Sulforaphane Suppresses Polycomb Group Protein Level via a Proteasome-Dependent Mechanism in Skin Cancer Cells

Chemicals and Reagents.

Sulforaphane was obtained from LKT Laboratories (St. Paul, MN). Trypsin and Dulbecco's modified Eagle's medium were purchased from Invitrogen (Carlsbad, CA). Lactacystin and mouse anti-β-actin were obtained from Sigma (St. Louis, MO). Mouse monoclonal anti-Bmi-1, rabbit anti-histone H3, and rabbit anti-UTX were obtained from Abcam (Cambridge, MA). Rabbit anti-procaspase 3, anti-procaspase 9, and mouse anti-procaspase 8 were from Cell Signaling Technology (Danvers, MA). Mouse monoclonal anti-Suz12 and rabbit polyclonal anti-H3K27me3 were purchased from Millipore (Billerica, MA). Mouse monoclonal antibody specific for poly(ADP-ribose) polymerase (PARP) was from BD Pharmingen (San Diego, CA). Rabbit polyclonal antibodies specific for cdc25c, cdk2, cdk4, and ubiquitin, and mouse monoclonal antibodies were specific for p21^Cip1, cyclin A, cdk1, and cyclin B1 were from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal antibody specific for Ezh2 was obtained from BD Transduction Labs (San Jose, CA). Rabbit anti-JMJD3 was purchased from Abgent, Inc. (San Diego, CA). Horseradish peroxidase-conjugated sheep anti-mouse IgG and donkey anti-rabbit IgG were obtained from GE Healthcare (Piscataway, NJ).

Cell Culture and Cell Cycle Analysis.

A431 and SCC-13 squamous cell carcinoma and HaCaT immortal but not transformed cell lines were obtained from American Type Culture Collection (Manassas, VA). The cells are maintained in a Dulbecco's modified Eagle's medium containing 5% fetal calf serum, antibiotics, d-glucose, l-glutamine, and sodium pyruvate (Balasubramanian et al., 2010). For cell proliferation assays, A431, HaCaT, and SCC-13 cells were plated at low density in 9.5-cm² dishes and allowed to attach for 24 h. Cells were then treated with fresh medium containing 0.1% DMSO or indicated concentration of SFN. After 48 h, the cells were harvested with 0.025% trypsin containing 1 mM EDTA, fixed in isotonic buffer and counted using Coulter counter (Beckman Coulter, Fullerton, CA). For cell cycle analysis, subconfluent A431, HaCaT, and SCC-13 cells were treated with SFN (0–20 μM) or vehicle for 24 h. The cells were then trypsinized, washed twice in ice-cold phosphate-buffered saline (PBS) and centrifuged (4°C). The cell pellets were resuspended in PBS and fixed in ice-cold methanol. The cells were then washed and suspended in PBS, incubated with RNase and stained with propidium iodide before flow cytometric analysis.

Bmi-1 Adenovirus Infection.

Construction of human Bmi-1 encoding adenovirus (tAd5-hBmi-1), empty virus (tAd5-EV), and the helper TA virus (Ad5-TA) are described elsewhere (Lee et al., 2008). SCC-13 cells were plated into 35-mm dishes at 20% confluence and allowed to attach. After 24 h, the cells were infected with 5 MOI of multiplicity of hBmi-1 encoding or empty virus along with 5 MOI of Ad5-TA helper virus in the presence of 2.5 μg/ml hexadimethrine (Polybrene).

Immunological Methods.

Subconfluent cells were treated in the presence of 20 μM SFN or 0.1% DMSO for 24 h. The cells were then washed in phosphate-buffered saline and suspended in lysis buffer containing 20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM glycerophosphate, 1 mM sodium vanadate, 1 μg of leupeptin, and 1 mM phenylmethylsulfonyl fluoride. Equal amounts of protein were electrophoresed on denaturing and reducing polyacrylamide gels (8–12%) and transferred to nitrocellulose membrane for immunoblot. The membrane was incubated in 10 mM Tris-HCl, pH 7.2, with 100 mM NaCl, 0.1% Tween 20, and 5% nonfat dry milk to block nonspecific binding and then with the primary antibody. Binding of the primary antibody was detected using a corresponding horseradish peroxidase-conjugated secondary antibody and chemiluminescent detection methods (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK).

Quantitative Real-Time RT-PCR.

Subconfluent cultures of SCC-13 cells were treated with 0, 15, or 20 μM SFN. After 24 h, total RNA was extracted using the RNeasy RNA isolation kit (QIAGEN, Valencia, CA). Superscript reverse transcriptase and oligo (dT) primers (Invitrogen) were used for reverse transcription from total RNA. The cDNA samples were amplified using primers specific for Ezh2 (forward, 5′-GCA TCT ATT GCT GGC ACC ATC TGA; reverse, 5′-TTG TTA CCC TTG CGG GTT GCA T), Bmi-1 (forward, 5′-TGT TCG TTA CCT GGA GAC CAG CAA; reverse, 5′-ATT GGC AGC ATC AGC AGA AGG ATG), and GAPDH (forward, 5′-TCC ACT GGC GTC TTC ACC, reverse, 5′-GGC AGA GAT GAT GAC CCT TT).

Detection of Ubiquitinated PcG Proteins.

Subconfluent SCC-13 cells were treated with 0 or 20 μM SFN for 0, 1, 4, 8, 16, and 24 h and total cells extract was prepared in lysis buffer. Total extract (30 μg) was electrophoresed for detection of total ubiquitin level. To detect specific ubiquitination of Bmi-1 and Ezh2, 75 μg of total extract was immunoprecipitated with anti-IgG, anti-Bmi-1, or anti-Ezh2, and the precipitate was electrophoresed for immunoblot with anti-ubiquitin.

SFN Suppresses SCC-13 Cancer Cell Proliferation, PcG Protein Level, and Histone H3K27me3 Formation.

SFN has been reported to suppress cancer cell survival in a number of systems (Singh et al., 2004a,b; Myzak and Dashwood, 2006). We began by examining the impact of SFN treatment on SCC-13 cell proliferation. SFN treatment suppresses SCC-13 cell number in a concentration-dependent manner and treatment with intermediate levels of SFN (10 μM) maintains cell number at the initial plating density (Fig. 1A). Because increased PcG protein expression is known to be associated with increased skin cancer cell survival (Balasubramanian et al., 2010), we determined whether SFN treatment is associated with reduced PcG expression. Figure 1B shows that the reduction in cell number is associated with a parallel reduction in Suz12, Ezh2, and Bmi-1 level. The end product of Ezh2 action is H3K27me3 formation (Pasini et al., 2004), and our studies confirm a reduction H3K27me3 level with the reduction in Ezh2 expression (Fig. 1B). Time course studies indicate that the suppression is complete within 24 h after additional of 20 μM SFN (Fig. 1C).

SFN Modulates Cell Cycle.

We next assessed the impact of SFN on cell cycle progression and apoptosis. SFN treatment causes PC-3 prostate cancer cells to arrest in G₂/M (Singh et al., 2004b), whereas G₁/S arrest is observed in DU145 and LNCaP cells (Wang et al., 2004). SFN-induced G₂/M cell cycle arrest has been reported to involve increased expression of p21^Cip1 and altered expression of cyclin B1, cdc25b, and cdc25c in different cancer cells (Gamet-Payrastre et al., 2000; Singh et al., 2004b; Herman-Antosiewicz et al., 2007). Figure 2A shows that SFN treatment of SCC-13 cells increases the number of G₂/M events from 11 to 34% and is associated with a concomitant decrease in G₁ events from 48 to 25%. In contrast, the number of S-phase cells is not altered. Figure 2B shows that this arrest is associated with a reduction in cdk1, cdk2, cyclin A, and cyclin B1. The level of cdc25c, the phosphatase that dephosphorylates and activates cdk1 (Donzelli and Draetta, 2003), is also reduced. In contrast, p21^Cip1 level is increased (Fig. 2B). These findings are consistent with cessation of cell proliferation.

SFN Affects Apoptosis.

Figure 2C shows a SFN-dependent decrease in the level of procaspase 3, 8, and 9 and an increase in PARP cleavage. The SFN impact is similar to that observed for the bioactive green tea polyphenol, (−)-epigallocatechin-3-gallate, in that it alters cell cycle regulatory protein expression to reduce cell proliferation and increases apoptosis (Balasubramanian et al., 2010). However, SFN alters the relative distribution of cells within the cell cycle, whereas (−)-epigallocatechin-3-gallate (EGCG) does not (Balasubramanian et al., 2010). Considering the strong caspase and PARP cleavage observed in 20 μM SFN treated cells, cell cycle analysis shows only a modest increase in sub-G₁ events (Fig. 2A). However, the sub-G₁ fraction is markedly increased when cells are treated for longer periods or with increased levels of SFN. Treatment with 40 μM SFN for 24 h, for example, increases G₁ content to 55%, reduces G₂ content to 28%, and increases sub-G₁ events to 11% (not shown).

Impact of Bmi-1 on SFN-Dependent Events.

The experiments in Fig. 1 demonstrate that SFN treatment reduces PcG protein level. Because PcG proteins generally promote cell survival, we examined whether maintaining PcG protein level during SFN challenge reverses the SFN-associated response. Cells were infected with tAd5-hBmi-1 or tAd5-EV for 24 h and then treated with DMSO or 15 μM SFN for 48 h. Figure 3A shows that total cell number increases during the 3-day growth period and that this is further increased by Bmi-1. Addition of 15 μM SFN to tAd5-EV-infected cells reduces cell number. It is interesting that this suppression is reversed by Bmi-1 expression. Figure 3B shows that SFN reduces Bmi-1, Ezh2, and H3K27me3 level and that these changes are reversed by maintaining Bmi-1. These studies indicate that suppression of PcG protein level and cell number is associated with SFN treatment, and these responses can be reversed by Bmi-1.

To understand the mechanism of Bmi-1 reversal of the SFN-dependent responses, we monitored the impact of Bmi-1 overexpression on key cell cycle and apoptotic endpoints. Figure 3C shows that SFN treatment increased p21^Cip1 and suppresses cdk1, cdk2, cyclin B1, and cdc25C levels, changes that are consistent with reduced proliferation. SFN also activates procaspase 3 and PARP cleavage. However, these changes are partially reversed in the presence of Bmi-1 (Fig. 3C).

SFN Increases Proteasome-Dependent PcG Protein Degradation.

A key consideration is the mechanism whereby PcG protein level is reduced by SFN treatment. The mechanism responsible for SFN-dependent loss of PcG protein expression in skin cancer cells is not known. The reduced level could be due to a direct impact on the PcG protein level or it could be associated with a reduction in mRNA level. To assess this, SCC-13 cells were treated with SFN and PcG protein and mRNA levels were monitored. As shown in Fig. 4A, SFN reduces Bmi-1 and Ezh2 protein level and reduces H3K27me3 formation. The graph compares the average protein level in control and SFN-treated cultures based on three independent experiments. Ezh2 and Bmi-1 level and H3K27me3 formation are significantly reduced in SFN treated cells. We next used quantitative PCR to compared Bmi-1 and Ezh2 mRNA level in control and SFN-treated cells. Figure 4B shows that Bmi-1 and Ezh2 mRNA levels are not altered by SFN treatment.

These findings suggest that the SFN-dependent protein reduction is due to direct effects on protein level and not mRNA level. The proteasome has been reported to be involved in PcG turnover in selected cell types (Tan et al., 2007; Dimri et al., 2010). To assess this possibility, we treated SCC-13 cells with SFN in the presence or absence of lactacystin and monitored the impact on Bmi-1, Ezh2, and H3K27me3 formation. Figure 5A shows that lactacystin inhibits the SFN-dependent loss of these markers. To provide additional evidence for proteasome involvement we treated cells with SFN for 0–24 h and monitored for the presence of ubiquitinated Ezh2 and Bmi-1. Figure 5B shows that SFN stimulates a general increase in ubiquitination. To determine whether Bmi-1 and Ezh2 are among the ubiquitinated proteins, we treated cells for 0–24 h with 20 μM SFN and prepared extract for immunoprecipitation with anti-IgG, anti-Bmi-1, and anti-Ezh2. Immunoblot of the immunoprecipitates with anti-ubiquitin reveals evidence for specific ubiquitination of Bmi-1 and Ezh2 (Fig. 5C). No ubiquitination was observed in the IgG immunoprecipitated group. The onset of increased ubiquitination was at 8 h after application of SFN.

SFN Regulation of PcG Protein Level in A431 and HaCaT Cells.

To assess whether proteasome regulation of PcG proteins is generally observed, we examined the impact of sulforaphane on Ezh2, Bmi-1, and H3K27me3 formation in A431 and HaCaT cells. We first examined the impact of SFN on cell survival to determine whether SFN reduces cell number in these cells. Figure 6A shows that A431 and HaCaT cells display dose-dependent decrease in cell survival and that this reduction is associated with decreased expression of Bmi-1 and Ezh2 and reduced H3K27me3 formation (Fig. 6B). We anticipated that these changes may also be due to proteasome-dependent degradation. We therefore treated cells with SFN in the presence or absence of lactacystin. As shown in Fig. 6C, lactacystin inhibits the SFN-dependent reduction in Bmi-1 and Ezh2 level. These results confirm that SFN-dependent suppression of PcG level is proteasome-dependent in these cell lines.

SFN Regulation of Cell Cycle and Apoptosis in A431 and HaCaT Cells.

A431 cells are tumor-forming skin cancer cells and HaCaT cells are immortalized nontumorigenic keratinocytes that have p53 mutations. SFN treatment of A431 cells increases G₂/M events from 7 to 24% with a parallel reduction in G₁ events (Fig. 7A). In HaCaT cells, SFN increases G₂/M events from 15 to 51% with a concomitant decrease in G₁ and S phase (Fig. 7A). These changes are associated with increased p21^Cip1 and a marked reduction in cdc25c (Fig. 7B). In contrast, cdk1 level is not regulated and cyclin B1 level is increased. We also examined the impact of SFN on apoptosis indicators, because sub-G1 events are increased in SFN-treated A431 and HaCaT cells (Fig. 7, A and B). As shown in Fig. 7C, we observe reduced levels of procaspase-3, -8, and -9 and also increased PARP cleavage in SFN-treated A431 and HaCaT cells.

SFN Impact on H3K27me3 Demethylases.

Jumonji domain protein 3 (JMJD3) and ubiquitously transcribed tetratricopeptide repeat, X protein (UTX) are histone demethylases that oppose PcG protein action and demethylate H3K27me3 (Agger et al., 2007, 2009; Barradas et al., 2009). We therefore examined the possibility that the change in H3K27me3 level may result, in part, from a change in expression level of these demethylases. SCC-13, A431, and HaCaT cells were treated for 24 h with 20 μM SFN before assessment of JMJD3 and UTX level. Figure 8 shows that SFN treatment reduces UTX level in all three cells lines and JMJD3 level in A431 cells. The loss of JMJD3 and UTX should lead to an increase in H3K27me3. Because the reverse is observed, it seems likely that the decrease we observe is due to loss of Ezh2.