Background
In normal tissues, the epithelium is separated from the underlying mesenchyme by the basement membrane (BM), a specialized sheet of the extracellular matrix. The BM is built from constituents produced by both the epithelial and the mesenchymal cells [
1,
2]. Whereas collagen IV is the most prominent mesenchymal derived component providing the structural scaffold of the BM sheet the epithelial derived laminins build the centerpiece of the network that harbors additional proteins including perlecan, nidogen and fibulin [
3]. The basement membrane has been recognized as a structural but also as an important functional component of tissues. In particular, the laminins mediate cellular functions including adhesion, migration, growth and tissue-specific gene expression [
4,
5].
The laminins are large heterotrimeric glycoproteins with at least 15 different isoforms composed of different combinations of one α-, one β- and one γ-chain, each, out of five α, three β and three γ-chains. The laminins are expressed in a tightly regulated development- and differentiation-specific pattern [
6‐
8]. In the adult human intestine, laminins-211 and -511 show complementary distributions along the crypt-villus axis, whereas laminin-332 is restricted to the villus regions. In premalignant stages of colorectal carcinogenesis, namely in different types of adenomas, normal expression and deposition of laminin-332 and -511 has been reported. The transition to malignancy is defined by breaking the basement membrane barrier. In colorectal carcinomas, this is associated with a lack of laminin-511 and with irregular deposition of laminin-332 at invasive edges [
9‐
11]. Relative overexpression of the laminin-γ2 (and β3) chain has often been described and represents one of the most impressive molecular markers for the invasive front of colorectal and other cancer entities (for review see [
12]). It specifically marks socalled budding tumor cells [
13,
14]. Laminin-γ2 has been described as a target gene of the Wnt/β-catenin pathway [
15]. Whereas β-catenin is constitutively activated through mutation of the tumor suppressor APC in the majority of adenomas the relative overexpression of γ2 at the invasive edge of carcinomas requires additional alterations. Overexpression of γ2 is believed to result from cellular responses to environmental signals illustrating that the regulation of laminin expression is subject to tumor cell intrinsic factors including the pattern of their respective genetic alterations and to extrinsic microenvironmental factors including signals from inflammatory cells in the tumor tissue.
We have recently identified laminin-332 as a target structure of the tumor suppressor Smad4 [
16]. We have shown that Smad4 functions as a positive transcriptional regulator of all three chains encoding laminin-332. Reexpression of Smad4 led to the increased expression of heterotrimeric laminin-332 and to its deposition in basement membrane-like structures at contact sites with fibroblasts. Loss of Smad4 in the carcinogenic process, in turn, is implicated in reduced or absent expression of laminin-332 in poorly differentiated carcinomas.
Smads are primarily characterized as transmitters of signals from the TGFβ superfamily of cytokines but also function as promiscuitive transcriptional coregulators that can interact with a variety of ubiquitous and tissue-specific transcription factors and coregulators in a context-dependent manner [
17,
18]. TGFβ, in Smad4-reexpressing cancer cells like in premalignant adenoma cells induces the expression of all three genes encoding heterotrimeric laminin-332 whereas Smad4-negative cells are non-responsive [
16,
19]. The underlying molecular mechanisms are surprisingly complex and involve transcription factor binding sites like AP1 which are targeted by various signaling cascades. Moreover, the modular composition of the three promoters significantly differs from each other; a functional smad binding element (SBE) is present exclusively in the LAMA3 promoter [
19]. Thus, we wonder if the consequences of Smad4 loss in response to extracellular signals other than TGFβ may differ between the three genes encoding laminin-332. As an approach towards modelling the cytokine environment in tumor tissues we here address effects of TNFα, a prominent inflammatory cytokine produced by tumor infiltrating macrophages, on laminin-332 expression in Smad4-positive and Smad4-deficient tumor cells.
We report, that Smad4-reexpressing human colorectal cancer cells like adenoma cells respond to TNFα with a moderate increase of all three chains encoding laminin-332 and with synergistic induction in response to the combination of TGFβ and TNFα. In contrast, their Smad4-deficient counterparts display uncoupled responses to TNFα: whereas the β3 chain and in particular the γ2 chain is strongly induced in Smad4-negative cells, induction of the α3 chain is Smad4-dependent and is mediated via an NF-κB site and downstream AP1 sites in the LAMA3 promoter. Of note, TNFα induction leads to the release of significant amounts of the γ2 chain in a monomeric form and in complex with (an) unknown protein(s) as shown by Western blotting under non-reducing conditions and confirmed by mass spectrometry. Ultimately, induced secretion of soluble γ2 by transient suppression of the α3 chain leads to induction of cell migration.
Discussion
Invading tumor cells are maximally exposed to growth factors and cytokines expressed by stromal cell types. Among these, macrophages have previously been shown to induce angiogenesis [
23] and to enhance invasion through the secretion of TNFα [
24]. TNFα in cooperation with TGFβ dramatically enhanced EMT [
24]. On the other hand normal intestinal epithelial cells respond to TNFα and TGFβ with an increase in the expression of heterotrimeric laminin-332 [
25]. We therefore focused on the analysis of laminin-332 expression in response to cytokines TGFβ and TNFα in cell models adequate to reflect the molecular progression of colorectal cancer in vitro. To this end we used pairs of cell clones derived from the human Smad4-deficient colorectal cancer cell lines SW480 and SW620, in which Smad4 expression was stably restored. Responses of Smad4-reexpressing cancer cells were compared to responses of LT97 cells, a cell line derived from a late adenoma and carrying inactivated APC as well as an activated Ki-ras oncogene [
20]. LT97 cells secreted increased amounts of heterotrimeric laminin-332 in response to TGFβ and TNFα, respectively, and showed extensive synergistic induction of laminin-332 in response to the combination of both cytokines. Smad4-reexpressing colon cancer cells displayed similar, although less pronounced effects. These results are consistent with observations in vivo, that colorectal adenoma cells in the vicinity of infiltrating inflammatory cells display thickening of the basement membrane with streak-like deposits of laminin-332 [
10]. Likewise, intestinal epithelial cells in patients afflicted with Morbus Crohn show increased levels of TNFα and induced expression of the constituents of laminin-332 [
26,
27]. The response of (normal and) premalignant cells to increase expression of laminin-332 may thus be interpreted as a defense mechanism against an inflammatory attack by strengthening the basement membrane barrier.
Synergistic induction of laminin-332 in response to TGFβ and TNFα was Smad4-dependent as it did not occur in Smad4-deficient SW480 and SW620 cells. Smad4-deficient cells can induce the expression of the (β3 and) γ2 chain of laminin-332 in response to TNFα whereas TNFα induction of the α3 chain is Smad4-dependent. These results indicate that loss of Smad4 may represent a genetic alteration in the carcinogenic process that can lead to uncoupled regulation of the three genes encoding laminin-332 in response to inflammatory cytokines. We have shown previously that the molecular mechanisms of Smad4-dependent regulation of the three promoters encoding laminin-332 are surprisingly complex. Concerning basal and TGFβ-induced expression levels Smad4 is essential for positive regulation of all three genes. The molecular mechanisms underlying this regulation, however, are significantly divergent between the LAMA3 promoter as compared to the LAMB3 and LAMC2 promoters [
19]. Here we show that Smad4 is essential for TNFα induction of LAMA3 but not of LAMB3 and LAMC2 and that AP1 and NF-κB sites are involved in TNFα-mediated Smad4-dependent LAMA3 induction. Unraveling transcription factor complexes built in response to cytokines and active at the three promoters will require further detailed analyses.
We here focus on functional consequences of uncoupled regulation of the three laminin-332 chains in response to TNFα. The prevailing view suggests that the β3 and γ2 chains first form a heterodimer intracellularly, which then binds to α3 followed by rapid secretion of the heterotrimer [
28]. Uncoupled induction at the mRNA level in response to TNFα therefore let us expect an intracellular accumulation of the γ2 chain in Smad4-deficient cells. Despite repeated attempts, however, intracellular γ2 could not be detected by Western blotting of cell homogenates. Rather, Smad4-deficient cells release the γ2 chain in a monomeric form and in two complexes with as yet unidentified proteins as shown by Western blotting and unequivocally confirmed by mass spectrometry.
What are the functional implications of the release of monomeric γ2? We have shown here, that increased amounts of γ2 are associated with increased migration and assume that secreted γ2 may somehow promote tumor invasion. The release of γ2 may impinge on the composition of the extracellular matrix, alter its functional characteristics and so indirectly affect cell adhesion and migration. Interaction of γ2 with various ECM molecules including collagen, perlecan and fibulin has been reported [
29,
30]. Alternatively, monomeric γ2 may directly affect the tumor cells by interacting with cellular receptors followed by effects on cell signaling which subsequently may result in cell migration. The predominant laminin-receptors are the integrins. Whereas interaction of laminin-332 with cells is predominantly mediated via integrins α3β1 or α6β4 through binding to the laminin-like globular domains of the α3 laminin chain, the γ2 chain can bind to α2β1 integrin [
30]. Interestingly, it is known that domain III of the γ2 chain can also directly interact with the EGF-receptor [
31]. EGF signaling is a major stimulus for cell migration.
Clues to the functional relevance of γ2 secreted from cells may come from further analysis of γ2 complexes in conditioned media. Western blotting as well as results from mass spectrometry indicated that similar amounts of γ2 are present under non-reducing conditions at 140 kDa corresponding to the monomeric form and at 240 kDa corresponding to a γ2 protein complex; thus, we assume that the laminin-γ2 chain may interact with another protein of approximately 100 kDa in size. The proteomic analysis surprisingly provided evidence for another as yet unknown γ2 complex of about 400 kDa in size. As soon as alternative γ2 binding partners will be identified their functional relevance for tumor cell migration and invasion can be addressed.
In summary, our results provide evidence for a sequence of events, in which loss of the Smad4 leads to induction of the laminin γ2 chain in response to TNFα followed by the release of monomeric laminin γ2 which exerts a proinvasive effect. In conclusion, we here suggest a novel mechanism that may underlie the switch to invasive tumor growth upon loss of the tumor suppressor Smad4.
A large variety of growth factors and cytokines can be expressed at the invasive margin of carcinomas; some of them have previously been suspected to underlie relative overexpression of γ2. For example, Olsen at al. investigated the involvement of HGF and found synergistic induction of γ2 but not α3 by HGF and TGFβ [
32]. LAMC2 is an established β-catenin target gene and nuclear β-catenin has been reported to correlate with intracellular accumulation of γ2 at invasive margins and in budding tumor cells [
15]. Thus, upstream ligands of the wnt gene family induced upon cancer progression may also represent putative inducers of overexpressed γ2. Interestingly, the expression of wnts 2 and 5 has specifically been found in macrophages associated with colon tumors [
33]. Activated macrophages can indirectly promote Wnt signaling through TNFα [
34,
35]. We here present data showing that tumor cell responses to TNFα and to the combination of TNFα and TGFβ critically depend on Smad4. As extensive crosstalk mechanisms exist between Wnt/β-catenin and TGFβ/Smad pathways [
36‐
38] the detailed understanding of laminin regulation will require future investigations based on an integrated view of signaling networks in normal and oncogenically programmed cells and their respective responses to a dynamic cytokine milieu.
Methods
Cell culture and conditioned media
The human colorectal carcinoma cell lines SW480 and SW620 were obtained from the American Type Culture Collection, the human colon adenoma cell line LT97-2 was kindly provided by M Marian (Vienna, Austria). LT97 cells were maintained in Ham's F12 medium with supplements as described [
20]. All other cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal calf serum, 2 mM glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin. When indicated, cells were incubated with 5 ng/mL of TGFβ
1 (R&D systems, Minneapolis, MN, USA) and 30 ng/mL of TNFα (Pan Biotech, Aidenbach, Germany) in serum reduced medium (0.5% FCS). Preparation of proteins from serum-free conditioned media was performed as described previously [
39].
Production of polyclonal antibody 2140
GST-tagged recombinant laminin-γ2 was produced by expressing pGEX γ2lam5 (kindly provided by Dr. M. Failla, IDI-IRCCS, Roma, Italy) in E.coli. Affinity-purified laminin γ2-GST protein was confirmed by mass spectrometry (LSMDO, CNRS-EPCM, UMR7509, Strasbourg, France) and injected into rabbits. Antibodies were verified by immunoblotting on HT29-MTX cells and by immunofluorescence on human intestines (not shown) giving identical but stronger signals than MAB19562 (Chemicon, Hampshire, UK).
Western blotting and RNA analyses
For laminin Western blots, samples were subjected to SDS-PAGE on either 8% polyacrylamide gels or on NuPAGE Novex 3-8% Tris-Acetate gels (Invitrogen, Karlsruhe, Germany) and run under conditions either with (reducing conditions to analyze single chain expression) or without dithiothreitol (non-reducing reducing conditions). Heterotrimeric laminin-332, dimeric and monomeric laminin-γ2 under non-reducing conditions were detected with monoclonal antibody MAB19562 (Chemicon) and polyclonal antibody 2140. The blots were incubated with a secondary antibody directly coupled with a fluorescent dye (Alexa Fluor 680; Alexa Fluor 800; Invitrogen and Rockland). Signals were detected using the Odyssey Infrared Imaging System (LI-COR Biosciences, Bad Homburg, Germany) which allows for a digital quantification of signals over a wide linear range of signal intensities. RNA analyses were performed according to standard procedures by Northern blot hybridization and by qRT-PCR as described previously [
16,
19,
21].
Mass spectrometry and mass spectrometric data analysis
Mass spectrometry and mass spectrometric analysis are described in detail in the additional information. In brief, tryptic digest were analyzed by nano-HPLC/ESI-MS/MS using the UltiMate™ 3000 HPLC system (Dionex LC Packings, Idstein, Germany) online coupled to an LTQ Orbitrap XL instrument (Thermo Fisher Scientific, Bremen, Germany). Reversed-phase (RP) capillary HPLC separations were performed as described previously [
40].
Peak lists of MS/MS spectra were imported into ProteinScape (version 1.3, Bruker Daltonics, Bremen, Germany) and subsequently correlated with the human International Protein Index database (human IPI v3.41,
http://www.ebi.ac.uk) containing 72155 protein entries using MASCOT (release version 2.2) [
41]. To enable the estimation of a false discovery rate (FDR), the database was concatenated with a duplicate of itself in which the amino acid sequence of each protein entry was randomly shuffled [
42]. Protein hits up to an accumulated FDR of 5% were considered as true positive identifications.
Migration assay and transient LAMA3 knockdown
Cells in 500 μL media were added to the upper compartment of 12 well plates supplemented with inserts (8 μm pore size; BD Falcon). Cytokines were added one day later and cells incubated for another 72 hours at 37°C. Cells which had passed the pore membrane were quantified using Cell Titer Glo (Promega, Madison, WI, USA) in accordance with the manufacturer's recommendations.
For siRNA experiments cells were grown to a confluency of 50% and transfected with ON-TARGETplus siRNA (LAMA3, J-011071-05, Dharmacon, Lafayette, CO, USA) or Dharmacon ON-TARGETplus Nontargeting siRNA as a control, respectively, using Dharmafect (Dharmacon). For migration assays cells were plated into transwells 24 h after transfection.
Promoter construction and transient transfections were performed as previously described [
16,
19] with minor modifications. Cells were grown to a confluency of approximately 50-70% in 96-well plates, medium was changed to serum reduced medium (0.5% FCS) and cells were transfected using Effectene (Qiagen, Hilden, Germany). The next day cytokines were added and cells were harvested 4 h and 24 h after transfection. Luciferase assays were carried out as triplicates and quantified using a luminometer (GloMax™ 96 Microplate, Promgea, Madison, WI, USA) and the Dual-Luciferase-Reporter Assay System (Promega).
Acknowledgements
We thank B Marian (Vienna) for providing LT97 cells, M Failla (IDI-IRCCS, Roma) for the pGEX γ2lam5 expression vector, M Vanier (U682, Strasbourg) for the production of the laminin γ2 antibodies (#2140), JM Strub (LSMDO, CNRS, Strasbourg) for mass spectrometry analysis of the laminin γ2-GST protein, and S Hoppe and A Schöneck for excellent technical assistance. This work was supported by grants from the Ruhr-University Bochum, FoRUM Program (F510-2006) and from INCa (Canceropôle Grand-Est to PSA).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
DZ designed the study, carried out Western Blot, RNA and promoter analyses, performed transient knockdown and migration experiments and drafted the manuscript. BW performed mass spectrometry analyses. MBM and HB participated in expression analyses. SK-S, HEM and WS contributed to the design of the study. PS-A established the novel polyclonal laminin-γ2 antibody 2140 and contributed to the manuscript. IS-W is the PI, designed the study and drafted the manuscript. All authors have read and approved the final manuscript.