Divergent mechanisms underlie Smad4-mediated positive regulation of the three genes encoding the basement membrane component laminin-332 (laminin-5)

The human colorectal carcinoma cell line SW480 and the human pancreatic carcinoma cell line BxPC3 cells 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 [32]. 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.

The full-length coding sequence of Smad4 was cloned into the pBK-CMV expression vector (Stratagene) as previously described [17] and Smad4 re-expressing SW480 cell clones and negative control transfectants were established by retroviral transduction.

Expression of the Smad4 protein product was analyzed by Western blotting of lysates. Cells were lysed in NP-40 lysis buffer (25 mM Tris HCl, pH 7.4, 0.5% NP-40, 100 mM NaCl, 1 mM EDTA) containing a protease inhibitor cocktail (Roche) and 1 mM PMSF. Proteins were resolved by SDS PAGE and transferred to Immobilon membranes (Millipore). The blots were incubated with monoclonal antibodies against Smad4 (anti-Smad4 B8; dilution 1:500, Santa Cruz), washed with PBS containing 0.05% Tween 20 and incubated with a secondary antibody which was coupled with the fluorescent dye Alexa Flour 680. Signals were detected using the Odyssey Infrared Imaging System (LI-COR Biosciences).

Promoter fragments upstream of exon1 of the human LAMA3A; LAMB3 and LAMC2 genes were amplified by genomic PCR with primers listed in additional file 1 and cloned into the pGL3-basic vector (Promega). All constructs were sequence-verified. The 0.8 kb deletion construct of the LAMC2-promoter was generated by double-digestion of the LAMA3-2 kb construct with NheI (within theMCS of pGL3 basic) and EcoRI. Point mutations were introduced using the Quick-Change Site directed mutagenesis kit (Stratagene) with the primer oligonucleotides listed in additional file 1. DNA used for transient transfections was prepared with a plasmid Midi Kit (Qiagen, endotoxin-free).

For transient transfections, cells were grown to a confluency of approximately 50–70% in 24-well plates and transfected with 200/400 ng of the respective promoter construct (pGL3-basic, Promega) and 2/4 ng internal control plasmid (phRL-SV40, Promega) using Effectene (Qiagen) in accordance with the manufacturer's recommendations. TGFβ was added when indicated 5 h after transfection at a final concentration of 5 ng/mL. The cells were harvested after 24 h and the luciferase assays were carried out as triplicates using a luminometer (GloMax™ 96 Microplate, Promgea) and the Dual-Luciferase-Reporter Assay System (Promega).

ChIP assays were performed as described in Zapatka et al. [19] with minor modifications. BxPC3 cells were grown to confluence in 150 mm culture dishes and treated with recombinant TGFβ1 (R&D Systems) at a concentration of 5 ng/mL for 90 min. Proteins and DNA were crosslinked by incubating the cells in 1% (v/v) formaldehyde at room temperature for 10 min and the reaction was stopped by the addition of glycine to a final concentration of 0.125 M. Cells were washed twice with Tris-buffered saline (20 mM Tris, pH 7.4, 150 mM NaCl) and lysed in SDS buffer (50 mM Tris, pH 8.1, 0.5% (v/v) SDS, 100 mM NaCl, 5 mM ethylenediaminetetraacetic acid (EDTA), protease inhibitors, Roche). Chromatin was pelleted by centrifugation and resuspended in 0.5 mL of immunoprecipitation buffer (two parts SDS buffer, one part Triton dilution buffer (100 mM Tris, pH 8.6, 100 mM NaCl, 5% (v/v) Triton X-100, 5 mM EDTA) and protease inhibitors (Roche). Chromatin was sonicated (four times for 20 s, using a Bandelin sonoplus sonicator (Bandelin electronic GmbH&Co KG, Germany) to yield genomic DNA fragments with a bulk size of 300–500 bp. Immunoprecipitaton was performed overnight at 4°C with polyclonal antibodies specific for Smad4 (3 μg H-552 and C-20, Santa Cruz). Immune complexes were recovered by adding 50 μL magneto-beads (Dynal, Invitrogen) and incubated for 2 h at 4°C. Then beads were washed successively as described previously [19]. Washed precipitates were incubated overnight at 65°C in elution buffer (TE, 1% SDS, 0.1 M NaHCO3) to reverse the formaldehyde crosslinking. DNA fragments were purified with a QiaQuick Spin Kit (Qiagen) according to the manufacturer's recommendations except that the samples were first mixed with agitation for 30 min with PB buffer. The promoter regions were amplified from 2 μL of the extracted DNA per reaction and using primers listed in additional file 1 in 32 cycles of amplification to yield fragments of ~200 bp length.

RNA was isolated with a commercial kit (Qiagen). Northern blots and hybridizations were performed as described previously [17]. AP1 probes were prepared by RT-PCR from cell line RNA, using primers listed in additional file 1. For loading control blots were stripped and reprobed for GAPDH. Signal intensities were quantified by PhosphorImage analysis (Packard).

We are using gene transfer techniques to address functions of the tumor suppressor gene Smad4 in human cancer cells. Previously, we have established Smad4-positive derivatives from the colon carcinoma cell line SW480 by stable transfection. This cell model proved that reexpression of Smad4 at subphysiological levels was adequate to suppress tumor growth whereas TGFβ resistance of the cells was retained. Rather, Smad4 induced mesenchymal to epithelial reversion through induction of E-cadherin [15, 17]. We also found that Smad4-positive SW480 cells in contrast to Smad4-negative clones deposited an adhesive matrix in vitro on tissue culture plastic. Based on this observation we identified the heterotrimeric BM component LM-332 as a novel target structure of Smad4. In pancreatic carcinoma cells BxPC3 and CFPAC-1, Smad4 reexpression also increased constitutive LM-332 expression levels and additionally restored TGFβ induction of LM-332 [19].

In SW480 cells TGFβ responses through the canonical pathway are restricted by very low expression levels of the TGFβ type II receptor; Smad4-positive SW480 cells displayed TGFβ responsiveness of the p3Tplux promoter in cotransfections with a constitutively active TGFβ type I receptor construct, only [17]. As these stably transfected SW480 cell clones express "subphysiological" Smad4 levels (roughly one third of "normal" endogenous levels in Smad4-positive cell lines), we asked if higher Smad4 levels as previously obtained through retroviral gene transfer in Smad4-deficient pancreatic carcinoma cells (i.e. BxPC3) were adequate to overcome the limiting receptor levels in SW480 cells. In fact, a novel set of Smad4-positive retrovirally transduced SW480 derivatives showing moderate overexpression of Smad4 (roughly threefold of "normal" endogenous levels in Smad4-positive Paca44 cells, Fig. 1A) displayed strong TGFβ responsiveness in transient transfection assays with p3Tplux (5–6 fold induction) and p6SBE (40 fold induction) promoter reporter constructs (Fig. 1B).

Smad4-positive BxPC3 cell clones display significantly higher activities of both reporter constructs in the absence and in the presence of exogenous cytokine when compared to Smad4-positive SW480 cells. This may be due to different levels of autocrine TGFβ (family) cytokines expressed by both cell lines. Moreover, we have shown previously in SW480 cells, unlike BxPC3, that very low expression levels of the TGFβ type II receptor restrict TGFβ responsiveness [15].

We then analyzed TGFβ responses of the endogenous genes encoding LM-332 through Northern blot analysis. The results confirmed that moderate overexpression of Smad4 was adequate to restore TGFβ responsiveness of LM-332 in SW480 and BxPC3 cells (Fig. 1C).

To unravel molecular mechanisms and pathways involved in Smad4-mediated positive regulation of the LAMA3, LAMB3 and LAMC2 genes, we here set up detailed promoter analyses. An in silico sequence analysis using MatInspector software [33, 34] confirmed that the previously analyzed SBE element at -1.5 kb is the single SBE site in a 5 kb fragment upstream of the transcription start site of the LAMA3 gene [19] and indicated that both, the LAMB3 and the LAMC2 promoters harbor three putative SBE sites approximately at -1.41; 2.66 and 3.67 and at -1.59; 2.67 and 3.85 kb, respectively. Thus, we amplified 4 kb promoter regions from the three LM-332 promoters and cloned them into the pGL3 basic luciferase promoter vector. Next, the activities of the promoter-reporter constructs were tested in Smad4-negative and Smad4-positive clones of the SW480 and BxPC3 cell lines as well as in the Smad4-positive colorectal adenoma cell line LT97 in the absence and in the presence of TGFβ. All three constructs displayed increased constitutive activities in the Smad4-positive derivatives and all of them mediated Smad4-dependent TGFβ induction (Fig. 2) thus reflecting responses of the endogenous genes (Fig. 1C, [19]).

The 4 kb (Fig. 2) and the 2 kb LAMA3 promoter constructs displayed very similar responses in all three cell lines analyzed (data not shown). Thus, we used the 2 kb construct for further analyses, which contains the single SBE site (Fig. 3, construct a). Mutational inactivation of the SBE site at -1.5 kb (construct b) did not affect the Smad4-dependent increase of constitutive promoter activity. Moreover, the mutant construct still mediated approximately half of the TGFβ response as compared to the wildtype construct. Thus, we asked for additional promoter sequences involved in Smad4-dependent transcriptional regulation of LAMA3 expression. Three AP1 binding sites close to the transcription start site at positions -90, -146 and -272 in the human promoter confer epithelial specific expression [35] and have also been implicated in the TGFβ response of the mouse promoter [31]. Moreover, AP1 transcription factors are known to interact with Smad proteins [36‐38]. A construct with all three AP1 sites mutated displayed a reduction of basal promoter activity as expected (construct c). The reduction was moderate in SW480 cells and very pronounced in LT97 and BxPC3 cells which both display epithelial morphology. Levels of Smad4-dependent constitutive and TGFβ-induced activity were reduced correspondingly. Interestingly, when both, mutation of the SBE site and mutation of the AP1 sites were combined, the TGFβ response was completely abolished (construct d), suggesting that Smad4 cooperates with AP1 transcription factors in TGFβ-induced LAMA3 induction.

To the best of our knowledge functional analyses of the LAMB3 promoter except for methylation studies have not yet been published. As the in silico sequence analysis indicated a putative SBE site at position -1413 we first cloned a 2 kb promoter fragment into the reporter vector (Fig 4A, construct a). This construct reflected the Smad4-dependent increase in basal expression levels but did not display TGFβ induction neither in SW480 cells (Fig. 4) nor in BxPC3 cells (data not shown). Moreover, mutation of the putative SBE site at -1.41 kb (construct b) did not alter reporter responses neither in SW480 cells (Fig. 4) nor in BxPC3 cells (data not shown) suggesting that this site is not functional. Next, as two additional putative SBE sites were indicated at positions -2.66 and -3.67 kb, we cloned a reporter construct harboring approximately 4 kb of the LAMB3 promoter (construct c). This construct did show induction in response to TGFβ in all TGFβ responsive cell lines analyzed and thus reflected endogenous gene responses (Fig. 1C, Fig. 2). Surprisingly, however, both of these SBE sites also proved non-functional; as for the SBE site at -1.41 kb mutation of the SBE sites at positions -2.66 and -3.67 kb did not alter reporter responses in transient transfections of SW480 cells (constructs d, e) and other TGFβ responsive cell lines (Smad4-reconstituted BxPC3, LT97, Hacat; data not shown).

The 4 kb promoter construct (Fig 4B, construct a') of the LAMC2 promoter also reflected endogenous gene responses to Smad4, namely Smad4-dependent increase of constitutive expression and a less pronounced Smad4-dependent TGFβ induction of LAMC2 expression (Fig. 1C, 2). The 4 kb region of the LAMC2 promoter harbors three putative SBE sites, located at -1.59; -2.67 and -3.85 kb. Functional regulatory sites were delimited by deletion and mutation constructs (constructs b'-d'). Surprisingly, we found, that a 0.8 kb promoter fragment still showed similar activities in SW480 cells as compared to the 4 kb fragment suggesting that all relevant regulatory sites reside within this part of the promoter (construct e').

As all three putative SBE in the LAMB3 promoter did prove non-functional, we then asked if AP1 sites may mediate Smad4 effects on the LAMB3 promoter. AP1 sites reside at positions -737 and -3520; both were mutated separately and in combination. Mutation of the promoter-proximal AP1 site reduced basal promoter activity (in Smad4-negative cells) as well as Smad4-dependent constitutive and TGFβ-induced promoter activity (Fig. 5A, construct b vs. a). Mutation of the 3.5 kb AP1 site in contrast inactivated TGFβ responsiveness but did not alter constitutive expression levels (construct c). Still, when both mutations were combined, the Smad4-dependent difference in basal activity of the LAMB3 promoter was retained (construct d). Ultimately, we tested the Sp1 binding site indicated at position -99 by in silico sequence analysis. Mutation of this site on its own had no significant effect (construct e). When combined with the mutated AP1 sites, however, the Smad4 effect on the LAMB3 promoter activity was nearly completely abolished (construct f vs. d).

On the LAMC2 promoter two neighboring AP1 sites at positions -50 and -91 have been reported to mediate synergistic effects of HGF and TGFβ. Of note, these results were obtained in HT29 colon carcinoma cells, which are Smad4-negative [29]. Here we show, that in particular the AP1-site at position -91 is involved in the Smad4 effect on constitutive and on TGFβ-induced expression levels (Fig. 5B, construct a'-c'). As Smad4-dependent differences in the construct with combined AP1 mutations are retained (construct d'), we additionally analyzed the involvement of the Sp1 site located at position -345 in the LAMC2 promoter. Again, mutation of the Sp1 site on its own had no significant effect (construct e'). When combined with the mutated AP1 sites, however, all promoter activities were strongly suppressed (construct f').

In summary, Smad4effects on constitutive and TGFβ-induced LAMB3 and LAMC2 promoter activity in SW480 cells are mediated through AP1 sites and an Sp1 site whereas SBE sites are non-functional in these promoters.

The comparison of promoter activities in Smad4-negative and Smad4-reconstituted cells does not allow to differentiate between direct effects (Smad4 binds to the respective promoter region) and indirect effects (Smad4 may alter expression patterns of AP1, Sp1 and other transcription factors). To get more insight into the underlying mechanisms, we next we performed transient cotransfections of a Smad4 expression construct with all three wild-type promoter-reporter constructs into Smad4-deficient SW480 cells (data not shown). This led to increased promoter activities to a similar extent as determined in the stable Smad4-positive derivatives suggesting that all three LM-332 genes are direct Smad4 target genes. Thus, we continued to assess direct binding of Smad4 to the respective sites by chromatin IP.

We have shown in our previous work that LAMA3 is a direct target gene of Smad4. Chromatin IP has confirmed Smad4 binding to the SBE region in the endogenous LAMA3 promoter [19]. In this work we demonstrated that the promoter-proximal AP1 sites are additionally involved in conferring TGFβ responsiveness to the LAMA3 promoter in a Smad4-dependent manner (see Fig. 2, 3 construct c). Consistent with this result, Smad4 binding to this region could also be demonstrated by chromatin IP (Fig. 6). PCR amplification of an upstream region performed as a negative control failed.

Putative SBE sites in the LAMB3 and LAMC2 promoters, investigated with transient transfections, proved non-functional. Correspondingly, all attempts to show Smad4 binding to the respective promoter regions failed. In contrast, direct binding of Smad4 – presumably in complex with AP1 family transcription factors [36, 38] – to the promoter regions with functional AP1 sites could be demonstrated (Fig. 6). In conclusion, results from chromatin IP experiments are consistent with promoter analyses through transient transfections of reporter constructs throughout. The involvement of functional binding sites in basal promoter activity, Smad4-dependent constitutive gene expression and Smad4-mediated TGFβ responses is summarized in Figure 7.

Genes encoding AP1 family members have previously been characterized as TGFβ-responsive [39]. Thus, we analyzed transcriptional responses to TGFβ in Smad4-deficient and Smad4-reexpressing SW480 cells as well as in Smad4-positive LT97 colorectal adenoma cells (Fig. 8). Expression levels of c-jun, junB, junD, c-fos, fosB and fra-1 remained unaffected by the addition of recombinant TGFβ in Smad4-deficient SW480 cells (fra-2 not expressed). Smad4-reexpressing SW480 cells, however, displayed increased transcript levels of c-jun, junB, junD, fosB and fralin response to TGFβ as did LT97 cells (with the exception of fosB and fra1). Interestingly, expression of c-fos was downregulated in Smad4-positive SW480 cells but strongly upregulated in LT97 cells. Whereas TGFβ-induced binding of Smads to promoter elements in the c-jun and junB promoters has been reported earlier [37, 40] the involvement of Smad4 in transcription regulation of other AP1 family members is a novel finding. Our findings suggest, that "indirect" effects of Smad4, namely altered expression patterns of AP1 family members, may impinge on Smad4 target genes, among them the genes encoding LM-322.