Programmed expression of pro-apoptotic BMCC1 during apoptosis, triggered by DNA damage in neuroblastoma cells

Human NB-derived SK-N-AS, NBL-S, and NLF cells were obtained from the Children’s Hospital of Philadelphia cell line bank and were maintained in RPMI 1640 culture medium (Wako, Osaka, Japan), supplemented with 10% heat-inactivated fetal bovine serum (FBS, Thermo Fisher Scientific, Waltham, MA) and 100 U/mL Penicillin-Streptomycin (Thermo Fisher Scientific). LNCaP cells were provided from Dr. Ueda [5, 25] and were cultured in RPMI 1640 medium supplemented with 5% heat-inactivated FBS and 100 U/mL Penicillin-Streptomycin. The cells were grown in 5% CO₂ and humid condition at 37 °C [5]. Cisplatin (CDDP, Sigma-Aldrich, St. Louis, MO) was used to induce DNA damage, and 2-Morpholin-a-yl-6-thianthren-1-yl-pyran-4-one (ATM kinase inhibitor, Calbiochem, San Diego, CA) was used to inhibit ATM activity. To inhibit the activity of caspases, SK-N-AS cells were treated with 40 μM of CDDP and were cultured in the presence of 50 μM of a pan-caspase inhibitor (Z-VAD-FMK, Promega, Madison, WI) or a caspase-9-specific inhibitor (Z-LEHD-FMK, R&D systems, Minneapolis, MN). After 48 h of CDDP treatment, cell viability was measured using WST-8 assay kit (Dojindo, Kumamoto, Japan), and yielded absorbance on 450 nM was detected with Microplate Reader MTP-310 (HITACHI, Tokyo, Japan). For knockdown of E2F1 using lentiviral shRNA delivery system, MISSION shRNA plasmid DNA clones (shE2F1#1 (TRC 0000000251) and shE2F1#2 (TRC 0000000252)) were purchased from Sigma-Aldrich.

Total RNA was extracted from NB cell lines using RNeasy Mini Kit (Qiagen, Hilden, Germany). The cDNA was reverse-transcribed from total RNA using SuperScript II reverse transcriptase and random primers (Thermo Fisher Scientific) and subjected to PCR-based amplification using rTaq DNA polymerase (TaKaRa, Shiga, Japan). The primer sets used for PCR were as follows:

A GeneAmp PCR 9700 and a Veriti thermal cycler (Thermo Fisher Scientific) were used for PCR, and the amplified DNA fragments were separated using agarose gel electrophoresis. Then, they were stained with ethidium bromide and detected with a UV-transilluminator (ATTO, Tokyo, Japan). The expression levels of BMCC1 in SK-N-AS cells were measured by quantitative real-time PCR (Q-PCR) using ABI PRISM 7500 system (Thermo Fisher Scientific). TaqMan probes for BMCC1 (Hs00322421_m1) and GAPDH (4310884E) were purchased from Thermo Fisher Scientific.

Whole cell lysate were immunoblotted as previously described [5]. The primary antibodies used for immunoblotting were described as follows. The rabbit antibody against human BMCC1 was generated by MBL (Nagoya, Japan) [5]. The Anti-E2F1 (#3742), anti-phospho-ATM (Ser1981) (#4526), anti-phospho-Chk2 (Thr68) (#2661), anti-phospho-H2A.X (Ser139) (#9718), anti-phospho-FOXO3a (Thr32) (#9464), anti-FOXO3a (#2497), anti-PARP (#9542, Fig. 5a and b) and anti-caspase-9 (#9502), anti-Bim (#4582), HRP-conjugated anti-Rabbit (#7074) and anti-Mouse (#7076) secondary antibodies were purchased from Cell Signaling Technology (Danvers, MA). Anti-Human phospho-E2F1 (pS364) was purchased from ROCKLAND (Limerick, PA) and anti-p73 (ab-3) was obtained from Millipore (Billerica, MA), whereas anti-ATM (sc-23,921), anti-PARP-1 (sc-8007), and anti-Actin (sc-8432) were bought from Santa Cruz Biotechnology (Dallas, TX).

Luciferase reporter plasmids harbored human BMCC1 promoter region from − 1727 to + 78 (pGL3-BMCC1-luc promoter luciferase vector), a deletion mutant of BMCC1 promoter (pGL3-BMCC1-luc △S1/S7 (△ − 1727/− 532)), and two point mutants of putative E2F1 binding constructs (pGL3-BMCC1-luc-△S1/S7 + S8 mut (− 329/− 322) and pGL3-BMCC1-luc-△S1/S7 + S9 mut (− 9/− 2)) (Fig. 3a). The luciferase reporter assay follows a previously described protocol [6].

Proteins were synthesized from the T7 promoter on pcDNA6.2-BMCC1 or pcDNA3.1-MDM2 plasmids using in vitro transcription and translation (TNT) kit (Promega). To label this protein with biotin, TNT reaction was performed in the presence of biotinylated lysine-tRNA (Promega). Biotinylated proteins reacted with recombinant active caspase-9 (Millipore, Billerica, MA) or active caspase-3 (Millipore), supplied for SDS-PAGE, transferred onto a PVDF membrane, and detected using HRP-conjugated Streptavidin. Developed signals were captured using a LAS-4000 imager.

We previously reported that BMCC1 facilitates apoptosis by neglecting multiple steps in Akt-survival pathway, such as anti-apoptotic Bcl-2 binding, inhibiting Akt and FOXO3a phosphorylation, and inducing Bim using the C-terminus BNIP-2 homology domain in human cells [5]. We also reported that acute accumulation of BMCC1 in human cells as a result of administration of DNA-damaging drugs, such as Cisplatin (CDDP) and Adriamycin (ADR), was needed for subsequent apoptosis progression [5]. However, expression profile of BMCC1 and its regulatory mechanisms during DNA damage-triggered apoptosis were unknown. Therefore, we first analyzed protein expression patterns of BMCC1 in SK-N-AS cells for up to 48 h after CDDP treatment (Fig. 1a). Consistent with our previous findings, rapid induction of BMCC1 was accompanied by initiation of DNA damage response, including phosphorylation of ATM at Ser-1981 (p-ATM), Chk2 at Thr-68 (p-Chk2), and H2A.X at Ser-139 (p-H2A.X) (Fig. 1a). Similar results were obtained in NBL-S cells (Additional file 1: Figure S1). In SK-N-AS cells with elevated expression of BMCC1, Bim induction was used to monitor CDDP-initiated progression toward apoptosis. BMCC1 protein expression decreased in SK-N-AS cells at a later step of apoptosis with elevated cleavage of caspase-9 and PARP-1 (24 to 48 h). It should be noted that a reduced expression of BMCC1 occurred along with accumulation of small fragments of pro-apoptotic Bim (15- and 12-kDa) that was mediated through cleavage by caspase-3 [26]. BMCC1 expression at mRNA levels measured by Q-PCR was also significantly increased in cells shortly after CDDP treatment (2 and 4 h) (Fig. 1b). In contrast to reducing amount of BMCC1 at protein levels, sustained expression of BMCC1 at mRNA levels was observed in apoptosis-progressed SK-N-AS cells at 36 and 48 h of CDDP treatment (Fig. 1b). Similarly, BMCC1 down-regulation was observed in NLF and LNCaP cells at 48 h of CDDP treatment in a dose-dependent manner (Additional file 2: Figures S2a and b). Furthermore, up-regulated expression of BMCC1 at mRNA levels in apoptotic cells was also confirmed in LNCaP cells (Additional file 2: Figure S2b). It should be noted that reduction in the viabilities of SK-N-AS, NLF, and LNCaP cells after 48 h of CDDP treatment was confirmed by WST assay (Additional file 2: Figure S2c).

During the early response to DNA damage, acute induction of BMCC1 was accompanied by initiation and accumulation of ATM signal pathway that includes p-ATM, p-Chk2, and p-H2A.X in SK-N-AS cells (Figs. 1a and 2a) [5]. On the other hand, induction of BMCC1 was cancelled when ATM activity was blocked using an ATM specific inhibitor, even in the presence of CDDP (Fig. 2b and c) [5]. Further research revealed that BMCC1 was trans-activated in an E2F1-dependent fashion in NB cells under normal culture conditions [6]. Therefore, these observations prompted us to examine whether the transactivation of BMCC1 could be mediated by E2F1 even in apoptotic human cells initiated by CDDP-dependent DNA damage.

E2F1 promoted apoptosis, triggered by DNA-damaging agents, was mediated via accumulation through protein stabilization, not by transcriptional induction [8‐12]. In addition to stabilization of E2F1, phosphorylation of E2F1 at Ser-364, catalyzed by Chk2 after phosphorylation by ATM/ATR, was associated with its transcriptional activity [13]. Indeed, up-regulation of E2F1 protein levels (and not mRNA levels) was detected in SK-N-AS cells shortly after CDDP treatment (Fig. 2a and b). Furthermore, phosphorylation of E2F1 at Ser-364 catalyzed by p-Chk2 was observed in these cells along with increased amounts of p-Chk2 (Fig. 2a and b). Again, when ATM-dependent DNA damage signal pathway was blocked in NB cells with an ATM inhibitor, there was a reduction in the immediate induction of BMCC1 both at mRNA and protein levels, normally followed by accumulation of E2F1 in response to DNA damage (Fig. 2b, c, Additional file 3: Figures S3a and b). On the other hand, ATM inhibitor influenced marginal effect on E2F1 mRNA expressions in NB cells with DNA damage (Fig. 2c and Additional file 3: Figure S3b). These findings suggest that transcriptional induction of BMCC1 mediated by DNA damage is closely associated with ATM-dependent stabilization of E2F1.

Our earlier study demonstrated that an E2F-binding site-9 (S9) (Fig. 3a) is essential for E2F1-dependent activation of BMCC1 promoter among nine putative E2F-binding sequences within a 1.7 kb upstream of the coding region [6]. Therefore, we wondered whether DNA damage could enhance the activity of BMCC1 promoter through putative E2F-binding sequence(s) in NB cells. To this end, the promoter activity was measured using a luciferase assay [6]. Thus, SK-N-AS cells were transfected with luciferase reporter plasmid pGL3-BMCC1-luc and Renilla luciferase reporter plasmid pRL-TK and cultured in the presence or absence of CDDP and/or ATM inhibitor.

CDDP treatment enhanced the luciferase activity driven by BMCC1 promoter, but the enhancement was completely blocked in the presence of the ATM inhibitor (Fig. 3b). These results imply that CDDP-dependent activation of BMCC1 promoter is strongly associated with ATM activity. In addition, by using deletion and mutant constructs of BMCC1 promoter (ΔS1/S7, ΔS1/S7&S8-mut, and ΔS1/S7&S9-mut, Fig. 3a), it was demonstrated that the effect of CDDP treatment on BMCC1 promoter depends on site-9 (Fig. 3c). This result was consistent with our previous finding that E2F binding site-9 was essential for BMCC1 promoter activity [6]. Therefore, these results indicate that transcriptional induction of BMCC1 mediated by DNA damage is regulated by E2F1-binding site S9 in its promoter region.

As shown in Fig. 4a, silencing of E2F1 using shRNAs (shE2F1#1 and shE2F1#2) greatly limited the accumulation of E2F1, which was initiated by CDDP-mediated DNA damage. Consistent with E2F1 suppression, induction of BMCC1 facilitated by DNA damage was undetectable in E2F1-depleted cells. Similarly, the luciferase activity driven by BMCC1 promoter increased in E2F1-non-depleted cells following CDDP exposure in comparison with that in untreated cells, whereas CDDP treatment had almost no effect on the luciferase activity in E2F1-depleted cells (Fig. 4b).

We further monitored DNA damage using elevated amounts of p-ATM, p-Chk2, and p-H2A.X. A significant attenuation of DNA damage caused by CDDP in E2F1-knockdown SK-N-AS cells was found (Fig. 4a). Because the accumulation of p-H2A.X is a marker for DNA double-strand breaks, the results suggest that knockdown of E2F1 can attenuate the cellular response toward DNA double-strand breaks.

In addition to acute induction of BMCC1 in an E2F1-dependent manner during early stages of apoptotic process initiated by DNA damage, a subsequent down-regulation of BMCC1 protein in SK-N-AS cells undergoing apoptosis was also observed, due to accumulation of cleaved forms of caspase-9 and PARP-1 (Figs. 1a and 5a).

A previous in vitro study found that variants of BMCC1, known as BNIP family (e.g., BNIP-2 and BNIP-XL), could be cleaved efficiently by several caspases [24]. Therefore, it was assumed that full-length BMCC1 can also be cleaved by caspases during apoptosis. The potential amino acid sequences within full-length BMCC1 protein cleaved by caspase activity were searched, and two caspase-9 cleavage sites, LEID (660–663) and LEED (2380-2383), were identified (Additional file 4: Figure S4). The active caspase-9 can cleave proteins with LEXD sequence [27, 28], and thus the cleavage of full-length BMCC1 by caspase-9 in the apoptotic cells was verified. A pan-caspase inhibitor (z-VAD-FMK) and caspase-9-specific inhibitor (z-LEHD-FMK) were used to inhibit caspase activation during apoptosis initiated by CDDP. SK-N-AS cells were treated with 40 μM of CDDP and cultured in the presence or absence of caspase inhibitors for 24 h (Fig. 5b). Immunoblotting results revealed that pan-caspase and caspase-9-specific inhibitors efficiently block the accumulation of cleaved forms of caspase-9 and PARP-1 in SK-N-AS cells treated with CDDP. At the same time, we found that the reduced expression of BMCC1 in the cells treated with CDDP was significantly reversed by inhibition of caspase-9. These observations indicate that reduced expression of full-length BMCC1 within apoptotic cells was strongly associated with caspase-9 activation.

Finally, we investigate whether caspase-9 could directly cleave full-length BMCC1. To this end, biotin-labeled full-length BMCC1 was synthesized using TNT system [29] and subjected to in vitro caspase proteolysis assay (Fig. 5c). As expected, the full-length BMCC1 (340 kDa) was reduced efficiently in a dose-dependent manner following incubation with a recombinant active caspase-9. The smaller fragments of 70, 60, and 45 kDa bands were observed as a result of the treatment with active caspase-9. In addition, the recombinant active caspase-3 could not cleave full-length BMCC1, while it cleaves its target MDM2 [29]. Therefore, it was affirmed that the full-length BMCC1 is proteolytically cleaved by caspase-9 during apoptosis.