Background
Functional inactivation of the tumor suppressor Smad4 in colorectal and pancreatic carcinogenesis occurs coincident with the transition of premalignant precursor lesions – adenomas and pancreatic intraepithelial neoplasias (PanINs), respectively, to invasive and metastatic growth [
1‐
4]. The hallmark of invasive growth is loss of the basement membrane (BM) barrier. BMs are specialized sheet-like structures of the extracellular matrix that separate epithelia from the underlying mesenchyme. They are built through interaction of epithelial and mesenchymal cells, which both provide components, mainly the laminins and collagen IV [
5]. The epithelial-derived laminins constitute a family of at least 15 different isoforms in mammals, each an α
xβ
yγ
z heterotrimer derived from a combination of one, each, out of five α-, three β- and three γ-glycoprotein subunits [
6‐
8]. In the gastrointestinal tract a single-layered epithelial sheet is separated from the underlying mesenchyme through a BM containing laminins- 111, -211, -332, -511 and -521 (laminins 1, 2, 5, 10 and 11), which are expressed in a characteristic regional and development dependent pattern [
5,
9,
10]. Also, in the normal adult pancreas, a single layer of ductal cells – the presumptive precursors of pancreatic adenocarcinomas, is separated from mesenchymal cells through a BM containing laminin-332 (LM-332) [
11‐
13].
Loss of the BM barrier upon the transition to invasive growth can be due either to proteolytic degradation or to decreased synthesis of BM components [
14]. Proteolytic degradation of BM in carcinomas has been intensively investigated. It appears to be predominantly executed through proteases expressed mainly by stromal cell types like activated fibroblasts and inflammatory cells, which can be recruited through signals originating from the tumor cells. Molecular mechanisms underlying decreased synthesis of BM components, on the other hand, have rarely been addressed.
To unravel the mechanisms that underlie Smad4-mediated tumor suppression we have established derivatives of Smad4-deficient human colorectal and pancreatic carcinoma cells, in which Smad4 is stably restored through gene transfer. Using this approach we could proof Smad4's tumor suppressor function and could identify Smad4 target genes, among them VEGF and E-cadherin [
15‐
18]. Recently, we have unraveled, that LM-332, composed of α3-, β3- and γ2-chains, is another relevant target structure of Smad4. We have shown that all three genes encoding LM-332, LAMA3, LAMB3 and LAMC2, are under positive transcriptional control of Smad4. Smad4 increased basal and/or TGFβ-induced expression of LM-332 in Smad4-reexpressing colon and pancreatic cancer cells leading to a huge increase in the extracellular release of the heterotrimer and to the deposition in BM-like structures at contact sites with fibroblasts [
19].
LM-332 expression is tightly controlled in normal epithelia; adenomas consistently retain normal staining patterns for LM-332 in BMs [
20]. In colorectal carcinomas, in contrast, laminin deposition in BM structures becomes discontinuous or is absent suggesting that shut-down of laminin expression is associated with genetic alterations that mediate the transition to invasive growth [
8,
14,
21]. Thus, our finding that the tumor and invasion suppressor Smad4 can act as a positive regulator of LM-332 is consistent with current knowledge.
Here, we wished to further decipher the molecular mechanisms of how Smad4 acts as a positive transcriptional regulator of constitutive and of TGFβ-induced expression of the three genes encoding LM-332. Smad4 encodes an intracellular messenger common to all signaling cascades induced by members of the TGFβ superfamily of cytokines through the canonical pathway [
22,
23]. Cellular signaling from the TGFβ family is initiated by binding of the ligand to transmembrane receptor serine/threonine kinases, TβRI and TβRII. Activated TGFβ receptors stimulate the phosphorylation of receptor-regulated Smad (R-Smad) proteins, which in turn form complexes with Smad4 that accumulate in the nucleus. Here, the Smad complex can bind to DNA directly, at so-called SBE sites (Smad binding element), but with low binding affinity, only, and can also bind to and interact with a plethora of other transcription factors, coactivators or corepressors among them transcription factors of the AP1 and Sp1 families. High affinity binding of the Smad complex to a promoter is thought to occur through the incorporation of an additional transcription factor into the R-Smad-Smad4 complex, which binds to its respective cognate sequence [
22,
23]. Adding to the complexity of cellular signaling networks TGFβ besides the canonical pathway fuels into further signaling cascades like the MAP kinase pathway [
24,
25]. TGFβ has previously been identified as a positive regulator of LM-332 in diverse cell types among them epidermal keratinocytes and gastric adenocarcinoma cells [
26,
27]. AP1 sites were found to confer TGFβ responsiveness of the LAMA3 and the LAMC2 promoter -encoding for α3 and γ2-chains- in murine keratinocytes and human colon carcinoma cells, respectively [
28‐
31]. No reports, to our knowledge, have yet been published addressing molecular mechanisms of LAMB3 -encoding for β3-chain- promoter regulation.
We have reported earlier, that an SBE site is functional and is involved in Smad4-mediated TGFβ induction of LAMA3 expression [
19].
In silico analyses detected putative SBE sites also in the promoter regions of the LAMB3 and the LAMC2 genes but have not yet been functionally analyzed.
Here, we present detailed studies of the three promoters of the genes encoding LM-332 in human colorectal adenoma cells and in Smad4-deficient and Smad4-reexpressing colorectal and pancreatic carcinoma cells. We confirm that the SBE site at -1.5 kb confers one part of Smad4-dependent TGFβ induction of LAMA3 expression and that the downstream AP1 sites are additionally involved. On the other hand, whereas each, three, putative SBE sites were identified in the LAMB3 and the LAMC2 promoter through in silico analyses, each of these sites proved non-functional. Rather, TGFβ induction is conferred through AP1-sites and through a single Sp1 site in both of these promoters. In summary, our results show, that whereas Smad4 functions as a positive transcriptional regulator of all three genes encoding LM-332, the underlying mechanisms are surprisingly complex and substantially diverse.
Discussion
Molecular mechanisms and target genes through which Smad4 mediates its tumor suppressor function are still incompletely understood. We have previously reported that Smad4 is a positive regulator of the three genes which encode for the heterotrimeric Laminin-332 (LM-332) molecule, a prominent component of basement membranes (BMs). Reexpression of Smad4 in Smad4-deficient tumor cells led to secretion and deposition of the heterotrimeric molecule in BM-like structures and was associated with reversion from mesenchymal-like to epithelial morphology, with suppression of invasiveness
in vitro and suppression of tumor growth
in vivo [
15‐
19]. Expression control of an essential BM component thus constitutes an important function of the tumor suppressor Smad4.
Smad4 is the central mediator of TGFβ responses through the canonical TGFβ/Smad signal cascade. Smad4 is the single co-Smad that forms complexes with receptor-Smads, which then translocate into the nucleus where they bind to Smad binding elements (SBE) in the promoter region of target genes. In addition, Smad complexes can be targeted to DNA by interacting with ubiquitous sequence-specific DNA-binding transcription factors like AP1 and Sp1 [
22]. TGFβ, in addition to the canonical pathway activates further signal cascades like the MAPK/JNK pathways, which in turn can be modified through cross-talk with the Smads [
24,
25].
Addressing the molecular mechanisms through which Smad4 mediates transcriptional regulation of the three LM-332 genes we here provide evidence for three different mechanisms: First, Smad4 binds to a functional SBE site, this mechanism is operative in the LAMA3 promoter, exclusively. Secondly, Smad4 binds to AP1 (and Sp1) sites in all of the three promoters presumably via interaction with AP1 family components and lastly, Smad4 mediates transcriptional induction of AP1 factors. As AP1 sites are intimately involved in transcriptional regulation of laminin genes this Smad4 impact on AP1 factors may represent an important, though indirect mechanism of how Smad4 could affect target gene transcription even in the absence of direct binding at the respective promoters.
Whereas Smad4 increases constitutive expression levels and TGFβ responses of all three genes encoding LM-332, the underlying mechanisms are surprisingly complex and substantially diverse. To start to decipher mechanisms and pathways involved we here used transient transfection assays using promoter-reporter constructs and showed that their activities strictly reflected responses of the endogenous genes. The colorectal adenoma cell line LT97 used as a model for premalignant cells displayed increased expression of all three LM-332 genes in response to TGFβ. Likewise, SW480 colorectal and BxPC3 pancreatic carcinoma cells showed transcriptional induction of all three genes in a Smad4-dependent manner, whereas their Smad4-negative counterparts remained unaffected by TGFβ treatment. The LAMA3 promoter, only, harbored a functional SBE at -1.5 kb mediating Smad4-dependent TGFβ induction of LAMA3 promoter activity. In addition, downstream AP1 sites were shown to be involved in basal promoter activity and, in combination with the SBE, to mediate Smad4-dependent constitutive and TGFβ-induced promoter activity.
The LAMB3 promoter to our knowledge has not yet previously been characterized. We here report that three putative SBE sites surprisingly proved non-functional through the analysis of mutation and deletion constructs. Rather, mutation of an AP1 site at -737 in the LAMB3 promoter reduced basal promoter activity and mutation of a second AP1 site at -3.5 kb abrogated TGFβ responsiveness. Mutation of a promoter proximal Sp1 site did not show significant effects on its own, but combined mutations of the Sp1 and both AP1 sites fully suppressed Smad4 effects on basal and TGFβ-induced promoter activities. Likewise, Smad4-dependent regulation of the LAMC2 promoter was mediated through two AP1 sites and an Sp1 site residing within a 350 bp downstream promoter region. Of note, whereas Sp1 sites are involved in positive regulation of both the LAMB3 and LAMC2 promoters, the LAMA3 promoter with a downstream Sp1 site mutated displayed strongly increased activity suggesting that in the LAMA3 promoter the Sp1 site exerts transcriptional repression (data not shown).
Chromatin immunoprecipitation showed TGFβ-induced binding of Smad4 to the functional SBE in the LAMA3 promoter and to the promoter regions harboring functional AP1 sites in each of the three promoters, confirming that all three genes encoding LM-332 are direct Smad4 target genes. Binding of Smad4 to AP1 sites may work via interaction of Smads with AP1 transcription factors as previously shown in the regulation of collagenase-I [
38], MMP1 [
40], interleukin-11 [
41] and demonstrated by Liberati et al. [
37] and Yamamura et al [
42].
In addition, expression analysis of AP1 family transcription factors revealed Smad4-dependent transcriptional induction in response to treatment with TGFβ indicating that direct and indirect mechanisms may converge on these three promoters to regulate LM-332 expression. Of note, whereas TGFβ induced binding of Smads to promoter elements in the c-jun and junB promoters has been reported earlier [
37,
43] the involvement of Smads in transcriptional regulation of other AP1 family members is a novel finding in this work.
Interestingly, the dual interdependence of Smad4 and AP1 factors is not without precedence. Our data suggest that Smad4, both, impinges on the regulation of AP1 expression as well as depends on AP1 factors for binding to AP1 sites. Likewise, it has been shown earlier that the androgen receptor drives the expression of ETS transcription factors and can then be co-dependent on ETS factors for its recruitment to a subset of promoters [
44].
We do not yet know if all of these Smad4 effects are mediated through the canonical TGFβ/Smad signaling pathway. TGFβ can also activate the MAPK pathway among others [
24,
25]; phosphorylation and activation of AP1 proteins through the MAPkinase pathway has been described long ago [
45]. However, all of the TGFβ effects observed in our cell models are dependent on Smad4. So, if other signal cascades in addition to the canonical pathway were involved in transcriptional regulation of the laminin genes in SW480 and BxPC3 cells these also function in a strictly Smad4-dependent manner.
We are only beginning to decipher Smad4 functions in cellular signaling networks. Here, we addressed molecular mechanisms underlying Smad4-dependent regulation of constitutive (cell-autonomous) and of TGFβ-induced transcription of laminin genes. Of note, although Smad4 is a positive regulator of all three LM-332 chains the underlying mechanisms are surprisingly complex and binding sites involved are divergent for the LAMA3 promoter on the one hand and for the LAMB3 and LAMC2 promoters on the other side. We hypothesize that this divergence in modular regulation of the three promoters may lay the ground for uncoupled regulation of LM-332 at the invasive front of tumors where an intracellular accumulation of the γ2-chain can often be observed and represents an impressive molecular marker [
46‐
49]. Invading/budding tumor cells are maximally exposed to cytokines expressed by stromal cells at the invasive front. For example, monocytes/macrophages present in many tumors at high numbers are a major source not only for TGFβ but also for TNFα [
50,
51]. Interestingly, Korang et al. reported that TNFα in epidermal keratinocytes inhibited LAMA3 but not LAMB3 and LAMC2 transcription [
26]. We also observe uncoupled responses of the three LM-332 genes to TNFα in Smad4-deficient but not in Smad4-positive tumor cells (Zboralski et al., in preparation) suggesting that loss of Smad4 may also contribute to uncoupled regulation of LM-332 and consequently to an intracellular accumulation of the γ2-chain. NF-kB binding sites mediating TNFα responses through the canonical pathway are present in all three promoters but have not yet been functionally analyzed. In addition, TNFα can also signal through Sp1 [
52] and through AP1 binding sites [
45] which we have shown here to be implicated in Smad4-dependent transcriptional regulation of the laminin genes.
Conclusion
In summary, whereas Smad4 is a positive regulator of basal and of TGFβ-induced promoter activities of the three LM-332 genes, the underlying mechanisms are surprisingly complex and significantly differ between the three promoters. Uncoupled regulation, namely induction of the LAMC2 but not the LAMA3 gene in response to signals derived from the tumor stroma at the invasive front of tumors is an important issue in tumor biology. We here show that multiple transcription factors and binding sites are involved in transcriptional regulation of the LM-332 genes. At least some of them, i.e. components of the AP1 family, are well known to be targeted through cytokines other than TGFβ, cytokines, which may also be present in the microenvironment of invasive tumor cells. Thus, it will be interesting in the future to address the impact of the Smad4 status of tumor cells on transcriptional responses in the context of various environmental stimuli.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
DZ participated in the design of the study, performed plasmid construction experiments and luciferase assays, carried out Northern and ChIP analyses and drafted the manuscript, MB established the novel SW480 and MZ the BxPC3 Smad4 cell system, assisted by SH and AS, SAH sequenced the promoter reporter constructs and WS contributed to the design of the study. IS-W is the PI, designed the study and drafted the manuscript. All authors have read and approved the final manuscript.