Pentabromophenol suppresses TGF-β signaling by accelerating degradation of type II TGF-β receptors via caveolae-mediated endocytosis

Pentabromophenol (PBP), a brominated flame retardant (BFR), is widely used in various consumer products. BFRs exert adverse health effects such as neurotoxic and endocrine-disrupting effects. In this study, we found that PBP suppressed TGF-β response by accelerating the turnover rate of TGF-β receptors. PBP suppressed TGF-β-mediated cell migration, PAI-1 promoter-driven reporter gene activation, and Smad2/3 phosphorylation in various cell types. Furthermore, PBP abolished TGF-β-mediated repression of E-cadherin expression, in addition to the induction of vimentin expression and N-cadherin and fibronectin upregulation, thus blocking TGF-β-induced epithelial–mesenchymal transition in A549 and NMuMG cells. However, this inhibition was not observed with other congeners such as tribromophenol and triiodophenol. TGF-β superfamily members play key roles in regulating various biological processes including cell proliferation and migration as well as cancer development and progression. The results of this in vitro study provide a basis for studies on the detailed relationship between PBP and modulation of TGF-β signalling. Because PBP is similar to other BFRs such as polybrominated diphenyl ethers (PBDEs), additional laboratory and mechanistic studies should be performed to examine BFRs as potential risk factors for tumorigenesis and other TGF-β-related diseases.

Brominated flame retardant (BFR) phenols include pentabromophenol (PBP), 2,4,6-tribromophenol (TBP), 2,4-dibromophenol, and tetrabrominated bisphenol (TBBP). PBP, TBP, and TBBP are precursors of four nonphenolic derivatives that are also used as BFRs 1 . PBP and TBP are used for developing epoxy resins and vinyl aromatic polymers and as intermediates of polyester resins 2 . BFRs and their metabolites induce potential endocrine-disrupting effects in humans and animals 3 , in addition to being detected in human milk and blood 4 . BFRs are one of the most widely used but least understood organohalogen compounds. Molecular mechanisms underlying the toxic effects of BFRs are largely unknown. In vitro studies have shown that PBP and TBP and their brominated phenol congeners interact with transthyretin, a human thyroxine transport protein, competing with thyroid hormone thyroxine or with oestrogen on oestrogen receptors [5][6][7] . An in vitro study also revealed that TBP markedly enhanced aromatase activity, whereas 6-OH-BDE99 and 6-OH-BDE47 considerably reduced aromatase activity 8 . In the present study, we determined that PBP suppressed transforming growth factor-beta (TGF-β ) signalling by accelerating TGF-β receptor degradation through caveolae-mediated endocytosis.
TGF-β superfamily proteins, including bone morphogenetic proteins, inhibins, activins, and TGF-β , regulate many physiological processes such as cell proliferation, development, and differentiation. Dysregulation of these proteins is associated with cancer development, vascular diseases, and fibrosis [9][10][11] . In a canonical pathway, binding of TGF-β to TGF-β receptors induces the assembly of type I and II TGF-β receptors (Tβ RI and Tβ RII, respectively) on the plasma membrane into heteromeric complexes for transducing signals to intracellular molecules
Cell surface TGF-β receptor biotinylation and endocytosis assays. Surface biotinylation was performed at 0 °C using 0.2 mM Sulfo-NHS-SS-biotin (ThermoFisher) according published procedures 32 . Biotinylated cell lysates were analyzed by 10% SDS-PAGE followed by immunoblotting analysis and quantification using ImageQuant. Mv1Lu cells grown to 90% confluence on 6-well cluster plates were treated with PBP for different time periods at 37 °C. After treatment, cells were washed with cold PBS and incubated with 0.2 mM Sulfo-NHS-SS-biotin for 30 min. Biotinylated cells were washed with TBS and the cells then were lysed in lysis buffer and incubated with streptavidin beads for 1 h at 4 °C. Strptavidin-precipitated Tβ RII protein was detected using immunoblotting. The biotinylated Tβ RII remaining on the cell surface should be compared to the total Tβ RII level before biotinylation.

Analysis of lipid raft/caveolae and non-lipid raft microdomains.
To separate and analyze the membrane microdomains, we performed sucrose density gradient ultracentrifugation according published procedures 33 without any modification. Mv1Lu were grown on 100 mm dishes (5 × 10 6 cells per dish). Cells were then incubated with or without 5 μ M PBP in low serum (0.1% FBS) DMEM at 37 °C for the time indicated 18 . After two washes with ice cold phosphate-buffered saline, cells were scraped into 0.85 ml of 500 mM sodium carbonate, Scientific RepoRts | 7:43206 | DOI: 10.1038/srep43206 pH 11.0. Homogenization was carried out by three 15-second bursts of an ultrasonic disintegrator (Qsonica, Newtown, CT, USA) to disrupt cell membranes, as described previously 18 . The homogenates were adjusted to 45% sucrose by addition of 0.85 ml of 90% sucrose in 25 mM 2-(N-morpholino) ethanesulfonic acid, pH 6.5, 0.15 M NaCl (MBS), and placed at the bottom of an ultracentrifuge tube. A discontinuous sucrose gradient was generated by overlaying 1.7 ml of 35% sucrose and 1.7 ml of 5% sucrose in MBS on the top of the 45% sucrose solution, and it was then centrifuged at 40,000 rpm for 16-20 h in an SW55 TI rotor. Ten 0.5-ml fractions were collected from the top of the tube, and a portion of each fraction was analyzed by immunoblotting using antibodies against Tβ RII. The relative amounts of Tβ RII on the blot were quantified by densitometry. Fractions 4-5, and fractions 7 to 10 contained flottlin-2 and EEA-1, respectively 18,33 . Immunoblotting analysis analysis. Cell lysates (~50 μ g protein) were subjected to 7.0%, 10%, or 12.5% SDS-PAGE under reducing conditions and then electrotransferred to PVDF membranes. After being incubated with 5% nonfat milk in Tris-buffered saline plus Tween 20 (TBST) (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween 20) for 1 h at room temperature, the membranes were further incubated with specific polyclonal antibodies to Tβ R-I and Tβ R-II in TBST/non-fat milk at 4 °C for 20 h and washed three times with TBST for 10 min each. Bound antibodies were detected using peroxidase-conjugated anti-rabbit or anti-mouse IgG and visualized using the ECL system.

Immunofluorescent staining.
To determine the effect of PBP in TGF-β -induced EMT, cells on 24 mm round coverslips (Paul Marienfeld, Germany) were pretreated with or without 2 μ M PBP for 2 h in low serum DMEM (0.1% FBS), cells were then continuingly stimulated with TGF-β (100 pM) for 48 h. Treated cells were washed with phosphate buffered saline (PBS) and fixed in cold methanol for 10 min. After washings with PBS, cells were blocked with 5% goat serum (Dako) in 1% BSA/PBS. After incubation with rabbit anti-E-cadherin, anti-vimentin, anti-N-cadherin, and anti-fibronectin antibodies (1:200) in 1% BSA/PBS for 18 h at 4 °C, cells were incubated with donkey anti-rabbit-Alexa Fluor ® 488 at RT for 1 h. Coverslips were mounted with mounting medium containing DAPI (ThermoFisher). Photomicrographs were taken with a Zeiss Axio Observer Z1 microscope equipped with a Photometrics HQ2 camera.
To determine the effect of PBP in subcellular localization of Tβ RII, Mv1Lu cells grown on 24 mm round coverslip were transiently co-transfected with Tβ RII-HA and caveolin-1-GFP plasmids using Lipofectamin 2000 (ThermoFisher) according to the manufacturer's protocol. Twenty-four hours after transfection, cells were changed to low serum medium (0.1% FBS) and treated with PBP 5 μ M for the time indicated. After treatment, cells were fixed in 4% paraformaldehyde solution containing 0.1% Triton-X100 for 30 minutes, washed with PBS and then blocked by 0.2% gelatin in PBS for 1 h. Cells were incubated overnight at 4 °C in a humidified chamber with a goat anti-HA-probe (F-7; Santa Cruz Biotechnology) at 1:100 dilutions. After extensive washing, cells were incubated with Alexa Fluor ® 594-conjugated donkey anti-goat antibody at a 1:50 dilution for 1 h. Images were acquired using a Nikon TCS SP confocal microscope (Nikon Ltd., Tokyo, Japan). The measurements of co-localization rate were analyzed using a Nikon Application Suite.
Transcriptional response assay. The procedurals for transcriptional assay were performed in Mv1Lu or MLE cells according to our recent report 21,34 and are described concisely as follow. Mv1Lu cells were transiently transfected with CMV-β gal, and Fibro-luc 35 or COL1A2-luc 35 reporter plasmids using electroporation. In a similar experiment, MLE cells (Mv1Lu cells stably expression 3TP-luc promoter plasmid) were also used. Cells grown in low serum medium were incubated with several concentrations of PBP for 1 h follow by TGF-β treatment for 4 h. Fifty micro liter cell lysates (approximately 20 μ g of protein) were then used to measure both luciferase and β -gal activities. The luciferase activity was normalised and the increment of luciferase activity was calculated against the experimental controls 21 .
Scratch wound assay. The procedural for cell migration assay was descripted in our previous work 34 .
Briefly, A549 cells grown in 4-chambered 35-mm dish (95% confluency) were serum-starved in DMEM containing 0.1% FBS for 2 h prior to wounding to ensure that no proliferation occurred during the experiments. A scratch wound was created by using a 200 μ l pipette tip on cells monolayers. The wounded cells were immediately treated with TGF-β (100 pM) in the presence or absence of 5 μ M PBP for 15 h. Digital images of the cells that had migrated into the wound area were taken by an Axio Observer Z1 inverted microscope fitted with a K heating stage and incubator (Carl Zeiss Inc., Oberkochen, Germany).

Statistical Analysis.
All experiments were conducted in triplicate. All data were shown as the mean ± standard deviation (S.D.). We used Student's t test for the comparison between two groups, and used One-way ANOVA when we compared more than two groups. The means were considered significant if P < 0.05 (*) or P < 0.01 (**).

Results
To rule out the cytotoxicity effects mediated by PBP in this study, we performed the toxicity assays and cell viability assays by testing plasma membrane integrity and mitochondria functions (i.e., MTT assay). Acute toxicity of PBP was determined by measuring G6PDH leakage (data not shown), and the IC 50 of PBP on NMuMG cells and A549 cells are more than 30 μ M ( Figure S1). Therefore, the doses of PBP used were between 1 μ M and 5 μ M in subsequent experiments.
The TGF-β-induced Smad phosphorylation and promoter activation were suppressed by PBP in cells. Tβ RI and Tβ RII are expressed in all normal cells, but they are not expressed in some cancer cells [9][10][11] . In the presence of TGF-β , Tβ RI and Tβ RII form a hetero-oligomeric complex that activates canonical (Smad-dependent) and noncanonical (Smad-independent) TGF-β signalling and is crucial for many cellular Scientific RepoRts | 7:43206 | DOI: 10.1038/srep43206 processes including cell growth, apoptosis, differentiation, extracellular matrix production, and EMT 9,11,13 . Mv1Lu cells have been widely used as a model for studying TGF-β signalling and relative cell responses [16][17][18][19] . Mv1Lu cells expressing 3TP-Lux luciferase promotor (termed as MLE cells) were used to evaluate the inhibitory effects of PBP in TGF-induced cellular responses. 3TP-Lux contains three consecutive TPA response elements (TREs) and a portion of the PAI-1 promoter region. In Fig. 1A, TGF-β stimulation resulted in a fivefold increase of luciferase activity in MLE cells harbouring the 3TP-Lux luciferase promoter. PBP attenuated TGF-β -induced luciferase activity in a dose-dependent manner, with the IC 50 value of PBP being approximately 3 μ M and the maximum inhibition being achieved using 10 μ M PBP (Fig. 1A, black columns). However, a structurally related congener of PBP, such as 2,4,6-triiodophenol (TIP), did not considerably affect TGF-β -stimulated PAI-1 promoter activity in MLE cells (Fig. 1A, grey columns). In addition to PAI-1 gene, collagen type I, α 2 and fibronectin genes are also important targets for canonical TGF-β signalling [28][29][30] . Mv1Lu transiently expressing COL1A2-luc or Fibro-luc were used to determine the effects of PBP on TGF-β signalling and β -galactosidase expression serving as an internal control. Figure 1B shows that TGF-β -induced transcription of collagen and fibronectin were inhibited by PBP in a dose-dependent manner. To further determine the specific target of PBP on canonical (Smad-dependent) TGF-β signalling, we performed immunoblotting to observe the levels of phosphorylated Smad2 in Mv1Lu cells treated with PBP in the presence of TGF-β . Smad2/3 proteins are the major signal transducers of TGF-β signalling. TGF-β stimulation activates Smad2/3 by phosphorylation at their C-terminal serine residues through Tβ RI-Tβ RII receptor complexes. Next, phosphorylated Smad2/3 complexes with Smad4 migrate into the nucleus and activate various target genes. In this study, Mv1Lu cells were pretreated with increasing concentrations (0 to 5 μ M) (Fig. 1C) or single concentration (5 μ M) (Fig. 1D) of PBP for 6 h, followed by TGF-β stimulation for 30 min. TGF-β treatment strongly stimulated Smad2 phosphorylation in Mv1Lu cells; however, PBP pretreatment inhibited TGF-β -induced Smad2 phosphorylation in a dose-dependent manner in Mv1Lu cells (Fig. 1C), with the IC 50 value of PBP being approximately 1.5 μ M and the maximum inhibition being achieved using 5 μ M PBP (Fig. 1C, luciferase promoter plasmids and then analysed by performing the luciferase assay. (C and D, right graph for quantification) Mv1Lu cells were treated with PBP for 6 h, and cell lysates were resolved by performing immunoblotting analysis to assess Smad2 phosphorylation. Smad2/3 served as an internal control. All experiments were repeated three times, and data are expressed as mean ± SD. Dual asterisks indicate significant differences (P < 0.01), as determined using one-way analysis of variance with SPSS statistical software. lanes 7-12, right graph for quantification). The level of phosphorylated Smad2 increased in a dose-dependent manner in the Mv1Lu cells treated with increasing concentrations of TGF-β (Fig. 1D, lanes 1-6); nevertheless, PBP treatment inhibited the Smad2 phosphorylation induced by all concentrations of TGF-β used (Fig. 1D, lanes 8-12, right graph for quantification). This finding was also validated in NMuMG cells, suggesting that PBP inhibits TGF-β -induced Smad phosphorylation, regardless of the cell type (Supplemental data, Figure S2). PBP attenuates TGF-β-induced EMT. TGF-β -Smad signalling strongly induces EMT 36 . NMuMG and A549 cells have been extensively used as in vitro models for studying EMT, and these cells undergo EMT discernible at 40 h after TGF-β stimulation 37,38 . To understand whether PBP could suppress TGF-β -induced EMT, expression of EMT markers including fibronectin, vimentin, N-cadherin, and E-cadherin in A549 and NMuMG cells were evaluated by immunoblotting analysis and immunofluorescence staining. EMT is characterised by E-cadherin disruption from cell junctions and by increased fibronectin, N-cadherin, and vimentin expression 39 . In immunoblotting analysis, the A549 and NMuMG cells were pre-treated with increasing concentrations ( Fig. 2A (Fig. 2B, lanes 7-12). In NMuMG cells, PBP inhibited TGF-β -stimulated fibronectin, N-cadherin, vimentin, and PAI-1 protein expression in a dose-dependent manner (Fig. 2C, lanes 4-6); conversely, PBP treatment slightly reversed the TGF-β -induced E-cadherin disruption by 18% (Fig. 2C, lane 4 versus lane 6). In addition, 2 μ M PBP inhibited TGF-β -induced fibronectin, N-cadherin, vimentin, and PAI-1 protein by more than 90% (Fig. 2D, lanes 4 to 6). Consistent with the immunoblotting results, the immunofluorescence staining results revealed that PBP reversed TGF-β -induced suppression of E-cadherin expression ( Fig. 3Al versus 3Ak) and reduced TGF-β -stimulated induction of fibronectin, N-cadherin, and vimentin expression ( Fig. 3Bl versus 3Bk, 3Cl versus 3Ck, and 3Dl versus 3Dk). Taken together with prior results in Fig. 1, PBP could suppress TGF-β -induced Smad phosphorylation, and causing the inhibition of EMT.

PBP inhibits TGF-β-induced cell migration.
In addition to inducing EMT in epithelial cells, TGF-β plays a crucial role in promoting cancer cell migration and invasion via a Smad-dependent pathway 39,40 . Inhibition of Tβ R-I with SB431542 has been shown to inhibit the function of TGF-β in cell migration 41 . To test if PBP inhibited TGF-β -stimulated cell migration, we determined the effect of PBP on TGF-β -induced increases in cell motility by performing a wound healing assay, as described previously by Lamouille et al. 39,42 . We observed that TGF-β -stimulated migration of A549 cells by inducing exhibited > 95% wound closure (Fig. 4Ag)   versus 4Ah, and 4B). In the experiment with PBP alone, PBP reduced the percentage of wound closure from 55% to 42% (Fig. 4Ae versus 4Af, and 4B). This result indicates that PBP suppresses TGF-β -induced cell migration. It is worth noting that A549 cell is responsive to TGF-β in both cell growth and wound healing. Our [ 3 H]-Thymidine incorporation assays and cell counting results (data not shown) show that A549 cells are growth-inhibited by approximately 50% and 35%, respectively. These suggest that proliferation is not involved in the migration of A549 cells induced by TGF-β . Furthermore, the results of MTT assay ( Figure S1C) show that 10 μ M PBP enhances cell viability by 40%, this suggest that the migratory inhibition of PBP is not due to cytotoxicity.
PBP accelerates the internalisation of TβRII and results in its rapid degradation. In the preceding sections, PBP attenuated TGF-β -stimulated cellular response including reporter gene activation, Smad2 phosphorylation, and EMT. These findings prompted us to investigate the detailed mechanism underlying the inhibitory effect of PBP on TGF-β . We conjectured that PBP may reduce TGF-β activity by increasing the endocytosis and degradation of TGF-β receptors. To test this conjecture, we examined the effect of PBP on the expression of TGF-β receptors on the surface of Mv1Lu cells by performing cell surface biotinylation. The Mv1Lu cells were pretreated with 5 μ M PBP for 0-2 h. At the indicated time, the cells were cooled rapidly, and proteins expressed on the surface of these cells were biotinylated. Biotinylated Tβ RII was pulled down by using streptavidin-Sepharose beads and was examined through immunoblotting. To determine whether PBP altered TGF-β receptor stability, we performed a parallel experiment by measuring the total receptor protein levels in the lysates of the cells treated with PBP. As expected, PBP treatment reduced Tβ RII protein levels both on the cell surface and in the cell lysates in a time-dependent manner (Fig. 5A). The reduction of Tβ RII in cell surface was started at 15-30 min (Fig. 5A,   Figure 3. PBP attenuated TGF-β-induced EMT in A549 cells. EMT was determined by immunostaining for epithelial marker E-cadherin (A), ECM protein fibronectin (B), and mesenchymal markers N-cadherin and vimentin (C and D, respectively). A549 cells cultured on a cover glass were treated with TGF-β (100 pM) in 0.1% FCS in the presence or absence of PBP (5 μ M) for 48 h. Cells were fixed with 4% paraformaldehyde and then incubated with primary antibodies against E-cadherin, fibronectin, N-cadherin, and vimentin. Fluorescence signals were visualised using Alexa Fluor 488-conjugated secondary antibodies. Nuclei of the cells were stained with DAPI. Scale bar = 200 μ m. lanes 2 and 3) and started from 60-120 min for total lysates (Fig. 5A, lanes 4 and 5). Since the PBP-induced disappearance of Tβ RII in cell surface was faster than in whole cell lysates, which suggest that PBP-induced Tβ RII internalisation is prior to its degradation. However, PBP treatment did not alter the mRNA levels of Tβ RII (Supplemental Data Figure S3). These results signify that PBP may reduce Tβ RII stability. To assess the effect of PBP on Tβ RII stability, we monitored Tβ RII turnover after the impeding of protein synthesis by cycloheximide and found that PBP reduced the half-life of Tβ RII in the Mv1Lu cells (Fig. 5B). To further confirm that PBP accelerates Tβ RII turnover, Mv1Lu cells expressing Tβ R-II-flag were treated with 5 μ M PBP for increasing time period or with increasing concentration of PBP for 4 h and were further detected by immunoblotting with the anti-flag antibody. As shown in Fig. 5C and D, PBP treatment enhanced Tβ R-II-flag degradation in both timeand dose-dependent manners.
Because PBP enhances Tβ RII turnover, and it has been recognized that Tβ RII turnover is dynamically regulated by clathrin vesicle-mediated ligand-triggered trafficking, recycling, and lysosome degradation, as well as caveolae vesicle-mediated proteasomal degradation 19 . We used lysosomal inhibitor NH 4 Cl and proteasome inhibitor MG132 to determine the pathways involved in Tβ RII degradation. Our results showed that MG132 (but not NH 4 Cl) reversed PBP-induced Tβ RII degradation ( Fig. 6A and C for quantification), signifying that proteasome-dependent degradation was primarily involved in PBP-induced Tβ RII degradation. Notably, PBP induced Tβ RII degradation without altering the EGFR, Tβ RI, and Cav-1 levels ( Fig. 6A and B for quantification). Because Tβ RII was targeted to the proteasome, we examined ubiquitination of Tβ RII but found no evidence of mono-or polyubiquitination ( Figure S5). Taken together, the PBP class of molecules comprises selective TGF-β inhibitors that function by diverting Tβ RII to the proteasome through an ubiquitin-independent mechanism.
Lipid rafts/Caveolae are essential for PBP-induced TβRII degradation. Tβ RII is internalised through both caveolae-and clathrin-mediated endocytosis 32,43 , and caveolae-mediated endocytosis attenuates TGF-β signalling by promoting Tβ RII degradation. These two endocytic pathways are maintained in a dynamic balance and the inhibition of one these pathways leads to the promotion of the other pathway 19,32 . Methyl-β -cyclodextrin (Mβ CD) and trifluoperazine (TFP) were used to inhibit lipid raft/caveolae-and clathrin-mediated endocytosis, respectively 27 . We observed that PBP induced Tβ RII internalisation and degradation mainly through lipid raft-/caveolae-mediated endocytosis and that PBP-induced Tβ RII internalisation and degradation was inhibited by Mβ CD, rather than TFP ( Fig. 6A and D for quantification). Consistent with this finding, treatment with Mβ CD, a cholesterol chelator and lipid raft disruptor, reversed PBP-inhibited TGF-β signalling including Smad2/3 phosphorylation (Fig. 7A, lower graph for quantification) and PAI-1 promoter activation (Fig. 7B). By contrast, treatment with TFP, the inhibitor of clathrin-mediated endocytosis, did not reverse the PBP-inhibited TGF-β signalling (data not shown). To define the chronologic sequence of Tβ RII internalisation and degradation after PBP treatment, we treated cells with PBP and/or Mβ CD follow by cell surface biotinylation. If PBP-induced Tβ RII degradation is secondary to its internalisation, the inhibitors of caveolae-mediated endocytosis will alleviate PBP-induced Tβ RII internalisation and degradation. Figure 7C indicates that Mβ CD, a caveolae disruptor, not only retain Tβ RII in the cell surface but also inhibits Tβ RII degradation. Echoing with prior result in Fig. 5A, this result also suggests that the proceeding of Tβ RII internalisation is prior to degradation.

PBP enhance TβRII internalisation and degradation via caveolae-mediated endocytosis.
Previous studies have suggested that lipid rafts/caveolae induce proteasome-mediated degradation of Tβ RII in the absence of a ligand 26 . Therefore, we examined whether PBP-induced Tβ RII degradation was dependent on lipid rafts/caveolae. In this study, caveolin-1 and flotillin-2 were used as markers for lipid-raft/caveolae. Flotillins are topologically similar but unrelated in sequence to caveolins 44 . In fact, they were thought to be present in caveolae 45 or to substitute for caveolae in cell types or tissues, such as leukocytes, which lack detectable caveolin-1 46 . Immunostaining assay results revealed that the overexpressed Tβ RII-HA was located on the cell surface and in cytoplasm. PBP treatment for 2 h markedly reduced the levels of Tβ RII-HA on the cell surface (Fig. 8Af versus Ac) and increased caveolin-1-GFP and Tβ RII-HA colocalisation (Fig. 8Af, as indicated by arrowheads). To corroborate these observations regarding PBP-induced Tβ RII translocation, we examined the effect of PBP on the subcellular localisation and degradation of Tβ RII in the Mv1Lu cells by performing sucrose gradient ultracentrifugation. In Fig. 8B, the results showed that Tβ RII was distributed in both lipid raft and non-lipid raft fractions in the control experiment (0 h); in the first hour after PBP treatment, Tβ RII in non-lipid raft fractions (fractions 7 to 10) not only slightly decreased, but also shifted to lipid-raft fractions (fraction 4 and 5) in the plasma membrane (marked with a red star), and it continued to turnover in prolonged treatment (2 h and 4 h). Conversely, PBP treatment induced neither translocation nor degradation of Tβ RI, EGFR, and caveolin-1 in this study. To further define the degradation route for PBP-induced Tβ RII turnover, we performed density gradient fractionation to determine the effects of inhibitors in PBP-induced Tβ RII translocation and degradation. In Fig. 9A, Tβ RII which found primarily in the lipid-raft fractions of Mv1Lu cells in control experiment and 4 hours PBP treatment induced Tβ RII degradation (Fig. 9A, denote as ▴ ). Mβ CD, a lipid-raft/caveolae disruptor, not only reversed PBP-induced Tβ RII degradation in lipid-raft but also moved the Tβ RII from lipid-raft to non-lipid raft fraction (Fig. 9A, denote as *, right graph for quantification). We also test whether clathrin-mediated endocytosis, another endocytic pathway for TGF-β receptor could confer PBP-induced Tβ RII turnover. In Fig. 9B, TFP (trifluoperazine), an inhibitor of clathrin-mediated endocytosis/recycling/lysosome route for Tβ RII, did not reverse Tβ RII turnover in any of the fractions (Fig. 9B, denote as ▴ ). In Fig. 9C, we use NH 4 Cl, a weak base that blocks lysosomal degradation by neutralizing proton accumulation in the process of lysosome maturation. NH 4 Cl does not prevent PBP-induced Tβ RII degradation in lipid-raft (Fig. 9C, denote as ▴ ). However, inhibition of lysosome maturation by NH 4 Cl treatment may cause accumulation of Tβ RII in pre-lysosomal compartments in high density fractions (Fig. 9C, denote as #) and slightly retard Tβ RII from PBP-induced degradation. It is noteworthy that only Mβ CD alter caveolin-1 partitioning between lipid-raft and non-lipid raft, which indicates that Mβ CD wreck caveolae and obstruct its function (Fig. 9A). Consistent with the preceding results (Figs 6 and 7C) of the present study, caveolae-mediated endocytosis inhibitor (Mβ CD) abolished PBP-induced Tβ RII degradation but not TFP and NH 4 Cl. These results suggest that PBP-induced Tβ RII degradation is through caveolae-mediated endocytosis. Graphs represent mean ± SD densitometry data from three independent experiments. (B) PBP induced the rapid degradation of Tβ RII but did not exert any effect on Tβ RI, EGFR, and caveolin-1. Dual asterisks indicate significant differences (P < 0.01) in comparisons between Tβ RII and Tβ RI and EGFR. (C) MG132, a proteasome inhibitor (but not NH 4 Cl, a lysosome inhibitor), abolished PBP-induced Tβ RII degradation. (D) Mβ CD, an inhibitor of caveolae-mediated endocytosis (but not TFP, an inhibitor of clathrin-mediated endocytosis), abolished PBP-induced Tβ RII degradation.

Discussion
In this study, the inhibitory effects of PBP on TGF-β signalling were characterised using Mv1Lu, A549, and NMuMG cells. This study also investigated the ability of PBP to induce the internalisation and turnover of Tβ RII; inhibit the migration of cells; and affect the expression of TGF-β -regulated proteins such as PAI-1, fibronectin, N-cadherin, vimentin, and E-cadherin. PBP is one of the most frequently used BFRs, extensively employed as an additive in resins and polyester polymers for improving their fire resistance. Other classes of BFRs, such as brominated bisphenols, may break down into PBP, which has higher bioavailability. Detailed information about the potential mechanisms underlying the biological and toxic effects of PBP is scarce. The results of the present study demonstrate, for the first time, that PBP inhibits TGF-β signalling by increases the clearance rate of Tβ RII from the cell surface, and by accelerating their turnover. The results also confirm our hypothesis that PBP promotes caveolae-mediated endocytosis of cell surface Tβ RII, resulting in the degradation of Tβ RII and subsequent termination of the signalling of TGF-β . These results are corroborated by the following findings. First, PBP inhibited all TGF-β responses examined in this study including Smad2 phosphorylation, PAI-1 promoter activation, EMT, and cell migration. Second, PBP treatment for 30 min reduced Tβ RII expression levels on the surface of the Mv1Lu cells by 58%, as determined by performing cell surface biotinylation (Fig. 5A). This reduction in cell surface Tβ RII expression is concurrent with drops in the total Tβ RII protein levels and TGF-β -induced cellular responses, suggesting that PBP suppresses TGF-β signalling by inducing the rapid internalisation and degradation of Tβ RII. The halogenated phenol 2,4,6-triiodophenol (TIP) is an analogue of PBP and has been used in this study. A recent series of experimental binding and computational studies have suggested that the TIP as an inhibitor of the ATPase activity of myosin VI 47 . Live cell image studies also suggested that TIP inhibits myosin VI-mediated vesicle secretion/recycling, with an IC 50 of approximately 2 μ M which is similar to PBP in this study 47,48 . However, TIP did not affect TGF-β signalling (Fig. 1A), which further implicate that PBP might sequester Tβ RII from cell surface by promoting Tβ RII internalization rather than inhibition of recycling. Therefore, additional studies will be necessary to characterize the binding sites and mechanism of PBP inhibition of the Tβ RII.
Ligand binding triggers TGF-β receptor endocytosis through clathrin-and caveolae-mediated pathways 19,[49][50][51][52][53] . Clathrin-mediated endocytosis transfers receptors into an early endosome. Such internalised receptors are then either recycled to the cell surface or sent to the lysosomes for degradation. Caveolae-mediated endocytosis involving lipid rafts is a crucial trafficking pathway for TGF-β receptor internalisation and its ubiquitin-mediated degradation in the absence of a ligand 54 . Depletion of membrane cholesterol disrupts lipid rafts/caveolae, thus inhibiting caveolae-mediated endocytosis. Hence, we used clathrin-mediated endocytosis inhibitor TFP and cholesterol chelator Mβ CD to determine the endocytic pathway involved in PBP-induced Tβ RII degradation. Our results reveal that PBP-induced Tβ RII degradation was considerably blocked by Mβ CD, rather than TFP (Fig. 6). Mβ CD not only attenuated PBP-induced Tβ RII degradation but also reversed the inhibitory effect of PBP on TGF-β signalling including Smad2 phosphorylation and reporter gene activation ( Fig. 7A and B). The results from cell surface labeling and sucrose gradient fractionation reveal that Mβ CD not only prevents PBP-enhanced caveolae-mediated endocytosis of cell surface Tβ RII (Figs 7C and 9A), but also moved the Tβ RII from lipid-raft to were treated with and without 5 μ M PBP (panels Ad, Ae, and Af and panels Aa, Ab, and Ac, respectively) at 37 °C for 2 h. Cells were then analysed by performing indirect immunofluorescence staining with anti-HA (panels Aa and Ad) and anti-caveolin-1 antibodies (panels Ab and Ae). Merged staining is shown in panels Ac and Af. Before PBP treatment, Tβ RII-HA was primarily present on the plasma membrane and caveolin-1-GFP was primarily present in the cytoplasm of Mv1Lu cells. Arrowheads in the inset of panel Ac indicate Tβ RII-HA on the plasma membrane (red colour). PBP treatment reduced the levels of Tβ RII-HA on the cell surface and transferred Tβ RII-HA into caveolin-1-positive vesicles in the cytoplasm. Arrowheads in the inset of panel Af indicate the colocalisation (yellow colour) of Tβ RII-HA and caveolin-1 on the plasma membrane (panels Af); scale bar = 10 microns. (B) Mv1Lu cells were treated with 5 μ M PBP at 37 °C for 0, 1, 2, and 4 h. Localisation of Tβ RII, Tβ RI, EGFR, caveolin-1, flotillin-2, and EEA-1 (early endosome antigen 1) in lipid rafts/caveolae and non-lipid raft microdomains in cells treated and not treated (control) with PBP were determined by performing sucrose gradient ultracentrifugation followed by immunoblotting with antibodies against Tβ RII, Tβ RI, EGFR, EEA-1, flotillin-2, and caveolin-1. Fractions 4 and 5, which mainly contained caveolin-1, represent the location of lipid rafts/caveolae (Lipid raft). Fractions 7, 8, 9, and 10 which contained EEA-1, represent the location of non-lipid raft microdomains (Non-lipid raft). Non-lipid raft contains small amounts of caveolin-1. This is due to the presence of mitochondria in these fractions 31,33,59 ). The *symbol indicates the slightly increased amount of Tβ R-II in the fraction of cells treated with PBP for 2 h as compared with that in control cells. For longer treatments with PBP (2 h and 4 h), the closed arrow heads indicate the decreased amount of Tβ R-II in the fraction of PBP-treated cells as compared to that in control cells. The relative total amount of Tβ R-II in lipid rafts/ cavelolae and non-lipid raft microdomains in control experiment (0 h) were taken as 100% (black bar + grey bar in 0 h). For example, the relative amounts of Tβ R-II in lipid-rafts in cells treated with PBP for 0. 1. 2. and 4 h were estimated to be 22%, 38%, 3%, and 2%, respectively; the relative amounts of Tβ R-II in non-lipid-rafts in cells treated with PBP for 0. 1. 2. and 4 h were estimated to be 78%, 40%, 13%, and 9%, respectively. non-lipid raft fraction (Fig. 9A, denote as *). In contrast, TFP (trifluoperazine), an inhibitor of clathrin-mediated endocytosis/recycling /lysosome route for Tβ RII, neither changes Tβ RII localization nor reverses PBP-induced Tβ RII turnover (Figs 6A,D and 9B, denote as ▴ ). These results demonstrate that PBP induces Tβ RII degradation through caveolae-mediated endocytosis. Lysosomes are expected to degrade internalised proteins more efficiently at low pH levels because lysosomal hydrolysis typically requires acidic pH. Increasing lysosomal pH levels by adding weak bases such as NH 4 Cl and chloroquine can considerably reduce protein degradation in lysosomes 50 . However, our results reveal that PBP-induced Tβ RII degradation was attenuated after treatment with the proteasome inhibitor MG132 but not after treatment with the lysosomal inhibitor NH 4 Cl (Figs 6A,C and 9C, denote as ▴ ). In fact, NH 4 Cl, prevents lysosomal maturation by neutralizing proton accumulation and NH 4 Cl treatment causes accumulation of Tβ RII in pre-lysosomal compartments in high density fractions (Fig. 9C, denote as #) and slightly retards Tβ RII from PBP-induced degradation. Therefore, we conclude that PBP regulates the proteasomal degradation of TGF-β receptors through caveolae-mediated endocytosis. Previous studies described equal proteasomal degradation of both Tβ RI and Tβ RII through the ubiquitin-dependent (ubiquitin ligase Smurf2 mediated) 19 or ubiquitin-independent pathway which is exclusively for Tβ RII 55 . Our data on PBP meets the , and NH 4 Cl (C) at 37 °C for 4 h. Localisation of Tβ RII and caveolin-1 in lipid rafts/caveolae and non-lipid raft microdomains in cells treated and not treated (control) with PBP were determined by performing sucrose gradient ultracentrifugation followed by immunoblotting with antibodies against Tβ RII and caveolin-1. Representative of three experiments are shown. Fractions 4 and 5, which mainly contained caveolin-1, represent the location of lipid rafts/caveolae (lipid-raft). Fractions 7, 8, 9 and 10, which represent the location of non-lipid raft microdomains (Non lipid-raft). The closed arrow head indicates the decreased amount of Tβ R-II in the fraction of cells treated with PBP as compared with that in control cells. The * symbol indicates the increased amount of Tβ R-II in the fraction of Mβ CD-treated cells as compared to that in control cells. The #symbol indicates the increased amount of Tβ R-II in the fraction of NH 4 Cl-treated cells as compared to that in control cells. The relative amounts of Tβ R-II in the microdomains in treated cells were quantified by densitometry using caveolin-1 as an internal control. The relative total amount of Tβ R-II in lipid rafts/cavelolae (fractions 4 and 5, black bar) and non-lipid raft microdomains (fractions 7, 8, 9 and 10, gray bar) in control cells was taken as 100% (lipid-raft + non-lipid raft in control experiment). For example, the relative amounts of Tβ R-II in lipid rafts/caveolae (lipid raft) and non-lipid raft microdomains (non-lipid raft) in cells treated with PBP were estimated to be 4~8%, and 1~5%, respectively. The experiments in all three panels (Fig. 9A-C) were performed independently.
later mechanism since PBP induces Tβ RII degradation without changing Tβ RI, EGFR, and caveolin-1 level; and the process is ubiquitin-independent. To exclude the possibility that PBP enhances Tβ RII degradation via ubiquitin-dependent pathway, we have tried to detect PBP-induced ubiquitination signal in endogenous Tβ RII or overexpressed Tβ RII-Flag, no ubiquitination signal was detected ( Figure S5). Therefore, we suggest that PBP may possess a third mechanism of specific degradation exclusive for Tβ RII 55 . Wells et al. have shown different half-lives for Tβ RI and Tβ RII, which also echo with the concept that PBP induces distinct degradation mechanisms may exist to remove Tβ RII from cell surface 56 .
In addition to the inhibitory effects of PBP in Tβ RII turnover and TGF-β signaling, this study has elicited an important question about intracellular trafficking of TGF-β receptors and their degradation routes. It has long been recognized that Tβ RI and Tβ RII form hetero-complexes and co-translocated (or co-internalised) into intracellular compartments. However, we show that PBP selectively induces Tβ RII translocation and further degradation without affecting Tβ RI and other receptor such as EGFR (Figs 6A and 8B). In the future, additional studies should be performed to determine the targets of PBP in the endocytic machinery and Tβ RII degradation pathways. We will use SPR (Surface Plasma Resonance) to study interaction between PBP and Tβ RII or alternatively employ NMR to test PBP-Tβ RII interaction by observing the changes of 1 H and 13 C chemical shift. Although the direct target of PBP remains to be elucidated, it is possible that PBP directly binds Tβ RII to drive its internalization and degradation. It is also possible that PBP directs Tβ RII sorting by affecting its companion proteins follow by internalization. Interestingly, Tβ RII appears to be exclusively downregulated in several human cancers such as renal carcinomas and this reduction has been attributed to increased proteasomal degradation 55,57,58 . PBP might therefore be useful as a probe to understand how the altered dynamics of Tβ RII trafficking contributes to cancer.
Analyzing selective TGF-β -suppressing effects of PBDEs is outside the scope of the present study. However, there is an enormous body of evidence which demonstrates that the availability and function of Tβ RII is crucial determinants of TGF-β signaling and aberrant TGF-β responses are frequent in human diseases, such as cancer, fibrosis, inflammation, and cardiovascular disease. Therefore, the bioaccumulative and TGF-β -inhibitory properties of PBP observed in the present study suggest the potential effects of PBP and PBDEs in cancer development and TGF-β -relative diseases in vivo.
In conclusion, we found that PBP negatively regulated TGF-β signalling by enhancing Tβ RII degradation. The biochemistry approach revealed that PBP acts by stimulating clearance of Tβ RII from the cell surface through caveolae-mediated endocytosis and subsequent proteasomal degradation. However, additional in vivo studies are required to elucidate the potential targets and toxic effects of PBP. Considering these adverse effects of PBP, conducting a systemic assessment of the potential ecotoxic and biological effects of phenolic BFRs and relative compounds is imperative.