Research ArticleCancer

TRAF6 Stimulates the Tumor-Promoting Effects of TGFβ Type I Receptor Through Polyubiquitination and Activation of Presenilin 1

See allHide authors and affiliations

Science Signaling  07 Jan 2014:
Vol. 7, Issue 307, pp. ra2
DOI: 10.1126/scisignal.2004207

Abstract

Transforming growth factor–β (TGFβ) can be both a tumor promoter and suppressor, although the mechanisms behind the protumorigenic switch remain to be fully elucidated. The TGFβ type I receptor (TβRI) is proteolytically cleaved in the ectodomain region. Cleavage requires the combined activities of tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6) and TNF-α–converting enzyme (TACE). The cleavage event occurs selectively in cancer cells and generates an intracellular domain (ICD) of TβRI, which enters the nucleus to mediate gene transcription. Presenilin 1 (PS1), a γ-secretase catalytic core component, mediates intramembrane proteolysis of transmembrane receptors, such as Notch. We showed that TGFβ increased both the abundance and activity of PS1. TRAF6 recruited PS1 to the TβRI complex and promoted lysine-63–linked polyubiquitination of PS1, which activated PS1. Furthermore, PS1 cleaved TβRI in the transmembrane domain between valine-129 and isoleucine-130, and ICD generation was inhibited when these residues were mutated to alanine. We also showed that, after entering the nucleus, TβRI-ICD bound to the promoter and increased the transcription of the gene encoding TβRI. The TRAF6- and PS1-induced intramembrane proteolysis of TβRI promoted TGFβ-induced invasion of various cancer cells in vitro. Furthermore, when a mouse xenograft model of prostate cancer was treated with the γ-secretase inhibitor DBZ {(2S)-2-[2-(3,5-difluorophenyl)-acetylamino]-N-(5-methyl-6-oxo-6,7-dihydro-5H-dibenzo[b,d]azepin-7-yl)-propionamide}, generation of TβRI-ICD was prevented, transcription of the gene encoding the proinvasive transcription factor Snail1 was reduced, and tumor growth was inhibited. These results suggest that γ-secretase inhibitors may be useful for treating aggressive prostate cancer.

INTRODUCTION

Transforming growth factor–β (TGFβ) family members play important roles in different cellular processes such as proliferation, differentiation, epithelial-mesenchymal transition (EMT), growth arrest, and apoptosis (1, 2). TGFβ exerts its cellular effects by binding to the constitutively active TGFβ type II receptor (TβRII), which then forms a heterotetrameric complex with the TGFβ type I receptor (TβRI). TβRII activates TβRI by phosphorylation (3). The activated TβRI phosphorylates the receptor-associated Smad proteins (R-Smads), which, in complex with Smad4, have key roles in TGFβ-induced gene transcription (48).

TGFβ also signals through non-Smad pathways. The tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6) is a ubiquitin E3 ligase that interacts with a conserved consensus motif in TβRI (9, 10). The TβRI-TRAF6 interaction leads to TGFβ-induced autoubiquitination of TRAF6 and subsequent Lys63-linked polyubiquitination of the mitogen-activated protein kinase kinase kinase TAK1 (TGFβ-associated kinase 1) (9, 1113).

Presenilin 1 (PS1) is a transmembrane protein that is the catalytic subunit of the γ-secretase complex (14, 15). The γ-secretase complex cleaves more than 60 transmembrane receptors, including the Notch receptor (16). The inactive PS1 holoprotein (42 to 43 kD) is cleaved by an unknown presenilinase to generate an N-terminal fragment (NTF; 27 to 28 kD) and a C-terminal fragment (CTF; 16 to 17 kD) (17). The structure of PS1 is currently under debate. PS1 has been proposed to contain nine transmembrane domains. The cytosolic N terminus is thought to comprise the first six hydrophobic regions, followed by a cytosolic loop domain and the C terminus, which comprises three transmembrane domains (1821). PS1 associates with nicastrin, presenilin enhancer 2 (pen-2), and the anterior pharynx defective-1 (aph-1) proteins, and these interactions lead to the formation of an active γ-secretase complex (22). PS1 has a conserved TRAF6 binding motif, and its association with TRAF6 leads to enhanced autoubiquitination of TRAF6 (23).

The γ-secretase complex cleaves various transmembrane proteins in their intracellular regions in a regulated series of events, termed regulated intramembrane proteolysis (RIP) (24). RIP releases the intracellular domain (ICD) of target transmembrane proteins such as Notch. The released ICD can then translocate to the nucleus, where it activates the transcription of target genes. Similarly, the TβRI is cleaved in the ectodomain region by TNF-α–converting enzyme (TACE, also known as ADAM17). This ectodomain is then shed from the membrane, which results in reduced cell surface availability of TβRI and reduced TGFβ signaling (25). We previously reported that extracellular cleavage of TβRI by TACE is preceded by ubiquitination of TβRI in a TRAF6- and Lys63-dependent manner (26). This cleavage leads to the generation of a C-terminal ICD, which can translocate to the nucleus, where it associates with the transcriptional coactivator p300 and regulates transcription of proinvasive target genes such as Snail1 and MMP2.

In the present study, we wanted to understand how TβRI-ICD is released from the cell membrane and its effects on tumor biology. Our results show that TβRI is cleaved in the transmembrane region by activated PS1. We found that TRAF6 mediated the activation of PS1 through TGFβ-dependent, Lys63-linked polyubiquitination. This γ-secretase–induced cleavage of TβRI promotes the nuclear translocation of TβRI-ICD. This translocation and ectopic expression of hemagglutinin (HA)–tagged TβRI-ICD induces invasive behavior in various types of cancer cells. Moreover, we found that treatment of cancer cells with γ-secretase inhibitors prevented TGFβ-induced cancer cell invasiveness in vitro, inhibited generation of TβRI-ICD, and inhibited tumor growth in a prostate cancer xenograft model.

RESULTS

PS1 cleaves TβRI

PS1 functions as the catalytic core of the γ-secretase complex, which cleaves target proteins in a transmembrane domain. We used Western blotting and confocal imaging to examine whether RNA interference [small interfering RNA (siRNA)] directed against PS1 or genetic ablation of PS1 affected the cleavage of TβRI in human prostate cancer cells (PC-3U) and in mouse embryonic fibroblasts lacking PS1 (PSI−/− MEFs). TGFβ treatment for 6 hours induced the generation of a 34-kD TβRI-ICD under control conditions, but not under PS1 knockdown conditions (Fig. 1A). Furthermore, the abundance of TβRI-ICD was also significantly reduced in PSI−/− MEFs (Fig. 1B). Moreover, in PS1 knockdown PC-3U cells and PS1−/− MEFs, TGFβ did not induce nuclear translocation of TβRI-ICD (fig. S1, A to C). Retransfection of PS1 in PSI−/− MEFs restored TGFβ- and PS1-dependent generation of TβRI-ICD (Fig. 1C).

Fig. 1 TGFβ stimulates PS1, which promotes TβRI cleavage.

(A and B) Representative immunoblots were probed for full-length TβRI (TβRI-FL), TβRI-ICD, and PS1 (PS1-NTF). PC-3U cells transiently transfected with control or PS1-specific siRNA (n = 6 independent experiments) (A) and wild-type (WT) MEFs or PS1−/− MEFs (−/−) (n = 5 independent experiments) (B) were treated as indicated. (C and D) Representative immunoblots probed for TβRI and PS1-NTF in WT (PSI+/+) MEFs, PS1−/− MEFs, or PS1−/− MEFs transiently transfected with PS1 (n = 5 independent experiments) (C), or probed for PS1 in PC-3U cells transiently transfected with Myc-PS1 (n = 5 independent experiments) (D), and treated as indicated. (E) qRT-PCR analyses of PS1 mRNA (n = 4 independent experiments). (F and G) Representative immunoblots probed for TβRI-ICD in PC-3U cells treated or not treated with the γ-secretase inhibitors L-685,458 (F) or compound E (CpdE) (G) (n = 5 independent experiments). (H) Representative images of TβRI-ICD (red) in PC-3U cells treated as indicated (n = 3 independent experiments); superimposed images (merged) show the localization of TβRI-ICD in the nucleus (blue). Scale bar, 20 μm; the percentage of cells in the nucleus is based on 200 total cells counted in each group (right). Bar graphs show the means ± SEM from four to six independent experiments; *P < 0.05, Mann-Whitney U test (two-tailed).

TGFβ regulates PS1 expression and processing

Presenilins are synthesized as holoproteins and are cleaved to form fragments that comprise the N terminus (NTF) and C terminus (CTF), which associate to form a heterodimeric protein (27, 28). In PC-3U cells that overexpressed recombinant PS1 with a Myc tag (Myc-PS1), the intensity of a 32-kD band that corresponded in size to PS1-NTF increased as the amount of Myc-PS1 transfected was increased (Fig. 1D). Moreover, TGFβ treatment for 0.5 hour enhanced the abundance of PS1-NTF (Fig. 1D), suggesting that TGFβ promoted the endoproteolysis of PS1. Quantitative real-time polymerase chain reaction (qRT-PCR) revealed that TGFβ induced an increase in PS1 mRNA expression in control PC-3U cells, but not in cells that lacked PS1 (Fig. 1E). We found that TGFβ-induced expression of PS1 required TACE and the TβRI protein, but not the kinase activity of TβRI; this result suggested that TGFβ-induced expression of PS1 required the non-Smad signaling pathway (fig. S1, D to F). We validated the specificity of the PS1 antibodies (fig. S1, G and H). TGFβ treatment also increased the abundance of full-length PS1 and induced the appearance of an 18-kD band that corresponded to PS1-CTF (fig. S1, I to K).

To examine the role of γ-secretase activity in the generation and nuclear translocation of TβRI-ICD, we treated PC-3U cells with the γ-secretase inhibitors L-685,458 and compound E. Treatment of PC-3U cells with these inhibitors prevented the TGFβ-induced generation of TβRI-ICD, as determined by immunoblotting (Fig. 1, F and G) and imaging (Fig. 1H). Together, these results suggest that TGFβ increases the abundance and activation of PS1, which results in the generation of TβRI-ICD and its translocation to the nucleus.

PS1 interacts with TβRI

Because PS1 was implicated in the generation of TβRI-ICD, we examined whether PS1 formed a complex with TβRI. In PC-3U cells, endogenous PS1 coimmunoprecipitated with TβRI, an interaction that was enhanced by stimulation with TGFβ (Fig. 2A). In addition, endogenous TβRI was immunoprecipitated with PS1 in lysate from wild-type MEFs, but not in lysate from PS1−/− MEFs (Fig. 2B). In addition, we performed an in situ proximity ligation assay (PLA), which enables the detection of protein complexes in cells with high sensitivity (fig. S2A) (29), and found significantly more endogenous TβRI and PS1 complex formation in TGFβ-stimulated PC-3U cells than in unstimulated cells (Fig. 2C and fig. S2B). Moreover, the interaction between ectopically expressed TβRI with a HA tag (HA-TβRI) and Myc-PS1 in PC-3U cells was enhanced upon TGFβ stimulation (Fig. 2D). Finally, confocal imaging showed that ectopically expressed HA-tagged, constitutively active TβRI and Myc-tagged PS1 colocalized in TGFβ-stimulated PC-3U cells (Fig. 2E). These results show that PS1 forms a complex with TβRI and that TGFβ stimulates the formation of this complex.

Fig. 2 Endogenous PS1 associates with TβRI.

(A, B, and D) Total cell lysates (TCL) were subjected to immunoprecipitation and immunoblotting. (A and B) Representative immunoblots probed for PS1-NTF in complex with TβRI in PC-3U cells (n = 4 independent experiments) (A), or in WT and PS1−/− MEFs (n = 4 independent experiments) (B), treated as indicated. (C) Representative confocal images of PC-3U cells treated as indicated. TGFβ induced the formation of PS1-TβRI complexes (red) (see fig. S2A). Bar graph shows the means and SD from four independent experiments; 350 cells were counted in each group. Negative control data for the PLA antibodies are shown in fig. S2B. (D) Representative immunoblots probed for Myc-PS1 in complex with HA-TβRI in PC-3U cells (n = 4 independent experiments), treated as indicated. (E) Representative confocal images of ectopically expressed Myc-PS1 and HA-TβRI in PC-3U cells, visualized with Myc (red) and HA antibodies (green) (n = 3 independent experiments, 0.5 × 105 cells/cm2). Superimposed images (merged) show colocalization of the proteins (yellow). In (C) and (E), cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; blue). Bar graphs in (A), (B), and (D) show the means ± SEM from four independent experiments. *P < 0.05, Mann-Whitney U test (two-tailed).

TGFβ promotes Lys63-linked polyubiquitination of PS1-NTF

To explore the mechanism by which TGFβ activates PS1, we performed a kinetic analysis in a cellular ubiquitination assay with untreated or TGFβ-treated PC-3U cells. TGFβ treatment induced the appearance of an ~40-kD band that corresponded to ubiquitinated PS1-NTF, suggesting that PS1-NTF underwent Lys63-linked polyubiquitination (Fig. 3A and fig. S3A). Furthermore, we found that TGFβ induced ubiquitination of PS1-NTF only in PC-3U cells transfected with a Lys63 (K63-only) ubiquitin mutant, but not a Lys48 (K48-only) ubiquitin mutant (Fig. 3B). In PC-3U cells, knockdown of TRAF6 by siRNA inhibited TGFβ-induced, Lys63-linked polyubiquitination of PS1-NTF (Fig. 3C). PLA revealed that significantly more PS1-NTF was complexed with Lys63-linked polyubiquitin in TGFβ-stimulated PC-3U cells than in unstimulated cells (Fig. 3D and fig. S2B). Together, these results demonstrate that TGFβ promotes Lys63-linked polyubiquitination of PS1-NTF in a TRAF6-dependent manner.

Fig. 3 TGFβ promotes Lys63-dependent polyubiquitination of PS1 in vitro.

(A to C) Representative immunoblots showing ubiquitinated PS1 in PC-3U cells, transfected with HA-tagged ubiquitin (HA-Ub; n = 4 independent experiments) (A); HA-tagged K63-only or K48-only ubiquitin mutants (n = 7 independent experiments) (B); or TRAF6 siRNA (n = 4 independent experiments) (C) and treated as indicated. Total cell lysate (TCL) proteins were immunoprecipitated with anti–PS1-NTF antibodies and immunoblotted (IB) with antibodies specific for K63-linked polyubiquitin (K63-Ub) (A and C) or with P4D1 antisera (B) to visualize polyubiquitin. (D) Representative confocal images of PS1 complexed with K63-ubiquitinated protein chains, detected with a PLA in PC-3U cells, treated as indicated. Quantification in lower panel shows the means ± SEM of four independent experiments; 350 cells were analyzed in each group. Cell nuclei were stained with DAPI. Scale bar, 20 μm. Bar graphs in (A) to (C) show the means ± SEM from four to seven independent experiments. *P < 0.05, ***P < 0.001, Mann-Whitney U test (two-tailed).

TRAF6 recruits PS1 to TβRI

Coimmunoprecipitation experiments demonstrated that endogenous PS1 interacted with TRAF6, an interaction that was enhanced by TGFβ (Fig. 4A). Furthermore, in transfected PC-3U cells, endogenous PS1 interacted with wild-type TRAF6, but not with the catalytically inactive C70A TRAF6 mutant (Fig. 4B and fig. S3, B to D). These results suggested that autoubiquitination of TRAF6 was required for the association between TRAF6 and PS1, and that PS1 was a substrate for TRAF6. Moreover, the interaction between TRAF6 and PS1 was enhanced by TGFβ treatment, which was confirmed with PLA (Fig. 4C and fig. S2B). Confocal imaging of transiently transfected PC-3U cells showed that TGFβ treatment induced colocalization of Flag-tagged TRAF6 (Flag-TRAF6) and Myc-PS1 (Fig. 4D).

Fig. 4 Active TRAF6 is required for an association between TβRI and PS1.

Total cell lysate (TCL) proteins were immunoprecipitated with anti–PS1-NTF antibodies (A, B, and F) or anti-TβRI antibodies (G) and immunoblotted. (A and B) Representative immunoblots (IB) were probed to detect TRAF6 (n = 5 independent experiments) (A) or the Flag-tagged TRAF6 or C70A TRAF6 mutant (n = 4 independent expeiments) (B). (C and D) Representative confocal images of PC-3U cells, treated as indicated. PLA was used to detect TRAF6-PS1 complexes (red) (C). Quantification in lower panel shows the means ± SEM of four experiments; 350 cells were analyzed in each group. Representative confocal images of ectopically expressed Myc-PS1 (red) and Flag-TRAF6 (green) in PC-3U cells, treated as indicated (D). Superimposed images (merged) show colocalization of Myc-PS1 and Flag-TRAF6 (yellow) (n = 3 independent experiments; 0.5 × 105 cells/cm2). Cell nuclei were stained with DAPI. Scale bars, 20 μm. (E) Representative immunoblots probed for PSI or TRAF6 in PC-3U cells, treated as indicated (n = 5 independent experiments). (F and G) PC-3U cells were transiently transfected with HA-tagged TβRI or mutant TβRI (E161A) and treated as indicated. Representative immunoblots of immunoprecipitated proteins were probed to detect HA (n = 4 independent experiments) (F) or PS1-NTF (n = 4 independent experiments) (G). Bar graphs show the means ± SEM from four to five independent experiments. *P < 0.05, Mann-Whitney U test (two-tailed).

Next, we investigated whether TRAF6 was required for the TGFβ-induced generation of PS1-NTF in PC-3U cells. Silencing of TRAF6 inhibited the TGFβ induction of PS1-NTF (Fig. 4E). Previous studies have shown that TRAF6 interacts with TβRI at a conserved consensus motif in TβRI, an interaction that is attenuated by a point mutation, E161A, in this motif (9). In transfected PC-3U cells, PS1 interacted with wild-type TβRI, but not with the TRAF6 binding–deficient TβRI E161A mutant (Fig. 4F). Furthermore, the interaction between TβRI and PS1 was lost upon silencing of TRAF6 (Fig. 4G). These findings indicated that TRAF6 promoted the interaction between TβRI and PS1 and, consequently, the cleavage of the TβRI and the generation of TβRI-ICD.

PS1 cleaves the transmembrane region of TβRI

We searched for a possible γ-secretase cleavage site in the transmembrane domain of TβRI. A β-peptide from amyloid precursor protein contains a Val-Ile-Ala motif within its transmembrane domain, and the activated γ-secretase complex cleaves amyloid precursor protein between Val and Ile (27, 28). We therefore explored the possibility that γ-secretase might cleave TβRI between Val129 and Ile130 in the transmembrane domain. We mutated these two residues to alanine residues (VI129-130AA) to create the VI129-130AA TβRI transmembrane mutant (TM-TβRI), which did not undergo cleavage in response to TGFβ stimulation (Fig. 5A). The HA-tagged TβRI transmembrane mutant (HA-TM-TβRI) phosphorylated Smad2 (Fig. 5A), thus indicating that its kinase activity was not affected. Immunoprecipitation experiments showed that endogenous PS1 interacted with both TβRI and the TβRI transmembrane mutant (Fig. 5B). The cleavage site in the transmembrane domain of TβRI was confirmed in transiently transfected human embryonic kidney (HEK) 293T cells, and the TβRI transmembrane mutant could mediate activation of the Smad and p38 signaling pathways (fig. S4A). Confocal imaging of transiently transfected PC-3U cells treated with TGFβ indicated that nuclear HA-TβRI-ICD was detected in cells transfected with wild-type TβRI, but not in cells transfected with the TβRI transmembrane mutant (Fig. 5C). These results suggest that the TβRI transmembrane mutant is not cleaved by the γ-secretase complex, and that the mutation prevents the generation and nuclear translocation of TβRI-ICD.

Fig. 5 Identification of the PS1 cleavage site in TβRI.

(A) Representative immunoblots probed to detect HA-tagged proteins, phosphorylated (p)–Smad2, and p-p38 in PC-3U cells transiently transfected and treated as indicated. Filters were reblotted to detect total Smad2 and p38 proteins (n = 6 independent experiments). (B) Total cell lysates (TCL) from (A) were immunoprecipitated with anti-PS1 antibodies. Representative immunoblots were probed to detect ectopically expressed HA-tagged TβRIs associated with PS1 (n = 4 independent experiments). (C) Representative confocal images of PC-3U cells transiently transfected with WT HA-TβRI or the HA-tagged TβRI transmembrane mutant (HA-TM-TβRI), treated as indicated. Both WT and TM-TβRIs were visualized with HA antibodies (red) (n = 5 independent experiments); bar graph shows the percentage of cells with HA-TβRI-ICD in the nucleus, based on 200 total cells counted in each group. Nuclei were stained with DAPI (blue). Scale bars, 20 μm. (D) qRT-PCR results show TβRI mRNA abundance in PC-3U cells transiently transfected and treated as indicated (n = 4 independent experiments). (E) Chromatin was immunoprecipitated with an antibody directed against TβRI in PC-3U cells, treated as indicated, and probed for the TβRI promoter (n = 5 independent experiments). (F and G) Representative images of invasive PC-3U cells transiently transfected with HA-TβRI or HA-TM-TβRI mutant and treated as indicated. Cells were stained with crystal violet. Bar graphs show the mean A560 values for invasive cells in n = 5 (F) and n = 3 (G) experiments. Bar graphs show the means ± SEM from three to six independent experiments. *P < 0.05, **P ≤ 0.005, Mann-Whitney U test (two-tailed).

We previously reported that TβRI can be cleaved by TACE at Gly120 (26). In the present study, we found that PS1 could cleave wild-type TβRI but not the Gly120 TβRI mutant (fig. S4B). Thus, it is likely that cleavage by PS1 occurs after an initial cleavage by TACE; this finding is consistent with current knowledge about the function of γ-secretase in the RIP of other transmembrane receptors (28).

Nuclear TβRI-ICD promotes its own expression and cell invasion

We previously reported that TβRI-ICD can promote the expression of genes involved in cancer cell invasion (26). To further determine the function of TβRI-ICD, we used qRT-PCR to investigate the potential effects of TβRI-ICD on the expression of the gene encoding TβRI. We found that the addition of TGFβ to PC-3U cells promoted the expression of the gene encoding TβRI in a PS1-dependent manner (Fig. 5D). Chromatin immunoprecipitation (ChIP) assays indicated that TβRI-ICD bound to its own promoter in TGFβ-stimulated PC-3U cells (Fig. 5E) and that transiently transfected HA-TβRI bound to the TβRI promoter in response to TGFβ (fig. S4C). Transient transfection of wild-type TβRI, but not the TβRI transmembrane mutant, promoted cell invasion behavior in PC-3U cells (Fig. 5F). Moreover, in PC-3U cells transfected with TβRI siRNA, reconstitution with wild-type TβRI, but not with the TβRI transmembrane mutant, conferred invasive behavior (Fig. 5G).

TGFβ- and γ-secretase–induced RIP of TβRI occurs at the plasma membrane

To investigate the TGFβ-induced activation of the γ-secretase complex further, we used an inhibitor of exocytosis, monensin, which prevents the transport of proteins from the Golgi to the plasma membrane (30, 31) and can prevent γ-secretase from cleaving Notch in its transmembrane domain (28). Confocal imaging and immunoblotting indicated that monensin treatment of PC-3U cells prevented TGFβ-induced generation and nuclear translocation of TβRI-ICD. Monensin treatment also prevented the colocalization of endogenous TβRI and PS1 (Fig. 6, A and B) and the proteolysis of HA-tagged TβRI and the nuclear translocation of TβRI-ICD without affecting the canonical TGFβ-Smad pathway (Fig. 6, A, C, and D). In monensin-treated cells, HA-tagged wild-type TβRI accumulated in the trans-Golgi compartment marked by syntaxin staining (Fig. 6D). In transiently transfected PC-3U cells, monensin treatment also prevented the colocalization of ectopically expressed TβRI and PS1 and the nuclear translocation of HA-TβRI-ICD (fig. S5). Together, these observations suggest that proteolytic release of TβRI-ICD and its nuclear translocation depend on γ-secretase–mediated cleavage, either at the plasma membrane or shortly after internalization of membrane-bound TβRI.

Fig. 6 TGFβ-dependent activation of the γ-secretase complex generates TβRI-ICD.

(A) Representative immunoblots probed to detect the TβRI-ICD fragment in PC-3U cells treated with TGFβ in the absence or presence of monensin. The filter was reprobed to detect phospho-Smad2 (p-Smad2) and total Smad2 (n = 6 independent experiments). (B) Representative confocal images of endogenous TβRI (red) and endogenous PS1 (green) in PC-3U cells treated as indicated (n = 3 independent experiments; 0.5 × 105 cells/cm2). (C) Representative immunoblots probed to detect the TβRI-ICD fragment in PC-3U cells transiently transfected with HA-TβRI and treated as indicated. The filter was reprobed to detect p-Smad2 and total Smad2 (n = 4 independent experiments). (D) Representative confocal images of transiently transfected, C-terminally tagged HA-TβRI (red) and syntaxin (green) in PC-3U cells treated as indicated. Syntaxin is located in the trans-Golgi (n = 3 independent experiments; 0.5 × 105 cells/cm2). Bar graphs show normalized quantification of the means and SEMs from four to six independent experiments. *P < 0.05, **P < 0.005, Mann-Whitney U test (two-tailed). (B and D) Nuclei were stained with DAPI (blue). Superimposed images (merged) show colocalization of the proteins (yellow). Scale bars, 20 μm.

TGFβ-induced cancer cell invasiveness depends on γ-secretase–generated TβRI-ICD

To determine the role of activated γ-secretase complex in TGFβ-driven cancer cell invasiveness, we next used γ-secretase inhibitors in cell invasion assays. The TGFβ-induced invasiveness of PC-3U cells, human lung carcinoma cells (A549), and breast carcinoma cells (MDA-MB-231) was prevented with the γ-secretase inhibitor L-685,458 (Fig. 7, A to D). Endogenous PS1 was required for TGFβ-induced PC-3U cell invasiveness (Fig. 7E). qRT-PCR analysis confirmed that the transmembrane mutant TβRI did not induce transcription of the genes encoding TβRI and Snail1. Jag1 encodes a Notch ligand that is implicated in TGFβ-driven EMT and metastasis (32, 33). We found that Jag1 mRNA transcription was also induced by TGFβ activation of this pathway (Fig. 7F).

Fig. 7 Inhibitors of γ-secretase prevent TGFβ-induced cancer cell invasiveness.

(A, B, and E) Representative images of invasive cells (crystal violet stain); PC-3U cells (n = 5 independent experiments) (A) and human lung carcinoma (A549) and human breast carcinoma (MDA-MB-231) cells (n = 5 independent experiments each) (B to D) were treated with TGFβ in the absence or presence of the γ-secretase inhibitor L-685,458. (E) Representative images of invasive PC-3U cells (crystal violet stain) transiently transfected with nontargeting (control) siRNA or PS1-specific siRNA (siPS1), treated as indicated (n = 4 independent experiments). (F) qRT-PCR results showing TβRI, Snail1, and Jag1 mRNA abundance in PC-3U cells transiently transfected with WT or the transmembrane mutant TβRI, stimulated with TGFβ as indicated (n = 4 independent experiments). (G to I) PC-3U cells were transiently transfected and treated as indicated. (G) Representative immunoblot probed to confirm protein abundance (n = 4 independent experiments). (H) Representative confocal images show HA-TβRI-ICD (red) translocation to the nucleus (blue) (n = 3 independent experiments; 0.5 × 105 cells/cm2). (I and J) Representative images of invasive cells (crystal violet stain) (n = 4 and n = 5 independent experiments, respectively); NT, not transfected; A560, absorbance of invasive cells. (K and L) qRT-PCR results showing Snail1 and Jag1 mRNA abundance in PC-3U or LNCaP cells transiently transfected and stimulated with TGFβ as indicated (n = 4 independent experiments). Means and SD are presented in (A), (C) to (E), and (I) to (L). Other bar graphs show the means ± SEM from three to four independent experiments. *P < 0.05, **P < 0.005, Mann-Whitney U test (two-tailed).

We generated a plasmid encoding HA-tagged TβRI-ICD (HA-TβRI-ICD) to determine whether TβRI-ICD could promote tumor invasion. Transiently transfected PC-3U cells showed nuclear localization of HA-TβRI-ICD (Fig. 7, G and H), which promoted invasiveness of PC-3U cells (Fig. 7I), which was significantly higher than that in untransfected PC-3U cells, either under basal conditions (not treated with TGFβ) or after TGFβ treatment. Treatment with a γ-secretase inhibitor, either L-685,458 or compound E, did not inhibit cell invasiveness driven by HA-TβRI-ICD. Transient overexpression of HA-TβRI-ICD also promoted LNCaP cell invasion (Fig. 7J). qRT-PCR analysis showed that HA-TβRI-ICD promoted transcription of Snail1 and Jag1 in transiently transfected PC-3U and LNCaP cells, when compared with nontransfected cells (Fig. 7, K and L).

Together, these results demonstrate that TGFβ stimulation activates the γ-secretase complex to generate TβRI-ICD, which promotes invasive behavior in different kinds of cancer cells. Moreover, ectopic overexpression of TβRI-ICD can promote invasive behavior in human prostate cancer cells (PC-3U and LNCaP).

TGFβ promotes the association between TβRI-ICD and Notch-ICD

Because Notch is a key substrate of the γ-secretase complex (34), we investigated whether there is crosstalk between the TGFβ and Notch signaling pathways at the level of TβRI. Immunoprecipitation assays with antibodies directed against TβRI or the Notch intracellular domain (NICD) indicated that endogenous NICD associated with TβRI-ICD, and that the association was enhanced by TGFβ stimulation (Fig. 8A), which also enhanced the colocalization of endogenous TβRI-ICD and NICD (Fig. 8B). Consistent with this observation, we found that ectopically expressed HA-TβRI and Myc-NICD also colocalized in transiently transfected PC-3U cells (Fig. 8C). PLA indicated that TGFβ stimulation significantly increased the formation of TβRI-ICD–NICD complexes in PC-3U cells (Fig. 8D). qRT-PCR indicated that transient transfection of Myc-NICD significantly enhanced transcription of the genes encoding TβRI, the proinvasive transcription factor Snail, and Jag1 in TGFβ-stimulated PC-3U cells (Fig. 8E). These results suggest that TGFβ enhances the interaction and colocalization of TβRI-ICD with NICD and that their association promotes cell-invasive behavior.

Fig. 8 TGFβ-dependent association between TβRI and NICD.

(A) Immunoprecipitated proteins were immunoblotted (IB) to detect endogenous TβRI associated with endogenous NICD in PC-3U cells treated or not treated with TGFβ. Similar results were obtained when NICD or TβRI was immunoprecipitated. Bar graphs show the means ± SEM of n = 4 independent experiments. *P < 0.05, Mann-Whitney U test (two-tailed). (B and C) Representative confocal images of endogenous TβRI (red) and endogenous NICD (green) (B) or ectopic HA-TβRI (red) and ectopic Myc-NICD (green) (C) in PC-3U cells treated with TGFβ as indicated. Images are representative of three independent experiments (0.5 × 105 cells/cm2). Nuclei were stained with DAPI (blue). Superimposed images (merged) show colocalization of the proteins (yellow). Scale bars, 20 μm. (D) Representative confocal images showing association of NICD with TβRI-ICD (red), detected by PLA in PC-3U cells treated as indicated. Nuclei were stained with DAPI (blue). Graph shows the mean ± SEM of five independent experiments; 350 cells were analyzed in each group. ***P < 0.001, Mann-Whitney U test (two-tailed). (E) qRT-PCR results showing TβRI, Snail1, and Jag1 mRNA abundance in PC-3U cells transiently transfected with Myc-NICD, or not transfected (–), and treated as indicated. Bar graphs show the means ± SEM obtained from four independent experiments; *P < 0.05, Mann-Whitney U test (two-tailed).

Treatment with the γ-secretase inhibitor DBZ prevents tumor growth and cleavage of TβRI in vivo

To explore the biological relevance of our findings, we used mouse prostate cancer cells (TRAMPC2) (35) in xenograft experiments in immunocompetent C57BL/6 mice. We found that the γ-secretase inhibitor L-685,458 prevented the generation and nuclear translocation of TβRI-ICD, and that either γ-secretase inhibitor (L-685,458 or compound E) prevented invasive behavior in TRAMPC2 cells (fig. S6).

We then measured tumor volume in C57BL/6 mice that had been subcutaneously injected with TRAMPC2 cells and treated with vehicle [dimethyl sulfoxide (DMSO)] or with the γ-secretase inhibitor DBZ {(2S)-2-[2-(3,5-difluorophenyl)-acetylamino]-N-(5-methyl-6-oxo-6,7-dihydro-5H-dibenzo[b,d]azepin-7-yl)-propionamide}. After 7 and 10 days, the tumor volume was significantly lower in DBZ-treated mice than in control mice (Fig. 9A and fig. S7A). After termination of the experiment on day 10, the weights of the tumors from DBZ-treated mice were lower than those from the control group (Fig. 9B and fig. S7B). The mean body weights in the control and DBZ-treated groups were not significantly different (Fig. 9C and fig. S7C). qRT-PCR analysis of total RNA from tumors indicated that the transcription of genes encoding TβRI, Snail1, and Jag1 was significantly reduced in DBZ-treated mice compared to control mice (Fig. 9, D to F). Immunoblotting analysis of tumor proteins revealed that a TβRI-ICD band of 34 kD was present in tumors from the control group, but not in tumors from the DBZ-treated group (Fig. 9G). Moreover, we found higher amounts of Snail1 protein in tumors from the control group than in tumors from the DBZ-treated group (Fig. 9H). These findings suggest that DBZ inhibits the generation of TβRI-ICD and reduces the expression of the proinvasive factor Snail1 in an in vivo mouse prostate cancer model.

Fig. 9 Inhibitors of γ-secretase prevent generation of TβRI-ICD and inhibit tumor growth.

(A) Mice with prostate cancer TRAMPC2 xenografts were treated with DMSO (vehicle) or DBZ for the indicated number of days. Graph shows the mean relative tumor volumes ± SEM from the indicated numbers (n) of mice. Day 1 was set at 100%. (B) Relative tumor weights (DMSO = 100%) in mice treated with DMSO or DBZ. (C) Relative body weights (day 1 = 100%) were measured in mice treated with DMSO or DBZ for the indicated number of days; no significant difference was detected between treatment and control groups. (D to F) qRT-PCR results showing TβRI, Snail1, and Jag1 mRNA abundance in the indicated numbers (n) of tumors treated with DMSO or DBZ. Bar graphs show the means ± SEM of five independent experiments. (G and H) Representative immunoblots of proteins extracted from tumors treated with DMSO or DBZ, which were probed to detect TβRI (G) or Snail1 (H). *P < 0.05, Mann-Whitney U test (two-tailed).

DISCUSSION

Presenilins are indispensable members of the γ-secretase complex that can cleave various receptors in their transmembrane regions. We have shown that ubiquitination of TβRI by TRAF6 leads to TβRI cleavage in the ectodomain region by TACE (26), which generates an ICD that translocates to the nucleus and binds to the transcriptional coactivator p300, which increases the transcription of several TGFβ target genes, such as Snail1, that encode factors that promote invasiveness (26). In the present study, we demonstrated that TGFβ enhanced the abundance and activation of PS1. The Lys63-linked polyubiquitination of PS1 promoted the activation of the γ-secretase complex. In turn, the activated PS1 cleaved the TβRI in the transmembrane region, which generated the TβRI-ICD. We also found that TRAF6 mediated ubiquitination of PS1 and facilitated the recruitment of PS1 to the TβRI complex. The ICD translocated to the nucleus, where it bound to several gene promoters, including the promoter of its own gene. We have previously shown that TβRI is cleaved by TACE between Gly120 and Ile121 (26), and showed in the present study that TβRI was cleaved in the transmembrane domain by γ-secretase after the TACE cleavage. Together, these results suggest that TβRI undergoes RIP in a manner similar to that described for Notch (fig. S8).

TGFβ treatment of PC-3U cells increased PS1 mRNA and protein abundance and increased the generation of PS1-NTF and PS1-CTF (Fig. 1 and fig. S1, D to K). In addition, TβRI interacted with the PS1-NTF (Fig. 2, A to D, and Fig. 4, F and G) in a TGFβ- and TRAF6-dependent manner. We thus conclude that stimulation by TGFβ increases PS1 abundance and activity in PC-3U cells.

The progression of various cancers, including breast and prostate cancers, is associated with high amounts of TGFβ1 and the Notch ligand Jag1 (36, 37). The Notch and TGFβ pathways are linked by direct protein-protein interactions between Smad3 and NICD (38) that promote EMT (33, 39). Moreover, the protumorigenic effects of TGFβ and Notch appear to be functionally coordinated in some forms of cancer (40, 41). In the present study, we found that NICD was associated with TβRI-ICD, and that NICD promoted the transcription of TβRI, Snail1, and Jag1 in a manner dependent on both TGFβ stimulation and an active γ-secretase complex (Fig. 8). This finding suggests that the two pathways are integrated at several levels, including the processing of TβRI and the association between NICD and TβRI-ICD (fig. S8).

TRAF6 was initially identified as an adaptor protein that activates nuclear factor κB (NF-κB) signaling in interleukin-1–stimulated cells (42). TRAF6 is also a ubiquitin E3 ligase that mediates Lys63-linked polyubiquitination of various proteins. The C-terminal TRAF domain mediates protein-protein interactions and the polyubiquitination of various proteins. Nerve growth factor (NGF) triggers the interaction of TRAF6 with PS1, which has a consensus TRAF6-binding motif, and leads to enhanced ubiquitination of TRAF6 upon NGF-mediated stimulation of the p75 neurotrophin receptor (23). In the present study, we found that PS1 interacted with TRAF6 through the RING domain of TRAF6 (Fig. 4B), and that TGFβ promoted the TRAF6-dependent Lys63-linked polyubiquitination of PS1 (Fig. 3C). These results suggest that the E3 ligase activity of TRAF6 is required for the interaction between TRAF6 and its substrate, PS1.

TRAF6 binds to TβRI and has a role in TGFβ signaling (9). The TβRI-TRAF6 association is required for the TGFβ-mediated autoubiquitination of TRAF6 and the subsequent activation of the TAK1-p38 pathway, which leads to apoptosis and EMT (8, 9, 43). Here, we showed that TRAF6 also has a role as an adaptor protein by enabling PS1 to interact with, and thus cleave, TβRI at the cell membrane (Fig. 4, F and G).

TRAF6 is an amplified oncogene in lung carcinoma; this amplification promotes Ras-driven activation of the proinflammatory transcription factor NF-κB and subsequent tumor-promoting responses, such as anchorage-independent growth (44). Further studies are required to determine whether the cell invasion pathway identified in this report, which was initiated by TGFβ- and TRAF6-induced activation of γ-secretase, contributes to a poor prognosis in patients with tumors that overexpress TGFβ or TRAF6.

Moreover, we found that treatment of various cancer cells with γ-secretase inhibitors in vitro prevents TGFβ-dependent proteolysis of TβRI and invasive behavior. Ectopic overexpression of the ICD of TβRI in two different prostate tumor cell lines efficiently promoted invasive behavior (Fig. 7, I and J). In addition, PC-3U cells treated with γ-secretase inhibitor did not show inhibition of TGFβ-stimulated invasive behavior (Fig. 7I).

Finally, in a xenograft prostate cancer model, we found that treatment of mice with the γ-secretase inhibitor DBZ prevented the generation of TβRI-ICD, prevented the transcription of Snail1 and Jag1, and reduced tumor growth (Fig. 9). These findings suggest that pharmacological interference of the TGFβ-induced oncogenic pathway may be a novel therapeutic strategy for combating aggressive cancer.

MATERIALS AND METHODS

Cell culture

Sub–cell lines of human prostate cancer (PC-3U) cells (45), androgen-sensitive human prostate adenocarcinoma cells (LNCaP), human lung carcinoma cells (A549), and breast carcinoma cells (MDA-MB-231) were cultured in RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 1% glutamine, and 1% penicillin-streptomycin (PEST). Transient transfection of PC-3U cells was performed with FuGENE 6 (Roche) according to the manufacturer’s instructions. We also used immortalized wild-type MEFs or MEFs deficient in presenilin1 or presenilin2 (PS1−/− and PS2−/−, respectively) (46, 47). The MEFs and HEK 293T cells were grown in Dulbecco’s modified Eagle’s medium containing 10% FBS and PEST. All cell cultures were incubated at 37°C in the presence of 5% CO2. FuGENE 6HD was used to transfect MEFs.

TGFβ1 (10 ng/ml) was added to starvation media (see below). PC-3U and LNCaP cells were starved for 18 hours in medium that contained 1% FBS. HEK 293T cells were starved for at least 16 hours in medium containing 3% FBS (further reductions in FBS increased cell apoptosis). MEFs were starved for at least 18 hours in medium containing 0.5% FBS.

Antibodies and other reagents

TGFβ1 was obtained from R&D Systems. Monoclonal antibodies including rabbit anti–phospho-p38, mouse anti-p38, rabbit anti–phospho-Smad2, rabbit anti-Smad2, rabbit anti–lamin A, rabbit anti–β-tubulin, and rabbit anti–Lys48-linked polyubiquitin were obtained from Cell Signaling. Mouse monoclonal anti–β-actin antibody and mouse anti–Flag M2 monoclonal antibody were from Sigma. Rabbit anti-TRAF6 (C-term) antibody was from Zymed Laboratories. Rat anti-PS1 (N-terminal) monoclonal antibody was from Millipore. Mouse anti-PS1 (C-terminal) monoclonal antibody and mouse anti-PS1 (N-terminal) monoclonal antibody were from Thermo Scientific. Mouse anti-Notch1 and rabbit anti-Snail1 were from Novus Biologicals. Mouse anti-polyubiquitin (Lys63 linkage–specific) monoclonal antibody was from Enzo Life Sciences. Rabbit anti-HA, rabbit anti-TβRI V22 (C-terminal), and mouse anti-ubiquitin (P4D1) antibodies were from Santa Cruz Biotechnology. The specificity of the anti-TβRI V22 antibody has been shown previously (26). Secondary horseradish peroxidase–conjugated anti-mouse, anti-rabbit, and anti-rat antibodies were from GE Healthcare. Light chain–specific anti-rabbit and anti-mouse antibodies (IgG L.C.) were from Jackson Laboratories. Secondary Alexa Fluor 555–conjugated donkey anti-rabbit, Alexa Fluor 488–conjugated goat anti-rabbit and goat anti-rat, and Alexa Fluor 555–conjugated donkey anti-mouse antibodies were from Invitrogen. Mounting medium with DAPI was from Vector Laboratories. Pefabloc was from Roche. L-685,458 was purchased from Sigma. Compound E was purchased from Merck.

Immunoblotting and protein interactions in cells

After treatment with TGFβ for the indicated times, cells were washed once with ice-cold phosphate-buffered saline (PBS) and lysed in ice-cold lysis buffer [150 mM NaCl, 50 mM tris (pH 8.0), 1% Triton X-100, 10% (v/v) glycerol, 1 mM aprotinin, 1 mM Pefabloc, and 1 mM sodium orthovanadate]. Protein concentration was measured with the bicinchoninic acid (BCA) protein assay kit (Thermo Scientific). Equal amounts of proteins were immunoprecipitated with the indicated antibodies and then subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE) in 10, 12, or 4 to 12% polyacrylamide gels or in 7% tris-acetate gels (Invitrogen). The proteins were then transferred from the gels onto nitrocellulose membranes with an iBlot apparatus (Invitrogen). A light chain–specific antibody was used to reduce cross-reactions with the IgG heavy chain.

We used the E161A mutant TβRI (9) to substantiate our finding that TRAF6 was required for the interaction between TβRI and PS1. We transfected PC-3U cells with HA-tagged wild-type TβRI or E161A mutant TβRI and subjected cell lysates to immunoprecipitation with a PS1-NTF antiserum, followed by immunoblotting with HA antibodies.

Small interfering RNA

We obtained ON-TARGETplus SMARTpool siRNA for the knockdown of human PS1 (PSEN-1), SMARTpool TRAF6 siRNA for the knockdown of TRAF6, and GENOME nontargeting siRNA #1 for the knockdown of TβRI from Dharmacon Research. PC-3U cells were transfected with a specific or control siRNA with Oligofectamine reagent (Invitrogen) according to the manufacturer’s protocol.

Plasmids

Positive clones were verified by sequencing the inserted complementary DNA (cDNA). A constitutively active TβRI with HA fused to the C terminus was a gift from P. ten Dijke (University of Leiden, the Netherlands). This plasmid was modified to create the HA-tagged, VI129-130AA transmembrane TβRI mutant by PCR, and the mutation was confirmed by sequencing. Construction of the HA-tagged TβRI-E161A plasmid has been described previously (9). The HA-TβRI-ICD fragment, which lacked amino acid residues 1 to 178 of the full-length TβRI, was PCR-amplified from the full-length TβRI cDNA. The PCR product was cloned into a pcDNA3 vector with Hind III and Not I. The HA tag was ligated to the C terminus of the TβRI-ICD cDNA with Not I and Xho I, yielding a construct of 1017 base pairs. The expression plasmids for 3xHA-tagged wild-type ubiquitin and K48- and K63-only ubiquitin mutants were gifts from V. M. Dixit (Genentech, San Francisco, CA). The Myc-tagged wild-type PS1 plasmid was a gift from B. De Strooper (VIB Center for the Biology of Disease, Leuven, Belgium). The PS1 plasmid was a gift from H. Karlström (Karolinska Institute, Stockholm, Sweden). The Myc-NICD plasmid was a gift from R. Kopan (Washington University, Seattle, WA). Expression vectors for the HA-tagged transmembrane mutants of TβRI were generated by PCR, and the mutations were confirmed by sequencing.

Immunofluorescence and confocal microscopy

PC-3U cells were seeded with a cell density of 0.5 × 105 cells/cm2, grown on coverslips, and starved in 1% fetal calf serum (FCS) for 18 hours. Then, they were stimulated with TGFβ for the indicated times. Next, they were washed once in PBS, fixed in 4% formaldehyde, permeabilized with 2% Triton X-100, and then blocked in 5% bovine serum albumin. Incubations with antibodies to TβRI (V22), TRAF6, HA, and PS1 were performed for 1 hour. Then, secondary antibodies Alexa Fluor 555 and 488 (Invitrogen) were added. The preparations were mounted with coverslips in mounting medium with DAPI according to the manufacturer’s recommendations. The slides were analyzed with a Zeiss LSM 710 confocal microscope equipped with a 63× lens (numerical aperture 1.4). Photomicrographs were obtained with a Zeiss 710 Meta (Carl Zeiss MicroImaging) equipped with a digital camera (RET-EXi-F-M-12-C, QImaging) and visualized with ZEN software. The specificities of primary TRAF6 and TβRI antibodies and of all secondary antibodies were determined, and no background staining was observed. The number of cells with TβRI in the nucleus was determined with confocal imaging, and these numbers were expressed as percentages of a total of 200 to 350 cells counted in each group. Representative results are shown from experiments repeated at least three times.

Ubiquitination assay

Cell lysates were heated in the presence of 1% SDS to disrupt noncovalent protein-protein interactions, then diluted with lysis buffer containing 0.5% NP-40. PS1-NTF was immunoprecipitated, subjected to immunoblotting, and probed with an antiserum to Lys63-linked polyubiquitin, as previously described (9).

RNA isolation and RT-PCR

Total RNA was isolated from cells with an RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. Purified total RNA (2 μg) served as a template for cDNA synthesis, which was performed with the ThermoScript RT-PCR system (Invitrogen) according to the manufacturer’s instructions.

Quantitative real-time PCR

Purified cDNA (2 μg) was amplified and measured in duplicate with a Stratagene RT-PCR system, and SYBR Green (Applied Biosystems) was used for detection of PCR products. The following primers were used for qRT-PCR: TβRI, forward primer (FP) TGTTGGTACCCAAGGAAAGC, reverse primer (RP) CACTCTGTGGTTTGGAGCAA; TRAF6, FP TCCACACAATGCAAGGAGAA, RP GGGCTTCCAGATGCATAAAA; PS1, FP ACTTTTGCAGCTTCCTTCCA, RP TTGACCTCGTCCCTAAATC; Snail1, FP GAGCATACAGCCCCATCACT, RP GGGTCTGAAAGCTTGGACTG; Jag1, FP CAGACGCTGAAGCAGAACAC, RP TTTTGTTGCCATTCTGGTCA; and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), FP TGATGACATCAAGAAGGTGGTGAAG, RP TCCTTGGAGGCCATGTGGGCCAT. GAPDH was used as an internal control.

Chromatin immunoprecipitation

Five biological replicates of each ChIP were performed according to the ChIP protocol provided by Abcam. Chromatin was precipitated with the V22 rabbit antibody to TβRI-ICD (Santa Cruz Biotechnology) or with the rabbit polyclonal HA ChIP-grade antibody (Abcam). After purification, the ChIP DNA was amplified with qRT-PCR in triplicate with the following primers: TβRI, FP GGGAGCAGGAGGAAATAGGA, RP AGCACATGCCACCCAGAT.

In situ PLA

PC-3U cells were untreated or treated with TGFβ for 30 min, fixed, and then probed with the primary antibodies indicated. The antibodies were raised in different species and directed to the ICDs of TβRI (V22) and PS-NTF. Next, the cells were blocked and incubated with two sets of secondary antibodies conjugated with tag sequences, according to the Duolink Detection Kit instructions (Olink Bioscience). Tag sequences were ligated with a bridging probe in a proximity-dependent manner, which allowed rolling-circle amplification. Finally, the product was visualized with complementary fluorescent probes.

Slides were mounted with Duolink mounting medium and evaluated with a confocal microscope (Carl Zeiss). Z-stack micrographs were acquired with 40× or 63× objectives. The number of heterodimers, visualized as bright fluorescent signals, was counted in 10 to 15 fields per well. Representative results are shown from experiments repeated at least three times. Cell images were exported in TIF format with AxioVision software (Carl Zeiss) for further analysis, and the number of heterodimers per cell was determined with Blob-Finder image analysis software (version 2.5), which was developed by the Center for Image Analysis, Uppsala University, Sweden.

Protein quantification

Protein bands were quantified with Bio-Rad Quantity One software. Quantifications were based on three or more independent experiments.

Invasion assays

We performed invasion assays with the CytoSelect Cell Invasion assay kit (Cell Biolabs). The upper chamber was rehydrated with RPMI 1640, 1% FCS, l-glutamine, and PEST for 30 min at room temperature. A suspension of 2 × 106 cells was seeded into the upper chamber in serum-free RPMI 1640, with or without TGFβ. The lower chamber was filled with 500 μl of RPMI 1640 supplemented with 10% FCS, 1% l-glutamine, and PEST. Noninvasive cells were removed from the upper chamber, and invasive cells in the lower chamber were stained with crystal violet cell staining solution. Photographs were taken with a Leica DMR light microscope. We performed calorimetric quantification of samples by collecting 200 μl of the solution from each sample (in duplicate) and reading the absorbance at 560 nm (A560). The A560 values were used for statistical analysis.

Mouse experiments

Ten-week-old C57BL/6 male mice used in the xenograft model were purchased from Taconic Europe A/S and were maintained at the animal facility of Umeå University. All experiments involving animals were approved by the local Animal Review Board (Umeå, Sweden) (approval ID: A110-12, date: 21 August 2012). Mice were injected subcutaneously with TRAMPC2 cells (4.2 × 105/ml) and were monitored up to 6 weeks, until a palpable tumor volume was reached. Initial tumor measurements were taken, and the mice were then injected with vehicle (DMSO; Sigma-Aldrich) or with the γ-secretase inhibitor DBZ (SYNCOM BV) (48). Both vehicle and DBZ (10 μmol/kg) were administered by intraperitoneal injection once a day for 10 days. Tumor volumes and body weights were measured twice a week, and physical activity was monitored. After completing the treatment, the mice were sacrificed according to ethical guidelines, and tumors were collected and weighed. A small portion of tumor tissue was used for protein and RNA extraction. The remaining tumors were fixed in formalin and embedded in paraffin. Tumor morphology was examined by light microscopy.

Protein and RNA analysis

Tumor protein and RNA were extracted from tumor tissue (20 mg) with the Qiagen Protein/DNA and RNA extraction kit according to the manufacturer’s instructions. The extracted protein (10 μl) was incubated in 1 ml of BCA solution (Thermo Scientific) at 37°C for 30 min, and the protein concentration was measured at 560 nm with a spectrophotometer. Equal amounts of proteins were loaded onto an SDS-PAGE gel for electrophoresis. β-Actin was used as an internal loading control. The proteins were transferred onto blots and probed with antibodies to Snail1 (Novus Biologicals), TβRI (Santa Cruz Biotechnology), and β-actin (Sigma-Aldrich).

The extracted tumor RNA served as a template for cDNA synthesis with the ThermoScript DNA synthesis kit (Invitrogen) according to the manufacturer’s instructions. Next, qRT-PCR was performed with the SYBR Green Master Mix (Applied Biosystems) and primers specific for Snail1, TβRI, Jag1, and GAPDH. The primers were synthesized with Primer3 software; their sequences are available upon request. The GAPDH sequence was used as an internal control. Experiments were repeated three or more times.

Statistical analysis

Tumor measurements (length, width) were performed with digital calipers. Tumor volume (TV) was calculated from the formula TV = (length × width2)/2. Tumor weights represent the mean for a particular group, with n = 8 or n = 6 in each group.

Unless otherwise indicated, quantitative data are presented as means ± SEM from at least three independent experiments. P values <0.05 were considered to be statistically significant. The data that are presented as either fold change or a percentage of the total were log-transformed before statistical differences between samples or treatment groups were examined. Values represent the means of three or more independent samples run on immunoblots, repeated three times.

After qRT-PCR, the cycle threshold (CT) values of respective genes were normalized to the CT of GAPDH. Then, the ΔCT values were obtained and plotted. This analysis was performed for each of three or more independent qRT-PCR assays.

Differences in the means ± SEM between samples or groups were analyzed with the Mann-Whitney U test (two-tailed), with the exception of Fig. 5G, where a one-tailed Mann-Whitney test was used. IBM SPSS Statistics 20 software was used throughout.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/7/307/ra2/DC1

Fig. S1. TGFβ promotes endoproteolytic cleavage of PS1.

Fig. S2. Schematic overview of the PLA.

Fig. S3. TGFβ promotes Lys63-dependent polyubiquitination of PS1 in cells.

Fig. S4. Identification of the PS1 cleavage site in TβRI.

Fig. S5. TGFβ promotes the colocalization of ectopically expressed TβRI and PS1, and treatment with monensin prevents activation of the γ-secretase complex and nuclear translocation of TβRI-ICD.

Fig. S6. TGFβ-induced invasive behavior of TRAMPC2 cells in vitro.

Fig. S7. Effects of a γ-secretase inhibitor on a xenograft prostate cancer model in vivo.

Fig. S8. Proposed simplified model of TGFβ-induced and TRAF6-dependent generation of TβRI-ICD.

REFERENCES AND NOTES

Acknowledgments: We are grateful to I. Dikic, V. M. Dixit, H. Hedman, R. Kopan, U. Lendahl, K. Miyazono, A. Moustakas, S. Souchelnytskyi, B. van der Strooper, and P. ten Dijke for their kindness in providing expression vectors, reagents, cells, and access to the confocal microscope. We used the confocal microscope at the Department of Oncology, Umeå University. We thank S. Kilter for expert technical assistance with the animal experiments. Funding: This work was supported by grants to M.L. from the Swedish Medical Research Council (K2010-67X-15284-01-3), the Swedish Cancer Society (100303), the Torsten and Ragnar Söderberg Foundation (MT29/09), ALF-VLL-224051, the Kempe Foundation (SMK.1132), the Knut and Alice Wallenberg Foundation (2012.0090), the Cancer Research Foundation in Northern Sweden, Lion’s Cancer Research Foundation, and Umeå University. Author contributions: All authors analyzed and discussed the data. S.K.G. carried out most of the experiments and prepared all the final figures for the paper. R.S. and Y.M. contributed to several experiments and helped with preparation of the figures. S.K.G. carried out the animal experiments together with G.Z. A.W. constructed the C-terminal–tagged HA-TβRI-ICD. S.K.G., A.B., C.-H.H., and M.L. wrote the manuscript. A.B., C.-H.H., and M.L. supervised the study. M.L. coordinated the study. Competing interests: A patent on the cleavage of the TβRI has been filed by Ludwig Institute for Cancer Research Ltd., and Y.M., R.S., S.K.G., C.-H.H., and M.L. The other authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article