Phosphorylations and Acetylations of Cytochrome c Control Mitochondrial Respiration, Mitochondrial Membrane Potential, Energy, ROS, and Apoptosis

Cytochrome c (Cytc) has both life-sustaining and cellular death-related functions, depending on subcellular localization. Within mitochondria, Cytc acts as a single electron carrier as part of the electron transport chain (ETC). When released into the cytosol after cellular insult, Cytc triggers the assembly of the apoptosome, committing the cell to intrinsic apoptosis. Due to these dual natures, Cytc requires strong regulation by the cell, including post-translational modifications, such as phosphorylation and acetylation. Six phosphorylation sites and three acetylation sites have been detected on Cytc in vivo. Phosphorylations at T28, S47, Y48, T49, T58, and Y97 tend to be present under basal conditions in a tissue-specific manner. In contrast, the acetylations at K8, K39, and K53 tend to be present in specific pathophysiological conditions. All of the phosphorylation sites and two of the three acetylation sites partially inhibit respiration, which we propose serves to maintain an optimal, intermediate mitochondrial membrane potential (ΔΨm) to minimize reactive oxygen species (ROS) production. Cytc phosphorylations are lost during ischemia, which drives ETC hyperactivity and ΔΨm hyperpolarization, resulting in exponential ROS production thus causing reperfusion injury following ischemia. One of the acetylation sites, K39, shows a unique behavior in that it is gained during ischemia, stimulating respiration while blocking apoptosis, demonstrating that skeletal muscle, which is particularly resilient to ischemia-reperfusion injury compared to other organs, possesses a different metabolic strategy to handle ischemic stress. The regulation of Cytc by these post-translational modifications underscores the importance of Cytc for the ETC, ΔΨm, ROS production, apoptosis, and the cell as a whole.

Keywords: cytochrome c, post-translational modifications, apoptosis, respiration, mitochondria, electron transport chain

In healthy mitochondria, the electron transport chain (ETC) together with ATP synthase produces the majority of cellular ATP via the oxidative phosphorylation (OxPhos) process [1]. Four complexes and two electron carriers form the ETC. Electrons enter through either complex I, via NADH, or complex II, via FADH2, which couples the citric acid cycle to the ETC. These electrons are transferred to the ubiquinone pool, generating ubiquinol, which reduces complex III. The electrons then reach cytochrome c (Cytc), which passes them to complex IV (cytochrome c oxidase, COX). The movement of Cytc between complex III and COX has been proposed to occur through 2D diffusion or restricted diffusion by Cytc sliding between the complexes, making it more efficient compared to the free diffusion model [2,3]. Finally, COX catalyzes the reduction of oxygen, the terminal electron acceptor, to water. Complexes I, III, and IV pump protons from the matrix into the intermembrane space (IMS). This generates the proton-motive force, which is composed of both a pH gradient and the mitochondrial membrane potential (ΔΨm) [4] allowing ATP synthase (complex V) to harnesses this electrochemical gradient to produce ATP.

Naturally, this process is tightly regulated to ensure that ATP production is equivalent to energy demand. In addition to the complexes being regulated at the transcriptional and translational level, other mechanisms, such as supercomplex formation, allosteric binding of ADP and ATP, and post-translational modifications, allow for direct control over complex activity [2,5,6,7]. There is proposed to be crosstalk between these regulatory mechanisms, as post-translational modifications of Cytc may optimize supercomplex functioning [3]. However, these processes can go awry, with the resulting mitochondrial dysfunction contributing to a variety of pathologies, including neurodegeneration, ischemia-reperfusion injury, and cancer [8]. In many disease conditions, mitochondrial dysfunction triggers increased reactive oxygen species (ROS) production, which then initiates the release of Cytc into the cytosol [9]. Cytc release from the mitochondria takes place via a two-step process. First, Cytc, which is typically associated with cardiolipin within the cristae, dissociates from cardiolipin following cardiolipin peroxidation [10,11]. Cristae remodeling and swelling via tBID allows for further Cytc clearance out of the cristae [12]. Second, Cytc must cross the outer mitochondrial membrane, which occurs via calcium-dependent mechanisms such as the permeability transition pore or calcium-independent mechanisms via the Bcl-2 family proteins [13,14,15]. Once released from the mitochondria into the cytosol, Cytc is rapidly reduced in otherwise healthy cells [16]. However, in apoptotic cells, it is the oxidized form that binds to apoptosis protease-activating factor-1 (Apaf-1), leading to formation of the apoptosome and activation of the caspase cascade, because Cytc is rapidly oxidized by COX to which it has access due to outer mitochondrial membrane permeabilization [17,18].

Pharmacological interventions, such as treatment with anthocyanins that reduce cytosolic Cytc, hold promise as tissue protective treatments by inhibiting apoptosome formation [19,20]. In cancer, it is worth noting that sublethal release of Cytc into the cytosol can desensitize the cancer cell to apoptosis [21]. In effect, the subcellular localization and regulation of Cytc controls the balance between cellular life and death. This review discusses how tissue-specific post-translational modifications of Cytc affect the functions of the protein as well as its impact on various disease pathologies. While reviews of post-translational modifications of Cytc exist [22,23], new discoveries since then demand updating the literature.

Cytc is a small, globular, 104 amino acid, evolutionarily optimized protein that carries out a variety of distinct functions. Cytc possesses a covalently attached heme group, which is crucial for carrying out its redox-based functions. The two major functions of Cytc were already introduced above: it operates a single electron carrier from complex III to COX of the mitochondrial ETC and initiates intrinsic apoptosis via binding to Apaf-1 when released into the cytosol. Cytc also supports other crucial, life-sustaining processes (Figure 1). For example, Cytc scavenges and detoxifies ROS including superoxide and hydrogen peroxide [24,25,26] and acts as an electron acceptor of the redox-coupled import pathway of proteins to the IMS via the Erv1-Mia40 pathway [27]. Additionally, a new area of investigation is the role of Cytc in nuclear-mitochondrial crosstalk. In response to DNA damage, Cytc has been shown to translocate to the nucleus, where it promotes both Protein Phosphatase 2A activity via interaction with histone chaperone ANP32A and the release of alternative reading frame protein from nucleophosmin, facilitating DNA damage repair [28,29,30]. However, it still has to be determined if this function of Cytc can rescue cells from apoptosis or only delay it, because binding to Apaf-1 in the cytosol competes with its nuclear function.

Moreover, Cytc is involved in a variety of pro-apoptotic processes beyond its interaction with Apaf-1, and dysfunction in programmed cell death is relevant to numerous disease processes. Cytc is able to produce ROS via p66Shc [31,32], which can damage the cell and trigger cell death when excessive ROS are generated. As an early step during apoptosis, Cytc catalyzes the formation of cardiolipin peroxide [11,33,34,35]. Cardiolipin remodeling is an on-going research topic, which is also related to supercomplex formation, and defects in this process have been implicated in different disease pathologies [36,37,38]. Interestingly, the cardiolipin peroxidase activity of Cytc may be driven by ROS produced during ischemia-reperfusion injury [39].

Overall, the functional diversity of Cytc underscores the need for tight regulatory control, particularly when cellular fate hangs in the balance. Various post-translational modifications of Cytc have been discovered, some of which have been functionally characterized (Figure 2). Some of these modifications, or the loss thereof, result in dysregulation of Cytc and have been associated with specific pathologies. Although other post-translational modifications, such as methylation, nitration, nitrosylation, and homocysteinylation, have been reported, this review focuses on the effects of phosphorylations and acetylations of Cytc that have been identified in vivo, along with any associated pathologies.

Phosphorylation of Cytc (Table 1) has been identified on threonine 28 (T28), serine 47 (S47), tyrosine 48 (Y48), threonine 49 (T49), threonine 58 (T58), and tyrosine 97 (Y97). These residues, except for T58, which is replaced with isoleucine in humans, are conserved in mammals. Cytc was not known to be regulated via cell-signaling mechanisms until the discovery of Y97 phosphorylation in 2006, which was mapped in bovine heart [40]. This discovery was made using protein purification protocols that preserve the in vivo phosphorylation state of mitochondrial proteins. Despite this advance, studying post-translational modifications of Cytc remains challenging. Protein purified from animal tissue exists as a mixture of modified and unmodified forms, and depending on the tissue, yields can be low. Therefore, these phosphorylation sites were also functionally characterized using phosphomimetic replacement strategies, most commonly by glutamate replacement, which mimics the negative charge of the phosphate group. Following Y97 phosphorylation, Y48 phosphorylation was identified in bovine liver [41], followed by T28 [42] and T58 [43] in bovine and rat kidney, respectively, S47 phosphorylation in rat and porcine brain [44], and, most recently, T58 in rat heart [45].

A general theme for phosphorylation sites identified on Cytc is slowed electron transfer kinetics to COX and, although not universally true, decreased ability to activate downstream caspases. The binding of Cytc, which is a highly basic protein with a pI of 9.6, with both COX and Apaf-1 is largely predicted to be driven by electrostatic interactions, with the positively charged Cytc binding to negatively charged binding pockets on both proteins [2,17,52,53,54,55,56,57,58]. Therefore, it can be expected that phosphorylation, which introduces a negative charge, generally reduces the ability of Cytc to interact with these two key partners.

Reducing ETC flux through phosphorylation of Cytc may at first sight seem disadvantageous. However, similar to batteries that work best at an intermediate charge, the same argument can be made for mitochondria. This is because of the relationship between ETC activity, ΔΨm, and mitochondrial ROS production [22,59,60]; mitochondria with optimal, intermediate ΔΨm levels in the range of about 100–130 mV generate ATP by ATP synthase at close to maximal capacity [61], whereas ROS production is minimal. At membrane potentials exceeding 140 mV, ROS production increases exponentially [62,63,64], which damages the cell and can result in cell death. Therefore, we propose that one major role of Cytc phosphorylation is maintenance of optimal intermediate ΔΨm values, given that these phosphorylation sites tend to partially inhibit ETC activity, thus limiting mitochondrial ROS production to low basal levels. During acute conditions of stress, including ischemia as seen in ischemic stroke and myocardial infarction, these protective phosphorylations are lost in an attempt by the cell to boost ATP production and because kinases cannot operate due to ATP depletion while phosphatases are still operational or even activated due to mitochondrial calcium influx. Since there is no oxygen, the ETC is idling but primed for hyperactivity. During this time, succinate accumulates due to a reversal of complex II [65]. When blood flow is reinstated by mechanical or enzymatic removal of the clot, oxygen reenters the affected tissue. ETC activity resumes at pathological, maximal speed, leading to ΔΨm hyperpolarization and excessive ROS production at complexes I and III, which is further amplified by succinate-driven reverse electron transport to complex I (Figure 3) [65,66,67].

We propose that this sequence mechanistically explains reperfusion injury, which is a significant contributor to cell death on top of that caused initially by ischemia. Therefore, understanding the effect of post-translational modifications of Cytc will provide a mechanistic understanding of pathophysiological processes that occur in a highly tissue-specific manner. Below we discuss in detail the functional effects of identified phosphorylations and acetylations of Cytc.

T28 phosphorylation was mapped in Cytc purified from bovine kidney [42]. For a post-translational modification to be biologically relevant, an appreciable portion of the protein pool must possess the modification. High resolution gel electrophoresis found that up to 83% of the protein pool purified from the entire kidney was phosphorylated. The Vmax of COX in reaction with phosphorylated T28 Cytc was 50% reduced compared to the maximal activity of COX in reaction with non-phosphorylated wild-type (WT) Cytc. Additionally, the Km decreased from 6.3 μM to 4.5 μM. Similar results were obtained using phosphomimetic T28E Cytc, which showed a 73% reduction in Vmax and a decrease in Km from 7.8 μM to 4.7 μM in the reaction with COX [42]. These profound results may be explained by the strategic position of T28, which is part of the negative classical γ turn of Cytc, which stabilizes the protein and is part of the positive epitope that facilitates Cytc binding to its negatively charged binding site on COX [68]. Additionally, during the interaction between Cytc and COX, T28 of Cytc comes into close proximity with D50 and D51 of COX subunit I, which are known to be flexible residues, with D51 proposed to be involved in proton pumping [69,70]. Altogether, this suggests that phosphorylation of T28 perturbs the interaction between Cytc and COX. Another study looked at T28D and found increased electron transfer to COX [46]. However, aspartate is not evolutionarily allowed at the site, while glutamate is allowed in certain species of plants, implying that T28E is the better phosphomimetic since it recapitulates the effects of phosphorylated Cytc [71]. The opposing effects of T28E and T28D on COX activity may be explained based on structural considerations of the Cytc-COX complex in which T28 localizes near K47 of COX subunit II [72].

Based on the charge–charge interaction model, both the T28E and T28D mutations should strengthen the interaction between Cytc and COX, slowing down dissociation and thus the overall reaction. However, considering salt bridge formation, it is possible that electron transfer depends on the length and flexibility of the residue, which would be larger with T28E compared to T28D. With T28D, the complex could be restrained in a conformation allowing faster electron transfer, thereby explaining the opposite effects of the two mutations, despite showing the same charge change. Similar considerations may also apply to S47E and S47D (Figure 4).

The ability to activate downstream caspase-3 activity was unchanged comparing the phosphomimetic T28E to WT. The redox potential of T28E was decreased from 246 mV to 217 mV, which is near the typical reported literature range of 220 mV to 270 mV [73]. Compared to WT, T28E Cytc showed an increased rate of reduction by ascorbate, sometimes considered a stand-in for Cytc superoxide scavenging capability, and no change in rate of oxidation by H2O2. Interestingly, T28E Cytc resisted degradation from high concentrations of H2O2 better than WT, tracked spectrophotometrically via decrease in heme absorbance. T28E also showed slightly reduced cardiolipin peroxidase activity [42].

The effects of T28 phosphorylation were also studied using Cytc double knockout mouse lung fibroblasts, where both the somatic and testes isoforms of Cytc are knocked out. Double knockout is necessary because knockout of just the somatic isoform induces expression of the testes isoform [74]. A construct for T28E was transfected into the double knockout cells. Clones expressing equal levels of WT or phosphomimetic T28E Cytc were selected for further experimentation [42]. Cells expressing phosphomimetic T28E Cytc showed no change in proliferation compared to cells expressing WT after trypan blue staining. It is known that mitochondrial respiration, ΔΨm, and mitochondrial ROS production are linked [62]. This relationship was observed as the cells expressing phosphomimetic T28E showed 60% reduced intact cellular respiration, reduced ΔΨm measured via JC-1 fluorescence, 40% reduced mitochondrial ROS production measured via MitoSOX fluorescence, and 28% reduced ATP levels. Importantly, this demonstrates that modification of a single residue on Cytc by phosphorylation or mutation can control electron flux through the entire ETC. Further supporting the protective role of this modification, cells expressing phosphomimetic T28E Cytc underwent lower levels of cell death after treatment with low levels of H2O2 [42].

The signaling pathway for T28 phosphorylation was also elucidated. In vitro experiments with commercial AMPK and Cytc revealed that AMPK selectively phosphorylated T28. AMPK and Cytc were shown to interact via co-immunoprecipitation and co-localized together in the IMS via submitochondrial fractionation. Chemical activation of AMPK resulted in increased threonine phosphorylation of Cytc, while pharmacological inhibition resulted in loss of threonine phosphorylation of Cytc. Similarly, incubation of kidney tissue homogenates with pharmacological activators or inhibitors of AMPK resulted in increased or decreased respiration, respectively [42]. The kidney is an energetically demanding organ [75], and in contrast to other tissues, the kidney has high basal activity of AMPK and its dysregulation is known to result in kidney disease [76]. Furthermore, pharmacological activation of AMPK has been shown to be an effective treatment in a rat model of kidney ischemia-reperfusion injury [77].

Altogether, these experiments show that T28 phosphorylation is essential for kidney bioenergetic regulation. By partially inhibiting mitochondrial activity under basal conditions, T28 phosphorylation maintains an optimal intermediate ΔΨm and prevents the generation of excessive ROS. When this mechanism is lost, it would facilitate maximal respiration rates and result in increased ROS production [78].

S47 phosphorylation was characterized in Cytc purified from both pig and rat brain tissue and characterized [44]. S47 is the only serine residue present in porcine and rat somatic Cytc; however, human Cytc has a second serine residue at S15 [71]. Interestingly, S47 phosphorylation was found to be fully lost in ischemic brain tissue. Phos-tag gel electrophoresis determined that 35% of the pig Cytc pool was phosphorylated, which probably is an underestimation because it takes time to open the skull and remove the brain, a period of time during which the brain is ischemic and likely starts losing the modification. Altered kinetics in reaction with COX were also observed when using S47 phosphorylated protein purified from control brain tissue. The activity of the Cytc-COX reaction was reduced by 48% compared to the dephosphorylated protein purified from ischemic brain tissue. Docking simulations of the Cytc-COX interaction predict that S47 of Cytc would interact with K58 of COX subunit VIIa [52], highlighting the importance of this residue. Looking at the ability to initiate intrinsic apoptosis, S47 phosphorylated Cytc showed 62% decreased caspase-3 activity compared to the dephosphorylated protein [44].

Using glutamate as a phosphomimetic replacement, activity of the Cytc-COX reaction was 54% reduced and downstream caspase-3 activity was 65% reduced when S47E Cytc was used compared to WT. These results are similar to those obtained using the in vivo phosphorylated protein purified from porcine brain, indicating that S47E reproduces functions of in vivo phosphorylation. Other findings using the S47E protein were that it was 33% more resistant to degradation at high concentrations of H2O2 and that the cardiolipin peroxidase activity was decreased [44].

Another study looked at phosphorylation via replacement with aspartate at this site [46]. Using this strategy, caspase-3 and cardiolipin peroxidase activities were decreased whereas Cytc-COX activity was increased, which is opposite to the behavior of in vivo phosphorylated Cytc. In addition, molecular dynamics simulations using the S47E crystal structure (6N10.pdb) demonstrate better spatial similarities between S47E and in silico phosphorylated S47 (Figure 4), implying that glutamate replacement is a better phosphomimetic replacement than aspartate [44].

The Cytc knockout cell culture system was also employed to characterize the phosphomimetic S47E [47]. The cell line expressing phosphomimetic S47E Cytc showed a 52% decrease in basal respiration with a concurrent increased extracellular acidification rate, a stand-in for glycolytic activity. As a result of the decreased ETC activity, cells expressing phosphomimetic S47E Cytc also showed a decreased ΔΨm, a 30% decrease in MitoSOX fluorescence indicating reduced mitochondrial ROS production, and a slight decrease in ATP levels. These experiments again highlight that a single modification or point mutation on Cytc affects electron flux through the entire ETC and is able to induce global changes in overall mitochondrial activity. Using a model of oxygen-glucose deprivation followed by reoxygenation (OGD/R), which simulates ischemia-reperfusion injury in cell culture, the cells expressing WT showed a doubling in mitochondrial ROS production following OGD/R compared to control cells not subjected to OGD/R. In contrast, cells expressing phosphomimetic S47E Cytc were unresponsive to OGD/R and did not show any differences in mitochondrial ROS production. Lastly, in agreement with data suggesting that phosphomimetic S47E replacement reduces the pro-apoptotic abilities of Cytc, the cells expressing phosphomimetic S47E showed less cell death after exposure to H2O2 or OGD/R compared to WT [47].

Akt was identified as the putative kinase mediating S47 phosphorylation. In vitro kinase assays revealed that all three isoforms of Akt were able to phosphorylate Cytc on S47 [47]. Akt, specifically Akt2, is known to translocate to the mitochondrial IMS [79,80,81,82]. Treatment of pig brain homogenates with activators or inhibitors of Akt induced or suppressed the phosphorylation of S47, respectively [47]. Altogether, these data show that S47 phosphorylation acts to maintain a healthy mitochondrial activity level and serves as a protective mechanism in brain tissue. Because this modification is lost during ischemia, and because the loss facilitates high rates of respiration during reperfusion, this causes ΔΨm hyperpolarization and thus mitochondrial ROS production [59], providing a molecular mechanism explaining, at least in part, reperfusion injury seen in brain tissue following stroke treatment [67].

Cytc purified from bovine liver was shown to be phosphorylated on Y48 [41]. The kinetics in reaction with COX were measured using the Y48 phosphorylated protein purified from bovine liver compared to protein dephosphorylated in vitro by shrimp alkaline phosphatase. The Vmax of the Cytc-COX reaction was 55% decreased compared to the dephosphorylated protein, and there was also a slight reduction in the Km from 3.8 μM for dephosphorylated Cytc to 3.0 μM for Y48 phosphorylated Cytc [41]. Using a phosphomimetic glutamate replacement, Y48E Cytc showed a similar change in Vmax, which was 30% decreased compared to WT [48]. Interestingly, phosphomimetic Y48E Cytc showed an increase in Km from 1.1 μM to 3.7 μM. The midpoint redox potential of the phosphomimetic Y48E Cytc was greatly reduced to 192 mV in multiple publications [48,49], which is below the 220 mV to 270 mV range reported in the literature for Cytc [73]. These results suggest significant perturbations to the heme crevice, altering the ability of the protein to carry out efficient electron transfer [49]. Furthermore, this redox potential value is below that of complex III, suggesting that electron transfer from complex III to Cytc may also be impaired [83]. In addition, perturbing Y48 can be expected to disrupt this interaction because Y48 was predicted to be a key residue mediating the interaction between Cytc and complex III [84].

Regarding pro-apoptotic functions, the phosphomimetic Y48E Cytc was unable to activate downstream caspase-9 or caspase-3 activities in multiple studies [48,49,50]. Additionally, the cardiolipin peroxidase activity of phosphomimetic Y48E Cytc was reduced, which can be explained by reduced binding to cardiolipin [48].

Y48 phosphorylation has also been studied using p-carboxymethyl-L-phenylalanine (pCMF) replacement, incorporated via the evolved tRNA synthetase method [85]. Work with the novel phosphomimetic Y48pCMF yielded some different results [50]. While there was agreement regarding decreased downstream caspase-3 activity, Y48pCMF showed increased activity with isolated COX and increased cardiolipin peroxidase activity. Interestingly, although Cytc-COX activity was increased, the overall electron transfer rate through the ETC was decreased, possibly due to the perturbation between the Cytc-complex III interaction.

Y48H is a known disease-causing mutation of Cytc in humans, resulting in mild thrombocytopenia [86]. Y48H resulted in reduced respiration rate and increased apoptotic activity. Another study found that the mutation also caused the heme group to change to a pentacoordinated form, which increased cardiolipin peroxidase activity [87]. Given the crucial functions of Cytc and its high evolutionary conservation, it is surprising that one of the few known disease-causing mutations in humans only causes mild thrombocytopenia. Perhaps only mild mutations of Cytc are compatible with life.

T49 phosphorylation (numbering based on the mature protein, which lacks the start methionine) was reported to increase in aged mouse heart after detection via high-throughput phosphoproteomics [45]. It was characterized using the AC16 cell line, a human cardiomyocyte cell line, which was transfected using lentiviral constructs containing the sequence for phosphomimetic T49E Cytc. Using this system, the authors reported lower levels of cell death and lower levels of caspase-9 and caspase-3 activities. However, the authors did not report knocking out endogenous Cytc in their cell lines prior to transfection with the phosphomimetic T49E Cytc, nor did they use the double Cytc knockout mouse lung fibroblast cell line generated by the Moraes lab [74]. Therefore, the transfected cells likely express an unknown combination of both endogenous Cytc in addition to phosphomimetic T49E Cytc and data interpretation is thus rather limited.

T58 phosphorylation was mapped in Cytc purified from rat kidney [43]. Unlike T28 phosphorylation, which was detected in all five kidney preparations tested [42], T58 phosphorylation was only detected in two of five kidney preparations [43]. T58 is known to be replaced with isoleucine in the testes isoform of Cytc as well as in the single human Cytc gene [71]. Using phosphomimetic T58E Cytc, the kinetics of the Cytc–COX reaction were interrogated. The Vmax was 45% decreased compared to WT while the Km was unchanged. This inhibition of Cytc-COX activity is consistent with that seen with the other Cytc phosphorylations and supports the model that Cytc phosphorylation partially inhibits respiration to maintain optimal mitochondrial functioning [22,59].

Downstream caspase-3 activity of T58E Cytc was 70% decreased compared to WT. The redox potential of T58E was reduced to 209 mV and the ROS scavenging ability of Cytc was also altered; the T58E Cytc demonstrated a 50% reduced rate of oxidation by H2O2 and a 50% increased rate of reduction by ascorbate. The heme group of the T58E Cytc was slightly more resistant to degradation by high H2O2 concentrations compared to WT. Furthermore, the cardiolipin peroxidase activity of T58E Cytc was reduced at high concentrations of cardiolipin. T58 has been reported to be part of the cardiolipin binding site on Cytc, which suggests that the phosphorylation may partially disrupt this interaction [88].