APT1-Mediated Depalmitoylation Regulates Hippocampal Synaptic Plasticity

Abstract

Palmitoylation may be relevant to the processes of learning and memory, and even disorders, such as post-traumatic stress disorder and aging-related cognitive decline. However, underlying mechanisms of palmitoylation in these processes remain unclear. Herein, we used acyl-biotin exchange, coimmunoprecipitation and biotinylation assays, and behavioral and electrophysiological methods, to explore whether palmitoylation is required for hippocampal synaptic transmission and fear memory formation, and involved in functional modification of synaptic proteins, such as postsynapse density-95 (PSD-95) and glutamate receptors, and detected if depalmitoylation by specific enzymes has influence on glutamatergic synaptic plasticity. Our results showed that global palmitoylation level, palmitoylation of PSD-95 and glutamate receptors, postsynapse density localization of PSD-95, surface expression of AMPARs, and synaptic strength of cultured hippocampal neurons were all enhanced by TTX pretreatment, and these can be reversed by inhibition of palmitoylation with palmitoyl acyl transferases inhibitors, 2-bromopalmitate and N-(tert-butyl) hydroxylamine hydrochloride. Importantly, we also found that acyl-protein thioesterase 1 (APT1)-mediated depalmitoylation is involved in palmitoylation of PSD-95 and glutamatergic synaptic transmission. Knockdown of APT1, not protein palmitoyl thioesterase 1, with shRNA, or selective inhibition, significantly increased AMPAR-mediated synaptic strength, palmitoylation levels, and synaptic or surface expression of PSD-95 and AMPARs. Results from hippocampal tissues and fear-conditioned rats showed that palmitoylation is required for synaptic strengthening and fear memory formation. These results suggest that palmitoylation and APT1-mediated depalmitoylation have critical effects on the regulation of glutamatergic synaptic plasticity, and it may serve as a potential target for learning and memory-associated disorders.

SIGNIFICANCE STATEMENT Fear-related anxiety disorders, including post-traumatic stress disorder, are prevalent psychiatric conditions, and fear memory is associated with hyperexcitability in the hippocampal CA1 region. Palmitoylation is involved in learning and memory, but mechanisms coupling palmitoylation with fear memory acquisition remain poorly understood. This study demonstrated that palmitoylation is essential for postsynapse density-95 clustering and hippocampal glutamatergic synaptic transmission, and APT1-mediated depalmitoylation plays critical roles in the regulation of synaptic plasticity. Our study revealed that molecular mechanism about downregulation of APT1 leads to enhancement of AMPAR-mediated synaptic transmission, and that palmitoylation cycling is implicated in fear conditioning-induced synaptic strengthening and fear memory formation.

Introduction

Palmitoylation is one of the reversible lipid modifications, S-acylations by introducing lipids to protein cysteine residues, via post-translational addition of saturated 16-carbon palmitic acid to cysteine residue through formation of a labile thioester linkage (Dietrich and Ungermann, 2004; Resh, 2006). Intracellular palmitoylation reactions are mediated by a family of palmitoyl acyl transferases (PATs), which have a common aspartate-histidine-histidine-cysteine domain (DHHC proteins), such as ZDHHC2, ZDHHC9, and ZDHHC15 (Greaves and Chamberlain, 2011; Abrami et al., 2017; Shah et al., 2019; Shimell et al., 2019; Jin et al., 2021). Palmitoylation reversibly modifies numerous of proteins, such as neurotransmitter receptors, synaptic scaffolding proteins, and signaling molecules, and also regulates the stability, trafficking, localization, and signaling pathways of these target proteins (El-Husseini and Bredt, 2002). In several studies, palmitoylation has been shown to regulate neuronal protein stability, sorting, and trafficking, and subsequently alter protein synaptic targeting and functional efficiency (Huang and El-Husseini, 2005; Greaves and Chamberlain, 2007). There are increasing amounts of evidence showing that palmitoylation might be relevant to the processes of disorders, such as X-linked intellectual disability and 22q11.2 deletion syndrome (Fukata and Fukata, 2010; Moutin et al., 2017; Shimell et al., 2019). KO PATs contribute to a deficiency in prepulse inhibition, and the KO mice display a decrease in exploratory activity in a new environment (Mukai et al., 2004). Emerging evidence has also revealed that palmitoylation may be involved in learning and memory, although little has been known about the mechanisms coupling them.

Regulation of glutamate receptors is essential and basal for glutamatergic synaptic transmission, and this is implicated in many (patho)physiological processes, such as learning and fear memory (Kim and Cho, 2017). However, mechanisms underlying synaptic plasticity and synapse protein modification need to be further explored. Postsynapse density-95 (PSD-95) is a neuronal scaffolding protein that is widely studied in the field of synaptic plasticity (Xu, 2011), and it regulates and maintains the strength and plasticity of glutamatergic synapse function and spine morphology by associating with the membrane receptors and ion channels, such as AMPARs and NMDARs (Elias and Nicoll, 2007; Vallejo et al., 2017). Accumulated studies suggest that the dynamics of PSD-95 are closely related to the expression of LTP, an important form of synaptic plasticity (Ehrlich and Malinow, 2004; Ehrlich et al., 2007). Moreover, the number of AMPAR slots and the level of PSD-95 in the PSD are upregulated by increased palmitoylation level of PSD-95 via depalmitoylation enzyme blockade (Jeyifous et al., 2016). However, direct evidence about whether changes in the palmitoylation state of PSD-95 mediate synaptic changes is lacking.

Depalmitoylation, mainly mediated by two kinds of depalmitoylating enzymes, acyl protein thioesterase (APT) and protein palmitoyl thioesterase (PPT) (Bachovchin et al., 2010; Lin and Conibear, 2015a), plays a pivotal role in the recycling and/or degradation of proteins that undergo palmitoylation modification (Greaves and Chamberlain, 2007; Salaun et al., 2010). As is the same with PATs, abnormality of depalmitoylating enzymes also contributes to many CNS diseases, such as infantile neuronal ceroid lipofuscinosis, caused by protein palmitoyl thioesterase 1 (PPT1) deficiency, which impairs the cleavage of thioester linkage in palmitoylated proteins and prevents the degradation and leads to accumulation of lysosomal ceroid (Kim et al., 2008; Sarkar et al., 2013). However, the regulatory mechanism and effect of (de)palmitoylation in glutamatergic synaptic transmission which is essential for learning and memory is still far from fully known.

Fear-related anxiety disorders, such as social phobia, post-traumatic stress disorder, and panic disorder, are the most prevalent psychiatric conditions (Amstadter et al., 2009; Carleton et al., 2009; Ege and Reinholdt-Dunne, 2016). Although post-traumatic stress disorder exhibits multiple molecular phenotypes (Johnson et al., 2012; Liberzon, 2018), limited understanding of the pathophysiology at cellular and molecular levels has been a major setback in developing effective therapeutic interventions. Fear induces persistent strengthening in glutamatergic synaptic transmission, and this occludes further transmission potentiation, which may hinder the capacity for new information storage (Kim and Cho, 2017). However, little is known about neuronal synaptic modifications underlying fear memory formation, and the mechanism of palmitoylation in learning, particularly the acquisition of fear memory, still needs fine research.

In this study, we first explored whether palmitoylation is necessary for glutamatergic synaptic transmission at a network level, and then detected whether homeostatic scaling (a form of in vitro synaptic plasticity with a compensatory mechanism that enables neurons to adjust the synaptic strength in response to persistent changes in the network activity) affects the palmitoylation levels and function of several important synaptic proteins. We examined whether synaptic strength is influenced by the regulation of depalmitoylation, and comparatively studied and determined which depalmitoylating enzyme is predominant in the palmitoylation manipulation of glutamate receptors and PSD-95 via silencing with shRNA and specific inhibition. We also tested on hippocampal slices of fear-conditioned rats to confirm whether (de)palmitoylation bears a part in fear conditioning-induced synaptic transmission enhancement. We found that palmitoylation is required for PSD-95 clustering and hippocampal glutamatergic synaptic transmission, and APT1-mediated depalmitoylation plays important role in the regulation of synaptic plasticity.

Materials and Methods

Agents

PAT inhibitors N-(tert-butyl) hydroxylamine hydrochloride (NtBuHA) and 2-bromopalmitic acid (2-BP), hydroxylamine (HA) hydrochloride, TTX, S-methyl methanethiosulfonate (MMTS), neocuproine, APT1 specific inhibitor ML348, and streptavidin-agarose were purchased from Sigma-Aldrich. N-(6-(Biotinamido) hexyl)-3′-(2′-pyridyldithio)-propionamide (HPDP)-biotin was supplied by Thermo Fisher Scientific. Green fluorescent protein (GFP)-tagging lentiviral vectors (LVs) expressing specific short hairpin RNAs (shRNAs) for APT1 and PPT1 knockdown were obtained from Shanghai GeneChem. Other agents were obtained from commercial suppliers.

Experimental animals

Male Sprague Dawley rats (200-250 g, age 7-8 weeks) were provided by Shanghai Laboratory Animal Center, Chinese Academy of Sciences. All animals were housed individually under a controlled 12 h light-dark cycle at constant temperature (22 ± 2°C) and relative humidity (60%-70%) with laboratory standard food and water available ad libitum. The use of animals for experimental procedures was conducted in accordance with the Guide for the care and use of laboratory animals as adopted and promulgated by the National Institutes of Health. All experimental procedures were approved by the Animal Welfare Committee of Fujian Medical University. Fear conditioning was conducted over 3 d, habituation day (day 1), fear conditioning day (day 2), and testing day (day 3). On day 1, rats were exposed to the chamber for 5 min. On day 2, after 3 min habituation in the chamber, rats were presented with six presentations of tone conditional stimulus (CS, 65 dB, and 30 s), each coterminated with a 0.75 mA foot shock lasting 1 s as unconditional stimulus (US), and the interval time between presentations was 30 s. After training, rats were removed from the chamber to the home cage immediately. On day 3, rats were placed into the chamber and observed for 5 min. The percentage of freezing time was measured during tone presentation, which was recorded by software of ANY-maze (Stoelting).

Stereotactic surgery and microinjection

Cannulas were implanted into both sides of hippocampus of rats. The rats were anesthetized with pentobarbital sodium (40 mg/kg, i.p.) and placed on a stereotaxic apparatus (Stoelting) with the bregma and posterior on the same level. The body temperature was maintained at 37.0°C by an electric incandescent lamp. A small hole was drilled into the bone to insert a stainless guide cannula (length 10 mm, OD 0.6 mm) into the hippocampal CA1 region (3.5 mm posterior to the bregma, 2.5 mm lateral to the midline, 2.0 mm vertical from the cranial theca). The cannula was fixed to the skull with the aid of dental acrylic resin and dental cement. Stainless-steel obturators (OD 0.35 mm) were placed into the guide cannula to prevent obstruction. Rats were individually housed and allowed to recover for at least 7 d after surgery. As for drug injections, for example, 2-BP (0.5 mm) was infused 60 min before testing into each of the sites at a dose of 1 nmol/site, with a microsyringe (5 μl) connected by a PE-10 polyethylene tubing to a needle (OD 0.35 mm). The injection volume was set 2 μl within a period of 2 min. Infusion sites were identified based on The rat brain in stereotaxic coordinates(Paxinos and Watson, 1997), and only data from rats with correct implants were included for analysis.

Hippocampal neuron cultures

Hippocampal neuron cultures were performed as described previously (Ming et al., 2006). Neonatal Sprague Dawley rats (day 0-3) of both sexes were obtained from the Experimental Animal Center of Fujian Medical University. Hippocampus of newborn rats was dissected and subsequently digested in PBS containing 0.125% trypsin (Amresco) at 37°C for 30 min. Hippocampal tissues were mechanically triturated by repeated passages through a 5 ml pipette. Neurons were collected by centrifugation (300 × g for 6 min at room temperature) and resuspended in DMEM and F-12 supplement (1:1) (Invitrogen) with 5% FBS (Invitrogen), 2 mm L-glutamine (Sigma), and penicillin (100 U/ml)-streptomycin (100 U/ml) for 24 h. Subsequently, the culture medium was changed to Neurobasal medium (Invitrogen) supplemented with 2% B-27 (Invitrogen) and 2 mm glutamine (Invitrogen). Astrocytes were minimized by treating the culture with cytarabine (10 μm) on day 3. The medium was replaced with fresh medium every 2 d.

Surface biotinylation assay

Surface biotinylation assay, Western blotting, acyl-biotin-exchange (ABE) assay, coimmunoprecipitation, and PSD purification procedures were similar with our previous reports (Mao et al., 2009; Fan et al., 2019) with minor modifications. Samples were collected into Eppendorf tubes with ice-cold aCSF (in mm as follows): 119.0 NaCl, 1.3 MgSO₄, 3.5 KCl, 1.0 NaH₂PO₄, 26.2 NaHCO₃, 2.5 CaCl₂, and 11.0 glucose, adjusted to pH 7.4 and 300 mOsm, containing EZ-link Sulfo-NHS-LC Biotin (1 mg/ml) (Pierce Biotechnology) for 2 h at 4°C with gentle shaking, and the reaction was terminated by quenching with 100 mm glycine on ice for 30 min, and then washed with cold aCSF for 3 times. Finally, it was homogenized with 10 μl/mg radio-immunoprecipitation assay (RIPA) buffer (50 mm Tris-HCl, pH 7.4, 1% NP-40, 0.5% Na-deoxycholate, 0.1% SDS, 150 mm NaCl, 2 mm EDTA, and 50 mm NaF), including a cocktail of protease and phosphatase inhibitors (Sigma-Aldrich), and then centrifuged at 12,000 × g for 15 min at 4°C. Protein concentration of the supernatants was measured via Coomassie Blue protein-binding assay (Nanjing Jiancheng Institute of Biological Engineering). Biotinylated protein (200-300 μg) was incubated with streptavidin (Pierce Biotechnology) for 4 h at 4°C. Streptavidin-protein complexes were washed 3 times with RIPA buffer and centrifuged at 4000 × g for 3 min at 4°C. Bound proteins were separated from the beads and denatured by SDS-loading buffer at 95°C for 10 min.

ABE assay

The cultured cells were lysed with ice-cold RIPA buffer containing protease inhibitor and homogenized by sonication for 10 s. After static settlement for 30 min on ice, centrifuged at 12,000 × g for 20 min at 4°C, the supernatant fraction was obtained. Four folds of volume of blocking buffer (1 vol 25% SDS solution and 9 vol of HEN buffer [250 mm HEPES-NaOH, pH 7.7, 1 mm EDTA and 1 mm neocuproine] adjusted to 20 mm MMTS) was mixed with all samples and homogeneously mixed, shaken gently at 50°C, and protected from light for 30 min to block free thiols. Four times volume of acetone was mixed with the samples and incubated at −20°C for 30 min and centrifuged to remove the residual MMTS at 12,000 × g for 20 min, and the precipitates were collected and resuspended in RIPA buffer again. The process was repeated twice. The RIPA buffer which resuspended the last precipitates was divided into two portions. One portion was incubated with RIPA buffer containing only HPDP-biotin to serve as controls; another portion was incubated in 4× volume of RIPA buffer with 0.7 mm HA to cleave thioester bonds and 1 mm HPDP-biotin to link biotin to newly exposed cysteine thiols and used to detect the palmitoylated protein level. All samples were placed in darkness with frequent vortexing at room temperature for 2 h. Acetone precipitates were used to remove the remaining HPDP-biotin and HA. The precipitates were resuspended in RIPA buffer. Coomassie Blue protein-binding assay was used, and global palmitoylation was detected by HRP-streptavidin with Western blotting steps. Protein samples after ABE were divided into two portions. One portion was directly added into SDS-PAGE sample loading buffer to serve as “input,” and another portion was purified with the streptavidin-agarose (Sigma-Aldrich). After this portion of sample was repeatedly purified 10 times, RIPA buffer was used to wash the streptavidin-agarose 3 times to affirm the reaction was completed. Then the biotinylated proteins were eluted and stored at −80°C after denatured by sample loading buffer at 95°C for 30 min. The palmitoylated protein samples were normalized to the input proteins (total proteins), which were internally controlled by β-actin.

Western blotting

The sample preparation and determination processes were the same as above. After denatured at 95°C in SDS-loading buffer for 5 min, the lysate samples were loaded in each line and separated via 10% SDS-PAGE, and transferred to nitrocellulose membranes (transfer buffer: 25 mm Tris, 190 mm glycine, 20% methanol, 0.5% SDS). The membranes were incubated in blocking buffer (5%, w/v, BSA in TBS, pH 7.4, and 0.5% Tween-20, TBST) at room temperature for 1 h before incubation in the primary antibody solution overnight at 4°C. Membranes were washed in TBST solution 3 times for 10 min and incubated in HRP-conjugated secondary antibodies solution for 1 h at room temperature, then were rinsed with double-deionized water, and immersed in enhanced chemiluminescence detecting substrate (Super Signal West Pico; Pierce Chemical) and captured by the Micro Chemiimaging system (DNR Bio-imaging systems). The optical densities of the immunoblots were measured by National Institutes of Health's ImageJ software. All results were presented as percentage of control following normalization. Detailed information about antibodies used is shown in Extended Data Figure 1-1. Some original images of Western blots for representative figures are also shown in Extended Data Figures 1-2, 2-1, 6-1, 7-1, 8-1, and 9-1.

Coimmunoprecipitation

Collected cells were lysed with 0.2 ml of extraction buffer (in mm) as follows: 150 NaCl, 0.2% NP-40, 50 NaF, 1 Na₃VO₄, 6 Na-deoxycholate, 3 Na-pyrophosphate, 1% cocktail of protease inhibitor, 1% phosphatase inhibitor; Roche), and homogenized on ice for 30 min, centrifuged at 12,000 × g for 15 min at 4°C. Subsequently, supernatants (500 μg) were incubated with 2 μg of stargazin or PSD-95 antibody overnight at 4°C with gentle shaking. Antibody-antigen complex beads were precipitated by incubation of the sample with 50 μl settled Protein A/G PLUS-Agarose (Santa Cruz Biotechnology) for 3-4 h at 4°C. The beads were collected by centrifugation for 3 min at 4000 × g and washed 3 times with washing buffer (10 mm HEPES, pH 7.5, 100 mm NaCl, 1 mm EDTA, 10% glycerol, and 0.1% Triton X-100). Finally, proteins were released from the beads with 50 μl of SDS-loading buffer, and then boiled at 95°C for 5 min before the eluents were analyzed by Western blotting steps.

Electrophysiological methods

Brain slice preparation, whole-cell patch-clamp recording, and field potential recording were performed as described previously (Li et al., 2018). After 1.5 h recovery, one slice was placed in the recording chamber where it was continuously perfused with oxygenated aCSF at the rate of 2 ml/min. For AMPAR-mediated miniature EPSCs (mEPSCs), recordings of neurons were made in a submersion chamber with patch electrodes (3-5 mΩ resistance) filled with a solution containing (in mm) as follows: 122.5 Cs-gluconate, 17.5 CsCl, 0.2 EGTA, 10.0 HEPES, 1.0 MgCl₂, 4.0 Mg-ATP, 0.3 Na-GTP, and 5.0 QX-314 (adjusted to pH 7.2 with CsOH, 290-320 mOsm). Whole-cell patch-clamp recording was performed at room temperature using a Muticlamp700B amplifier (Molecular Devices), filtered at 3 kHz, amplified 5 times, and then digitized at 20 kHz with a Digidata1550A digitizer and Clampex 10 software (Molecular Devices). To isolate mEPSCs mediated by AMPARs, recordings were performed in the presence of bicuculline (20 μm) and TTX (1 μm) in the bath solution. All neurons were clamped at −70 mV, and experiments with >20% changes in series resistance or with series resistance >15 mΩ, as monitored with a 10 mV pulse, were excluded. The frequency and amplitude of mEPSCs were analyzed by Mini Analysis Program (Synaptosoft). Field EPSPs (fEPSPs) in Schaffer collateral-CA1 pathway in response to high-frequency stimulus (HFS, consisting of three trains of 100 pulses at 100 Hz separated by 30 s and delivered at test intensity) were recorded using 3 m NaCl-filled glass electrodes (3-5 mΩ) placed within the CA1 region.

Statistical analysis

All analyses were performed using SPSS 18.0 software, and data are presented as mean ± SEM. All statistical results and tests used are included in the figure legends, and detailed statistical analysis results are shown in Table 1. Required sample size was estimated based on prior experience. Data were comparatively evaluated by independent samples two-tailed t test, one-way ANOVA, or two-way ANOVA with least significant difference post hoc tests, as appropriate. Differences at p value <0.05 were considered statistically significant.

Table 1.

Statistical analyses in each figure

Figure	Group	n	Mean ± SEM	F or t	p	Test
1 D	Ctrl TTX TTX + 2-BP	12 12 13	12.8 ± 0.96 18.8 ± 1.585 13.9 ± 1.538	F(2,35) = 5.962	TTX vs Ctrl, 0.0025 TTX + 2-BP vs TTX, 0.0379	One-way ANOVA
1 E	Ctrl TTX TTX + 2-BP	12 12 13	1.37 ± 0.169 1.677 ± 0.194 1.551 ± 0.209	F(2,34) = 0.623	NA	One-way ANOVA
1 H	GluA1 Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.171 1.765 ± 0.119 1.096 ± 0.119 1.092 ± 0.139	F(3,12) = 6.476	TTX vs Ctrl, 0.0106 TTX + 2-BP vs TTX, 0.0235 TTX + NtBuHA vs TTX, 0.0105	Two-way ANOVA
	GluA2 Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.009 1.599 ± 0.059 1.127 ± 0.044 0.971 ± 0.064	F(3,12) = 34.49	TTX vs Ctrl, 0.0056 TTX + 2-BP vs TTX, 0.0115 TTX + NtBuHA vs TTX, p < 0.001	Two-way ANOVA
	GluN1 Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.091 1.051 ± 0.054 1.018 ± 0.038 0.988 ± 0.059	F(3,12) = 0.297	NA	Two-way ANOVA
	GluN2A Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.038 0.963 ± 0.041 0.937 ± 0.019 1.026 ± 0.021	F(3,12) = 1.522	NA	Two-way ANOVA
	GluN2B Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.059 1.027 ± 0.059 1.058 ± 0.047 0.999 ± 0.054	F(3,12) = 0.238	NA	Two-way ANOVA
2 A	Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.128 1.894 ± 0.234 1.12 ± 0.163 1.072 ± 0.152	F(3,12) = 5.749	TTX vs Ctrl, 0.086 TTX + 2-BP vs TTX, 0.0353 TTX + NtBuHA vs TTX, 0.0261	Two-way ANOVA
2 B	Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.035 1.196 ± 0.06 0.788 ± 0.05 0.795 ± 0.008	F(3,12) = 20.1	TTX vs Ctrl, 0.0312 TTX + 2-BP vs TTX, 0.0071 TTX + NtBuHA vs TTX, p < 0.001.	Two-way ANOVA
2 C	Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.031 1.469 ± 0.031 0.916 ± 0.023 0.924 ± 0.033	F(3,12) = 70.51	TTX vs Ctrl, p < 0.001 TTX + 2-BP vs TTX, p < 0.001 TTX + NtBuHA vs TTX, p < 0.001	Two-way ANOVA
2 D	Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.042 1.408 ± 0.088 0.949 ± 0.061 0.959 ± 0.057	F(3,12) = 11.63	TTX vs Ctrl, 0.0059 TTX + 2-BP vs TTX, 0.0053 TTX + NtBuHA vs TTX, 0.0054	Two-way ANOVA
2 E	Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.034 1.545 ± 0.095 0.97 ± 0.036 0.994 ± 0.06	F(3,12) = 20.30	TTX vs Ctrl, 0.0017 TTX + 2-BP vs TTX, 0.0014 TTX + NtBuHA vs TTX, 0.0028	Two-way ANOVA
2 F	Ctrl TTX TTX + 2-BP TTX + NtBuHA	4 4 4 4	1.0 ± 0.0488 1.72 ± 0.085 1.14 ± 0.022 1.104 ± 0.038	F(3,12) = 35.84	TTX vs Ctrl, 0.0004 TTX + 2-BP vs TTX, 0.0006 TTX + NtBuHA vs TTX, 0.0006	Two-way ANOVA
3 G	Veh 2-BP	4 4	1.0 ± 0.085 0.959 ± 0.058	t = 0.385	NA	Student's t test
4 B	LV-eGFP LV-APT1 LV-PPT1	71 37 51	11.62 ± 0.364 15.76 ± 0.526 12.65 ± 0.603	F(2,155) = 16.82	LV-APT1 vs LV-eGFP, p < 0.001 LV-PPT1 vs LV-APT1, 0.004	One-way ANOVA
4 C	LV-eGFP LV-APT1 LV-PPT1	71 37 51	1.648 ± 0.096 2.292 ± 0.14 1.688 ± 0.109	F(2,155) = 8.518	LV-APT1 vs LV-eGFP, 0.002 LV-PPT1 vs LV-APT1, 0.041	One-way ANOVA
4 E	Veh ML348	39 31	10.49 ± 0.764 13.46 ± 0.694	t = 2.881	0.0075	Student's t test
4 F	Veh ML348	39 31	1.528 ± 0. 069 1.88 ± 0.095	t = 2.983	0.0046	Student's t test
4 G	Veh ML348	6 6	1.01 ± 0.128 1.627 ± 0.166	t = 2.977	0.0139	Student's t test
4 H	Veh ML348	6 6	1.479 ± 0.057 2.105 ± 0.156	t = 3.755	0.0038	Student's t test
5 B	LV-eGFP + Veh LV-APT1 + Veh LV-eGFp + 2-BP LV-APT1 + 2-BP	11 15 11 15	11.01 ± 0.74 15.08 ± 0.746 11.36 ± 0.348 11.29 ± 0.599	F(3,48) = 9.564	LV-APT1 + Veh vs LV-eGFP + Veh, 0.0009 LV-APT1 + 2-BP vs LV-APT1 + Veh, 0.005	Two-way ANOVA
5 C	LV-eGFP + Veh LV-APT1 + Veh LV-eGFp + 2-BP LV-APT1 + 2-BP	11 15 11 15	1.406 ± 0.107 2.078 ± 0.108 1.461 ± 0.13 1.357 ± 0.155	F(3,48) = 7.349	LV-APT1 + Veh vs LV-eGFP + Veh, 0.003 LV-APT1 + 2-BP vs LV-APT1 + Veh, 0.007	Two-way ANOVA
6 A	LV-eGFP LV-APT1	4 4	1.0 ± 0.135 0.305 ± 0.03	t = 4.998	0.0352	Student's t test
6 B	LV-eGFP LV-APT1	4 4	1.0 ± 0.0349 1.678 ± 0.117	t = 5.549	0.0014	Student's t test
6 C	LV-eGFP LV-APT1	4 4	1.0 ± 0.042 1.633 ± 0.097	t = 5.919	0.001	Student's t test
6 D	LV-eGFP LV-APT1	4 4	1.0 ± 0.035 1.525 ± 0.191	t = 2.691	0.0036	Student's t test
6 E	LV-eGFP LV-APT1	4 4	1.0 ± 0.0452 1.245 ± 0.065	t = 3.061	0.0222	Student's t test
6 F	LV-eGFP LV-APT1	4 4	1.0 ± 0.0328 1.445 ± 0.077	t = 5.29	0.0018	Student's t test
6 G	GluA1 LV-eGFP LV-APT1	4 4	1.0 ± 0.057 2.156 ± 0. 146	t = 7.368	0.0062	Student's t test
	GluA2 LV-eGFP LV-APT1	4 4	1.0 ± 0.085 2.201 ± 0.071	t = 10.05	p < 0.001	Student's t test
	GluN1 LV-eGFP LV-APT1	4 4	1.0 ± 0.022 0.9752 ± 0.042	t = 0.514	NA	Student's t test
	GluN2A LV-eGFP LV-APT1	4 4	1.0 ± 0.023 0.943 ± 0.032	t = 1.418	NA	Student's t test
	GluN2B LV-eGFP LV-APT1	4 4	1.0 ± 0.062 1.021 ± 0.033	t = 0.583	NA	Student's t test
7 A	Veh ML348	4 4	1.0 ± 0.1103 1.368 ± 0.078	t = 2.719	0.00347	Student's t test
7 B	Veh ML348	4 4	1.0 ± 0.151 1.525 ± 0.149	t = 2.47	0.0485	Student's t test
7 C	Veh ML348	4 4	1.0 ± 0.043 1.459 ± 0.097	t = 4.284	0.0052	Student's t test
7 D	Veh ML348	4 4	1.0 ± 0.024 1.166 ± 0.04227	t = 3.42	0.0141	Student's t test
7 E	Veh ML348	4 4	1.0 ± 0.03905 1.161 ± 0.082	t = 6.720	0.0005	Student's t test
7 F	GluA1 Veh ML348	4 4	1.0 ± 0.03 1.921 ± 0. 09	t = 9.679	p < 0.001	Student's t test
	GluA2 Veh ML348	4 4	1.0 ± 0.103 1.594 ± 0.045	t = 5.247	0.0019	Student's t test
	GluN1 Veh ML348	4 4	1.0 ± 0.022 0.975 ± 0.042	t = 0.514	NA	Student's t test
	GluN2A Veh ML348	4 4	1.0 ± 0.06 1.021 ± 0.058	t = 0.252	NA	Student's t test
	GluN2B Veh ML348	4 4	1.0 ± 0.041 0.9926 ± 0.018	t = 0.162	NA	Student's t test
8 A	LV-eGFP LV-PPT1	4 4	1.0 ± 0.098 0.324 ± 0.031	t = 6.514	0.0006	Student's t test
8 B	LV-eGFP LV-PPT1	4 4	1.0 ± 0.044 0.991 ± 0.044	t = 0.143	NA	Student's t test
8 C	LV-eGFP LV-PPT1	4 4	1.0 ± 0.087 1.092 ± 0.143	t = 0.547	NA	Student's t test
8 D	LV-eGFP LV-PPT1	4 4	1.0 ± 0.054 1.046 ± 0.057	t = 0.584	NA	Student's t test
8 E	LV-eGFP LV-PPT1	4 4	1.0 ± 0.0851 0.977 ± 0.072	t = 0.203	NA	Student's t test
8 F	LV-eGFP LV-PPT1	4 4	1.0 ± 0.105 0.95 ± 0.15	t = 0.271	NA	Student's t test
8 G	GluA1 LV-eGFP LV-PPT1	4 4	1.0 ± 0.043 1.162 ± 0.052	t = 2.36	NA	Student's t test
	GluA2 LV-eGFP LV-PPT1	4 4	1.0 ± 0.046 0.9756 ± 0.05	t = 0.354	NA	Student's t test
	GluN1 LV-eGFP LV-PPT1	4 4	1.0 ± 0.039 0.94 ± 0.03	t = 1.193	NA	Student's t test
	GluN2A LV-eGFP LV-PPT1	4 4	1.0 ± 0.034 0.954 ± 0.021	t = 1.135	NA	Student's t test
	GluN2B LV-eGFP LV-PPT1	4 4	1.0 ± 0.015 1.1 ± 0.023	t = 0.617	NA	Student's t test
9 D	Ctrl + Veh FC + Veh Ctrl + 2-BP FC + 2-BP	6 6 6 6	1.8 ± 0.057 3.358 ± 0.568 1.655 ± 0.112 1.448 ± 0.185	F(3,16) = 6.443	FC + Veh vs Ctrl + Veh, 0.04 FC + 2-BP vs FC + Veh, 0.0096	Two-way ANOVA
9 E	Ctrl + Veh FC + Veh Ctrl + 2-BP FC + 2-BP	6 6 6 6	9.838 ± 0.335 17.25 ± 1.595 12.42 ± 0.791 11.94 ± 0.51	F(3,16) = 8.933	FC + Veh vs Ctrl + Veh, 0.0062 FC + 2-BP vs FC + Veh, 0.0099	Two-way ANOVA
9 G	Ctrl + Veh Ctrl + 2-BP FC + Veh FC + 2-BP	4 4 4 4	159.7 ± 4.455 149.8 ± 2.038 175.4 ± 2.748 131.2 ± 3.027	F(3,12) = 33.69	FC + Veh vs Ctrl + Veh, 0.0244 FC + 2-BP vs FC + Veh, p < 0.001	Two-way ANOVA
9 H	FC + Veh FC + 2-BP	6 6	159.7 ± 4.455 149.8 ± 2.038	t = 4.27	0.0016	Student's t test
9 J	Ctrl + Veh Ctrl + ML348 FC + Veh FC + ML348	4 4 4 4	144.8 ± 3.281 146.4 ± 3.825 171.1 ± 4.488 168.7 ± 1.426	F(3,12) = 16.74	FC + Veh vs Ctrl + Veh, 0.0032 FC + ML348 vs Ctrl + ML348, 0.0016	Two-way ANOVA
9 K	FC + Veh FC + 2-BP	6 6	76.33 ± 4.187 89.18 ± 4.561	t = 2.075	0.06	Student's t test

Results

Palmitoylation is required for excitatory synaptic transmission in vitro

Previous researches have shown that long-term homeostatic scaling of neuronal networks regulates the palmitoylation of several synaptic proteins (Turrigiano, 2008; Brigidi et al., 2014), and thereby alters their synaptic targeting and functional efficiency (Fukata and Fukata, 2010). However, it is still unknown whether palmitoylation exerts beneficial effect on synaptic transmission at a network level. To examine this, we treated hippocampal neurons (day 14) with TTX (1 μm) to block neuronal activity for 48 h, to simulate an increase in synaptic network strength, and added palmitoylation inhibitors (day 16), 2-BP (10 μm) or NtBuHA (100 μm), together with TTX, for 8 h (as shown in Fig. 1A). ABE assay result of total biotin-labeled proteins from cultured neurons treated with TTX and 2-BP, without purification via streptavidin-agarose, showed that TTX treatment increased global palmitoylation level of total proteins, and this was reversed by 2-BP (Fig. 1B). To further verify the role of palmitoylation in synaptic transmission, we recorded mEPSCs from hippocampal neurons treated with TTX, and observed an obvious increase in the amplitude of mEPSCs, which was abolished by 2-BP (Fig. 1C,D), but without significant change in mEPSC frequency (Fig. 1E). These results show that palmitoylation plays an important role in synaptic transmission at a network level. In addition, surface biotinylation assay results showed that the membrane surface levels of AMPA-type glutamate receptor subunit GluA1-2, not NMDARs, were significantly increased through palmitoylation, which was identified using 2-BP and NtBuHA (Fig. 1F–H). These findings prove that palmitoylation is required for synaptic transmission.

Figure 1.

Synaptic scaling increases palmitoylation and surface expression of glutamatergic receptors in vitro. A, Schematic diagram of the experiment. Hippocampal neuron cultures, application of TTX (day 14-16) and 2-BP or NtBuHA (day 16), and differential testing approaches for palmitoylation. B, Western blotting analysis of streptavidin-labeled protein from cultured neurons treated with TTX and 2-BP. TTX increased global palmitoylation, which was prevented by 2-BP. C, Representative (left lane; calibration: 10 pA, 500 ms) and average (right lane) traces of mEPSC recordings from cultured neurons. D, Cumulative probabilities and average mEPSC amplitudes showing that 2-BP abolished the increase of the mEPSC amplitude induced by TTX. E, Cumulative probabilities and average mEPSC frequencies showing that TTX and 2-BP had no effect on frequency of mEPSC recording from hippocampal neuron cultures. Data are mean ± SEM. F, G, Western blots of surface-expressing GluA1-2 and NMDARs from cultured neurons (F), and Western blots of total expression of GluA1-2 and NMDARs from the same samples of the cultured neurons (G), with β-actin as internal reference. H, Quantification results of membrane surface expression (biotin-linked aggregated subunits) of glutamatergic receptors, showing that TTX increased the surface level of GluA1-2, not NMDARs, and 2-BP or NtBuHA abolished the increase. Data are mean ± SEM. One-way ANOVA (D,E); two-way ANOVA (G). n = 12, 12, 13 per group (D,E); n = 4 per group (B,G). *p < 0.05, **p < 0.01 versus Ctrl group. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 versus TTX group. For detailed statistical analyses, see Table 1. Detailed information about antibodies used for Figure 1 and other figures in this study are shown in Extended Data Figure 1-1. Original images of Western blots for Figure 1 are shown in Extended Data Figure 1-2.

Figure 1-1

Information of antibodies used in this study. Download Figure 1-1, DOCX file ^{(26.5KB, docx)} .

Figure 1-2

Full original images of Western blotting assays for Figure 1. Download Figure 1-2, TIF file ^{(7.2MB, tif)} .

Homeostatic scaling facilitates palmitoylation and function of glutamate receptors

To examine the effect of palmitoylation induced by synaptic strengthening on the synaptic proteins in hippocampal neurons, ABE assay was conducted. The ABE assay results showed that TTX induced a remarkable upregulation of the palmitoylation of PSD-95, GluA1-2, and AMPA-type glutamate receptor subunit GluN2A-B, and all of these upregulation of palmitoylation could be reversed by the inhibitors of palmitoylation, 2-BP (10 μm) or NtBuHA (100 μm) (Fig. 2A–E). Results of the coimmunoprecipitation assay also showed that TTX promoted the association between PSD-95 and stargazin via palmitoylation, and this was identified using 2-BP and NtBuHA (Fig. 2F). In order to ensure the reliability of ABE method and the conclusion, we conducted experiments with HA-free negative control groups, and examined the effect of 2-BP on the base-level palmitoylation of substrate protein PSD-95 without TTX treatment. As shown in Figure 3A–E, without HA treatment, there were no palmitoylation signals detected by the ABE method, and the results also showed that, without TTX treatment, 2-BP had no significant effect on the base-level palmitoylation of PSD-95 (Fig. 3F,G). Collectively, these results demonstrated that homeostatic scaling can facilitate the palmitoylation and function of synaptic proteins, including PSD-95 and glutamate receptors.

Figure 2.

Homeostatic scaling facilitates palmitoylation and function of glutamate receptors. A, B, Western blotting analysis showing that TTX increased PSD-95 and GluA1 palmitoylation, which was reversed by 2-BP or NtBuHA. C-E, Western blot analysis of streptavidin-labeled protein from cultured neurons treated with TTX and 2-BP. TTX significantly increased the palmitoylation of GluA2 (C), GluN2A (D), and GluN2B (E), which was reversed by NtBuHA or 2-BP. F, Cell lysates from cultured neurons were analyzed by coimmunoprecipitation using antibody against stargazin, then subjected to Western blotting analysis, and IgG was used as the isotype control. TTX promoted the association between PSD-95 and stargazin, which was canceled by 2-BP or NtBuHA. Data are mean ± SEM. Two-way ANOVA was used, and n = 4 per group for all (A-F). *p < 0.05, **p < 0.01, ***p < 0.001 versus Ctrl group. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 versus TTX group. For detailed statistical analyses, see Table 1. Original images of Western blots for Figure 2 are shown in Extended Data Figure 2-1.

Figure 3.

ABE method is reliable for palmitoylation detection, and 2-BP has no effect on basal palmitoylation of PSD-95. A, Palmitoylation of PSD-95 in cultured hippocampal neurons. B, C, Palmitoylation of GluA1 and GluA2 in cultured hippocampal neurons. D, E, Palmitoylation of GluN2A and GluN2B in cultured hippocampal neurons. F, G, Western blotting analysis of lysates prepared from cultured neurons treated with vehicle or 2-BP. Data show that treatment with 2-BP dose did not decrease the palmitoylation of PSD-95 on a basal level. Data are mean ± SEM. Student's t test was used, and n = 4 per group. For detailed statistical analyses, see Table 1.

Figure 2-1

Full original images of Western blotting assays for Figure 2. Download Figure 2-1, TIF file ^{(13.7MB, tif)} .

Synaptic transmission strength is regulated by inhibition of depalmitoylation

Palmitoylation is known to regulate protein stability, sorting, and localization, and depalmitoylation plays a pivotal role in recycling and/or degradation of proteins that undergo palmitoylation (Greaves and Chamberlain, 2007; Salaun et al., 2010). However, the effects of depalmitoylation on synaptic function remain poorly resolved, so we asked whether inhibiting depalmitoylation by lentivirus silencing the depalmitoylatases, APT1 and PPT1, could alter synaptic transmission and plasticity. Based on the expression level of target proteins, the degree of silencing APT1 and PPT1 with LV-shRNAs was calculated to be 83% and 68%, respectively. Interestingly, knockdown of APT1, but not PPT1, led to a significant upward shift in mEPSC amplitude and frequency (Fig. 4A–C). To further examine this, we used selective inhibitor ML348 to specifically inhibit the activity of APT1 to reduce depalmitoylation and thereby increase palmitoylation. Treating hippocampal neurons with ML348 (1 μm, 24 h) caused a significant overall increase in mEPSC amplitude and frequency (Fig. 4D–F). Paired-pulse facilitation, an indicator of presynaptic neurotransmitter release, was measured, and an increasing tendency of paired-pulse facilitation was observed in ML348-treated neurons (Fig. 4G), indicating that APT1 not only affects postsynaptic responsiveness, but also likely to ascribe to the alteration of presynaptic neurotransmitter release. Moreover, consistent with a previous study (Jeyifous et al., 2016), an elevated ratio of AMPAR/NMDAR EPSC amplitude was observed (Fig. 4H), showing that APT1 is more likely to alter the EPSCs mediated by AMPARs than NMDARs. In addition, treatment with 2-BP could successfully reverse the effect of APT1 knockdown on mEPSC amplitude and frequency (Fig. 5A–C). Together, these results suggest that inhibition of depalmitoylation can significantly enhance glutamatergic neuronal synaptic transmission.

Figure 4.

Synaptic strength is regulated by inhibition of depalmitoylation. A, Representative (left lane; calibration: 10 pA, 500 ms) and average (right lane) traces of mEPSCs recorded from cultured neurons transfected with LV-eGFP, LV-shAPT1, and LV-shPPT1. B, C, Cumulative probabilities and average mEPSC amplitudes and frequencies showing that knockdown of APT1, not PPT1, significantly enhanced the mEPSC amplitude and frequency. D, Representative (left lane; calibration: 10 pA, 500 ms) and average (right lane) traces of mEPSCs recorded from cultured neurons treated with vehicle or ML348. E, F, Cumulative probabilities and average mEPSC amplitudes and frequencies illustrating that ML348-treated neurons had a significant overall increase in mEPSC amplitude and frequency. G, Traces of paired pulse-evoked EPSCs from cultured neurons treated with vehicle or ML348 (left), and statistical analysis result of paired-pulse facilitation ratio (PPR), indicating that ML348 increased the PPR (right). Calibration: 50 pA, 25 ms. H, Traces of AMPAR- and NMDAR-meditated EPSCs from cultured neurons treated with vehicle or ML348 (left), and statistical analysis of AMPAR/NMDAR EPSC amplitude ratio validating ML348 elevated the AMPA/NMDA ratio (right). Calibration: 100 pA, 20 ms. Data are mean ± SEM. One-way ANOVA (B,C), Student's t test (others). n = 71, 37, 51 per group (B,C); n = 39, 31 per group (E,F); n = 6 per group (G,H). *p < 0.05, **p < 0.01, ***p < 0.001 versus LV-eGFP or Veh group. ^#p < 0.05, ^##p < 0.01 for LV-shPPT1 versus LV-shAPT1 group. For detailed statistical analyses, see Table 1.

Figure 5.

2-BP can reverse the effect of APT1 knockdown on mEPSCs. A, Representative (left lane; calibration: 10 pA, 500 ms) and average (right lane) traces of mEPSCs from cultured neurons transfected with LV-eGFP or LV-shAPT1, or treated with 2-BP. B, C, Cumulative probabilities and average mEPSC amplitudes and frequencies showing that knockdown of APT1 significantly enhanced the mEPSC amplitude and frequency, whereas these were reversed by 2-BP. Data are mean ± SEM. B, C, One-way ANOVA was used, and n = 11-15 per group from 5 rats. **p < 0.01, ***p < 0.001 versus LV-eGFP + Veh group. ^##p < 0.01 versus LV-shAPT1 + Veh group. For detailed statistical analyses, see Table 1.

Silencing APT1 increases palmitoylation of glutamate receptors and PSD-95

LV constructed with shRNA targeting APT1 (LV-shAPT1) was transfected into cultured hippocampal neurons to knockdown APT1 with success (silencing efficiency 83%; Fig. 6A). Then ABE assay indicated that knockdown APT1 with special silencing shRNA resulted in an elevating tendency on the palmitoylation levels of PSD-95 (Fig. 6B), GluA1 (Fig. 6C), GluA2 (Fig. 6D), and GluN2A-B (Fig. 6E,F). In addition, inhibiting the expression of APT1 this way increased the membrane surface expression levels of GluA1-2, with little influence on NMDARs surface expression (Fig. 6G). These findings show that glutamatergic synaptic strength can be significantly influenced by inhibition of depalmitoylation via silencing APT1.

Figure 6.

Silencing APT1 increases palmitoylation of glutamate receptors and PSD-95. A, Western blotting analysis of APT1 from cultured neurons treated with LV-eGFP and LV-shAPT1. Data show that the expression of APT1 was dramatically declined after knockdown with transfection. B-D, Western blotting analysis from cultured neurons transfected with LV-eGFP or LV-shAPT1. Data show that LV-shAPT1 had a promoting effect on the palmitoylation of PSD-95 (B), GluA1 (C), and GluA2 (D). E, F, Western blotting analysis of lysates prepared from cultured neurons treated with LV-eGFP or LV-shAPT1. Knockdown of APT1 upregulated the palmitoylation of GluN2A and GluN2B. G, Western blotting analysis from cultured neurons transfected with LV-eGFP and LV-shAPT1. Data show that treatment with LV-shAPT1 increased the surface level of GluA1-2, but without effect on the surface expression of NMDARs. Data are mean ± SEM. Student's t test was used, and n = 4 per group for all. *p < 0.05, **p < 0.01, ***p < 0.001 versus LV-eGFP group. For detailed statistical analyses, see Table 1. Original images of Western blots for Figure 6 are shown in Extended Data Figure 6-1.

Figure 6-1

Full original images of Western blotting assays for Figure 6. Download Figure 6-1, TIF file ^{(18.8MB, tif)} .

Downregulation of APT1 activity increases palmitoylation of glutamate receptors and PSD-95

To further confirm the findings described above, cultured hippocampal neurons were treated with ML348, a specific inhibitor targeting depalmitoylating enzyme APT1 (Virlogeux et al., 2021). Our ABE assay indicated that inhibiting the activity of APT1 resulted in increased palmitoylation levels of PSD-95 (Fig. 7A), GluA1 (Fig. 7B), GluA2 (Fig. 7C), and GluN2A-B (Fig. 7D,E). Moreover, inhibiting the activity of APT1 with ML348 (1 μm, 24 h) at cellular level at 37°C enhanced the membrane surface expressions of GluA1-2, but with no influence on the surface levels of NMDARs (Fig. 7F). Together, our findings show that glutamatergic synaptic strength can be efficiently regulated by preventing depalmitoylation via inhibition of APT1.

Figure 7.

Downregulation of APT1 activity increases palmitoylation of glutamate receptors and PSD-95. A, B, Western blotting analysis results from cultured neurons treated with vehicle and ML348. Data show that ML348 had a promoting effect on the palmitoylation of PSD-95 (A) and GluA1 (B). C-E, Western blotting analysis of lysates prepared from cultured neurons treated with vehicle or ML348. Data show that treatment with ML348 increased the palmitoylation of GluA2 and GluN2A-B. F, Western blotting analysis from cultured neurons treated with vehicle or ML348. Results showed that ML348 increased the surface level of GluA1-2, but without influence on the surface expression of NMDARs. Data are mean ± SEM. Student's t test was used, and n = 4 per group for all. *p < 0.05, **p < 0.01, ***p < 0.001 versus Veh group. For detailed statistical analyses, see Table 1. Original images of Western blots for Figure 7 are shown in Extended Data Figure 7-1.

Figure 7-1

Full original images of Western blotting assays for Figure 7. Download Figure 7-1, TIF file ^{(8.1MB, tif)} .

Silencing PPT1 has no effects on palmitoylation of glutamate receptors and PSD-95

Additionally, we transfected the cultured hippocampal neurons with LV containing PPT1 shRNA (LV-shPPT1; silencing efficiency 68%) and further examined the effect of this on the palmitoylation levels of glutamate receptors and PSD-95 (Fig. 8A). The ABE assay results indicated that, different with APT1, silencing PPT1 has little or no influence on the palmitoylation levels of PSD-95 (Fig. 8B), GluA1-2 (Fig. 8C,D), and GluN2A-B (Fig. 8E,F). Inhibiting the expression of PPT1 also had no influence on the plasma membrane surface expression of GluA1-2 and NMDARs (Fig. 8G). Together, our findings show that glutamatergic synaptic strength can be effectively regulated by inhibition of depalmitoylation enzyme APT1 but not PPT1, and further study is needed to reveal the underlying mechanism of this phenomenon.

Figure 8.

Silencing PPT1 has no effects on the palmitoylation of glutamate receptors and PSD-95. A, Western blotting analysis of PPT1 from cultured neurons treated with LV-eGFP and LV-shPPT1. Data showing the expression of PPT1 were significantly decreased. B-D, Western blotting results of lysates prepared from cultured neurons treated with LV-eGFP or LV-shPPT1. Results indicate that knockdown of PPT1 did not alter the palmitoylation of PSD-95 and GluA1-2. E-G, Western blotting analysis of lysates prepared from cultured neurons treated with LV-eGFP or LV-shPPT1. Knockdown PPT1 did not affect the palmitoylation of GluN2A-B and had no effect on the surface expression of GluA1-2 and NMDARs. Data are mean ± SEM. n = 4 per group for all. ***p < 0.001 versus LV-eGFP group (Student's t test). For detailed statistical analyses, see Table 1. Original images of Western blots for Figure 8 are shown in Extended Data Figure 8-1.

Figure 8-1

Full original images of Western blotting assays for Figure 8. Download Figure 8-1, TIF file ^{(16.5MB, tif)} .

Palmitoylation participates in fear conditioning-induced synaptic transmission strengthening

We had focused on the regulation of synaptic plasticity by palmitoylation at the cellular level and found that APT1 may be an important target for glutamatergic synaptic plasticity modification. Further, we examined the effect of palmitoylation on hippocampal synaptic plasticity in a classic learning memory model of rats. Timelines of the animal experiments are shown in Figure 9A. ABE assay for total biotin-labeled proteins without purification by streptavidin-agarose showed that the hippocampal global palmitoylation level at 2 h after fear conditioning training was significantly elevated compared with the control group (Fig. 9B). Results of mEPSC recording from hippocampal slices showed that the frequency and amplitudes of mEPSCs were both increased after fear training, and they were almost completely reversed by palmitoylation inhibitor 2-BP (10 μm, 8 h prior recording) at slice level (Fig. 9C–E). Our fEPSP recordings from rat hippocampal slices showed that fear conditioning enhanced fEPSP slopes, and 2-BP treatment at slice level significantly attenuated fear conditioning-induced SC-CA1 LTP enhancement (Fig. 9F,G). Fear memory acquisition testing on day 3 of the fear conditioning tasks showed that freezing behavior of rats was significantly reduced by treatment with 2-BP (1 nmol per side, 1 h before testing) compared with the control group (Fig. 9H). Moreover, while APT1 inhibitor ML348 (1 μm, 8 h prior recording) at slice level did not obviously affect LTP in the hippocampus of both control and fear-conditioned groups (Fig. 9I,J), it (0.1 nmol per side into the hippocampus, 1 h before testing) facilitated fear memory formation in vivo (Fig. 9K). Figure 10 is the illustrative synopsis of palmitoylation cycling in hippocampal synaptic plasticity. These findings suggest that palmitoylation is functionally involved in fear conditioning-induced enhancement of hippocampal synaptic transmission and the formation of fear memory.

Figure 9.

Palmitoylation is involved in fear conditioning-induced hippocampal synaptic strengthening. A, Timelines of the animal experiments with fear-conditioned rats. B, Western blotting analysis of streptavidin-labeled protein from the hippocampal tissue of the control or fear-conditioned (FC) rats. FC increased global palmitoylation level. C, Representative (left lane; calibration: 10 pA, 500 ms) and average (right lane) traces of mEPSC recordings from rat hippocampal slices at 2 h after fear conditioning training. D, Cumulative probabilities and average mEPSC frequencies showing that FC enhanced the frequency of mEPSC on hippocampal slices, which could be abolished by 2-BP treatment. E, Cumulative probabilities and average mEPSC amplitudes showing that 2-BP abolished the increase of mEPSC amplitude induced by FC. F, Time course of the SC-CA1 LTP induced by HFS in hippocampal slices from different groups, showing LTP recording before and 60 min after the application of HFS. Calibration: 2 mV, 10 ms. G, Histogram illustrating the mean increase in fEPSPs slopes averaged from the last 15 min in each group and showing 2-BP significantly attenuated FC-induced LTP enhancement with no obvious effect on LTP of control group. H, The freezing behavior of rats tested on day 3 was significantly reduced by 2-BP treatment compared with the vehicle, indicating that inhibition of palmitoylation impairs the formation of fear memory. I, Time course of SC-CA1 LTP in hippocampal slices from different groups, showing LTP recording before and 60 min after application of HFS. Calibration: 2 mV, 10 ms. J, Histogram illustrating the mean increase in fEPSPs slopes averaged from last 15 min in each group and showing that ML348 had no significant influence on LTP of both control and fear conditioning groups. K, The freezing behavior of rats tested on day 3 was slightly facilitated by ML348 treatment compared with the vehicle (p = 0.06), indicating that inhibition of APT1 promotes the formation of fear memory. Data are mean ± SEM. Two-way ANOVA (D,E,G,J); one-way ANOVA (H,K). n = 6 per group for (D,E,H,K); n = 4 for (G,J). *p < 0.05, **p < 0.01 versus Ctrl or Veh group. ^##p < 0.01, ^###p < 0.001 versus FC + Veh group. For detailed statistical analyses, see Table 1. Original images of Western blots for Figure 9 are shown in Extended Data Figure 9-1.

Figure 10.

Schematic illustration of palmitoylation cycling in hippocampal synaptic plasticity. During synaptic scaling, the activity of PATs is upregulated, which promotes the palmitoylation modification, translocation, and stabilization of synaptic proteins (PSD-95, glutamate receptors) in postsynaptic density. Knockdown with shRNA or inhibiting the activity of depalmitoylatase APT1 with ML348 can largely simulate the effect of synaptic scaling. Fear conditioning-induced enhancement in synaptic strength was also attenuated by palmitoylation inhibitor 2-BP.

Figure 9-1

Full original images of Western blotting assays for Figure 9. Download Figure 9-1, TIF file ^{(6.5MB, tif)} .

Discussion

In the present study, we demonstrated that palmitoylation modification is essentially required for hippocampal synaptic plasticity and memory acquisition after fear conditioning training, and APT1-mediated depalmitoylation is involved in the regulation of PSD-95 palmitoylation and clustering in the PSD and glutamatergic synaptic transmission at a network level. Downregulation of the activity of APT1 with shRNA silencing or pharmacological inhibitor significantly increased the palmitoylation levels and synaptic or membrane surface expression levels of PSD-95 and AMPARs, suggesting that (de)palmitoylation plays important roles in regulating glutamatergic synaptic plasticity and fear memory.

It has been reported that palmitoylation participates in regulating neuronal protein stability, sorting, and trafficking, and subsequently changes the synaptic targeting and functional efficiency of substrate proteins (Huang and El-Husseini, 2005; Greaves and Chamberlain, 2007). We here demonstrated that blocking neuronal activity with TTX to simulate increased transmission strength in synaptic network enhanced the global palmitoylation and PSD-95 and glutamate receptor palmitoylation levels, as well as the synaptic and membrane surface distribution of these PSD proteins. Consistent with our previous results (Shen et al., 2019; Xia et al., 2019), treating with palmitoylation inhibitors, 2-BP or NtBuHA, reversed the increased palmitoylation levels or synaptic expression of the target molecules, such as PSD-95, GluA1-2, and NMDARs. Additionally, depalmitoylation plays a pivotal role in recycling and/or degradation of proteins that undergo palmitoylation (Greaves and Chamberlain, 2007; Salaun et al., 2010); and in clinical research, analysis of the brains of patients with infantile neuronal ceroid lipofuscinosis compared with control samples revealed an obvious deficiency of PPT1, which impairs the cleavage of thioester linkage in palmitoylated proteins, and prevents the degradation and causes accumulation of lysosomal ceroid (Kim et al., 2008; Sarkar et al., 2013). However, we found that lentiviral silencing or ML348 selective inhibition of APT1 prominently elevated the amplitude and frequency of mEPSC, suggesting that APT1, not PPT1, is an important regulator in synaptic transmission.

PSD-95 is a commonly involved scaffolding protein in the context of synaptic plasticity (Chen et al., 2005; Xu, 2011), and it has been proposed to play an essential role in maintaining and regulating glutamatergic synaptic function and spine morphology (Béïque et al., 2006; Elias and Nicoll, 2007). In several studies, higher LTP was observed in PSD-95 mutant mice and increased PSD-95 occluded LTP in the cerebral cortex (Béïque and Andrade, 2003; Stein et al., 2003). In contrast, a growing field of research reported that overexpression of PSD-95 raises the number of postsynaptic AMPARs, and the LTP expression is supposed to be ascribed to this (El-Husseini et al., 2002; Ehrlich and Malinow, 2004; Ehrlich et al., 2007). Though controversial, these pertinent and correlative observations suggest that the dynamics of PSD-95 are closely related to synaptic plasticity and the expression of LTP. Moreover, in another study, the number of AMPAR slots, as well as the level of PSD-95 in the PSD, is upregulated by increasing PSD-95 palmitoylation via APT1 blockade (Jeyifous et al., 2016). There have been highly relevant studies addressing the association between PSD-95 palmitoylation and synaptic plasticity (including homeostatic upscaling and downscaling) (Noritake et al., 2009; Chowdhury et al., 2018). However, direct evidence whether changes in the palmitoylation state of PSD-95 mediate synaptic transmission changes is still lacking. Here, we revealed that the levels of global palmitoylation and PSD-95 palmitoylation were obviously raised after TTX treatment at the cellular level.

In this study, we explored the potential role of APT1-mediated palmitoylation in synaptic transmission but did not rule out the role of other palmitoylation-related enzymes. We have not yet discovered that PSD-95 is the direct substrate of APT1; however, it has been reported that PSD-95 is the substrate of ABHD17 (Yokoi et al., 2016). The palmitoylation of PSD-95 is a dynamic equilibrium by both palmitoylation and depalmitoylation processes. One nonselective inhibitor of depalmitoylatases, palmostatin B, may increase the palmitoylation level of PSD-95 by blocking the activity of ABHD17s (Lin and Conibear, 2015b). The role of ABHD17 depalmitoylases in regulating palmitoylation modification of synaptic proteins will be further analyzed in our follow-up studies. We have found in our ongoing research that the enzyme activity of DHHCs has association with their own palmitoylation; that is, the activity of DHHC enzymes increases with the increase of the palmitoylation levels of themselves; and DHHC enzymes are also the substrate of APT1. Therefore, we hypothesized that the APT1 selective inhibitor ML348 may induce the increase of the palmitoylation level of DHHCs by blocking the activity of APT1, and subsequently increases the palmitoylation level of PSD-95. APT1 is mainly distributed in cytoplasm and mitochondria (Kathayat et al., 2018), so effects of APT1 knockdown and ML348 on synaptic proteins, including PSD-95 and GluRs, may be indirect. It is likely that APT1 plays roles in the regulation of synaptic-related proteins and synaptic transmission via affecting the palmitoylation and/or depalmitoylation process of other unknown substrate proteins.

It has been found that congenital complete loss of PPT1 has effect on the expression of GluN2A and PSD-95 in mouse visual cortex neurons from >33 d after birth (Koster et al., 2019). However, there are at least the following differences between that study and ours, which may be the causes for the different effects of PPT1 silencing. First, there are differences in the origin and types of neurons that were studied. Second, the purposes of study were different. The aim of Koster et al. (2019) was to investigate the expression of synaptic proteins during the visual development of PPT1 KO mice; our study focused on the palmitoylation cycling of hippocampal synaptic proteins. Third, the gene interferences were conducted in different ways and at different points in time. Fourth, in the study of Koster et al. (2019), only a small reduction (∼20%) in GluN2A and PSD-95 expression was observed under the condition of complete KO of the PPT1 gene; differently, the method of shRNA knockdown in our study cannot silence the expression of target gene completely (the knockdown efficiency of PPT1 was ∼68%), so it may result in no obvious change in expression of downstream targets and no significant effect on palmitoylation modification of synaptic proteins.

Strength and plasticity of excitatory synaptic transmission can be regulated by PSD-95 through associating with the membrane receptors and ion channels (Vallejo et al., 2017), such as AMPARs and NMDARs. However, we revealed that the change of PSD-95 palmitoylation only alters the surface expression of AMPARs, but not NMDARs. A possible explanation for this might be that the spatial structure of PSD-95 after palmitoylation is dynamic when associated with AMPARs and more stable when it interacts with NMDARs. Our findings suggest that palmitoylation and APT1-mediated depalmitoylation of PSD-95 and AMPARs have important influence on the expression of LTP of hippocampal glutamatergic synapses.

Fear conditioning enhances LTP induction, but the underlying molecular basis responsible for this phenomenon is not fully understood (Abrari et al., 2009; Chen et al., 2014; Morel et al., 2018). Our study results after fear conditioning indicate that palmitoylation is required for the strengthening of synaptic transmission and fear memory acquisition. In combination with our in vitro study, it can be supposed that (de)palmitoylation may play a key functional role in the regulation of insertion of PSD-95 and AMPARs into the glutamatergic synapses, which is a basic molecular component for synaptic plasticity and learning and memory. However, the animal studies in this study were not in-depth; and because 2-BP may cause changes in palmitoylation modification levels of various proteins after in vivo administration, thus the connection between animal level experiment and mechanism exploration at cellular level is insufficient. Together, these findings suggest that the APT1/PSD-95/AMPARs pathway and palmitoylation cycling are involved in selective membrane surface insertion of synaptic proteins and regulation of LTP expression, as well as memory formation.

In conclusion, the results of this study showed an important effect of palmitoylation and APT1-mediated depalmitoylation on the dynamical functional modification and trafficking of PSD-95 and AMPARs, and hippocampal glutamatergic synaptic transmission. Our findings indicate that palmitoylation cycling and APT1/PSD-95/AMPARs pathway-mediated selective membrane surface stability of glutamatergic receptors may be critical for memory formation, and it may offer new explanation for the impairment of synaptic plasticity in learning and memory deficient disorders, such as aging-related cognitive decline.