2.2. Apoptosis

The term “apoptosis” was derived from a Greek word that means “falling off” and was first coined by Kerr et al. [1] to describe a morphological manifestation of cellular demise. The complex cellular morphology known as apoptosis can be confidently recognized by a series of morphological changes at the microscopy level (Figure 1). Apoptosis is characterized by cell shrinkage, membrane blebbing, condensation and margination of nuclear chromatin, degradation of DNA into nucleosomal units (200bp), and formation of apoptotic bodies. However, the hallmark of an apoptotic process is its dependence on caspase activation [1, 8, 12].

Cells undergo apoptosis in response to extrinsic or intrinsic pathways that are regulated by various antiapoptotic and proapoptotic proteins [8, 13]. The extrinsic pathway is mediated by binding of the tumor necrosis factor (TNF) to its receptor (TNFR-1) which is followed the formation of two TNFR complexes [14]. TNFR-1 complex I comprises the adaptor protein TNFR1-associated death domain protein (TRADD), the death domain- (DD-) containing protein kinase receptor-interacting protein 1 (RIPK1), the ubiquitin E3 ligases TNFR-associated factor 2 (TRAF2), and cellular inhibitor of apoptosis protein 1 (cIAP1). TNFR-1 complex II comprises the adaptor FAS-associated death domain protein (FADD), caspase-8, RIPK1 (IIa), and/or RIPK3 (IIb). E3 ubiquitin ligases, including cIAP-1 and -2 and deubiquitinases, including CYLD (cylindromatosis), A20, Cezanne, HOIL-1/HOIP/Sharpin (LUBAC ubiquitin ligase complex), modify the balance between TNFR-1 complex I and complex II [13–15] leading to either cell survival via activation of nuclear factor-κB (NF-κB) or alternative cell death signaling pathways.

Homodimerization and activation of caspase-8 within either TNFR complex I or II propagate the activation of effector caspases-3, -6, and -7, which then cause cellular destruction by apoptosis without mitochondria participation (known as type I intrinsic pathway). Both caspase-8 and cFLIP (FADD-like IL-1β-converting enzyme-inhibitory protein) possess two DED (death effector domain) domains. Their heterodimerization will block the apoptotic signaling cascade, leading to cellular survival and NF-κB-mediated proinflammatory response. Thus the ratio of different c-FLIP isoforms (long and short isoforms) bound to procaspase-8 is a critical regulator of both procaspase-8 dimerization/activation and cell death by apoptosis and necroptosis [16].

The intrinsic pathway is also called the mitochondrial pathway and is absolutely dependent on BAX and BAK protein activation by BH3-only molecules (BH3s) of The BCL-2 (B-cell lymphoma/leukemia 2) family proteins. Various BH3s proteins of the pro-apoptotic family, including BAD, BIK, BID, BIM, BMF, NOXA, and PUMA participate in this process. These proteins interact and dislocate the BH3s proteins of the anti-apoptotic family, including BCL-2, BCL-XL and MCL-1, allowing conformational changes and association of BAX/BAK proteins at specific sites of the mitochondrial outer membrane. Cytochrome c is an abundant protein of the mitochondrial inner membrane and acts as an electron transport intermediate. BAX and BAK promote apoptosis by perturbing the permeability of the mitochondrial outer membrane, referred to as MOMP, which lead to the release of cytochrome c [17]. The cytochrome c released into the cytoplasm stimulates the formation of the apoptosome, a scaffold for activation of caspase-9, which, in turn, cleaves and activates the effector caspases-3, -6, and -7. Mitochondrial membrane permeabilization leads to the release of other apoptogenic proteins, including apoptosis inducing factor (AIF), Smac/Diablo, HtrA2/Omi (serine protease), and endonuclease G, which execute distinct functions in the downstream apoptosis-signaling pathway. Accordingly, new small-molecule inhibitors of Bcl-2 proteins, including ABT-199 and ABT-263, are viewed as promising new anticancer agents.

In vivo, apoptotic cells maintain their plasma membrane integrity. However, apoptotic cells under certain conditions promote the exposure or release of ecto-calreticulin, phosphatidylserine, HSP70, HSP90, opsonins, thrombospondin, and high mobility group box 1 (HMGB1), which are known as cDAMPs [18]. The uptake of apoptotic cells by macrophages promotes the release proinflammatory cytokines [19, 20] and cell growth and wound healing through the release of vascular endothelial growth factor (VEGF) and transforming growth factor-β (TGF-β), respectively [21]. Several cytokines and lipid mediators produced by immune cells and surrounding tissue are involved in the resolution of acute and chronic inflammation [22].

3.1. Regulation of Mitochondrial Osmotic Balance

Mitochondria function in a cytosolic milieu containing Na⁺, K⁺, and Ca²⁺ as well as other anions and cations [25]. The inner mitochondrial membrane, however, is impermeable to these ions and their flux and concentrations in the mitochondrial matrix are controlled by different subtypes of pores, channels, and exchangers. The mitochondrial K⁺ balance is controlled by ATP-dependent (KATP) and Ca²⁺-dependent (KCa) K⁺ channels responsible for influx and by K⁺/H⁺ exchanger responsible for removal of excess of matrix K⁺ [26, 27]. Sodium balance is governed by Na⁺/Ca²⁺ (influx) and Na⁺/H⁺ (efflux) exchangers and calcium balance by Ca²⁺ channel (influx) and Na⁺/Ca²⁺ exchangers (efflux). The balance of the main cytoplasmic anions, phosphate and chloride is regulated by numerous carriers and channels [25, 27]. Cell size decreases due to water loss from increased K⁺ and Cl⁻ ionic effluxes. The mitochondrial inner membrane also has unique water channels named aquaporins, which facilitate water or other small-uncharged molecule transport between the cytoplasm and matrix [28]. The rate of water flux in or out of the mitochondrion is determined not only by the osmotic gradient that acts as the driving force for water transport but also by the water permeability of the inner membrane [28].

The mechanisms by which Ca²⁺ ions modulate both oxidative phosphorylation and mitochondrial matrix swelling have been a matter of great debate for decades. Halestrap and colleague [29] demonstrated that an increase in intramitochondrial Ca²⁺ concentration can increase the flux of K⁺ into mitochondria. Since then, several mechanisms have been proposed to explain this finding. First, it was proposed that elevated matrix Ca²⁺ could open a putative mitochondrial large conductance Ca²⁺-activated K⁺ channel (KCa channel) [26, 30]. Activation of calcium efflux pathways via Na⁺/Ca⁺ and Na⁺/H⁺ exchangers (mitochondrial NHE1) during intramitochondrial Ca²⁺ overload causes dissipation of the mitochondrial membrane potential and loss of proton gradient [27]. Loss of proton gradient suppresses the activity of K⁺/H⁺ exchange leading to the mitochondrial swelling. Finally, if the Na⁺/Ca²⁺ exchanger becomes saturated, a high intramitochondrial Ca²⁺ could rise to the level sufficient to facilitate entry of potassium as well as other ions and solutes and inducing massive swelling of the matrix [26]. An increase of Ca²⁺ concentration up to the micromolar level inhibits pyrophosphatase activity in the matrix, which, in turn, may transiently displace adenine nucleotides from adenine nucleotide translocase (ANT) and convert the latter into a potassium channel. Most likely the total net flux of ions is directed toward the mitochondrial matrix and this ion movement, accompanied by osmotically mediated entrance of water, which leads finally to mitochondrial swelling [31]. The swelling could lead to a rupture of the mitochondrial outer membrane allowing further expansion of the matrix. Thus, rupture of the mitochondrial outer membrane has been proposed as alternative way of permeating the release of cytochrome c in the apoptotic process, but this is debatable [32].

3.2. ROS-Induced Mitochondrial Permeability

ROS (reactive oxygen species) overproduction can lead to severe mitochondrial dysfunction commonly observed in the necrotic cell death process [2]. Reactive oxygen species induce mitochondrial structural oxidation and opening of the permeability transition pores, channels, and exchangers [33]. Under normal metabolic conditions, electron-transporting complexes I, II, III, and IV plus a nonredox H⁺-translocating complex, ATP synthase (also called complex V, FoF1-ATP synthase), coenzyme Q, and cytochrome c carry out oxidative phosphorylation [34]. The a- and b-type cytochromes are inaccessible components of large complexes, but cytochrome c is monomeric, freely diffusible in the inner membrane, and in equilibrium between the inner membrane, intermembrane space and cristae. A small percentage of the total O₂ consumed by the mitochondrial electron transport chain in healthy tissues becomes ROS, such as superoxide (O₂ ^·−), hydrogen peroxide (H₂O₂), and hydroxyl radical (OH−) [33]. This ROS production occurs primarily in complex I (NADH dehydrogenase) and complex III (ubiquinone-cytochrome c reductase). O₂ itself is also a free radical because it has two unpaired electrons in its outer orbit that make it reactive. Increase in the level of products of the one-electron reduction of O₂ is known to induce the mitochondrial permeabilization transition. The hypothesis is that pore formation is involved in the organization of a defense system preventing ROS formation. It is proposed that an ROS-induced nonspecific pore opening lowers ROS production due to (a) maximal stimulation of mitochondrial O₂ consumption and, hence, intracellular lowering of ROS and (b) complete dissipation of mitochondrial membrane potentials and, as a consequence, maximal oxidation of such respiratory chain carriers such as coenzyme Q, which serve as one-electron O₂ reductant [34].

Earlier studies have shown that OH^· oxidize thiol (–SH) groups of sensor proteins which directly promote the activation or opening of the mitochondrial permeability transition pore [33, 34]. The mitochondrial permeability transition pore complex (PTPC) is a nonspecific pore in the inner mitochondrial membrane (IMM) whose opening is triggered by high concentration Ca²⁺ in the matrix [35–37]. Ca²⁺ loading increases the production of ROS, including superoxide. PTP is considered to function as a point of no return for both apoptosis and necrosis. This supramolecular complex may contain or be regulated by ANT (adenine nucleotide translocator), the mitochondrial ADP/ATP nucleotide exchanger, VDAC (the voltage-dependent anion channel), also known as porin, protein cyclophilin D (CypD), the peripheral benzodiazepine receptor, the phosphate carrier, and the FoF1 ATP synthase [38, 39]. CyPD binds to the FoF1 ATP synthase into lipid bilayers to form a Ca²⁺-activated channel that displays features of the mPTP [38]. In this physiological configuration, PTP functions together with other channels and exchangers in the regulation of mitochondrial Ca²⁺ homeostasis [2, 38, 39].

The roles of MOMP and MPT have been mostly considered to participate in the two independent pathways for release of apoptotic and necrotic factors from mitochondria [17]. Bax and Bak interact with the MPT to induce a permeability transition and cytochrome c release in isolated mitochondria [40, 41]. Interestingly, intermediate structures common to membrane fusion or fission machineries have also been described to participate in mitochondrial permeabilization phenomenon [17, 42]. New ideas and experiments concerning structures and functions of membrane-permeation channels, exchanger, and pores are under way to confirm whether they act together or separately to promote the mitochondrial events common to apoptotic and necrotic intrinsic pathways [41].

5.1. Mice Models for Lethal Genes

Knockout mice for caspase-8, caspase-8−/−, FADD, and fadd−/− or double-knockouts for both show an embryonically lethal phenotype due to uncontrolled apoptosis or necroptosis [11, 82, 83]. The embryonic lethality in mice lacking FADD or caspase-8 is due to massive necrosis and can be rescued by RIPK1 or RIPK3 deletion, respectively. Ripk1−/− mice die at birth of systemic inflammation and the large area of necrosis in the liver and thymus [84, 85]. On the contrary, knockout of RIPK3 in mice did not cause any measurable defect in development, fertility, NF-κB activation, and apoptosis [86]. Ripk3−/− mice were resistant to necrotic pancreatitis (cerulein-induced) and vaccinia virus-induced hepatic necrosis [55]. The rescue of a lethal phenotype in RIPK1 knockout is often used as an argument for the implication of necroptosis. Mice deficient in FADD or caspase-8 die during embryogenesis; however, mice with triple deletion of FADD, caspase-8, and RIPK3 are viable [55, 62, 87, 88]. The cardiac, vascular, and hematopoietic defects that occur during the heart development of caspase-8 and FADD knockout mice are caused by RIPK3-mediated necrosis [87, 89]. Therefore, FADD and caspase-8 act as prosurvival factors that suppress the deleterious effects of necrosis by promoting the cleavage and inactivation of RIPK1 and RIPK3. On the other hand, double-knockout mice for ripk3−/− and flip−/− die during embryonic development due to uncontrolled apoptosis driven by active caspase-8 [89]. These experiments demonstrate that FLIP is an important brake on both apoptotic and necrotic cell death in vivo.

Caspase-8 and RIPK3 are all essential for clonal expansion for T and B cell clones [63]. Mice deficient in Casp8−/− fail to develop and die in utero, which may ultimately be due to failing to maintain the proliferation of T and B cells [90]. After stimulation, T cells lacking caspase-8 or its adaptor protein FADD developed hyperautophagic morphology and die by necroptosis [90]. Proliferation in caspase-8-deficient T cells is fully rescued by crossing with ripk3−/− mice, although such rescue ultimately leads to lymphadenopathy [63].

Although TNF-α was initially identified by its ability to kill tumor cells, most normal and tumor cells do not undergo cell death in response to this cytokine [91]. The treatment of wild type murine embryonic fibroblasts (MEFs) with TNF does not induce cell death. Overexpression of the ripk3 gene causes necroptosis in MEFs stimulated with TNF in the absence of RIPK1, caspase-8, Bax, and Bak [92]. RIPK3 activation may be driven by spontaneously high RIPK3 levels or by another RHIM-containing protein such as DAI or TRIF [92]. MLKL is a RHIM-containing protein that is activated in cells with elevated RIPK3 consistent with its important role in necroptosis [92]. RIPK3 can partner also with TLR3 and TLR4 to induce macrophage necrosis and NF-kB activation but this requires TRIF [93].

MLKL is one essential effector of necroptosis since it causes cell death by a caspase- and Bax/Bak-independent mechanism [72]. Mice deficient in Mlkl gene are anatomically normal, viable, and fertile [94]. Cells from Mlkl−/− mice failed to undergo TNF-mediated necroptosis, as expected. Double deficient mice of caspase-8 and Mlkl genes, Casp8−/−; Mlkl−/− mice, were normal but showed pronounced splenomegaly, thrombocytopenia, and lymphadenopathy after a few months of age [94]. This phenotype is similar as described by mice lacking functional FAS (lpr) or FAS ligand (gld mutant mice) [11].

5.2. Mice Models for Inflammatory Genes

Various pathological processes such as ischemic brain injury, myocardial infarction, organ transplantation, and virus replication are accompanied by strong inflammatory response. This inflammatory process is characterized by extensive necrosis of tissues. Studies on mouse models with FADD-TNFR1 and FADD-MyD88 deficiency revealed that both TNF and TLR signaling partially contribute to progression of inflammation [55, 88, 95]. Studies on mouse models have also shown that TNF-induced systemic inflammatory response syndrome (SIRS) and CLP-induced peritoneal sepsis are driven by both RIPK1 and RIPK3-dependent necroptosis [12, 58]. The regulated necrosis is also associated with tissue damage and inflammation driven by activation of sterile inflammatory response, including ischemia-reperfusion injury, Alzheimer's disease, atherosclerosis, and toxic insult to liver and lung [9, 96]. A recent study showed that loss of the RIPK3 or MLKL can provide protection to Ripk1−/− mice from systemic inflammation but fails to protect these mice from lethal intestinal inflammation [85]. Ripk1−/− and Ripk3−/− mice have significantly lower rates of death and inflammation, whereas the Ripk1−/− and Myd88−/− mice have reduced inflammation, however, not reduced mortality [85]. Together, these studies demonstrated that only Ripk1−/−, Ripk3−/−, and Casp8−/− mice are protected from inflammation and intestinal disruption, reach adulthood, and are viable and fertile [83, 85, 89]. Future studies using Ripk1−/− Ripk3−/− and Ripk1−/− Mlkl−/− mice will be important to understand the role of these proteins in inflammation and the release of cytokines and CDAMPs.

Mice with chronic proliferative dermatitis mutation (cpdm mutant) develop TNF-dependent multiorgan inflammation, which is characterized by dermatitis, liver inflammation, splenomegaly, and loss of Peyer's patches [97]. These mice are deficient in SHARPIN, a protein that, together with HOIL-1, is a key regulator of HOIP (LUBAC), an ubiquitin ligase complex that catalyzes the addition of linear ubiquitin to target proteins [15]. SHARPIN is required for the TNFR signaling complex and for prevention of cell death in various cells, including epidermal keratinocytes [15]. A recent study has demonstrated that RIPK3 or MLKL deficiency is able to prevent liver inflammation in shpn−/− mice as well as restoring splenic architecture splenic phenotype and leukocytosis [97]. Only combined caspase 8 heterozygosity and RIPK3 deficiencies were able to almost completely prevent chronic proliferative dermatitis in shpn−/− mice. The role of SHARPIN in the skin-inflammation phenotype is mainly due to keratinocyte mitochondria-dependent apoptosis [85]. Since RIPK3 and MLKL deletion markedly reduced leukocytosis, it is suggested that the hematopoietic phenotype is mediated by necroptosis [97].

The E3 ligases, cIAP-1 (Clap 1 gene), cIAP-2 (Clap 2 gene), and Xiap, are responsible for RIPK1 polyubiquitination, but knockout mice of these genes are surprisingly phenotypically normal. Double-knockout Xiap−/− Clap2−/− mice are also phenotypically normal. However, Xiap−/− Clap1−/− and Clap1−/− Clap2−/− double-knockout mice are embryonic lethal and this can be rescued by crossing these animals with Ripk1−/− and Ripk3−/− mice [98].

Mice deficient in Cylindromatosis (CYLD) gene are viable and fertile during adulthood life. CYLD deficiency leads to hyperubiquitinated RIPK1 in the necrosome and impaired phosphorylation of RIPK1 and RIPK3, thereby blocking caspase-8 activation. CYLD is a caspase-8 substrate and when activated it cleaves CYLD and prevents necroptosis, similar to what occurs with the cleavage of RIPK1 and RIPK3 [99]. CYLD is required for TLR3 or TLR4 receptor-induced necroptosis. The crossing of cyld−/− mice with other members of the NLR and TLR families promises to deliver further experimental evidence on its crucial role in innate response. Finally, deficiency of both RIPK3 and caspase-8 or FADD completely abrogated Yersinia-induced cell death and caspase-1 activation. Mice ablated of RIPK3 and caspase-8 genes in their hematopoietic compartment showed high susceptibility to Yersinia infection as well as displayed a very low production of proinflammatory cytokines by monocyte and neutrophils [100]. We are waiting for the participation of additional components in these pathways and more complex interactions among them in the intracellular resolution of infections.

The authors thank Peter Vandenabeele, Douglas Green, and Ricardo Weinlich for contributions and helpful discussions. The work in our lab is sponsored by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Fundaçao Amparo a Pesquisa do Estado de Sao Paulo (FAPESP).