The Role of SnoN in Transforming Growth Factor β1-induced Expression of Metalloprotease-Disintegrin ADAM12*

Increased expression of metalloprotease-disintegrin ADAM12 is a hallmark of several pathological conditions, including cancer, cardiovascular disease, and certain inflammatory diseases of the central nervous system or the muscoskeletal system. We show that transforming growth factor β1 (TGFβ1) is a potent inducer of ADAM12 mRNA and protein in mouse fibroblasts and in mouse and human mammary epithelial cells. Induction of ADAM12 is detected within 2 h of treatment with TGFβ1, is Smad2/Smad3-dependent, and is a result of derepression of the Adam12 gene. SnoN, a negative regulator of the TGFβ signaling pathway, is a master regulator of ADAM12 expression in response to TGFβ1 stimulation. Overexpression of SnoN in NIH3T3 cells reduces the magnitude of ADAM12 induction by TGFβ1 treatment. Down-regulation of SnoN expression by short hairpin RNA enhances TGFβ1-induced expression of ADAM12. In a panel of TGFβ1-responsive cancer cell lines with high expression of SnoN, induction of ADAM12 by TGFβ1 is significantly impaired, suggesting that the endogenous SnoN plays a role in regulating ADAM12 expression in response to TGFβ1. Identification of SnoN as a repressor of the ADAM12 gene should contribute to advances in the studies on the role of ADAM12 in tumor progression and in the development of other pathologies.

Increased expression of metalloprotease-disintegrin ADAM12 is a hallmark of several pathological conditions, including cancer, cardiovascular disease, and certain inflammatory diseases of the central nervous system or the muscoskeletal system. We show that transforming growth factor ␤1 (TGF␤1) is a potent inducer of ADAM12 mRNA and protein in mouse fibroblasts and in mouse and human mammary epithelial cells. Induction of ADAM12 is detected within 2 h of treatment with TGF␤1, is Smad2/Smad3-dependent, and is a result of derepression of the Adam12 gene. SnoN, a negative regulator of the TGF␤ signaling pathway, is a master regulator of ADAM12 expression in response to TGF␤1 stimulation. Overexpression of SnoN in NIH3T3 cells reduces the magnitude of ADAM12 induction by TGF␤1 treatment. Down-regulation of SnoN expression by short hairpin RNA enhances TGF␤1-induced expression of ADAM12. In a panel of TGF␤1-responsive cancer cell lines with high expression of SnoN, induction of ADAM12 by TGF␤1 is significantly impaired, suggesting that the endogenous SnoN plays a role in regulating ADAM12 expression in response to TGF␤1. Identification of SnoN as a repressor of the ADAM12 gene should contribute to advances in the studies on the role of ADAM12 in tumor progression and in the development of other pathologies.
ADAM12, a member of the metalloprotease-disintegrin family of proteins, has been implicated in the progression of cancer, cardiovascular disease, osteoarthritis, and neurological disorders (1). The ADAM12 gene is frequently mutated in human breast cancers (2,3), and cancer-associated mutations cause mislocalization of the ADAM12 protein in cells and alter its function (4). Missense single nuclear polymorphism in the ADAM12 gene shows strong association with osteoarthritis (5,6). In addition to changes in its amino acid sequence, expression levels of ADAM12 are significantly increased in many pathological states. For example, ADAM12 expression levels are 20 -30-fold higher in human breast tumors than in normal mammary epithelium (7)(8)(9)(10)(11)(12). ADAM12 expression is also markedly up-regulated in cancers of the liver, lung, stomach, colon, prostate, bladder, and in glioblastoma (13)(14)(15)(16)(17)(18). Increased ADAM12 expression levels are found in the cardiac tissue of patients with hypertrophic obstructive cardiomyopathy (19) and in mice with angiotensin II-induced hypertension and cardiac hypertrophy (20,21). During inflammatory responses and aseptic osteolysis associated with loosened hip replacement implants, ADAM12 is up-regulated in the interface tissue around loosening implants (22). In experimental autoimmune encephalomyelitis, an animal model of multiple sclerosis, ADAM12 level is markedly increased in the T cells that infiltrate spinal cords (23).
The mechanisms regulating ADAM12 expression, in particular those that may be responsible for altered levels of ADAM12 in various pathological states, are poorly understood. Previous studies employing hepatic stellate cells, a mesenchymal cell type in hepatic parenchyma, have indicated that ADAM12 expression is induced by transforming growth factor ␤ (TGF␤) 2 (13,24). The TGF␤ signaling pathway is initiated when one of the family members, e.g. TGF␤1, -␤2, or -␤3, binds to a complex of TGF␤ type I and type II serine/threonine kinase receptors (T␤RI and T␤RII, respectively) and induces phosphorylation and activation of T␤RI by T␤RII. T␤RI then phosphorylates receptor Smads (R-Smads), Smad2 and Smad3. Phosphorylated Smad2/3 associate with the common partner Smad4 and translocate to the nucleus, where they regulate transcription of target genes (25,26). In addition, receptor activation in certain cell types leads to Smad-independent responses via the activation of mitogen-activated protein kinases (MAPKs), phosphoinositide 3-kinase, and Rho family members (27,28).
SnoN and the related Ski protein are negative regulators of TGF␤ signaling. They bind to nuclear Smad complexes and repress their transcriptional activities (29 -31). In response to TGF␤ stimulation, SnoN (and to a lesser extent Ski) undergoes ubiquitination and rapid proteasomal degradation (32,33). The ubiquitin ligases implicated in ubiquitination of SnoN, the anaphase promoting complex, Smurf2, and Arkadia, are recruited to SnoN via the phosphorylated R-Smads (34 -38).
Previous study on the regulation of ADAM12 expression by TGF␤ in hepatic stellate cells used rather long (24 -72 h) stimulation times and showed that ADAM12 induction was par-* This work was supported, in whole or in part, by National Institutes of Health tially blocked by inhibitors of MAPKs, phosphoinositide 3-kinase, or p70S6 kinase (13,24). Based on these results, it was postulated that induction of ADAM12 expression by TGF␤ might be Smad-independent, but a direct role of R-Smads in the regulation of ADAM12 expression has not been tested. In this report, we investigate short term (0 -24 h) effects of TGF␤ on ADAM12 mRNA and protein levels in mouse fibroblasts. We find that TGF␤ causes derepression of the Adam12 gene in a Smad2/3-dependent manner, and that the repressor responsible for the negative regulation of ADAM12 expression is SnoN. Our studies uncover a new mechanism of ADAM12 regulation by TGF␤ that may contribute to aberrant expression of ADAM12 in various diseases.
Viral Transduction-Human SnoN cDNA was transferred from pCI-Neo HA-hSnoN plasmid vector into the retroviral pBMN-I-GFP vector (both from Addgene). Phoenix Eco cells were transfected with SnoN retroviral vector (15 g of plasmid DNA/100-mm plate) using the calcium phosphate precipitation method. Viral supernatants were harvested 48 h later, supplemented with 5 g/ml of Polybrene, and used without further dilution for infection of NIH3T3 cells. For SnoN knockdown, NIH3T3 cells were incubated with MISSION TM Lentiviral shSnoN Transduction Particles (Sigma, clone ID TRCN0000088306) or with MISSION TM Non-Target shRNA Control Transduction Particles (Sigma, SHC002V), according to the manufacturer's instructions. After 1 day, medium containing retroviral particles was replaced with fresh medium, and after an additional 24 h, stably transduced cells were selected with 2 g/ml of puromycin for 7 days.
Gene Reporter Assays-Fragments of the genomic region located upstream of the ADAM12 translation initiation site (see Fig. 4A) were amplified from mouse genomic DNA using Pfu Turbo DNA polymerase and inserted into the multiple cloning site of pGL4.10luc2 vector (Promega) to generate pA12.Luc reporters. NIH3T3 cells grown in 6-well plates were transfected at 50% confluence with 0.5 g of pA12.Luc reporters or empty pGL4.10luc2 vector, together with 0.05 g of Renilla luciferase vector (pRL-TK), using FuGENE 6 transfection reagent. After 24 h, firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega).
To study the effect of TGF␤1 on the Adam12 promoter, 24 h after transfection cells were stimulated with 2 ng/ml of TGF␤1 for an additional 24 h. The p3TP-Lux plasmid (Addgene) was used as positive control. The effect of TGF␤1 was also studied in NIH3T3 cells stably transfected with pA12.Luc1 vector. Stable transfectants were selected for 7 days in the presence of 2 g/ml of puromycin; pooled populations of cells were used without isolation of individual clones. All assays were performed in duplicates.
Statistical Analysis-Paired t test was used to compare values of two groups. When fold-change in ADAM12 expression was calculated (stimulus-or inhibitor-treated cells versus vehicle control), data were analyzed by one sample t test (GraphPad Prism Software, San Diego, CA). Linear regression analysis was performed using the Linear Fit tool, with direct error weighting (Origin 8.0 software, OriginLab, Northampton, MA).

RESULTS
Previous reports have shown that treatment of human hepatic stellate cells with TGF␤1 induces expression of ADAM12 (13,24). Here, we extended the analysis of TGF␤1 effects on ADAM12 expression to other cell types. We observed that treatment of mouse NIH3T3 fibroblasts or normal mouse mammary epithelial cell line NMuMG for 24 h with TGF␤1 led to dramatic increase in ADAM12 protein levels ( Fig. 1, A and  B). The levels of two other ADAMs, ADAM9 and ADAM15, were not changed after similar TGF␤1 treatment (Fig. 1C), demonstrating that among the three ADAMs tested, the effect of TGF␤1 was specific for ADAM12.
Induction of ADAM12 protein in cells that were starved in 0.5% serum for 24 h prior to adding TGF␤1 was comparable with the induction observed in the presence of 10% serum ( Fig.  2A). Therefore, all subsequent experiments were performed using medium supplemented with 10% FBS, without starvation. It has to be stressed that when cells were incubated for prolonged times (ϳ48 h) without adding fresh medium, the basal level of ADAM12 expression was significantly elevated, and it was efficiently reduced by adding SB-431542, an inhibitor of T␤RI ( Fig. 2A). This result suggests that the increase in the basal level of ADAM12 expression was most likely due to the autocrine/paracrine effect of the endogenous TGF␤1 produced in NIH3T3 cells that was accumulating over time in cell medium. The induction of ADAM12 protein in NIH3T3 cells by exogenously added TGF␤1 was dose-dependent and reached a maximum at 2 ng/ml of TGF␤1 (Fig. 2, B and C). This concentration of TGF␤1 was used in the remaining part of this study. The up-regulation of ADAM12 protein was evident after 8 h of stimulation of cells with TGF␤1 (Fig. 2, D and F). Semiquantitative RT-PCR analysis further demonstrated that TGF␤1 treatment increased the level of ADAM12 mRNA, and the changes in ADAM12 mRNA preceded the changes in ADAM12 protein levels (Fig. 2, E and F).
Pre-treatment of cells with actinomycin D, an inhibitor of transcription, completely blocked the up-regulation of ADAM12 protein (Fig. 3A), indicating that induction of ADAM12 expression by TGF␤1 required new transcription. Consistently, qRT-PCR (Fig. 3B) or Northern blot analysis of NIH3T3 cells treated with TGF␤1 ( Fig. 3C) showed that the induction of ADAM12 mRNA by TGF␤1 was also blocked by pre-treatment of cells with actinomycin D. As TGF␤1 did not have any effect on the stability of ADAM12 mRNA (Fig.  3D), we have concluded that TGF␤1 activates the transcription of the Adam12 gene.
Smad2 and Smad3 are the main mediators of the transcriptional responses to TGF␤1. Smad3-Smad4 DNA binding elements contain a repeated AGAC sequence or its complement, GTCT. Activated Smad2-Smad4 complexes are recruited to the activin response elements with the help of FoxH1a or FoxH1b (40,41). The ϳ5.1-kilobase genomic region upstream of the translation initiation site of mouse Adam12 contains a promoter activity and comprises multiple AGAC (or GTCT) sequences, including a 9ϫGTCT repeat, and an inverted activin response element, AATAAACA (Fig. 4A). Despite the presence of these motifs, gene reporters containing different fragments of the Adam12 promoter were not responsive to TGF␤1 (Fig.  4B). pA12.Luc1 reporter stably transfected into NIH3T3 cells was also unresponsive to TGF␤1 (result not shown), excluding the possibility that activation of the Adam12 promoter by TGF␤1 occurs only when the promoter is integrated into chromatin. Nevertheless, the induction of ADAM12 by TGF␤1 in Smad2 Ϫ/Ϫ or Smad3 Ϫ/Ϫ MEFs was significantly impaired when compared with the wild-type MEFs (Fig. 4, C and D).
These results indicate that Smad2/3 do play a role in regulation of ADAM12, but TGF␤-responsive elements are located outside the ϳ5.1-kilobase promoter region in the Adam12 gene.
Two lines of evidence further suggest that induction of ADAM12 expression by TGF␤1 involves derepression of the Adam12 gene. First, pretreatment of cells with cycloheximide, an inhibitor of translation, did not abolish the induction of ADAM12 mRNA by TGF␤1 (Fig. 5A). In fact, cycloheximide alone increased the level of ADAM12 mRNA (Fig. 5B). These results suggest that de novo protein synthesis is not required for the up-regulation of ADAM12 by TGF␤1 and that cycloheximide might block synthesis of a transcriptional repressor acting on the Adam12 promoter/gene. Second, pre-treatment of cells with MG132, a proteasomal inhibitor, efficiently blocked the induction of the ADAM12 protein and mRNA by TGF␤1 (Fig.  5, C and D). This result, together with the effect mediated by cycloheximide, indicates that TGF␤1 signaling leads to proteasomal degradation of a repressor of the Adam12 gene. SnoN, a negative regulator of TGF␤1 signaling, is known to be rapidly degraded in response to TGF␤1 stimulation, and thus may be a good candidate for a repressor of the Adam12 gene. Indeed, analysis of SnoN protein levels in cells treated with TGF␤1 confirmed that SnoN, a 75-kDa protein, was degraded within 30 min of TGF␤1 treatment and remained at a low level for up to 4 h after adding TGF␤1 (Fig. 5E). As SnoN expression is up-regulated by TGF␤1 signaling as a part of the negative feedback loop mechanism that limits the duration and strength of the TGF␤1 signals (32), the level of SnoN began to rise and eventually returned to its initial level after 8 -16 h of TGF␤1 treatment (Fig. 5E).
To determine whether SnoN is directly involved in TGF␤1-induced up-regulation of ADAM12, we first studied the effect of SnoN overexpression on the level of ADAM12 in TGF␤1-treated cells. NIH3T3 cells were transduced with SnoN or control retroviruses, and 24 h later they were incubated for 16 h with or without TGF␤1. As shown in Fig. 6, A and B, overexpression of SnoN caused ϳ50% reduction in ADAM12 expression in response to TGF␤1, and this effect was statistically significant. The lack of a stronger inhibition of ADAM12 expression by SnoN may be caused by the fact that the efficiency of viral transduction was only ϳ50 -60% (as determined by GFP fluorescence, result not shown).
Next, we examined the effect of knocking down the expression of the endogenous SnoN in NIH3T3 cells on the expression levels of ADAM12. NIH3T3 cells were infected with lentiviral SnoN shRNA or control shRNA particles, and cells with stable incorporation of shRNA vectors were selected in the presence of puromycin. The level of the endogenous SnoN protein in shSnoN cells was reduced to an undetectable level (Fig. 7A). Induction of ADAM12 mRNA by TGF␤1 was more potent and occurred with faster kinetics in shSnoN cells than in shRNAControl, as revealed by qRT-PCR (Fig. 7B).
As the induction of ADAM12 by TGF␤1 was very efficiently blocked by MG132 (see Fig. 5, C and D), suggesting that degradation of a transcriptional repressor is required for ADAM12 expression, we asked whether MG132 would be equally potent in blocking the induction of ADAM12 in SnoN-deficient cells. If SnoN is the repressor of the Adam12 gene, then in the absence of SnoN, MG132 should have little effect on ADAM12 expression. As shown in Fig. 7C, right panel, incubation of shSnoN cells with TGF␤1 in the presence of MG132 led to the induction of ADAM12 expression, whereas no induction was observed in control cells under the same conditions (Fig. 7C, left  panel). The induction of ADAM12 in shSnoN cells in the presence of MG132 was more modest than in the absence of the inhibitor (Fig. 7D), but this can be explained by the actual levels of SnoN in these cells. Although no SnoN was detected in shSnoN cells in the absence of the inhibitor, MG132 treatment resulted in a significant accumulation of SnoN (Fig. 7C). This most likely was the result of incomplete double stranded mRNA degradation, active synthesis of SnoN protein, and inhibition of its degradation. The accumulation of SnoN in MG132-

FIGURE 2. Characterization of TGF␤1-induced expression of ADAM12.
A, the effect of serum on the induction of ADAM12 by TGF␤1. NIH3T3 cells were preincubated for 24 h in the presence of 0.5 or 10% FBS, as indicated, followed by treatment with 2 ng/ml of TGF␤1 for an additional 16 or 48 h. In some cases, 10 M SB-431543, an inhibitor of T␤RI, was added 30 min prior to TGF␤1 treatment. B and C, TGF␤1 dose-response of the induction of ADAM12. B, NIH3T3 cells were incubated for 24 h with the indicated doses of TGF␤1, and the levels of ADAM12 protein were analyzed as described in the legend to Fig. 1. Quantitative differences in the abundance of the full-length ADAM12 (short arrow) are best shown after short exposure of the immunoblot, the mature ADAM12 (long arrow) is visualized after long exposure. C, the intensities of the bands corresponding to the full-length ADAM12 protein in panel B were quantified using densitometry and ImageJ software. The data represent the mean Ϯ S.E. (error bars) from three independent determinations. D-F, time course of ADAM12 induction by TGF␤1. NIH3T3 cells were incubated without or with 2 ng/ml of TGF␤1 for the indicated times. D, the levels of ADAM12 protein were analyzed by Western blotting, as described in the legend to Fig. 1. E, the levels of ADAM12 mRNA were analyzed by semi-quantitative RT-PCR, GAPDH is a gel-loading control. F, the intensities of the bands corresponding to the full-length ADAM12 protein in D and ADAM12 mRNA in E were quantified using densitometry and ImageJ software. The data represent the mean Ϯ S.E. (error bars) from three independent determinations. treated shSnoN cells might be the reason why SnoN knockdown by shRNA did not fully bypass the inhibitory effect of MG132 on the induction of ADAM12 by TGF␤1. Alternatively, SnoN might not be the sole repressor of the Adam12 gene. A possible candidate for ADAM12 repression may be a SnoN-related protein Ski, which has also been reported to undergo proteasomal degradation in response to TGF␤1 stimulation (42).
To determine whether Ski plays a role in repressing Adam12, we examined the effect of TGF␤1 on the level of ADAM12 protein in Ski Ϫ/Ϫ cells, in the presence or absence of MG132 (Fig. 8). As shown in Fig. 8C, inhibition of protein degradation completely blocked the induction of ADAM12 by TGF␤1 in Ski Ϫ/Ϫ cells. Thus, Ski does not appear to play a major role in repressing the Adam12 gene.
To determine whether the role of SnoN in TGF␤1-induced expression of ADAM12 observed in NIH3T3 cells can be extended to other cell types, we examined the effect of TGF␤1 on ADAM12 expression in human cancer cell lines that typically express higher levels of SnoN than untransformed cells. For our analysis, we selected several cell lines in which the major components of the canonical TGF␤1 signaling pathway remain intact and which are responsive to TGF␤1 signals: MDA-MB-231, MDA-MB-435S, DU145, and HT1080 (43)(44)(45). The basal level of ADAM12 mRNA in all four cancer cell lines was higher than in MCF10A cells, but it was not inhibited by SB-431542 (results not shown), indicating that this was not a result of an elevated autocrine/paracrine TGF␤1 signaling. Upon TGF␤1 treatment, there was a ϳ7-fold increase in ADAM12 expression in MCF10A cells and a much lower (Ͻ2fold) increase in MDA-MB-231, MDA-MB-435S, DU145, and HT1080 cells (Fig. 9A). As expected, no induction of ADAM12 was detected in Smad4-deficient MDA-MB-468 cells, or in T47D and MCF7 cells, which express very low levels of T␤RII and do not respond to TGF␤1 signals (46) (results not shown). In agreement with previous reports, the level of SnoN in cancer cells was consistently higher than in the normal mammary epithelial cell line MCF10A, whereas the level of Ski showed more variation (Fig. 9B). Thus, it appears that among TGF␤1-responsive cells, there is an inverse correlation between the extent of ADAM12 induction by TGF␤1 and the level of SnoN, but not Ski (Fig. 9C). These results further support the role of SnoN in repression of the ADAM12 gene.

DISCUSSION
The induction of ADAM12 expression by TGF␤1 was first reported in hepatic stellate cells (13,24). Our studies demonstrate that induction of ADAM12 by TGF␤1 is a more general phenomenon and it takes place in fibroblasts and in epithelial cells. Recently, it has been shown that ADAM12 can enhance TGF␤ signaling through its interaction with T␤RII (47). Thus, TGF␤1-mediated induction of ADAM12 may be part of a positive feedback loop mechanism, leading to self-amplification of TGF␤1 signals.
TGF␤1 does not appear to cause general up-regulation of ADAM proteins, as at least two other ADAMs, ADAM9 and ADAM15, are not affected by the cytokine treatment (Ref. 13 and this report). According to our knowledge, the only other ADAM reported to be up-regulated by TGF␤1 is ADAM19, whose mRNA levels increase in alveolar epithelial cells treated with TGF␤1 (48). The same study showed that TGF␤1 treatment leads to down-regulation of the ADAM28 mRNA level, further highlighting the specificity of TGF␤1 in regulating the expression of individual ADAMs.
According to previous reports, the induction of ADAM12 by TGF␤1 in hepatic stellate cells occurred with slow kinetics, as the level of ADAM12 mRNA was not significantly changed before 24 h of TGF␤1 treatment (13). Furthermore, FIGURE 3. Induction of ADAM12 expression by TGF␤1 occurs at the transcriptional level. A, NIH3T3 cells were pretreated for 15 min with 5 g/ml of actinomycin D or with DMSO, then incubated for 8 h with or without 2 ng/ml of TGF␤1, followed by Western blotting with anti-ADAM12 antibody, as described in the legend to Fig. 1. Tubulin is a gel loading control. The experiment was repeated 3 times with similar results. B, qRT-PCR analysis of ADAM12 expression. NIH3T3 cells were pretreated for 15 min with 5 g/ml of actinomycin D or DMSO, and then incubated without or with 2 ng/ml of TGF␤1 for 4 h. The level of ADAM12 mRNA was normalized to GAPDH. The data represent the mean Ϯ S.E. (error bars) from three independent experiments. Asterisk indicates the statistically significant effect (p Ͻ 0.05) of inhibitor treatment. C, Northern blot analysis of ADAM12 expression. NIH3T3 cells were pretreated for 15 min with 5 g/ml of actinomycin D or with DMSO, and then treated without or with 2 ng/ml of TGF␤1 for the indicated times. Total mRNA was extracted and hybridized with ADAM12 and ␤-actin probes. The analysis was repeated 3 times with similar results. D, the effect of TGF␤1 on the stability of ADAM12 mRNA. NIH3T3 cells were incubated for 16 h with 2 ng/ml of TGF␤1 (squares) or with vehicle alone (triangles), and then 5 g/ml of actinomycin D was added (time 0). Total mRNA was extracted at various time points and ADAM12 mRNA levels, normalized to GAPDH mRNA, were determined by qRT-PCR. The data represent changes in ADAM12 mRNA relative to the levels at time 0 (mean values Ϯ S.E. (error bars), n ϭ 3). The half-life of ADAM12 mRNA in the presence or absence of TGF␤1 was estimated as 6.5 Ϯ 1.0 h Ϫ1 (solid line) and 6.6 Ϯ 1.7 h Ϫ1 (dashed line), respectively. the effect of TGF␤1 on ADAM12 expression was partially blocked by inhibitors of MAPKs, phosphoinositide 3-kinase, or p70S6 kinase (24), indicating that TGF␤1 might up-regulate ADAM12 via Smad-independent pathways. In this report, we show that the induction of ADAM12 by TGF␤1 in mouse fibroblasts is more rapid, and increased levels of ADAM12 mRNA are detected within ϳ2 h after TGF␤1 treatment. ADAM12 induction is completely blocked upon inhibition of transcription or proteasomal degradation, but is not blocked by cycloheximide, an inhibitor of translation. Based on these observations, we conclude that TGF␤1 induces ADAM12 expression by relieving a transcriptional repression of the Adam12 gene. We show that SnoN, a transcriptional repressor that is efficiently degraded after TGF␤1 stimulation, is involved in the regulation of ADAM12 expression. Overexpression of SnoN reduces ADAM12 induction, and down-regulation of SnoN expression by shRNA leads to increased induction of ADAM12 by TGF␤1. Importantly, whereas pre-treatment of shControl cells with MG132, a proteasomal inhibitor, completely blocks the induction of ADAM12 by TGF␤1, the same pre-treatment of shSnoN cells still allows ADAM12 to be induced by TGF␤1. The extent of induction of ADAM12 by TGF␤1 in shSnoN cells in the presence of MG132 is not as potent as in the absence of the proteasomal inhibitor, but this is most likely due to the fact that MG132 treatment eventually leads to build up of SnoN protein in these cells. In contrast, incubation of Ski Ϫ/Ϫ MEFs with MG132 fully abolishes the effect of TGF␤1 on ADAM12 expression. Clearly, these results indicate that SnoN, rather than Ski, is a repressor that needs to be degraded to derepress the Adam12 gene.
Although SnoN was first described as a negative regulator of TGF␤1 signaling, it is not a universal repressor of TGF␤1 responsive genes. Rather, SnoN acts in a gene-specific manner, and inhibition of the proteasome leads to abrogation of certain TGF␤1 target gene regulation, without any effect on other TGF␤1 target genes (49). Furthermore, recent reports indicate that under certain circumstances SnoN can act as a positive The translation initiation site is at nucleotide position ϩ1, the putative transcription start site (TSS) is located at position Ϫ220, the 5Ј-untranslated region is shown in black, the genomic upstream sequence is in gray, the 9ϫGTCT repeat at Ϫ1368 to Ϫ1333 is indicated by the asterisk, the activin response element at Ϫ2685 to Ϫ2678 is indicated by the circle. NIH3T3 cells were co-transfected with individual ADAM12 reporters and pRL-TK, and analyzed 24 h later. After normalization to Renilla luciferase, firefly luciferase activity relative to that of pGL4 was calculated for each reporter. The data represent the mean Ϯ S.E. (error bars) from two to three independent experiments. The core Adam12 promoter is located between residues Ϫ407 and Ϫ129, and additional positive regulatory elements may be present between residues Ϫ5090 and Ϫ3752. B, the effect of TGF␤1 on the activity of the Adam12 promoter; the p3TP-Lux vector is positive control. C, induction of ADAM12 in Smad2 Ϫ/Ϫ cells. Wild-type (WT) or Smad2 Ϫ/Ϫ MEFs were incubated for 24 h with or without 2 ng/ml of TGF␤1, followed by analysis of ADAM12 expression by Western blotting, as described in the legend to Fig. 1; tubulin is a loading control. The full-length ADAM12 protein is indicated by a short arrow, the mature form lacking the pro-domains is indicated by long arrow, the asterisk denotes a nonspecific band. Fold-induction of ADAM12 by TGF␤1 was quantified by densitometry and ImageJ; the data represent the means from three different determinations Ϯ S.E. (error bars); the asterisk indicates statistically significant effect (p Ͻ 0.05). D, induction of ADAM12 in Smad3 Ϫ/Ϫ cells. The experiments and quantification of the results were performed as in C.
mediator of transcription (50). Indeed, a microarray analysis of human lung cancer A549 cells demonstrated that a large set of genes is down-regulated in cells lacking SnoN, suggesting that SnoN may function as a transcriptional activator, in addition to acting as a transcriptional repressor of the Smad proteins (43). Thus, the effect of SnoN on a particular target gene is not easily predictable, and a negative or positive regulation is possible. As shown in this study, in the case of ADAM12, SnoN is a negative regulator of its expression.
ADAM12 expression is dysregulated in many cancers, cardiac hypertrophy, during aseptic loosening of hip replacement implants, and in experimental autoimmune encephalomyelitis (1, 19 -23). Each of these pathological conditions is accompanied by increased levels of TGF␤1 and/or abnormal expression of SnoN (31,(51)(52)(53)(54). It is tempting to speculate that up-regulation of ADAM12 expression in cardiac hypertrophy, in inflammatory responses related to osteolysis, or in experimental autoimmune encephalomyelitis is directly linked to activation of TGF␤1 signaling. Furthermore, during fibrotic kidney disease after obstructive injury, SnoN is down-regulated due to enhanced ubiquitin-mediated degradation, which in turn is a result of TGF␤1-induced expression of Smurf2 ubiquitin ligase (55,56). It is possible that a similar chain of events leads to degradation of SnoN during fibrosis associated with cardiac FIGURE 5. Induction of ADAM12 expression by TGF␤1 involves derepression of the Adam12 gene. A and B, ADAM12 expression is induced in the absence of protein synthesis. In A, NIH3T3 cells were pretreated for 2 h with 5 g/ml of cycloheximide (CHX), an inhibitor of translation, and then treated with or without 2 ng/ml of TGF␤1 for the indicated amounts of time, with the continuous presence of CHX. Total RNA was extracted and hybridized with ADAM12 or ␤-actin probes. In B, NIH3T3 cells were pre-treated for 15 min with 5 g/ml of actinomycin D or DMSO, and then treated for 4 h with or without 5 g/ml of CHX, as indicated. The level of ADAM12 mRNA was evaluated by semi-quantitative RT-PCR, GAPDH is an internal control. The experiments shown in A and B were repeated 3 times with similar results. C and D, induction of ADAM12 by TGF␤1 is blocked by a proteasomal inhibitor. NIH3T3 cells were pre-treated for 1 h with or without 10 M MG132, and then treated for 16 (C) or 4 h (D) with or without 2 ng/ml of TGF␤1, in the presence or absence of MG132, as indicated. The level of ADAM12 protein (C) was analyzed by Western blotting, as described in the legend to Fig. 1; the level of ADAM12 mRNA was quantified by qRT-PCR (D). In D, ADAM12 mRNA was normalized to GAPDH mRNA; the data represent the mean Ϯ S.E. (error bars) from two independent experiments. Asterisk indicate statistically significant effect (p Ͻ 0.05) of inhibitor treatment. E, the level of SnoN repressor at the indicated times after adding TGF␤1 was evaluated by Western blotting. FIGURE 6. Overexpression of SnoN partially inhibits the induction of ADAM12 by TGF␤1. A, NIH3T3 cells were transduced with SnoN or control retroviruses, and 24 h later they were treated for an additional 16 h with or without 2 ng/ml of TGF␤1. ADAM12 and SnoN levels were examined by Western blotting. The full-length ADAM12 (short arrow) and the mature ADAM12 (long arrow) are best visualized at short and long film exposures, respectively. B, the experiment shown in panel A was repeated three times, and the relative changes in the level of ADAM12 protein were quantified by densitometry and ImageJ; the data represent the mean Ϯ S.E. (error bars); asterisk indicates statistically significant effect (p Ͻ 0.05) of SnoN. FIGURE 7. Knockdown of SnoN expression increases ADAM12 expression in response to TGF␤1 stimulation. NIH3T3 cells were stably transduced with SnoN shRNA or control shRNA lentiviruses. A, the level of SnoN protein in shSnoN or shControl cells was evaluated by Western blotting. B, shSnoN and shControl cells were incubated with 2 ng/ml of TGF␤1 for the indicated times. ADAM12 mRNA levels, normalized to GAPDH mRNA, were measured by qRT-PCR; fold-changes over the levels at time 0 are shown. C, shSnoN and shControl cells were incubated for 16 h with or without 2 ng/ml of TGF␤1, in the absence or presence of 10 M MG132, followed by evaluation of ADAM12 and SnoN protein levels. D, the intensities of the bands corresponding to the full-length ADAM12 protein in panel C were quantified using densitometry and ImageJ software. The data represent the mean Ϯ S.E. (error bars) from three independent experiments. Asterisks indicate statistically significant differences (p Ͻ 0.05).
hypertrophy, which would explain high levels of ADAM12 expression in hypertrophic myocardium. In cancer, the situation is more complex, as both TGF␤1, a positive ADAM12 regulator, and SnoN, a negative ADAM12 regulator, are elevated. Our studies in cancer cells show that increased basal expression of ADAM12 is not due to the autocrine/paracrine effects of TGF␤1 produced by some of these cell lines. Rather, it is an inherent feature of cancer cells and may be a result of genetic or epigenetic changes associated with the oncogenic transformation. Increased basal levels of ADAM12 expression also do not correlate with increased SnoN levels in cancer cells. However, cancer cells have many other components of the TGF␤1 pathway aberrantly expressed, including TGF␤ receptors and SMADs, and direct correlations between absolute expression levels of ADAM12 and SnoN in different cell types may be difficult to establish. Importantly, our results indicate that there is an inverse correlation between the level of SnoN in cancer cells and the ability of TGF␤1 to induce ADAM12 expression.
TGF␤ signaling plays dual roles during tumor development. During early phases of tumorigenesis, TGF␤ acts as a tumor suppressor by limiting cancer cell proliferation and enhancing differentiation. In later stages, TGF␤ promotes tumor growth by stimulating cell migration, invasion, and metastasis, by modifying tumor microenvironment, and by modulating host immune responses (51,57). High expression of SnoN makes cancer cells resistant to the anti-proliferative effects of TGF␤ (a pro-oncogenic role of SnoN) and limits their metastatic potential (an anti-tumorigenic activity of SnoN) (43). Interestingly, both tumor-promoting and tumor-suppressing roles have been postulated for ADAM12. Transgenic expression of ADAM12⌬cyt (lacking the cytoplasmic domain) in mammary tumors of murine mammary tumor virus-PyMT mice accelerates tumor growth, which is consistent with the tumor-promoting function of ADAM12 (8). On the other hand, ADAM12 is frequently mutated in breast cancer (2,3), and cancer-associated mutations cause mislocalization of the ADAM12 protein in cells and interfere with its function at the cell surface, suggesting that the wild-type ADAM12 may also play an anti-tumor role (4). We believe that identification of SnoN as a repressor of the ADAM12 gene will contribute to advances in studies The intensities of the bands corresponding to the full-length ADAM12 protein were quantified using densitometry and ImageJ software (right panels). The data represent the mean Ϯ S.E. (error bars) from three independent experiments. on the role of ADAM12 in tumor progression and in the development of other pathological conditions.