Transforming growth factor ß (TGFß) is an essential paracrine mediator of androgen-withdrawal-induced apoptosis (AWIA) of prostate epithelium.1 Binding of TGF-ß to its receptors leads to phosphorylation of receptor-associated R-Smads (such as Smad2) which complex with Smad4, translocate to the nucleus and bind to Smad response elements, mediating gene expression. Although 3 TGFßs (1, 2 and 3) exist, and all are detected in the prostate, most studies have focused on TGFß1.2 In rodents, AWIA is accompanied by an increase in TGFß1 mRNA, and the kinetics of mRNA induction closely parallel those of AWIA.3 Additionally, there is a concomitant rise in the expression of the RI and RII subunits of the TGFß receptor4, 5 as well as phosphorylated Smad2, a key downstream mediator of TGFß.6 A dominant negative form of TGFß-RII blocks TGFß-induced differentiation and cell death.7, 8 Finally, instillation of TGFß1 into rat prostate results in regression of the gland because of apoptosis.3, 9
TGF-ß negatively regulates the proliferation of many adult epithelia, exerting its effects by arresting cell cycle progression or by inducing apoptosis. The mechanism of cell cycle arrest is Smad-dependent and well characterized,10 but the molecular mechanisms controlling apoptosis remain uncertain. There is support for a variety of molecular mechanisms11 including transcriptional regulation,12 which can be both Smad-dependent and Smad-independent.13 One potential downstream target is the signaling network regulated by the family of death receptor (DR) signaling ligands.14, 15
Consistent with the growth inhibitory activity of this pathway, TGFß signaling components function as tumor suppressors in many early stage cancers.11, 13 For example, human prostate cancer specimens frequently display reduced levels of TGFß receptors5, 16–18 and Smad4.18, 19 However, 2 of the well-characterized human prostate cancer cell lines (DU145 and PC3) have intact TGFß signaling pathways,20–22 yet have been reported to undergo only limited apoptosis when treated with TGFß.21, 23, 24 A rat prostate cancer cell line that is not normally growth inhibited by TGFß is inhibited when grown at low density in serum-free media,25 and this inhibition is reversed by growth factor treatment. Thus, autocrine growth factor signaling may account for the resistance of DU145 and PC3 to the apoptotic effects of TGFß. In our report, we identify the conditions that allow induction of TGFß1-induced apoptosis in the DU145 human prostate cancer cell line. We then test the hypothesis that TGFß1 acts by lowering the endogenous level of FLICE-like inhibitory protein (FLIP), thereby inducing caspase-8 activity and triggering apoptosis.
TGFß1 (R&D Systems, Minneapolis, MN), anti-FLIPL/S (used at 1:250 for immunoblotting, NF6, Alexis, San Diego, CA), anti-ß-actin (1:5,000, AC-15, Sigma, St. Louis, MO), anti-eEF2 (1:20,00026), anti-PARP (1:10,000, C2.10, Enzyme Systems, Livermore, CA), anti-caspase-8 (1:1,000, 1C12, Cell Signaling, Danvers, MA) and anti-caspase-10 (1:1,000, M059-3, Woburn, MA) were purchased from the indicated suppliers.
Cell culture and apoptosis assays
DU145 cells were grown in RPMI 1640 supplemented with 10% heat-inactivated fetal calf serum (FCS), 15 mM HEPES pH 7.4 and 1 mM sodium pyruvate. Apoptosis assays were performed in low serum media (substituting 1% heat-inactivated calf serum (CS) for FCS) using 1 ng/ml TGFβ1. For quantification of live DU145 cells after TGFß1 treatment, plates were washed with PBS, adherent cells were released by trypsin treatment and then counted using a Coulter counter. For fluorescence-activated cell sorting (FACS) analysis of apoptosis, DU145 cells (2–4 × 105) were detached from the dishes by pipetting and resuspended in PBS with 10 mM EDTA. Cells were subsequently washed in Hank's balanced salt solution (HBSS) and incubated in 0.1 ml HBSS containing R-phycoerythrin-conjugated Annexin-V (1:100, Invitrogen) and 7-amino-actinomycin D (7-AAD, 1:200, Invitrogen) for 15 min at room temperature in the dark. Cells were washed 3 times, resuspended in HBSS and analyzed by 2-color FACS using CellQuest Pro (BD Biosciences, San Jose, CA). Data were analyzed using FloJo software (Tree Star, Ashland, OR). Statistical analysis was performed by paired Student's t-test.
For small interfering RNA (siRNA) treatment, cells were seeded 1 day prior to transfection at a density of 2.5 × 105/35-mm dish in 1 ml RPMI 1640 medium without antibiotics. Cells were then transfected with either a control (10 nM) or a pool (10 nM total) of gene-specific oligomers using Oligofectamine (Invitrogen) as per the vendor's protocol. Four hours after transfection, cells were supplemented with normal medium containing 1% CS and further incubated for 3 days before analysis or TGFβ1 treatment. Specific siRNAs targeting luciferase, FLIPL/FLIPL+S, caspases-8 and caspases-10 were purchased from Dharmacon (Chicago, IL). Sequences of siRNAs are available upon request.
RNA (5 μg) was isolated using TRIzol reagent (Invitrogen) and reverse transcribed using SuperScript II MMLV reverse transcriptase (Invitrogen). Real-time PCR was performed with SYBR Green (Molecular Probes) using a Lightcycler (Idaho Technologies) as described.27 The primer pairs, which spanned at least 1 intron, are as follows: for β-actin, 5′-GAAGGATTCCTATGTGGG-3′ and 5′-ATCACGATGCCAGTGGTA-3′; and for FLIPL, 5′-GAGCAGGACAAGTTACA-3′ and 5′-CTTGTATCTCTCTTCAGG-3′. The mRNA copy numbers were determined by comparing the amplification profiles (log fluorescence versus cycle number) of the SYBR Green signal recorded 2°C below the duplex melting point with serially diluted internal standards of the same target gene. FLIPL copy number was normalized to ß-actin copy number for each sample. Each reverse transcription reaction was performed in triplicate, and real-time PCR reactions were carried out in triplicate for each reverse transcription sample. Statistical analysis was performed using Student's t-test.
Protein extracts were prepared by lysing cells in NP40 lysis buffer (1% NP40, 1 mM NaVO4, 100 mM NaF, 2.5 μM ZnCl2, 10% glycerol, 0.2 mM phenylmethylsulphonylfluoride and protease inhibitor cocktail (Sigma) in Tris-buffered saline solution (TBS)) followed by centrifugation. The supernatant was used as cytoplasmic extract and the pellet (dissolved in SDS sample buffer) as nuclear extract. Immunoblots were sequentially incubated in primary antibody, peroxidase-conjugated secondary antibody and SuperSignal West Pico chemiluminescense substrate (Pierce). Blots were imaged using a cooled charged-coupled digital camera (Kodak 4000R) and specific signal was quantitated using Molecular Imaging software (Kodak).
Low serum media slows growth and increases FLIP levels
Despite the apoptotic response of normal prostate epithelial cells to TGFß,1, 28, 29 we did not detect apoptosis in the human prostate cancer cell line DU145 grown as a monolayer in media containing 10% FCS (data not shown; DNS). The presence of growth factors in FCS may mask the apoptotic effects of TGFβ1 on DU145 cells. Because growth in 1% FCS slightly sensitized the cells to apoptosis,23 we characterized the growth pattern in a media with even lower levels of growth factors (1% CS). Under these conditions, cell growth slowed markedly relative to cells grown in 10% FCS (Fig. 1a). Doubling time increased from ∼32 hr in 10% FCS to 140 hr in 1% CS. DU145 cells exhibited increased basal apoptosis when grown in CS, but ∼80% of the DU145 cells were viable after 8 days in CS (DNS). Interestingly, FLIPL protein may be cytoprotective under these conditions, as its expression increased in the presence of low mitogen media (Fig. 1b), similar to an increase in FLIPL expression seen upon low mitogen-induced differentiation of the immortalized rat prostate epithelial cell line, NRP-152.29
TGFß1 induces apoptosis in DU145 cells
Because differentiating NRP-152 cells in low growth factor-containing media sensitizes them to apoptosis in response to TGFß1, DU145 cells grown in CS were treated with 2 concentrations of TGFβ1 and assayed for apoptosis. Although there was no significant difference in detached cells floating in the media, an indicator of apoptosis (DNS), there was a decrease in cell growth of 13% at day 3 and 22% at day 7, as assessed by counting live adherent cells (Fig. 2a). Nuclear lysates extracted from these cells exhibited biochemical evidence of apoptosis. Specifically, poly(ADP-ribose)polymerase (PARP), a substrate of caspase-3, was cleaved in cells treated with TGFβ1 for 7 days (Fig. 2b). There was no difference between 1 and 5 ng/ml TGFβ1 treatment, so all subsequent experiments were conducted at the lower concentration. DU145 cells cultured in 1% CS apoptosed at a low rate in the absence of treatment, and the extent of cell death increased with time, as assessed by FACS analysis using Annexin V and 7-AAD staining (Fig. 2c). Treatment with TGFß1 significantly enhanced apoptosis in these cultures, resulting in greater than 20% increase in apoptosis, relative to untreated time matched cultures at 10 and 14 days after treatment (Fig. 2d).
TGFß1-induced apoptosis is dependent on cell density
Because lowering exogenous growth factors was permissive for TGFß1 to exert its apoptotic effects, we hypothesized that DU145 cells might be secreting additional growth factors that inhibit apoptosis. We therefore cultured DU145 cells in 1% CS at a variety of densities, such that media conditioning by the cells is minimized,30 and then treated these cells with or without TGFß1 and assayed them for evidence of apoptosis. FACS analysis 7 days after TGFß1 treatment shows that the extent of DU145 apoptosis is inversely proportional to cell density (Fig. 3a). As we observed in the time course analysis (Fig. 2d), apoptosis progressively increased as plating density decreased, even in the absence of TGFß1 (Fig. 3b). Again, the addition of TGFß1 to these cultures significantly increased apoptosis relative to density-matched, untreated control cultures. This suggests that DU145 cells either produce growth inducing and/or apoptosis inhibiting factors, or alternatively that cell-to-cell contact is protective against cell death. To optimize both TGFß1 effects and cell recovery, all subsequent experiments were performed at an intermediate cell density (103/cm2).
TGFβ1 downregulates FLIP
FLIP is an inhibitor of cell death, and growth in low mitogen media upregulates FLIP protein in DU145 prostate cancer cells (Fig. 1b). We therefore examined TGFß1-induced changes in FLIP levels in DU145 prostate cancer cells cultured under standardized conditions (7 days in 1% CS, 103 cells/cm2; see earlier). Although culture under these conditions in the absence of TGFß1 does lead to increased levels of FLIPL protein during the first 2 days, it is maximal by day 2 and remains constant at day 4 (Fig. 1b). By day 7 in CS-containing media, the level of FLIPL protein is less than half of that at day 4. In addition, there is a time-dependent increase in caspase activity by day 4, as evidenced by the increased detection of the partially cleaved 43 kD FLIPL cleavage product. Specifically, relative to the intact 55 kD FLIPL isoform, the 43 kD protein increases 5-fold, from 3% at day 2 to ∼15% at days 4 and 7 (Fig. 4a).
We have previously shown that TGFß1 decreases FLIPL levels in differentiated NRP-152 cells.29 When DU145 cells grown in CS are treated with TGFß1, the total level of FLIPL protein capable of inhibiting caspase-8 (the 55 and 43 kD isoforms) decreases ∼50% by day 4 and ∼90% by day 7. As the amount of FLIPL protein decreases, the caspase-8 generated 43 kD isoform increases from ∼20% at 2 days to ∼30% by 4 and 7 days of TGFß1 treatment relative to the intact 55 kD FLIPL protein (Fig. 4a). There is a 5-fold increase in the 43-kD cleavage product relative to the untreated sample by 2 days TGFß1 treatment, which later declines to 2-fold, suggesting that activation of caspase-8 is an early event in TGFß1-induced apoptosis.
Finally, we sought to determine whether the increase in FLIPL protein in CS grown DU145 cells is due to an increase in FLIPL-encoding mRNA. RNA from cells grown in CS and treated with TGFß1 was analyzed by quantitative RT-PCR. The level of FLIPL-encoding mRNA remains constant in untreated cells cultured 3, 7 and 10 days in CS (Fig. 4b, black columns), suggesting the initial increase in FLIP protein (Fig. 1b) is due to changes in protein stability or mRNA translation. In contrast, the reduction of FLIPL protein in TGFß1-treated cells (Fig. 1b) appears to be secondary to a reduction in FLIPL-encoding mRNA (Fig. 4b).
FLIP expression silencing induces apoptosis in DU145 cells
The reduction of FLIP mRNA levels suggests that TGFß1 induces apoptosis at least in part by reducing the levels of FLIP, which inhibits apical caspases. We tested this hypothesis by reducing endogenous FLIP protein levels using siRNAs. DU145 cells were transfected with siRNA targeting the mRNA for the FLIPL isoform only (FLIP#1), both the FLIPL and FLIPS isoforms (FLIP#2) or #1 and #2 in combination. Annexin staining of the transfected cells increased when FLIPL targeting siRNAs were used, compared to luciferase targeting control siRNAs (Fig. 5a). When the cells were transfected and allowed to grow for 2 days, there was a 38% reduction in live cells because of FLIP knock down. When allowed to grow for 7 days, the same relative killing was seen, although total apoptosis was higher in all samples compared to 2 days (Fig. 5b). The transfection itself did not induce any increase in cell death (compare Fig. 3b, 103 cells/cm2versus Fig. 5b, control 7 days). Either siRNA alone or the combination effectively eliminated FLIPL protein at both 2 and 7 days (Fig. 5c). Therefore, reducing FLIPL protein levels, whether by siRNA knock down or TGFß1 treatment, is sufficient to induce apoptosis of DU145 cells cultured in low growth factor conditions.
Caspase gene silencing blocks TGFß1-induced apoptosis of DU145 cells
Because FLIPL inhibits the apical caspases, and because we see evidence of caspase activation by TGFß1 (Fig. 4a), we hypothesized that TGFß1 may be inducing apoptosis via caspase-8 or caspase-10. To test this, we used siRNA pools directed against either caspase alone or both in combination to reduce the apical caspase levels prior to TGFß1 treatment. DU145 cells were transfected with siRNA and 3 days later split into 2 dishes, one of which received TGFß1 and were allowed to grow for 7 days. FACS analysis shows that TGFß1 treatment of the luciferase or caspase-10 siRNA-transfected DU145 cells had decreased AnnexinV/7-AAD-staining, suggesting normal TGFß1-induced apoptosis, whereas knocking down caspase-8 blocks TGFß1-induced apoptosis (Fig. 6a). After 7 days of TGFß1 treatment, both the control luciferase siRNA and caspase-10 siRNA exhibited more than 20% reduction in live cells, whereas caspase-8-transfected DU145 cells were not significantly affected (Fig. 6b). Interestingly, caspase-8 siRNA transfection was able to block TGFß1-induced apoptosis despite only reducing the total caspase-8 protein level by 70% (Fig. 6c). These knockdown experiments suggest that caspase-8 is necessary for TGFß1-induced apoptosis, whereas caspase-10 is dispensable.
The commonly used human prostate cancer cell lines are growth inhibited by TGFß when grown in soft agar, but fail to respond to TGFß treatment when grown in monolayer culture.31 Although some prostate cancers resist TGFß-induced apoptosis because of defects in the TGFß receptor,1 growth factors have also been shown to counteract TGFß effects on prostate epithelial cell lines.25, 29, 32 In our report, we show that DU145 cells cultured in media containing CS, which has reduced growth factor levels relative to FCS, demonstrate reduced proliferation (Fig. 1a). This resembles the effects of low mitogen treatment on the rat prostate epithelial cell line NRP-152.29 Once DU145 cells have been shifted to low mitogen conditions, TGFß1 induces apoptosis as assessed by multiple criteria, including cell viability, Annexin V staining and PARP cleavage (Fig. 2). When we reduced the cell density to minimize the effect of secreted growth factors on adjacent cells,30 we observed a significant increase in the rate of TGFß1 apoptosis of DU145, but only modest affects on the survival of untreated cells (Fig. 3). Morton and Barrack have shown that FGF (and to a lesser extent EGF) is able to block TGFß-induced apoptosis, and that FGF-neutralizing antibody restores TGFß-induced apoptosis when the rat prostate cancer MatLyLu cells are grown at high density.25 Taken together, our data suggest that growth factors present in FCS stimulate DU145 cells to resist TGFß1-induced apoptosis and/or DU145 cells secrete one or more factors that inhibit apoptosis. Removing or diluting these factors causes DU145 cells to apoptose in response to TGFß1 akin to the response seen in normal prostate epithelial cells.
We have previously shown that an inhibitor of the DR signaling pathways, FLIP, is downregulated in the rat prostate after castration.27, 33 FLIP is an inactive homologue of caspase-8, an apical caspase activated in DR signaling, that acts as a dominant inhibitor of caspase-8 and thereby prevents the activation of distal caspases and cell death.34 In addition to the TGFß effects on rat prostate in vivo described in the Introduction, immortalized rat prostate NRP-152 cells (previously differentiated in mitogen-poor media) apoptose in vitro in response to TGFß1 treatment.28, 29 Similar to AWIA in the intact rat prostate gland, FLIP is downregulated in these cells by TGFß1 treatment, while enforced FLIP expression blocks apoptosis.29 Furthermore, silencing of FLIP expression using siRNA induces apoptosis, even in the absence of TGFß.29 NRP-152 cells maintained in mitogen-rich media are resistant to TGFß1-induced apoptosis.29, 35 This is due at least in part to activation of the Akt/mTOR pathway, which suppresses Smad3 activation32 and raises FLIP levels.29 FLIP is overexpressed in human prostate cancers,36 and loss of expression is associated with AWIA of the androgen-responsive human prostate cancer cell line LNCaP, when grown as a xenograft.37 In addition, in LNCaP cells, we have found that resistance to TRAIL-induced apoptosis depends on FLIP levels and that this effect is regulated by androgens as well as AKT acting via FOXO3a.33
Thus, we investigated FLIP as a regulator of TGFß1-induced apoptosis in DU145 cells. First, we observed that FLIP levels increase in DU145 cells grown in the low mitogen conditions (Fig. 1b) that enhance DU145 cell sensitivity to TGFß1-induced apoptosis. The mechanism that accounts for increased FLIP levels under these conditions is not known, but a similar increase in FLIP is seen when the NRP-152 cell line undergoes differentiation after a shift into low mitogen media.29 Notably, the kinetics of low mitogen-induced FLIP upregulation (a gradual increase over a period of days) is similar in both DU145 and NRP-152 cell lines.
Because FLIP levels are reduced after TGFß1 treatment of NRP-152 cells,29 we hypothesized that FLIP would be similarly reduced after TGFß1 treatment of DU145 cells. Indeed, there is a dramatic reduction in FLIP protein levels and this was mirrored by a large reduction in FLIP-encoding mRNA (Fig. 4). TGFß1-treated NRP-152 cells display a similar reduction in FLIP mRNA.29 Interestingly, in cells grown in CS but not treated with TGFß, we did not observe an increase in FLIP mRNA levels corresponding to the 4-fold increase seen in FLIP protein (compare Figs. 1b and 4b). This suggests that the increase in FLIP protein that accompanies the shift into low mitogen media is due to increased protein stability or mRNA translation efficiency, rather than a change in the rate of transcription or the half-life of the mRNA. Conversely, the TGFß1-induced decrease in FLIP protein is tied more closely to the abundance of FLIP mRNA.
When we silenced the expression of the FLIP gene using siRNAs, FLIP protein rapidly fell to undetectable levels and 40% of the DU145 cells apoptosed, despite the absence of added TGFß. It is important to note that FLIP siRNA mediated knockdown produced levels of cell death that were comparable to TGFß1 treatment alone (Fig. 5bversus 3b). Thus, although the data in Figure 4 provide correlative support for our hypothesis that TGFß acts via FLIP modulation, the knockdown results in Figure 5 represent strong functional evidence that FLIP plays a critical role in the induction of DU145 apoptosis. The extent of knockdown-mediated death in DU145 cells closely matches the 35% spontaneous apoptosis seen in lung carcinoma A549 cells after FLIP knockdown.38 In contrast, knockdown of FLIP in pancreatic and mesothelial cells, as well as in the human prostate cancer cell line PC3-TR, results in little spontaneous apoptosis, although it does sensitize these cells to death ligand-induced apoptosis.39–41 In addition, we previously found that in the NRP-152 cells, FLIP knockdown leads to spontaneous apoptosis, which is also enhanced by TGFß treatment.29 Taken together, these observations suggest that in prostate epithelium, FLIP could regulate apoptosis entirely by inhibiting a pre-existing death signal(s). However, TGFß might also induce DR activation via, for instance, an increase in the levels of a death ligand, such as TNF, which could synergize with FLIP downregulation. We are currently exploring this latter possibility in prostate cells.
Our observation that the decline in FLIP expression correlates with the onset of apoptosis, and that FLIP knockdown is sufficient to induce apoptosis suggests that the molecular mechanism of TGFß-induced death involves one or more of the death ligand/receptor signaling pathways. To test this idea, we used siRNA to silence caspases-8 and -10, 2 related proximal caspases that are activated by DRs,42–44 and inhibited by FLIP.45 We found that knockdown of caspase-8, but not caspase-10, abrogated TGFß1-induced death of DU145 cells (Fig. 6), providing additional functional support for our hypothesis that the TGFß induces apoptosis via FLIP. TGFß1-induced apoptosis of rat prostate epithelial NRP-152 cells also depends on caspase-8 activation, as crmA overexpression blocks cell death.29 The caspase-8 requirement also suggests that one or more of the DR ligands might be produced/secreted in response to TGFß. In our previous studies on TGFß1-induced apoptosis of NRP-152 cells, we were not able to identify a death ligand that mediated the effects of TGFß, although we did observe an increase in TNF mRNA.29 The latter observation remains unexplained, but might reflect a novel mechanism (Fig. 7). Two other cell lines that apoptose in response to TGFß, derived from a gastric carcinoma and a B-cell lymphoma, display a similar phenotype.14, 15 In both cases, death is caspase-8-dependent, but does not apparently require a secreted DR ligand. Along the same lines, it is notable that FLIP knockdown also induces apoptosis in a death ligand-independent, but caspase-8 and/or FADD dependent manner in colorectal cancer cells.46, 47 Thus, novel mechanisms that utilize caspase-8 but not secreted death ligands may also mediate TGFß-induced cell death.
Androgens regulate the growth of normal prostate and prostate cancer. Castration results in regression of the normal prostate, and androgen signaling blockade is the primary treatment for late-stage prostate cancer.48 We propose that FLIP may play a crucial role in the death and survival of prostate epithelial cells (Fig. 7).49 We and others have reported that FLIP mRNA27 and protein33, 36 levels decline in rat prostates after castration. Studies in LNCaP prostate cancer cells, which express the androgen receptor, demonstrate that the FLIP promoter contains androgen response elements and that FLIP transcription (and protein expression) is upregulated by androgens.33, 37 In addition, low-level AKT signaling, which induces FOXO3a translocation to the nucleus, is required to mediate FLIP transcription by androgens33 (not shown in Fig. 7). In this report and in our previous studies of NRP-152 rat prostate epithelial cells,29 we demonstrate that TGFß, which is produced in a paracrine manner after castration,3, 50 inhibits FLIP mRNA and protein expression. Integrating all of this data, we present a model where androgen withdrawal and paracrine TGFß provide distinct but cooperative signals to repress FLIP transcription (Fig. 7). Low levels of FLIP are permissive for the activation of caspase-8 and facilitate the induction of prostate epithelial cell apoptosis.
The authors thank members of the laboratory of Dr. D. Fruman (UC Irvine, Irvine, CA) for assistance with FACS analysis.