IGFBP‐3 and TGF‐β inhibit growth in epithelial cells by stimulating type V TGF‐β receptor (TβR‐V)‐mediated tumor suppressor signaling

Abstract The TGF‐β type V receptor (TβR‐V) mediates growth inhibition by IGFBP‐3 and TGF‐β in epithelial cells and loss of TβR‐V expression in these cells leads to development of carcinoma. The mechanisms by which TβR‐V mediates growth inhibition (tumor suppressor) signaling remain elusive. Previous studies revealed that IGFBP‐3 and TGF‐β inhibit growth in epithelial cells by stimulating TβR‐V‐mediated IRS‐1/2‐dependent activation and cytoplasm‐to‐nucleus translocation of IGFBP‐3‐ or TGF‐β‐stimulated protein phosphatase (PPase), resulting in dephosphorylation of pRb‐related proteins (p107, p130) or pRb, and growth arrest. To define the signaling, we characterized/identified the IGFBP‐3‐ and TGF‐β‐stimulated PPases in cell lysates and nucleus fractions in Mv1Lu cells treated with IGFBP‐3 and TGF‐β, using a cell‐free assay with 32P‐labeled casein as a substrate. Both IGFBP‐3‐ and TGF‐β‐stimulated PPase activities in cell lysates are abolished when cells are co‐treated with TGF‐β/IGFBP‐3 antagonist or RAP (LRP‐1/TβR‐V antagonist). However, the IGFBP‐3‐stimulated PPase activity, but not TGF‐β‐stimulated PPase activity, is sensitive to inhibition by okadaic acid (OA). In addition, OA or PP2Ac siRNA reverses IGFBP‐3 growth inhibition, but not TGF‐β growth inhibition, in Mv1Lu and 32D cells. These suggest that IGFBP‐3‐ and TGF‐β‐stimulated PPases are identical to PP2A and PP1, respectively. By Western blot/phosphorimager/immunofluorescence‐microscopy analyses, IGFBP‐3 and TGF‐β stimulate TβR‐V‐mediated IRS‐2‐dependent activation and cytoplasm‐to‐nucleus translocation of PP2Ac and PP1c, resulting in dephosphorylation of p130/p107 and pRb, respectively, and growth arrest. Small molecule TGF‐β enhancers, which potentiate TGF‐β growth inhibition by enhancing TβR‐I–TβR‐II‐mediated canonical signaling and thus activating TβR‐V‐mediated tumor suppressor signaling cascade (TβR‐V/IRS‐2/PP1/pRb), could be used to prevent and treat carcinoma.


K E Y W O R D S
IGFBP-3, IRS-1/2, PP1 c , PP2A c , TGF-β, TβR-V early stage of carcinogenesis and a tumor promoter in latestage cancer. 19 As a tumor suppressor, TGF-β suppresses carcinogenesis by potently inhibiting growth in epithelial cells for maintaining normal squamous epithelial morphology and physiology. 21 We previously demonstrated that IGFBP-3 and TGF-β inhibit growth in epithelial cells by stimulating TβR-V-mediated tumor suppressor signaling which involves IRS-1/2-dependent activation and cytoplasm-to-nucleus translocation of IGFBP-3-or TGF-β-stimulated protein phosphatase (PPase), and dephosphorylation of retinoblastoma family proteins in the nucleus, resulting in cell growth arrest. 7,10,22,23 In this communication, we demonstrate the identification of IGFBP-3-and TGF-β-stimulated PPases as PPase 2A (PP2A) and PPase 1 (PP1), which are the master regulators of the eukaryotic cell cycle, respectively, based on the distinct sensitivity of these PPase activities to okadeic acid (OA) and PP2A c siRNA. By [Methy-3 H] thymidine incorporation/Western blot/phosphorimager/immunofluorescence-microscopy analyses, we also demonstrate that IGFBP-3 and TGF-β stimulate IRS-2dependent activation and cytoplasm-to-nucleus translocation of PP2A c and PP1 c , resulting in dephosphorylation of pRbrelated proteins (p130 or p107) and pRb (p105) in the nucleus, respectively, in epithelial cells and growth arrest.
Alexa Fluor 488-and 594-conjugated secondary antibodies were purchased from Thermo Fisher. Secondary antibodies conjugated with horseradish peroxidase (Millipore, USA) and enhanced chemiluminescence (ECL) kit (Perkin-Elmer Life Sciences) were used to develop immunoblots. TGF-β peptide antagonist [β 1 25  The PPase activity assay mixtures were composed of 50 mM Tris-HCl, pH 7.0 containing 10% glycerol, 1 mM benzamidine, 0.1 mM PMSF, 14 mM mercaptoethanol, 0.1 mg of bovine serum albumin (BSA), PPase-containing sample (cell lysates or nucleus extracts containing 5 µg protein), and 32 P-labeled substrate in a final volume of 0.05 ml. Reactions in triplicates were initiated with the 32 P-labeled casein at 30°C, and after a 10 min reaction period, 0.1 ml of 10% trichloroacetic acid (TCA) was added. The mixture was centrifuged at 12,000 g for 2 min in a microcentrifuge. About 0.1 ml of the supernatant was then added to 1 ml scintillation counting liquid, and radioactivity was determined.
The lysates from cells treated with vehicle only exhibited non-specific PPase activity (IGFBP-3-or TGF-β-independent PPase activity with certain ~10 2 -10 3 cpm; 200 cpm/pmol phosphate). This non-specific PPase activity was subtracted from the total PPase activity in the cell lysates from cells treated with IGFBP-3 or TGF-β in order to estimate IGFBP-3-stimulated or TGF-β-stimulated PPase activity. For this reason, the mean (±SD) of the non-specific PPase activity from triplicates was taken as 0 cpm in cells treated with vehicle only.

| Immunofluorescence microscopy
One milliliter of culture media containing approximately 5,000-10,000 Mv1Lu cells was added to a 35 mm culture dish containing a square coverslip. Mv1Lu cells grown on coverslips were treated with IGFBP-3 or TGF-β. Cells were then fixed in 4% paraformaldehyde for 15 min followed by permeabilization. Fixed cells were blocked with 5% BSA in PBS for 20 min at room temperature (RT) and then incubated with an appropriate primary antibody solution overnight at 4°C. Fixed cells were incubated with Alexa Fluor-conjugated secondary antibodies for 1 hr at RT. Samples were observed with a Zeiss AxioObserver Z1 microscope (Zeiss), and images were captured using AxioVision Rev 4.6 software. To determine the nuclear localization and the colocalization of PP1 c and hyperphosphorylated pRb (P-Rb), the images were analyzed in three dimensions using an AxioObserver Z1 Apotome microscope (Zeiss). Colocalization was evaluated in single optical planes taken through the entire z-axis of each cell. All images were acquired using identical intensity and photodetector gain to allow quantitative comparisons of relative levels of immunoreactivity between samples. All images were cropped and sized using ImageJ.

| Nucleus fractionation for PPase activity assay
Nuclear extracts of the cells were prepared by hypotonic lysis followed by high salt extraction. Briefly, cell pellets were homogenized in 0.5 mL of ice-cold lysis buffer, composed of 10 mM HEPES pH 7.9, 10 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol (DTT), and 0.5 mM phenylmethylsulfonyl fluoride (all from Sigma Chemical Co.). The homogenates were centrifuged for 30 s at 500 g at 4°C to eliminate any unbroken tissue. The supernatants were incubated on ice for 20 min, vortexed for 30 s after the addition of 50 μL of 10% Nonidet P-40 (Sigma Chemical Co.), and then centrifuged for 1 min at 5,000 g at 4°C. The crude nucleus pellet was suspended in 200 μL of ice-cold extraction buffer (20 mM HEPES pH 7.9, 420 mM NaCl, 1.5 mM MgCl 2 , 0.1 mM EDTA, 1 mM DTT, and 0.5 mM PMSF) and incubated on ice for 30 min, mixed frequently, and centrifuged at 12,000 g at 4°C for 15 min. The supernatants were collected as nucleus extracts for Western blot and PPase activity assay. Protein concentration was determined using a bicinchoninic acid assay kit with BSA as the standard (Pierce Biochemicals).

| siRNA interference
Murine PP2A c siRNA oligonucleotide corresponding to nucleotide sequence 5'-xxx-3' (ON-TARGETplus SMARTpool Cat #: L-040657-00) and negative control siRNA were obtained from Dharmacon. PP2A c siRNA and negative control siRNA were resuspended in in RNase-free water and stored at −80°C. Transfection of siRNA was carried out using electroporation (Bio-Rad Gene Pulser Xcell Total System). Three million cells in 600 µl of RPMI 1640 were incubated with siRNA in a 0.4 cm cuvette for 5 min on ice before electroporation (260 V, 950 µF). After additional 5-min incubation on ice, cells were re-suspended in 12 ml of RPMI 1640 supplemented with glutamine and 10% FCS (fetal calf serum) without antibiotic. Antibiotics (1% penicillin/streptomycin) were added at 6 hr after electroporation. All measurements were performed at 24 or 72 hr after transfection.

| Metabolic labeling and immunoprecipitation
Mv1Lu and 32D cells (3 × 10 6 cells) grown in 6-well plate were washed and incubated in phosphate-free DMEM for 1 hr to deplete intracellular phosphate. After 2 hr of incubation with [ 32 P] orthophosphate at 37°C in a CO 2 incubator, cells were treated with 1 µg/ml of IGFBP-3 and/or OA (and RAP) for 16 hr. Cell lysates were prepared by suspending cells in 600 µl of lysis buffer and p130 or p107 was immunoprecipitated with a rabbit polyclonal antibody against the N-terminal domain of p130 or p107. The p130 or p107 antibody complex was captured with a protein G-coated agarose beads. The immunoprecipitated proteins were resolved using 7.5% SDS-PAGE. The gel was dried and autoradiographed by a phosphorimager.

| Statistical analysis
Two-tailed unpaired Student's t-test was used for determining the significance of a difference between two (vehicle only and sample) means. It was mainly used to compare the means between two groups (vehicle only and one specific concentration of IGFBP-3 or TGF-β). The values were presented as mean ± SD. p < 0.05 was considered significant.

| IGFBP-3-and TGF-β-stimulated PPase activities are distinct in the sensitivity to OA inhibition in Mv1Lu cells
We previously proposed a model for the mechanisms by which IGFBP-3 and TGF-β inhibit growth in epithelial cells by stimulating TβR-V/IRS-1/2/PPase signaling. 7 However, in this model, the identity of IGFBP-3-or TGF-β-stimulated PPase was unknown. To characterize and identify the IGFBP-3-and TGF-β-stimulated PPases, we developed a cell-free PPase activity assay. In this assay, 32 P-phosphorylated casein, which was generated by 32 P-phosphate-labeling ( 32 P-labeling) of casein (dephosphorylated) with protein kinase A in the presence of γ-32 P-ATP, was incubated with cell lysates of Mv1Lu cells treated with or without IGFBP-3 or TGF-β1 (TGF-β). After incubation, 32 P-phosphate released from 32 P-casein via the action of stimulated PPase in cell lysates and nucleus extracts were separated from remaining 32 P-casein by 10% trichloroacetic acid (TCA) precipitation in the presence of a carrier protein (BSA). The 32 P-phosphate released was recovered in the supernate of the 10% TCA solution. The IGFBP-3-and TGF-β-stimulated PPase activities were estimated by subtracting the radioactivity of 32 P-phosphate released by cell lysates or nucleus extracts of cells treated without IGFBP-3 or TGF-β from that released by cell lysates or nucleus extracts of cells treated with IGFBP-3 or TGF-β. Using this assay, we characterized the kinetics, IGFBP-3 or TGF-β concentration dependence and OA sensitivity of the IGFBP-3-or TGF-β-stimulated PPase activity in Mv1Lu cells. As shown in Figure 1, IGFBP-3 and TGF-β stimulated the PPase activities in a time-and concentration-dependent manner. The IGFBP-3-stimulated PPase activity in the cell lysates appeared to be linear with time up to 3 hr treatment in these cells treated with 1 µg/ml of IGFBP-3 ( Figure 1A). The TGF-β-stimulated PPase activity in the cell lysates also exhibited a linear relationship with the treatment time for 3 hr in Mv1Lu cells treated with 40 pM TGF-β (data not shown). IGFBP-3 and TGF-β stimulated the PPase activities in a concentration-dependent manner ( Figure 1B and Figure 1C, respectively). The halfmaximum concentration of IGFBP-3 to stimulate the PPase activity was estimated to be ~10 nM (0.3 μg/ml) ( Figure 1B) which is close to the half-maximum concentration of IGFBP-3 for binding to the IGFBP-3 receptor (TβR-V) and for inhibiting cell growth in Mv1Lu cells. 4,5 TGF-β also stimulated a PPsse activity in Mv1Lu cells in a concentration-dependent manner with a half-maximum concentration of ~40 pM which is close to the K d (50 pM) of TGF-β binding to TβR-V 2,3 in these cells ( Figure 1C). TGF-β at 10 pM stimulated a significant level of PPase activity (0.8 × 10 3 cpm; 200 cpm/pmol phosphate) ( Figure 1C). However, the IGFBP-3-stimulated PPase activity is distinct from the TGF-β-stimulated PPase activity in its greater sensitivity to OA inhibition. OA at 0.5 nM completely inhibited the IGFBP-3-stimulated PPase activity ( Figure 1D). OA at 1 nM did not significantly affect the TGF-β-stimulated PPase activity ( Figure 1D). These results suggest that IGFBP-3 and TGF-β stimulate PPase activities by interaction with TβR-V in Mv1Lu cells and that IGFBP-3-and TGF-β-stimulated PPases are different enzymes with distinct sensitivity to OA inhibition in these cells.
To analyze the presence of IGFBP-3-stimulated PPase-IRS-2 complexes in nucleus extracts of Mv1Lu cells, the IGFBP-3-stimulated PPase activity associated with IRS-2 in nucleus extracts was then determined. As shown in Figure 3C, IGFBP-3 stimulated a PPase activity in cell lysates of treated Mv1Lu cells. Approximately 50% of it was present in nucleus extracts and could be immunoprecipitated by antibodies to IRS-2 ( Figure 3D vs. Figure 3C). Insulin completely abolished the IGFBP-3-stimulated PPase activity in cell lysates ( Figure 3C). These results suggest that IGFBP-3 stimulates complex formation and cytoplasm-to-nucleus translocation of IGFBP-3-stimulated PPase and IRS-2 in Mv1Lu cells.

| IGFBP-3-induced growth inhibition, but not TGF-β-induced growth inhibition, is reversed by OA and PP2A c siRNA in Mv1Lu and 32D cells
Since OA blocked the activity of the IGFBP-3-stimulated PPase and the TGF-β-stimulated PPase activity was relatively resistant to inhibition by 0.2-1 nM OA ( Figure 1D), OA should be able to reverse IGFBP-3-induced growth inhibition, but not TGF-β-induce growth inhibition, in Mv1Lu cells. To test this, we determined the effects of OA on growth inhibition (as measured by [Methy-3 H] thymidine incorporation) induced by IGFBP-3 and TGF-β. As shown in Figure 4A, OA reversed the growth inhibition induced by different concentrations of IGFBP-3. OA reversed IGFBP-3-induced growth inhibition in a concentration-dependent manner ( Figure 4A). At 5 nM, OA reversed IGFBP-3 (0.1 μg/ml)-induced growth inhibition by ~80% ( Figure 4A) but did not significantly affect TGF-β-induced growth inhibition in these cells ( Figure 4B). IGFBP-3-stimulated PPase is sensitive to OA inhibition, suggesting that IGFBP-3-stimulated PPase is likely to be identical to PP2A which is known to be highly sensitive to OA. 26,27 To test this, we used murine myeloid cells which stably expressed human IR and IRS-2 (32D cells) and responded to IGFBP-3-induced growth inhibition. 22  scaffolding A subunit, and a 56-kDa substrate-recognizing B subunit (PP2A-B56). 27 We examined the effect of PP2A c siRNA transfection on IGFBP-3-induced growth inhibition in these murine 32D cells. This PP2A c siRNA was developed based on the murine sequence. 32D cells were transfected with control siRNA, 2 and 4 nM PP2A c siRNA by electroporation and treated with IGFBP-3. As shown in Figure 4C,D, PP2A c siRNA (2 and 4 nM) reversed the growth inhibition induced by IGFBP-3 ( Figure 4C), but not by TGF-β ( Figure 4D), in a dose-dependent manner in murine 32D cells. Four nM PP2A c siRNA reversed IGFBP-3-induced growth inhibition at 0.1 μg/ ml by ~78% in these murine cells ( Figure 4C). This degree of inhibition is comparable to the ~70% downregulation of PP2A c protein by murine PP2A c siRNA transfection (vs. control siRNA transfection) ( Figure 4E)  vs. lane 1 and bottom panel, quantitative analysis in three independent experiments). Murine PP2A c siRNA was unable to reverse IGFBP-3-induced growth inhibition in mink Mv1Lu cells (data not shown). This is consistent with the inability of murine PP2A c siRNA to significantly downregulate mink PP2A c ( Figure 4E, top panel, lane 4 vs. lane 3). These results suggest that IGFBP-3-induced growth inhibition is reversed by OA in Mv1Lu cells and by murine PP2A c siRNA in murine 32D cells.

| IGFBP-3 stimulates cytoplasm-tonucleus translocation of PP2A in Mv1Lu cells
As described above, IGFBP-3-stimulated PPase activity and IGFBP-3-induced growth inhibition are blocked or reversed by co-treatment with very low concentrations of OA in Mv1Lu and 32D cells and by transfection with PP2A c siRNA in murine 32D cells. These suggest that the IGFBP-3-stimulated PPase is identical to PP2A. We hypothesized that IGFBP-3 induces growth inhibition by stimulating IRS-2-dependent activation and cytoplasm-to-nucleus translocation of PP2A c in Mv1Lu cells. To test this, Mv1Lu cells were treated with 0, 2, and 10 nM (or 0.06 and 0.3 µg/ml, respectively), IGFBP-3 for 2 hr. The cytoplasm and nucleus fractions in treated cells were then isolated and subjected to 7.5% SDS-PAGE followed by Western blot analysis. As shown in Figure 5A, IGFBP-3 at 2 and 10 nM increased accumulation of PP2A c in the nucleus fraction by 1.5-to 1.7-fold (n = 3) as compared to that in cells treated with vehicle only (0 nM IGFBP-3). These results suggest that IGFBP-3 promotes cytoplasm-to-nucleus translocation of PP2A c (likely as the IRS-2 complex) in Mv1Lu cells.

| IGFBP-3 inhibits growth by inducing dephosphorylation of pRb-related proteins, p130 and p107, in Mv1Lu and 32D cells, respectively
PP2A plays a critical multi-faceted role in the regulation of the cell cycle. It has been implicated in dephosphorylation of two retinoblastoma protein (pRb)-related proteins, p130 and p107, which interact primarily with E2F4 and E2F5 and are most active in G0-the quiescent phase of the cell cycle. 26,27 Moreover, pRb (p105) interacts primarily with E2F1-3 and is most active at the G1-to-S phase transition. 26,27 These suggest that IRS-2-PP2A complexes may dephosphorylate pRb-related proteins (p130 and p107) in the nucleus of target F I G U R E 5 IGFBP-3 stimulates cytoplasm-to-nucleus translocation of PP2A c in Mv1Lu cells (A) and inhibits growth by inducing dephosphorylation of pRb-related proteins, p130 and p107, in Mv1Lu (B) and 32D (C) cells, respectively. Mv1Lu cells were treated with 0, 2, and 10 nM (or 0, 0.06 and 0.3 µg/ml, respectively), IGFBP-3 for 2 hr. The cytoplasm and nucleus fractions were separated by centrifugation and analyzed by 7.5% SDS-PAGE followed by Western blot analysis using antibodies to PP2A c , β-actin, and lamin B. The final volume of the total cytoplasm fraction was 10 times higher than that of the total nucleus fraction. However, an equal volume of cytoplasm and nucleus fractions was analyzed by 7.5% SDS-PAGE followed by Western blot analysis. At 2 and 10 nM, IGFBP-3 appeared to increase cytoplasm-to-nucleus translocation of PP2A c by ~2 fold. Western blot analysis was the representative of a total of three experiments. Lamin B and β-actin served as nuclear and cytoplasmic internal standards, respectively. (B and C) Mv1Lu (B) and 32D (C) cells were pre-incubated with [ 32 P]-orthophosphate for 2 hr, washed and incubated in the culture medium with excess phosphate. 32 P-labeled cells were treated with vehicle only or 0.3 µg/ml of IGFBP-3 in the presence and absence of OA (5 nM) and RAP (60 µg/ml). After 2 hr at 37°C, cell lysates were immunoprecipitated with antibodies to p130 and p107. The immunoprecipitates were analyzed by 7.5% SDS-PAGE and quantified by a Perkin Elmer phosphorimager (B and C, panels a and b). Phosphorimager analysis was the representative of a total of three experiments. IGFBP-3 appeared to stimulate dephosphorylation of p130 and p107 in Mv1Lu and 32D cells, respectively (B and C, panels a, lane 2 vs. lane 1 and panel b, quantitative data). OA and RAP inhibited IGFBP-3-stimulated dephosphorylation of p130 and p107 (B and C, panel a, lanes 3 and 4 vs. lane 2 and panel b, quantitative data) in Mv1Lu and 32D cells, respectively. The quantitative data from three independent analyses were shown. The data are mean ±SD *Significantly lower than that of cells treated with vehicle only (control): p < 0.01 cells. To test this possibility, Mv1Lu and 32D cells were prelabeled with 32 P-orthophosphate at 37°C for 1 hr, washed, and incubated with 0.3 µg/ml (10 nM) IGFBP-3 in the presence of excess unlabeled orthophosphate in the medium. After 2 hr at 37°C, 32 P-labeled cell lysates were immunoprecipitated with specific antibodies to p130 and p107 and analyzed by 7.5% SDS-PAGE and quantified by a phosphorimager (panels a and b, respectively). As shown in Figure 5B and Figure

| TGF-β induces colocalization of TβR-V and PP1 c at the plasma membrane and accumulation of PP1 and decreased levels of hyperphosphorylated pRb (P-Rb) in the nucleus in Mv1Lu cells
The TGF-β-stimulated PPase involved in TGF-β-induced growth inhibition has been identified as PP1 in human keratinocytes. 28 PP1 is responsible for dephosphorylating of pRb (p105) which is linked to TGF-β-induced growth inhibition in Mv1Lu cells. 29 The mechanism by which TGF-β stimulates PP1 activity is not clear. Since TGF-β induces growth inhibition by stimulating complex formation of TβR-V, IRS-1/2, and likely PP1 at the plasma membrane in Mv1Lu cells, we hypothesize that PP1 should be activated by its interaction with IRS-1/2 in the formation of the TβR-V-IRS-1/2-PP1 ternary complexes in TGF-β-treated cells. PP1 enzyme contains both a 37-kDa catalytic subunit (PP1 c ) and at least one regulatory subunit which directs PP1 c to different substrates or sites. To test this, we performed immunofluorescence microscopy in Mv1Lu cells treated with 40 pM TGF-β at 37˚C for 0 and 1 hr using antibodies to TβR-V (LRP-1) and PP1 c ( Figure 6A). TGF-β-stimulated colocalization of TβR-V and PP1 c at the plasma membrane, as indicated by arrowheads in Mv1Lu cells treated with TGF-β at 37˚C for 1 hr (Figure 6Af). In contrast, Mv1Lu cells treated with 40 pM TGF-β for 0 hr did not exhibit colocalization of TβR-V and PP1 c in these cells (Figure 6Ac).
Retinoblastoma protein (pRb) present in the cytoplasm and nucleus fractions are identified as hyperphosphorylated (as a slow-migrating form) and hypophosphorylated (as a fastmigrating form) forms of pRb, respectively, based on its mobility on 7.5% SDS-PAGE. [29][30][31] Cytoplasmic pRb is known to be mainly the hyperphosphorylated form. 32 These suggest that TGF-β stimulates cytoplasm-to-nucleus translocation of PP1 c and correspondingly increases the amount of hypophosphorylated pRb (as a fast-migrating form of pRb on 7.5% SDS-PAGE), which is the PP1 c -dephosphorylated product of pRb in the nucleus. To demonstrate the subsequent cytoplasm-to-nucleus translocation of PP1 c and its effect on dephosphorylation of pRb in the nucleus, we performed immunofluorescence analysis in Mv1Lu cells treated with 40 pM TGF-β at 37˚C for 0, 1, and 2 hr using specific antibodies to PP1 c and hyperphosphorylated pRb (P-Rb). We reasoned that TGF-β promotes nucleus accumulation of PP1 c and should accordingly decrease the amount of P-Rb, its target substrate, in the nucleus. PP1 specifically dephosphorylates pRb in the nucleus of target cells. 28,29 After treatment of cells with TGF-β and immunofluorescence staining, six images, which consist of 6-8 cells/image, were taken at different areas of cells grown on a coverslip. The image shown in the data was the representative of the six images. As shown in Figure 6B, after treatment of cells with TGF-β at 37˚C for 1 or 2 hr, approximately 40%-50% cells on a coverslip exhibited significantly decreased yellow fluorescence (co-localization) in the nucleus, whereas ~90% cells (treated with TGF-β at 37˚C for 0 hr) on a coverslip exhibited yellow fluorescence (colocalization) in the nucleus (Figure 6Bf,i and Figure 6Bc, respectively). TGF-β treatment of cells for 1 and 2 hr decreased the amount of P-Rb and colocalization of PP1 c and P-Rb in the nucleus (Figure 6Be,h and Figure 6Bf,i, respectively). These results support the suggestion that TGF-β promotes cytoplasm-to-nucleus translocation of PP1 c , resulting in dephosphorylation of pRb in the nucleus, which leads to cell growth arrest.

| TGF-β stimulates cytoplasm-tonucleus translocation of PP1 c and increases formation of dephosphorylated pRb (Rb) in the nucleus in Mv1Lu (A) and A549 (B) cells
To further support the hypothesis that TGF-β stimulates cytoplasm-to-nucleus translocation of PP1, we determined the subcellular localization of PP1 c , PP2A c , pRb (Rb), phosphorylated Smad2 (P-Smad2), phosphorylated IRS-1/2 (P-IRS-1/2, phosphorylation at Ser 270), lamin B, and β-actin using 7.5% SDS-PAGE and quantitative Western blot analysis with specific antibodies to PP1 c , PP2A c , and others after subcellular cytoplasm/nucleus fractionation of Mv1Lu and A549 cells treated with 40 pM TGF-β for 0, 1, and 2 hr. As shown in Figure 7, TGF-β increased the amounts of PP1 c ( Figure 7A,B), dephosphorylated Rb (as a fast-migrating form of Rb on 7.5% SDS-PAGE) ( Figure 7A,B), P-IRS-2 ( Figure 7A), and P-Smad2 ( Figure 7A,B) in the nucleus fraction in a time-dependent manner in these cells. Both P-IRS-1/2 contain Ser 270 but only P-IRS-2 entered the nucleus (Figure 3Be,h). After 2 hr, TGF-β increased the amounts of PP1 c , dephosphorylated Rb (as a fast migrating form on 7.5% SDS-PAGE), P-IRS-2 and P-Smad2 by 1.5-to 2-fold (n = 3) in the nucleus fraction in Mv1Lu and A549 cells. In contrast, TGF-β did not significantly increase the amount of PP2A c in the nucleus fraction in these cells.
Interestingly, Rb present in the cytoplasm and nucleus fractions were identified as phosphorylated (as a slow-migrating form) and dephosphorylated (as a fast-migrating form) forms of Rb, respectively, based on its mobility on 7.5% SDS-PAGE. [29][30][31] Cytoplasmic Rb is known to be mainly the hyperphosphorylated form. 32 These results suggest that TGF-β stimulates cytoplasmto-nucleus translocation of PP1 c and P-IRS-2, and correspondingly increases the amount of dephosphorylated Rb (as a fast-migrating form of Rb), which is the PP1-dephosphorylated product of Rb, in the nucleus.

| DISCUSSION
Here, we have provided evidence revealing that IGFBP-3 inhibits growth in epithelial cells by stimulating the F I G U R E 6 TGF-β induces co-localization of TβR-V and PP1 c at the plasma membrane (A) and accumulation of PP1 c and decreased levels of hyperphosphorylated pRb (P-Rb) in the nucleus (B) in Mv1Lu cells. Mv1Lu cells were grown to 50% confluence on coverslips in 35 mm culture dishes at 37˚C for 24 hr. Mv1Lu cells were then treated with 40 pM TGF-β at 37°C. After 0 and 1 hr (A) or 0, 1, and 2 hr (B), cells were fixed and stained by immunofluorescence using antibodies to TβR-V and PP1 c (A) or using antibodies to PP1 c and P-Rb (B). (A) After immunofluorescence staining, cells on coverslips were counted. Cells treated with TGF-β at 37°C for 0 hr did not exhibit colocalization of TβR-V and PP1 c at the plasma membrane (Ac). However, cells treated with TGF-β at 37°C for 1 hr exhibited colocalization of TβR-V and PP1 c at the plasma membrane as indicated by arrowheads (Af). TβR-V (LRP-1) is known to undergo constitutive endocytosis and recycling in cells. Perinuclear labeling is likely to be endocytic vesicles which are often seen in juxtanuclear regions. This appearance of endocytic vesicles might be due to longer-time cell culture before the experiment. (B) After treatment of cells with TGF-β for 1 or 2 hr at 37°C, approximately 40%-50% cells on a coverslip exhibited significantly decreased yellow fluorescence (co-localization) in the nucleus, whereas ~90% cells (treated with TGF-β at 37°C for 0 hr) on a coverslip exhibited yellow fluorescence (colocalization) in the nucleus (Bf,i and Bc, respectively). Arrowheads indicate decreased colocalization (as marked by decreased yellow fluorescence) of PP1 c and P-Rb (Bf,i) due to decreased P-Rb (as marked by decreased red fluorescence) in the nucleus (Be,h) p130 and p107 possess high-affinity PP2A-B56 docking motifs of LsgIlE (residues 519-524) and LinIfE (residues 412-417), respectively. It is important to note that p130 and p107 do not possess specific PP1 c docking motifs (FxxR/ KxR/K), 34 suggesting that p130 and p107 are the PP2A target substrates in the nucleus. In this communication, we also demonstrate that IGFBP-3 stimulates cytoplasm-to-nucleus translocation of IRS-2 but not IRS-1. IRS-2 has been shown to undergo nuclear translocation in normal and cancer cells. 35 It possesses a putative nuclear localization signal (NLS) motif of KKwRsK (residues 80-85). Moreover, after dissociation from the cytoplasmic tail of TβR-V, low-affinitybound dephosphorylated IRS-1-PP2A complexes are mainly present in the cytoplasm and do not have known functions in cells treated with IGFBP-3. 7 However, IGFBP-3 is known to inhibit phosphorylation of c-raf-MEK-ERK and p38 kinase in insulin-secreting cells. 36 PP2A is also known to inhibit the kinase activities of the kinases involved in TβR-I-activated non-Smad pathways, which include JNK, 37 TAK1-p38/ JNK, 38 PI3K-AKT, 39 and RhoA-Rock 40 signaling, by dephosphorylating these kinases. 41,42 It is likely that IRS-1-PP2A is responsible, at least in part, for IGFBP-3-induced inhibition of non-Smad signaling. 36 Several lines of evidence suggest that IGFBP-3 acts as a potential tumor suppressor gene. First, aberrant promoter methylation of IGFBP-3 gene, which silences its expression, is detected in human gastric cancer, colorectal cancer, breast cancer, and malignant mesothelioma cancer. 20 Second, low F I G U R E 7 TGF-β stimulates cytoplasm-to-nucleus translocation of PP1 c in Mv1Lu (A) and A549 (B) cells and increases dephosphorylated pRb (a fast-migrating form of pRb) levels in the nucleus in these cells. Mv1Lu (A) and A549 (B) cells were treated with 40 pM TGF-β for 0, 1, and 2 hr. The cytoplasm and nucleus fractions were separated by centrifugation and analyzed by 7.5% SDS-PAGE followed by quantitative Western blot analysis using antibodies to PP1 c , PP2A c , pRb (Rb), P-IRS-1/2 (phosphorylated IRS-1/2, Ser 270), P-Smad2 (phosphorylated Smad2), lamin B, and β-actin. pRb (Rb) (present in the nucleus) migrated as a fast-migrating form (dephosphorylated pRb) on 7.5% SDS-PAGE. pRb (Rb) (present in the cytoplasm) migrated as a slow form (phosphorylated pRb). The final volume of the total cytoplasm fraction was 10 times higher than that of the total nucleus fraction. An equal volume of cytoplasm and nucleus fractions was then analyzed by 7.5% SDS-PAGE followed by Western blot analysis. Western blots were representatives of a total of three experiments. Lamin B and β-actin served as nuclear and cytoplasmic internal standards, respectively | 723 CHEN Et al. levels of IGFBP-3 expression in cancer tissues are correlated with poor prognosis for patients with esophageal squamous cell carcinoma 43 and hepatocellular carcinoma. 44 Third, low IGFBP-3 expression correlates clinically with higher tumor grade, advanced stage, and poor survival in ovarian endometrioid adenocarcinoma patients. 45 Here, we demonstrate that IGFBP-3 inhibits cell growth by stimulating the TβR-V-mediated tumor suppressor signaling pathway (TβR-V/ IRS-1/2/PP2A/p130, p107). Among the components of this signaling cascade, TβR-V and PP2A have been proved to be tumor suppressor genes by that stable transfection of human carcinoma cells and CHO-LRP-1 −/− cells with LRP-1 (TβR-V) cDNA restores the growth inhibitory response to IGFBP-3 and TGF-β, and normal epithelial morphology 10,15 and by that loss of PP2A regulatory subunit B56δ promotes spontaneous tumorigenesis in vivo. 46 In addition, knockout of the TβR-V (LRP-1) gene and both p130 and p107 genes in mice has been shown to cause embryonic and neonatal lethality, respectively. 47,48 In normal epithelial cells, IGFBP-3 induces growth inhibition by stimulating TβR-V-mediated and IRS-1/2-dependent activation of PP2A. PP2A may serve as an important down-stream effector for mediating other known Insulin and IGF-I antagonize TGF-β-stimulated TβR-V/IRS-1/2/ PP1 signaling by stimulating tyrosine phosphorylation of IRS-1/2 via interaction with their cognate receptors, insulin receptor (IR), and IGF-1 receptor (IGF-1R). The tyrosine phosphorylation of IRS-1/2 catalyzed by IR and IGF-1 receptor, which occurs more rapidly than the Ser/Thrspecific dephosphorylation of IRS-1/2 by PP1 or PP2A, leads to multiple IR/IGF-1R downstream signaling pathways and prevents the formation of TβR-V-IRS-1/2-PP1 complexes. Tyrosine phosphorylation and Ser/Thr-specific dephosphorylation of IRS-1/2 are mutually exclusive. 7 In diabetes, insulin or insulin signaling defects potentiate TGF-β-induced growth inhibition in target cells. 7 In addition, high glucose in the plasma and tissues of diabetic patients may enhance TβR-V and TβR-I/TβR-II signaling via increasing TGF-β production and TβR-I/TβR-II expression. 7,80 Increased ECM synthesis (which is mediated by TβR-I/TβR-II/Smad2/3/4 signaling) further attenuates insulin signaling and enhances TGF-β-induced growth inhibition, 7 resulting in alopecia, impaired wound healing, accelerated glomerulopathy, and tissue fibrosis in diabetic patients. The signaling cascades potentiated in diabetic patients are indicated by red arrows cell biological activities of IGFBP-3 and other high-affinity IGFBPs. 1,36 TβR-V is the only cell surface IGFBP-3 receptor identified by I 125 -labeled IGFBP-3 affinity labeling (binding/crosslinking) followed by immunoprecipitation using antiserum to TβR-V in epithelial cells and other cell types. [4][5][6] TβR-V also binds IGFBP-4 and IGFBP-5 but not IGFBP-1, IGFBP-2, and IGFBP-6, as determined by I 125 -labeled IGFBPs affinity labeling. 5,6 It exhibits the highest binding affinity toward IGFBP-3 with a K d of 10 nM. 4-6 A TGF-β peptide antagonist β 1 25 , which contains a minimal active site motif of WS/ CXD in TGF-β and IGFBP-3 molecules, blocks TGF-β and IGFBP-3 binding to TGF-β receptors in epithelial cells and reverses growth inhibition induced by either TGF-β or IGFBP-3 in these cells. 4,5,11,12 The transmembrane protein TMEM219 (25 kDa) was also identified as an IGFBP-3 receptor (termed IGFBP-3R) in 2010, using the yeast twohybrid screening and a human breast cancer cell cDNA library. 49 In contrast to the IGFBP-3 receptor (TβR-V/LRP-1), IGFBP-3R/TMEM219 does not bind other high-affinity IGFBPs 49 and have known function of a tumor suppressor gene. IGFBP-3R/TMEM219 was identified as a cell death receptor mediating IGFBP-3-induced anti-tumor effects in cancer cells 49,50 and as an autophagy-activation receptor mediating IGFBP-3-activated autophagy in Vero cells. 51 In addition, the K d of IGFBP-3 binding to TMEM219/IGFBP-3R has been estimated to be 125 nM. 51 IGFBP-3 (1 µM) is utilized to stimulate TMEM219/IGFBP-3R-mediated autophagy activation in Vero cells (kidney epithelial cells). 51 It is important to note that carcinoma cancer cells do not express the IGFBP-3 receptor (TβR-V) and loss of TβR-V confers cancer malignancy. 7 Many lines of evidence suggest that the IGFBP-3 receptor (TβR-V) is the primary IGFBP-3 receptor in normal epithelial cells and other cell types. 7 The transcriptional activation and growth inhibition activities of TGF-β have generally been thought to be mediated by the canonical TβR-I/TβR-II/Smad2/3/4 signaling cascade. 13 However, these two activities appear to segregate in several cell types and under various conditions. [52][53][54] This suggests that other signaling pathways must be involved in mediating the TGF-β activities. Although TGF-β-stimulated canonical TβR-I/TβR-II/Smad2/3/4 signaling can be modulated by other signaling pathways, 55,56 it is mainly responsible for mediating the transcriptional activation of ECM synthesis-related genes. Smad2/3/4 responsive elements exist in the promoter regions of all responsive genes. In contrast, the signaling involved in TGF-β-induced growth inhibition in target cells is unknown. Other signaling pathways, in addition to the well-known canonical TβR-I/TβR-II/Smad2/3/4 signaling pathway, are suggested to be involved in the growth inhibitory response to TGF-β. 57,58 The Ras/ERK signaling and PP2A are involved in mediating TGF-β-induced growth inhibition in certain cells. 59,60 However, the main signaling pathway, in concert with canonical TβR-I/TβR-II/Smad2/3/4 signaling, 13 mediates the growth inhibitory response to TGF-β in epithelial cells remains unknown. TβR-I, TβR-II, TβR-III, and TβR-V co-express in all normal cell types studied. Since the TβR-III null mutation in mice does not affect the growth regulatory response to TGF-β in embryonic fibroblasts derived from these mice, 61 the remaining candidate is TβR-V. Many carcinoma cells and primary tumors express no or very low levels of TβR-V expression. 2,7,[16][17][18] Growth of these cells is not inhibited by either TGF-β or IGFBP-3. In the absence of TβR-V in late-stage cancer, TGF-β induces EMT (epithelial-mesenchymal transition), autoinduction, and increased invasiveness by stimulating TβR-I-activated or TβR-I-mediated non-Smad signaling pathways 37-40 as well as canonical Smad signaling (TβR-I/TβR-II/Smad2/3/4 signaling). 13 TβR-II is apparently not involved in TGF-β-stimulated TβR-I-activated/mediated non-Smad signaling pathways [37][38][39][40][41][42] which are involved in cell survival, migration, proliferation, malignant transformation, and tumor growth. In fact, TGF-β stimulates tumor promoter signaling toward EMT is mediated by both non-Smad and Smad pathways. While no LRP-1 (TβR-V) is detected in hepatoma in human patients, it is present in the normal parenchymal tissue surrounding the hepatomas. 16 These suggest that TβR-V is involved in mediating the growth inhibitory response to TGF-β in normal epithelial cells and that its loss contributes to the malignant phenotype in cancer cells. 7 These are also consistent with the notion that TβR-V acts as a tumor suppressor gene and controls cell growth in normal epithelial cells by mediating TGF-β-induced growth inhibition in these cells. The loss or deficiency of TβR-V leads to the development of carcinoma cancer. Although no mutation in the LRP-1 (TβR-V) gene has been found related to cancer initiation or progression, the T allele of the C766 T polymorphism in the LRP-1 (TβR-V) gene is associated with an increased risk of breast cancer. 62 We previously proposed a model for the mechanism by which TGF-β inhibits growth in epithelial cells by binding to a site between cell surface subdomains I and II of TβR-V in target cells. In this model, TGF-β stimulates sequential association of IRS-1 or IRS-2 and a Ser/Thr-specific PPase with the cytoplasmic tail of TβR-V by inducing TβR-V dimerization via its covalently linked homodimeric structure. In the ternary complexes, the Ser/Thr-specific PPase becomes activated and dephosphorylate IRS-1/2. Dephosphorylated IRS-1-PPase or IRS-2-PPase binary complexes dissociate from the cytoplasmic tail of TβR-V and undergo IRS-1/2dependent translocation from cytoplasm to the nucleus where the PPase dephosphorylates pRb (retinoblastoma protein) or pRb-related proteins, resulting in growth arrest. This model lacked the identity of PPase, IRS-1/2, and retinoblastoma family proteins which are involved in TGF-β-stimulated TβR-V-mediated tumor suppressor (growth inhibition) signaling. 7 Here, we provide several lines of evidence to suggest that PP1, IRS-1/2, and pRb (p105) are involved in the TGF-βstimulated TβR-V-mediated tumor suppressor signaling cascade (TβR-V/IRS-2/PP1/pRb). These include: (1) TGF-β stimulates colocalization of TβR-V and IRS-1/2 at the plasma membrane in Mv1Lu cells, as demonstrated by immunofluorescence staining. (2) TGF-β stimulates cytoplasm-tonucleus translocation of IRS-2 but not IRS-1 in Mv1Lu cells, as demonstrated by immunofluorescence staining. (3) PP1 is known to be responsible for mediating TGF-β-stimulated dephosphorylation of pRb in keratinocytes and Mv1Lu cells. 28,29 (4) TGF-β-stimulated PPase (PP1) activity is abolished in cells co-treated with RAP (LRP-1/TβR-V antagonist) or insulin 7,22,23 in Mv1Lu cells. (5) TGF-β-stimulated PPase (PP1) activity is distinct from IGFBP-3-stimulated PPase (PP2A) activity in its relative insensitivity to OA inhibition, which appears to be the biochemical character of PP1 activity. 26,27 PP2A is completely inhibited at 1 nM OA, compared to greater than 1 µM OA for PP1. 26 (6) OA at 0.5 nM completely inhibits IGFBP-3-stimulated PPase (PP2A) activity, but not TGF-β-stimulated PPase (PP1) activity, in Mv1Lu cells treated with IGFBP-3 and TGF-β. (7) TGF-β stimulates cytoplasm-to-nucleus translocation of PP1 c , resulting in dephosphorylation of pRb (p105) in the nucleus, as demonstrated by Western blot analysis following subcellular fractionation (to yield cytoplasm and nucleus fractions) and immunofluorescence analysis. (8) PP1 as well as PP2A are the master regulators of the eukaryotic cell cycle. 27 The above evidence supports an updated model ( Figure 8B) in which TGF-β induces growth inhibition in target cells by stimulating TβR-V-mediated signaling (TβR-V/IRS-2/PP1) which leads to dephosphorylation of pRb (p105) in the nucleus, resulting in cell growth arrest.
In this model ( Figure 8B), TGF-β, a covalently associated homodimeric cytokine, interacts with TβR-V at a site between subdomains I and II, 7,10,13 resulting in dimerization of TβR-V and sequential recruitment of IRS-1 or IRS-2 and PP1 to the cytoplasmic tail of dimeric TβR-V to form TβR-V-IRS-1-PP1 or TβR-V-IRS-2-PP1 ternary complexes. IRS-1 and IRS-2 possess PP1 c docking motifs of FrssfR (residues 438-443) and FefRpR (residues 298-303), respectively (PP1 c docking affinity: IRS-2 > IRS-1). IRS-2 appears to comprise of overlapping high-affinity PP1 and PP2A docking motifs of FefRpR (residues 298-303) and LkeLfE (residues 294-299), respectively, suggesting that PP1 and PP2A docking to IRS-2 are mutually exclusive. After dephosphorylation of IRS-2 by activated PP1 in the ternary complex, dephosphorylated IRS-2-PP1 binary complexes dissociate from the cytoplasmic tail of TβR-V and enter the nucleus via the nucleus-targeting function of IRS-2. In the nucleus, PP1 dephosphorylates pRb (p105), resulting in cell growth arrest. In normal epithelial cells, TGF-β potently inhibits cell growth (~100% growth inhibition at 1-5 pM) by stimulating the TβR-V-mediated signaling cascade (TβR-V/IRS-2/PP1/pRb) to dephosphorylate (activate) pRb by PP1 in concert with canonical TGF-β signaling (TβR-I/TβR-II/Smad2/3/4). 7 TGF-β-stimulated canonical signaling potentiates TβR-V-mediated growth inhibition in these epithelial cells, at least in part, by transcriptional activation of cyclin-dependent kinase (CDK) inhibitors which maintain pRb unphosphorylated (active) in the nucleus. 7,13 Moreover, upon TGF-β stimulation in cells, dephosphorylated IRS-1-PP1 complexes also dissociate from TβR-V and bind (or anchor) to a high-affinity PP1 docking motif FesfKR (residues 393-398) in the cytoplasmic domain of TβR-I in the TβR-V-TβR-I complex. 9 Since PP1 c itself possesses a high-affinity PP2A-B56 docking motif of LlrLfE (residues 82-87), on the way to bind to TβR-I ( Figure 8B), PP1 c also recruits PP2A to form the high-affinity-bound TβR-1-PP c -PP2A complex. PP2A recruited by TβR-1 as the high-affinity-bound PP1-PP2A complex 63,64 effectively suppresses or silences TGF-β-stimulated TβR-I-mediated non-Smad pathways by dephosphorylation of the kinases involved in non-Smad pathways in normal epithelial cells. However, in cancer cells lacking TβR-V expression, 7 TβR-I is transiently localized in lipid rafts due to its specific interaction with caveolin-1, a structural component of lipid rafts. 65 Lipid rafts serve as major platforms for non-Smad signaling regulation in cell migration and proliferation. 66 In these cancer cells, TβR-I-mediated non-Smad signaling pathways are activated by TGF-β due to defective recruitment of PP-1 c -PP2A by TβR-I to suppress non-Smad signaling. 42 Furthermore, in carcinoma cells, loss or very low levels of TβR-V expression do not affect TGF-β-stimulated canonical signaling (TβR-I/TβR-II/Smad-2/3/4), as evidenced by TGF-β-stimulated expression of PAI-1 in these cells. 10 Thus, as a tumor promoter, TGF-β stimulates both Smad and non-Smad signaling (termed tumor promoter signaling), leading to epithelial mesenchymal transition (EMT), autoinduction, invasiveness, and chemoresistance in these cancer cells. PP2A plays a pivotal role in suppressing the development of cancer malignancy via TGF-β-induced TβR-I recruitment of PP1-PP2A complexes 41,42 to suppress non-Smad signaling pathways. [37][38][39][40] The expression and activity of PP2A are commonly reduced in cancer tissues. Small molecule PP2A activators have been developed to treat cancer. 67 TGF-β is known to act as a tumor suppressor and a tumor promoter during tumorigenesis. The mechanism of switching TGF-β from a tumor suppressor to a tumor promoter in the process of tumorigenesis remains unclear. 68 We hypothesize that the presence and absence of TβR-V expression in target cells appear to be critical in determining whether TGF-β is a tumor suppressor or a tumor promoter. 7,69 In normal epithelial cells which express TβR-V, TGF-β suppresses carcinogenesis by potently inhibiting cell growth via stimulating TβR-V-mediated IRS-2-dependent tumor suppressor