Mitochondrial Bioenergetics and Dynamics in Secretion Processes

Secretion is an energy consuming process that plays a relevant role in cell communication and adaptation to the environment. Among others, endocrine cells producing hormones, immune cells producing cytokines or antibodies, neurons releasing neurotransmitters at synapsis, and more recently acknowledged, senescent cells synthesizing and secreting multiple cytokines, growth factors and proteases, require energy to successfully accomplish the different stages of the secretion process. Calcium ions (Ca2+) act as second messengers regulating secretion in many of these cases. In this setting, mitochondria appear as key players providing ATP by oxidative phosphorylation, buffering Ca2+ concentrations and acting as structural platforms. These tasks also require the concerted actions of the mitochondrial dynamics machinery. These proteins mediate mitochondrial fusion and fission, and are also required for transport and tethering of mitochondria to cellular organelles where the different steps of the secretion process take place. Herein we present a brief overview of mitochondrial energy metabolism, mitochondrial dynamics, and the different steps of the secretion processes, along with evidence of the interaction between these pathways. We also analyze the role of mitochondria in secretion by different cell types in physiological and pathological settings.

About 20% of the proteins synthesized by eukaryotic cells are secreted to the extracellular space either as soluble or membrane bound proteins (1). The secretory pathway in eukaryotic cells is responsible for biogenesis and proper distribution of a wide range of extracellular proteins, as well as complex carbohydrates and lipids. This pathway is highly dynamic and responsive to specific cellular demands and stimuli (2). Eukaryotic cells also secrete many amino acids and amino acid derivatives, such as neurotransmitters, that play key roles in intercellular communication (3).

As many other complex cellular processes secretion of proteins consumes energy, therefore requires the support of functional mitochondria. In the conventional pathway proteins are transported into the endoplasmic reticulum (ER), and folding and quality control of proteins in the ER (4–8) consumes ATP. Likewise the assembly, transport and fusion of the vesicles carrying proteins from the ER to the Golgi and to the plasma membrane requires the hydrolysis of ATP, as well as GTP (2, 9, 10). Besides, several steps in the secretion process depend on calcium ion (Ca2+) as a cofactor or signaling molecule (11, 12). Calcium pumps are required to maintain the appropriate concentrations of Ca2+ inside the ER and near the sites of exocytosis, and mitochondria play a relevant role providing energy to transport Ca2+ against its concentration gradient (13, 14). Mitochondria also modulate Ca2+ concentrations in the cytosol, sequestering the ion in the mitochondrial matrix (14, 15). Secretion of proteins by unconventional pathways also depends on mitochondria, not only as a source of ATP, but also of activation signals such as mitochondrial reactive oxygen species (ROS) and oxidized mtDNA (16, 17), as well as a structural platform for the assembly of inflammasomes (17, 18).

Work by others and us shows that impairment of mitochondrial catabolism and dynamics affects the secretion processes (6, 7, 19–26) and can underlie pathology in endocrine and neurodegenerative diseases (27–30). Relevant roles for mitochondria in protein secretion by immune and senescent cells have been recently described as well (19–21, 24–26, 31), underscoring the relevance of these interactions. This prompted us to explore the existing literature on mitochondrial interactions with the secretion machinery.

While extensive literature can be found regarding both secretion and mitochondrial bioenergetics and dynamics, the connection between these processes and the organelles involved are still largely unexplored. In this manuscript secretion pathways, mitochondrial metabolism and dynamics per se have not been extensively reviewed. We have relied on reviews by others to present an overview, and centered our efforts in exploring the requirement for mitochondrial ATP and fusion and fission proteins to sustain secretion in metazoans, excluding the events linked to ER stress. We expand on three main roles of mitochondria: (1) providing ATP for multiple steps of the secretion process; (2) buffering Ca2+ concentrations; (3) providing signals and structural scaffold for the activation of the inflammasome (Figure 1).

Secreted proteins can reach the extracellular media through the conventional (classic) pathway or through unconventional pathways (32). In the conventional pathway proteins are transported into the ER, either cotranslationally or postranslationally; then travel from the ER to the Golgi complex, from which they then migrate to the trans-Golgi network and finally to the plasma membrane (33, 34). Transport from one compartment to another takes place by sequential budding and fusion of vesicles (35) and the microtubule and actin cytoskeleton plays a relevant role in vesicle transport (33, 36) (Figure 2). Unconventional secretion pathways, on the other hand, include proteins that do not present a leader sequence and proteins that by-pass the Golgi apparatus while traveling to the plasma membrane (32).

Most of the secreted proteins in the conventional pathway are translocated to the rough ER during translation by the ribosome (37). Proteins are targeted to the ER by a hydrophobic signal peptide located in the N-terminus, which is later cleaved by a signal peptidase present in the ER lumen. The signal peptide is recognized by the signal recognition particle (SRP) that then binds to the SRP receptor (SR) (33, 38). Both SRP and SR are GTPases and the docking and release of SRP at the ER membrane requires GTP (37, 39). Docking is followed by the translocation of the nascent protein to the ER lumen through the translocon, a pore that in mammals is composed of Sec61 proteins, and associated proteins including BiP, Sec63, Sec62, translocating-chain-associated protein (TRAM), translocon-associated protein (TRAP), and ribosome-associated membrane protein (RAMP) (37, 40, 41). BiP plays a dual role during protein translocation, it seals de pore and provides the driving force to transfer of the nascent protein into the ER (4, 40). This protein is an ATP-dependent chaperone that belongs to the heat shock protein 70 (Hsp70) family. During protein translocation BiP interacts with co-chaperones Sec63 and RAMP, these ER-resident J-domain proteins (ERdjs) stimulate BiP binding to the nascent protein and ribosomes as well as ATP hydrolysis (4, 41). BiP also interacts with the nucleotide exchange factors GRP170 and Sil1 that promote the exchange of ADP for ATP (41, 42). Though ATP hydrolysis by BiP is the main driving force for translocation, GTP hydrolysis during the elongation stage of protein synthesis may also contribute to the process (4, 40) (Table 1 and Figure 2).

In the ER, the oligosaccharyl transferase complex (OGT) catalyzes the transfer of the oligosaccharide (Glc3Man9GlcNAc2) from dolichol phosphate to an asparagine residue in the protein (52, 53). Exoglucosidases I and II then catalyze the removal of two terminal glucose moieties producing a monoglucosylated structure (Glc1Man9GlcNAc2) that is recognized and bound by lectin chaperones calnexin and calreticulin. These chaperones, in collaboration with BiP, GRP-94 and protein disulfide isomerase A3 (PDIA3, Erp57), help the protein acquire its native conformation (8, 38). Removal of the remaining glucose moiety by exoglucosidase II prevents further binding of the glycoprotein to calnexin and calreticulin, allowing it exit to the ER. However, if the native structure is not achieved the protein undergoes reglucosylation catalyzed by UDP-glucose:glycoprotein glucosyltransferase (GT), an enzyme that preferentially acts on incompletely folded glycoproteins. Thereby favoring a new cycle of interactions with calnexin and calreticulin and the other components of the folding machinery (8, 54, 55). Protein folding requires ATP, since chaperones BiP and GRP-94 hydrolyze ATP during their catalytic cycles (41, 44). Besides, calnexin and calreticulin bind ATP though they do not hydrolize it (56). Calnexin and calreticulin also bind Ca2+, therefore Ca2+ homeostasis must be maintained in the ER lumen (56). The sarcoplasmic/endoplasmic reticulum calcium ATPase (SERCA) is responsible of coupling ATP hydrolysis to Ca2+ transport from the cytosol to the ER (4, 57) (Figure 2).

In the ER proteins undergo quality control pathways for the detection, remotion, and degradation of misfolded proteins that did not attain their native structures. Degradation by the proteasome takes place in the cytosol in a pathway known as ER-associated degradation (ERAD) (8, 45). The first step in this pathway is the recognition of the misfolded protein and several luminal and cytoplamatic factors are involved in this process (46). Among them we find ER mannosidases (ER α1,2-mannosidase and EDEMs, 1, 2, and 3) that trim the mannose residues from glycoproteins, preventing glucosylation by GT and therefore withdrawing the protein from the quality control cycle (8, 58). Mannose trimmed glycoproteins bind OS-9 and XTP-3B and associate with SEL1L that channels the misfolded protein to the dislocon where proteins are retrotranslocated to the cytosol (45). Though the dislocon is not well-characterized it is thought to include Sec61, the E3 ligase HRD1, Derlins 1-3, and ATPase VCP/p97, which converts the energy released by ATP hydrolysis into mechanical force for the extraction process (4, 45). BiP and PDIs (ERdj5 and ERp90), also play a role, at this stage, binding and oxidizing the proteins to be transported (41, 59). In the cytosol, ubiquitin activating enzymes (E1), conjugating enzymes (E2), and E3 ligases (primarly HRD1 and AMFR) transfer ubiquitin to the protein in an ATP-dependent process (46). Finally, ubiquitinated proteins are degraded by the 26S proteasome in a pathway that also consumes ATP (45, 46). Overall the ERAD is an extremely costly process that consumes ATP during the extraction, ubiquitination and degradation of misfolded proteins (4, 41, 45) (Table 1 and Figure 2).

Secretory proteins that achieve the correct folded and assembled conformation are then transported from the ER to the Golgi complex in coat protein complex II (COPII) carrier vesicles (47). Vesicles for anterograde (forward) transport are formed in ER exit sites (ERES) and key proteins involved in their assembly include SEC12, SEC16, secretion-associated Ras-related 1 (SAR1) GTPase and the two major coat subunits SEC23- SEC24 and SEC13- SEC31 (47, 60). Proteins enter the vesicles by diffusion (bulk-flow), or undergo selective uptake (cargo capture) by SEC24, and other receptors (2, 48, 61). Upon leaving the ERES COPII vesicles shed their SEC13- SEC31 coats in a process that involves the hydrolysis of GTP bound to SAR1 (62), and fuse with one another or with the ER-Golgi intermediate compartments (ERGIC) (2, 47). Rab-family small GTPases and Rab effectors are involved in vesicle tethering to the membranes. After tethering vesicles fuse with the acceptor compartment delivering the cargo to the Golgi (2, 48). Retrieval of misplaced ER luminal proteins from the ERGIC or from the Golgi is achieved by retrograde transport in coat protein complex I (COPI) carrier vesicles. Formation of COPI vesicles involves the ADP-ribosylation factor 1 (ARF) GTPase, that recruits the coatomer subunits to the membrane (48). Vesicle formation, coat shedding and tethering to target membranes are energy consuming processes that consume GTP (2). SAR1 and ARF GTPases switch between inactive GDP-bound and active GTP-bound forms. Guanine nucleotide exchange factors (GEFs) catalyze the release of GDP (allowing the union of a new GTP molecule); while GTPase-activating proteins (GAPs) promote the hydrolysis of bound GTP, regulating enzyme activity and therefore vesicle trafficking (2, 63) (Table 1).

From the cis-Golgi compartment proteins are transported via cisternal maturation to the trans-Golgi network (48), where proteins to be secreted are sorted in secretory vesicles and delivered to the plasma membrane, to the endosomal system or to immature secretory granules (33, 48). Motor proteins, kinesin, and dynein transport the cargo vesicles along microtubules. Kinesins transport secretory vesicles from the organelles where they arise to the cortical region, while dynein mediates their transport in the opposite direction (49, 64). Myosin V and II are responsible for the movement of vesicles along actin filaments in the cortical area (49, 50). The cortical actin network, at the cell borders, acts as a diffusion barrier that prevents the access of granules to secretory sites until the arrival of the triggering stimuli (50). Motor proteins, kinesin, dynein and myosin use the energy released during ATP hydrolysis to move vesicles along microtubules and actin filaments (9) (Table 1 and Figure 2).

Upon arrival to the plasma membrane secretory proteins are released to the extracellular space by exocytosis. Constitutive exocytosis occurs in all cell types to release extracellular proteins and maintain plasma membrane homeostasis and cell polarity. While regulated exocytosis occurs in specialized secretory cells and is triggered by secretion signals (13), such as increases in Ca2+ concentrations (65). Calcium ions bind sensor molecules and promote the fusion of vesicles with the plasma membrane (65). Upon the arrival of the secretion stimuli myosin II drives the movement of vesicles to the fusion sites where they dock to the plasma membrane (66). SNARE complexes (formed by syntaxin, SNP-25, and VAMP), and SM proteins (such as Munc18-1), are in charge of the fusion of vesicles with the plasma membrane (10, 13, 51), while the actin and myosin network contributes to the release of vesicle contents in an ATP-dependent manner (50, 66, 67). Fusion of vesicles with target membranes also consumes ATP. SNARE proteins localized in vesicles and target membranes interact forming a complex that pulls the two membranes together, exerting the force required for fusion. The dissociation of the SNARE complex involves soluble NSF attachment protein (SNAP) and the ATPase N-ethylmaleimide-sensitive fusion protein (NSF) that hydrolyses ATP (10, 51) (Table 1 and Figure 2).

Overall the concerted action of many cellular components is required to ensure the selective and efficient secretion of proteins. Many of these steps require energy and mitochondria play a relevant role, fulfilling the ATP requirements of the pathway (Figures 1, 2). Besides, mitochondria help to modulate the intracellular concentrations of Ca2+ (13, 14), an essential cofactor in the folding process and a potent trigger of exocytosis (65) (Figure 1).

Though the conventional protein secretion pathway was considered for a long time as the only mechanism for protein secretion, we now know that many proteins use alternative pathways. These include the secretion of cytosolic proteins that do not have a signal peptide (leaderless proteins) or a transmembrane domain; and proteins that contain a signal peptide or a transmembrane domain and enter the ER but are not transferred to the Golgi complex (68).

Proteins without a leader sequence are secreted along three different pathways: Type I pathways involve the translocation from the cytoplasm to the extracellular space through a pore in the plasma membrane. In Type II secretion processes the ATP-binding cassette transporter is responsible for the secretion of the protein present in the cytoplasm while in Type III secretion the proteins use autophagosomes and endosomes to reach the extracellular space. The Golgi-bypassing route is also known as Type IV secretion pathway (68). Some proteins are constitutively secreted through unconventional pathways, however these pathways are usually induced in stressful conditions, such as nutrient starvation, inflammation, and mechanical or ER stress (32, 68).

One of the most studied proteins secreted by the unconventional pathways is interleukin-1β (IL-1β) (68, 69). During inflammation IL-1β is produced as an immature and inactive form that is cleaved by caspase 1 upon recruitment and activation by the Nod-like receptor family pyrin domain containing 3 (NLRP3) inflammasome (70), in a process that will be discussed below in detail. The assembly and activation of the complex is an energetically costly event that requires the hydrolysis of ATP by NLRP3 (71). The posterior secretion of mature IL-1β involves the formation of a pore that increases the permeability of the plasma membrane (69) or occurs through the autophagosome/endosmal pathway (68).

Many neurotransmitter molecules are amino acids or amino acid derivatives and their secretion occurs through exocytosis. The pathway starts with the synthesis of the neurotransmitters by enzymes, followed by their loading into of synaptic vesicles. The loading process involves transporters in the vesicle membrane and happens at the expense of an electrochemical proton gradient, generated by the vacuolar H+-ATPase. This pump uses the energy released by ATP hydrolysis to transport protons into the vesicle lumen (72). Loaded vesicles can then be sequestered in a reserve pool or travel to specialized release sites, known as active zones, at the presynaptic terminal where vesicles become docked at the plasma membrane. Increases in Ca2+ concentrations promote the fusion of the vesicle with the plasma membrane and discharge their content into the synaptic cleft (73). Calcium sensing, vesicle docking and exocytosis follow the same steps described for the conventional secretion pathway of proteins (51). These steps are followed by the recovery of the synaptic vesicle membranes that are recycled for refilling in preparation for the next cycle of exocytosis (74).

Mitochondria are cell organelles defined by a double membrane, the outer mitochondrial membrane (OMM) and the inner mitochondrial membrane (IMM). The OMM separates the mitochondria from the cytosol, however the voltage-dependent anion channel (VDAC or porin) allows the passage of metabolites and ions. In the OMM are also located proteins involved in apoptosis, mitochondrial dynamics, and tethering to other organelles. In the IMM, two functional and structurally different regions are described: the inner boundary membrane and the cristae, where electron transport and ATP synthesis take place in a process known as oxidative phosphorylation (75). In the cristae we find the electron transport chain (ETC) complexes: NADH-ubiquinone oxidoreductase (complex I), succinate-ubiquinone oxidoreductase (complex II), ubiquinol-cytochrome c oxidoreductase (complex III), and cytochrome c oxidase (complex IV), as well as the FoF1-ATP synthase (ATP synthase or complex V) and two mobile electron carriers, ubiquinone and cytochrome c (75).

The ETC complexes are assembled forming supercomplexes that optimize electron transport and proton shuttling through the IMM (76, 77). ATP synthase localizes on the edges of cristae forming dimers, which optimize ATP production (78). At the cristae junctions we find the mitochondrial contact site and cristae-organizing system (MICOS), a large oligomeric complex that interacts with both the IMM and OMM (79). Cristae structure influences the ETC, complex and supercomplex formation, and can present different morphologies depending on the metabolic state of the cell (80).

Mitochondria are the main source of ATP in most cells, since many catabolic pathways converge in this organelle and result in the production of ATP by oxidative phosphorylation. The catabolism of metabolites such as glucose, proteins, and fatty acids produces acetyl-CoA, which is in turn oxidized in the tricarboxylic acid cycle generating the reduced electron donors nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FADH2). Electrons are then transferred to the ETC and flow through the complexes to molecular oxygen (O2). This thermodynamically favorable electron transport releases energy, which is used to pump protons from the matrix to the intermembrane space, at complexes I, III, and IV, creating an electrochemical gradient (81). The energy released by the dissipation of the electrochemical proton gradient is then coupled to the synthesis ATP from ADP and Pi, catalyzed by ATP synthase (82, 83).

Mitochondrial catabolism of nutrients and electron transport in the respiratory chain involves many redox reactions that have as by-products reactive oxygen species (ROS) (84). In particular, superoxide anion radical is formed by several flavoproteins and at complex I and III of the respiratory chain, and its dismutation gives rise to hydrogen peroxide (85). These mildly oxidant species can result in the formation of higher oxidant species such as peroxynitrite, formed in the reaction between superoxide and nitric oxide, oxo-metal complexes, nitrogen dioxide, and hydroxyl radical (86).

Under physiological conditions mitochondria produce controlled levels of oxidants, many of which participate in signaling processes. However, ROS formation can increase during cellular stress or in pathological conditions. Although mitochondrial oxidants can be detoxified by enzymatic and non-enzymatic antioxidants (87, 88), an imbalance in mitochondrial redox status may lead to mitochondrial damage (84, 89, 90). In fact, oxidants impair the activity of enzymes of the tricarboxylic acid cycle, the ETC, and ATP synthase (84, 90).

Mitochondrial dynamics consists of fusion and fission events driven mainly by dynamin-related GTPases (91). Optic atrophy protein 1 (OPA1) and mitofusins 1 and 2 (MFN1 and MFN2) are involved in mitochondrial fusion. The long OPA1 isoform is anchored to the inner membrane, where it promotes IMM fusion. The soluble and short OPA1 isoform, found in the intermembrane space, maintains mitochondrial cristae structure (92). MFN1 and MFN2 participate in tethering and fusion of the outer membranes of two different mitochondria through the formation of homo and heteroligomeric complexes (93, 94). Phospatidic acid, a fusogenic lipid formed in a reaction catalyzed by mitochondrial phospholipase D, is also required for OMM fusion (95).

In turn, mitochondrial fission is carried out by the GTPase dynamin related protein 1 (DRP1), a cytosolic protein that is recruited to the OMM (96) by receptor proteins, such as mitochondria fission factor (97), mitochondrial dynamics protein of 49 and 51 kDa (98) and fission 1 protein (FIS1) (99). DRP1 oligomerizes around mitochondria into a ring-like structure constricting the organelle (96). ER-mitochondrial contacts play a role in mitochondrial fission, determining the position of the fission events (100) and ER tubules wrap around mitochondria and constrict them (100). Components of the cytoskeleton such as actin, myosin II and septin 2 are required for mitochondrial fission (101–104), as well as the ER protein inverted formin 2 (105) and a mitochondrial Spire1 isoform (103). The latter two cooperate to induce localized actin polymerization at the constriction sites (103). Both fusion and fission are highly regulated processes, and changes in protein levels, GTPase activity and post-translational modifications of proteins involved in mitochondrial dynamics affect not only mitochondrial morphology but also cellular bioenergetics and homeostasis (106).

Multiple studies linking mitochondrial morphology and bioenergetics can be found in the literature (107–109). At the molecular level several reports support that MFN2 is required for correct mitochondrial function (110); since its silencing results in an impairment of oxidative phosphorylation, while overexpression enhances mitochondrial metabolism (19, 111–113). Less evidence supports a role for MFN1. Although there are reports on MFN1 silencing decreasing mitochondrial respiration (114) this does not seem to be the case in all models (19). OPA1 deletion also affects mitochondrial membrane potential, respiration and cristae structure and reduces assembly, and stability of supercomplexes (80, 114–117).

In order to face increases in energy demands, such as those imposed by secretion, mitochondria also modify their cellular distribution. Motor proteins, kinesin, and dynein, transport mitochondria over long distances along microtubules toward the cell periphery or the cell center, respectively (49). While myosin 19, mediates short-range movement along the actin filaments near the cell periphery (49, 118). The mitochondrial Rho GTPase (MIRO) and the adaptor proteins trafficking kinesin-binding proteins 1 and 2 (TRAK1 and 2) support bidirectional mitochondrial movement along microtubules by binding kinesin and dynein (119–121) (Figure 2). MIRO proteins also engage myosin 19, linking mitochondria to actin filaments (122) and interact with components of the MICOS complex, linking the transport machinery to cristae organization and ensuring the appropriate provision of energy to the regions where mitochondria are delivered (123).

Mitochondria can interact with several cellular compartments, including organelles involved in protein secretion such as the Golgi complex and ER (124–126). In particular, mitochondrial interactions with the ER have been extensively studied and many components and functions of the ER–mitochondrial contact sites (ERMCs)1 have been identified. ERMCs are crucial sites for the synthesis and exchange of lipids, Ca2+ transport, mitochondrial fission and inflammasome formation (4, 125, 127).

Among the proteins found in ERMCs is MFN2, and though some controversy exists (128, 129), this protein is proposed to link the ER membrane to mitochondria through interactions with MFN1 and MFN2 in the OMM (130, 131). The mitochondrial proteins MITOL and mitostatin, as well as the ER protein presenilin 2 modulate these MFN2-dependent interactions between organelles (132–134). Other relevant proteins linking mitochondria to the ER are VDAC, the OMM chaperone GRP-75 and the inositol-1,4,5-trisphosphate-sensitive channel or receptor (IP3R) in the ER. Another tethering complex is formed between the ER protein vesicle-associated membrane protein- associated protein B (VAPB) and the OMM protein tyrosine phosphatase-interacting protein 51(PTPIP51) (135–137) (Figure 2).

Physical tethering of mitochondria and the ER by mitofusins (130, 131), VDAC/IP3R/GRP-75 (14, 138), and VAPB/ PTPIP51 (135) is required for efficient transport of Ca2+ between organelles. Besides, several regulatory proteins modulate Ca2+ release from the ER and uptake by mitochondria, through interactions with IP3R or VDAC (123, 138, 139). Of particular interest is the chaperone calnexin, whose translocation to different ER domains is regulated by palmitoylation of two conserved cysteine residues (140) and interaction with regulatory proteins (138). Calnexin migration to the ERMCs is required for Ca2+ transfer from the ER to the mitochondria (141).

Calcium ion leaves the ER through the I3PR and enters the mitochondria through VDAC and the mitochondrial calcium uniporter (MCU) complex present in the OMM and IMM, respectively (14, 138). The complex is formed by the MCU, a pore that allows the passage of the ions through the lipid membrane (142, 143), essential MCU regulator (ERMES) and mitochondrial calcium uptake 1 and 2 proteins (MICU1 and 2) that act as gatekeepers regulating Ca2+ uptake (144). Calcium entry to mitochondria occurs at the expense of the H+ gradient in the IMM (144). In the matrix, Ca2+ activates several dehydrogenases (including enzymes of the tricarboxylic acid cycle) and ATP synthase enhancing oxygen consumption rate and ATP synthesis (15, 144). Thus, MERCs are extremely important for mitochondrial function. On the other hand Ca2+ uptake by mitochondria prevents or “buffers” the increase of Ca2+ concentrations near the mouth of the IP3R affecting its activity as we will describe in detail in the next section (14).