The Phosphatidylinositol 3′-kinase Pathway Is a Dominant Growth Factor-activated Cell Survival Pathway in LNCaP Human Prostate Carcinoma Cells

It is well established that androgenic hormones are potent and obligatory survival factors in vivo for secretory epithelial cells of the prostate gland and other male accessory sex glands. This dependence becomes apparent in the dramatic wave of apoptosis that is rapidly induced in the prostate in response to removal of circulating androgens by castration (reviewed in Ref. 1). Because androgenic hormones play a similar trophic role in hormone-dependent PCa,² androgen ablation has long been the basis for prostate cancer therapy. Androgens exert their survival function in cells of prostatic origin in part by transcriptional up-regulation of anti-apoptotic proteins such as Bcl-2 (2, 3); however, the specific mechanisms of cell survival used by prostatic cells are still poorly understood.

Cell survival signals can be transmitted by other intercellular signaling molecules, such as diffusible growth factors (4, 5, 6, 7). It is possible that the androgen-response pathway in PCa cells intersects with other survival pathways that operate in non-androgen-dependent cell types. This cross-talk may play a role in mechanisms of normal prostate growth control and in apoptosis regulation as well as in the development of the resistance of prostate cancer to androgen ablation therapy. The emergence of androgen-independent PCa, in which mechanisms of androgen-dependent cell survival must no longer play a role, is a common outcome of endocrine therapy and generally results in progression to end-stage disease.

The PI3K pathway is an important intracellular mediator of survival signals originating from trophic factors (4, 7, 8, 13). PI3K is recruited and activated by cell surface receptors for peptide growth factors and cytokines, providing a mechanism whereby exogenous peptide growth factors can directly activate intracellular cell survival pathways. PI3K phosphorylates inositol lipids that act as second messengers for several targets, including the serine-threonine Akt kinase and the ribosomal S6 kinase, pp70^S6K. The cell survival function of the PI3K pathway is mediated in part by Akt-dependent inactivation of pro-apoptotic proteins, such as the Bcl-2-related phosphoprotein, BAD, by site-specific phosphorylation (9).

In this report, we demonstrate that, in LNCaP human PCa cells, constitutive signaling through the PI3K pathway is required to avoid spontaneous activation of an apoptotic signal. LNCaP cells were not similarly dependent on the androgen-response pathway or on the Erk/MAPK pathway, an established mediator of growth factor-activated survival signals. The LNCaP cell line resembles clinical PCa in many important respects. Consequently, these data support a critical role for PI3K-dependent survival signals in human PCa.

Wortmannin and LY294002 were purchased from Biomol (Plymouth Meeting, PA). Rapamycin was from Calbiochem (San Diego, CA). DHT was from Sigma (St. Louis, MO). Casodex (Ref. 10; Bicalutamide; ICI 176,334) was a gift from Zeneca Pharmaceuticals (Cheshire SK10 2NA, United Kingdom). Charcoal-dextran absorbed serum was from HyClone (Logan, Utah). pMMTV-luciferase was a gift from Dr. Myles Brown (Dana-Farber Cancer Institute, Boston, MA). The pSV-β-galactosidase plasmid was purchased from Promega (Madison, WI). Polyclonal anti-erbB1, anti-erbB3, and anti-Akt₁ antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal anti-p85 (clone UB93–3) was from Upstate Biotechnology (Lake Placid, NY).

The human PCa cell lines LNCaP, PC3, and DU145 were purchased from the American Type Culture Collection (Rockville, MD). They were cultured in RPMI 1640 supplemented with 10% heat-inactivated FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Inc., Grand Island, NY). These supplements were used in all of the media unless otherwise indicated. Normal human prostate epithelial cells (PrEC, strain 4428) were obtained at first passage (Clonetics, San Diego, CA) and expanded in serum-free growth medium (PrEBM, Clonetics) supplemented with bovine pituitary extract, recombinant human EGF, hydrocortisone, gentamicin-amphotericin (GA-1000), tri-iodothyronine, insulin, transferrin, retinoic acid, and epinephrine. Experiments were performed on cells between passages 4 and 6. All of the cells were cultured in a humidified atmosphere at 37°C, 5% CO₂.

LNCaP cells were transiently transfected using the calcium phosphate method (11). The pMMTV-luciferase and pSV-β-gal plasmids were cotransfected for 24 h in DMEM and 10% FBS. Cells were harvested and replated in 24-well plates in RPMI 1640 and 5% FBS for 24 h. Cells were then treated with Casodex for 36 h in RPMI medium containing 5% charcoal dextran-treated FBS. Cell lysates were prepared in reporter lysis buffer (Promega). Luciferase activity was determined using a commercially available substrate (Promega). Chemiluminescence was measured with a liquid scintillation counter (LKB, Wallac, Gaithersburg, MD) for 30 s at 25°C immediately after the addition of luciferase substrate (luciferin). β-galactosidase activities were determined using the β-galactosidase Enzyme Assay System (Promega).

Cells (1.2–1.5 × 10⁶) were seeded in 6-cm dishes for 24 h and treated with the various agents in RPMI medium for different times with or without 5% FBS. For androgen depletion experiments, LNCaP cells were cultured in 5% charcoal-dextran-treated FBS for various times (at least 3 days). Cells were harvested by scraping, and low molecular weight DNA was isolated according to Herrmann et al. (12) with only slight modifications. The precipitated DNAs were separated in 1.8% agarose gels and visualized by ethidium bromide staining.

A quantitative sandwich ELISA was used to measure mono- and oligonucleosomes in the cytoplasmic fraction of cell lysates according to the manufacturer’s protocol (Cell Death Detection ELISA, Boehringer Mannheim, Indianapolis, IN). Briefly, 1.5–2.0 × 10⁴ cells/well were seeded in poly-l-lysine-coated 96-well plates for 2 days and cell lysates were collected after stimulation with growth factors and inhibitors under serum-free conditions as indicated. The amount of histone-associated DNA fragments was quantified by spectrophotometric measurement of peroxidase activity retained in the immunocomplex (415 nm). The apoptosis enrichment factor was calculated as follows:

To determine whether PI3K associates specifically with ErbB1 or ErbB3 after cell activation by EGF, 95%-confluent LNCaP cells were serum-starved for 24 h and stimulated with 10 ng/ml EGF for 10 min on ice, followed by incubation at 37°C for 5 min. Cell lysis was performed in NP40 lysis buffer [1% NP40, 10 mm Tris (pH 7.4), 10 mm NaF, 10 mm NaPP, 1 mm Na₂VO₃, and protease inhibitor cocktail (Boehringer Mannheim)]. Cell lysates (500 μg of protein) were incubated with 5 μl of monoclonal anti-p85 antibody (UBI) at 4°C overnight with gentle shaking, and immunocomplexes were captured on Protein A/G Plus (Santa Cruz). Immunoprecipitates were electrophoresed through 6% SDS-PAGE, and associated proteins were identified in Western blots using monospecific primary antibodies; proteins were visualized with the ECL Detection System (Amersham) according to the manufacturer’s specifications.

The effects of PI3K inhibitors on PI3K and Akt₁ kinase activities were tested under the same conditions as those used for cell death determinations. We chose one time point to assess kinase activity. Briefly, LNCaP cells were plated on poly-l-lysine-coated 60-mm dishes in RPMI 1640 and 10% FBS medium for 2 days. The cells were then treated with 50 nm wortmannin for 45 min followed by 10-ng/ml EGF stimulation for 10 min on ice and at 37°C for 5 min under serum-free conditions. Cell lysis was performed in NP40 lysis buffer as above. Cleared lysates were immunoprecipitated by monoclonal anti-p85 antibody (UBI). The kinase activity was measured by TLC using l-α-phosphatidylinositol (Avandi Polar Lipids, Inc., Alabaster, AL) as substrate.

Akt kinase activity was measured by the method of Crowder and Freeman (8). Briefly, the cell lysates were immunoprecipitated by anti-Akt₁ polyclonal antibody (Santa Cruz). The immune complexes were washed three times with lysis buffer, once with ice-cold H₂O, and twice with kinase buffer [20 mm HEPES (pH 7.4), 10 mm MgCl₂, 10 mm MnCl₂, 1 mm DTT, and 0.2 mm EGTA with 5 μm of ATP]; washed complexes were incubated for 30 min at 30°C in 30 μl of kinase buffer containing 1 μm protein kinase A inhibitor H89 (Biomol), 0.1 mg/ml histone H2B (Boehringer Mannheim), and 10 μCi/reaction [γ-P³²]ATP (specific activity = 3000 Ci/mmol; Amersham). Phosphorylation of histone H2B was analyzed by 15% SDS-PAGE, followed by autoradiography.

Statistical evaluation of differences between values was determined by paired Student t test.

Three human PCa cell lines—LNCaP, PC-3, and DU145—were tested for evidence of oligonucleosomal DNA fragmentation as a marker for apoptosis after treatment with two structurally distinct and specific PI3K inhibitors, wortmannin and LY294002. PI3K inhibition by wortmannin is irreversible. Extensive DNA fragmentation (laddering) was detected in androgen-sensitive LNCaP cells within 4–6 h after treatment with wortmannin or LY294002 in the presence or absence of serum (Fig. 1). DNA laddering was observed at lower concentrations of inhibitors when serum-free medium was used in place of serum-containing medium (not shown), consistent with previous reports that showed PI3K levels are reduced in cells cultured under serum-free conditions (4). DNA laddering results were verified by DNA fragmentation ELISA, which quantifies mono- and oligonucleosomes in the cytosol (Fig. 2). In contrast, apoptosis was not observed when LNCaP cells were treated with rapamycin (Fig. 1 and Fig. 2,A), an inhibitor of the kinase pp70^S6K, which acts downstream of PI3K (14); with PD98059, a specific inhibitor of the Erk/MAPK kinase, MEK (15); with culture in androgen-depleted medium for 7 days; or with Casodex, a specific inhibitor of the AR (Fig. 1). Casodex inhibits the AR—expressed by LNCaP cells—which contains a mutation in the ligand-binding domain that alters the ligand-binding properties of the receptor (10). Apoptosis was also not observed when the androgen-insensitive PC-3 or DU145 cells (Fig. 1), or when normal human prostate epithelial cells (Fig. 2 B), were treated with the PI3K inhibitors. Taken together, these results identify PI3K as an essential mediator of cell survival in LNCaP cells.

To determine whether activation of the androgen-regulated survival pathway is capable of inhibiting apoptosis induced by PI3K inhibition, LNCaP cells were treated with DHT 3 days before treatment with the PI3K inhibitors. Apoptosis was inhibited by DHT concentrations as low as 0.1 nm (Fig. 3), which indicates that survival signals are mediated by AR activation. The protective effect of DHT was not observed when wortmannin and DHT were added simultaneously or when DHT pretreatment was for less than 12 h (not shown), which suggests that the survival function of DHT is dependent on increases in gene expression, consistent with AR’s role as a transcription factor. To verify AR function in LNCaP cells, cells were transfected with an androgen-responsive pMMTV-luciferase plasmid. DHT treatment (1 nm) activated promoter activity approximately 24-fold (not shown), confirming that the AR was activated by DHT in the experiment shown in Fig. 3. Promoter activation by DHT was inhibited to basal levels by ≥ 0.5 μm Casodex, demonstrating that in the Casodex experiment shown in Fig. 1, AR function was suppressed. These results indicate that androgen exerts a protective effect downstream from the PI3K pathway; however, AR function is not essential for LNCaP cell survival.

Several EGF-like growth factors, which serve as activating ligands for the ErbB1 tyrosine kinase (EGF receptor), are expressed in the human prostate. These include EGF and TGF-α, which are expressed predominantly by ductal epithelial cells and PCa cells (16), and HB-EGF, which is expressed predominantly by smooth muscle cells of the fibromuscular stroma (17). To determine whether these growth factors can attenuate the apoptotic effects of PI3K inhibition, EGF, TGF-α, and HB-EGF were each added simultaneously with LY294002, and apoptosis was measured quantitatively by DNA fragmentation ELISA, 24 h after treatment. Apoptosis was inhibited between 50 and 70% with all of the three ligands (Fig. 4,A). A similar protective effect was observed when cells were treated with wortmannin, an irreversible PI3K inhibitor, followed by a treatment of the cells with EGF 1 h later (Fig. 4,B). Results obtained by DNA fragmentation ELISA were confirmed independently in DNA laddering experiments (Fig. 4 C). These data indicate that ErbB1 activation by ligand is capable of significantly attenuating the apoptotic effect of PI3K inhibition.

Next we sought to determine whether the cytoprotective effect observed above was due to PI3K pathway activation by the EGF-like growth factors. Previous results from our laboratory have demonstrated that EGF and HB-EGF each activate ErbB1 and ErbB3 in LNCaP cells (17). PI3K activation by ErbB1 ligands may occur by transactivation of ErbB3 by ErbB1, inasmuch as the ErbB1 cytoplasmic domain does not contain consensus binding sites for the p85 regulatory subunit of PI3K (18). Consistent with this prediction, p85 was observed to specifically associate with ErbB3, and not ErbB1, after treatment of LNCaP cells with EGF (Fig. 5, A and B). The serine/threonine kinase Akt is activated by phospholipid products of PI3K and is a prominent downstream target of PI3K in cell survival signaling (reviewed in Ref. 19). Akt activation was observed in LNCaP cells in response to EGF, TGF-α, or HB-EGF (Fig. 5 C), which suggests that the PI3K survival pathway can be activated by EGF-like growth factors in this cell line.

The above results demonstrate that EGF-like growth factors, acting through ErbB1 and ErbB3, can activate the PI3K pathway in LNCaP cells. Therefore, it is possible that activation of ErbB1 by ligand reverses the apoptotic effect of PI3K inhibition by over-riding the effects of the PI3K inhibitors. To determine whether this is the case, Akt and PI3K assays were performed after treatment of LNCaP cells with wortmannin, with wortmannin + EGF, and with EGF alone. As anticipated, wortmannin significantly suppressed both kinase activities, as seen in Fig. 6, A and B. However, differences in both the PI3K and Akt kinase activities between the wortmannin alone and the wortmannin + EGF conditions were reproducibly small or not evident (Fig. 6). These data suggest that, under conditions of irreversible PI3K inhibition, ErbB1 activation by ligand activates a PI3K-independent cell survival pathway.

The Erk/MAPK pathway is a prominent signaling pathway activated by receptor tyrosine kinases, including ErbB1 (20, 21). To determine whether survival signals under conditions of PI3K inhibition can be transmitted through the Erk/MAPK pathway, apoptosis was quantitatively evaluated by DNA fragmentation ELISA in the presence of PD98059, a specific inhibitor of MEK (a MAPK kinase) and activator of the MAPKs, Erk1 and Erk2. When LNCaP cells were treated with both wortmannin and PD98059, the cytoprotective effect of EGF was abolished (Fig. 7). Consistent with the results shown in Fig. 1, treatment with PD98059 alone did not stimulate apoptosis significantly. These results indicate that, although signaling through the Erk/MAPK pathway is not obligatory for cell survival in LNCaP, cytoprotective signals can be transmitted through this pathway in response to ErbB1 activation.

In this study, we have identified the PI3K pathway as an essential mediator of cell survival in LNCaP human PCa cells. Attenuation of signaling through this pathway by pharmacological PI3K inhibition triggered a rapid and extensive apoptotic response. Induction of apoptosis was not simply the result of a nonspecific sensitivity of LNCaP cells to treatment with kinase inhibitors because apoptosis did not result after treatment with PD98059, a potent inhibitor of the MAPK kinase, MEK, or with rapamycin, an inhibitor of the pp70 S6 kinase. Both of these kinases are important mediators of intracellular signaling. In addition, apoptosis did not occur when cells were treated with a potent and specific inhibitor of the AR or when they were cultured in androgen-depleted medium for 7 days. This finding demonstrates a lack of dependence on AR-mediated mechanisms for cell survival in this cell line, a property previously noted by others (2). We also demonstrated the existence of two receptor-dependent pathways capable of antagonizing the apoptotic effect of PI3K inhibition in LNCaP cells: (a) AR activation by the endogenous prostatic androgen, DHT; and (b) ErbB1/EGF receptor activation by three ErbB1 ligands previously shown to be expressed in the human prostate. ErbB1 activation by ligand resulted in the association of PI3K with the related receptor, ErbB3, thereby linking these two growth factor receptors to the PI3K cell survival mechanism in LNCaP cells. Recruitment of PI3K to ErbB3 after receptor activation has been observed previously (18). Involvement of ErbB3 in the transmission of cytoprotective signals is potentially physiologically relevant to PCa progression because ErbB3 overexpression has been linked to a less favorable prognosis in analyses of clinical PCa specimens (22). The three ErbB1 ligands used in this study—EGF, TGF-α, and HB-EGF—have all been shown to be present endogenously in the human prostate, as synthetic products either of ductal epithelial cells and carcinoma cells or of smooth muscle cells within the interstitial fibromuscular stroma (16, 17). These studies have demonstrated that these diffusible growth factors are PCa survival factors in addition to their role as epithelial and carcinoma cell mitogens.

Importantly, however, analysis of PI3K activity and Akt kinase activity—after wortmannin and EGF treatment—indicated that survival signals stimulated by ErbB1 ligands are also likely to be transmitted through a PI3K-independent pathway. This was demonstrated by experiments in which the protective effect of ErbB1 activation by ligand was abolished when wortmannin and PD98059 were used in combination. This finding indicates that ErbB1-activated survival signals can also be transmitted through the Erk/MAPK pathway, despite the fact that the MEK inhibitor, when used alone, did not induce apoptosis. Our findings indicate that multiple cell survival pathways, distinct from the well-known androgen-mediated survival pathway, coexist in human PCa cells and can be activated by endogenous autocrine and paracrine prostatic growth factors as well as by steroid hormones.

The PI3K pathway has been shown previously to be an essential survival pathway for a number of cell types. PI3K and downstream effectors can be activated by ligands for multiple receptor tyrosine kinases, including receptors for insulin-like growth factors, PDGF, and EGF-like factors (6, 7, 23, 24, 25). The mechanism of cytoprotection in this pathway is not fully understood; however, it seems to involve phosphorylation of the pro-apoptotic protein, BAD, by the serine-threonine protein kinase, Akt (9, 26). Phosphorylation prevents BAD from heterodimerizing with cytoprotective proteins, such as Bcl-2 and Bcl-x_L, thereby inhibiting its pro-apoptotic activity. However, as shown previously by others (4) as well as by data presented in this report, a dependence on PI3K for cell survival is a cell-specific characteristic. In fact, PI3K activation mediates ligand-dependent apoptotic signals in some cells (5, 27, 28). In our study, the androgen-independent prostate cell lines, PC-3 and DU145, and normal prostate epithelial cells obtained from a young donor, were insensitive to the apoptotic trigger. These observations suggest that the sensitivity of LNCaP cells to the PI3K inhibitors is an intrinsic property of this cell line and that normal prostate epithelial cells and some androgen-independent PCa cells are resistant to this particular apoptotic signal.

The LNCaP cell line has been of interest to prostate cancer researchers for many years. It remains the only continuous human PCa cell line that expresses a variety of properties characteristic of the human disease (reviewed in Ref. 29). PC-3 and DU145 cells do not express definitive prostatic markers, and their relevance to PCa seen clinically has been questioned by some investigators. LNCaP cells can be grown and studied as androgen-dependent, prostate-specific antigen-secreting xenografts in immune-deficient animals (30, 31, 32, 33). Recently, several groups have reported the development of novel LNCaP variants that show more aggressive in vivo behavior than the parent line, including alterations in androgen-responsiveness, metastasis to the skeleton, and formation of osteoblastic lesions at bony sites (31, 33). These recent reports have greatly expanded the capability of the LNCaP system to model the human disease. Several groups have also noted that LNCaP cells can be induced to differentiate along a neural or neuroendocrine pathway in response to a variety of signals (34, 35, 36). This is significant because there is some evidence that neuroendocrine differentiation is a marker of a more aggressive phenotype in PCa, particularly in the case of androgen-independent tumors (37). Hence, a tendency toward neural/neuroendocrine differentiation by LNCaP cells may be a reflection of in vivo properties exhibited by some clinically aggressive PCa. Interestingly, the apoptotic sensitivity of LNCaP cells to PI3K inhibitors is similar to that reported for some neural cells. Neurons, oligodendrocytes, and oligodendrocyte precursors rapidly undergo programmed cell death in response to the inhibition of PI3K using pharmacological approaches similar to the ones we used in this study (8, 38, 39). This similarity suggests the possibility that a dependence on constitutive activation of the PI3K pathway is a reflection of the neuroendocrine properties of LNCaP cells. In oligodendrocytes, neurotrophic factors were found to be incapable of inhibiting the apoptotic response to PI3K inhibition (39). In contrast, our results indicate that such a reversal is possible in LNCaP by ligand-mediated receptor tyrosine kinase activation. It is interesting to speculate that LNCaP cells, which originate from a PCa lymph node metastasis, have acquired atypical means of bypassing apoptotic signals as a result of malignant progression. Resistance to apoptosis has been observed to correlate with increasing metastatic potential in some LNCaP variants (40).

Treatment of LNCaP cells with IL-6 induces a neuroendocrine phenotype by a PI3K-dependent mechanism (36). This finding is consistent with the conclusion that the PI3K pathway plays a prominent role in the mechanisms of cell growth and differentiation in these cells. IL-6 has also been shown to activate ErbB2 and ErbB3 in LNCaP cells by a novel mechanism involving heterodimerization of the IL-6 receptor and the ErbB2 tyrosine kinase (41). Taken together with the results of our study, these findings suggest the possibility that anti-apoptotic signals may be triggered in PCa cells by inflammatory cytokines through ErbB receptor-activatable survival mechanisms. Prostate tumors are often infiltrated with inflammatory cells capable of secreting potent bioactive factors into the local environment (42).

In summary, our results have shown that several cell survival mechanisms operate in LNCaP human prostate cancer cells, a physiologically relevant cell line that resembles the human disease in important respects. These pathways can be activated by several diffusible biological mediators, including growth factors and androgenic hormones, which are present in hormonally intact and in androgen-depleted individuals. An important finding is that cell survival signals mediated by PI3K in these cells are obligatory for cell survival. It is uncertain as yet how broadly these findings can be applied to prostate cancers seen clinically; however, they suggest the possibility that targeting the PI3K pathway may prove to be an effective means of chemotherapeutic intervention in vivo.

This study was funded by the National Institutes of Health (RO1 DK47556 and RO1 CA77386) and by a grant from the Hershey Program for Prostate Cancer Research.