Keywords

1 Introduction

Although the major mechanism of programmed cell death (PCD) has been characterized by activation of caspases and packaging of cells into apoptotic bodies, several types of cell death have been described with distinct features: apoptosis, necrosis, pyroptosis, and autophagic cell death [15, 47]. Prominent morphological features of apoptosis include cell shrinkage, nuclear condensation, nuclear fragmentation, chromatin condensation, membrane blebbing, retention of intact organelles, vacuole formation, and DNA fragmentation [83, 189]. Necrosis, now recognized as an alternate form of programmed cell death, often involves prominent immune responses. Recent studies suggest that by initiating inflammatory and reparative responses, necrosis may also serve to maintain tissue homeostasis and organismal integrity (reviewed in [178, 203]). In addition, autophagy plays important roles in cell death signaling in both promotional and inhibitory manners (for reviews: [55, 176]). Cell death can also be induced by microbial and viral infection. Infection induced cell death, termed "pyroptosis", involves an inflammatory response with neighboring cells [15].

A common feature of cell death is the activation of proteases followed by DNA degradation. Similar to the activation of caspases during apoptosis, other types of PCD are triggered by specific proteases. Cells undergoing necrotic death activate cathepsins and calpains, whereas pyroptotic cells activate caspase-1 and caspase-7. Furthermore, many downstream DNases have been discovered, such as Caspase-activated DNase (CAD), EndoG, LEI/L-DNase II, and Granzyme A-activated DNase (GAAD). PCD is a complex phenomenon that engages more than 400 proteins with diverse functions ranging from PCD receptors to the execution proteins.

Regulated cell death processes are critical for the development and homeostasis of multi-cellular organisms. Disruption of cell death regulation plays an important role in the pathogenesis of a wide spectrum of human diseases. Cell death regulation occurs at multiple levels and involves many different PCD associated proteins (for recent reviews, [15, 33, 60, 176]). One important step of this regulation occurs during pre-mRNA splicing because many cell death genes are expressed as functionally distinct or even antagonistic isoforms as a result of alternative splicing. Genes implicated in autophagy have been reported to undergo alternative splicing, including autophagosome markers (reviewed in [188]), although the functional significance remains to be investigated.

In this chapter, we review recent molecular and biochemical studies on alternative splicing of genes involved in regulating PCD and illustrate the current PCD paradigms. These genes range from extracellular signals and death receptors to the intracellular components of the PCD machinery. It is possible that alternative splicing profoundly contributes to the life-or-death decision of cells. Because of the large number of different PCD genes that undergo alternative splicing, it is difficult to cover all relevant aspects of these genes within this short chapter. We will describe selected examples to illustrate the extreme complexity of alternative splicing in cell death regulation and its impact on cancer pathogenesis and treatment.

2 Pre-mRNA Splicing and Alternative Splicing Regulation

In mammals, nascent transcripts (messenger RNA precursor, or pre-mRNAs) produced from the vast majority of protein-encoding genes contain intervening sequences or introns, that must be accurately removed to form functional messenger RNAs. In addition, it is now estimated that more than 90 % of human protein-coding genes undergo alternative splicing to generate distinct transcripts [114, 142], providing a robust post-transcriptional mechanism for increasing genetic diversity. Mammalian gene regulation is a highly complex process, involving multiple interconnected networks of regulation [184]. Disruption of such a regulated process leads to a wide range of diseases including cancer [36, 165].

Alternative splicing is regulated by intricate interactions between cis-regulatory elements and splicing machinery. There are four types of cis-acting regulatory sequence elements: splicing enhancers (SE) in introns (ISE) or exons (ESE) and splicing silencers (SS) in introns (ISS) or exons (ESS) [177]. Splicing enhancers recruit splicing activators to the spliceosome to stimulate splicing, whereas splicing silencers block the interaction between the spliceosome and the corresponding splice site. Multiple families of splicing factors have been identified with characteristic sequences such as the RNA-recognition motifs (RRM) and serine/arginine-rich (SR) proteins among others [124].

3 Alternative Splicing Regulation of Cell Death Genes

Cell death is a tightly regulated process. The components of the cell death machinery that control and execute this process are under strict regulation as well. At the post-transcriptional level, alternative splicing is one of the most powerful and diverse mechanisms that regulate the expression and function of many PCD genes. A large number of players function together in a carefully orchestrated manner to control cell death.

3.1 Alternative Splicing Isoforms of Genes Encoding Caspases and Other PCD-Related Proteases

Caspases are a family of cysteine proteases that play important roles in cell death regulation and cytokine secretion. Activation of pro-apoptotic caspase (caspase 2, 3, 6, 7, 8, 9, and 10) is a central pathway for different death signals (for reviews; [104, 156]). Each caspase has different sets of isoforms as a result of alternative splicing ([74, 75, 76, 95, 128], for reviews see [185, 190]). Many caspase genes produce pro- and anti-PCD isoforms as a result of alternative splicing, further support that alternative splicing is a common mechanism for regulating caspase gene expression [74, 85].

Caspase-1 is also known as IL-1β converting enzyme (ICE) that processes pro-inflammatory cytokine precursors pro-IL-1β and pro-IL-18. Recent studies show that caspase-1 also activates pro-apoptotic caspase-7 by processing pro-caspase-7 ([3, 105], for review [55]). Caspase-1 is activated by various cell death stimuli, such as bacterial toxin, LPS, microbial/viral nucleic acids, viral infection, reactive oxygen species (ROS), and asbestos. Five splicing isoforms of caspase-1 (caspase 1α − 1ε) have been reported. Among these, caspase-1δ and -1ε display an anti-apoptotic effect [6, 52].

Alternative splicing of caspase-2 generates five splicing isoforms: caspase-2L, caspase-2S, caspase-2β, caspase-2L-Pro, and caspase-2S-Pro. Caspase-2β, a possible negative regulator of caspase-2 activity, contains a deletion downstream of the first potential aspartic proteolytic cleavage site between the large and small subunits [5]. In addition, caspase-2S has anti-apoptotic activity [14, 46]. Interestingly, caspase-2S is abundant in the developing brain, suggesting that caspase-2 alternative splicing is developmentally regulated [14].

In caspase-8, retention of intron 8 results in the formation of caspase-8L mRNA and the caspase-8L protein that lacks the C-terminal half of the proteolytic domain. Patients with systemic lupus erythematosus (SLE) also display alternative splicing of caspase-8 [76]. Overexpression of caspase-8L inhibits Fas-mediated apoptosis, suggesting dominant negative effects of caspase-8L [75].

In addition to caspase genes, other protease genes involved in cell death also undergo alternative splicing. Calpains are Ca2+-regulated proteases important for apoptotic, necrotic, and pyroptotic cell death. M- and μ-calpain are expressed ubiquitously and share the regulatory small subunit called calpain 4. Cells lacking calpain 4 exhibit resistance against certain PCD stimuli. In addition, an intrinsic calpain inhibitor, calpastatin, has at least eight splicing isoforms, some of which exhibit tissue-specific expression [66]. The role of alternative splicing in regulating calpains and calpastatin remains to be further characterized.

Cathepsins are a family of lysosomal proteases participating in the hydrolysis of macro-molecules in lysosomes. Cathepsins are released into the cytosol by various cell death signals and activate downstream PCD signaling. For example, Bid, a pro-apoptotic Bcl-2 family member, is cleaved by cathepsins released from lysosomes and translocates from the cytosol to the mitochondria, inducing mitochondrial death. Cathepsins activate other PCD associated DNases. Cathepsin leakage from lysosomes is observed during necrosis and pyroptosis as well [192].

The cathepsin B gene generates at least 3 splicing isoforms. Exon 2, which is an alternative cassette exon located at the 5′ untranslated region (UTR) of cathepsin B full-length mRNA, down-regulates cathepsin B expression [203]. On the other hand, alternative splicing also modulates the subcellular localization of cathepsin B. Exclusion of exons 2 and 3 results in a protein with a different N-terminus that serves as a mitochondrial targeting sequence [135]. This isoform contains a catalytic domain but does not have protease activity possibly because of misfolding. Overexpression of this isoform changes the mitochondrial morphology and induces cell death. Therefore, alternative splicing may trigger cell death by changing the cellular localization of cathepsin B.

3.2 Bcl-2 Superfamily

The Bcl-2 gene family of proteins mediates a complex network of interactions among different pro- and anti-apoptotic proteins as well as downstream molecules [99, 112]. They play important roles in controlling mitochondrial permeability, cytochrome C release, and caspase activation. The human Bcl-2 family has multiple members including pro- and anti-apoptotic genes. Prototypical anti-apoptotic members including Bcl-2, Bcl-w, Bfl-1/A1, Mcl-1, and Boo/DIVA usually contain four Bcl2-homology (BH) (BH1-4) domains. Pro-apoptotic members, on the other hand, contain two or more BH domains (BH1-3) (Bax, Bak, Bok/Mtd, Bcl-G, Boo/Diva, and Bfk) or only the BH3 domain (BH3-only protein family). The list of BH3-only protein family members has continuously grown (Bad, Bid, Bik, Bim, Blk, Hrk, Noxa, BNIP1, BNIP3, BNIP3L, Bmf, and Beclin-1). Furthermore, many proteins containing BH3-like domains have been reported (APO L1, APO L6, BRCC2, HER2, HER4, MAP-1, MULE, SPHK2, Rad9, and TGM2). These proteins have death inducing activity as well as the ability to interact with the anti-apoptotic Bcl-2 family as well as BH3-only proteins. Beclin-1, for example, is a critical effecter of autophagy. Bad and Bik can induce autophagic cell death.

The role of alternative splicing in the Bcl-2 family gene regulation has not yet been fully elucidated. Several genes of this superfamily including Bcl-x, Bak, Mcl-1, and Bid encode for anti-apoptotic (Bcl-xL, N-Bak, Mcl-1L, BidS) and pro-apoptotic (Bcl-xS, Bak, Mcl-1S, Bid) isoforms as a result of alternative splicing [10, 18, 154]. Bid is a prominent member of the BH3-only family that bridges the death receptor signaling pathway to the mitochondrial signaling pathway, which is mediated by other Bcl-2 family member proteins. Alternative splicing of Bid pre-mRNA generates isoforms lacking proteolytic cleavage sites (BidES, lacking all cleavage sites; BidS, lacking the granzyme B cleavage site) [110, 117]. In addition, some of Bcl-2 family proteins have a transmembrane domain that can be removed by alternative splicing, leading to differential subcellular distribution. Therefore, alternative splicing regulation of Bcl-2-related proteins has a considerable impact on cell death signaling pathway.

3.3 Death Ligands and Death Receptors

Death ligands and their receptors have been studied since the discovery of FAS, a member of the TNF (Tumor Necrosis Factor) superfamily. TNF-α, TNFSF1/Lynphotoxin-α, TNFSF3/Lynphotoxin-α, TRAIL, VEGI, TWEAK, and LIGHT proteins mediate apoptosis. TRAIL (TNF-related apoptosis-inducing ligand) has three splicing isoforms: TRAIL-α, β, and γ. TRAIL-α, the full-length form, contains five exons and promotes apoptosis. In contrast, apoptotic activity is not observed when exon 3 (TRAIL-β) or exons 2 and 3 (TRAIL-γ) are skipped [100]. In addition to TRAIL, splicing isoforms of TNFSF1 and VEGI are reported [26, 170]. In naïve T cell lymphocytes, TNF-α transcripts do not undergo splicing and are instead stored as pre-mRNA. TNF-α pre-mRNA splicing is initiated to produce TNF-α protein following the engagement of T cell receptors [195].

In addition to the TNF superfamily, other ligands, such as growth factors, neurotrophic factors, and cytokines participate in cell death regulation. Several neurotrophic factor genes of the TGF-β superfamily contain introns in their prodomains, including persephin, neurturin (NTN), and glial cell line–derived neurotrophic factor (GDNF). The GDNF gene produces a short isoform by removing 78 bases at the end of exon 1. The short isoform GDNF∆78 accumulates in the Golgi apparatus and secretion of GDNF∆78 is repressed [182]. It is conceivable that activation of this splicing event occurs in response to certain stimuli, providing an efficient post-transcriptional mechanism controlling the production of functional mRNAs for trophic factors such as NTN and GDNF.

Alternative splicing of transmembrane receptors may lead to the formation and secretion of the soluble receptors of cell death related ligands. In fact, many transmembrane receptors have splicing isoforms that encode soluble proteins capable of modifying the downstream ligand effects in antagonistic or agonistic manners [108]. For example, FAS has multiple alternative splicing isoforms lacking the transmembrane domain [25, 145]. Alternative splicing of Fas exon 6 produces a soluble (lacking exon 6) or a membrane-bound (containing exon 6) protein product. Soluble FAS proteins suppress PCD induced by FAS ligand and participate in T cell regulation, immune diseases, and cancer development. A membrane-bound form of the FAS receptor found in thymocytes lacks exon 7 and is missing the death domain (DD) due to the frame shift caused by exon 6 retention. This FAS alternative splicing isoform acts as a FAS decoy receptor (FDR) because it blocks PCD induced by FAS ligands as well as soluble FAS.

Other death receptors have multiple splicing variants: tumor necrosis factor receptor 1 (TNF-R1), lymphocyte-associated receptor of death (LARD, also named as DR3, Apo-3 or TRAMP), death receptor-4 (DR4), and TRAIL receptor inducer of cell killing-2 (TRICK-2) [101]. Additionally, many cytokine receptor genes are alternatively spliced to generate functionally diverse receptor isoforms. For example, alternative splicing generates truncated interleukin 7 (IL-7) receptor isoforms associated with leukemia [180]. Therefore, alternative splicing modulates the expression and function of genes encoding both death ligands and death receptors.

PCD signals from extracellular ligands are often transduced by proteins containing death domains (DDs) and death effector domains (DEDs). DD and DED are homophilic protein interaction motifs that allow association between different proteins containing DDs or DEDs. For example, FAS recruits FADD through an interaction between their respective DDs. The DED of FADD in turn recruits caspase 8 via interaction between DEDs of the FADD and caspase 8. In addition, some DD and/or DED containing proteins modulate cell death signaling. Two DEDs are located at N terminus of caspase 8 and 10. Viral proteins named v-FLIPs have been found in equine herpesvirus-2, bovine herpesvirus-4, herpesvirus saimili human herpesvirus-8, and human poxvirus. These v-FLIP proteins show homology to the N-terminus of caspase 8/10 but lack the cysteine protease domain. Viral-FLIP proteins bind to FADD and/or caspase 8/10 and interfere with their recruitment to Fas-FADD receptor complex. Cellular FLIP (c-FLIP, also known as Casper/I-FLICE/FLAM/CASH/CLARP/MRIT/usurpin) was discovered based on viral protein sequence information. The c-FLIP gene generates three splicing isoforms—FLIPL, FLIPS, and FLIPR by alternative splicing [44, 65]. The sequence of FLIPL protein is similar to caspase 8, including the two aspartic acid residues at the auto-cleavage site, although it lacks caspase enzymatic activity. FLIPS and FLIPR lack a cysteine protease-like region, similar to v-FLIP. FLIPL is expressed in many tissues, whereas FLIPS is detectable mainly in T lymphocytes. FLIPS expression in T cells activated by CD3 and CD28 has been shown to correlate with the cellular resistance against FAS-mediated apoptosis [94]. On the other hand, IL-2 upregulates FAS expression and represses the transcription and expression of FLIPS in T cells, sensitizing T cells to apoptosis [152]. These results suggest important roles of FLIPS in activation-induced cell death (AICD) and self-tolerance. A similar expression pattern of FLIPR and FLIPS in CD3 and CD28-activated T cells has been reported [65]. Although the anti-apoptotic effects of FLIPL and FLIPS are reported, FLIPL also stimulates apoptosis under certain conditions. FLIPL can activate caspase 8 after the formation of a heterodimer on FADD and death receptor complexes. It has been proposed that activation of caspase 8 by FLIPL results in partial autocleavage of caspase 8 into the p43/41 and p12 subunits. This mechanism tethers caspase 8 on the death-induced signaling complex (DISC) by FADD and processes receptor-interacting kinase (RIP) within the FAS-signaling complex [125]. FLIPL may generate proliferation signals from DISC by processing DISC-proximal substrates, as a result of activation of ERK and NFkB by FLIPL in T cells [91].

The caspase activation and recruitment domain/caspase recruitment domain (CARD) is engaged in homophilic interactions with CARD containing caspases. TUCAN/CARD8/CARDINAL, cloned as an anti-apoptotic CARD containing protein, is overexpressed in colon cancer [147]. TUCAN can interact with caspase-9 via the CARD domains and interfere with the binding between Apaf-α and caspase-9. Interestingly, TUCAN also induces cell death when overexpressed in MCF7 and VERO cells. A TUCAN splicing isoform, TUCAN-54, has been recently reported as a cancer-specific protein expressed in gastric, colon, and breast cancer tissues. TUCAN-54 exhibits anti-apoptotic activity by inhibiting caspase-9 and caspase-8 activation. TUCAN along with caspase-9 and DRAL (a LIM domain protein) are recruited by the Shh receptor, Patched, to enhance cell death [127]. Therefore, TUCAN alternative splicing regulation may be important both in normal development and in cancer pathogenesis.

MAPK-activating death domain-containing protein (MADD) is a Rab3 GTP/GDP exchange protein containing a DD motif. There are six reported isoforms: IG20pa, MADD, IG20-SV2, DENN-SV, KIAA0358, and IG20-SV4 [111]. IG20pa render cells more susceptible to PCD by TNF-α, TRAIL, γ-irradiation, and cancer drugs; however, DENN-SV has an opposite effect, promoting cancer cell suvival [7, 48, 49, 102, 136, 137].

3.4 Intrinsic Cell Death Signals

The death domain mediates not only receptor-mediated cell death signals, but also intrinsic cell death signals. PIDD (P53-Induced Protein with Death Domain), which is induced by p53, interacts with caspase 2 by RAIDD bridging, and activates caspase-2, forming a protein complex known as PIDDosome. Binding between PIDD and RAIDD is mediated by a homophilic interaction of their DDs. Caspase 2 is recruited via an interaction between the CARDs of caspase 2 and RAIDD. Among the five reported splicing isoforms of PIDD, isoform 2 (also known as LRDD) counteracts the pro-apoptotic activity of full-length PIDD isoform 1 [35]. Similarly, Apaf-1 recruits caspase 9 with cytochrome C and dATP and forms apoptosome. Apaf-1 gene also generates pro- and anti- apoptotic splicing isoforms [140].

Similar to PIDDosome, NALP1 (NACHT leucine-rich-repeat protein 1) interacts with caspase 1 via ASC (Apoptosis-associated speck-like protein containing a CARD). The PYD motifs of NALP1 and ASC and the CARD motifs of ASC and caspase 1 interact together to comprise inflammasomes. The NALP family is composed of at least 14 genes (NALP1-14). NALP proteins consist of PYD, NACHT domain, and leucine-rich repeats (LRRs). The NACHT domain has a predicted P-loop NTPase sequence and is extracted across different proteins: NAIP (Neural Apoptosis Inhibitory Peptide), CIITA (class II transactivator), HET-E (incompatibility locus protein from Podospora anserine), and TP1 (telomerase-associated protein). Interestingly, NACHT domain shares homology with the dATP binding domain of Apaf-1. The LRRs of NALPs are thought to be able to sense various microbial molecules in intracellular space to form inflammasomes. Many members of NALP family have multiple alternative splicing isoforms (JYW, unpublished data).

PYD (pyrin domain) containing proteins and caspase 1 are involved in processing pro-IL-1/18 and pro-apoptotic caspase 7. PYD is a homophilic adapter domain as well as for DD and DED. Familial Mediterranean fever (MEFV) encodes PYD containing proteins interacting with ASC. The human MEFV gene produces multiple isoforms including full-length (MEFV-fl) and MEFV-d2 [42]. MEFV-fl encodes a cytoplasmic protein, whereas MEFV-d2 protein is concentrated in the nucleus [144]. Multiple components of inflammasome, including TUCAN, have distinct alternative splicing isoforms [2]. Alternative splicing may regulate the pyroptotic cell death signaling.

3.5 Inhibitor of Apoptosis Proteins (IAPs)

IAPs, first identified in baculovirus, contain 1–3 BIR motifs and one RING domain. BIR domains bind caspases and inhibit protease activity and RING domains recruit E2 ubiquitin conjugating enzyme and transfer ubiquitins onto itself or binding partners. Survivin, a member of the IAP family, contains a BIR domain but lacks a RING domain. Interestingly, Survivin still retains anti-apoptotic activity (for reviews, [141]). The Survivin gene generates several splicing isoforms. Among them, Survivin 2α and 2β uniquely display pro-apoptotic activity. Survivin 2β splicing is upregulated by Wilms Tumor 1 associated protein (WTAP) [169]. The alternative splicing pattern of Survivin pre-mRNA changes dynamically in cancer. Survivin alternative splicing has been considered as a diagnostic marker and therapeutic target (reviewed in [158]).

Two different splicing isoforms of livin, another member of the IAP family, livin α and β, have different tissue expression patterns and show distinct properties in inhibiting apoptosis induced by different signals [9].

3.6 Cell Death-Related DNases and Their Regulators

Caspase-activated DNases (CAD) are critical nucleases in PCD. Other notable DNAses are Endo G, LEI/L-DNase II, and GAAD. ICAD/DFF45 is a chaperonin for Caspase-activated DNase (CAD/DFF40) and inhibits the CAD/DFF40 activity. ICAD/DFF45 is one of substrates for capsases. Cleavage of ICAD/DFF45 by caspases releases CAD/DFF40 from the ICAD/DFF45-CAD/DFF40 complex and activates CAD/DFF40 DNase activity. There are at least two splicing isoforms encoded by the ICAD/DFF45 gene. The short isoform is generated by the retention of intron 5 which contains an in-frame stop codon. Although ICAD/DFF45 (ICAD-L) is necessary for CAD/DFF40 peptide to be folded properly, the short isoform ICAD-S/DFF35 does not have the ability to function as a chaperonin. However, ICAD-S binds to CAD/DFF40 more strongly than ICAD/DFF45 and can inhibit the DNase activity of CAD/DFF40. ICAD/DFF45 contains a nuclear localization signal and is expressed in nuclei whereas ICAD-S/DFF35 is also distributed to the cytosol. Aberrant splicing products of CAD/DFF40 have been detected in human hepatoma cells, but biological functions of those isoforms have not yet been characterized [78].

Endonuclease G (Endo G) is another DNase that is released from the mitochondria during PCD. Because the yeast orthologue of Endo G participates in caspase independent PCD, Endo G is thought to contribute to caspase independent PCD in mammalian cells [23, 34]. LEI/L-DNase II and GAAD also contribute to caspase independent PCD. There are many predicted alternative splicing isoforms for Endo G, LEI/L-DNase II, and GAAD isoforms; however, their biological roles and splicing regulation are still unclear.

3.7 Mitochondrial Cell Death Proteins

Since the discovery of the role of cytochrome C in apoptosis, many other mitochondrial factors have been reported to regulate cell death, such as Smac, Omi, and AIF.

Initially identified as an inhibitor of IAPs, Smac/DIABLO has a number of splicing isoforms including Smac-β (Smac S), γ, δ, and Smac 3 [56, 172]. As a result of alternative splicing, Smac-β lacks the mitochondrial targeting sequence at its N terminus and displays a distinct distribution in cells, rather than the mitochondrial localization. Although Smac-β does not interact with IAPs, Smac-β is pro-apoptotic. Smac 3 is missing exon 4, but still contains the mitochondria targeting sequence and IAPs binding domain. Smac 3 disrupts caspase 9 binding to XIAP, promotes caspase 3 activation, and accelerates the auto-ubiquitination of XIAP, whereas the full-length Smac does not accelerate XIAP auto-ubiquitination. Smac 3 localizes to the mitochondria and is released into the cytosol following apoptosis signaling [56].

AIF induces cell death after being released from mitochondria. However, AIF induced cell death does not show oligonucleosome fragmentation, a hallmark of caspase-dependent cell death induced by CAD activation. In fact, cathepsins and calpains can release AIF from the mitochondria to trigger necrotic cell death [37]. Human AIF gene generates multiple alternative splicing isoforms: AIF, AIF-exB, AIFsh, AIFsh2, and AIFsh3 [38, 39]. AIFsh retains pro-apoptotic activity and is expressed in the cytosol due to a missing mitochondria targeting sequence caused by alternative splicing. Therefore, alternative splicing may induce cell death by producing cytosolic AIFsh.

3.8 Autophagy, Cell Death, and Alternative Splicing of Autophagy Regulatory Genes

Autophagy is an essential cellular process mediating the degradation of damaged or degenerated cellular materials. During this process, an isolated membrane sequesters unwanted macromolecules and organelles, such as aggregation-prone proteins and malfunctioning mitochondria. The formed double-membraned vacuoles are called autophagosomes, which in turn fuse with lysosomes forming autolysosomes to degrade their contents [96, 107]. The autophagy pathway is highly conserved from yeast to humans, playing important roles in cellular homeostasis [11, 153]. Under physiological conditions, autophagy degrades altered proteins and organelles, eliminating from the cell malfunctioning components and simultaneously recycling molecular components for the regeneration of new organelles. During nutrient deprivation, autophagy plays critical roles in the adaptation of organisms to new environmental conditions, providing nutrients from degraded cellular contents to maintain cellular metabolism [129].

In addition to cell survival, autophagy mediates cell death, although the underlying molecular mechanisms remain to be elucidated. The phenomenon of autophagic cell death (also known as type II programmed cell death) was observed in the 1960s. At the ultrastructural level, autophagic cell death is characterized primarily by the formation of numerous autophagic vacuoles in dying cells [27, 160, 161]. Autophagy genes Atg7 and Beclin1 are required for cell death in certain types of cells [197, 198]. However, the role of autophagy in regulating cell death remains unclear.

The mammalian target of rapamycin (mTOR) is a kinase that plays important roles in cellular metabolism, cell growth, cell proliferation, and autophagy [41, 43, 71, 188]. mTOR is initially inhibited during starvation, which triggers autophagy. A recent study shows that the mTOR signaling can be reactivated by prolonged starvation, which in turn forms an evolutionarily conserved cycle that maintains energetic homeostasis during cellular starvation [199]. Two splicing isoforms of mTOR have been identified: mTORα (the full-length protein) and mTORβ. mTORβ is capable of regulating cell cycle and cell proliferation. Notably, mTORβ may act as a proto-oncogene, because overexpression of mTORβ leads to immortalization of cells and is tumorigenic in nude mice [143]. However, the role of mTORβ in autophagy remains to be investigated.

Microtubule-associated protein light chain 3 (LC3), a mammalian homologue of yeast Atg8, is another essential component of autophagy [88]. LC3-I can be converted to LC3-II and then processed to bind tightly to the autophagosomal membrane. In rats, two alternative splicing isoforms of LC3 are produced, LC3A and LC3B, which generate proteolytic II forms from precursor I forms and colocalize with LC3. In different rat tissues LC3A, LC3B, and LC3 show different expression patterns, suggesting possible regulation of LC3 by alternative splicing [186].

Tumor protein 53-induced nuclear protein 1, TP53INP1, shows sequence similarity to TP53INP2, a protein essential for autophagy in mammalian cells by interacting with VMP1 and recruiting LC3 and/or LC3-related proteins to initiate the autophagosome [139]. Alternative splicing of TP53INP2 appears to be important for cell invasion, although its role in autophagy remains unclear [134].

Many other autophagy regulatory genes have splicing isoforms. For instance, ULK (uncoordinated-51 like kinase) protein is the mammalian counterpart of yeast Atg1, a Ser/Thr protein kinase involved in the initial step of autophagosome formation in collaboration with its regulators, Atg13, Atg17, Atg29, and Atg31 [61, 86]. Human ULK2 has two alternatively spliced transcript variants that differ in the 3′ UTR that may have different mRNA stability. However, the functional difference among the splicing variants is still unknown. Further studies are necessary to understand the role of alternative splicing in regulating autophagy.

4 Alternative Splicing: A Versatile Mechanism for Regulating Expression and Function of Cell Death Genes

Emerging evidence, some of which is summarized above, supports that a wide range of cell death genes undergo alternative splicing that impact their activities in regulating programmed cell death. Many of these PCD genes express alternative splicing isoforms in a tissue- or development stage-specific manner. In naïve T lymphocytes, TNF-α transcripts do not undergo splicing and are stored as pre-mRNA. TNF-α pre-mRNA splicing is only initiated to produce TNF-α protein following the engagement of T cell receptors [194]. In na B and T cells, LARD-1 (lymphocyte-associated receptor of death 1) is expressed as alternative isoforms with very little full-length mRNA [163]. The full-length LARD-1 becomes the predominant product after T cell activation.

Bcl-x pre-mRNA is alternatively spliced to produce two isoforms with opposing functions. The two isoforms are produced by using different 5′ splice sites (ss) in exon 2. The use of the upstream 5′ ss produces a shorter product, Bcl-xS and the downstream site gives rise to Bcl-xL. Bcl-xS is pro-apoptotic while the longer form is anti-apoptotic. Adult neurons predominantly express the Bcl-xL mRNA; whereas immature thymocytes express a relatively high level of Bcl-xS transcript [18]. Adding further to the complexity, several critical PCD regulators have been found to use alternative splicing to generate gene products that have antagonistic activities in cell death. The mechanism by which alternative splicing may regulate function of these PCD genes can be summarized in at least three aspects:

  • regulating subcellular localization of PCD gene products

  • modulating functional activity of PCD gene products

  • influencing stability of mRNA and/or translational control

4.1 Regulation of Subcellular Localization

Many cell death genes, including the death receptor family, Bcl-2 superfamily, and cell death associated proteases, have both membrane associated as well as soluble isoforms (see review [185]). Alternative splicing of these cell death genes produce proteins that contain or lack their transmembrane domains, thus generating gene products with different subcellular localization. These different isoforms may have distinct functions in PCD. For example, several Fas splicing variant mRNAs encode soluble proteins that block Fas-mediated apoptosis induced by the agonistic antibody [25] and by the Fas ligand [145]. Another Fas isoform, FasEX08Del, is generated by exon 8 skipping. This isoform contains a premature termination codon and thus lacks the entire intracytoplasmic death domain. FasEX08Del is only expressed in resistant tumor clones [25]. It is likely that different alternative splicing products of the death receptor may have distinct roles in the propagation of the cell death or survival signals.

Genes from the Bcl-2 superfamily undergo alternative splicing to generate different isoforms with distinct intracellular localization (reviewed in [4]). For example, both membrane-bound and soluble isoforms of Bcl-2 gene members are produced as a result of alternative splicing. As described previously, different splicing isoforms of mitochondrial cell death proteins lacking or containing their mitochondrial targeting sequences may have distinct function in cell death. Finally, alternative splicing may regulate the terminal events in cell death such as chromosome fragmentation and DNA degradation. For example, the nuclear localized ICAD-L and cytoplasmic ICAD-S are generated by alternative splicing which removes the nuclear localization signal in ICAD-L to form ICAD-S [157]. ICAD-L and ICAD-S may have different regulatory activities in cell death.

4.2 Modulating Functional Activities

Interestingly, many PCD associated genes generate splicing isoforms that have antagonistic activities, such as caspase-2L versus caspase-2S [31, 59]. Recent analyses indicate that a majority of the known caspase family members have alternative splicing at or near the regions encoding peptide sequences contributing to their active sites for enzyme activity [74]. Alternative splicing events that generate truncated peptides provide an additional mechanism to quantitatively regulate gene expression. The delicate balance of different pro-apoptotic and anti-apoptotic products of these PCD genes likely plays a critical role in determining cellular susceptibility to death signals. Alternative splicing can regulate each step of cell death induction or execution by generating distinct gene products with different or even antagonistic activities in cell death.

4.3 Altering mRNA Stability or Translational Efficiency

Alternative splicing can generate mRNAs with differential turnover rates or with differential properties in translational control (reviewed in [79]). A large number of alternative splicing events occur at 5′ or 3′ untranslated regions of cell death genes. For example, exon 2 of cathepsin B is an alternative exon and encodes the 5′ UTR regulating the protein expression level [135]. Caspase-2 mRNA undergoes nonsense-mediated decay (NMD) under certain conditions, and its protein expression may be tightly regulated by combination of alternative splicing and NMD [29]. A conserved AU-rich element has been identified in the 3′ UTR of Bcl-2 mRNA. This element interacts with AU-rich binding proteins and is associated with bcl-2-down-regulation during apoptosis [45, 159]. IL-1RI-associated kinase-1 (IRAK1) is a serine-threonine kinase important for IL-1 signaling and has different splicing isoforms. The IRAK1b isoform is kinase inactive and more stable than IRAK1 isoform. It is conceivable that different splicing isoforms of critical cell death genes that contain distinct elements for controlling their mRNA or protein stability could have a significant impact on the expression and function of these cell death genes.

5 Molecular Mechanisms Regulating Alternative Splicing of Cell Death Genes

5.1 Splicing Signals, Splicing Machinery, and Alternative Splicing Regulators

With combined approaches, significant progress has been made in identifying the molecular components of splicing machinery and in understanding the mechanisms that control alternative splicing [17, 24, 73, 123, 132, 133, 166]. Through highly dynamic RNA–RNA, protein-RNA, and protein–protein interactions, components of the splicing machinery assemble onto the pre-mRNA and form the catalytically active spliceosome in which biochemical reactions of splicing take place.

The basic splicing signals include the 5′ splice site, branch site, and polypyrimidine track-AG at the 3′ splice site. These signals are initially recognized by the U1 snRNP, U2 snRNP, and U2 snRNP Auxiliary Factor (U2AF), respectively. In mammals, these basic splicing signals tend to be degenerate and are not sufficient by themselves to confer the specificity required to achieve accurate splice site selection. A number of other factors also contribute to splice site recognition and influence splicing efficiency. These include the distance between two splice sites, exon size, as well as local secondary structures in the pre-mRNAs. In addition, various types of exonic and intronic elements have been identified that modulate the usage of nearby splice sites. For example, exonic or intronic splicing enhancers (ESEs or ISEs) and exonic or intronic splicing silencers (ESEs or ISEs and ESSs or ISSs respectively) in different genes have been described (for reviews see Hertel et al. 2008; [70, 183]). These splicing enhancers or silencers can promote or inhibit the use of either upstream 3′ splice sites or downstream 5′ splice sites. Many splicing enhancer elements function by interacting with different members or combinations of SR proteins and hnRNP proteins. A general theme has begun to emerge that the alternative splicing events of a given gene are regulated by a number of different splicing activators and repressors. In most cases, precise molecular mechanisms underlying the splicing inhibition remain to be elucidated.

The specific recognition of splice sites and proper association between authentic 5′ and 3′ splice sites is the central issue for pre-mRNA splicing and alternative splicing regulation. Spliceosomal snRNPs including U1 (or U11), U2 (or U12), U4/U6 (or U4atac/U6atac), and U5 play important roles in splice site recognition and association. In addition to the snRNPs, several families of accessory proteins are also important in regulating alternative splicing. These factors include heteronuclear ribonucleoproteins family (hnRNP proteins), proteins containing serine-arginine-rich sequences (SR proteins) and other RNA-binding proteins ([16, 187]; see Chap. 3). In many cases, SR proteins function as splicing activators by binding enhancer sequences, whereas hnRNP proteins often function as splicing respressors by binding splicing silencer sequences [113]. Other splicing regulators include KH-domain-containing proteins, CUGBP proteins, and proteins containing other sequence motifs (for example, helicase, RGG, zinc finger) [87]. There are also a number of proteins that can act as either splicing activators or splicing repressors depending on the splicing substrates and their binding sites [186, 187].

Aberrant pre-mRNA splicing has been implicated in human diseases associated with either excessive or insufficient cell death, although underlying molecular mechanisms remain to be elucidated (for reviews, [17, 89, 123, 149, 168, 183].

5.2 Mechanisms Underlying Alternative Splicing Regulation of PCD Genes

In recent years, a number of cell death genes have been characterized in detail for mechanisms underlying their alternative splicing regulation. Only a few examples are shown here, including FAS, Bcl-x, and caspase-2. As discussed before, alternative splicing of FAS exon 6 leads to the formation of the membrane-bound full-length FAS-L isoform and exon 6-skipped soluble FAS-s isoform. Two related splicing regulators TIA1 and TIAR stimulate exon 6 inclusion by binding to a U-rich sequence downstream of the 5′ splice site of exon 6 and promoting U1 snRNP interaction with this 5′ splice site [53, 54]. A protein kinase, Fas-activated serine/threonine kinase (FAST K) synergizes with TIA1 and TIAR to enhance FAS exon 6 inclusion [83]. On the other hand, Polypyrimidine Tract Binding protein (PTB), promotes exon 6 skipping by binding to an exonic splicing silencer [82]. RNA binding protein SPF45, which has a U2AF homology motif (UHM), interacts with the SF3b155 ULM (UHM-ligand motif) domain to enhance inclusion of Fas exon 6. This binding plays a critical role in FAS exon 6 inclusion [30]. During the early stages of apoptosis, U2AF65 is cleaved and the N-terminal fragment of U2AF65 has a dominant negative effect on FAS exon 6 splicing [81] thus leading to formation of the Fas-s isoform. Hu antigen R (HuR) binds to the ESS on FAS pre-mRNA and inhibits the association of U2AF65 with the 3′ss [84]. RBM5 excludes exon 6 of FAS pre-mRNA. Interestingly, RBM5 does not affect U1 and U2 snRNP assembly on FAS pre-mRNA but inhibits the transition from pre-spliceosome to mature spliceosome via interaction between U4/5/6 trisnRNP and the OCRE domain of RBM5 [20].

The production of anti- and pro-apoptotic isoforms of Bcl-x involves alternative selection of two competing 5′ splice sites. Both exonic and intronic elements have been identified in Bcl-x that regulate its alternative splicing [109, 120]. Trans-acting factors involved include SAP155, Sam68, hnRNA A1, hnRNP F/H, SRp30c, and RBM25 [28, 62, 146, 201]. An apoptotic agent Ro-31-8220 inhibits PKC and activates JNK, and concomitantly, Bcl-xL splicing is inhibited. This effect is repressed by okadaic acid, an inhibitor against PP1 and PP2A. Okadaic acid splicing regulation is mediated via a 16-nt G-tract element (Gt16) on pre-mRNAs [69]. Emetine, a protein synthesis inhibitor, also upregulates Bcl-xS splicing. However, emetine upregulation is blocked by calyculin A by inhibiting PP1 and PP2A [21]. Various stimuli are reported to change Bcl-xL/Bcl-xS ratio. For example, IL-6, GM-CSF, and TPA all upregulate Bcl-xL splicing in erythroleukemia and glioma. The Bcl-xS isoform is induced by S-adenosylmethionine (SAMe), whose synthesis is impaired by liver injury. β-adrenergic receptor activation upregulates Bcl-xS and induces cell death of cardiomyocytes. Consistently, β-blockers inhibit Bcl-xS induction. Ceramide responsive elements CRCE1 and 2 have been identified in Bcl-x pre-mRNA. Ceramide appears to regulate Bcl-x splicing via SAP155 which binds to CRCE1 [121, 122]. Although Bcl-x splicing responds to a number of stimuli, it still remains unclear how these signals transduce to the splicing machinery. A recent study using genomic siRNA screening for Bcl-x splicing regulators has uncovered a complex network of splicing regulators that link cell cycle and cell death controls [133].

In the case of caspase-2, several splicing factors have been identified to regulate the formation of anti-apoptotic (caspase-2S) and pro-apoptotic (caspase-2L) splicing isoforms. Interestingly, SR proteins including SC35 and ASF/SF2 promote exon skipping to produce caspase-2L, whereas, hnRNPA1 facilitates exon inclusion to produce caspase-2S [85]. The effects of these SR proteins and of hnRNP A1 on caspase-2 splicing are opposite to their effects on other model splicing substrates. A caspase-2 mini-gene model system has been used to dissect the cis-elements and trans-acting factors involved in caspase-2 alternative splicing ([85, 32]). An evolutionarily conserved 100 nt intronic element, In100, has been identified as an intronic splicing silencer element responsible for exon 9 exclusion between exon 9 and 10. The In100 element contains two domains: an upstream decoy 3′ splice site and a downstream PTB binding domain 32. The decoy 3′ splice site engages the 5′ splice site of the alternative exon 9 in a non-productive manner, effectively suppressing the use of this 5′ splice site without reducing U1 snRNP binding. PTB plays a role in regulating alternative splicing of a number of other genes (reviewed in [16, 171, 181]). The regulatory mechanism of PTB in caspase-2 alternative splicing appears to be distinct from that involved in other genes. Furthermore, our recent survey of human genome suggests that In100-like (Intron 100-like) intronic elements (i.e., decoy 3′ splice site juxtaposed to PTB binding domains) may represent a general intronic splicing regulatory motif and that such elements may play a role in the regulation alternative splicing of other cell death genes [74]. Recently, we reported that caspase 2 splicing is regulated by RBM5, which is frequently deleted in lung cancer [59]. RBM5 binds to the U/C-rich region immediately upstream of In100. These results suggest that splicing regulation plays an important role in cancer development via PCD-related gene products. However, crucial questions still remain unanswered. How does splicing machinery differentially recognize the decoy 3′ splice site inside In100 as a regulatory element as opposed to a true 3′ splice site? How are splicing regulators and cis-elements coordinated to regulate alternative splicing? Molecular dissection of cis-acting elements and trans-acting splicing regulators involved in caspase-2 alternative splicing has provided a good beginning point to understand the mechanisms underlying the complex regulation of alternative splicing of important PCD genes. The involvement of such a pseudo- or decoy splice site in alternative splicing regulation may provide an explanation for the phylogenic conservation of sequences containing pseudo-splicing signals in mammalian introns. Further studies are necessary to test whether pseudo-splice sites proximal to splicing repressor binding sites represent general splicing regulatory motifs.

5.3 Complex Networks Linking Alternative Splicing, Cell Death, and Other Processes

Recent efforts using systems biology approaches have begun to reveal the complex networks that link splicing regulation, cell death, and other important cellular processes [133]. Genome-wide analyses of splicing patterns can help identify specific gene products that are tumorigenic as well as involved in other cellular pathways. Gene regulation by alternative splicing plays critical roles in cellular differentiation, cell proliferation, and cell death. Therefore, imbalance in the splicing pathway can lead to tumorigenic events [63]. High-throughput microarray analyses and next-generation sequencing assays have been used to identify alternative splicing events or factors involved in specific pathways [191]. Such studies will help to decipher the splicing codes that dictate normal cell development and subsequently, how mutations affect these events to give rise to cancer. Given the importance of alternative splicing in generating antagonistic isoforms of pro- or anti-apoptotic proteins, it is hopeful to target alternative splicing machinery for future cancer therapeutics.

6 Cell Death Regulation, Pre-mRNA Splicing, and Cancer

Mounting evidence supports that alternative splicing of genes involved in cell cycle control, cell proliferation, apoptosis, angiogenesis, motility, and invasion are associated with tumor progression and metastasis [116, 200]. In cancer, aberrant alternative splicing has been associated with mutations affecting cis-elements that regulate splicing. Such mutations may alter the abundance, localization, or post-translationally modify trans-acting factors that determine splice site selection [123]. The precise mechanisms by which alternative splicing controls expression of genes related to cancer remain poorly understood.

6.1 Splicing Factors, Splicing Variants, and Cancer

A number of factors regulating alternative splicing of cell death genes show oncogenic properties [68]. Splicing factors, such as hnRNP A1 [22], hnRNP A2, hnRNP B1, polypyrimidine tract binding protein (PTB), [148] and HuR, are frequently overexpressed in tumors. Splicing factor overexpression can trigger malignant transformation.

In addition, cancer-related splicing factor isoforms could alter function of these splicing factors, resulting in aberrant alternative splicing. Depletion of splicing factors prevents the generation of cancer-associated isoforms, suggesting splicing factors as potential therapeutic targets for cancer therapies. Tumor-specific variations in splicing may also generate new epitopes that can serve as anticancer agents.

6.2 Death Receptors and Cancer

Fas (Apo-1/CD95), a transmembrane death receptor, has a soluble isoform (sFas) generated by alternative splicing of Fas mRNA. sFas lacks the transmembrane domain and antagonizes cell-surface Fas function. sFas is detected in patients with different types of leukemia and solid tumors [90, 126], such as adult T cell leukemia (ATL), large granular lymphocyte (LGL), leukemia and renal cell carcinoma.

TNF-related apoptosis-inducing ligand (TRAIL) and its receptors, namely DR4, DR5, DcR1, and DcR2 have become attractive targets for anti-cancer therapies, because they seem to trigger apoptosis selectively in cancer cells but not in normal cells. Thus far, several compounds and biologics (such as agonistic TRAIL antibodies) have gained attention due to their anti-tumor efficacy [131, 151, 196].

In addition, c-FLIP modulates the activation of procaspase-8 and thereby prevents induction of apoptosis mediated by death receptors. There is evidence for increased c-FLIP expression in various types of tumor cells, including colorectal cancer [97], gastric cancer [138], Hodgkin’s lymphoma [138], and ovarian cancer [1]. Thus, downregulating c-FLIP induced by pharmacological agents, such as proteasome inhibitors, protein or RNA synthesis inhibitors [57], or chemotherapeutic agents [103], may have therapeutic value for these types of cancer.

6.3 BCL-2 Family and Cancer

The Bcl-2 family of proteins are crucial players in regulation of apoptosis. Aberrant expression of members of this family has been associated with different cancers. Over-expression of Bcl-2 was originally observed in B cell lymphomas ([130]; reviewed in [98]). Bcl-2 overexpression has since been detected in other solid tumors in the lung, kidney, stomach, and brain [67, 119]. Interestingly, the relationship between Bcl-2 expression levels and cancer prognosis is cell type-dependent. For example, high levels of Bcl-2 were correlated with poor prognosis in certain lymphomas; however, low Bcl-2 expression as correlated with a poor prognosis in breast cancer. Experiments using knockout mice have advanced our understanding of the role of Bcl-2 family members in tumorigenesis. Bad-knockout mice develop B cell lymphoblastic leukemia/lymphoma when exposed to sub-lethal doses of γ-irradiation, whereas Bid-knockout mice show chromosomal aberrations and develop leukemia [202].

Furthermore, the imbalances between apoptosis-promoting and apoptosis-inhibiting members of the Bcl-2 family are also common in various human cancers. Myeloid cell leukemia-1 (MCL-1) has three splicing variants: anti-apoptotic MCL-1L and pro-apoptotic MCL-1S and MCL-1ES. There is an imbalance between the expression levels of MCL-1L, MCL-1S, and MCL-1ES in the skin basal cell carcinoma (BCC) cell line [167] and renal cancer [92]. Bcl-x mRNAs encoding a long isoform, Bcl-xL, predominates in various types of malignant lymphomas and may be involved in lymphomagenesis [190]. Bcl-xL was also expressed in human hepatoma cell lines at high levels and its down-regulation activated apoptosis [173]. Another Bcl-x alternative splicing product is Bcl-xAK, which contains the Bcl-2 homology domains, BH2 and BH4, as well as the transmembrane domain but lacks BH1 and BH3. Bcl-xAK is expressed in melanoma and other tumor cells and its overexpression results in significant induction of apoptosis in melanoma cells [77]. Another Bcl-2 family member, Bfk undergoes alternative splicing to produce four isoforms, out of which two are pro-apoptotic. In the transition from normal tissue to tumor, pro-apoptotic Bfk isoform expression is substantially reduced in tumors isolated from the human gastrointestinal tract [40].

6.4 Caspase Alternative Splicing and Cell Death Regulation in Cancer

Caspases play important roles in the regulation of physiological cell death, therefore, the disturbance of the caspases expression or function may contribute to the cancer formation. Caspase-9 initiates apoptosis and has two distinct protein isoforms generated as a result of alternative pre-mRNA splicing: pro-apoptotic caspase-9a and anti-apoptotic caspase-9b. A recent study demonstrated that hnRNP family member L (hnRNP L) is specifically phosphorylated in non–small cell lung cancer cells (NSCLC cells) and also associated with caspase-9 pre-mRNA. The interaction of hnRNP L with caspase-9 pre-mRNA in NSCLC cells promotes preferential expression of the 9b isoform of caspase-9, which is antiapoptotic, and promotes tumor growth [64].

Another example is the caspase-2 tumor suppressor, which is alternatively spliced to generate multiple isoforms (discussed in Sect. 3.1). RBM5, by virtue of binding to the U/C-rich region in In100 splicing repressor element (see Sect. 5.2), promotes production of the proapoptotic caspase-2L isoform and regulates the ratio of caspase-2 isoforms in HeLa cells [59, 155]. Fas can lead to cell death and also be alternatively spliced to produce shorter isoforms. Exclusion of exon 6 in Fas pre-mRNA generates FasDelE6 which can inhibit Fas-mediated cell death. Recently, RBM5 [20] and HuR [81] have been identified to play an important role in Fas exon 6 inclusion.

Caspase-8L, generated by alternative splicing of caspase-8, can suppresses caspase 8-dependent apoptosis. The imbalanced expression of the caspase-8L splicing isoform has been associated with cancer. Suppressing the formation of caspase-8L splice variant renders cells more sensitive to apoptosis-induced neuroblastoma cell death [128]. In addition, a splice variant of IG20 gene regulating the activation of caspase-8 was implicated during tumorigenesis [111, 137].

6.5 IAPS and Cancer

Survivin, a member of the IAPS (inhibitor of apoptosis proteins) family, functions as a key regulator of mitosis and programmed cell death. It regulates cell death by interrupting multiple cell cycle-related proteins, such as INCENP and Aurora B kinase. Studies have shown survivin overexpression results from several polymorphisms in the survivin gene promoter [19, 150], which also correlate with tumorigenesis and prognosis [115, 172]. Therefore, survivin has become a target for cancer therapeutics [174].

Livin, an IAPS-related protein, has two different functional splicing variants that are characterized as anti-apoptotic [9]. Livin interacts with downstream caspases, such as caspase-3, caspase-7, and caspase-9, leading to their inactivation and degradation. Aberrant expression of Livin has been reported to be associated with tumorigenesis in many different cancer types including melanoma [179], breast cancer [193], and lung cancer [72]. Therefore, Livin is believed to be a new target for immunotherapy and gene therapy for treatment of cancer.

6.6 Cell Death-Related DNases and Their Regulators

Previous studies have demonstrated that the activity of alkaline (DNase I; EC 3.1.21.1) and acidic DNases (DNase II; EC 3.1.22.1) was inhibited in non-necrotic cells in malignant tumors. In subsequent studies, vitamin C and K3 administration reactivated these DNases to induce cancer cell death [175].

Endonuclease G (EndoG) cleaves chromatin DNA into nucleosomal fragments in the nucleus and participates in the caspase independent apoptotic pathway [80]. As a pro-apoptotic protein, decreased expression of EndoG has been found in several cancers, such as hepatocellular carcinomas and breast cancer [12].

6.7 Mitochondrial Cell Death Proteins and Cancer

Apoptosis can be activated through two pathways: the extrinsic pathway (mediated by death receptors) or the intrinsic pathway (mediated by mitochondria). As mentioned previously, mitochondrial factors such as Smac and AIF can adhere to IAPs and inhibit their caspase-binding activity, thereby regulating cell death and tumorigenesis. For example, in a malignant glioma xenograft mice model, co-administration of Smac/DIABLO peptides and TRAIL sensitized glioma cells lead to apoptotic death and induced malignant glioma regression [58]. Furthermore, several Smad analogs can induce cancer cell death and have shown potential as cancer therapies [106, 164].

6.8 Defective Autophagy and Cancer

Autophagy is an evolutionarily conserved mechanism for protein degradation and maintains homeostasis. Studies have implicated autophagy in tumorigenesis and tumor progression. Autophagy deficiency predisposes cells to tumor development. To this end, haploinsufficiency of autophagy genes increased tumor formation in mouse [118]. Conversely, once tumors were established, autophagy may enable cancer cell survival. For example, autophagy increases cancer drug resistance to Imatinib in chronic myeloid leukemia [13] and facilitates resistance trastuzumab for HER2 positive breast cancer cells [178]. Conversely, autophagy abrogation by autophagy inhibitors re-sensitizes the resistant cancer cells to the chemotherapy or radiation [8, 13].

7 Concluding Remarks

Many genes involved in PCD undergo alternative splicing to produce multiple isoforms with distinct functional activities. Alternative splicing not only generates products with different subcellular localization (membrane associated versus soluble proteins; nuclear versus cytoplasmic) but also produces proteins with different, and often antagonistic functional activities. Molecular analyses of these cell death genes suggest fundamental importance of alternative splicing in regulating PCD. However, molecular mechanisms controlling the alternative splicing of these PCD genes remain unclear. Recent studies using model systems have initiated molecular dissection of the link between alternative splicing and PCD regulation. These studies are only the beginning, given the wide variety of functionally distinct proteins generated by alternative splicing. A global landscape of alternative splicing patterns as well as molecular mechanisms involved in cell death regulation await further investigation using multidimensional and combinatorial approaches. Due to the complex regulatory networks that work in harmony to control cell fate and cell differentiation, mutations affecting one pathway can have far reaching consequences at the cellular, multicellular, and tissue levels. Therefore, elucidating regulatory mechanisms underlying functionally important alternative splicing events will not only help us understand pathogenetic mechanisms of human diseases caused by splicing defects but also provide molecular insights into designing new cancer therapies by targeting aberrant or defective splicing.