Abstract
We recently reported that the ER stress kinase PERK regulates ER-mitochondria appositions and ER– plasma membrane (ER-PM) contact sites, independent of its canonical role in the unfolded protein response. PERK regulation of ER-PM contacts was revealed by a proximity biotinylation (BioID) approach and involved a dynamic PERK-Filamin A interaction supporting the formation of ER-PM contacts by actin-cytoskeleton remodeling in response to depletion of ER-Ca2+ stores. In this report, we further interrogated the PERK BioID interactome by validating through co-IP experiments the interaction between PERK and two proteins involved in Ca2+ handling and ER-mitochondria contact sites. These included the vesicle associated membrane (VAMP)-associated proteins (VAPA/B) and the main ER Ca2+ pump sarcoplasmic/endoplasmic reticulum Ca ATPase 2 (SERCA2). These data identify new putative PERK interacting proteins with a crucial role in membrane contact sites and Ca2+ signaling further supporting the uncanonical role of PERK in Ca2+ signaling through membrane contact sites (MCSs).
Introduction
The endoplasmic reticulum (ER) is a major site of protein folding in the cell, handling roughly one-third of all cellular proteins and all proteins destined for transport towards the plasma membrane (PM) or extracellular matrix (Brodsky & Wojcikiewicz, 2009). To cope with this demand, the ER has evolved to possess an intricate folding machinery and an ideal protein folding environment. However, when the ER can no longer match cellular protein folding demand, the resulting accumulation of unfolded proteins causes ER stress (Ron & Walter, 2007). The unfolded protein response (UPR) consists of the activation of a conserved signal transduction pathway, ultimately eliciting a transcriptional program that operates as a principal safeguard against the loss of ER homeostasis caused by ER stress. The UPR is launched by the activation of three ER membrane proteins. The ER stress kinase PKR-like endoplasmic reticulum kinase (PERK) is one of these three mediators and is activated upon ER stress (Ron & Walter, 2007). In homeostatic conditions, PERK is kept inactive through the binding of the ER chaperone BiP to its luminal domain, and is activated upon its release, resulting in the oligomerization of PERK, followed by its autophosphorylation. Activated PERK is then able to phosphorylate eukaryotic initiation factor 2 alpha (eIF2α), leading to a protein translation pause and giving the ER folding machinery time to deal with its protein burden (Ron & Walter, 2007).
In our previous studies (van Vliet et al., 2017; Verfaillie et al., 2012), we uncovered that independent of its UPR function, PERK moonlights at the ER-mitochondria contacts and aids apoptotic cell death by the transfer of ROS signals from the ER to the mitochondria. More recently, we showed that the activation of PERK can occur independently of its luminal domain and canonical ER stress, instead of being activated by a rise in cytosolic Ca2+.
To uncover the roles of PERK we carried out an unbiased proximity biotinylation (BioID) screen, using the promiscuous biotinylation enzyme BirA* tagged to PERK on its cytosolic side (Roux et al., 2012; van Vliet et al., 2017). Using this technique, we discovered the cytoskeletal protein Filamin A (FLNA) as a novel PERK interactor. The Ca2+ -mediated PERK-FLNA axis was found to be required to support the formation of ER-PM contacts and store-operated Ca2+ entry (SOCE) (van Vliet et al., 2017). These findings together hint at a broader cross-talk between ER stress proteins and the regulation of membrane contact sites (MCSs). However, many hits uncovered through our BioID experiment warrant further investigation in order to get a better picture of the potential additional roles of PERK. In this report, we reveal the full dataset of PERK interacting proteins identified through BioID analysis. We further confirmed by IP/Co-IP analysis, the physical interaction between PERK and sarcoplasmic/endoplasmic reticulum Ca ATPase 2 (SERCA2) and VAMP-associated protein A / B (VAPA/B) and show that for these interactions, PERK kinase activity is dispensable.
Results and Discussion
The PERK Proximity Interactome
We generated a C-terminally tagged PERK-BirA* construct that was well expressed in HEK293-T cells and showed the expected ER localization (van Vliet et al., 2017). Our BioID approach was designed and performed to closely match the original study reporting BioID (Roux et al., 2012). Similar to that study, our approach relied on the expression of PERK-BirA* in HEK293-T cells, using mock-transfected parental HEK293-T cells treated in the same way (50 µM biotin for 24 h), as a control. Only biotinylated protein hits identified by LC-MS/MS in the streptavidin pulldown from PERK-BirA* transfected cells and not from mock-transfected cells were taken into consideration as putative interaction partners. Table 1 shows a list of PERK proximity interactors showing no spectral counts in the pulldown from mock-transfected cells. Interestingly, while our previous research focused on one of the most prominent hits, Filamin A, which we further validated functionally, the BioID dataset showed several other potential interacting proteins (Table 1), ranging from more proteins involved in actin cytoskeleton maintenance (cofilin, profilin,...), proteins involved in ER trafficking (syntaxin 5, coatomer subunits), proteins involved in membrane contact sites (VAPA/B, E-Syt1, junctophilin) to proteins linked with metabolism (ATP-citrate synthase, acyl-protein thioesterase 2). Interestingly, PERK has been linked recently with metabolic regulatory networks (Balsa et al., 2019; Moncan et al., 2021; Sorge et al., 2020), although whether these potential interactors uncovered by our BioID are relevant remains to be tested. To gain more insight into the various groups of proteins in the dataset, we performed a gene ontology analysis using WEB-based GEne SeT AnaLysis Toolkit (http://www.webgestalt.org/, Figure 1). This analysis yielded a set of enriched biological processes in our dataset. As a validation of our approach, two of these groups are linked to PERK's main role in protein folding and ER stress response (Figure 1A and B).
(A) Gene ontology analysis of the protein hits discovered using PERK-BirA* obtained using WEB-based GEne SeT AnaLysis toolkit (http://www.webgestalt.org/). The minimum number of IDs in each category was required to be 5, the maximum was 2000, the false-discovery rate (FDR) correction used was Benjamini and Hochberg (BH). The categories were first ranked based on FDR and after that the topmost significant categories were selected. 10 categories of biological processes were identified, many related to the function of PERK. The enrichment ratio indicates the ratio between the fraction of proteins belonging to each biological process in our dataset over the fraction of these proteins expected If our dataset was completely random. (B) Two specific categories are highlighted, response to endoplasmic reticulum stress, and protein folding, indicating an enrichment in the dataset of proteins and processes linked to PERK's role in the unfolded protein response and ER stress.
List of identified proteins (Scaffold, FDR < 1%) resulting from the BioID interactome screen using PERK-BirA as bait. Protein hits were only detected using PERK-BirA as bait and not in control. Parental cells are shaded in yellow. Relative quantification of proteins is based on spectral counts (‘Total spectra’). Only proteins with atleast 2 exclusive unique peptides per protein are listed.
Considering our previous and ongoing studies, we decided to focus on validating the interactions of PERK with proteins with a known tethering role and/or with a relevant function in membrane contact sites and ER Ca2+ homeostasis. Two hits that were interesting in this regard were SERCA1/2 and VAPA/B, found at positions 47 and 62, respectively, on our list (Table 1).
PERK Interaction With VAPB
The isoforms VAPA and B are members of a small VAP protein family and have broadly similar structures and functions (Murphy & Levine, 2016). They are tail-anchored ER membrane proteins that are central to the formation of MCSs. VAPs act as tethers for a growing group of proteins, including protein tyrosine phosphatase interacting protein 51 (PTPIP51), StAR Related Lipid Transfer Domain Containing 3 (STARD3), oxysterol-binding protein (OSBP), Nir2, and Sorting nexin 2 (SNX2), among others (Alpy et al., 2013; Amarilio et al., 2005; De Vos et al., 2012; Dong et al., 2016; Wyles et al., 2002). VAPB is an important mediator of both ER-mitochondria contact sites and ER-plasma membrane contact sites, where it can interact with various proteins responsible for tethering and lipid trafficking at these contact sites, including PTPIP51 on mitochondria and Nir2 at the ER-PM contacts. Given the role of PERK in modulating both ER-mitochondria and indirectly ER-PM contacts, we then explored the possibility that VAPA/B is a bona fide interacting partner of PERK.
In our dataset, we picked up unique peptides for both VAPA and VAPB, indicating that both proteins might interact with PERK. Because VAPB has traditionally been the more studied of the two at MCS, and because both proteins are highly similar, for simplicity, here we focused on VAPB. We tested the possible interaction between VAPB and PERK by co-immunoprecipitation (co-IP) experiments and successfully confirmed that immunoprecipitation of wild type PERK-myc (PERKWT) expressed in HEK293-T cells, pulled down VAPB (Figures 2 and S2). In previous studies, we showed that the function of PERK required at the ER-mitochondria and ER-PM contacts was independent of its kinase function (van Vliet et al., 2017; Verfaillie et al., 2012). Here, to investigate whether PERK's kinase activity was dispensable for PERK-VAPB interaction, we performed the IP in cells expressing either a PERKWT or a kinase dead mutant of PERK (PERKKD). Interestingly, the expression of the PERKKD mutant yielded no difference in VAPB binding (Figures 2 and S2). Together these observations suggest that PERK interacts with VAPB constitutively, and in the absence of a signal, evoking its UPR activation. It is possible that VAPB could recruit PERK to specific MCS upon certain stresses, or vice versa; further research would be needed to validate this possibility.
Identification of VAPB and SERCA2 as putative PERK interactors. HEK293-T cells were transfected with myc-tagged expression vectors for PERK WT (PERKWT) and PERK Kinase dead (PERKKD). After 48 h of transfection, PERKWT and PERKKD were immunoprecipitated (PERK IP) and interactors were detected using antibodies against VAPB and SERCA2 by immunoblot. Input was 10% of the total protein amount used for the IP (50 µg of protein was loaded as an input versus 500 µg of total protein used for the IP). Unspecific anti-mouse antibody was used as a negative control (Mouse IP). Data show are representative of N = 3 (VAPB), N = 2 (SERCA2) biologically independent experiments.
Proteins that interact with VAPs usually do so through a specific domain, termed a FFAT motif (two phenylalanines (FF) in an acidic tract) that binds the major sperm protein domain (MSP) on VAP (Murphy & Levine, 2016). However, using a previously published algorithm (Murphy & Levine, 2016) we did not find any robust FFAT motif in the PERK primary sequence. This suggests that the PERK-VAPB interaction may involve either another FFAT containing protein or may be mediated through another binding in cis, as they are both ER membrane proteins. This would leave the MSP of VAPB free to bind FFAT motifs in other proteins. A recent publication has explored the human “VAPome” in a systematic way using BioID and found that PERK was a hit for at least VAPA (Cabukusta et al., 2020). This same study also reports the putative interaction between PERK and MOSDP1 and 3. These proteins are VAP-like but bind to a slight variation of the FFAT motif, the FFNT motif (two phenylalanines (FF) in a neutral tract). Their analysis indicates that PERK contains an FFNT motif at location 283 of the protein, which is located in the lumen of the ER. This study, along with ours, does give support for a functional role between PERK and VAPA/B and other MCS proteins.
PERK Interaction With SERCA2
SERCA1/2 are ATPases located in the ER membrane that pump Ca2+ ions from the cytosol into the ER lumen, counteracting the Ca2+ leak from various sources (translocon, inositol 1,4,5-trisphosphate receptor) and maintaining a high level of Ca2+ in the ER lumen with a steady-state Ca2+ concentration of approximately 1 mM (de la Fuente et al., 2013).
In contrast to SERCA2, no unique peptides for SERCA1 could be identified in the BioID data set, so the presence of the latter isoform in the PERK interactome could not be proven. Neither SERCA1 nor SERCA2 have been detected at the ER-mitochondria contact sites, but a truncated version of SERCA1, S1 T (truncated after amino acid 395) has been shown to play a role at ER-mitochondria contact sites (Chami et al., 2008). As with VAPB, we confirmed the interaction of SERCA2 with PERK through co-IP, which was again detected independently of its kinase activity (Figures 2 and S2). Our previous results indicated that PERK is strongly activated by increases in cytosolic Ca2+ levels, and we speculated that this is linked to the depletion of ER luminal Ca2+ leading to ER stress, as shown also in a recent study (Preissler et al., 2020). Furthermore, we reported that PERK is involved in SOCE, by regulating the ER-PM contacts through its binding to FLNA (van Vliet et al., 2017). SOCE is induced by ER Ca2+ depletion, which leads to oligomerization of STIM1 and its translocation to the PM where it interacts with the PM localized ORAI1 protein. The interaction of STIM1 with ORAI1 then allows Ca2+ to enter the cytosol from the extracellular medium through the opening of the ORAI1 channel (Zhang et al., 2005). PERK's interaction with SERCA2 is an intriguing link since, although ORAI1 allows Ca2+ to enter the cytosol, the refilling of ER-Ca2+ store requires the activity of the Ca2+ pump SERCA. A previous study has indicated a close relationship between STIM1/ORAI1 driven SOCE and SERCA (Courjaret & Machaca, 2014). We can then speculate that PERK may be located close to the site of STIM1/ORAI1 contact, playing a part in the mechanism that ensures Ca2+ entering the cytosol is rapidly internalized into the ER lumen through SERCA2. Further study is needed to investigate a possible role of PERK in regulating SERCA2 activity, and therefore, influencing this mechanism.
Conclusions and Limitations of the Study
In this report, we show a BioID dataset obtained by tagging the promiscuous biotin ligase BirA* to PERK, in order to map its close interactors. We highlight and validate two of these, VAPB and SERCA2, which lead us to further speculate about the role of PERK in ER MCS formation and Ca2+ signaling.
Our study was performed when BioID was just emerging as a tool, and at the time of our planning, only the original study had been published (Roux et al., 2012). We, therefore, closely modeled our experimental setup on this study. It is important to note when interpreting this dataset that later studies using BioID (and derivatives of BioID like TurboID, where the BirA* ligase has been mutated to biotinylate at a much higher rate), have implemented more stringent controls to detect false positives (inherent with using BioID as an approach to screen protein-protein interactions). For example, the initial study reporting BioID used only parental cells (mock-transfected) as a control, while some recent studies used BirA* (or TuboID) not tagged to their protein of interest to control for non-specific biotinylation events (Szczesniak et al., 2021; Vermehren-Schmaedick et al., 2021; Zhang et al., 2019). Our reasoning for using only parental cells as control was that by using free BirA* as a control, there was a risk that this enzyme might biotinylate bona fide hits randomly, leading to false negatives. Since the initial study reporting BioID, the original authors have published numerous extensive updates on how to set up a BioID study, incorporating more appropriate or bespoke controls and comparing the different new techniques (May et al., 2020; May & Roux, 2019; Roux et al., 2018; Sears et al., 2019). In summary, we have identified 129 potential interacting partners of PERK, of which we have tested and confirmed two, SERCA2 and VAPB. We hope that this dataset can yield new insights concerning PERK and cellular signaling.
Materials and Methods
Cell Lines and Transfection:
HEK293-T cells have been maintained in Dulbecco's modified Eagle's medium containing 4.5 g/l glucose and 0.11 g/l sodium pyruvate and supplemented with 2 mM glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, and 10% fetal bovine serum (FBS) (all added, AA medium). Cells were transiently transfected with different PERK constructs encoding PERKK618A (Addgene plasmid 21815) or PERKFL (Addgene plasmid 21814), both myc tagged, using Trans-IT X2 transfection reagent (Mirus Bio LLC, Science Dr.Madison, WI USA).
BioID, Biotinylation Assay, and Mass Spectrometry
BioID was performed as described previously (Roux et al., 2012), with minor modifications. Overexpression of PERK-BioID was achieved by transiently transfecting Hek293-T cells using X-tremegene 9 (Roche, Germany). A 6 μl X-tremegene9 was mixed with a 2 μg PERK-BioID plasmid and dispersed on cultured cells. A 30 h post-transfection medium containing 50 μM biotin was added to the cells for 24 h. After incubation with biotin, cells were lysed in 1 mL lysis buffer (50 mM Tris, pH 7.4, 500 mM NaCl, 0.4% SDS, 5 mM EDTA, 1 mM DTT, and 1x complete protease inhibitor [Roche]) and sonicated. Triton X-100 was added to a 2% final concentration. After further sonication, an equal volume of 4 °C 50 mM Tris (pH 7.4) was added before additional sonication (subsequent steps at 4 °C) and centrifugation at 16,000 g. Supernatants were incubated with 600 μl Dynabeads (50% slurry) (MyOne Streptavidin C1; Invitrogen) overnight. Beads were collected and washed twice for 8 min at 25 °C (all subsequent steps at 25 °C) in 1 mL wash buffer 1 (2% SDS in dH2O). This was repeated once with wash buffer 2 (0.1% deoxycholate, 1% Triton X-100, 500 mM NaCl, 1 mM EDTA, and 50 mM HEPES, pH 7.5), once with wash buffer 3 (250 mM LiCl, 0.5% NP-40, 0.5% deoxycholate, 1 mM EDTA, and 10 mM Tris, pH 8.1), and twice with wash buffer 4 (50 mM Tris, pH 7.4, and 50 mM NaCl). For western blot analysis, bound proteins were removed from the magnetic beads with 50 μl of Laemmli SDS-sample buffer saturated with biotin at 98 °C. For mass spectrometry analysis, beads were washed repeatedly in MQ water containing 50 mM ammonium bicarbonate (AmBic) before being incubated for 30 min at 37 °C in 50 mM AmBic containing 5 mM DTT. Beads were further washed in 50 mM AmBic and incubated with 25 mM iodoacetamide in the dark for 30 min at 37 °C. After reduction/alkylation, beads were washed in 50 mM AmBic and incubated with 1:20 (w/w) modified trypsin (Pierce) ON at 37 °C in 50 mM AmBic containing 5% acetonitrile. After removal of beads by magnetic separation, formic acid was added to the peptide solution (to 2%) before desalting by C18 Micro Spin Columns (Harvard Apparatus). The resulting peptide mixture was analyzed by nano LC-MS on a hybrid quadrupole-orbitrap mass spectrometer (Q Exactive, Thermo Fisher Scientific). Peptides were identified by MASCOT 2.2 (Matrix Science) using the SwissProt database (taxonomy Homo sapiens, 20231 entries) adopting the following MASCOT search parameters: trypsin/P, two missed cleavages allowed, variable modification oxidation (M), fixed modification carbamidomethylation (C). The mascot was searched with a fragment ion mass tolerance of 0,02 Da and a parent ion mass tolerance of 10 ppm.
Scaffold 4 (Proteome Software Inc.) was used to validate MS/MS-based peptide and protein identifications. Peptide and protein identifications were accepted to achieve an FDR less than 10%. Standard protein grouping was adopted. The presence of at least 2 exclusive unique peptides per protein was required.
Immunoprecipitation
After 48 h of transfection with selected plasmids cells, they were collected through scraping and lysed in lysis buffer (1% CHAPS, 100 mM KCl, 150 mM NaCl, 1x protease inhibitor (Pierce Protease Inhibitor Tablets, Thermo Fisher Scientific Inc.)) for 30 min at 4 °C. Cells were centrifuged at 13.000 g for 15 min to remove debris and unbroken cells. From the supernatant, 500 μg of proteins were combined with primary antibodies overnight (ON) at 4 °C, against PERK and using a non-specific Mouse IgG as a control. Protein-antibody complexes were captured by the addition of Protein AG magnetic beads (Pierce) for 1.5 h at room temperature (RT). Protein AG magnetic beads with captured protein-antibody complexes were washed three times with lysis buffer. Proteins were eluted with sample buffer (62.5 μM Tris-HCl, 10% glycerol, 2% SDS, 1x protease inhibitor, 1x phosphatase inhibitor (Pierce phosphatase Inhibitor Tablets, Thermo Fisher Scientific Inc.) in MQ water) and loaded on a gel for western blot analysis.
Western Blotting
Samples were separated by SDS-PAGE on the Criterion system (Bio-Rad Laboratories, Hercules, CA, USA) on a 4%–12% Bis-TRIS gel and electrophoretically transferred to Protran 2 μm-pored nitrocellulose paper (PerkinElmer, Wellesley, MA, USA). The blots were blocked for 1 h at RT in TBS-T buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 0.1% Tween-20) containing 5% nonfat dry milk and then incubated with selected antibody solutions. Samples were processed and enhanced chemiluminescence using Pierce ECL Western Blotting Substrate was used for western blot detection and membranes were scanned using the Bio-Rad Chemidoc Imager (Bio-Rad Laboratories N.V.3, Winninglaan, Temse, Belgium).
Antibodies
Antibodies used were mouse monoclonal anti-c-Myc (Sigma, Cat# M4439), Control normal mouse IgG (sc-2025), anti-ATP2A2/SERCA2 (Cell signaling 388S), anti-VAPB (Invitrogen, PA5-53023), and an HRP-based detection using as a secondary antibody, the Veriblot antibody (#ab131366, Abcam).
Footnotes
Author Contributions
Experiment design, analysis, and manuscript writing: MLS, PA, ARVV. Performing and analyzing western blots: MLS. Mass spectrometry analysis: RD, EW. Project and funding: PA.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the EOS consortium (grant number 30837538, FAF-F/2018/1252, G049817N, G070115N, G076617N).
