].

The mitogenic effect of Ras and related signal pathways are quite complex and continue to be intensively studied, as are Sp1/Sp3 protein post-translational modifications. It has been shown that ERK2, one of the kinases known to be activated by Ras, phosphorylates Sp1 and stimulates its DNA binding activity in vitro [57]. However, a difference in the Sp1/Sp3 DNA binding activity was not observed in the study of RECK gene downregulation by activated Ras through the Sp1 site [15]. Therefore, it is possible that, using different pathways, activated Ras inhibits mouse RECK gene expression by modulating Sp1/Sp3 post-translational modifications and, in turn, affects its transactivation ability. Another possibility is that Ras signaling affects the interaction between Sp1/Sp3 and their regulatory protein(s). Notably, a 74,000 Mr protein which binds to the transactivation domain of Sp1 and substantially inhibits Sp1-mediated transactivation in vivo was identified [78]. Possible mechanisms of Ras-mediated dowregulation of the mouse RECK gene through the Sp1(B) site, by modulating the transactivation ability of Sp1/Sp3 transcription factors are shown in Fig. 2. Further studies to test these models should provide clues on how oncogenic signals suppress gene expression and lead to manifestation of the malignant phenotypes.

Genes whose expression is downregulated in the process of carcinogenesis are important candidates as tumor and invasion/metastasis suppressors. Analysis of their transcriptional regulation has been the object of intense study, since understanding of the molecular mechanism of malignant transformation may lead to the development of new strategies to detain cancer progression. Hypermethylation of CpG islands in the promoter regions of many cancer-related genes, which results in silencing of their expression, has been documented in various types of tumors [79]. Tumor and invasion/metastasis suppressor genes that have been found to be hypermethylated in human cancer cells and primary tumors include: p15 INK4B (cyclin kinase inhibitor) [80], p16 INK4A (cyclin kinase inhibitor) [81], [82], [83], p73 (p53 homology) [84], ARF/INK4A [85], Wilms tumor [86], von Hippel Lindau (VHL) [87], retinoic acid receptor-β (RARβ) [88], [89], estrogen receptor [90], androgen receptor [91], mammary-derived growth inhibitor [92], hypermethylated in cancer (HIC1) [93], retinoblastoma (Rb) [94], E-cadherin [95], [96], mts-1 [97], CD-44 [98], [99] and tissue inhibitor of metalloproteinase-3 (TIMP-3) [100]. However, not all tumor and invasion/metastasis suppressor genes expression were found to be inhibited by DNA hypermethylation. Clinical analysis of SMAD4(DPC4), a candidate tumor suppressor gene, suggests that hypermethylation of the SMAD4(DPC4) promoter region is not a frequent event in colorectal tumorigenesis [101]. It was reported that methylation of a CpG island within the promoter region of KAI1 metastasis suppressor gene is not responsible for the downregulation of KAI1 expression in invasive cancers or cancer cell lines [102], suggesting that events other than promoter hypermethylation, are responsible for reduced KAI1 expression in invasive bladder tumors and tumor cell lines.

Expression of maspin, a serine protease inhibitor with tumor/metastasis-suppressing activity in the mammary gland, is lost during tumor progression as a result of decreased transactivation through the Ets and Ap1 sites [103]. It has recently been reported that maspin expression is directly regulated by p53 [104]. Another report showed that aberrant cytosine methylation and heterochromatinization of the maspin promoter may silence maspin gene expression, thereby contributing to the progression of human mammary cancer [105]. Study of the nm23-H1 metastasis suppressor gene expression revealed that the metastasis-preventing effect of all trans-retinoic acid may partly result from upregulation of nm23-H1, and the metastasis-promoting effects of EGF and c-erbB-2/neu were probably mediated, in part, by the downregulation of this same gene [106]. BRCA1, a tumor suppressor gene, is transcriptionally repressed in a large portion of sporadic breast cancer patients by aberrant cytosine methylation, histone hypoacetylation and chromatin condensation of the BRCA1 promoter [107]. Alpha2-integrin also plays an important role in the malignant behavior of tumor cells, being downregulated by Erb-B2 and v-ras through the Sp1/Sp3 binding site present in the promoter region, as mentioned above [58]. In addition to RECK, another protease inhibitor gene that is negatively regulated by the ras oncogene product has recently been found [10]. It will be interesting to investigate whether a common mechanism underlies the ras downregulating effect of these genes.

Ras controls DNA methylation by inducing the expression of DNA methyl transferase (MeTase) [108], which is responsible for methylation of cytosine residues localized in the CpG dinucleotide sequences [108]. Therefore, Ras-mediated downregulation of the RECK gene could involve hypermethylation of the RECK promoter region. To test this hypothesis, we analyzed the effect of Ras overexpression in the methylation pattern of the mouse RECK promoter, a region displaying a high CpG content (that results in a large number of HapII restriction sites). Southern blot assays suggested that the mechanism of mouse RECK gene downregulation, mediated by Ras, does not involve hypermethylation of this promoter region (Sasahara RM, Takahashi C and Noda M, unpublished results). Several transformed cell lines (HeLa, HT1080, DT and B16) that do not express the RECK gene were treated to check whether DNA hypermethylation was the mechanism underlying RECK gene inactivation. To this end, these cell lines were treated with a demethylating agent (5-azacytidine) and subjected to Northern blot assays. The results suggest that DNA hypermethylation is not involved in RECK transcription inactivation in these cell lines (Sasahara RM, Takahashi C and Noda M, unpublished results).

It has been shown that the Sp1 transcription factor interacts with histone deacetylase 1, resulting in transcriptional repression of the thymidine kinase promoter through Sp1 sites [109]. The repressor action of histone deacetylase through Sp1 sites was also observed in the p21^WAF1 gene promoter in MG63 cells (derived from osteosarcoma) and in the NIH-3T3 cell line [110], [111]. Our finding that Ras-mediated downregulation of the RECK gene occurs through an Sp1 site [15] raises the possibility that the histone deacetylation mechanism may play a role in this process as well. This hypothesis was tested in NIH-3T3 cells using Trichostatin A (TSA), a specific inhibitor for histone deacetylase, but Northern blot analysis suggests that this mechanism is not involved in the control of the RECK gene transcription in NIH-3T3 cells (Brochado SM, Sasahara RM and Sogayar MC, unpublished results). It would be interesting to test this hypothesis in transformed cell lines which do not express the RECK gene.