Inhibition of Phosphatidylinositol 3-Kinase/Protein Kinase B Signaling Is Not Sufficient to Account for Indole-3-Carbinol–Induced Apoptosis in Some Breast and Prostate Tumor Cells

Indole-3-carbinol is well known for its chemopreventive activity in vitro and in vivo, comprehensively reviewed in ref. (1), and has shown promising results in the clinic (2–4), although some concerns have been raised regarding possible tumor-promoting effects (1, 5). Despite much interest in this agent, and more recently also in its acid condensation product diindolylmethane, definitive mechanisms whereby indole-3-carbinol exerts its tumor-suppressing activity remain elusive, although several potential cellular targets have been proposed.

We showed previously that indole-3-carbinol decreased phosphorylation of protein kinase B (PKB/Akt) in MDA MB468 breast and LNCaP prostate tumor cell lines but not in the HBL100 breast or the prostate tumor DU145 cell lines. This inhibition of PKB coincided with induction of apoptosis by indole-3-carbinol. Treatment of LNCaP cells with the phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002 also resulted in apoptosis (6). We therefore suggested that inhibition of PKB may be involved mechanistically in the induction of apoptosis in these indole-3-carbinol–sensitive cell lines. Recent reports from Sarkar et al. have suggested that inhibition of PKB by indole-3-carbinol is key to the induction of apoptosis in PC3 prostate tumor cells, and together with inhibition of nuclear factor-κB signaling, is also an important event in the induction of apoptosis in MCF10CA1a tumorigenic breast cells (7, 8). However, there is no direct evidence that PKB down-regulation is functionally involved in the chemopreventive activity of indole-3-carbinol.

The PI3K/PKB signaling pathway is widely regarded as critical to cell survival and is involved in the regulation of a range of proapoptotic and antiapoptotic proteins (9–16). Deregulation of this pathway is also implicated in oncogenesis (11, 17–20) and therefore represents an attractive potential target for chemopreventive agents.

Indole-3-carbinol is known to exert cell type specific effects and to act via different mechanisms in cell lines even from the same tissue of origin such as breast and prostate (6, 8, 21–23). The aim of this study was therefore to show conclusively whether or not inhibition of the PI3K/PKB pathway that we observed previously is a critical event in the induction of apoptosis by indole-3-carbinol in the MDA MB468 breast tumor cell line.

Materials. Indole-3-carbinol, epidermal growth factor (EGF), complete protease cocktail inhibitor, phosphatidylinositol substrate for the PI3K assay, and all cell culture media and reagents were obtained from Sigma-Aldrich Co. Ltd. (Poole, United Kingdom). Medium was supplemented with FCS from Invitrogen Ltd. (Paisley, United Kingdom). LY294002 was from Promega (Southampton, United Kingdom). Antibodies against phosphotyrosine (PY99), total PKBα, and EGF receptor (EGFR) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The anti-p-PKB (Ser⁴⁷³) was obtained from Biosource (Nivelles, Belgium), and the hemagglutinin (HA) tag antibody was from New England Biolabs (UK) Ltd. (Hitchin, United Kingdom). FITC-conjugated Annexin V was from Bender MedSystems (Vienna, Austria). Genejuice transfection reagent was obtained from Novagen (Madison, WI). TLC silica gel plates (60F254) were obtained from Merck (Darmstadt, Germany).

Cell lines and treatments. The human-derived breast carcinoma cell line MDA MB468, and the immortalized nontumorigenic cell line HBL100 were cultured as described previously (24). PC3 cells were cultured in HAMS F12K supplemented with 1.5 g/L sodium bicarbonate and 10% FCS. MCF10CA1 cells (supplied by S. Santner, Karmanos Cancer Institute, MI) were cultured in DMEM/Ham's F12 in a 1:1 ratio and supplemented with 10 mmol/L HEPES, 0.029 mol/L sodium bicarbonate, and 5% horse serum. Indole-3-carbinol and LY294002 were prepared in DMSO, and cells were treated in such a way that all control and treated cells received equal volumes of DMSO, which did not exceed a final concentration of 0.1%. EGF was prepared in PBS.

Measurement of phosphatidylserine externalization. Cells were seeded at 2.5 × 10⁵, 5 × 10⁵, or 1 × 10⁶ onto 9-cm plates depending on treatment time and treated as indicated in the figures. Phosphatidylserine externalization was determined by FITC-Annexin V staining, and necrotic cells were identified by propidium iodide staining and detected by flow cytometry as described previously (6).

Measurement of caspase-3/7 activity. Caspase-3/7 activity was measured in the four cell lines following a 48-hour treatment with LY294002 or indole-3-carbinol, using the Caspase-Glo 3/7 assay kit supplied by Promega (Madison, WI) according to the manufacturer's instructions. In brief, cells were seeded at 1 × 10⁴ on a white 96-well Viewplate (Perkin-Elmer, Milan, Italy) and left to adhere overnight. Medium was removed and replaced with 50 μL medium containing 50 μmol/L LY294002, 100 or 500 μmol/L indole-3-carbinol and incubated (37°C, 5% CO₂) for 48 hours before the addition of 50 μL Caspase-Glo substrate equilibrated to room temperature. Following mixing, the plate was incubated at room temperature for 1 hour in the dark before reading luminescence on a Fluostar Optima (BMG Labtech, Offenburg, Germany). Luminescence was proportional to caspase activity and expressed in relative light units.

Immunoblotting. Cells were lysed using standard procedures, and cell extracts were analyzed by SDS-PAGE and immunoblotting followed by visualization using enhanced chemiluminescence (Amersham Life Science Ltd., Little Chalfont, United Kingdom). To detect p-EGFR, treated cells were lysed and subjected to immunoprecipitation (500 μg protein) using a phosphotyrosine antibody, immunoprecipitated proteins were then separated and detected as above using an antibody against total EGFR as described previously (25). Equal amounts of protein were loaded in each lane as determined by a Bio-Rad (Hercules, CA) protein assay. Blots were quantified by densitometry (Molecular Dynamics, Sunnyvale, CA.) using Image Quant software.

Measurement of phosphatidylinositol 3-kinase activity. PI3K activity was measured using an in vitro lipid kinase assay, based on the method described by Hawkins et al. (26). PI3K was immunoprecipitated from 500 μg cell lysate using an anti-phosphotyrosine antibody. In brief, the immunoprecipitates were then washed and resuspended in kinase assay buffer followed by addition of indole-3-carbinol (50 or 500 μmol/L) or LY294002 (50 μmol/L) directly into the kinase assay 30 minutes before the addition of 20 μL phosphatidylserine/phosphatidylinositol substrate mix (3 mg/mL). The mix was incubated at 37°C for 5 minutes before addition of 40 μL ATP mix (3 μmol/L cold ATP, 7.5 mmol/L MgCl₂, and 0.37 MBq γ³²P-ATP) and further incubation for 30 minutes at 37°C. The reaction was terminated by the addition of 450 μL chloroform/methanol (1:2) followed by 150 μL chloroform and 150 μL of 0.1 mol/L HCl. A further 150 μL chloroform and 150 μL of 0.1 mol/L HCl were added to the organic phase, which was then dried using a Savant DNA speed vac (Thermo Electric Corp., Milford, CT). The residue was resuspended in 50 μL of a chloroform/methanol/0.1 mol/L HCl mix (200:100:1) containing 0.46 KBq [³H]phosphatidyl-3-inositol. Samples were separated by TLC (silica gel plates) in methanol/chloroform/concentrated ammonia/distilled water (20:14:3:5) buffer. Plates were air-dried, and products detected by autoradiography. Identification and quantification of products was achieved both by dual scintillation counting of silica gel scraped from the chromatography plate (Beckman Coulter LS6500 multipurpose scintillation counter) and densitometric analysis of autoradiographs.

Overexpression of constitutively active or kinase-dead protein kinase B. Expression vectors encoding either myristoylated/palmitylated HA-tagged PKBα (m/p-HA-PKBα; containing the NH₂-terminal membrane localization sequence from Lck, attached to the NH₂ terminus of HA-PKBα) or a kinase-dead mutant (m/p-HA-PKBα-S473A; mutation of Ser⁴⁷³ to alanine), as described in ref. (27), were a kind gift from Prof. Dario Alessi. Cells were seeded in six-well plates at 2.5 × 10⁵ per well and allowed to adhere overnight before transfection with an expression vector (1 μg/3 μL transfection reagent) using Genejuice; control cells were incubated with transfection mix lacking plasmid DNA. Transfection was done 24 hours before treatment with indole-3-carbinol or LY294002 for 5 hours (Westerns) or 24 hours (Annexin binding). Cells were harvested for determination of PKB and HA-tag protein levels, or for the flow cytometric apoptosis assay as described above.

Statistical analysis. Statistical significance was assessed using ANOVA balanced or general linear models followed by Fisher's least significant difference post hoc test.

Effect of indole-3-carbinol on phosphatidylinositol 3-kinase activity in vitro. Following our earlier observation that indole-3-carbinol inhibited phosphorylation of PKB in MDA MB468 cells (6), we examined whether PI3K was the upstream target. Addition of indole-3-carbinol (50 μmol/L) directly into an in vitro kinase assay resulted in >70% inhibition of recombinant PI3K (p110 catalytic subunit of PI3Kγ) activity. Screening a panel of kinases as described in ref. (28), excluding SAPK3/p38γ and SAPK4/p38γ but with the addition of PI3K, CDK2/cyclin A, and dual specificity tyrosine phosphorylation-regulated kinase 1, revealed that PI3K inhibition by indole-3-carbinol was quite specific. Only two other kinases of the 25 tested showed >10% inhibition [i.e., p38-regulated/activated kinase (13%) and dual specificity tyrosine phosphorylation-regulated kinase 1A (19%); data kindly provided by Prof. Sir Philip Cohen]. The screen also confirmed our previous finding that 50 μmol/L indole-3-carbinol did not inhibit PKB activity directly (PKBα; 7% inhibition; ref. 6).

In support of these data, we showed that indole-3-carbinol directly inhibited activity of PI3K that had been immunoprecipitated from HBL100 or MDA MB468 cell lysates, using a phosphotyrosine antibody (Fig. 1). PI3K activity was reduced to 59 ± 21%, 52 ± 15%, and 21 ± 9% (P < 0.05) and 44 ± 23%, 53 ± 20%, and 23 ± 5% (P < 0.05) control levels by 50 and 500 μmol/L indole-3-carbinol and 50 μmol/L LY294002 in immunoprecipitates from HBL100 and MDA MB468 cell lysates, respectively.

Effect of LY294002 on apoptosis in MDA MB468 cells. To determine whether PI3K/PKB signaling is essential for survival of the MDA MB468 cell line, we measured induction of apoptosis following treatment with the PI3K inhibitor LY294002. Treatment with 10 μmol/L for 5 hours decreased p-PKB levels to 45 ± 11%, which is equivalent to ∼500 μmol/L indole-3-carbinol (5 hours) treatment in MDA MB468 cells (Fig. 2; Table 1). However, in the tumor cell line after 24 hours, 10 μmol/L LY294002 had no significant effect on apoptosis or necrosis (Fig. 2; Table 1). This is in marked contrast to the effect of 500 μmol/L indole-3-carbinol, which caused induction of apoptosis from 6 hours reaching 31 ± 8% by 24 hours (plus 19 ± 8% necrosis; Fig. 2; ref. 6). LY294002 (500 μmol/L) resulted in only ∼15% apoptosis at this time, despite almost eliminating p-PKB. At longer time points, 10 μmol/L LY294002 caused modest induction of apoptosis (15% at 72 hours; Table 1). These data show that down-regulation of PI3K does not cause extensive apoptosis in the MDA MB468 cell line.

Effect of epidermal growth factor receptor signaling on induction of apoptosis by indole-3-carbinol. We next investigated whether up-regulation of PKB affected induction of apoptosis by indole-3-carbinol. EGFR phosphorylation following EGF stimulation was not blocked by indole-3-carbinol in either cell line (Fig. 3). EGF increased p-PKB levels in both cell lines, albeit to a lesser extent in the MDA MB468 line with much higher basal p-PKB. Indole-3-carbinol did not prevent EGF-induced phosphorylation of PKB.

We therefore investigated whether EGF might block induction of apoptosis by indole-3-carbinol in this cell line. Treatment with 2 or 15 nmol/L EGF for 30 minutes caused only minor induction of apoptosis (<10%, measured after 24 hours). Treatment with EGF did not reverse or prevent induction of apoptosis by indole-3-carbinol (Fig. 3), and in fact, the combination of agents induced greater apoptosis than with indole-3-carbinol alone. The increase in total cell death (apoptosis plus necrosis) of the combination treatment indole-3-carbinol plus 15 nmol/L EGF was greater than additive (P < 0.05). We construe from this that down-regulation of p-PKB levels is not required for apoptosis to proceed in the MDA MB468 cell line.

Induction of apoptosis by indole-3-carbinol despite overexpression of active protein kinase B. To verify that levels of p-PKB did not correlate with the extent of apoptosis in this tumor cell line, we used another approach. MDA MB468 cells were transfected with constructs expressing an active (m/p-HA-PKBα) or a kinase-dead (m/p-HAPKBα-S473A) form of PKBα. The expressed protein contains a membrane-targeting domain, which in the case of the active form, leads to expression of high levels of PKB activity (27). Following transfection, cells showed increased expression of total PKB (6- and 5.5-fold for the active and mutant constructs, respectively). Equivalent transfection efficiency between samples is shown by the presence of comparable levels of total PKB (Fig. 4; confirmed by the presence of an HA-tag; data not shown). Twenty-nine hours after introduction of the plasmids, levels of p-PKB were raised to 210 ± 25% (P < 0.05) of nontransfected control levels in the m/p-HA-PKBα (active)–transfected cells but not in cells transfected with the mutant construct (104 ± 3%; Fig. 4). Indole-3-carbinol reduced levels of p-PKB in the m/p-HA-PKBα expressing cells by ∼25%, but levels remained considerably higher than in nontransfected cells (154 ± 9%; Fig. 4). As expected, presence of the kinase-dead protein did not affect p-PKB levels in the cells. Consistent with previous reports (27), LY294002 reduced levels of p-PKB in both transfected and nontransfected cells. Overexpression of active PKBα did not prevent the induction of apoptosis or necrosis in the MDA MB468 cell line in response to indole-3-carbinol (Fig. 4).

Inhibition of protein kinase B phosphorylation by LY294002 failed to induce significant apoptosis in other tumor cell lines. To determine whether these observations were specific to the MDA MB468 cells, we examined the effect of LY294002 in two other tumor cell types reported to respond to indole-3-carbinol. In the case of MCF10CA1a cells, significant inhibition of PKB phosphorylation occurred following treatment with 20 to 50 μmol/L LY294002 but without apoptosis as measured by Annexin V binding or caspase-3/7 activities. Treatment with indole-3-carbinol resulted in only very modest inhibition of PKB phosphorylation and very little apoptosis (Fig. 5A and C). Dose response data showed effective inhibition of PKB phosphorylation from 5 μmol/L LY294002 or 250 μmol/L indole-3-carbinol in PC3 cells but only significant cell death following indole-3-carbinol treatment (Fig. 5B). Induction of apoptosis in this cell line was more apparent with the caspase activity assay (Fig. 5C). In MDA MB468, the only cell line where significant levels of apoptosis were observed using Annexin binding, caspase activity was increased around 4-fold by LY294002 but by >10-fold with 100 μmol/L indole-3-carbinol. At the higher dose of indole-3-carbinol, only a small increase in caspase activity was indicated, but under these conditions, <10% of the cell population would have remained viable.

We have shown, using two separate assays, that indole-3-carbinol can directly inhibit PI3K activity in vitro. It is plausible that this accounts for the inhibition of PKB phosphorylation that we previously observed in the MDA MB468 breast tumor and the LNCaP prostate tumor lines (6). However, as PI3K activity was determined in vitro using cell lysates, we cannot rule out the possibility that additional mechanisms are involved in the decreased phosphorylation of PKB. If PI3K activity can be inhibited in immunoprecipitates from both MDA MB468 and HBL100 cell lines, as the in vitro data indicate, then the lack of a decrease in p-PKB in the HBL100 line presents an intriguing inconsistency, which could relate to differing levels of PI3K activity between the cell lines, differing mechanisms of PKB activation, or possibly to different uptake and processing of indole-3-carbinol by the cell lines.

The phosphatase PTEN (PTEN/MMAC1/TEP1) tumor suppressor is implicated in the regulation of PKB activity in cells. It is interesting to note that the PTEN status of a number of breast and prostate cell lines seems to correlate with their sensitivity to induction of apoptosis and inhibition of PKB by indole-3-carbinol. The more sensitive cell lines, MDA MB468 and LNCaP (6), are negative or low expressers of PTEN, whereas the less sensitive lines, HBL100, MCF10A, and DU145 are PTEN positive (6, 8). However, two sublines derived from the PTEN-positive MCF10A cell line, with premalignant (MCF10DCIS.com) and malignant (MCF10CA1a) phenotypes, were also reported to show indole-3-carbinol–induced inhibition of PKB (8). In the results presented here, inhibition of PKB phosphorylation was much more readily observed in the PC3 (lacking PTEN) compared with the MCF10CA1a cell line, but for the PC-3 cells, this did not correlate with sensitivity to indole-3-carbinol–induced apoptosis.

We previously hypothesized that the inhibition of PKB by indole-3-carbinol might contribute to the induction of apoptosis observed in the MDA MB468 and LNCaP cell lines (6), because we showed that the latter also underwent apoptosis after treatment with LY294002. However, on more detailed investigation of the MDA MB468 cell line, it was revealed that the modest down-regulation of phosphorylated PKB could not be responsible for the large amount of apoptosis observed. In these breast cells, inhibition of PI3K activity with LY294002 induced only low levels of apoptosis, indicating that this mechanism was clearly insufficient to account wholly for the much greater induction of apoptosis seen in response to an equivalent (in terms of inhibition of p-PKB), concentration of indole-3-carbinol. Failure of LY294002 to induce much apoptosis in this cell line is consistent with a previous report (29).

In support of this conclusion was the continued induction of apoptosis by indole-3-carbinol despite stimulation of p-PKB levels by EGF. The EGFR pathway has been shown to signal through the PI3K pathway in a number of cell lines and to prevent induction of apoptosis under certain conditions, primarily by up-regulating PI3K/PKB survival signaling (30–33). However, the MDA MB468 cell line, which overexpresses EGFR, has been shown to undergo apoptosis in response to treatment with EGF alone (10ng/mL, equivalent to ∼1.6 nmol/L, over 48 hours; refs. 34–36). In this study, we used 30-minute treatments of EGF, which resulted in a low level of apoptosis 24 hours later. Stimulation of PKB by EGF did not reduce the ability of indole-3-carbinol to induce apoptosis in the MDA MB468 cell line, and in fact, when both agents were given in combination (5-hour indole-3-carbinol + 30-minute EGF), induction of apoptosis was significantly enhanced compared with either agent alone, the total increase in cell death being greater than additive.

We nevertheless considered the possibility that inhibition of the PI3K/PKB pathway in the absence of EGF may still be necessary to prime cells for induction of apoptosis by indole-3-carbinol. We investigated this using a different approach, by overexpressing a membrane-targeted, constitutively activated, PKBα isoform (27). The reasoning behind this approach was that if inhibition of PKB is required for induction of apoptosis by indole-3-carbinol, overexpression of an active form should result in less apoptosis.

However, overexpression of the constitutively active form and also of a kinase-dead mutant form of PKBα had no effect on induction of apoptosis or necrosis by indole-3-carbinol in either cell line. These data led us to conclude that inhibition of PI3K/PKB signaling is not required for induction of apoptosis by indole-3-carbinol in the MDA MB468 breast tumor cell line. Our data also indicate that this pathway is not critical to the survival of this cell line.

Two recent publications proposed that inhibition of PKB signaling is a key event in the induction of apoptosis by indole-3-carbinol in PC3 prostate cells, and also, together with inhibition of nuclear factor-κB, is important in the malignant MCF10CA1a cell line (7, 8). However, it should be noted that whereas these reports show clearly that indole-3-carbinol induces apoptosis (as measured by 7-AAD staining, DNA ladder analysis, or histone/DNA ELISA) and inhibits PKB in those cell lines, together with modulation of other proapoptotic and antiapoptotic proteins, in neither study did the authors show a causative link between inhibition of PI3K or PKB and induction of apoptosis. In the present study, we subjected both these cell lines to treatment with LY294002, but while we observed good inhibition of PKB phosphorylation, this did not result in induction of apoptosis. Thus, our results indicate that inhibition of PKB signaling is not a generic mechanism by which indole-3-carbinol induces apoptosis in all tumor cell lines. In further support of a lack of a ubiquitous role for PKB in preventing apoptosis, levels of p-PKB were greatly reduced in HBL100 cells upon serum withdrawal or treatment with LY294002, without ensuing apoptosis (data not shown; ref. 25).

EGFR has also been implicated in the response to indole-3-carbinol. Rahman et al. (8) showed that indole-3-carbinol inhibits signaling through the EGFR in MCF10CA1a breast cells, whereas Chinni and Sarkar (7) showed a decrease in EGFR protein levels in PC3 cells. In the MDA MB468 breast cell line, indole-3-carbinol did not prevent EGF-stimulated PKB phosphorylation. In fact, we have recently shown that indole-3-carbinol causes a transient decrease and then an increase in phosphorylation of the EGFR followed by degradation of the receptor in these cells (21).

In conclusion, we present clear evidence that inhibition of the PI3K/PKB pathway is not sufficient to explain the induction of apoptosis by indole-3-carbinol in MDA MB468 cells. The same seems to hold true for PC3 prostate and MCF10CA1a breast tumor cells, whereas LNCaP prostate tumor cells do undergo apoptosis in response to LY294002. These findings are in contrast to the conclusion drawn in recent cited reports (7, 8). However, our data regarding sensitivity of prostate cell lines (present study; ref. 25) are in agreement with studies by Zhang et al. and Lin et al. (37, 38). A more detailed understanding of the specific nature of indole-3-carbinol activity is essential for the correct interpretation of future mechanistic studies, for the development of chemopreventive strategies involving this agent, and for understanding the tumor-preventing or tumor-promoting effects of indole-3-carbinol observed in vivo (1).

We thank Prof. Sir Philip Cohen (Medical Research Council Signaling Unit, Dundee) for the in vitro kinase assay data, Prof. Dario Alessi (School of Life Sciences Research Biocentre, University of Dundee) for the PKB constructs, and Dr. Elena Moiseeva (Cancer Biomarkers and Prevention Group, University of Leicester) for helpful advice and discussion during the article preparation.