Cytochrome c speeds up caspase cascade activation by blocking 14-3-3ε-dependent Apaf-1 inhibition

Elena-Real et al. Cell Death and Disease (2018) 9:365 DOI 10.1038/s41419-018-0408-1 Cell Death & Disease ARTICLE Open Access Cytochrome cspeeds up caspase cascade activation by blocking 14-3-3ε-dependent Apaf-1 inhibition Carlos A. Elena-Real 1 , Antonio Díaz-Quintana 1 , Katiuska González-Arzola 1 , Adrián Velázquez-Campoy 2 ,MarOrzáez 3 , Abelardo López-Rivas 4 , Sergio Gil-Caballero, Miguel Á. De la Rosa 1 and Irene Díaz-Moreno 1 Abstract Apoptosis is a highly regulated form of programmed cell death, essential to the development and homeostasis of multicellular organisms. Cytochrome cis a central figure in the activation of the apoptotic intrinsic pathway, thereby activating the caspase cascade through its interaction with Apaf-1. Our recent studies have revealed 14-3-3ε(a direct inhibitor of Apaf-1) as a cytosolic cytochrome ctarget. Here we explore the cytochrome c/ 14-3-3εinteraction and show the ability of cytochrome cto block 14-3-3ε-mediated Apaf-1 inhibition, thereby unveiling a novel function for cytochrome cas an indirect activator of caspase-9/3. We have used calorimetry, NMR spectroscopy, site mutagenesis and computational calculations to provide an insight into the structural features of the cytochrome c/ 14-3-3ε complex. Overall, these findings suggest an additional cytochrome c-mediated mechanism to modulate apoptosome formation, shedding light onto the rigorous apoptotic regulation network. Introduction Apoptosis, a specific form of programmed cell death (PCD), is an essential process for development and homeostasis of multicellular organisms 1 . There are two main apoptotic pathways in mammalian cells: the extrinsic or death receptor-initiated pathway and the intrinsic or mitochondria-dependent pathway 2,3 . The last is triggered by signals, including DNA damage, oxidative stress, and growth factor deprivation, which converge on the release of pro-apoptotic proteins from the mitochondria into the cytosol; one such protein is cytochrome c(Cc). Under homeostasis, Ccacts as an electron carrier in the mitochondrial respiratory chain 4,5 . Upon its release from mitochondria, Ccinteracts with WD40 domains of the apoptotic protease activating factor-1 (Apaf-1) in the cytosol, enabling the assembly of the apoptosome 6 . This protein platform activates caspase-9 and later caspase-3, initiating the caspase cascade, which executes cell death in an orchestrated way 7,8 . Recent findings suggest Ccfulfills multiple roles in apoptosis, beyond Apaf-1 activation and apoptosome assembly. Indeed, Ccinteracts with various cytosolic and nuclear partners along the onset of PCD 9,10 . Hence, the full scope of Ccrole in apoptosis remains un-elucidated. Recently, our group has reported that Ccinhibits the histone chaperone activity of SET/TAF-Iβin the nucleus, impairing the formation of core histone-SET/TAF-Iβ complexes under DNA damage 11 . However, the novel functions of cytosolic Ccstay unveiled, despite a complex network of interactions mediated by Ccduring apoptosis has been suggested 12 . Therefore, we focused on the interaction between Cc and protein 14-3-3ε, a novel cytosolic Cctarget under DNA damage 10 . This protein belongs to the 14-3-3 © The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a linktotheCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Correspondence: Miguel A. De la Rosa ([email protected]s) or Irene Díaz-Moreno ([email protected]) 1 Instituto de Investigaciones Químicas (IIQ) —Centro de Investigaciones Científicas Isla de la Cartuja (cicCartuja), Universidad de Sevilla –Consejo Superior de Investigaciones Científicas (CSIC), Sevilla, Spain 2 Institute of Biocomputation and Physics of Complex Systems (BIFI), Joint Unit IQFR-CSIC-BIFI, Universidad de Zaragoza, Zaragoza, Spain Full list of author information is available at the end of the article Edited by A. Finazzi-Agrò Official journal of the Cell Death Differentiation Association 1234567890():,; 1234567890():,; family 13,14 , which comprises seven conserved isoforms (β, γ,ε,η,σ,τ/θ, and ζ), arranged as homo- and heterodimers. Each monomer contains nine α-helices that form a conserved concave groove, used by 14-3-3 proteins to bind their phosphorylated targets 15 (Supplementary Figure S1). Furthermore, they are also involved in phosphorylationindependent interactions 16–18 . 14-3-3 proteins participate in several cell processes related to metabolism, signal transduction, cell cycle control, apoptosis, transcription, and stress responses 19–24 . Among 14-3-3εfunctions, its ability to inhibit Apaf-1 stands out because it prevents apoptosome assembly and caspase activation 25 . Such inhibition is enhanced by phosphorylation of Apaf-1 at Ser268 by the p90kDa ribosomal S6 kinase-1 (Rsk-1) when the mitogenactivated protein kinases (MAPK) cascade is active. Hence, the interaction of Ccwith 14-3-3εcould modulate such inhibition somehow. Herein, we show that Cchinders 14-3-3ε-mediated Apaf-1 inhibition. Indeed, our results indicate a competition between Ccand 14-3-3εfor binding to Apaf-1, which enhances caspase activation. Furthermore, this new regulatory mechanism is modulated by phosphorylation of Apaf-1, which enhances its inhibition by 14-3- 3ε. We further show that Ccbinds to both the 14-3-3ε concave groove and the convex face, thereby providing a molecular basis for this novel modulation of apoptosome assembly. Results Ccinteracts with 14-3-3εin the cytosol under apoptotic conditions To elucidate the extra-mitochondrial function of Cc,we explored the interaction of Ccwith 14-3-3εwhen apoptosis is triggered. To this aim, HeLa cells were treated with the topoisomerase I inhibitor camptothecin (CPT), to induce DNA damage. Then, subcellular fractionation was performed, and Ccwas detected in the cytosol (Fig. 1a, lane 3). However, it remained inside the mitochondria under homeostasis (Fig. 1a, lane 2). The in cell Cc/ 14-3-3εinteraction was established as immunoprecipitation (IP) of cytosolic proteins associated with Cc yielded intrinsic 14-3-3εin CPT-treated cells (Fig. 1b, lane 6). To further confirm the IP specificity, untreated and CPT-treated cytosolic lysates were probed with a 14-3-3ε antibody (Fig. 1b, lanes 1 and 4, respectively). Negative controls using IgG (Fig. 1b, lanes 2 and 5) did not display any band. Immunoblotting against the anti-Ccantibody confirmed CcIP (Fig. 1b, lane 6). Ccblocks 14-3-3ε-mediated caspase inhibition Following its release into the cytosol, Cctargets Apaf-1 to assemble the apoptosome 6 . As 14-3-3εbinds Apaf-1 to prevent caspase activation 25 , we investigated whether Cc modulates Apaf-1 inhibition by 14-3-3ε. First, we checked the ability of Ccto activate caspase-3 in HEK293T cytosolic cell extracts. Caspase-3 activity was monitored upon Ccaddition (Fig. 2a, white columns), resulting in a substantial increase of such activity, as the hemeprotein triggered the apoptosome formation and, subsequently, caspase-9, -3 activation. MAPK proteins modulate Apaf-1 activity, leading the activation of Rsk-1, which phosphorylates Apaf-1 at Ser268, favoring the interaction between Apaf-1 and 14-3- 3ε, thereby increasing Apaf-1 inhibition 25 . Therefore, we tested the effect of phorbol 12-myristate 13-acetate (PMA), which triggers the MAPK cascade, on the caspase- 3 activity. Apaf-1 phosphorylation under PMA treatment was confirmed by Phos-tag TM SDS PAGE (Supplementary Figure S2). PMA-treated cells extracts showed a lower caspase-3 activity than untreated extracts in the presence of Cc(Fig. 2a, gray columns). This is consistent with Apaf- 1 phosphorylation at Ser268, enhancing its inhibition by 14-3-3ε 25 . Fig. 1 Cclocalization in the cytosol upon cell CPT-treatment. a Subcellular fractioning showing the Cclocation upon cell treatment with 20 μM CPT for 4 h. Purity of fractions was verified by western blot using anti-α-Tubulin and anti-CoxIV antibodies for detecting cytosolic and membrane proteins, respectively. bCc-IP of cytosolic fractions from non-treated (lane 3) and CPT-treated (lane 6) along with the detection of 14-3-3εby western blot (upper). Mouse IgG was used as control (lanes 2 and 5). Verification of immunoprecipitation of Ccupon CPT treatment is shown (lower) Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 2 of 12 Official journal of the Cell Death Differentiation Association Then, we tested the ability of recombinant 14-3-3ε (herein named as 14-3-3εfull length, 14-3-3εFL) to inhibit caspase-3 activity in non-treated and PMA-treated cell extracts. Addition of 14-3-3εFL made extracts to exhibit significantly lower Cc-mediated caspase-3 activity values (Fig. 2a). These were normalized to identify the particular effect of added 14-3-3εFL but, unexpectedly, the normalization evinced that the inhibition exerted by 14-3-3εFL was similar in non-treated and PMA-treated extracts (Fig. 2b). One of the plausible explanations for this observation is endogenous 14-3-3εalready abolished the caspase-3 activity by preferentially binding to phosphorylated Apaf-1, thus preventing any additional effect of exogenous 14-3-3εFL. Additional caspase-3 assays were carried out using extracts from 14-3-3εknockdown HEK293T cells (Fig. 2c), with the caspase-3 activity in PMA-treated extracts resulting more sensitive than nontreated extracts to the addition of 14-3-3εFL (Fig. 2d). In summary, these findings demonstrate how 14-3-3ε FL inhibits caspase activity, and how this inhibitory mechanism can be modulated by Apaf-1 phosphorylation. Nevertheless, the inhibitory effect of 14-3-3εFL on caspase-3 could be caused by either a direct interaction with Apaf-1 or by hindering the complex between the latter and Cc. Caspase-9 can indeed be phosphorylated by the ERK MAPK pathway, with the resulting modulation of its function and consequently, of caspase-3 activation 26 . To discard the direct interaction between Ccand Apaf- 1, as well as the ERK-mediated inhibition of caspase-9, we performed caspase-9 activity assays using a recombinant, truncated form of Apaf-1 lacking the Cc-binding WD40 domains (Apaf-1 ΔWD40 ) 27 . Apaf-1 ΔWD40 activates caspase- 9 in a constitutive manner in the absence of Cc(Fig. 3). The inability of Ccto alter the Apaf-1 ΔWD40 -mediated caspase-9 activity was likewise determined (Supplementary Figure S3). Then, the addition of 14-3-3εFL inhibited caspase-9 activity strongly, due to the binding of Apaf- 1 ΔWD40 to 14-3-3εFL (Fig. 3). However, the sequentially titration with Ccrestored caspase-9 activity, indicating a competition between Ccand Apaf-1 ΔWD40 for the binding Fig. 2 14-3-3εFL inhibits caspase-3 activity in HEK293T cytosolic cell extracts. a Relative caspase-3 activity in non-treated (white columns) and PMA-treated (gray columns) HEK293T cytosolic cell extracts were measured upon addition of recombinant Ccand increasing concentrations of 14-3- 3εFL. bNormalized data shown in a.cHEK293T cells were transfected with empty pSUPER (Control) or with pSUPER-shRNA against 14-3-3εand immunoblotted with an anti-14-3-3εantibody to examine the amount of remaining endogenous 14-3-3ε. An immunoblotting with anti-α-tubulin was used as loading control. dNormalized caspase-3 activity in 14-3-3ε-silenced non-treated and PMA-treated cell extracts upon addition of recombinant 14-3-3εFL. Normalized caspase-3 activity of non-14-3-3ε-silenced, non-treated cell extracts is included as a control Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 3 of 12 Official journal of the Cell Death Differentiation Association site of 14-3-3εFL, impairing Apaf-1 inhibition. Nuclear magnetic resonance (NMR) experiments corroborated such a competition. Specifically, the ε-methyl group signal of Met80 in reduced Ccsubstantially broadened upon binding of the hemeprotein to 14-3-3ε(Supplementary Figure S3e). Further addition of Apaf-1 ΔWD40 yielded partial recover of the signal intensity, indicating that Apaf-1 ΔWD40 competes with Ccfor the 14-3-3εFL binding site. The inability of Apaf-1 ΔWD40 to bind Ccwas corroborated (Supplementary Figure S3f). As phosphorylation of Apaf-1 at Ser268 enhances its interaction with 14-3-3εFL, we tested the ability of Ccto modulate the 14-3-3εFL / Apaf-1 ΔWD40 complex upon phosphorylation. To this end, we designed the recombinant Apaf-1 ΔWD40 S268D mutant, which mimics Ser268 phosphorylation. As a control of unspecific effects caused by mutations at such position, the Apaf-1 ΔWD40 S268A variant was also tested. Notably, the ability of phosphomimetic Apaf-1 ΔWD40 S268D mutant to trigger caspase-9 activity was similar to that exhibited by Apaf-1 ΔWD40 . Nevertheless, the Apaf-1 ΔWD40 S268D-mediated caspase- 9 activity was slightly more sensitive to 14-3-3ε(Fig. 3, gray columns). Indeed, the recovery of enzymatic activity by Ccwas substantially lower than that measured using Apaf-1 ΔWD40 (Fig. 3, gray columns). Such a finding indicates a larger affinity of the phosphorylated Apaf-1 species toward 14-3-3εthat displaces Cc 25 . On the other hand, the caspase-9 activation in the presence of the S268A mutant behaved in a way similar to that in the presence of Apaf-1 ΔWD40 (Supplementary Figure S3b). Hence, the negative charge at position 268 in Apaf-1 is responsible for the increase in affinity of Apaf-1 toward 14-3-3εFL. Our results thus support a model in which Ccactivates Apaf-1 while blocking the inhibition exerted by 14-3-3ε on the apoptotic factor (Fig. 4). Ccrecognizes two different binding sites of 14-3-3εfull length The dimerization state of 14-3-3εFL was confirmed by analytic ultracentrifugation combined with dynamic light scattering (Supplementary Figure S4a-c). Sedimentation velocity experiments confirmed the dimeric state of 14-3- 3εFL with a molecular mass of 66.4 kDa (Supplementary Figure S4b). Dynamic light scattering measurements corroborated that 14-3-3εFL behaves as a dimer, with a particle size of 9 ± 2 nm (Supplementary Figure S4c). Isothermal titration calorimetry (ITC) experiments in Supplementary Figure S5 revealed that Ccbinds to 14-3- 3εFL homodimer with a 2:1 stoichiometry and a dissociation constant (K D ) of 2.3 μM (Supplementary Table S1). To reveal 14-3-3εresidues implicated in the interaction with Cc, 14-3-3εFL D21K, K50E, S59E, E92K, D99K, and S187D mutants were designed. Residues were selected according to their role in the interactions of 14-3- 3 isoforms with their respective partners 28–30 and the impact of each mutation in the electrostatic surface potential of the protein. The dimeric state of these mutants was confirmed by analytic ultracentrifugation and dynamic light scattering (Supplementary Figure S4c and d). According to ITC analysis (Supplementary Table S1 and Figure S5), the six mutants bound to Ccwith slightly larger K D values than 14-3-3εFL WT. Further, the complexes involving 14-3-3εFL D21K, K50E, S59E, or E92K showed a 1:1 stoichiometry instead of the 2:1 found for the WT species (Supplementary Table S1), suggesting that one of the two Cc-binding sites of 14-3-3εFL has been compromised in these mutants. Therefore, Asp21, Lys50, Ser59, and Glu92 residues may be essential in Cc recognition. We also performed NMR experiments to further characterize the interaction between reduced Ccand 14-3-3ε FL. We followed Ccsignals upon titration of 14-3-3εFL, which caused an overall broadening of Ccresonances Fig. 3 Ccimpairs 14-3-3εFL-mediated caspase-9 inhibition. Cc and 14-3-3εFL were added to caspase-9 and different Apaf-1 ΔWD40 mutants Fig. 4 Model of Apaf-1 regulation by Ccand 14-3-3ε.Model of Apaf-1 regulation by Ccand 14-3-3ε. During homeostasis, Apaf-1 is available to interact with 14-3-3ε 25 . Upon its release into the cytosol, Cccompetes with Apaf-1 for the binding to 14-3-3ε, impairing the Apaf-1 / 14-3-3εcomplex. Thus, Apaf-1 is available to interact with Cc, thereby activating the caspases cascade 6 . Upon Apaf-1 phosphorylation, the equilibrium of the reaction is strongly shifted toward the Apaf-1 / 14-3-3εcomplex formation. This would hinder the competition performed by Cc, thereby partly impairing the activation of the caspases Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 4 of 12 Official journal of the Cell Death Differentiation Association because Cc/ 14-3-3εFL complex formation (Fig. 5a and Supplementary Figure S6a, left). Specifically, the ε-methyl group signal of the Met80 Ccresidue was largely broadened at 1:0.5 Cc:14-3-3εFL ratio (Fig. 5a), in agreement with the stoichiometry inferred from ITC data. Moreover, specificCcbackbone amide groups underwent singularly large changes in their 1 H line widths ( 1 HΔΔν 1/2 Binding ) upon binding with 14-3-3εFL (Supplementary Figure S6b). These resonances were expected to be at or in the proximity of the area of Ccinvolved in the recognition of 14-3-3εFL. Moreover, several signals displayed substantial chemical-shift perturbations (CSPs) in the presence of 14-3-3εFL (Supplementary Figure S6a, right). Specifically, Gln16, Thr19, Ser47, Lys72, Gly77 and Val83 signals showed significant CSPs (Δδ avg > 0.035 ppm.). Interestingly, the signal corresponding to Val83 (Supplementary Figure S6a, right) seems to be in conformational exchange in the free Cc,fixing its conformation upon the interaction with 14-3-3ε. These effects are consistent with an intermediate exchange rate at the NMR time scale, in agreement with the transient nature of this complex and the K D values in the μM range. Both broadening and CSPs maps on Ccsurface (Fig. 5b) indicate that Ccmainly recognizes 14-3-3εFL by a surface patch surrounding the heme cleft, as it does in its homeostatic interactions in the electron transport chain 31–33 . The C-terminal tails of 14-3-3εare important in the interaction with Cc To decipher the role of the unstructured C-terminal tail of 14-3-3εFL during Ccrecognition, the interaction between Ccand the 14-3-3εcore protein, lacking C- terminal tails, was measured by ITC. The recorded thermograms showed very similar thermodynamic parameters for both 14-3-3εFL and 14-3-3εcore proteins interacting with Cc, although a slight increase in the K D value was observed in the interaction between Ccand the 14-3-3ε core (Supplementary Table S1). This suggests that the C- terminal tails of 14-3-3εmay participate in the interaction with Cc, despite being unessential. To confirm that the C-terminal tails of 14-3-3ε(named 14-3-3ε 234-255 ) are involved in the interaction with Cc,2D [ 1 H, 15 N] HSQC were recorded along a titration of 14-3- 3ε 234-255 peptide onto 15 N-labeled reduced Cc Fig. 5 Binding of Ccto 14-3-3εas analyzed by NMR. a 1D 1 H NMR spectra of Cc(Fe 2+ ) along the titration with 14-3-3εFL. The complex formation is monitored by line broadening of the Met80-εCH 3 signal, included in the black square. bMapping of the Ccresidues perturbed upon binding to 14- 3-3εFL. The Cc:14-3-3εratio is 1:0.75. Residues are colored according to their 1 HΔΔν 1/2Binding (Supplementary Figure S6b). Residues with average chemical-shift perturbations (Δδ avg ) larger than 0.035 p.p.m. are represented in orange. Asterisks stand for residues showing both significant 1 HΔΔν 1/ 2Binding and Δδ avg . Residues without substantial 1 HΔΔν 1/2Binding and Δδ avg are in blue. cDetail of superimposed 2D [ 1 H, 15 N] HSQC spectra of 15 N- labeled Ccalong titration with 14-3-3ε 234-255 . Signals corresponding to distinct titration steps are colored according to the code in the panel. Arrow indicates direction of CSPs. dCSP map of reduced Ccupon addition of 14-3-3ε 234-255 . Residues are colored according to Δδ avg categories: insignificant Δδ avg < 0.020 ppm., blue; small 0.020 ppm. ≤Δδ avg < 0.025 ppm. yellow ; and large Δδ avg > 0.025 ppm., red. In b,dprolines and nonassigned residues are colored in gray; whereas, the heme group in green. e1D 19 F NMR spectra of free 5-fluorindole (upper) and when it is conjugated with 14-3-3ε 234-255 peptide (lower). f1D 19 F NMR spectra of 14-3-3ε 234-255 whose Lys244 is covalently bound to 5-fluorindole free and in the titrations with Ccand BSA. 19 F CSPs are colored according to the panels Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 5 of 12 Official journal of the Cell Death Differentiation Association (Supplementary Figure S6c). Such titration resulted in significant CSPs of the Ccamide signals (Fig. 5c). The resulting interaction surface (Fig. 5d) indicates that 14-3- 3ε 234-255 binds to the rim of the heme groove of Cc,as with 14-3-3εFL protein. However, unlike the 14-3-3εFL interaction, most of the Ccresonances were in fast exchange on the NMR time scale upon binding to the 14- 3-3ε 234-255 peptide. 1 HΔδ bind was fitted using the best 1:1 binding model, obtaining a K D of 49 μM (Supplementary Figure S6d). Moreover, ITC experiments in which the 14- 3-3ε 234-255 peptide was titrated onto Ccyielded a K D value of 16 μM (Supplementary Figure S5 and Table S1). Both NMR and ITC K D values are marginally higher than the value obtained for the interaction between 14-3-3εFL and Cc. Furthermore, the ITC analysis also confirmed a 1:1 stoichiometry. Then, to easily determine a residue which could be included in the latter docking calculations, the only one positively charged residue (Lys244) of 14-3-3ε 234-255 peptide was tagged with 5-fluorindole molecule through a conjugation reaction. To prevent the reaction of 5- fluorindole with the N-terminal −NH 3+ group, the peptide was modified by acetylation. To confirm that the acetylation in the N-terminal residue did not drastically modify the interaction with Cc, ITC experiments were performed with the acetylated 14-3-3ε 234-255 . Both the non-acetylated and acetylated peptides interacted with Cc in a similar way, despite the K D value showed a slight increase when the peptide was acetylated (Supplementary Table S1). Proper folding of 14-3-3ε 234-255 before and after the conjugation reaction was unambiguously assessed by 1D 1 H NMR spectra (Supplementary Figure S6e). Second, 1D 19 F NMR spectra of both free 5-fluorindole and fluorinated 14-3-3ε 234-255 were registered. The spectrum of fluorinated 14-3-3ε 234-255 peptide showed a signal at −125.53 ppm. corresponding to free 5-fluorindole, and another at −123.62 ppm. that was assigned to 5- fluorindole conjugated to the epsilon −NH 3+ group of Lys244 (Fig. 5e). The addition of increasing concentration of Ccresulted in significant CSPs of this second signal (Fig. 5f, left panel). Notably, the 19 F NMR signal also experienced a significant broadening consistent with the larger correlation time of the peptide bound to Cc. As a negative control, bovine serum albumin (BSA) protein was titrated into fluorinated 14-3-3ε 234-255 peptide, with the negligible resulting 19 F CSPs, suggesting that BSA cannot significantly bind to the peptide. (Fig. 5f, right panel). These results agree with Lys244 of 14-3-3ε FL taking part in Ccrecognition. Therefore, Lys244 was included as an active residue in further docking calculations (see below). Structural model of the interaction between Ccand 14-3- 3εfull length To further understand the molecular features of the interaction between Ccand 14-3-3εFL, we performed restrain-driven docking computations using HADDOCK software 34 . For these computations, we used the structural model of 14-3-3εFL in its open conformation inferred from MD as the input structure (Supplementary Information and Figure S7), to allow the entrance of Ccinto the concave groove of the 14-3-3εdimer. Line widths and CSPs obtained from NMR analysis of the Cc/ 14-3-3εFL complex were used as Ccambiguous interaction restrains (AIRs). The 14-3-3εFL residues that resulted in the impairment of one of the Cc-binding sites by ITC were used as 14-3-3εAIRs. The 500 solutions obtained in the computation showed two main Cc-binding sites on 14-3- 3εFL dimer, in agreement with the stoichiometry inferred from ITC data. These solutions located Ccin both the concave and convex faces of 14-3-3ε; these interacting regions will be referred hereafter to as the concave and convex binding sites, respectively. The convex site resulted more populated, although some of the lowest energy solutions locate within the concave site (Fig. 6, left panel). Statistical classification of solutions yielded three clusters in the convex site of 14-3-3εand only one in the concave site (Supplementary Figure S8a). Further analyses indicate that electrostatics is the major contributing factor to Fig. 6 HADDOCK molecular docking of Ccand 14-3-3εFL. Mass centers of Ccon the structural model of 14-3-3εFL on its open conformation, including (right) or not (left) Lys244 as an active residue (AIR). Mass centers of Ccare colored as a function of intermolecular energy (E inter ) according to the scales represented in their corresponding panels Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 6 of 12 Official journal of the Cell Death Differentiation Association binding energy in the three clusters (Supplementary Table S2). However, the concave cluster was more energetically favored than the convex clusters, mainly due to a lower electrostatic energy (E elec ). Nevertheless, convex clusters presented lower AIRs violations than the concave cluster. Furthermore, to study the role of the 14-3-3εC- terminal tails in the interaction, a docking computation was also performed including Lys244 (located in the C- terminal tail), as an active residue. This computation showed an increase in the number of solutions locating Cc in the concave binding site of 14-3-3ε(Fig. 6, right panel), suggesting that the C-terminal tails contribute to the entrance of Ccinto the concave groove of 14-3-3ε. Clustering analysis exhibited three clusters of solutions presenting Ccbound to the concave site of 14-3-3ε, and only one displaying Ccbound to the convex site (Supplementary Figure S8b). Again, electrostatic interactions constituted the major contribution in all clusters, concave site clusters displaying lower energies because of their E elec term (Supplementary Table S2). Structural models of Cc/ 14-3-3εFL complex were obtained for computations excluding and including (Supplementary Figure S8c and d, respectively) Lys244 as an active residue of 14-3-3ε. For this, the best two complexes with minimal E inter at each binding site of 14-3-3ε were selected from each computation. Both structural models present a Ccmolecule located in the concave groove of 14-3-3ε, contacting at least one of the C- terminal tails of 14-3-3εdimer. A second Ccmolecule is placed at the convex side of 14-3-3ε, specifically at the rim of the dimerization surface thereby binding both 14-3-3 monomers. In addition, the two Ccmolecules from both computations present their heme clefts oriented to 14-3- 3ε, according to NMR data. Discussion Due to the apoptosis significance in multicellular organisms, its regulation relies on a subtle balance between inhibition and activation pathways. In fact, signaling cascades convert continuous stimuli in an all-or- none response to avert physiological fluctuations to trigger a response accidentally 35 . Feedbacks are needed to lead response ultra-sensitivity: a large stimulus is required to trigger a response, which changes dramatically within a narrow range of stimulus intensity. During homeostasis, apoptotic factors remain inhibited, either blocked by other proteins or sequestered in a particular cell compartment. Along apoptosis, several apoptotic factors are released from the mitochondria to the cytosol. Among these factors is Cc, which promotes the apoptosome assembly through its association with Apaf-1, thereby initiating the caspase cascade 6 . On the other hand, prosurvival pathways, as the inhibition of Apaf-1 by 14-3- 3ε 25 , reduce the sensitivity to the apoptosis signaling network. Therefore, interactions between pro-apoptotic and prosurvival proteins, such as that between Ccand 14-3-3ε, are of particular relevance as they provide additional regulation loops. In this study, we have shown that Cc interacts with 14-3-3εin the cytosol under apoptotic conditions, preventing the binding of 14-3-3εto a specific site in Apaf-1. Hence, this interaction could be capable of regulating caspase activity. Indeed, Ccenhanced caspase activation in presence of 14-3-3εinhibitor and a truncated, constitutive form of Apaf-1 unable to bind the hemeprotein. Thus, Ccplays a double role in the activation of caspase cascade directly by interacting with Apaf- 1, and indirectly by blocking 14-3-3εthereby releasing Apaf-1. This double role enables Ccto inhibit pro-survival pathways while apoptotic routes have been triggered (Fig. 4). Further phenomena modulate the cell fate signaling, including protein phosphorylation events, as those which regulate Ccfunctions, including its ability to activate caspases 36–40 . In addition, such a phosphorylation enhances the inhibition of Apaf-1 by 14-3-3ε 25 , making caspase activation less sensitive to Cc. In fact, activation of MAPK cascade favors the Apaf-1 inhibition by 14-3-3ε. Similarly, the phosphomimetic Apaf-1 mutant is more susceptible to 14-3-3ε-dependent inhibition, which consequently becomes less sensitive to the presence of Cc. Hence, this phosphorylation of Ser268 in Apaf-1 may attenuate the effect of Ccrelease when caspase activation is undesirable (Fig. 4). Ccdoes lack the consensus sequence which 14-3-3ε usually recognizes in its various targets. Our biophysical analysis has revealed that two molecules of Cccan interact with 14-3-3εat two different sites: the groove of 14-3-3ε where its targets bind, and the convex side of the dimer. Additionally, the data herein suggest that flexible C- terminal 14-3-3εmay recognize Ccwithout being essential for the binding. These tails are flexible and mostly unstructured, although MD trajectories indicate the presence of residual helical elements. Moreover, NMR data show that Ccrecognizes 14-3-3εby residues surrounding the heme cleft, as reported for other interactions involving the hemeprotein 31–33 . In summary, Ccis revealed as a double Apaf-1 activator via two mechanisms: by directly binding to the apoptosis factor and impairing its inhibition by 14-3-3ε. Indeed, such a cooperative behavior of Cc, triggering effector pathways while modulating feedbacks loops make cell response ultrasensitive to its release. The analysis of the various additional interactions of Ccduring the onset of programmed cell death recently unveiled may confirm if this conclusion can be extended as a general mechanism by which Cctriggers apoptosis 9–12,41 . Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 7 of 12 Official journal of the Cell Death Differentiation Association Materials and methods Cell cultures HeLa and Human Embryonic Kidney 293T (HEK293T) cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, and maintained at 37 °C in a humidified 5% CO 2 atmosphere. For caspase-3 assays, 4.5 × 10 6 HEK293T cells were cultured in 140 mm Petri dishes in DMEM containing FBS until 70% confluence was reached, and cells were then kept in DMEM without FBS for 4 h. In some experiments (indicated in the main text) cells were treated with 50 ng/mL phorbol-12-myristate-13-acetate (PMA, Sigma-Aldrich) or treated with the same volume of DMSO for 120 min. shRNA knockdown In order to knockdown 14-3-3εexpression in HEK293T cells, a construct containing the shRNA sequence (5′-GATCCCCTGTTCAA- TACTGCTGATTATTCTCCTTCAAGAGAGGAGAATAATCAGCAGTATTGAACATTTTTA-3′) was cloned into a pSUPER vector using BgIII–HindIII restriction sites. An empty pSUPER vector was used as a control. Cells were transfected with the corresponding pSUPER plasmid by calcium phosphate method, as described elsewhere 42 . Transfected cells were selected by adding puromycin (1.5 μg/mL) to the culture medium. Four days after transfection, cells were collected and cytosolic extracts were obtained. Subcellular fractionation and extract preparation For immunoprecipitation assays, 6 × 10 6 HeLa cells were seeded in 140 mm Petri dishes and treated with an apoptosis inducing agent (20 μM camptothecin, CPT) for 4 h. HeLa cells were fractionated into cytosolic, membrane/organelles and nuclear fractions using a ProteoExtract Subcellular Proteome Extraction Kit (Calbiochem) following the manufacturer’s indications. The purity of subcellular fractions obtained was verified by western blot analysis, using anti-α-tubulin and anti-COX IV for detecting cytosol and mitochondria, respectively. For caspase-3 activity assays, cytosolic extracts were obtained from HEK293T cells as previously described with briefly modifications 43 . Briefly, cells were trypsinized and collected by centrifugation (5 min at 1500×g) and washed twice with phosphate buffer saline (PBS) and once with Cellular Extract Buffer (CEB, 20 mM HEPES at pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM Na-EDTA, 1 mM EGTA, 1 mM dithiothreitol [DTT], 100 μM PMSF). Two volumes of CEB were added to cells, and they were incubated for 15 min on ice, allowing the hypotonic swell of cells. The cells were then disrupted by 20–25 strokes with a Douncer homogenizer and lysates were centrifuged at 15,000×gfor 15 min at 4 °C to remove organelles. Immunoprecipitation Cytosolic fractions from subcellular fractionation of control and CPT-treated HeLa cells were used to immunoprecipitate Cc. Quantity of 300 μg of cytosolic samples were incubated with 50 μL Sepharose 6B (Sigma-Aldrich) for 2 h at 4 °C under agitation as a pre-clearing step to reduce non-specific binding. Samples were centrifuged and 20 μL of rabbit anti-human cytochrome c(Cc) were added to supernatants and incubated overnight at 4 °C under rotation. As a negative control, 300 μg of cytosolic lysates were incubated with 1 μg of anti-mouse IgG under the same conditions. Afterwards, 50 μL Protein A Sepharose (GE Healthcare) was added to each lysate for 4 h at 4 °C under rotation. The protein-Sepharose complexes were washed extensively, collected by centrifugation and boiled in freshly prepared reducing loading buffer. Controls including 30 μg of cytosolic lysates were run concurrently with the immunoprecipitated (IP) samples in the western blot assays. Apaf-1 immunoprecipitation was performed from total non-treated and PMA-treated lysates. Apaf-1 IP samples were analyzed by western blot using a Phos-tag TM SDS PAGE (Wako) —which reduce the electrophoretic mobility of phosphorylated proteins. Antibodies Rabbit anti-human Ccserum was obtained after immunizing male rabbits with full length recombinant Cc suspended in a 0.85% solution of NaCl (20 mg/mL). The suspension was then incorporated into an equal amount of complete (Freund) adjuvant obtained from Difco (BD Biosciences, US). Mouse monoclonal anti-α-Tubulin (catalogue number T8328), as well as secondary horseradish peroxidase (HRP)-conjugated anti-mouse IgG (catalogue number A4416), rabbit polyclonal anti-14-3-3ε (catalogue number SAB4503100), and anti-rabbit IgG (catalogue number A0545) were obtained from Sigma- Aldrich. Rabbit polyclonal antibody to cytochrome c oxidase subunit IV (COX IV) was from Abcam (catalogue number ab16056). Mouse monoclonal anti-Apaf-1 (catalogue number sc-135836) was obtained from Santa Cruz Biotechnology, INC. Mouse monoclonal antiphosphoserine (Millipore clone A4) was obtained from Merck. Western blot analysis For the immunoblot detection of Cc,α-Tubulin, COX IV, and 14-3-3εin the subcellular fractions or IP samples, protein content was measured using the DC TM protein assay (Bio-Rad Laboratories, US). Proteins were resolved by sodium dodecyl sulphate-polyacrylamide gel Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 8 of 12 Official journal of the Cell Death Differentiation Association electrophoresis (SDS-PAGE) in 12% gels and then transferred onto polyvinylidene fluoride (PVDF) membranes (EMD Millipore) using a Mini Trans-Blot electrophoretic transfer cell (Bio-Rad). Membranes were blocked in 5% non-fat dry milk in PBS with Tween-20 (TPBS) and immunoblotting was performed with primary antibodies. HRP-conjugated secondary antibodies were used for detection. The immunoreactive bands were detected using Amersham ECL Plus Western Blotting Detection Reagents (GE Healthcare Life Sciences). Cloning, expression, and purification of recombinant proteins Ccwas expressed in Escherichia coli BL21 (DE3) strain as previously described 32 . First, 25 mL Luria-Bertani (LB) medium, supplemented with 100 μg/mL ampicillin, were grown overnight at 37 °C under agitation. Volume of 2.5 mL of pre-culture was used to inoculate 2.5 L of LB medium. The culture was incubated at 30 °C under agitation for 24 h. Then, cells were centrifuged at 6000 r.p.m. for 10 min in an Avanti J-25 centrifuge (Beckman Coulter). Following this, cells were re-suspended in 1.5 mM borate buffer (pH 8.5), sonicated for 4 min and, finally, cellular debris was eliminated by centrifugation (20,000 r. p.m. for 30 min). For NMR measurements, 15 N-labeled Cc was produced in minimal media with 15 NH 4 Cl as a unique nitrogen source. Ccpurification was carried out by ionic chromatography with a carboxy-methylcellulose matrix. Fractions containing Ccwere concentrated in an Amicon (3 kDa cutoff) until correct concentration was achieved. 14-3-3εfull length (14-3-3εFL) was cloned into the pET-28a vector under the T7 promotor using NdeI– XhoI restriction sites and containing a N-terminal hexahistidine tag. For cloning, the following primers were used: 5′-AGCCATATGATGGATGATCGAGAGGAT-3′ and 5′-GTGCTCGAGTCATTCCTGATTTTCGTC-3′. The 14-3-3εcore was cloned with a Stop codon introduced after residue at position 233. The primers used for this mutagenesis were 5′-ACACTATGGACTTCATA- GATGCAGGGTGACGGT-3′and 5′- ACCGTCACCCTGCATCTATGAAGTCCATAGTGT- 3′. 14-3-3εFL D21K, K50E, S59E, E92K, D99K, and S187D mutants were obtained from Mutagenex Inc. (USA). The expression of 14-3-3εspecies were performed in the E. coli BL21 (DE3) strain. Cultures were grown in LB medium at 37 °C until an O.D. 600 nm of 0.6–0.8 was reached. Protein expression was induced by adding 1 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) and cultures were grown for 24 h at 30 °C. The cells were collected by centrifugation and re-suspended in 20 mM Tris- HCl (pH 8), 800 mM NaCl, 10 mM imidazole. Cells were then broken by sonication and insoluble debris was removed by centrifugation, as indicated above. 14-3-3ε species were purified by affinity chromatography, using a Ni Sepharose 6 Fast Flow (GE Healthcare) column and an imidazole gradient ranging from 10 mM to 300 mM imidazole. Protein containing fractions were concentrated in an Amicon (3 kDa cutoff) until the desired proper protein concentration was reached. Proteins were dialyzed against 10 mM sodium phosphate (pH 7.4) buffer for ITC measurements or 5 mM sodium phosphate (pH 6.3) buffer for NMR titrations. For caspase assays, protein was dialyzed against 20 mM HEPES (pH 7.5) buffer. 14-3-3εC-terminal peptide (14-3-3ε 234-255 ) was purchased from Genosphere Biotechnologies (France). Apaf-1 ΔWD40 S268D and S268A mutants were obtained by mutagenic PCR from pET28a-Apaf-1 ΔWD40 , using 5′- ACCAGAGACAAGGATGTTACAGATT-3′and 5′- AATCTGTAACATCCTTGTCTCTGGT-3′primers for the S268D mutant and 5′-ACCAGAGA- CAAGGCTGTTACAGATT-3′and 5′-AATCTGTAA- CAGCCTTGTCTCTGGT-3′primers for the S268A mutant. All Apaf-1 ΔWD40 species were expressed as previously described with small modifications 6 . Protein expression was carried out in E. coli BL21 (DE3) at 20 °C after induction with 1 mM IPTG. Cells were broken by freeze/heat cycles and cellular debris was removed by centrifugation. The soluble fraction was purified by affinity chromatography (His·Trap column, GE Healthcare) followed by anionic-exchange (Hi·Trap column, GE Healthcare). Recombinant caspase-9 was expressed in E. coli BL21 (DE3) at 30 °C and purified as indicated above for Apaf-1 ΔWD40 . Protein quantification was assessed using the Bradford protein assay 44 . Caspase-9 assays Caspase-9 activity was measured as previously described 6 . Recombinant Apaf-1 ΔWD40 species and 14-3-3εFL were mixed in buffer (40 mM HEPES, pH 7.5, 20 mM KCl, 1 mM DTT) to a final concentration of 100 nM and 1 μM, respectively. Ccwas added at different concentrations (ranging from 1 to 5 μM). After incubation for 15 min at room temperature, recombinant caspase-9 was added to a final volume of 100 μL and final concentration of 500 nM. After incubation for another 15 min at 37 °C, the caspase- 9 substrate LEHD peptide coupled to 7-amino-4- trifluoromethylcoumarin (Ac-LEHD-AFC, TOCRIS) was added at a final concentration of 50 μM. The increase in fluorescence resulting from Ac-LEHD-AFC cleavage was determined in a Cary Eclipse (Varian) fluorescence spectrophotometer, using an excitation wavelength of 390 nm and an emission wavelength of 510 nm. Each experimental data were the mean ± SD value of at least three independent measurements. Elena-Real et al. Cell Death and Disease (2018) 9:365 Page 9 of 12 Official journal of the Cell Death Differentiation Association