To die, or not to die: E2F1 never decides by itself during serum starvation

Abstract

The adenovirus E2 promoter-binding factor-1 (E2F1) induces apoptosis in response to DNA damage and serum starvation. After DNA damage, E2F1 is phosphorylated by ataxia telangiectasia-mutated (ATM) kinase to promote apoptosis. However, precisely how serum starvation stimulates E2F1-induced apoptosis is unclear. We recently found that serum starvation reduces E2F1 poly(ADP-ribosyl)ation, thereby releasing a proapoptotic protein, bridging integrator-1 (BIN1), into the cytoplasm.

Abbreviations

APAF1

apoptotic protease-activating factor-1

ATM

ataxia telangiectasia-mutated kinase

BAR

BIN-Amphiphysin-Rvs-related

BIN1

bridging integrator-1

CHK2

checkpoint kinase-2

E2F1

adenovirus E2 promoter-binding factor-1

PAR

poly(ADP-ribose)

PARP

poly(adenosine diphosphate [ADP]-ribose) polymerase

RB1

retinoblastoma-susceptible protein-1

Rvs

reduced viability on starvation

siRNA

small interfering RNA

TP53

tumor protein 53 (or p53)

TP73

tumor protein 73 (or p73)

In normal or untransformed cells, serum starvation induces hypophosphorylation of retinoblastoma-susceptible protein-1 (RB1), which then directly interacts with and inhibits adenovirus E2 promoter-binding factor-1 (E2F1).¹ Because E2F1 normally acts as a G₁/S cell cycle-promoting transcription factor, serum starvation induces G₁ arrest. Under certain stress conditions, including DNA damage and serum starvation, deregulation of E2F1 (by either overexpression of E2F1 or loss of RB1) induces apoptosis. However, in cancer cells, which are frequently deficient in RB1, deregulated E2F1 actively enhances cell cycle progression but does not naturally promote apoptosis, implying that an RB1-independent and cancer-associated mechanism selectively prohibits E2F1 from inducing apoptosis.

Stevens et al. found that, in response to DNA damage, E2F1 is phosphorylated by checkpoint-2 (CHK2) kinase after the activation of ataxia telangiectasia-mutated (ATM) kinase. Subsequently, E2F1 is stabilized and triggers apoptosis.² Carnevale et al. demonstrated that, even in the presence of RB1, DNA damage-induced phosphorylation and acetylation of E2F1 allows this transcription factor to selectively induce apoptosis.³ The link between DNA damage and E2F1-induced apoptosis has since been proved and established. However, serum starvation never activates ATM or CHK2 kinase, thus the precise mechanism by which E2F1 induces apoptosis during serum starvation remained unknown.

Bridging integrator-1 (BIN1), a member of the BIN-Amphiphysin-Rvs-related (BAR) protein family, was originally identified as a MYC oncoprotein-interacting tumor suppressor.⁴ Because of its structural similarity with Rvs167 (Rvs: reduced viability on starvation), a BIN1 homolog of Saccharomyces cerevisiae,⁴ and because there is no functional or structural homolog of MYC in yeast cells, we hypothesized that BIN1 negatively regulates cell growth in response to serum starvation by a MYC-independent mechanism. BIN1-dependent suppression of the oncogenic transformation mediated by adenovirus E1A also implied a MYC-independent function of BIN1.⁴

The first clue was provided by Gu et al., who reported a new role of p107, one of the RB1-related pocket proteins, in the suppression of MYC-dependent transactivation.⁵ Weinberg also described a functional redundancy between E2F and MYC, particularly when they drive progression all the way through G₁ in the cell cycle.⁶ Therefore, we wondered whether corepressors of these functionally overlapping transcription factors reciprocally compensate for one another to some extent. To test this theory, we investigated whether the MYC-interacting corepressor BIN1 acts as an E2F corepressor. Surprisingly, the E2F1–BIN1 physical interaction in vivo was much more stable than the BIN1–c-MYC interaction. We also found that RB1 was dispensable for the E2F1–BIN1 interaction and that heterologous expression of BIN1 repressed E2F1 activity even in the absence of RB1. Conversely, transient transfection of BIN1 siRNA re-activated endogenous E2F1 activity even in the presence of RB1. These results suggested that BIN1 acts as an RB1-independent E2F1 corepressor.⁷

However, the story is a little more complicated. The presence of several E2F-binding consensus sites on the human BIN1 gene promoter prompted us to test whether E2F1 directly activates BIN1 expression. We demonstrated that the BIN1 gene, which encodes an E2F1 corepressor (see above), is a transcriptional target of E2F1.⁸ This finding suggested that deregulated E2F1 activity must be curbed by a BIN1-mediated negative feedback mechanism. Because RB1 is not required for BIN1-dependent E2F1 repression, we propose that the E2F1–BIN1 interaction serves as an RB1 fail-safe apparatus to suppress deregulated cell proliferation in RB1-deficient cells.⁷

Using a proteomic approach, we pulled down poly(ADP-ribose) polymerase-1 (PARP1), a nuclear enzyme that controls genomic stability, chromatin remodeling, gene transcription, and apoptosis, as a new BIN1-interacting protein.⁹ Because E2F1 also physically interacts with PARP1,¹⁰ we wondered whether PARP1 might have any positive or negative impact on the E2F1–BIN1 interaction in vivo. We found that PARP1 stabilizes the E2F1–BIN1 association on an E2F-sensitive gene promoter, and that PARP1 post-translationally modifies E2F1 by poly(ADP-ribosyl)ation.⁷ To determine whether E2F1 is poly(ADP-ribosyl)ated in a cell cycle-dependent manner, we synchronized growing cells in the G₁ phase by serum depletion.

Of note, serum starvation massively reduced PARP1 abundance and thus E2F1 poly(ADP-ribosyl)ation. Therefore, serum starvation alone or PARP1 inactivation alone was sufficient to disrupt the E2F1–BIN1 negative-feedback loop, thereby leading to continuous expression of E2F1-dependent proapoptotic genes including BIN1 and APAF1, which encodes apoptotic protease-activating factor-1 (APAF1). Intriguingly, once activated in PARP1-depleted cells, BIN1 never again acted as an E2F1 corepressor, primarily because E2F1 was hypo-poly(ADP-ribosyl)ated and therefore BIN1–E2F1 binding was destabilized. Moreover, in the absence of PARP1, BIN1 predominantly existed in the cytoplasm. However, because of the lack of PARP1-dependent G₂/M transition in the cell cycle, activated E2F1 did not efficiently promote the cell cycle in PARP1-deficient cells, even though E2F1 was able to induce its own cell cycle-promoting genes.⁷

Robust induction of BIN1 transcription by PARP inhibition is functionally relevant because transient transfection of either BIN1 small-interfering RNA (siRNA) or dominant-negative BIN1 significantly compromised the proapoptotic property of PARP inhibitors. We therefore conclude that release of BIN1 from hypo-poly(ADP-ribosyl)ated E2F1 into the cytoplasm is a mechanism by which serum starvation (or PARP1 inhibition) induces apoptosis (Fig. 1). Nonetheless, it should also be noted that serum depletion-induced apoptosis does not naturally occur if RB1 is fully functional, suggesting that RB1-dependent E2F1 inhibition is dominant over hypo-poly(ADP-ribosyl)ation-dependent E2F1 activation.⁷

Figure 1.

Adenovirus E2 promoter-binding factor-1 (E2F1) promotes apoptosis through two distinct pathways. (A) DNA damage activates ataxia telangiectasia-mutated kinase (ATM), which phosphorylates checkpoint kinase-2 (CHK2). E2F1 is then phosphorylated by CHK2 and induces expression of several proapoptotic genes,² including TP53 (best known as p53), TP73 (also known as p73), APAF1 (apoptotic protease-activating factor-1 gene), and BIN1 (bridging integrator-1 gene),⁸ leading to apoptosis. P: phosphorylated, T68: threonine-68, S364: serine-364. (B) Inhibition of PARP1, either chemically or through serum starvation, causes loss of poly(ADP-ribosyl)ation of E2F1, allowing the transcription factor to trigger apoptosis through the induction of cellular proapoptotic genes.⁷ PAR: poly(ADP-ribose).

Like hypoxia, serum starvation commonly occurs in solid cancer tissues, particularly after chemo- and radiation therapies or antiangiogenic therapy. Given that cancer cells eventually acquire resistance to conventional therapies, the sensitivity to apoptosis after serum starvation could be reduced in late-stage (i.e., chemo- and radiation-resistant) cancer cells. Although a number of questions remain about the actions of cytoplasmic BIN1, nuclear E2F1, and PARP1 in apoptosis during serum starvation, our study may provide a mechanistic rationale for combining standard anticancer therapy with PARP inhibitors in a wide range of various human malignancies, including triple-negative breast cancers.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported in part by the grant from the US National Institutes of Health (NIH) (R01CA140379) (to D.S.).