Coccia et al. Cell Death and Disease (2020)11:82 https://doi.org/10.1038/s41419-020-2282-x Cell Death & Disease ARTICLE Open Access SIVA-1 regulates apoptosis and synaptic function by modulating XIAP interaction with the death receptor antagonist FAIM-L Elena Coccia 1,2,3, Laura Planells-Ferrer1,2,3, Raquel Badillos-Rodríguez1,2,3, Marta Pascual2,4,5, Miguel F. Segura 6, Rita Fernández-Hernández7, Joaquin López-Soriano1,2,3, Eloi Garí7, Eduardo Soriano2,4,5,8, Bruna Barneda-Zahonero1,2,3, Rana S. Moubarak 1,2,3,9, M. Jose Pérez-García 1,2,3 and Joan X. Comella 1,2,3 Abstract The long isoform of Fas apoptosis inhibitory molecule (FAIM-L) is a neuron-specific death receptor antagonist that modulates apoptotic cell death and mechanisms of neuronal plasticity. FAIM-L exerts its antiapoptotic action by binding to X-linked inhibitor of apoptosis protein (XIAP), an inhibitor of caspases, which are the main effectors of apoptosis. XIAP levels are regulated by the ubiquitin-proteasome pathway. FAIM-L interaction with XIAP prevents the ubiquitination and degradation of the latter, thereby allowing it to inhibit caspase activation. This interaction also modulates non-apoptotic functions of caspases, such as the endocytosis of AMPA receptor (AMPAR) in hippocampal long-term depression (LTD). The molecular mechanism of action exerted by FAIM-L is unclear since the consensus binding motifs are still unknown. Here, we performed a two-hybrid screening to discover novel FAIM-L-interacting 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; proteins. We found a functional interaction of SIVA-1 with FAIM-L. SIVA-1 is a proapoptotic protein that has the capacity to interact with XIAP. We describe how SIVA-1 regulates FAIM-L function by disrupting the interaction of FAIM-L with XIAP, thereby promoting XIAP ubiquitination, caspase-3 activation and neuronal death. Furthermore, we report that SIVA-1 plays a role in receptor internalization in synapses. SIVA-1 is upregulated upon chemical LTD induction, and it modulates AMPAR internalization via non-apoptotic activation of caspases. In summary, our findings uncover SIVA-1 as new functional partner of FAIM-L and demonstrate its role as a regulator of caspase activity in synaptic function. Introduction neuronal pruning, axon guidance, and synaptic plasticity3. Apoptosis plays a crucial role during neural develop- The activation of caspase-3 via mitochondria is required ment and adult life1,2. The main orchestrators of apoptosis for AMPA receptor (AMPAR) internalization in NMDA are caspases, effector proteases that mediate the proteo- receptor-dependent long-term depression (LTD). Several lytic cascade that ultimately leads to cell death. Several studies have shown that LTD is abolished in caspase-3 non-apoptotic functions of caspases have been reported in knock-out mice4, and caspase-3-deficient neurons in cul- neurons2. Localized activation of caspases has been ture fail to show spine shrinkage in response to NMDA described as necessary in the physiological context of stimulation (chemical LTD)5. FAIM-L is the neuron-specific isoform of FAIM, and it harbors 22 additional amino acids at the N-terminal as Correspondence: M. Jose Pérez-García (maria.perez@vhir.org) or compared to FAIM-S. We have previously shown that Joan X. Comella (joan.comella@vhir.org) 1 Cell Signaling and Apoptosis Group, Vall d’Hebron Research Institute (VHIR), FAIM-L regulates caspase activation in neurons. Moreover, 08035 Barcelona, Spain 2 decreased FAIM-L levels have been associated with the Centro de Investigación Biomédica en Red sobre Enfermedades progression of Alzheimer’s disease (AD), an irreversible and Neurodegenerativas (CIBERNED), ISCIII, 28031 Madrid, Spain Full list of author information is available at the end of the article. progressive neurodegenerative disorder that is the most Edited by B. Joseph © The Author(s) 2020 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 link to the Creative Commons 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/. Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 2 of 19 common cause of dementia6. AD is characterized by neu- Yeast two-hybrid screen ronal death and the dysregulation of synaptic plasticity7, The pACT2-mouse cDNA library in Escherichia coli mechanisms in which FAIM-L and apoptotic pathways are (Clontech cat# ML4008AH/cat# 638841) was pre- involved and have been shown to play crucial roles. transformed in the yeast AH109 strain (more than 107 FAIM-L modulates death receptor-induced cell death by independent clones). The full-length FAIM-L and the 22 interacting with death receptors8 and by interacting and additional amino acids at the N-terminal FAIM isoform stabilizing the X-linked inhibitor of apoptosis protein (FAIM-L) bait proteins were subcloned into pGBKT7 (XIAP). XIAP is a key member of the inhibitor of apoptosis vector and transformed in Y187 yeast strain. The two- protein family that directly binds to and inhibits effector hybrid selection was performed by mating, following the caspases. XIAP levels are regulated by a process of auto- matchmaker two-hybrid system 3 protocol (cat# K1612-1 ubiquitination and proteasome degradation. Given its anti- Clontech). Positive colonies were selected in drop out apoptotic role, modification of XIAP levels is a decisive step medium lacking leucine, tryptophan, and histidine and in cell death induction9. FAIM-L confers neurons additional containing 20 mM aminotriazole. Colonies were analyzed protection against cell death. In fact, FAIM-L interacts with by polymerase chain reaction (PCR). cDNA was XIAP and inhibits its ubiquitination and degradation by the sequenced and transformed in E. coli (WB Cat# OP50), proteasome, therefore maintaining the role of XIAP in and interactions of bait and prey were confirmed by back preventing the cleavage and activation of caspases10. FAIM- transformation in yeast. L also participates in processes where caspases play a non- apoptotic role. By stabilizing XIAP levels, FAIM-L prevents Cell culture AMPAR internalization after chemical induction of LTD HEK293T cells (ATCC Cat# CRL-3216) were grown (chLTD) in vitro and protects axons from degeneration in DMEM supplemented with 10% heat-inactivated induced by growth-factor deprivation11. fetal bovine serum (iFBS) (Invitrogen), 20 U/ml peni- Structural prediction algorithms for FAIM-L do not cillin and 20 µg/ml streptomycin. Rat pheochromocy- identify any functional domain that could predict potential toma PC12 cells (ATCC Cat# CRL-1721) were grown in protein association, and apart from XIAP, no other FAIM- DMEM supplemented with 6% iFBS, 6% heat inacti- L-interacting protein has been described to date. Here, vated horse serum (iHS), 10 mM HEPES, 20 U/ml using a yeast two-hybrid screening approach, we screened penicillin and 20 µg/ml streptomycin. Cultures were for potential FAIM-L partners of relevance for its function. maintained at 37 °C in a 5% CO2 atmosphere in a Using either the full-length or the specific N-terminal humidified incubator. amino acid sequence of FAIM-L as bait, we identified the proapoptotic SIVA-1 protein as a FAIM-L binding partner. Primary neuron cultures SIVA-1 has been described as a proapoptotic protein, Neuron cultures were prepared from wild-type C57BL/ capable of inducing cell death in several cellular models12. 6J mice (Envigo, France) at embryonic day 15–16 Moreover, SIVA-1 has been reported to interact with (E15–16). Cerebral cortices and hippocampi were dis- XIAP13. Apoptotic induction mediated by SIVA-1 has been sected in phosphate-buffered saline (PBS) pH 7.4. After proposed to occur through several pathways, and it is trypsin and DNase treatment, tissues were mechanically essential for p53-induced apoptosis in granular neurons14. dissociated and filtered through a 40-µm nylon mesh. Here, we studied the contribution of SIVA-1 to the roles of Cells were resuspended in DMEM supplemented with 5% FAIM-L and XIAP in caspase-dependent apoptotic and iFBS, 5% iHBS, 20 U/ml penicillin and 20 µg/ml strepto- non-apoptotic functions in neurons. We found that SIVA-1 mycin. Cells were then plated in poly-D-lysine-coated blocks the antiapoptotic function of FAIM-L by displacing plates at a density of 3 × 105 cells/ml or on coverslips at the XIAP/FAIM-L interaction, thereby inducing XIAP 1.5 × 105 cells/ml for immunocytochemistry experiments. degradation, caspase-3 activation and subsequent neuronal Four hour after seeding, medium was replaced by Neu- death or AMPAR internalization. Our results point to robasal medium supplemented with B27, glutaMAX (Life SIVA-1 as one of the proteins of the apoptotic machinery Tech), 20 U/ml penicillin and 20 µg/ml streptomycin. able to modulate caspase-3 activation in response to Culture medium was partially replaced every 3–4 days synaptic plasticity. On the basis of our findings, SIVA-1 with fresh medium. Cultures were kept at 37 °C in a 5% emerges as a novel synaptic modulator. CO2 atmosphere in a humidified incubator. When pan- caspase inhibitor quinolyl-Val-Asp-OPh (Q-VD) treat- Material and methods ment was performed, Q-VD was added directly to culture Reagents media at a final concentration of 10 µM. All experimental Unless otherwise specified, all biochemical reagents protocols were approved by the Vall d’Hebron Institu- were purchased from Sigma-Aldrich. tional Review Board. Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 3 of 19 Plasmids buffer (60 mM Tris/HCL 1 M, pH 6.8, 2% sodium dodecyl The constructions used for this study, namely 3xHA- sulfate (SDS), 10% Glycerol, 0.01% Bromophenol Blue) SIVA-1, 3xHA-ΔSIVA-1, 3xFLAG-FAIM-L, 3xHA- without dithiothreitol and boiling samples at 95 °C for FAIM-L, 6xMyc-XIAP and YFP, were expressed under 5 min. Whole-cell lysate and immunoprecipitated pro- the control of a cytomegalovirus constitutive promoter in teins were then analyzed by western blot. the pcDNA3 expression vector (Invitrogen). 3xFLAG- SIVA-1 plasmid was kindly provided by Dr. Ulrike Resch Immunohistochemistry (Medical University of Vienna). Lentiviral plasmids for E16, P0, P5, P15, and adult OF-1 mice (Charles River, this study were cloned into pEIGW. Short hairpin RNA Lyon; France) were used for immunohistochemistry (shRNA) targeting SIVA-1 was cloned into the pLVTHM experiments. The day the vaginal plug was detected was vector, and a scrambled sequence was used as a control. considered embryonic day 0 (E0), and the day of birth as postnatal day 0 (P0). First, the animals were deeply Lentiviral production anesthetized with a Ketolar (Parke-Davies/Pfizer New Lentiviruses were produced as described previously by York, NY; USA)/Rompun (Bayer AG, Leverkusen; Ger- Segura et al.8. For infection, lentiviruses were added to the many) mixture and perfused with 4% paraformaldehyde host cell medium. Infection efficiency was monitored by in 0.1 M phosphate buffer. Brains were dissected, cryo- counting green fluorescent protein (GFP)-positive cells. protected and frozen. Next, 30–50 μm coronal sections were obtained. After blocking, sections were incubated Cell transfection and infection overnight with specific rabbit antibody against SIVA-1 HEK293T (ATCC Cat# CRL-3216) or PC12 (ATCC (dilution 1:100, Santa Cruz Biotechnology Cat# sc- Cat# CRL-1721) cells were transfected with the desired 48768) or mouse anti-SIVA-1 (dilution 1:100, Sigma- expression plasmid using the calcium phosphate method Aldrich Cat#SAB1400393). These primary antibodies or Lipofectamine 2000 (Invitrogen), following the manu- were visualized by sequential incubation with biotiny- facturer’s instructions. The total amount of transfected lated secondary antibodies (dilution 1:200, Vector Labs DNA was kept constant by adding empty pcDNA3 Burlingame, CA; USA) and the streptavidin–peroxidase expression vector. Primary neurons were transfected with complex (dilution 1:400, Amersham Biosciences Pitts- Lipofectamine 2000, as described in Dalby et al.15. burgh, PA, USA). The peroxidase reaction was devel- oped with diaminobenzidine (DAB) and H2O2. The Immunoprecipitation sections were mounted onto gelatinized slides, and then After 24–48 h of transient transfection for ectopic dehydrated and cover-slipped with Eukkit (Panreac). expression, or after 24 h in culture, HEK293T and PC12 Sections from E16, P5, P15, and adult mice were used for cells were rinsed in PBS 1× and lysed in immunopreci- double immunofluorescence studies. Sections were incu- pitation lysis buffer (IP lysis buffer) containing 20 mM bated overnight with the specific antibody detecting SIVA-1 Tris/HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 10% (dilution 1:100, Santa Cruz Biotechnology, Cat# sc-48768) Glycerol, 1% Triton X-100, and supplemented with a combined with antibodies against neuron-specific nuclear protease inhibitor cocktail (Roche). Samples were lysed protein (NeuN, dilution 1:100, Millipore Cat# MAB377), for 30 min on ice and centrifuged at 4 °C at 12,000 × g. PV (dilution 1:3000, Swant Antibodies Cat# PVG-214, One milligram of total protein was used for immuno- RRID:AB_2313848) or GFAP (dilution 1:500, Millipore precipitation overnight at 4 °C in an orbital shaker. Spe- Cat# AB5804). Primary antibodies were visualized using cific FLAG M2 Affinity monoclonal agarose beads or anti- secondary Alexa Fluor-conjugated antibodies. Sections were ubiquitin-conjugated agarose beads (Santa Cruz Bio- mounted with Prolong diamond mounting medium technology Cat# sc-8017 AC) were used for IP:FLAG and (Thermo Fisher) and viewed under a confocal microscope. IP:Ubiquitin respectively. For IP:Myc, lysates were pre- cleared by 30 min of incubation with 20 µl conjugated Immunocytochemistry protein G suspension and 0.25 µg/µl control mouse IgG. Hippocampal or cortical primary neurons were cul- Afterwards, 1 µg Myc antibody (Santa Cruz Biotechnology tured on glass coverslips pre-coated with 10 µg/ml poly- Cat# sc-40) was added to each sample and incubated for D-lysine. After 7–10 days in vitro (DIV), cells were fixed 4 h at 4 °C in an orbital shaker. Finally, 20 µl conjugated with 4% paraformaldehyde for 30 min at room tem- protein G suspension was added and incubated overnight perature. Permeabilization and blocking were carried on an orbital shaker. After incubation, agarose beads were out with a permeabilization solution (PS) containing 5% washed five times with IP lysis buffer and eluted. Elution FBS, 5% bovine serum albumin, and 0.1% Triton-X for was performed with a competitor peptide for IP:FLAG 1 h at room temperature, followed by incubation with and IP:Myc, following the manufacturer’s protocol. For IP: primary antibodies in PS overnight at 4 °C. The follow- Ubiquitin, elution was performed by adding Laemmli ing primary antibodies were used: anti-SIVA-1 (dilution Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 4 of 19 1:500 Santa Cruz Biotechnology Cat# sc-48768; Sigma- at 3000 × g for 10 min to pellet the plasma membrane Aldrich Cat# SAB1400393); anti-SIVA-1 (dilution 1:25, (PM) and collect cytosolic fractions (supernatant). Nuclei Sigma-Aldrich Cat# SAB3500697), anti-XIAP (dilution were extracted from the 600 × g pellet with centrifuga- 1:25, BD Cat# 610762), anti-HA (dilution 1:100, Roche tions at 500 × g for 15 min. The supernatants were cen- Cat# 1-867-423); anti-Rab5 (dilution 1:500, Cell Sig- trifuged twice at each speed, and pellets were washed naling Technology Cat# 2143); anti-Calnexin (dilution twice by resuspension in homogenization buffer and 1:500, Cell Signaling Technology Cat# 2433S); anti-βIII recentrifugation. Pellets were lysed in SET buffer (10 mM Tubulin (dilution 1:1000, Covance Research Products Tris-HCl pH7.4, 1 mM EDTA, 150 mM NaCl, 1% SDS). Inc Cat# SIG-3840-1000); anti-TrkA (dilution 1:1000, in Protein concentration was quantified by a modified Lowry house); anti-PSD95 (dilution 1:200, Cell Signaling Cat# assay (DC protein assay; Bio-Rad, Hercules, CA). Samples 3450); anti-Synapsin II (dilution 1:200, Enzo Life Sci- were heat-denatured in Laemmli buffer and subjected to ence Cat# ADI-VAS-SV061-E); and anti-GFP (1:2000, SDS polyacrylamide gel electrophoresis. Abcam Cat# ab6556). Subsequently, cells were rinsed with PBS and incubated with fluorescent-conjugated Western blot secondary antibodies (dilution 1:400–1:1000, Alexa- Cells were harvested and lysed in SET buffer. Protein Fluor Thermo Fisher Scientific Cat# A-11001 Cat# A- concentration was quantified by a modified Lowry assay 11912) diluted in PS for 1 h at room temperature. Prior (DC protein assay; Bio-Rad, Hercules, CA, USA). Lysates to the last wash, cells were incubated for 30 min at room were prepared with Laemmli buffer (60 mM Tris/HCl temperature with Hoechst 33258 (0.05 µg/ml) to stain 1 M pH 6.8, 2% SDS, 100 mM glycerol, 0.01% Bromo- nuclear DNA. MitoTracker staining (Invitrogen) was phenol Blue), resolved in SDS polycramide gels, and used, following the manufacturer’s instructions. Cover- transferred onto polyvinylidene fluoride Immobilon-P slips were finally mounted on slides with Prolong dia- membranes (Merck Millipore). After blocking with TBS mond mounting medium (Thermo Fisher). 1× −0.1% Tween-20 containing 5% nonfat dry milk for 1 h at room temperature, membranes were probed with Cell death quantification the appropriate primary antibodies, prior to incubation At 4 days post infection, primary neurons were fixed for 1 h with the appropriate specific peroxidase- with 2% paraformaldehyde, permeabilized with 0.1% Tri- conjugated secondary antibody. Membranes were devel- ton™ X-100 and stained with Hoechst 33342 (0.05 µg/ml). oped using the EZ-ECL chemiluminescence detection kit Apoptosis was assessed by counting viable and dead cells, (Biological Industries, Kibbutz Beit Haemek, Israel). The discriminating condensed and fragmented nuclei (apop- following primary antibodies were used: anti-FAIM-L totic nuclear morphology type II)16. For each experiment (dilution 1:2000, in house); anti-caspase-3 (dilution quantification was performed in blind testing, and at least 1:1000, Cell Signaling Technologies Cat# 9662); anti-pan 500 cells were counted per condition. ERK (dilution 1:10000, BD Bioscience Cat# 610123); anti- β-actin (dilution 1:2000, Santa Cruz Biotechnology Cat# Caspase activity assay sc-47778); anti-alpha-tubulin (dilution 1:50000, Sigma- The caspase activity assay was performed as previously Aldrich Cat# T5168); anti-XIAP (dilution 1:5000, BD described by Galenkamp et al.17. Briefly, primary neurons Bioscience Cat# 610762); anti-FLAG (dilution 1:10000, were harvested and lysed in caspase activity buffer (20 mM Sigma-Aldrich Cat# F3165); anti-HA (dilution 1:5000, HEPES-NaOH, pH7.2, 10% sucrose, 150 mM NaCl, 5 mM Roche Cat# 11867423001); anti-SIVA-1 (dilution 1:500, EDTA, 1% Igepal CA-630, 0.1% CHAPS, and 1× EDTA-free Sigma Cat# SAB3500697); anti-GluA2 (1:1000, Millipore complete protease inhibitor cocktail). Next, 20 µg protein Cat# MAB397); anti-cytochromec (1:1000, BDbioscience was incubated at 37 °C in caspase activity buffer supple- Cat# 556433); anti-calnexin (1:1000, Abcam Cat# mented with 10 mM DTT and 50 µM of the fluorogenic ab31290); anti-histone-3 (1:1000, Cell Signaling Tech- substrate Ac-DEVD-Afc (Merck Millipore). Plates were nologies Cat# 9715); and anti-GAPDH (1:2000, Abcan read in a fluorometer using excitation and emission wave- Cat# ab9485). lengths of 405 and 535 nm, respectively. Chemical LTD induction Subcellular protein fractionation Primary neurons were treated at 12–14 DIV with 50 µM Subcellular fractionation was performed as previously N-methyl-D-aspartate (NMDA) for 15 min at 37 °C to described18. Adult mouse brains were homogenized in induce chLTD as described in Li et al.4. Fifty micromolar buffer containing 10 mm HEPES, pH 7.4, 2 mm EDTA, of BAPTA-AM was used as calcium chelator and 1 µg/ml 0.32 M sucrose, and protease inhibitor cocktail (Roche). of cycloheximide (CHX) as an inhibitor of protein trans- Cell homogenates were centrifuged sequentially at 600 × g lation. Cells were pretreated with BAPTA-AM or CHX for 10 min to remove nuclei (N) and unbroken cells, then during 30 or 60 min, respectively. Treatments were added Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 5 of 19 to media for the pretreatment and maintained during For GluA2 internalization and surface staining, a z-stack chLTD induction. of images was obtained through various filter channels (Alexa-488, Alexa-568 and Alexa-674). Typically, 25 serial GluA2 internalization assay and surface staining 2D images were recorded at 45-nm intervals. Image Internalization assays were performed as described in acquisition settings were identical in each experiment. FIJI Martinez-Marmol et al.11, with minor modifications. software was used to make 2D projections from the z- Hippocampal neurons at 12–14 DIV were incubated with stack of images. We measured the total integrated antibodies against the N-terminus of GluA2 (2 µg/ml, intensity of internalized GluA2 and surface-remaining mouse monoclonal, clone 6C4, Millipore Cat# MAB397) GluA2 in the same region of infected (GFP/HA-positive) for 60 min at 20 °C. They were then either stimulated with neurons. For each experiment at least 15–20 cells were medium containing NMDA (50 µM) for 15 min at 37 °C counted per condition. or left unstimulated. Subsequently, they were fixed for 5 min at room temperature in paraformaldehyde:sucrose Cell surface biotinylation in PBS. Surface-remaining antibody-labeled receptors Cell surface proteins were biotinylated and isolated using were visualized by means of a 1 h incubation with satu- the Pierce® Cell Surface Protein Isolation Kit (Thermo rated AlexaFluor-647 secondary antibody (15 µg/ml, Fisher Scientific), following the manufacturer’s instructions. Abcam Cat# ab150115). Neurons were then permeabi- Before immunoprecipitation, the amount and concentra- lized for 2 min with methanol (−20 °C), and internalized tion of protein were equalized by Lowry quantification. antibody-labeled GluA2 was detected by a 1 h incubation with Alexa 568-conjugated secondary antibody (1 µg/ml; RNA extraction and quantitative reverse transcription PCR Molecular Probes Cat# A-11004). Simultaneously, infec- Total RNA was isolated from cells using the RNeasy ted GFP-positive or HA-positive neurons were stained Mini Kit (Qiagen), following the manufacturer’s protocol. with antibodies against GFP (5 µg/ml, Abcam; Cat# cDNA was reverse-transcribed using the High-Capacity ab6556) or HA (10 µg/ml Roche Cat# 11867423001), cDNA Reverse Transcription Kit (Thermo Fisher). SIVA- respectively. Thus, the GFP fluorescence of infected 1 mRNA expression was measured with SYBR green neurons was enhanced by a 30 min incubation with Alexa (Applied Biosystems) and normalized against L27 using 488-conjugated secondary antibody (4 µg/ml; Thermo the following primers: SIVA-1 forward: 5′-GATCACAT Fisher Scientific Cat# A27034). After three washes in PBS ATCGAGCGAAGA-3′, reverse: 5′-GCCTTCCCATCCA 1×, the coverslips were mounted on slides with Prolong CAGATCT-3′; and L27: forward: 5′-AGCTGTCATCG diamond mounting medium (Thermo Fisher). TGAAGAA-3′, reverse: 5′-CTTGGCGATCTTCTTCTT GCC-3′. Analysis was performed using the 7900HT Image acquisition and analysis Sequence Detection System 2.3 Software (Applied Bio- Images were acquired with a confocal laser scanning systems). Relative expression fold change was determined microscope (spectral FV1000; Olympus). The digital by the comparative 2(−ΔΔCT) method21. images were processed using the FIJI software (Fiji)19. Protein co-localization in the dual-color confocal ima- Experimental design and statistical analysis ges was measured quantitatively using the JACoP Plug-in All results were confirmed in at least three independent (http://fiji.sc/wiki/index.php/Colocalization_Analysis). experiments. Representative images and western blots are Co-localization was determined by Pearson’s and Man- shown in figures, while measurements of all repeated ders co-localization coefficients. Briefly, Pearson’s coeffi- experiments are measured in graphs. Data sets were cient values describe the relationship between two image plotted and analyzed using GraphPad Prism v5 (GraphPad signals calculated by linear regression. Values can range Software; La Jolla, CA, USA). Values were excluded using from 1 to −1, with 1 standing for complete positive cor- the analysis to identify outliers included in the software. relation and −1 for complete negative correlation. Man- Values are expressed as mean ± SEM. Statistical sig- ders co-localization coefficients (M1 and M2) describe the nificance was set at p < 0.05 and was determined by two- contribution of two selected channels to the pixels of tailed unpaired t test for comparisons between two interest, in this case Channel 1 being SIVA-1 and Channel groups, or by two-way ANOVA for multiple groups. Each 2 the other markers. Each coefficient represents the specific test is indicated in figure legends. fraction of one channel signal that coincides with the signal of the other channel. Manders co-localization Results coefficients values range from 0 (uncorrelated distribu- SIVA-1 interacts with FAIM-L tions of two probes with one another) to 1 (perfect co- To identify the functional partners of FAIM-L, we localization of two images)20. At least 15 cells were con- performed a yeast-two-hybrid screening using as bait the sidered for quantification. full-length sequence of FAIM-L or the 22 N-terminal- Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 6 of 19 Table 1 Yeast two-hybrid analysis. List of protein–protein interactions with full-length FAIM-L and the 22 N-terminal- specific amino acids of the FAIM-L isoform. Yeast two- Bait Sequence hybrid analysis pGBKT7 pGBKT7- full pGBKT7-N-term (control) length FAIM-L FAIM-L -(MASGDD SPIFEDDESPPYSLEK) SIVA-1 (ΔSIVA-1) − + + MPKRSCPFADAAPLQLKVHVGLKELSHGVFAERYSREV FERM RhoGEF − + + HELKKDLIGIDNLVTPGREFIRLGSLSKLSGKGLQQRMFFLFNDVLLYTS RGLTASNQFKVHGQLPLYGMTIEESEEEWGVPHCLTLRGQRQSIIVAA SSRSEMEKWMEDIQMAIDLAEKSNGPTPELLASSPPDNKSPDEA GAPDH − + + YSNRMVDLMAYMASKE COMMD1 − + + EGGKSLSGLLSGLAQNAFHGHSGVTEELLHSQLYPEVPPEEFRPFLA KMRGLLKSIASADMDFNQLEAFLTAQTKKQGGITSEQAAVISKFWKSH KIKIRESLMKQSRWDNGLRGLSWRVDGKSQSRHSTQIHSPVAIIELEFG KNGQESEFLCLEFDEVKVKQILKKLSEVEESINRLMQAA Calcyclin − + + DKFVKIYITLTGVHQVPTENVQVHFTERSFDLLVKNLNGKNYSMIVNN LLKPISVESSSKKVKTDTVIILCRKKAENTRWDYLTQVEKECKEKE AP2m1 − + + PKRACQFNRTQLGDCSGIGDPTHYGYSTGQPCVFIKMNRVINFYAGAN QSMNVTCVGKRDEDAENLGHFVMFPANGSIDLMYFPYYGKKFHVNY TQPLVAVKFLNVTPNVEVNVECRINAANIATDDERDKFAGRVAFKLRINKT specific amino acids of this isoform. Among the top six interactions between SIVA-1, XIAP and tagged FAIM-L hits shown in Table 1, the N-terminal fragment of SIVA-1 takes place in neuronal cytoplasm. was detected (ΔSIVA-1) (Table 1 and Fig. 1a). We validated this interaction in HEK 293T cells by co- SIVA-1 is expressed in neurons immunoprecipitation experiments using full-length FLAG- FAIM-L expression is restricted to neurons8. To under- tagged FAIM-L, HA-tagged N-terminal SIVA-1 (ΔSIVA-1), stand the biological context in which SIVA-1 may modulate or HA-tagged full-length SIVA-1. We confirmed the FAIM-L functions, we first examined SIVA-1 expression in interactions between FAIM-L and both full-length and mouse brain tissue. Histological analysis showed that SIVA- ΔSIVA-1 forms (Fig. 1b, lanes 3 and 6). Interaction with the 1 expression occurred in most brain regions throughout truncated form of SIVA-1 was detected only in the presence development (Fig. 2a–h, stages E16-adult mouse). High of the proteasome inhibitor MG132 (Fig. 1b, lane 6), levels were detected in the telencephalon (cerebral cortex otherwise it was not stable and was consequently degraded. and hippocampus) and in the cerebellum (Fig. 2e–h). In the Our group reported XIAP as a FAIM-L-interacting embryonic cerebral cortex, labeled cells were located in the partner10. XIAP is an inhibitor of caspases and its endo- cortical plate (CP) and in the subplate (SP), corresponding to genous levels are crucial for FAIM-L protection against postmitotic neurons of these layers (Fig. 2a). At postnatal caspase activation and consequent apoptotic or non- stages (P0-P15), SIVA-1-positive cells were found through- apoptotic functions10,11. Since Resch and colleagues out the cortical layers, the strongest staining being detected described an interaction between SIVA-1 and XIAP13, we in layers II–III (Fig. 2b, c). At adult stages, the expression of performed co-immunoprecipitation experiments and cor- SIVA-1 in the cerebral cortex appeared to be restricted to roborated that XIAP does indeed interact with both FAIM- layer V cortical neurons (Fig. 2d). The cerebellum also L and SIVA-1 (Fig. 1c). showed intense SIVA-1-immunolabeling (Fig. 2e–g). SIVA- To detect the interaction in a semi-endogenous condi- 1 was expressed in migratory Purkinje cells in the embryonic tion, we transfected primary cortical neurons with 3xHA- cerebellum (data not shown). At postnatal P10 and FAIM-L. We then performed immunocytochemistry P15 stages, cell bodies and dendrites of Purkinje neurons against HA, endogenous SIVA-1 and XIAP (Fig. 1d). Co- showed intense SIVA-1-immunostaining (Fig. 2e). These localization analysis revealed partial correlation of signals neurons maintained SIVA-1 expression at later stages and of SIVA-1 with both XIAP (Pearson’s correlation value also during adulthood (Fig. 2g). At embryonic stages, cells in R = 0.412 ± 0.115) and HA-FAIM-L (Pearson’s correlation the hippocampal plate expressed SIVA-1 (data not shown). R = 0.607 ± 0.060). We could also observe that the At postnatal stages, pyramidal neurons in CA1–3 and some Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 7 of 19 Fig. 1 SIVA-1 interacts with FAIM-L and XIAP. a Schematic representation of the two-hybrid technique results. The N-terminal of SIVA-1(ΔSIVA-1) interacted with both baits, namely full-length FAIM-L and FAIM-L-specific N-terminal 22 amino acids. b Immunoprecipitation in HEK 293T cells of FLAG-FAIM-L using M2 Affinity FLAG. FAIM-L was immunoprecipitated and membrane was blotted with anti-HA to detect full-length SIVA-1 and ΔSIVA-1. Ten micromolar of MG132 (proteasome inhibitor) was used to impair the degradation of ΔSIVA-1. c Immunoprecipitation of Myc-XIAP using anti-Myc antibody. Cells were transiently transfected with Myc-XIAP, FLAG-FAIM-L, and FLAG-SIVA-1. The membrane was blotted with anti-FLAG to detect both FAIM-L and SIVA-1. XIAP was used as a control of the immunoprecipitation and actin as loading control. d Representative confocal images of immunocytochemistry in cortical neurons. Neurons were transfected with HA-tagged FAIM-L at 4 days in vitro. Forty-eight hour after transfection, immunofluorescence was performed staining with anti-SIVA-1 (green), anti-HA (to detect HA-FAIM-L, red), and anti-XIAP (blue). Scale bar 10 µm. Graph reports Pearson’s correlation coefficients obtained in co-localization analysis (SIVA-1/HA-FAIM-L R = 0.607 ± 0.060; SIVA-1/XIAP R = 0.412 ± 0.115). Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 8 of 19 Fig. 2 (See legend on next page.) Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 9 of 19 (see figure on previous page) Fig. 2 SIVA-1 is expressed in neurons during the development of the mouse brain. a At embryonic day 16 (E16), neurons in the cortical plate (CP) and subplate (SP) contained SIVA-1. Scale bar 100 μm. b At postnatal day 5 (P5), SIVA-1-immunopositive cells were located throughout all cortical layers. Scale bar 200 μm. c At postnatal day 15 (P15), SIVA-1-immunoreactive cells were present in all cortical layers. Note that cells located in layers II–III showed intense SIVA-1-immunolabeling. Scale bar 200 μm. d In the cerebral cortex of adult mice, expression of SIVA-1 was restricted to cortical cells in layer V. Scale bar 200 μm. e–g In the cerebellum, the soma and dendrites of Purkinje cells showed SIVA-1 immunostaining at all postnatal stages analyzed. Scale bar 200 μm. h In the hippocampus, pyramidal neurons and hippocampal interneurons expressed SIVA-1. Some hippocampal interneurons scattered in all hippocampal layers showed a strong immunostain (arrows in (h)). Scale bar 200 μm. i–k Double immunofluorescence of SIVA-1 and NeuN (arrows) in cortical layer V at P5, showing an almost complete co-localization of the two markers. Scale bar 25 μm. l–n Immunofluorescence co-localization of SIVA-1 and GFAP in cerebellum at P15 shows SIVA-1 expression in Purkinje cells (asterisks) but not in astrocytes (arrowheads) in the cerebellum. Scale bar 25 μm. o–q Immunofluorescence co-localization of SIVA-1 and PV shows expression of SIVA-1 in hippocampal interneurons (arrows). MZ marginal zone, IZ intermediate zone, SVZ subventricular zone, VZ ventricular zone, so stratum orines sp stratum pyramidale, sr stratum radiatum, ML molecular layer, PCL Purkinje cell layer, GL granule cell layer. interneurons throughout the hippocampal layers showed Taken together, our data indicate that SIVA-1 is SIVA-1-immunostaining (Fig. 2h, o–q). To confirm the expressed in the cytoplasm of neurons during develop- localization of SIVA-1 in neurons, P5 and P15 sections were ment and adult stages. Co-expression of FAIM-L and double immuno-labeled with SIVA-1 and the neuron- SIVA-1 in adult neurons in basal conditions suggests specific marker NeuN. This approach revealed an almost that the latter participates in physiological nonlethal complete co-localization of both markers at all the devel- processes. opmental stages analyzed (Fig. 2i–k). Double immuno- fluorescent studies with the astroglial cell marker Glial SIVA-1 induces XIAP ubiquitination fibrillary acidic protein (GFAP) did not show co-localization XIAP is a potent inhibitor of effector caspases. Its ubi- with SIVA-1 (Fig. 2l–n). Finally, by double immunodetec- quitination and degradation by the proteasome is crucial tion of SIVA-1 with a specific interneuron subtype marker, for caspase activation and induction of apoptosis9. FAIM- parvalbumin (PV), we confirmed the presence of SIVA-1 in L is a XIAP-interacting protein that executes its anti- hippocampal neurons (Fig. 2o–q). We concluded that SIVA- apoptotic function by inhibiting XIAP ubiquitination, 1 protein is predominantly expressed in developing and hence stabilizing its levels and inhibition on caspases. adult neurons. Having reported SIVA-1 interaction with both proteins To determine the subcellular localization of SIVA-1, we we proceeded to address whether SIVA-1 interferes with performed immunocytochemistry in primary cultures of the FAIM-L/XIAP interaction and XIAP ubiquitination. hippocampal neurons. We analyzed SIVA-1 distribution in To this end, we analyzed the FAIM-L/XIAP interaction in the cell body, performing co-localization analysis with sev- the presence of increasing amounts of overexpressed SIVA- eral organelle markers (i.e., MitoTracker for mitochondria, 1 in PC12 cells (Fig. 4a, lanes 4–6). We performed an Rab5 for early endosome, Calnexin for endoplasmic reticu- immunoprecipitation of Myc-XIAP and found the amount lum; Fig. 3a). Manders co-localization coefficients were of co-immunoprecipitated FAIM-L to be inversely corre- obtained for each marker (Fig. 3b) and only values higher lated to SIVA-1 overexpression. Our results thus suggest than 0.5 were considered to indicate co-localization. SIVA-1 that SIVA-1 reduces the interaction of FAIM-L with XIAP. showed a mainly cytosolic distribution and co-localized with Figure 4b shows ubiquitin immunoprecipitation in markers of the cytoplasmic membrane (TrkA; M1 = 0.68 ± HEK293T cells overexpressing HA-tagged SIVA-1, 0.02), endocytic vesicles (Rab5; M1 = 0.66 ± 0.15), and FAIM-L, or a combination of both proteins. XIAP pre- cytoskeleton (βIII-tubulin; M1 0.71 ± 0.14). SIVA-1 did not sented basal endogenous ubiquitination, which was localize to mitochondria (MitoTracker; M1 = 0.41 ± 0.04) or enhanced in the presence of SIVA-1 overexpression. to the endoplasmic reticulum (Calnexin; M1 = 0.22 ± 0.15). FAIM-L overexpression on the other hand reduced XIAP SIVA-1 expression was clearly excluded from the nucleus ubiquitination levels, thereby confirming our previous (Hoechst staining; M1 = 0.06 ± 0.03) (Fig. 3a, b). A sub- findings10. Moreover, overexpression of FAIM-L partially cellular protein fractionation assay confirmed the localiza- restored SIVA-1-induced XIAP ubiquitination. Overall, tion of SIVA-1 in cytosolic and light membrane fractions. As our results suggest that SIVA-1 and FAIM-L exert expected FAIM-L also was found in cytosolic fraction (Fig. opposite effects on XIAP ubiquitination levels. 3c). We also examined SIVA-1 co-localization with synaptic markers in the neurites of 14 DIV hippocampal neurons and SIVA-1 overexpression induces neuronal death by caspase- found that SIVA-1 can also locate in presynaptic and post- 3 activation synaptic terminals as indicated by a positive correlation of SIVA-1 was first reported as a proapoptotic protein, and SIVA-1 immunostaining with both PDS95 and synapsin II its overexpression has been shown to induce apoptosis in (Fig. 3d). various cell lines12,22–24. Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 10 of 19 Fig. 3 (See legend on next page.) Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 11 of 19 (see figure on previous page) Fig. 3 SIVA-1 has a mainly cytosolic distribution in neurons. a Representative confocal images of hippocampal neurons. At 7 days in vitro, immunofluorescence was performed by staining with anti-SIVA-1 (green), Hoechst (nucleus marker, blue). MitoTracker (mitochondria marker), anti- Rab5 (early endosome marker), anti-calnexin (ER marker), anti-β III Tubulin (cytoskeleton marker), and anti-TrkA (cytosolic membrane marker) were stained in red. Scale bar 10 µm. b Graph represents Mander’s coefficients M1 corresponding to the channel 1 (SIVA-1) signal that co-located with the channel 2 (cellular markers) signal. Only values higher than 0.5 were considered to indicate co-localization. c Subcellular protein fractionation assay of adult mouse brain. Histone 3 antibody was used as a nuclear marker, GluA2 as a plasma membrane marker, anti-calnexin as reticular membrane protein, cytochrome C as a mitochondrial protein, and GAPDH as a cytosolic protein. N nuclear fraction, HM heavy membrane fraction, LM light membrane fraction, C cytosolic fraction. d Representative confocal images of hippocampal neurites. At 14 days in vitro, immunofluorescence was performed by staining with anti-SIVA-1 (green), presynaptic markers synapsin II (red) postsynaptic marker PSD95 (blue). The graph represents Mander’s coefficient corresponding to co-localization between SIVA-1 and the synaptic markers. SIVA-1 was found to locate in some of the synapses (arrows and asterisks). Scale bar 2 µm. neurons (Fig. 5a, b) respect to the empty vector, strong cleavage of caspase-3 (Fig. 5c), and an evident increase in caspase-3 activity (Fig. 5d). Treating cells with the pan- caspase inhibitor Q-VD consistently abrogated SIVA-1- induced apoptosis (Fig. 5a, b), thereby indicating that SIVA-1 induces caspase-dependent cell death. Western blot analysis of HA-SIVA-1 (Fig. 5e) and immunocy- tochemistry analysis by counting GFP-positive cells (Fig. 5f) confirmed SIVA-1 overexpression by lentiviral vector in murine hippocampal neurons. SIVA-1 modulates GluA2 internalization in hippocampal neurons XIAP and FAIM-L modulate the non-apoptotic function of caspases, such as AMPAR GluA2 subunit internalization, the main mechanism of chLTD4,11,25. We found that SIVA- 1 interacts with both XIAP and FAIM-L, and is an activator of caspases. We therefore proceeded to analyze whether SIVA-1 also regulates this plasticity process. We performed a GluA2 internalization assay. Although no change was detected in SIVA-1-overexpressing neu- rons once NMDA treatment had been applied to induce chLTD, we found that the sole ectopic expression of SIVA-1 was sufficient to induce an increase in the inter- nalization levels of the endogenous AMPAR subunit Fig. 4 SIVA-1 promotes XIAP ubiquitination. a PC12 cells were GluA2 (Fig. 6a–c) in neurons. SIVA-1-induced inter- transiently transfected with Myc-XIAP, FLAG-FAIM-L, and two nalization was blocked by the overexpression of FAIM-L, concentrations of HA-SIVA-1. Myc-XIAP was immunoprecipitated thereby confirming a functional interaction between these using anti-Myc antibody. The membrane was blotted with anti-FLAG to detect FAIM-L and anti-HA to detect SIVA-1. The antibody against two proteins. As previously described in Martinez- XIAP was used as a control of the immunoprecipitation and anti- Marmol et al.11, FAIM-L overexpression abrogated tubulin as loading control. b Cells were transfected with FAIM-L and GluA2 internalization induced by chLTD. SIVA-1 (HA-SIVA-1), and endogenous ubiquitin was Given that caspases are essential for GluA2 inter- immunoprecipitated using anti-ubiquitin antibody. The membrane nalization, we addressed whether the decrease in GluA2 was blotted with FAIM-L, XIAP, and HA to detect SIVA-1, and tubulin was used as a loading control. basal surface level in SIVA-1-overexpressing neurons was caspase-dependent. The presence of the pan-caspase inhibitor Q-VD totally abrogated SIVA-1-induced inter- Here, we explored the effects of ectopic expression of nalization of GluA2 (Fig. 7a, b), indicating that SIVA-1 SIVA-1 in primary hippocampal neurons. modulates postsynaptic receptor levels in a caspase- The overexpression of SIVA-1 was sufficient to induce dependent manner. the typical hallmarks of apoptosis, such as chromatin To further examine the role of SIVA-1 in the regula- condensation and/or fragmentation, in 10% of the tion of synaptic receptor internalization, we knocked Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 12 of 19 Fig. 5 SIVA-1 induces caspase-3-dependent cell death in neurons. Primary hippocampal neurons were infected for 72 h with lentiviral particles carrying HA-SIVA-1 plasmid. a Representative images of Hoechst staining in control and overexpressing cells in the absence or presence of Q-VD, scale bar 20 µm. b Quantification of the percentage of nuclei with apoptotic morphology by Hoechst staining. SIVA-1 induced apoptosis, Q-VD treatment rescued cells from SIVA-1-induced cell death. Two-way ANOVA (empty vector vs. SIVA-1, F (1, 36) = 8.537 in untreated condition; F (1, 36) = 7.787 in Q-VD condition) *p ≤ 0.05. c Treatment with 10 µM of pan-caspase inhibitor Q-VD abrogated caspase-3 cleavage. The membrane was immunoblotted with anti-caspase-3 and anti-tubulin as a loading control. d DEVDase activity was measured in cells overexpressing SIVA-1 or control empty vector. SIVA-1 induced the activation of caspase-3 (F (1, 8) = 38.69). Q-VD treatment abrogated caspase-3 activity (F (1, 8) = 289.9) UT untreated, Q-VD quinolyl-Val-Asp-OPh. Two-way ANOVA *p ≤ 0.05; ***p ≤ 0.001. e SIVA-1 expression verified by western blot. HA was used to detect SIVA-1; tubulin was used as a loading control. f Immunocytochemistry of SIVA-1 and GFP to confirm lentiviral transfection and overexpression. Hoechst staining was used as a nuclear marker. Scale bar 10 µm. down SIVA-1 levels and analyzed GluA2 internalization. SIVA-1 levels are upregulated in chLTD induction A decrease in SIVA-1 levels was observed when a Our findings suggest a role of SIVA-1 in caspase- shRNA was used (Fig. 7e). Although no change was dependent GluA2 internalization. The expression of sev- observed in untreated cells, we did detect a decrease in eral modulators of synaptic plasticity is required for the GluA2 internalization in shSIVA-1-infected neurons correct induction and maintenance of plasticity changes26. after the induction of chLTD. GluA2-induced inter- Analysis of SIVA-1 protein levels during chLTD induction nalization was marked in shSIVA-1 neurons, but sig- showed that they were rapidly induced after 5 min of nificantly decreased compared to control infected NMDA treatment (Fig. 8a, b), while SIVA-1 mRNA levels neurons. were unaltered (Fig. 8c). The inhibition of protein Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 13 of 19 Fig. 6 (See legend on next page.) Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 14 of 19 (see figure on previous page) Fig. 6 SIVA-1 promotes GluA2 internalization. a Representative confocal images of neurons transfected with SIVA-1, FAIM-L, or empty vector. GluA2 internalization assay was performed in neurons treated with NMDA to stimulate LTD and in untreated neurons. Only GFP-positive cells (first column) were considered for quantification. Internalized GluA2 (second column, red in merge) and surface GluA2 (third column, green in merge) were measured. Scale bar 10 µm. b Results were normalized to empty vector, untreated cells. Induction of chemical LTD induced GluA2 internalization in empty vector condition and SIVA-1 transfected cells (F (1, 632) = 15.85). Non-stimulated cells overexpressing SIVA-1 showed an increase in GluA2 internalization. FAIM-L overexpression blocked LTD induction (F (3, 632) = 15.17), and its overexpression with SIVA-1 in untreated cells restored basal levels of receptor internalization (F (3, 632) = 15.85). Each point represents an independent experimental repeat in which 15–20 cells were analyzed. UT untreated, NMDA N-methyl-D-aspartate. c GluA2 surface receptor was isolated with the Biotin surface assay in hippocampal cells transfected with SIVA-1 or empty vector as a control. Lower panel shows quantification of GluA2 surface receptor (t (2) = 3,060). Two-way ANOVA *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.0001. translation by cycloheximide blocked the increment of degradation. By stabilizing XIAP levels, FAIM-L regulates SIVA-1 (Fig. 8d), thereby suggesting posttranscriptional caspase-3 activation in apoptosis and exerts non- regulation. apoptotic functions11. To characterize FAIM-L in depth, Induction of chLTD by NMDA stimulation leads to an we performed a two-hybrid assay to detect interacting increase in Ca2+ inflow in neurons, followed by calci- proteins. neurin and caspase-3 activation, and the internalization of We identified and confirmed SIVA-1 to be a functional AMPAR. Blockage of calcium by BAPTA-AM treatment partner of FAIM-L. SIVA-1 was first reported as a CD27- inhibited the increase in SIVA-1 expression after chLTD. interacting protein12. SIVA-1 plays a critical proapoptotic This observation thus suggests that the first steps of the role in several cellular types and is overexpressed in var- NMDA stimulation pathways are essential for SIVA-1 ious pathological circumstances, such as ischemic injury induction (Fig. 8d). and Coxsakie virus B3 infection14,30–32. Resch et al.13 In conclusion, we propose SIVA-1 as a new modulator demonstrated that SIVA-1 interacts with XIAP, and of synaptic plasticity. We demonstrate that SIVA-1 is suggested that the interaction leads to mutual interference increased in chLTD, where it activates caspase-3. We with their action. propose that SIVA-1, by modulating the FAIM-L/XIAP Our analysis of SIVA-1 expression and distribution in interaction, induces XIAP degradation and consequent brain tissues at embryonic and adult stages shows that this activation of caspases. SIVA-1 therefore emerges as a protein is expressed in both developing and adult neu- novel regulator of caspase-dependent apoptotic and non- rons. Nervous system development requires a precise apoptotic functions in neurons (Fig. 8e). regulation of programmed cell death, and many apoptotic proteins have been found to be crucial. Among these, Discussion SIVA-1 appears to play a key role during development, as Apoptosis is an essential process for the homeostasis of its deletion has been reported to result in embryonic organisms. It is subjected to multiple and tight mechan- lethality in mice33. The presence of SIVA-1 expression is isms of regulation, but it can also be unscheduled trig- maintained in adult cortical and hippocampal neurons, gered in pathological processes such as degenerative thereby suggesting that this protein contributes to neu- diseases. In these pathologies, the mechanisms of cellular ronal physiological functions. protection (i.e., antiapoptotic mechanisms) fail. Under- We examined the subcellular distribution of SIVA-1 standing the underlying components of the dysregulation in primary hippocampal neurons. Previous studies car- may provide new opportunities for therapeutic interven- ried out in various cell types reported controversial tion. The disassembly of neuronal integrity during cell findings on the subcellular localization of this protein. death is regulated by cysteine-aspartic acid proteases Some described that the ectopic expression of SIVA-1 called caspases. These proteins have been widely char- accumulates in the nucleus24,34,35 and that it can inter- acterized for their role in apoptosis27 and in non- act with cell surface membrane receptors12,23,24,34. apoptotic functions. In neurons, caspases locate in den- Moreover, Jacobs et al.14 reported membrane localiza- drites, axons, and in presynaptic and postsynaptic term- tion of SIVA-1 upon DNA-damage induced in the brain. inals. At these sites, they can be locally activated4 and However, most of the functions exerted by SIVA-1 in participate in synaptic plasticity and growth cone motility, apoptosis take place in the cytoplasm36. Our analysis among other non-apoptotic functions28. shows that, in cultured hippocampal neurons, SIVA-1 is FAIM-L is a neuron-specific death receptor antagonist located in cellular bodies, neurites and also in synapses, that protects neurons against a variety of neurotoxic and that it presents a diffuse cytosolic pattern that insults6,8,10,29. We have reported that FAIM-L protects excludes the nucleus, similar to the localization pattern XIAP, an inhibitor of caspases, from ubiquitination and described for FAIM-L. Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 15 of 19 Fig. 7 SIVA-1 modulates AMPAR internalization in a caspase-dependent manner. a Representative confocal images of neurons transfected with SIVA-1 or empty vector and treated with 10 μM of Q-VD (caspase inhibitor). Only GFP-positive cells (first column) were considered for quantification (left panel). Scale bar 10 µm. b The results of the quantification of internalized GluA2 vs. surface GluA2 were normalized to empty vector, untreated cells (right panel). Each point represents an independent experimental repeat in which 15–20 cells were analyzed. SIVA-1-transfected cells showed greater internalization of GluA2 receptor (F (1, 97) = 3.840), Q-VD treatment restored internalization to basal levels (F (1, 97) = 7.345). Two-way ANOVA ***p ≤ 0.001. c Representative confocal images of neurons transfected with vectors carrying shSIVA-1 or shScramble as control. The GluA2 internalization assay was performed in neurons treated with NMDA to stimulate LTD and in untreated neurons. Only GFP-positive cells (first column) were considered for quantification. Internalized GluA2 (second column, red in merge) and surface GluA2 (third column, green in merge) were measured. Scale bar 10 µm. d The results of the quantification of internalized GluA2 vs. surface GluA2 were normalized to shScrambled, untreated cells. Each point represents an independent experimental repeat in which 15–20 cells were analyzed. NMDA treatment induced GluA2 internalization in both shScrambled and shSIVA-1 cells (F (1, 84) = 167.1). However, shSIVA-1 cells treated with NMDA showed a significant decrease in receptor internalization (F (1, 84) = 33.5). Two-way ANOVA **p ≤ 0.01; ***p ≤ 0.001. shScr shScrambled. Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 16 of 19 Fig. 8 (See legend on next page.) Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 17 of 19 (see figure on previous page) Fig. 8 Levels of SIVA-1 increase after NMDA treatment. a Primary neuronal cells were treated with 50 μM of NMDA at different time points (0–30 min) and protein expression of SIVA-1 was analyzed by SDS-PAGE. Anti-tubulin was used as a loading control. b SIVA-1 expression increased with 5 min, and 15 min treatment of NMDA. t test (t (15) = 2427; t(15) = 2755). *p ≤ 0.05. c qPCR revealed no change in SIVA-1 mRNA expression after treatment. d The increase in SIVA-1 levels (two-way ANOVA F(2, 6) = 1712) was blocked by pretreatments with 50 µM of BAPTA-AM for 30 min or 1 µg/ml of cycloheximide for 1 h (two-way ANOVA (F (3, 9) = 1326)). *p ≤ 0.05 for comparison between time points in the same treatment, #p ≤ 0.05 for comparison between treatments. e Schematic representation of SIVA-1 function. In neurons, SIVA-1 was increased after chemical LTD induction. SIVA-1 destabilized XIAP by increasing its ubiquitination and impairing its interaction with FAIM-L. We propose that this is one of the mechanisms through which SIVA-1 triggers caspase-3 activation in neurons and consequent apoptosis and caspase-3-dependent GluA2 internalization. CHX cycloheximide, BAPTA-AM 1,2-Bis(2-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid tetrakis(acetoxymethyl ester), chLTD chemical long-term depression. We then analyzed the possible functions of SIVA-1 in chLTD is induced. A rapid increase in calcium levels is relation to the previously described FAIM-L function on triggered by NMDA receptor activation, followed by an XIAP modulation10,11. Our data show that SIVA-1 over- involvement of calcineurin and caspase-3. The activation expression displaces the FAIM-L/XIAP interaction. Loss of culminates in AMPAR internalization and therefore interaction with FAIM-L can be sufficient to promote XIAP synaptic weakening40. When measuring surface AMPAR instability, thereby inducing ubiquitination-mediated pro- levels in our cellular primary cultures, we observed that teasomal degradation. We reveal that SIVA-1 over- SIVA-1 induces an increase in GluA2 internalization in expression also has a direct effect on promoting XIAP non-stimulated neurons in a caspase-dependent manner. ubiquitination. Given that SIVA-1 is a XIAP-interacting Moreover, SIVA-1 knockdown significantly decreases protein, it might confer structural changes to XIAP that receptor internalization following chLTD. Therefore SIVA- permit auto-ubiquitination, or it may serve as an ubiquitin 1 emerges as another regulator of apoptosis that is also ligase on XIAP. In fact, SIVA-1 has been described to have required for functions not related to cell death. FAIM-L E3 ubiquitin ligase activity and to direct E3 ubiquitin ligases overexpression abrogates GluA2 internalization following to promote the ubiquitination of other proteins and con- chLTD and also restores SIVA-1 induction of GluA2 sequent changes in their function or stability37,38. In a internalization in hippocampal neurons. complex process such as apoptosis, which involves hun- Protein composition at synapses is modulated in response dreds of molecules, the regulation between proapoptotic to neuronal activity, through new synthesis and and antiapoptotic proteins is common and can often be removal41,42. We show an increase in SIVA-1 protein levels mediated by ubiquitination39. in primary cultures after a few minutes of NMDA treat- Here, we report the proapoptotic activity of SIVA-1 in ment. The increase is blocked by cycloheximide and does primary cultures of hippocampal neurons. SIVA-1 over- not affect SIVA-1 mRNA levels, thereby indicating specific expression is sufficient to induce apoptosis through translational regulation. The increase in SIVA-1 could be effector caspase activation. We propose that the promo- directly associated with NMDA receptor stimulation after tion of cell death induced by SIVA-1 overexpression treatment, since a calcium chelating agent, BAPTA-AM, occurs through XIAP destabilization and consequent blocks this increase. BAPTA-AM pretreatment has been inhibition of XIAP action on effector caspases. Here we reported to block LTD induction, thus eliminating available provide evidence that SIVA-1 participates in XIAP mod- calcium ions to start a stimulation response after NMDA or ulation and apoptosis and that it exerts the opposite electrical application43,44. The enhanced SIVA-1 expression function to FAIM-L. in chLTD supports the relevance of this protein in the We addressed whether SIVA-1 has a neuron-specific modulation of this synaptic plasticity mechanism. non-apoptotic role. We demonstrate that SIVA-1 is In summary, our results reveal novel key roles of SIVA- expressed in cultured hippocampal neurons in the absence 1 in neurons. We show that SIVA-1 interacts with both of proapoptotic stimuli. This finding points to a non- FAIM-L and XIAP and that it is also a caspase activator. apoptotic role of the protein. We studied LTD, a well We associate SIVA-1-induced caspase activation with characterized caspase-dependent mechanism, which neural apoptosis, which is consistent with its previously requires a controlled activation of the same mitochondrial described functions. Moreover, we demonstrate a novel pathway and that involves the same proteins that can lead function of SIVA-1 as regulator of synaptic function. to apoptotic cell death. Many apoptotic regulatory proteins modulate caspase activation in synaptic depression, and Acknowledgements among these we find both FAIM-L and XIAP as inhibitors This work was funded by grants awarded by the Spanish “Ministerio de Economía y Competitividad” (SAF2013-47989-R, SAF2016-80236-R, CIBERNED of the process. We used as a model for our study the CB06/05/1104 and PIE13/00027), the Generalitat de Catalunya (2014SGR1609), extensively described chLTD. After NMDA stimulation, and the Fundació La Marató de TV3 (201414-30) to J.X.C. E.C. is supported by a Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 18 of 19 predoctoral fellowship from the Vall d´Hebron Research Institute (VHIR). R.B. is 12. Prasad, K. V. et al. CD27, a member of the tumor necrosis factor receptor supported by a predoctoral fellowship from the Spanish “Ministerio de family, induces apoptosis and binds to Siva, a proapoptotic protein. Proc. Natl Economía y Competitividad” (BES-2018-080846). Acad. Sci. USA 94, 6346–6351 (1997). 13. Resch, U. et al. Siva1 is a XIAP-interacting protein that balances NFkappaB and Author details JNK signalling to promote apoptosis. J. Cell Sci. 122, 2651–2661 (2009). 1 14. Jacobs, S. B., Basak, S., Murray, J. I., Pathak, N. & Attardi, L. D. Siva is an apoptosis- Cell Signaling and Apoptosis Group, Vall d’Hebron Research Institute (VHIR), 08035 Barcelona, Spain. 2Centro de Investigación Biomédica en Red sobre selective p53 target gene important for neuronal cell death. Cell Death Differ. Enfermedades Neurodegenerativas (CIBERNED), ISCIII, 28031 Madrid, Spain. 14, 1374–1385 (2007). 3 15. Dalby, B. et al. Advanced transfection with Lipofectamine 2000 reagent: primary Institut de Neurociències, Departament de Bioquímica i Biologia Molecular, Facultat de Medicina, Universitat Autònoma de Barcelona, 08031 Bellaterra, neurons, siRNA, and high-throughput applications. Methods 33, 95–103 (2004). Spain. 4Institut de Neurociències, Universitat de Barcelona, Bellaterra, Spain. 16. Susin, S. A. et al. Two distinct pathways leading to nuclear apoptosis. J. Exp. 5 Med. 192, 571–580 (2000). Department of Cell Biology, Physiology and Immunology, Institut de Neurociències, Universitat de Barcelona, 08031 Barcelona, Spain. 6Group of 17. Galenkamp, K. M. et al. TNFalpha sensitizes neuroblastoma cells to FasL-, Translational Research in Child and Adolescent Cancer, Vall d’Hebron Research cisplatin- and etoposide-induced cell death by NF-kappaB-mediated expres- Institute (VHIR)-UAB, 08035 Barcelona, Spain. 7Cell Cycle Laboratory, Institut de sion of Fas. Mol. Cancer 14, 62 (2015). Recerca Biomèdica de Lleida (IRBLleida), and Departament de Ciències 18. Urresti, J. et al. Lifeguard inhibits Fas ligand-mediated endoplasmic reticulum- Mèdiques Bàsiques; Facultat de Medicina, Universitat de Lleida, 25198 Lleida, calcium release mandatory for apoptosis in type II apoptotic cells. J. Biol. Chem. Catalonia, Spain. 8ICREA Academia, Barcelona, Spain. 9Present address: 291, 1221–1234 (2016). Department of Pathology, NYU Langone Health, New York 10016 NY, USA 19. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012). Author contributions 20. Dunn, K. W., Kamocka, M. M. & McDonald, J. H. A practical guide to evaluating E.C., L.P.-F., B.B.-Z., E.S., R.S.M., J.X.C., and M.J.P.-G. designed the research; E.C., L. colocalization in biological microscopy. Am. J. Physiol. Cell Physiol. 300, P.-F., R.B.-R., M.P., M.F.S., R.F.-H., E.G., and R.S.M. performed the research; E.C., M. C723–C742 (2011). P., L.P.-F., and E.G. analyzed the data; and E.C., J.X.C, and M.J.P.-G. wrote 21. Livak, K. J. & Schmittgen, T. D. Analysis of relative gene expression data using the paper. real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25, 402–408 (2001). 22. Yoon, Y., Ao, Z., Cheng, Y., Schlossman, S. F. & Prasad, K. V. Murine Siva-1 and Conflict of interest Siva-2, alternate splice forms of the mouse Siva gene, both bind to CD27 but The authors declare that they have no conflict of interest. differentially transduce apoptosis. Oncogene 18, 7174–7179 (1999). 23. Spinicelli, S. et al. GITR interacts with the pro-apoptotic protein Siva and induces apoptosis. Cell Death Differ. 9, 1382–1384 (2002). Publisher’s note 24. Py, B., Slomianny, C., Auberger, P., Petit, P. X. & Benichou, S. Siva-1 and an Springer Nature remains neutral with regard to jurisdictional claims in alternative splice form lacking the death domain, Siva-2, similarly induce published maps and institutional affiliations. apoptosis in T lymphocytes via a caspase-dependent mitochondrial pathway. J. Immunol. 172, 4008–4017 (2004). Received: 7 June 2019 Revised: 8 January 2020 Accepted: 9 January 2020 25. Li, Z., Okamoto, K., Hayashi, Y. & Sheng, M. The importance of dendritic mitochondria in the morphogenesis and plasticity of spines and synapses. Cell 119, 873–887 (2004). 26. Klann, E. & Dever, T. E. Biochemical mechanisms for translational regulation in synaptic plasticity. Nat. Rev. Neurosci. 5, 931–942 (2004). References 27. Riedl, S. J. & Shi, Y. Molecular mechanisms of caspase regulation during 1. Fuchs, Y. & Steller, H. Live to die another way: modes of programmed cell apoptosis. Nat. Rev. Mol. Cell Biol. 5, 897–907 (2004). death and the signals emanating from dying cells. Nat. Rev. Mol. Cell Biol. 16, 28. Gilman, C. P. & Mattson, M. P. Do apoptotic mechanisms regulate synaptic 329–344 (2015). plasticity and growth-cone motility? Neuromol. Med. 2, 197–214 (2002). 2. Hyman, B. T. & Yuan, J. Apoptotic and non-apoptotic roles of caspases in 29. Liu, X. et al. SRT1720 promotes survival of aged human mesenchymal stem neuronal physiology and pathophysiology. Nat. Rev. Neurosci. 13, 395–406 cells via FAIM: a pharmacological strategy to improve stem cell-based therapy (2012). for rat myocardial infarction. Cell Death Dis. 8, e2731 (2017). 3. Hollville, E. & Deshmukh, M. Physiological functions of non-apoptotic caspase 30. Singaravelu, K. & Padanilam, B. J. p53 target Siva regulates apoptosis in activity in the nervous system. Semin Cell Dev. Biol. 82, 127–136 (2018). ischemic kidneys. Am. J. Physiol. Ren. Physiol. 300, F1130–F1141 (2011). 4. Li, Z. et al. Caspase-3 activation via mitochondria is required for 31. Padanilam, B. J. & Lewington, A. J. & Hammerman, M. R. Expression of CD27 long-term depression and AMPA receptor internalization. Cell 141, 859–871 and ischemia/reperfusion-induced expression of its ligand Siva in rat kidneys. (2010). Kidney Int. 54, 1967–1975 (1998). 5. Erturk, A., Wang, Y. & Sheng, M. Local pruning of dendrites and spines by 32. Henke, A. et al. Apoptosis in coxsackievirus B3-caused diseases: interaction caspase-3-dependent and proteasome-limited mechanisms. J. Neurosci. 34, between the capsid protein VP2 and the proapoptotic protein siva. J. Virol. 74, 1672–1688 (2014). 4284–4290 (2000). 6. Carriba, P. et al. Amyloid-beta reduces the expression of neuronal FAIM-L, 33. Van Nostrand, J. L. et al. The p53 target gene SIVA enables non-small cell lung thereby shifting the inflammatory response mediated by TNFalpha from cancer development. Cancer Discov. 5, 622–635 (2015). neuronal protection to death. Cell Death Dis. 6, e1639 (2015). 34. Lin, F. T., Lai, Y. J., Makarova, N., Tigyi, G. & Lin, W. C. The lysophosphatidic acid 2 7. Tu, S., Okamoto, S., Lipton, S. A. & Xu, H. Oligomeric Abeta-induced synaptic receptor mediates down-regulation of Siva-1 to promote cell survival. J. Biol. dysfunction in Alzheimer’s disease. Mol. Neurodegener. 9, 48 (2014). Chem. 282, 37759–37769 (2007). 8. Segura, M. F. et al. The long form of Fas apoptotic inhibitory molecule is 35. Balci-Peynircioglu, B. et al. Pyrin, product of the MEFV locus, interacts with the expressed specifically in neurons and protects them against death receptor- proapoptotic protein, Siva. J. Cell Physiol. 216, 595–602 (2008). triggered apoptosis. J. Neurosci. 27, 11228–11241 (2007). 36. Chu, F. et al. The Siva-1 putative amphipathic helical region (SAH) is sufficient 9. Scott, F. L. et al. XIAP inhibits caspase-3 and -7 using two binding sites: to bind to BCL-XL and sensitize cells to UV radiation induced apoptosis. evolutionarily conserved mechanism of IAPs. EMBO J. 24, 645–655 (2005). Apoptosis 9, 83–95 (2004). 10. Moubarak, R. S. et al. FAIM-L is an IAP-binding protein that inhibits XIAP 37. Wang, X. et al. Siva1 inhibits p53 function by acting as an ARF E3 ubiquitin ubiquitinylation and protects from Fas-induced apoptosis. J. Neurosci. 33, ligase. Nat. Commun. 4, 1551 (2013). 19262–19275 (2013). 38. Han, J. et al. SIVA1 directs the E3 ubiquitin ligase RAD18 for PCNA mono- 11. Martinez-Marmol, R. et al. FAIM-L regulation of XIAP degradation modulates ubiquitination. J. Cell Biol. 205, 811–827 (2014). synaptic long-term depression and axon degeneration. Sci. Rep. 6, 35775 39. Lee, J. C. & Peter, M. E. Regulation of apoptosis by ubiquitination. Immunol. Rev. (2016). 193, 39–47 (2003). Official journal of the Cell Death Differentiation Association Coccia et al. Cell Death and Disease (2020)11:82 Page 19 of 19 40. Benarroch, E. E. Glutamatergic synaptic plasticity and dysfunction in Alzheimer 43. Mulkey, R. M. & Malenka, R. C. Mechanisms underlying induction of homo- disease: Emerging mechanisms. Neurology 91, 125–132 (2018). synaptic long-term depression in area CA1 of the hippocampus. Neuron 9, 41. Sajikumar, S. & Frey, J. U. Anisomycin inhibits the late maintenance of long-term 967–975 (1992). depression in rat hippocampal slices in vitro. Neurosci. Lett. 338, 147–150 (2003). 44. Artola, A. & Singer, W. Long-term depression of excitatory synaptic transmis- 42. Kauderer, B. S. & Kandel, E. R. Capture of a protein synthesis-dependent com- sion and its relationship to long-term potentiation. Trends Neurosci. 16, ponent of long-term depression. Proc. Natl Acad. Sci. USA 97, 13342–13347 (2000). 480–487 (1993). Official journal of the Cell Death Differentiation Association