Beyond effector caspase inhibition

Abstract

Malignant gliomas are the most common and lethal primary central nervous system cancer. Glioblastoma mutliforme (GBM), the most aggressive of these neoplasms, are generally lethal within two years of diagnosis due in part to the intense apoptosis resistance of its cancer cells, hence poor therapeutic response to conventional and targeted therapies. Twenty years of research has uncovered key genetic events involved in disease initiation and progression, foremost the Tp53 tumor suppressor that is mutated or deleted in 35% of GBM. The prime importance of p53 signaling for gliomapathogenesis is further evidenced by epistatic genetic events targeting additional pathway components including deletion of p14^Arf (CDKN2A) and amplification of the p53-degrading ubiquitin ligases MDM2 and MDM4. Recent studies have identified and validated Bcl2-Like 12 (Bcl2L12) as a potent glioma oncoprotein with multiple strategic points in apoptosis regulatory networks, i.e. effector caspases and the p53 tumor suppressor. Bcl2L12 resides in both the cytoplasm and nucleus. In the cytoplasm, Bcl2L12 functions to inhibit caspases 3 and 7, in the nucleus, Bcl2L12 forms a complex with p53, modestly reduces p53 protein stability and prevents its binding to selected target gene promoters (e.g. p21, DR5, Noxa and PUMA), thereby inhibiting p53-directed transcriptomic changes upon DNA damage. Proteomic and multidimensional oncogenomic analyses confirmed a Bcl2L12-p53 signaling axis in GBM, as Bcl2L12 exhibited predominant genomic amplification, elevated mRNA and protein levels in GBM tumors with uncompromised p53 function. On the cell biological level, Bcl2L12 exerts robust inhibition of p53-dependent senescence and apoptosis processes in glioma cells. These multi-leveled studies establish Bcl2L12 as an important oncoprotein acting at the intersection of nuclear p53 and cytoplasmic caspase signaling and point to pharmacological disruption of the Bcl2L12:p53 complex as a promising novel therapeutic strategy for the enhanced treatment of GBM.

Introduction

Neoplasms of glial cells represent the most common tumors in the central nervous system (CNS).^1–3 They can be stratified into four different grades with distinct biological behaviors and clinical outcomes.⁴ GBM, or grade IV malignant glioma, represent the most prevalent and aggressive brain cancer with a highly progressive clinical course that culminates in death after only 1–2 years post diagnosis. GBM can manifest ‘de novo’ as primary GBM in older patients without any evidence of prior clinical disease or present as secondary GBM in younger patients evolving from progression of low-grade tumors.^1–3 Aggressive surgical resection and advanced radiation and chemotherapy have minimally impacted the median survival of GBM patients; radiation, together with the alkylating agent temozolomide, is the current standard care and imparts only modestly on median survival.⁵

The feeble impact of anti-glioma drug regimens relates to many factors. In particular, there is imperfect surgical eradication due to the invasive nature of GBM throughout the brain, the presence of an intact blood-brain-barrier, which limits drug penetration at sites distant from the primary tumor, the high degree of inter- and intratumoral genomic heterogeneity and sustained genome instability providing opportunity for the development of additional resistance mechanisms. These factors, coupled with the presence of a therapy-resistant glioma stem population, have conspired to make GBM an incurable disease.

Recent astounding progress in functional genomics and genetic model systems has positioned the field to make meaningful advances. In the area of genomics, new genome scanning and computational technologies have resulted in the identification of numerous additional glioma oncogenes and tumor suppressors as novel, putative drug targets. Advances in RNAi and viral vector technology and a growing collection of cDNAs have greatly facilitated the validation of myriad genetic elements of interest. These rich genomic and functional validation efforts have provided a host of signature genetic events that define GBM subsets and enabled the construction of refined mouse models of human GBM that have deepened our understanding of how key glioma genes operate in disease pathogenesis. For example, multidimensional oncogenomics⁶ and sophisticated genetically engineered mouse models with concomitant CNS-specific deletion of p53 and Pten^7,8 have produced penetrant high-grade malignant gliomas with clinical, pathological and molecular resemblance to primary GBM of the proneural subtype. This model afforded a system to define the cooperative interactions of p53 and Pten, revealing important roles in the regulation of Myc-directed stemness and differentiation. This model system and others, together with genomic profiles showing frequent inactivating mutations of Tp53 across multiple glioma grades,^{6, 9–11} establish p53 as a universal key tumor suppressor not only in low-grade astrocytoma and secondary GBM, but also in primary GBM tumors.

Functional genomics and subsequent cell biological validation efforts have identified a plethora of novel glioma oncoproteins and tumor suppressors that enhanced our knowledge of disease mechanisms. Here, we summarize our recent work that identified Bcl2L12, an atypical Bcl-2 family oncoprotein with potent cytoplasmic and nuclear anti-apoptotic activities, as an inhibitor of the tumor suppressive activity of p53.¹² We review experimental paradigms to assess and quantify Bcl2L12-mediated p53 inhibition in glial cells, compare Bcl2L12's modus operandi to canonical Bcl-2 family proteins and discuss future anti-glioma strategies to disrupt the Bcl2L12:p53 complex that may enhance p53-dependent apoptosis, senescence and proliferative arrest.

Bcl2L12 is a Glioma Oncoprotein Governing the Apoptosis and Necrosis Balance in Glial Cells

Bcl2L12 is a proline-rich protein characterized by a C-terminal 14 amino acid sequence with significant homology to the BH (Bcl-2 Homology) 2 domain found in several members of the Bcl-2 family.^13,14 Tissue microarray analysis validated robust Bcl2L12 protein expression in 96% of human primary GBM specimens with low or undetectable levels in cells of glial origin in normal brain surrounding tumor tissue or in low-grade astrocytoma.¹⁴ Enforced expression of Bcl2L12 in primary cortical astrocytes and transformed glioma cell lines enhanced cellular growth and in vivo tumorigenicity, conferred marked apoptosis resistance, yet engendered cellular necrosis and effected malignant transformation in cooperation with other glioma genes. On the biochemical level, Bcl2L12's oncogenic actions stemmed in part from its capacity to inhibit apoptosis by neutralizing effector caspase activity downstream of mitochondrial dysfunction and apoptosome activity. This caspase-inhibitory activity of cytoplasmic Bcl2L12 is associated with direct binding and inhibition of caspase-7¹⁴ as well as with transcriptional up-regulation of the αB-crystallin gene, encoding a small heat shock protein, which potently inhibits caspase-3^15,16 (Fig. 1, right panel). This dual inhibition of effector caspase-3 and caspase-7 downstream of mitochondrial membrane disintegration is reminiscent of Inhibitor-of-Apoptosis (IAP) molecules.^17,18 Intriguingly, while Bcl2L12 contributes to intense apoptosis resistance of GBM, the post-mitochondrial block at the level of effector caspases can shift GBM cells towards a necrotic fate. That is, under apoptosis-inducing conditions, mitochondrial dysfunction and extensive cytochrome c release impair oxidative phosphorylation and ATP production, rendering cells unable to maintain ion homeostasis and provoking cellular edema, dissolution of organelles and plasma membranes. Indeed, analysis of plasma membrane integrity and subcellular organelle morphology comprehensively demonstrated a pro-necrotic activity of Bcl2L12 in response to apoptotic stimulation supporting the general notion that apoptosis and non-apoptotic death paradigms are intertwined in GBM.^14,15 By blocking apoptosis signaling at the post-mitochondrial level and thereby redirecting the death program to necrosis, the molecular profile of Bcl2L12 provides a rational explanation for a prime paradox in GBM—apoptosis resistance yet florid necrosis—and points to Bcl2L12 up-regulation as a key progression event in malignant glioma. Together, these observations support a model where Bcl2L12 up-regulation, together with hypoxic stresses and nutrient deprivation in the tumor microenvironment caused by vascular regression/occlusion and intravascular thrombosis, promote neurologically debilitating necrosis.

Figure 1.

Nuclear and cytoplasmic anti-apoptotic activities of Bcl2L12. Bcl2L12 inhibits p53's transactivational activity and consequently abrogates transcription of selective cell cycle and apoptosis modulators, such as p21 and PUMA (left, nuclear, transcription-dependent functions). Bcl2L12's impact on p53-insigated autophagy, necrosis and metabolism-related pathways (representative targets highlighted in blue) require further studies. In the cytosol (right panel), p53 has direct apoptogenic activities at the level of mitochondria. Here, PUMA can displace p53 from an inhibitory p53:Bcl-x_L complex. Released p53 can act as a ‘BH3’-only activator of Bax/Bak to induce mitochondrial outer membrane permeabilization (MOMP) and subsequent caspase activation. Besides impacting p53, Bcl2L12 is a well-characterized inhibitor of postmitochondrial effector caspase activation, as it binds to and inhibits caspase-7 (Casp-7) and upregulates the small heat shock protein and caspase-3-specific inhibitor αB-crystallin (CRYAB). *, of note, Bcl2L12 selectively impacts p53 transcription and promoter occupancy.

While Bcl2L12's impact on effector caspases proved to be an important aspect of its oncogenicity, we considered the possibility that Bcl2L12 may have additional molecular functions. Specifically, Bcl2L12's nuclear localization and the capacity of other Bcl-2 family proteins (e.g., Bax, Bak, and Bcl-x_L) to interact with and functionally impact p53^19–23 prompted us to assess the potential for physical and functional Bcl2L12-p53 interactions.

Bcl2L12 Inhibits p53-Directed Replicative Senescence and DNA Damage-Instigated Apoptosis

Replicative senescence in serially passaged primary mouse and human cells is strongly dependent on a functional p53 pathway. Mouse embryonic fibroblast (MEF) culture have served as a murine model system to study the molecular regulators of senescence and have established the essentiality of either p53 or p19^Arf deficiencies in bypassing passage-induced senescence and maintenance of indefinite proliferative potential.^24,25 To assess whether Bcl2L12 can bypass p53-dependent senescence programs, Bcl2L12 was retrovirally transduced into low-passage wt MEFs. While vector control cultures senesced after 10–15 passages, Bcl2L12-expressing cultures showed unabated growth and attained an immortal phenotype in the absence p53 or p19^Arf extinction.¹²

p53 has many distinct functional activities, including its prominent role in apoptosis signaling. In particular, p53 acts as a cellular sentinel for DNA damage and orchestrates complex pro-apoptotic responses via transcription-dependent and independent mechanisms (see below). In this cellular program, p53 can potently transactivate the expression of a variety of apoptosis inducers, such as death receptors and pro-apoptotic Bcl-2 family proteins, and directly facilitates cytochrome c release through physical interaction with several Bcl-2 family proteins and mitochondrial membranes.²⁶ This knowledge prompted us to assess Bcl2L12's impact on p53-mediated apoptosis. We surveyed effector caspase activation in cortical astrocytes treated with the DNA-intercalating drugs doxorubicin and actinomycin D, and found robust inhibition of p53-dependent apoptosis in response to DNA damage. Together, these initial cell culture-based assays demonstrated that Bcl2L12 potently inhibited the cardinal p53-ochestrated biological processes of replicative senescence and apoptosis and pointed to p53 pathway inactivation as an additional dimension of Bcl2L12-directed gliomagenesis.¹² How then does Bcl2L12 neutralize p53 activity?

Bcl2L12 and its Distant Relatives of the Bcl-2 Family Meet p53—Once a Black Sheep, Always a Black Sheep

p53 is a transcription factor that orchestrates the expression of more than 2500 target genes to restrain tumor growth.²⁷ In recent years, this classical view of p53 action has been expanded significantly to encompass transcription-independent cyotplasmic interactions with Bcl-2 family proteins to regulate apoptosis. Specifically, transactivation-incompetent p53 mutants have been shown to promote apoptosis and suppress oncogene-induced cellular transformation of rodent fibroblasts and human osteosarcoma cells.^28,29 Notably, p53-directed apoptosis can proceed in the absence of de novo RNA and protein synthesis,³⁰ operate in enucleated cells,³¹ and can be triggered by p53-reactivating drugs under conditions of complete transcriptional or translational blockade.^31,32 Importantly, fibroblasts derived from mice engineered with a transactivation-competent p53 knock-in allele that lacks domains essential for cytoplasmic functions can undergo senescence, but fail to undergo apoptosis.³³

Mechanistically, p53 exerts its cytoplasmic and transcription-independent functions by intricately impacting mitochondrial physiology and canonical Bcl-2 family proteins as important regulators of mitochondrial membrane integrity. Anti-apoptotic members, such as prototypic Bcl-2 and Bcl-x_L, inhibit mitochondrial outer membrane permeabilization (MOMP) by sequestering their pro-apoptotic relatives Bax and Bak to prevent their oligomerization into the outer mitochondrial leaflet and subsequent formation of MOMP-triggering supramolecular complexes. ‘Activator’ BH3-only proteins, such as Bid or Bad, bind to and directly activate Bax/Bak to induce cytochrome c release and subsequent caspase-9, -3 and -7 activation. ‘Enabler’ BH3-only proteins form complexes with anti-apoptotic Bcl-2/Bcl-x_L to displace activator BH3-only proteins or even Bax or Bak themselves from an inhibitory interaction with Bcl-2/Bcl-x_L.³⁴

Chipuk et al.^19,20 recently identified the p53-PUMA signaling axis as an important modulator of MOMP. In the absence of cellular stresses/oncogene activation, p53 is not stabilized and its low expression levels are insufficient to induce PUMA transcription and consequently apoptosis. In addition, the small quantities of cytoplasmic p53 are sequestered and incapacitated by Bcl-x_L.²² Upon genotoxic stresses, nuclear p53 transcriptionally induces PUMA, which subsequently binds to Bcl-x_L, thereby liberating p53 from a p53:Bcl-x_L. Released p53 can then directly bind to and activate Bax.^19,20 Alternatively, p53 may also function as an enabler that binds Bcl-2/Bcl-x_L to release the Bax/Bak activator Bid²² or as an activator that forms a complex with Bak, thereby disrupting a Bak-inhibitory Bak:Mcl-1 heterodimer.²¹

Interestingly, nuclear magnetic resonance (NMR) together with molecular structure-function analyses identified the DNA-binding domain (DBD) of p53 as a critical binding interface for Bcl-2 proteins.²² Consequently, oncogenic hot-spot mutations affecting the DBD of p53 not only abrogate DNA binding and transactivational activity, but also hamper p53's direct apoptogenic role as a mitochondrial effector. Interestingly, cytoplasmic and nuclear functions of p53 are intertwined on different levels. The p53 transcriptional target Mdm2 is important to regulate its cytoplasmic localization through mono-ubiquitinylation³⁵ and p53-induced PUMA can release Bax and/or p53 from inhibitory Bax/Bcl-x_L and p53/Bcl-x_L complexes,¹⁹ suggesting that p53's nuclear functions are essential for its activities in the cytoplasm.

Given such prominence of Bcl-2 family proteins in regulating p53 activity, we dissected the molecular mechanisms, by which Bcl2L12 compromises p53 function. Bcl2L12 is an atypical Bcl-2-like protein with only focal homology to canonical family members. The polypeptide contains a single Bcl-2 homology domain 2 (BH2), but lacks additional BH as well as transmembrane motifs and consequently is not inducibly and constitutively associated with intracellular, most importantly mitochondrial membranes.¹⁴ Reflecting its atypical domain structure and lack of phylogenetic resemblance to canonical Bcl-2 proteins, Bcl2L12 does not safeguard mitochondrial membrane integrity but instead, as noted above, inhibits postmitochondrial effector caspase activation in an IAP-like fashion (Fig. 1, right panel).^14–16 Further reflecting such functional distinctiveness relative to prototypic Bcl-2 family proteins, Bcl2L12 limits p53 binding to selective target gene promoters and consequently reduces transcription of important p53 downstream effectors such as Puma, Noxa, DR5, p21 and cyclin G1, but not of Mdm2 (Fig. 1, left panel).¹²

The precise molecular mechanism of Bcl2L12's selective impact on the p53 transcriptome is not known. Intriguingly, the capacity of p53 to differentially impact target genes and selectively bind certain promoters is based on (1) the affinity of their p53 binding sites, (2) complex formation of p53 with cofactors, such as Mdm2/Mdm4, p63/p73, Bbp and ASPP proteins, and (3) the post-translational status of the p53 polypeptide.³⁶ Future studies aiming to further characterize the Bcl2L12:p53 complex and define essential cofactors driving selective p53 binding to promoter elements will provide important clues for understanding how Bcl2L12 impact the p53 transcriptome. In addition, it will be imperative to understand in more detail, how Bcl2L12 controls p53 post-translational modifications to drive selective p53 promoter binding, in particular acetylation/deacetylation, methylation, sumoylation and neddylation. Here, additional studies should probe further for functional and physical interactions between Bcl2L12 and post-translational modifiers of p53 such as the histone acetyltransferases CBP/p300, p300/CBP associated factor (PCAF), Tip60, hMof, acetyltransferases of the MYST family, histone deacetylase complexes such as HDAC and Sir2α/Sirt1, the methyltransferases Set7/9, Smyd2, Set8/PR-Set7, the demethylase LSD1 and PRMT, and the neddylating enzyme FBXO11.³⁶

Interestingly, Bcl2L12-driven cDNA complementation and RNAi loss-of-function studies in primary and transformed glial cells demonstrated that Bcl2L12 modestly impacts p53 protein stability. Enforced Bcl2L12 expression reduced p53 abundance with limited p53 posttranslational modifications (i.e. phosphorylation and acetylation), and stable Bcl2L12 protein knockdown triggered enhanced p53 induction with augmented acetylation and Ser/Thr-directed phosphorylation.¹² We hypothesize that reduced p53 post-translational modifications, in particular N-terminally directed Ser/Thr phosphorylation in the setting of intact Mdm2 might contribute to enhanced p53 protein degradation. While multiple in vitro studies demonstrated that p53 can be phosphorylated by a broad range of kinases, such as ATM, ATR, DNA-PK and Chk1/2 [in particular at Ser15 (mouse Ser18) and Ser20 (mouse Ser23)] to block binding to and degradation by Mdm2, it is important to note, that in vivo studies with Ser18A/Ser23A knock-in mouse models revealed only minor impact of Ser18/23 phosphorylation on p53 stability and consequently argued against phosphorylation as a general requirement for p53 stabilization.³⁶ Additional, more elaborate regulatory networks are likely operative to further regulate p53 stability. Such signaling pathways that are potentially impacted by Bcl2L12, include Mdm2 and its regulators HAUSP, the ribosomal proteins L5, 11 and 23, RASSF1A, Daxx, YY1, HLI98, and additional E3 ubiqutin ligases, such as Mdm4, COP1, Prih2 and Arf-BP1 that all can impact p53 protein stability via different modes of action.³⁶ Future studies aim to analyze Bcl2L12's impact on these multi-faceted signaling circuits that regulate p53 protein abundance.

Relationship of Bcl2L12 Expression and p53 Status in the TCGA GBM Dataset

Beyond direct mutation/deletion of p53, the tumor suppressive function of p53 can be neutralized by the E3 ubiquitin ligase Mdm2 acting to limit p53 action through ubiquitinylation and subsequent proteosomal degradation,³⁷ and the cell cycle inhibitor p14^Arf that binds Mdm2 and prevents formation of a p53-degrading Mdm2:p53 complex.^38–41 To genetically verify a p53-Mdm2-p14^Arf signaling axis in GBM, several oncogenomic and immunohistochemical surveys assessed co-amplification and -expression of p53, Mdm2 and p14^Arf in clinical samples. Specifically, studies in GBM^39,42 and other cancer types, such as sarcoma^43,44 demonstrated that Mdm2 amplification and TP53 mutation/deletion are mutually exclusive events. Additionally, p14^Arf under- and Mdm2 over-expression are inversely correlated in neuroendocrine and non-small cell lung carcinoma with an Mdm2/p14^Arf ratio > 1 indicative of a high-grade tumor phenotype.⁴⁵ Such studies genetically verified that Mdm2 and p14^Arf act in common pathway(s) to regulate p53 function. Mirroring and expanding these studies, TCGA-based oncogenomic and small-scale proteomic analyses of Bcl2L12 copy number alterations (CNA, seen as non-focal gain of chromosome 19q), mRNA and protein expression revealed higher Bcl2L12 levels in specimens with uncompromised p53 signaling and less robust chromosomal gains and expression in tumors with p53 pathway inactivation.¹²

Based on GBM's expression profiles and genetic aberrations, GBM tumors are subclassified into proneural, neural, classical and mesenchymal subtypes.⁴⁶ Deletions and somatic mutations of p53 occur predominantly in the proneural subtype. In this subtype, Bcl2L12 expression was also anti-correlated with the presence of p53 mutation further supporting a Bcl2L12-p53 signaling axis.

Conclusions and Future Directions

Continued study of Bcl2L12 has revealed a rich network of molecular interactions that impact key glioma molecules, such as effector caspases and p53 and control the hallmark tumor biological features of apoptosis resistance and pro-necrotic tendencies.^12,14–16 At the same time, many unanswered question remain, as to how Bcl2L12 expression is regulated, how it impacts BTSC biology, whether its CNS-specific deletion or transgenic expression in glioma-prone mouse strains impact gliomapathogensis in vivo, how it selectively neutralizes p53 targets, and whether it operates in controlling autophagic processes and MOMP-triggering apoptogenic functions.

The nuclear, transactivational activity of p53 represents one important facet of a rich spectrum of tumorsuppressive activities. In addition to transcriptional activation and repression of genes governing apoptosis, senescence and growth arrest, p53 also regulates autophagy, necrosis and mitochondrial membrane integrity. While a role for Bcl2L12 in autophagy and MOMP induction has not yet been established, a number of observations make these connections an intriguing possibility. Autophagy has diametrically opposing roles in cancer initiation/early progression and regulation of therapy responsiveness in established tumors. As a tumor suppressive mechanism, autophagy limits oxidative stress and chromosomal instability,⁴⁷ but represents a survival mechanism in response to anti-neoplastic agents.⁴⁸ Fascinatingly, p53 can induce and suppress autophagy: nuclear p53 can transactivate genes that induce autophagy (such as DRAM and sestrins-1 and-2), ^49–51 while cytoplasmic p53 represses autophagy via inhibition of the AMP-dependent kinase (AMPK), a positive regulator of autophagy, and activation of mTOR as a negative regulator of autophagy.⁵² Given the selective impact of Bcl2L12 on p53 promoter occupancy and gene transactivation, it will be important to determine whether and how Bcl2L12 modulates p53-directed trancriptomic changes regulating the autophagic process. Specifically, ChIP and mRNA expression analyses may illuminate whether Bcl2L12 inhibits p53-mediated induction of the autophagy inducers DRAM and sestrins. Although Bcl2L12 and p53 preferentially co-localize in the cell nucleus, it remains to be seen whether or not cytoplasmic Bcl2L12 may impact p53's ability to inhibit AMPK and mTOR and perhaps modulate p53's propensity to disrupt outer mitochondrial membranes. In this context, it will be important to assess whether PUMA can displace p53 from a cytoplasmic Bcl2L12:p53 complex in analogy to the PUMA-p53-Bcl-x_L interplay described above.

Recent studies implicated p53 in programmed necrosis, as DNA-damaged Bax/Bak-deficient cells underwent necrotic cell death via p53-induced cathepsin Q that cooperated with reactive oxygen species (ROS) to execute necrotic cell death.⁵³ Given the importance of Bcl2L12 as a post-mitochondrial apoptosis modulator with potent anti-apoptotic and pro-necrogenic functions (see above), future studies aim to decipher how Bcl2L12 impacts p53-instigated programmed necrosis. Specifically, promoter occupancy and mRNA expression analyses will determine whether and how Bcl2L12 impacts the p53-cathepsin Q signaling axis: Does Bcl2L12 modulate p53 binding to the cathepsin Q promoter? Besides cathepsin Q, does Bcl2L12 impact other p53-induced genes implicated in programmed necrosis, such as the post-mitochondrial effector XIAP that could drive necrotic cell death by blocking effector caspase activation similar to Bcl2L12?

p53 is arguably the most important tumor suppressor and its reactivation in established murine tumors could transiently halt malignant growth and, depending on the tumor type, triggers apoptosis, growth arrest or senescence.^54–56 Several therapeutic strategies have been develop to restore p53 function in glioma cell, including adenoviral gene therapy, p53-stabilizing small molecules (CP-31398, PRIMA-1, MIRA-1, Nutlins, RITA, geldamycine), and substances that induce mitochondrial translocation of p53 to trigger MOMP (CP-31398).⁵⁷ Along similar lines, the above actions of Bcl2L12 on p53 function raises the possibility of therapies that disrupt this interaction either via Bcl2L12-targeting RNAi or small molecules that may disrupt the Bcl2L12:p53 complex.