All experimental procedures were approved by the South China Normal University Animal Care and Use Committee. Cyld^+/− mice were generously provided by Dr. Shao-cong Sun (University of Texas MD Anderson Cancer Center, Texas, USA). In all experiments, both male and female 3-to 5-month-old Cyld^+/+ mice and their Cyld^−/− littermates, generated from Cyld^+/− mice, were used according to international and university ethical standards. Genotyping was performed by PCR, as previously described (Reiley et al., 2007), using tail DNA and the following primers: Cyld forward primer CCAGGCACTTTGAATTGCTGTC; Cyld reverse primer 1 CGTTCTTCCCAGTAGGGTGAAG; Cyld reverse primer 2 GCATGCTCCAGACTGCCTTGG. Animals were given ad libitum access to food and water and housed on a 12 h light/dark cycle at 22–25°C.

HEK293 cells were cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% fetal bovine serum. HEK293 cells were transfected with the indicated plasmids with calcium phosphate (Clontech). At 48 h after transfection, HEK293 cells were washed with ice-cold phosphate buffered saline (PBS), centrifuged at 5000 rpm at 4°C for 10 min, and the precipitate was collected and homogenized using ice-cold sodium dodecyl sulphate (SDS) lysis buffer containing protease inhibitors (Beyotime, China) and phosphatase inhibitors (Beyotime, China), and then centrifuged at 14000 rpm for 10 min at 4°C.

Mouse brains were dissected on ice and the DLS tissues were separately sectioned in a cryostat as previously described (Peng et al., 2021). Tissues were homogenized in ice-cold lysis buffer containing (in mM) 50 Tris (pH 7.5), 5 EDTA, 150 NaCl, 1% SDS, protease inhibitors and phosphatase inhibitors and kept at 4°C for 40 min before cellular debris was removed by centrifugation at 14000 rpm for 10 min at 4°C. The supernatant was collected, its protein concentration determined and subsequently denatured for 20 min at 75°C. To extract cell membrane proteins, we followed the manufacturer’s instructions for the Membrane and Cytosol Protein Extraction Kit (P0033, Beyotime, China) (Li et al., 2018; Jiang et al., 2022). Briefly, DLS membrane proteins were extracted at 0–4°C using 1 ml solution A from the kit containing complete a proteinase and phosphatase inhibitor cocktail (Complete Mini; Roche) with 1 min homogenization. Cellular debris and nucleus were removed by centrifugation at 700× g for 10 min at 4°C, and then cell surface debris were collected by centrifugation at 14,000× g for 30 min at 4°C. The DLS membrane proteins were extracted by adding 200 μl solution B and incubating for 10 min at 4°C, and then centrifuging at 14,000× g for 5 min at 4°C. The supernatant was collected for denaturation at 75°C for 20 min. The protein samples were either stored or mixed with 25% (by volume) 5× SDS loading buffer at 75°C for 20 min prior to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), then electrotransferred onto a nitrocellulose membrane. The membrane was blocked with 5% non-fat dry milk in Tris-buffered saline (TBS) containing 0.5% Tween-20 (TBST) for 1 h at room temperature (RT) and incubated overnight at 4°C with the appropriate primary antibodies in TBST; anti-β-actin (AF0003, Beyotime, China), anti-α-tubulin (AF0001, Beyotime, China) or anti-sodium potassium ATPase (AF1864, Beyotime, China) was used as loading control (see Table 1 for details). Following washing and incubation with suitable secondary antibodies for 1 h at RT, and three 10-min washes with TBST, protein bands were visualized using an Immobilon Western ECL system (Bio-Rad, USA) and analyzed with Gel-Pro Analysis software (Media Cybernetics, USA).

Mouse brain tissue and cell lysates were obtained and prepared as previously described (Chen et al., 2017). Lysates were incubated with 5 μg CYLD antibody (sc-74,435, Santa Cruz, USA), GluA1 antibody (182,011, Synaptic Systems, USA), GluA2 antibody (ab206293, Abcam, UK) or Myc antibody (ab32, Abcam, UK) (or IgG as a control) overnight at 4°C with rotation. Protein A + G agarose (P2012, Beyotime, China) was added to each sample at 4°C for 3–4 h. Subsequently, three 5-min washes were made with cold lysis buffer. The precipitates were eluted from the beads by adding the same volume of 2 × SDS loading buffer and heating for 20 min at 75°C. Then, the samples were used for western blotting as described above.

The in vitro deubiquitination assay was performed as described previously (Huo et al., 2015; Ma et al., 2017). GluA1 and GluA2 were purified from denatured lystes of HEK293 cells overexpressing Myc-GluA1 or Myc-GluA2 and HA-K63Ub by immunoprecipitation with anti-Myc and elution with Myc peptide (P9805, Beyotime, China). CYLD was purified from lysates of HEK293 cells overexpressing CYLD. Deubiquitination reactions were performed in deubiquitination buffer (in mM) (40 Tris, pH 7.1, 100 NaCl, 4 Tris (2-carboxyethyl) phosphine hydrochloride and 25% glycerol). Myc-GluA1 with K63Ub was incubated with or without CYLD for 2 h at 37°C. 1× SDS loading buffer was then added and the samples were heated for 20 min at 75°C prior to western blotting.

Mice were anesthetized with chloral hydrate and transcardially perfused with prechilled and oxygenated modified artificial cerebrospinal fluid (aCSF) containing the following (in mM): 93 N-methyl-D-glucamine, 93 HCl, 2.5 KCl, 1.2 Na₂HPO₄, 30 NaHCO₃, 20 HEPES, 25 D-glucose, 5 sodium L-ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO₄, 0.5 CaCl₂ (pH 7.3) saturated with 95% O₂/5% CO₂. Coronal 320 μm striatal slices were cut using a vibratome (VT1000S, Leica, Germany). Slices containing the cortex and striatum were continuously bathed in incubation aCSF (in mM): 92 NaCl, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 D-glucose, 5 sodium L-ascorbate, 2 thiourea, 3 sodium pyruvate, 2 CaCl₂, and 2 MgSO₄ (pH 7.3). For electrophysiology recordings, the slices were recovered at RT for at least 1 h before a single slice was transferred to a submersion chamber perfused with 95% O₂/5% CO₂-saturated recording aCSF containing (in mM): 124 NaCl, 2.5 KCl, 2 CaCl₂, 1.2 NaH₂PO₄, 24 NaHCO₃, 2 MgSO₄, 12.5 D-glucose, 5 HEPES (pH 7.3). Recording aCSF was passed through an in-line heater at a flow rate of 1.5–2.5 ml/min, and the temperature was held constant at 30°C.

Whole-cell responses were recorded as described previously (Chen et al., 2021). Briefly, whole-cell patch-clamp recordings were made from MSNs in the DLS. MSNs were morphologically and electrophysiologically identified. To assess the intrinsic membrane and AP properties of MSNs, glass pipettes (5–6 MΩ resistance) filled with internal solution containing (in mM) 110 K-gluconic acid, 10 NaCl, 1 MgCl₂, 10 EGTA, 40 HEPES, 2 Mg-ATP, 0.3 Na-GTP (pH 7.3; 280–300 mOsm) were used; in some cases, 0.1% biocytin (B4261, Sigma, USA) was added to the intracellular solution to visualize patched cells. The voltage responses were measured at steady state; a series of 20-or 1,000-ms depolarizing currents (from +40 pA to +440 pA with increments of 40 pA) was applied for the remaining measurements. Single AP properties, such as AP threshold, amplitude, half-width, rise and decay time, and AHP amplitude, were measured for the second evoked AP in a 20-ms current-clamp series. Finally, a series of 1,000-ms hyperpolarizing and depolarizing steps was used to assess repetitive AP firing. For recording of miniature excitatory postsynaptic currents (mEPSCs) and evoked excitatory postsynaptic currents (eEPSCs), the recording aCSF was supplemented with 1 μM tetrodotoxin (to block sodium current) and 50 μM picrotoxin (P-1675, Sigma, USA; to block GABA_A receptors). The synaptic response was recorded with glass pipettes (5–6 MΩ resistance) filled with internal solution containing the following (in mM): 100 CsMeSO₄, 10 NaCl, 10 TEA-Cl, 1 MgCl₂, 10 EGTA, 40 HEPES, 2 Mg-ATP, 0.3 Na₂-GTP, 3 QX-314 (pH 7.3; 280–300 mOsm). To assess eEPSCs, electrical field stimulation was achieved using a bipolar stimulation electrode placed in the dorsolateral corpus callosum to evoke glutamate release from the corticostriatal pathway. The MSNs to be recorded were located 300 ~ 400 μm away from the stimulation site. To construct input–output (I-O) curves for AMPAR-mediated eEPSCs, electrical stimulation current intensity was varied from 0.05 to 0.4 mA in 0.05 mA steps for each neuron. Paired pulses of AMPAR-mediated eEPSCs were delivered at 50 ms inter-pulse intervals. To construct stimulation-response curves for N-methyl-D-aspartate receptor (NMDAR)-mediated eEPSCs, electrical stimulation current intensity was varied from 0.1 to 0.6 mA in 0.1 mA steps for each neuron. To isolate AMPAR-and NMDAR-mediated eEPSCs, MSNs were voltage-clamped at −70 mV and + 40 mV in the presence of 50 μM picrotoxin. The AMPAR-mediated eEPSC amplitude was measured as the peak eEPSC amplitude at −70 mV, while NMDAR-mediated eEPSCs were estimated as the eEPSC amplitude 50 ms after the stimulation artifact at +40 mV. eEPSC paired-pulse ratios (PPRs) were measured by dividing the EPSC amplitude evoked by the second stimulus by the EPSC amplitude evoked by the first stimulus (R₂/R₁). For (RS)-3, 5-dihydroxyphenylglycine (DHPG)-induced LTD whole-cell experiments, MSNs were voltage-clamped at −70 mV in 95% O₂/5% CO₂-saturated recording aCSF with 50 μM picrotoxin. After recording 15–20 min of stable baseline, DHPG-induced LTD was induced by bath application of 100 μM DHPG for 10 min. The recording was continued for a further 30 min to monitor the induction and maintenance of LTD. All data were acquired with a Digidata 1440A interface and pClamp 10.4 software (Molecular Devices, USA) and a MultiClamp 700B amplifier (Molecular Devices, USA). Data were analyzed using Clampfit 10.4 (Molecular Devices, USA), Mini Analysis program (Synaptosoft, USA) and GraphPad Prism 8.0.2 (GraphPad, USA).

The procedures for field potential recording were described previously (Nazzaro et al., 2012; Misrani et al., 2019). Briefly, a single slice was transferred to a chamber perfused with 95% O₂/5% CO₂-saturated recording aCSF without 50 μM picrotoxin. The field potential recording glass pipette (3–4 MΩ resistance) filled with aCSF was placed in the DLS and a bipolar electrode was placed in the corpus callosum. After recording 15–20 min of stable baseline, LTD was induced by four trains of HFS (100 Hz, 1 s) with an inter-train interval of 10 s, and recording was continued for a further 60 min to monitor the induction and maintenance of LTD in the corticostriatal pathway. All data acquisition was done using a Digidata 1,550 interface and pClampex 10.4 software. Data were analyzed using Clampfit 10.4, SPSS software and GraphPad Prism 8.0.2.

Patched MSNs were visualized using biocytin for morphological analyses as described (Jiang et al., 2022). After whole-cell voltage-clamp recording, slices containing biocytin-filled MSNs were transferred to 4% paraformaldehyde (PFA) and fixed overnight at 4°C. Slices were washed (three times, 5 min each) with PBS and permeabilized in 1% Triton-X100/PBS overnight at 4°C. Subsequently, sections were incubated at 4°C overnight in Streptavidin Alexa 488 (S11223, Invitrogen, USA) diluted 1:1000 in 0.3% Triton-X100/PBS, then given three 5-min washes with PBS and mounted on glass slides. Images were collected with a confocal microscope (LSM-800, Zeiss, Germany) using 20× (2 μm Z-step) and 40× oil immersion (0.2 μm Z-step) objectives at 1024 × 1,024 pixels by sequential scanning and then processed using ZEN (Zeiss, Germany) and ImageJ software (NIH, USA). For the morphological analysis, we selected 9–11 animals per experimental group (blind to genotype), 2–3 MSNs per animal and 4–5 individual secondary dendrites (at least 50 μm from the soma; >20 μm long) per cell of comparable diameter. Dendrite arborization and spine density were quantified by an independent investigator. Dendritic spines were classified based on the spine length and spine head width, as follows: mushroom (with spine head, head width/length ratio > 0.5); thin (with spine head, length > 1.2 μm, head width/length ratio < 0.5); stubby (length < 1.2 μm, with spine head, head width/length ratio < 0.5; without head, not applicable); filopodia (without spine head, length > 1.2 μm) (Llano et al., 2015; Yasuda et al., 2020). In addition to the four classical spine subtypes, we define varicosity as dendritic segments displaying varicose swellings (dendritic diameter > 2 μm) with spine loss (Westrum et al., 1964; Isokawa, 1997; Swann et al., 2000; Maiti et al., 2015). Dendritic spine density was calculated as the number of spines per unit length.

Brain slices were prepared as for electrophysiological recordings. Slices were then transferred to a treatment chamber containing recording aCSF and treated with 100 μM DHPG or aCSF alone for 10 min at RT. For immunofluorescence, slices were allowed to recover for 10 min in recording aCSF and then fixed in 4% PFA/0.2% glutaraldehyde in PBS for 24 h, after which they were transferred to 30% sucrose in PBS until saturated. The slices were embedded in Tissue-Tek O.C.T Compound (4,583; SAKURA, USA) and stored at-80°C before being sliced into 20 μm coronal cryostat sections at-18°C (CM3050, Lecia, Germany).

The procedures for immunohistochemistry were described previously (Zhou et al., 2018; Chen et al., 2021). Briefly, mice were anesthetized with 10% chloral hydrate and subjected to cardiac perfusion with 95% O₂/5% CO₂-saturated recording aCSF followed by 4% PFA/0.2% glutaraldehyde in PBS. Sections were washed three times with PBS, permeabilized with 0.05% Triton X-100 in PBS for 2 h, blocked with 10% donkey serum, incubated with primary antibodies overnight at 4°C and then incubated with appropriate secondary antibodies and DAPI at RT for 2 h. Sections were subsequently washed and mounted on glass slides. For confocal microscopy, images were acquired using a Zeiss LSM-800 fitted with a 40 × oil immersion (2 μm Z-step) objective at a pixel resolution of 1,024 × 1,024. The image acquisition settings were the same for all scans when fluorescence intensity was compared. Images were analyzed using ZEN (Zeiss) and ImageJ software (NIH).

All data are expressed as the mean ± standard error of the mean (SEM) and are statistically evaluated by Student’s t-test, ANOVA (one-way, two-way, two-way repeated measures) with Tukey’s or Sidak’s post-hoc multiple comparison test or two-sample Kolmogorov–Smirnov test using GraphPad Prism and SPSS. p < 0.05 was considered significant (*p < 0.05, **p < 0.01, ***p < 0.001). Mean ± SEM values, sample size, p-values and statistical methods are defined in the respective results and figure legends.

CYLD is highly expressed in the dorsal striatum and regulates the morphogenesis of dendrites and spines. To determine whether CYLD plays a role in neurite complexity and spine morphology of MSNs in the DLS, we examined MSN morphology by filling the neurons with biocytin during whole-cell patch-clamp recordings. We measured the complexity of neuronal dendritic arborization in MSNs using three parameters: the length of dendrites, the surface area of dendrite coverage and the number of dendrite intersections at various distances from the cell body (Sholl’s analysis) (Ferreira et al., 2014; Wang et al., 2019). We found that CYLD deficiency induced a significant reduction in dendrite length (Figures 1A–C, Cyld^+/+: 3.005 ± 0.120 mm, Cyld^−/−: 2.282 ± 0.120 mm, t = 4.163, p = 0.0001, unpaired Student’s t-test) and dendrite surface area (Figure 1D, Cyld^+/+: 4.298 × 10⁴ ± 2.203 × 10³, Cyld^−/−: 3.294 × 10⁴ ± 1.851 × 10³, t = 3.497, p = 0.001, unpaired Student’s t-test), as well as the number of branches (Figure 1E, interaction between genotype and distance, F_{(17, 814)} = 2.312, p = 0.002; main effect of genotype, F_{(1, 48)} = 14.450, p = 0.0004; main effect of distance, F_{(3.202, 153.3)} = 202.200, p = 0.188, two-way repeated measures ANOVA).

To determine whether the dendritic spines, minute protrusions on dendrites, were altered, we segregated dendritic protrusions into different groups: mushroom spine, thin spine, stubby spine, varicosity and filopodia, based on the spine length, head width and dendrite diameter (see Section 2) (Swann et al., 2000; Maiti et al., 2015; Bello-Medina et al., 2016). We found abnormal spine loss with dendritic beading (Figures 1F,G, Cyld^+/+: 0.877 ± 0.027, Cyld^−/−: 0.427 ± 0.023, t = 12.640, p < 0.0001, unpaired Student’s t-test), as well as spine paramorphia in Cyld^−/− MSNs. The mature mushroom spine ratio was also significantly decreased (Figure 1H, Cyld^+/+: 38.096 ± 2.010, Cyld^−/−: 22.518 ± 1.756, t = 5.746, p < 0.0001, unpaired Student’s t-test). In contrast, the proportion of immature stubby spine and pathological varicosity, was significantly increased in Cyld^−/− MSNs (Figure 1H, stubby spine, Cyld^+/+: 25.451 ± 1.733, Cyld^−/−: 33.132 ± 1.908, t = 2.801, p = 0.006, unpaired Student’s t-test; varicosity, Cyld^+/+: 4.999 ± 1.2056, Cyld^−/−: 14.386 ± 2.060, t = 3.443, p = 0.0008, unpaired Student’s t-test). These data indicate that stable synaptic contacts were decreased in Cyld^−/− MSNs.

APs are the fundamental electrical signals used by the central nervous system to relay information (Bean, 2007). Therefore, we measured APs in MSNs to investigate how the above structural alterations are associated with functional modification. First, we analyzed the intrinsic electrical properties of a single AP. The AP threshold, half-width and rise time, as well as the amplitude of the after-hyperpolarization potential (AHP), were significantly increased in Cyld^−/− MSNs, suggesting that excitability is subnormal in CYLD knockout MSNs (Table 2). Furthermore, we measured intrinsic tonic firing activity in MSNs by assessing their response to depolarizing current injections (ranging from +40 to +440 pA, 40 pA steps, 1,000 ms) and found that AP firing was decreased in Cyld^−/− MSNs (Figures 1I,J, interaction between genotype and current injection, F _{(10, 430)} = 1.607, p = 0.102; main effect of genotype, F _{(1, 43)} = 6.784, p = 0.013; main effect of current injection, F _{(2.004, 86.17)} = 21.500, p < 0.0001, two-way repeated measures ANOVA). These data suggest that CYLD has a role in MSN intrinsic tonic firing activity under basal conditions.

We further asked whether the function of CYLD is also critical for excitatory synaptic connective function in DLS. To address this question, we carried out whole-cell patch-clamp experiments to record mEPSCs and eEPSCs (Figure 2A). We found that the amplitude and frequency of mEPSCs in MSNs were significantly decreased in Cyld^−/− mice, indicating that there were fewer functional synapses in Cyld^−/− MSNs, contributing to a decreased glutamate receptor and synaptic drive onto MSNs (Figures 2B–D, cumulative frequency distributions of amplitude, D = 0.19, p < 0.0001, two-sample Kolmogorov–Smirnov test; amplitude, Cyld^+/+: 12.370 ± 0.375, Cyld^−/−: 11.060 ± 0.370, t = 2.382, p = 0.022; cumulative frequency distributions of frequency, D = 0.241, p < 0.0001, two-sample Kolmogorov–Smirnov test; frequency, Cyld^+/+: 1.931 ± 0.210, Cyld^−/−: 1.477 ± 0.110, t = 2.091, p = 0.042, unpaired Student’s t-test).

AMPARs and NMDARs are two major types of glutamate receptors involved in excitatory synaptic transmission. AMPARs mediate the postsynaptic depolarization that initiates neuronal firing and also mediate the majority of fast synaptic transmission, whereas NMDARs mediate a slow component of excitatory potentials at glutamatergic synapses (Bredt and Nicoll, 2003; Popescu and Auerbach, 2003). As a further test for the effects of CYLD deficiency on MSN basal AMPAR-or NMDAR-mediated synaptic properties, we examined the I-O curves with incremental stimulus intensities: a significant decrease in the amplitude of AMPAR-mediated eEPSCs at −70 mV in Cyld^−/− MSNs was observed (Figures 2E,F, interaction between genotype and stimulation, F _{(7, 595)} = 4.640, p < 0.0001, main effect of genotype, F _{(1, 91)} = 7.936, p = 0.006; main effect of stimulation, F _{(7, 595)} = 61.710, p < 0.0001, two-way repeated measures ANOVA), while NMDAR-mediated eEPCSs at +40 mV showed no difference between the genotypes (Figures 2I,J, interaction between genotype and stimulation, F _{(5, 104)} = 0.077, p = 0.996, main effect of genotype, F _{(1, 21)} = 0.237, p = 0.631; main effect of stimulation, F _{(5, 104)} = 23.460, p < 0.0001, two-way repeated measures ANOVA). Therefore, the AMPAR/NMDAR ratio was significantly reduced in Cyld^−/− mice (Figures 2K,L, Cyld^+/+: 3.101 ± 0.380, Cyld^−/−: 2.216 ± 0.173, t = 2.222, p = 0.035, unpaired Student’s t-test) as a result of the decrease in AMPAR-mediated excitatory synaptic transmission.

Given that both the probability of presynaptic glutamate release and the number of functional synapses contribute to the frequency of mEPSCs, we measured PPRs, determined by the amplitude of the response to the second stimulus divided by the amplitude of the first one (R₂/R₁). This reflects neurotransmitter release probability, with a low PPR signifying a high initial release probability (Dobrunz and Stevens, 1997; Zinebi et al., 2001). The PPR of AMPAR-mediated eEPSCs at 50-ms inter-stimulus intervals was significantly increased in Cyld^−/− mice, suggesting a reduced probability of evoked presynaptic glutamate release from cortical terminals (Figures 2G,H, Cyld^+/+: 1.200 ± 0.030, Cyld^−/−: 1.342 ± 0.052, t = 2.426, p = 0.017, unpaired Student’s t-test). Thus far, the data suggest that CYLD deficiency alters both morphology and synaptic activity of MSNs.

To investigate whether there were any molecular alterations at synapses that parallel the electrophysiological and morphological impairments noted above, we evaluated total and plasma membrane excitatory synaptic protein levels in DLS using western blotting. We found that although the total amount of GluA1 and GluA2 expressed did not change significantly between genotypes (Figures 3A,B, GluA1, Cyld^+/+: 1.000 ± 0.372, Cyld^−/−: 0.8749 ± 0.3416, t = 0.248, p = 0.8126; GluA2, Cyld^+/+: 1.000 ± 0.244, Cyld^−/−: 1.120 ± 0.287, t = 0.318, p = 0.761, unpaired Student’s t-test), levels of surface-expressed GluA1 and GluA2 were significantly reduced in Cyld^−/− mice (Figures 3C,D, GluA1, Cyld^+/+: 1.000 ± 0.123, Cyld^−/−: 0.4646 ± 0.08106, t = 3.635, p = 0.0066; GluA2, Cyld^+/+: 1.000 ± 0.0618, Cyld^−/−: 0.535 ± 0.037, t = 6.455, p < 0.0001, unpaired Student’s t-test), suggesting a role for CYLD in AMPAR trafficking. We also measured levels of total calcium/calmodulin-dependent protein kinase II (CaMKII), pCaMKII (phosphor T286), which regulates the delivery and removal of AMPARs (Opazo et al., 2010; Hoerndli et al., 2015). However, there were no significant differences between genotypes (Figures 3A,B, CaMKIIα, Cyld^+/+: 1.000 ± 0.061, Cyld^−/−: 1.296 ± 0.162, t = 1.715, p = 0.117; CaMKIIβ, Cyld^+/+: 1.000 ± 0.337, Cyld^−/−: 1.022 ± 0.148, t = 0.059, p = 0.954; pCaMKIIα, Cyld^+/+: 1.000 ± 0.211, Cyld^−/−: 1.586 ± 0.433, t = 1.218, p = 0.251; pCaMKIIβ, Cyld^+/+: 1.000 ± 0.511, Cyld^−/−: 1.130 ± 0.268, t = 0.226, p = 0.826). Given that mGluRs, especially mGluR5, are highly enriched in the striatum and mediate long-term synaptic plasticity (Moussawi et al., 2009; Knackstedt et al., 2014), we tested levels of total and surface mGluR5, but found no significant differences between genotypes (Figures 3A–D, total amount of mGluR5, Cyld^+/+: 1.000 ± 0.205, Cyld^−/−: 0.973 ± 0.088, t = 0.119, p = 0. 909; surface-expressed mGluR5, Cyld^+/+: 1.000 ± 0.114, Cyld^−/−: 1.053 ± 0.076, t = 0.337, p = 0. 743, unpaired Student’s t-test). Consistent with the unchanged amplitude of NMDAR-mediated eEPSCs noted above, we observed unchanged levels of surface-expressed NMDAR1 and NMDAR2B, key subunits of NMDARs (Figures 3C,D, NMDAR1, Cyld^+/+: 1.000 ± 0.136, Cyld^−/−: 1.124 ± 0.239, t = 0.452, p = 0. 675; NMDAR2B, Cyld^+/+: 1.000 ± 0.128, Cyld^−/−: 1.079 ± 0.191, t = 0.344, p = 0.734, unpaired Student’s t-test).

To understand the molecular mechanism mediating levels of surface-expressed GluA1 and GluA2 in Cyld^−/− mice, we investigated the potential physical and functional associations of GluA1, GluA2 and CYLD. We first assessed the interaction of exogenously expressed GluA1 and GluA2 with CYLD in HEK293 cells. We co-transfected HEK293 cells with CYLD and Myc-GluA1 or Myc-GluA2. Cell lysates were immunoprecipitated using either anti-Myc or anti-CYLD antibodies, and immunoblotted with anti-CYLD and anti-GluA2. The results show that CYLD interacts with GluA1 and GluA2 (Figure 4A). To test whether these interactions occur endogenously, we performed co-immunoprecipitation (co-IP) experiment with CYLD using brain tissue lysates and obtained co-precipitation of both GluA1 and GluA2. Both GluA1 and GluA2 efficiently co-precipitated CYLD, suggesting interaction between endogenous CYLD and GluA1 or GluA2 in the converse IP experiment (Figure 4B).

We further evaluated the functional relevance of GluA1 and GluA2 interacting with CYLD. Given that K63Ub chains are the primary posttranslational modification of GluA1 and GluA2 (Widagdo et al., 2017), and that CYLD specifically cleaves K63Ub chains (Sato et al., 2015), we tested whether CYLD deficiency affects the levels of GluA1 and GluA2 K63-linked ubiquitination. HEK293 cells were co-transfected with Myc-GluA1 or Myc-GluA2 and HA-K63Ub, with or without CYLD. We precipitated GluA1 and GluA2 using anti-Myc antibody and immunoblotted with anti-K63Ub. For both GluA1 and GluA2, K63-polyubiquitination was diminished in the presence of CYLD, suggesting that CYLD reduced K63Ub conjugation to GluA1 and GluA2 in HEK293 cells (Figures 4C,D, GluA1-K63Ub, Cyld^+/+: 1.000 ± 0.179, Cyld^−/−: 1.826 ± 0.140, t = 3.145, p = 0.035; GluA2-K63Ub, Cyld^+/+: 1.000 ± 0.097, Cyld^−/−: 1.972 ± 0.276, t = 2.848, p = 0.0467, unpaired Student’s t-test). To further test whether K63-linked ubiquitination of GluA1 or GluA2 was defective in Cyld^−/− mice, we separately immunoprecipitated GluA1 and GluA2 using anti-GluA1 and anti-GluA2 antibody and immunoblotted with anti-K63Ub from Cyld^+/+ and Cyld^−/− mice. As expected, we observed that CYLD knockout increased K63-linked ubiquitination of GluA1 and GluA2 (Figures 4E,F, GluA1-K63Ub, Cyld^+/+: 1.000 ± 0.230, Cyld^−/−: 2.272 ± 0.320, t = 3.226, p = 0.018; GluA2-K63Ub, Cyld^+/+: 1.000 ± 0.062, Cyld^−/−: 1.231 ± 0.049, t = 2.922, p = 0.043, unpaired Student’s t-test). Using in vitro deubiquination assays, we co-expressed Myc-GluA1 or Myc-GluA2 with HA-K63Ub. GluA1 and GluA2 were purified by IP using anti-Myc antibody, eluted with Myc peptide, then incubated with or without CYLD. As previously, we observed that CYLD removed GluA1 and GluA2 K63Ub chains (Figure 4G).

To elucidate whether CYLD is involved in the regulation of AMPAR removal, we examined the trafficking behavior of GluA1 and used immunohistochemical staining to analyze the amount of surface GluA1 on acute striatal slices with or without (RS)-3, 5-dihydroxyphenylglycine (DHPG) (100 μM, 10 min) treatment, which is known to activate mGluRs and trigger endocytosis of GluA1 (Zhang et al., 2008; Pick et al., 2017; Sanderson et al., 2018; van Gelder et al., 2020). First, we assessed the amount of surface protein using an antibody directed to the N-terminal domain (NTD) of GluA1, and found that GluA1 surface expression was significantly reduced in Cyld^−/− mice under basal conditions, indicating that CYLD is involved in the regulation of GluA1 surface expression. Furthermore, we found that GluA1 surface expression in response to DHPG was significantly affected in Cyld^−/− mice, further suggesting that CYLD contributes to DHPG-stimulated AMPAR trafficking (Figures 5A,B, interaction between genotype and treatment, F _{(3, 57)} = 13.390, p = 0.0006; main effect of genotype, F _{(1, 57)} = 3.135, p = 0.082; main effect of treatment, F_{(1, 57)} = 0.017, p = 0.898; Cyld^+/+-Ctrl vs. Cyld^−/−-Ctrl, p = 0.002; Cyld^+/+-Ctrl vs. Cyld^+/+-DHPG, p = 0.049; Cyld^−/−-Ctrl vs. Cyld^−/−-DHPG, p = 0.065; Cyld^+/+-DHPG vs. Cyld^−/−-DHPG, p = 0.541; two-way ANOVA with Tukey’s post-hoc test). This suggests that CYLD is essential for the removal of GluA1 by DHPG-induced mGluR activation.

Synaptic plasticity is mainly a consequence of changes in synaptic transmission due to structural changes in the shape and number of spines. Glutamatergic synapses exhibit a long-term weakening of synaptic efficacy, known as LTD. The property and abundance of glutamate receptors play a crucial role in determining the impact of excitatory synaptic transmission and plasticity (Matsuzaki et al., 2004; Mabb and Ehlers, 2010; Maiti et al., 2015; Helm et al., 2021; Kasai et al., 2021). To further investigate whether impairment of DHPG-induced LTD in the DLS parallels the synaptic and molecular alterations noted above, LTD of excitatory corticostriatal transmission, induced by bath application of the mGluR agonist DHPG (100 μM, 10 min) (Zhu et al., 2018), was evaluated in whole-cell recordings. We found that DHPG-induced LTD deficits in Cyld^−/− mice (Figures 5C,D, interaction between genotype and time, F _(39，580) = 1.881, p = 0.001; main effect of genotype, F _{(1, 15)} = 4.764, p = 0.045; main effect of time, F _{(3.277, 48.74)} = 4.403, p = 0.007, two-way repeated measures ANOVA; Figure 5E, Cyld^+/+: 69.518 ± 4.457, Cyld^−/−: 93.743 ± 10.779, t = 2.167, p = 0.047, unpaired Student’s t-test). These results are consistent with the hypothesis that defective GluA1 removal in Cyld^−/− mice results in a DHPG-induced LTD deficit.

In the striatum, both DHPG-and HFS-induced LTD involve the activation of mGluRs and the removal of AMPARs (Sung et al., 2001; Skiteva et al., 2018). Therefore, we checked whether HFS-induced LTD is also impaired in DLS of Cyld^−/− mice. Using extracellular field potential recording of population spikes (PSs), we observed that CYLD deficiency impaired the induction and maintenance of HFS-induced field LTD in the corticostriatal pathway (Figures 5F,G, interaction between genotype and time, F _{(73, 2,190)} = 2.256, p < 0.0001; main effect of genotype, F _{(1, 30)} = 8.107, p = 0.008; main effect of time, F _{(2.879, 86.38)} = 14.680, p < 0.0001, two-way repeated measures ANOVA; Figure 5H, Cyld^+/+: 80.320 ± 2.979, Cyld^−/−: 90.860 ± 2.218, t = 2.837, p = 0.008, unpaired Student’s t-test).