Insulin-induced palmitoylation regulates the Cardiac Na + /Ca 2 + exchanger NCX1 Cell Calcium

The cardiac Na + /Ca2 + Exchanger (NCX1) controls Ca2 + extrusion from the cytosol by mediating bidirectional exchange of Na + for Ca2 + , and therefore controls cardiac relaxation. Insulin regulates Ca2 + handling in cardiac tissue through NCX1, however how insulin changes NCX1 activity is poorly understood. Palmitoylation is the only post-translational modification identified to alter NCX1 activity. Here we show that insulin triggers local structural re-arrangements within existing NCX1 dimers by inducing their palmitoylation, thus tunes NCX1 inactivation through a zDHHC5-dependent mechanism in multiple cell types. By activating fatty acid and fatty acyl CoA synthesis insulin promotes palmitoylation of the zDHHC5 active site, which leads to enhanced NCX1 palmitoylation. Our findings represent a new mechanism to regulate the palmitoylation of numerous zDHHC5 substrates.

Insulin was found to act on intracellular Ca2+ through NCX1 in failing myocardium from patients with and without diabetes [25,50]. Moreover, inhibiting reverse (Ca influx) mode of NCX1 with either KB-R7943 [50] or SEA0400 [25] eliminated insulin dependent positive inotropic effect, suggesting that insulin activates reverse mode of NCX1. Later, Villa-Abrille and colleagues dissected the effect of insulin on NCX1 in isolated cardiomyocytes using whole cell patch clamp. The amplitude of both forward and reverse NCX1 currents measured in isolated myocytes was significantly increased in the presence of insulin [49].
Blocking Protein Kinase C (PKC) but not phosphatidylinositol-3-kinase (PI3K) or nitric oxide (NO) synthase eliminated the effect of insulin on NCX1 current. Nor did disruption of caveolae affect insulin action on NCX1. Interestingly, PIP2 and XIP are involved in insulin regulation of the exchanger function [49]. This latter finding suggests that insulin tunes the ability of NCX1 to transit from activated to inactivated state via its intracellular regulatory domains; however, this is still poorly characterized.
Insulin stimulates a broad range of cellular signaling pathways including post-translational modifications such as phosphorylation [47] [59] and palmitoylation [52]. Unlike the rich background on phosphorylation cascade stimulated by insulin, our knowledge on insulin mediated palmitoylation is limited [52]. Palmitoylation is a type of post-translational modification in which palmitate is dynamically and reversibly conjugated to cysteine (Cys-) residue(s) of a target protein. NCX1 is palmitoylated at Cys-residue positioned at 739 within its intracellular loop, and palmitoylation of NCX1 is crucial for its inactivation [60]. Palmitoylation of NCX1 is directed by a helical structure with an amphipathic nature that is between residues 740 and 756 [61]. This secondary structure element directly interacts with the palmitoylating zDHHC-PAT enzymes [14]. Palmitoylation promotes local Fig. 1. Insulin regulates NCX1 activity by inducing its palmitoylation. A NCX1 structure tagged with FRET probe; either CFP or YFP, at position 266. B Mean CFP and YFP signals measured in NRVMs in the absence (left) and presence (right) of insulin. C Insulin treatment increased NCX1-NCX1 FRET (n:11 for NCX1 and n:15 for NCX1+insulin; p-value: **<0.01, unpaired t-test). D, E, F Effect of insulin on NCX1 palmitoylation in NRVMs (D), ARbMs (E) and Tet-NCX1 (F). Insulin elevated the palmitoylation of NCX1 (n:9 for both NRVMs and NRVMs+insulin; n:7 for both ARbMs and ARbMs+insulin; n:6 for both Tet-NCX1 and Tet-NCX1+insulin, p-value: **<0.01, ****<0.0001; unpaired t-test). Palmitoylation of the lipid raft resident protein flotillin 2 is presented as an assay control. G, H, I, J Measuring NCX1 current using whole cell patch clamp. Example traces from Tet-NCX1 cells in the absence (upper left, in black) and presence (upper right, in red) of insulin and from cells in which NCX1 expression was not induced (lower left, lower right) (G). The amplitude of whole cell currents was increased upon insulin treatment only in cells in which palmitoylatable-NCX1 expression was induced (H). Subtraction of whole cell currents recorded without tet induction of NCX1 from those recorded with tet induction of NCX1 reveals the functional impact of insulin on NCX1 (I). Unpalmitoylatable-NCX1 (C739A) is not regulated by insulin (J) (n:9 for Tet-NCX1 and n:7 for Tet-NCX1+insulin; n:5 for uninduced cells and n:5 for uninduced cells treated with insulin; n:11 for Tet-C739A and n:7 for Tet-C739A+insulin; p-value: *<0.05, **<0.01, ****<0.0001; one-way ANOVA with Sidak's multiple comparisons test). structural changes within the intracellular loop of the exchanger and modulates the interaction between the XIP domain and its binding site, which is near to C739 [13,15]. Hence palmitoylation tunes exchanger activity by modifying the ability of XIP to bind and inactivate NCX1.
Despite the well-established role of insulin in modulation of NCX1 function, to-date the underlying mechanism is still unknown. Herein we provide a thorough analysis of insulin effect on NCX1 function using intersectional approaches at molecular and cellular level. We found that (1) insulin induces local structural re-arrangement within the intracellular loop of the exchanger by modulating its palmitoylation, (2) insulin tunes NCX1 inactivation, (3) zDHHC5; a surface-membrane localized palmitoylating enzyme, is required for insulin induced palmitoylation of NCX1 (4) insulin elevates zDHHC5 palmitoylation at its active site in an acyl CoA-synthase dependent manner. Our findings bring a mechanistic insight into the insulin action on NCX1 function. This knowledge is particularly critical to devise pharmacological tools targeting NCX1 palmitoylation to mimic insulin action on NCX1 activity; therefore, to improve cardiac function.

Insulin mediated palmitoylation induces local conformational changes in the intracellular loop of NCX1, hence alters NCX1 inactivation
The physiological regulation of NCX1 (for example activation and inactivation mechanisms), is controlled by the structural elements that lie within its large intracellular loop. Therefore, we first hypothesized that insulin alters NCX1 function by re-structuring this intracellular loop. To test our hypothesis, we employed an intermolecular FRET approach using full length canine NCX1 with either YFP or CFP at position 266 (Fig. 1A). Cultured Neonatal Rat Ventricular Myocytes (NRVMs) were co-transfected with CFP-and YFP-inserted NCX1 plasmids [15], and the FRET behavior of NCX1-NCX1 pairs was monitored (Fig. 1B, C). Pre-incubation of NRVMs with 1 µM insulin in serum free buffer for 15 min increased NCX1-NCX1 FRET (Fig. 1C), suggesting that insulin either affects NCX1 dimerization or initiates local conformational changes within the existing NCX1 dimer. Insulin regulates ~10% of putative palmitoylated proteins in human endothelial cells (HUVEC) [52]. We first checked if the palmitoylation status of NCX1 was changed upon insulin treatment, and the broader physiological relevance of this by determining whether any effect of insulin on NCX1 palmitoylation was reproduced across multiple cell types. Strikingly, incubation of NRVMs (Fig. 1D), adult rabbit myocytes (ARbM) (Fig. 1E) and tetracycline (Tet-) inducible engineered cells stably expressing WT-NCX1 (Tet-NCX1) ( Fig. 1F) with 1 µM insulin for 15 min caused a modest increment in NCX1 palmitoylation. To address whether insulin mediates functional changes in NCX1 via palmitoylation, we next recorded NCX1 current in cells stably expressing Tet-inducible palmitoylatable-(Tet-NCX1). These experiments used the whole-cell patch clamp with voltage steps from − 120 mV to +100 mV (with 20 mV increments) from a holding potential of − 80 mV (Fig. 1G). We first measured the impact of insulin on whole cell currents in the presence and absence of tet induction of NCX1. Pre-incubation of the cells with insulin increased the magnitude of NCX1 current, while background currents in these cells (in the absence of tetracycline induction of NCX1 expression) were small and not regulated by insulin (Fig. 1H). Subtraction of background currents revealed the substantial increase in NCX1 current mediated by insulin (Fig. 1I). In contrast, in cells stably expressing Tet-inducible unpalmitoylatable-NCX1 (Tet-C739A) insulin elicited no change in NCX1 currents (Fig. 1J). We conclude from these experiments that insulin regulation of NCX1 occurs only when NCX1 is palmitoylatable.
Palmitoylation ultimately governs NCX1 inactivation by modifying XIP binding to its interacting region at 709-728. To further validate our findings of insulin action on NCX1 currents, we set out to antagonize NCX1 inactivation by introducing Lys-(K) to Gln-(Q) mutation at position 229 (K229Q) in the endogenous XIP domain ( Fig. 2A). A XIP domain bearing mutation K229Q is still able to bind to its binding site [15]; however, it is unable to inactivate NCX1 [38]. We therefore engineered a cell line expressing Tet-inducible uninactivatable NCX1 (Tet-K229Q). Importantly, uninactivatable NCX1 (Tet-K229Q) was equally palmitoylated compared to wild type NCX1 (Fig. 2B), but NCX1   current in Tet-K229Q cells was not regulated by insulin (Fig. 2C, D). Overall, our data demonstrate that insulin plays a modulatory role in NCX1 activation/inactivation by manipulating its palmitoylation.

zDHHC5 participates in the insulin-mediated NCX1 palmitoylation
Covalent attachment of palmitate to protein via a labile thioester bond is catalysed by a family of protein acyltransferases. Our results suggest that insulin delivers an immediate impact on NCX1 palmitoylation through acyltransferases. NCX1 is a substrate for multiple acyltransferases; plasma membrane, Golgi and Endoplasmic Reticulum (ER) localized zDHHC-PAT(s) [14]. Since the plasma membrane is the first compartment that is exposed to insulin, we began our search within plasma membrane resident zDHHC-PAT(s); zDHHC5, zDHHC20 and zDHHC21. Amongst these three palmitoylating enzymes, it is established that zDHHC5 palmitoylates NCX1 (Fig. 3A) [14,15]. We first examined NCX1-NCX1 FRET in zDHHC5-KO cells in response to insulin. FRET signals between NCX1 dimers expressed in zDHHC5-KO cells exposed to insulin were not significantly different from untreated KO cells (Fig. 3B, C). Furthermore, NCX1 palmitoylation in zDHHC5-KO cells was also unaffected by insulin, but overexpression of HA-tagged zDHHC5 in zDHHC5-KO cells noticeably rescued upregulation of NCX1 palmitoylation by insulin (Fig. 3D, E), indicating that zDHHC5 contributes to insulin-mediated palmitoylation of NCX1. We also noted that insulin treatment elevated palmitoylation of HA-tagged zDHHC5 overexpressed in zDHHC5-KO cells (Fig. 3F). To extend our findings, we probed the palmitoylation of the endogenous zDHHC5 from NRVM (Fig. 3G), ARbM (Fig. 3H) and Tet-NCX1 (Fig. 3I) cells in the absence and presence of insulin. An increase in palmitoylation of endogenous zDHHC5 was observed in three different cell types when they were treated with insulin.
insulin promotes zDHHC5 palmitoylation, we overexpressed either HA-tagged zDHHC5 or catalytically inactive zDHHS5 (C134S) in HEK293 cells, and measured the palmitoylation levels of the HA tagged zDHHC5 and zDHHS5 in the presence or absence of insulin (Fig. 4A). Palmitoylation of zDHHC5 but not zDHHS5 was upregulated upon insulin treatment (Fig. 4B), meaning that the zDHHC5 catalytic cysteine becomes palmitoylated following insulin treatment.
Since the insulin-induced increase in the palmitoylation of zDHHC5 was completely eliminated when it is catalytically inactive, we rule out that insulin regulates zDHHC5 palmitoylation by zDHHC20. As insulin regulates numerous cellular signaling pathways; particularly the activation of the fatty acid (FA) synthesis, we hypothesized that insulin activates FA synthesis that facilitates zDHHC5 palmitoylation by increasing the supply of fatty acyl-CoA (FA-CoA) to the active site cysteine of the enzyme. To examine this further, we pre-incubated NRVMs with 3 µM Triacsin C; a long FA-CoA synthetase inhibitor, for 2 h to prevent FA-CoA synthesis evoked by insulin treatment. Thereafter we probed and quantified both NCX1 and zDHHC5 palmitoylation in the absence and presence of insulin (Fig. 5A). Disrupting FA-CoA synthesis with Triacsin C inhibited the effect of insulin on the palmitoylation of the exchanger and its palmitoylating enzyme; zDHHC5 (Fig. 5B). We confirmed that other insulin signaling pathways remained intact following Triacsin C incubation by checking Akt phosphorylation (Fig. 5C), confirming the specificity of the inhibitor on FA synthesis.

Discussion
The effect of insulin on NCX1 activity and its consequences on Ca2+ handling in cardiac tissue has been shown by others in different experimental settings [25,49,50], however, the underlying mechanism was obscure. Here we bring the first mechanistic insight into the regulatory role for insulin in NCX1 physiology. Our findings introduce a three-step signaling model: (1) insulin stimulates FA-CoA synthesis that elevates the level of free acyl-CoAs. (2) This promotes FA loading to zDHHC5 at the catalytic cysteine. (3) Increased palmitate loading of the zDHHC5 active site triggers NCX1 palmitoylation which tunes its structure and physiology.

Insulin mediated palmitoylation as a regulatory process of NCX1 inactivation
Given that the intracellular loop of the exchanger is the main control unit for its function, how this structural component governs the exchanger function has been always of interest. With the help of intermolecular FRET strategy, we first noted that insulin alters FRET between exchanger proteins tagged with either CFP or YFP. Palmitoylation and Ca2+ dependent movements within the intracellular loop are established to modify NCX1-NCX1 FRET [15,32].
We therefore probed NCX1 palmitoylation, and found this to be increased upon insulin treatment. Further experiments on palmitoylatable-and unpalmitoylatable-NCX1 using whole-cell patch clamp showed that insulin is unable to regulate NCX1 current when its palmitoylation is abolished; providing direct evidence for insulin mediated palmitoylation and its regulatory role in NCX1 activity. We highlight that the functional impact of insulin on NCX1 activity appears relatively modest at resting membrane potential, when NCX1-mediated transsarcolemmal Ca fluxes are small. During systole in cardiac muscle (NCX1 acts in Ca influx mode at the start of the action potential and in Ca efflux mode later in the action potential) membrane potential and prevailing ion gradients favor much greater NCX1-mediated Ca fluxes, making the NCX1-mediated impact of insulin on Ca handling and contractility substantial [49,50].
Previously, global profiling of palmitoylated proteins in endothelial cells using palmitoyl-proteomics and SILAC labeling combined with acyl-biotin exchange, identified ~380 putative palmitoylated proteins. Palmitoylation of less than 10% of these (several of which have a role in controlling vascular function) was changed in the presence of insulin [52]. Our results expand the concept of insulin mediated palmitoylation to cardiac tissue, as well. We suggest that there are likely other palmitoylated insulin targets than NCX1 in the heart. The next question is "what is the physiological property of NCX1 being tuned by insulin?". It is established that palmitoylation alters the inactivation of the exchanger by modifying the interaction of autoinhibitory XIP domain with its docking site (residues 709-728aa near the palmitoylation site), which controls intracellular Ca2+ [15];. This being said, in whole-cell patch clamp experiments, NCX1 current remained unchanged following insulin pre-incubation when the exchanger is unable to inactivate due to unfunctional XIP domain (K229Q). Overall, in the light of these findings, it is evident that insulin-induced palmitoylation triggers local conformational changes within the intracellular loop that affects inactivation of the exchanger.
Supporting our findings herein, previous work by Villa-Abrille and colleagues reported a requirement for XIP and PIP2 in insulin dependent regulation of NCX1 current [49]. Palmitoylation is the only known cellular signaling process which ultimately modifies XIP dependent inactivation of the exchanger [60] [15]. The paradox that insulin receptor activation (which mobilises PIP2) stimulates an ion transporter that is activated by PIP2 has been noted by others [49] and is not resolved in this investigation. Indeed, palmitoylation restructures the NCX1 intracellular loop to enhance its sensitivity to XIP [15], yet the net result of insulin-induced NCX1 palmitoylation is enhanced NCX1 activity. While we don't rule out the involvement of other signaling molecules to modulate the activity of XIP domain in the presence of insulin, our data clearly indicate that palmitoylation is required for insulin to change NCX1 function by modifying its XIP dependent inactivation.

zDHHC5 and FA-CoA synthesis vs insulin
zDHHC5 plays a vital role in the cardiac tissue where it is abundantly expressed [63], and this unique palmitoyl-transferase is the only zDHHC-PAT established to palmitoylate the exchanger [15]. In this study we identified that the effect of insulin on NCX1-NCX1 FRET and the palmitoylation of the exchanger relies on zDHHC5. Furthermore, zDHHC5 itself becomes more palmitoylated at its catalytic cysteine upon insulin treatment. By blocking long FA-CoA synthetase activity with Triacsin C, we determined that the FA-CoA biosynthesis evoked by insulin, leads to enhanced palmitoylation of the zDHHC5 active site.
Despite the considerable importance of zDHHC5 in various biological events, to date, our knowledge on how its enzymatic activity is regulated is limited. Protein palmitoylation occurs in a two-step mechanism; (1) auto-palmitoylation of zDHHC-PATs; which is "the intermediate form", and (2) transfer of palmitate to proteins [11,27,43]. zDHHC5 contains three palmitoylation sites. It is auto-palmitoylated at the catalytic site (C134) between TM2 and TM3 and at the start of the C-tail by zDHHC20, to regulate substrate recruitment [62]. Hao and colleagues demonstrated that phosphorylation of zDHHC5 triggered by oleate during FA uptake in 3T3-L1 adipocytes inactivates zDHHC5 [16]. Taken together, our proposed mechanism is that insulin activates zDHHC5 by auto-palmitoylation and this, next, enhances NCX1 palmitoylation; which leads to changes in structure and function of the exchanger.
Over a decade ago, FA-CoAs were reported as endogenous activators of NCX1 activity. Longer chain saturated acyl moieties were determined as the most effective activators [45]. It is also known that zDHHC5 activity is partially controlled by acyl-CoA availability [21]. Consisted with this, in our experiments, abolishing acyl-CoA synthesis inhibited insulin-dependent zDHHC5, thus NCX1 palmitoylation. Notably, Plain and colleagues found that long-chain acyl-CoA synthetase (ACSL) isoforms are in close proximity to zDHHC5 using proximity biotinylation of zDHHC5 interacting proteins [62]. Hence we suggest that their presence in proximity to zDHHC5, allows ACSL isoforms to 'feed' substrate to the enzyme. In such a scenario the presence of an ACSL would specifically render zDHHC5 sensitive to insulin through insulin's ability to control FA-CoA synthesis locally. This is consistent with the concept that FA-CoA availability and zDHHC5 autopalmitoylation are rate limiting steps in the activity of the enzyme, and hence that the activity of a proximal ACSL isoform controls zDHHC5 activity. Whether FA-CoA availability controls the activity of other zDHHC-PATs likely depends on the affinity of these PATs for this FA-CoA. This, and the structural and/or sequence determinants of FA-CoA interaction with the zDHHC-PATs merit further investigation.

Ethics
Primary ventricular myocytes from rats and rabbits were utilized in this investigation. The protocols dealing with animals were approved by the University of Glasgow Animal Welfare and Ethics Review Board. Cardiac tissue samples from neonatal rats were collected postmortem after sacrificing animals using a method designated Schedule 1 by the Animals (Scientific Procedures) Act 1986. Rabbit hearts were excised from terminally anaesthetized, heparin-treated animals under the authority of a Project License granted by the UK Home Office.

Drugs and reagents
Porcine insulin was purchased from Sigma, and Triacsin C from Streptomyces sp was obtained from Sigma (T4540).

Plasmids and tetracycline (Tet-) inducible engineered cell lines
The experiments in this study were performed in HEK293 cells, zDHHC5 KO cells and HEK293 derived Tet-inducible FT293 cells expressing wild-type (WT)-NCX1, K229Q-NCX1 and C739A-NCX1. Tetinducible cells were engineered using the Invitrogen Flip-In T-Rex System as described elsewhere [60] [15].  5. Insulin stimulates zDHHC5 palmitoylation via FA synthesis. A Example blots showing endogenous NCX1, zDHHC5 and FLOT2 in NRVMs in following conditions: no treatment, Triacsin C and Triacsin C+insulin, B Triacsin C prevented insulin effect on both NCX1 (NCX1: n:6 for no treatment, n:7 for Triacsin C and n:9 for Triacsin C+insulin; zDHHC5: n:6 for no treatment, n:7 for Triacsin C and n:9 for Triacsin C+insulin) C Example western blots showing p-Akt (left) and total-Akt (center) and quantification (right) upon treatment of NRVMs with Triacsin C and insulin, confirming no gross impact of Triacsin C on Akt phosphorylation; a component of insulin-signaling cascade (n:6 for no treatment, n:7 for Triacsin C and n:9 for Triacsin C+insulin, p value: ****<0.0001; unpaired ttest).
We used the following plasmids in this investigation: HA-tagged zDHHC5 and zDHHS5 [62], NCX1-FRET sensors; full length canine NCX1 tagged with either CFP or YFP at position 266 [32]. FRET plasmids were a kind gift from Prof Michela Ottolia (UCLA, USA). Using Quickchange Lightning Site-Directed Mutagenesis kit (Agilent), K229Q and C739A mutations were introduced to both NCX1-FRET sensors. Lipofectamine2000 (Invitrogen) for zDHHC5-KO cells and Lipofectamine LTX (Invitrogen) for neonatal myocytes was used for plasmid transfection according to the manufacturer's instructions.

FRET imaging
Following overnight expression of NCX1 FRET sensors in NRVMs and zDHHC5-KO cells, FRET experiments were performed in Tyrode's buffer (120 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1.8 mM CaCl2 and 10 mM HEPES; pH:7.4) at room temperature [15]. In short, FRET activity between NCX1 pairs was monitored by an inverted camera; Olympus 1 × 71, with PlanApon, 60X, NA 1.42 oil immersion objective, 0.17/FN 26.5 (Olympus, UK), and recorded by a CCD camera (cool SNAP HQ Monochrome, Photometrics) and a beam splitter optical device (Dual-channel simultaneous imaging system, DV2mag biosystem (ET-04-EM)). Image acquisition and analysis were done in MetaFluor 7.1 (Meta Imaging System). FRET ratio was measured as the changes in the background subtracted 480/545 nm fluorescent emission intensity on excitation at 430 nm.

Electrophysiology
NCX1 current was measured in HEK293 derived Tet-inducible FT293 cells; Tet-NCX1, Tet-K229Q and Tet-C739A, using whole cell patch clamp experiments [36,49]. Whole cell patch clamp experiments were performed at room temperature, in an extracellular buffer containing 128 mM NaCl, 10 mM CsCl, 1 mM CaCl2, 1 mM MgCl2,10 mM Na-HEPES, pH 7.4 with CsOH. Patch pipettes were pulled from borosilicate glass capillaries and had a tip resistance of 2.5-4 MΩ. Pipette solution contained 120 mM CsCl, 3 mM CaCl2, 0.5 mM MgCl2, 20 mM Na-HEPES, 5 mM Mg-ATP, 5 mM K-BAPTA, pH 7.25, with 200 nM free Ca2+ calculated by MaxChelator. To measure NCX1 current, a standard voltage-step protocol was applied. Cells were held at − 80 mV (the calculated NCX1 reversal potential based on these intracellular and extracellular solutions is − 68.7 mV) before stepping to the test potential between − 120 mV and +100 mV (in 20 mV increments) for 300 ms at 0.5 Hz. Whole cell patch clamp recordings from Tet-NCX1 cells in which NCX1 expression was not induced revealed an insulin-insensitive background cation current with a reversal potential of approximately − 70 mV, very close to the calculated potential for cesium (− 63 mV) based on the intracellular and extracellular solutions used. Recordings were made using Axopatch200B amplifier with Digidata 1440A interface (Axon Instruments).

Western blotting
This investigation used antibodies raised against Flotillin-2 (BD Biosciences, 610,384), the HA epitope tag (Sigma, clone 3F10), NCX1 (Swant, R3F1), zDHHC5 (Sigma, HPA014670), Phospho-Akt (Ser473, Cell Signaling Technology, clone 193H12) and total-Akt (Cell Signaling Technology, clone 40D4). Images of western blots were acquired using a ChemiDoc-XRS imaging system (BioRad) and band intensities measured using Quantity One (BioRad). Palmitoylation levels of target proteins were calculated as the band intensity in the palmitoylated fraction (Palm) purified by acyl-RAC, relative to the band intensity in the corresponding unfractionated (UF) cell lysate. To account for day-to-day variations in palmitoylation stoichiometry, individual data points were normalized to the experimental mean for that experimental day. All palmitoylation assays analysed the capture of the constitutively palmitoylated protein flotillin 2 as a control for the integrity of the assay. However, no quantitative data regarding the degree of enrichment of flotillin 2 in the assay was used when calculating the relative palmitoylation of proteins of interest.

Statistics
All data herein are presented as mean±standard error of the mean (SEM). Quantitative differences between groups were evaluated using unpaired t-test, one-way or two-way ANOVA and multiple comparison tests where applicable. The values of p and n are provided in individual figure legends.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.