Rabbit polyclonal anti-protein-interacting with C kinase 1 (PICK1) (ab3420), anti-dynamin1 (ab3456), anti-dynamin1 phospho-Ser-774 (ab55324) and pitstop2 were obtained from Abcam Biotechnology (Cambridge, UK). Sheep polyclonal anti-dynamin phospho-Ser-778 (NB300-210) was obtained from Novus Biotechnology (Littleton, CO, USA). Rabbit monoclonal anti-GluK2/3 (clone NL9, #04-921) antibodies were purchased from Millipore Biotechnology (Temecula, CA, USA). Rabbit polyclonal anti-actin (sc-10731) was obtained from Santa Cruz Biotechnology (Dallas, TX, USA). Rabbit polyclonal Na${}^{+}$/K${}^{+}$-ATPase (#3010) was purchased from Cell Signaling Biotechnology. Kainate receptor agonist was purchased from Enzo life sciences (Farmingdale, NY, USA). Chlorpromazine hydrochloride (CPZ), genistein, methyl-$\beta$-cyclodextrin (M$\beta$CD), dynasore (D7693), dantrolene (Dan), ryanodine (Rya), FK506, and Fura-2/AM were from Sigma-Aldrich Biotechnology (St. Louis, MO, USA). Cyclosporine A (CsA) was obtained from EMD Millipore (Billerica, MA, USA). Alexa Fluor 488-conjugated transferrin (T13342) and EZ link sulfo-NHS-SS-Biotin (21331) was from Thermo Scientific (Rockford, IL, USA).

HEK293T cells were obtained from the American Type Culture Collection (Manassas, VA, USA) and were cultured in DMEM supplemented with 10% FBS. The recombinant plasmids encoding GluK2 (the wild-type protein and the tyrosine phosphorylation-deficient mutant Y590F) were introduced into the cells with polyethyleneimine (PEI; Invitrogen). After transfection (24 h), drugs were administered to the transfected cells.

Intracellular Ca${}^{2+}$ ([Ca${}^{2+}$]${}_i$) was measured using the Ca${}^{2+}$-sensitive fluorescent dye Fura-2/AM (Sigma), which is a UV-excited Ca${}^{2+}$ indicator that allows for ratiometric measurements. Fura-2/AM was dissolved in DMSO as a stock solution and directly added to the cell culture medium at a final concentration of 2 $\mu$M. After GluK2 transfection (24 h), HEK293T cells were loaded with Fura-2/AM for 30 min at room temperature and then washed three times to remove the extracellular dye. The Fura-2 fluorescence was alternately excited at 340 and 380 nm using the Lambda DG-4 Ultra-High-Speed Wavelength Switcher, and the excitation light was focused on the cells via a 20$\times$ objective. The emitted fluorescence was collected at 510 nm by a high-speed EMCCD camera. The ratio of the fluorescence at 340 nm to 380 nm was directly recorded with Meta Fluor software, indicating [Ca${}^{2+}$]${}_i$ fluctuations. The intracellular free calcium concentration ([Ca${}^{2+}$]${}_i$) was calculated as follows: [Ca${}^{2+}$]${}_i$ = Kd[(R-Rmin)/(Rmax-R)]$\beta$ (Grynkiewicz et al., 1985; Xu et al., 2012), where R is the ratio of fluorescence emitted at 510 nm after excitation at 340 and 380 nm; Rmin is the ratio of fluorescence at zero calcium, and Rmax is the ratio at saturating calcium. Kd is the effective dissociation constant of Fura-2 for calcium and is equal to 224 nmol/L. $\beta$ is the ratio of fluorescence of Fura-2 at 380 nm in the presence of EGTA to that in the presence of ionomycin and is equal to 1.53. Rmax and Rmin were determined by the ionomycin method.

Primary cortical neurons from 18-d-old embryonic Sprague Dawley rats were seeded in poly-D-lysine-coated culture dishes at 0.8 $\times$ 105 cells/cm${}^2$ as previously described (Xu et al., 2009). The culture medium (Neuro-basal medium supplemented with 2% B27 and 0.5 mM glutamine) was replaced every 3 days. After 14 days of in vitro culture, the neurons were harvested and lysed in ice-cold homogenization buffer [in mmol/L: MOPS (pH 7.4) 50, sucrose 320, KCl 100, MgCl${}_2$ 0.5, and protease and phosphatase inhibitors ($\beta$-glycerophosphate 20; sodium pyrophosphate 20; NaF 50; 1 mmol/L each of EDTA, EGTA, phenylmethylsulfonyl fluoride, benzamidine, sodium orthovanadate, and p-nitrophenyl phosphate; and 5 $\mu$g/mL each of aprotinin, leupeptin, and pepstatin A)] to be used for the experiments.

The transferrin-endocytosis assay was performed as previously described (Fu et al., 2011; Lu et al., 2016; Mackenzie et al., 2017). After 7 days of in vitro culture, the primary cortical neurons were starved in a neuro-basal medium for 2 h. For the uptake assays, the starved cells were transferred to a binding media (neuro-basal medium with 0.2% bovine serum albumin, 20 mM HEPES-NaOH [pH 7.5]) containing 20 $\mu$g/ml human transferrin-Alexa Fluor 488 (Thermo, Waltham, MA USA) and placed in an incubator with 95% O${}_2$, 5% CO${}_2$ for 30 min. The neurons were washed with ice-cold PBS three times and then fixed with 4% paraformaldehyde in 1$\times$ PBS. Confocal images were acquired using a Zeiss LSM710 laser-scanning confocal microscope (Zeiss, Germany).

Live transfected HEK293T cells, and primary cortical neurons were surface-biotinylated on ice using EZ link sulfo-NHS-SS-Biotin (Pierce) (Rockford, IL) after kainate stimulation with or without inhibitor treatments. The harvested cells were lysed in TNE buffer [50 mmol/L Tris$\cdot$HCl (pH 8.0), 50 mmol/L NaF, 1% Nonidet P-40, 20 mmol/L EDTA, and 0.1% SDS]. The cell lysates (200 $\mu$g) were incubated with 30 $\mu$L of immobilized NeutrAvidin (Pierce), which binds specifically to sulfo-NHS-SS-Biotin, at 4 ${}^{\circ }$C for 2 h to precipitate the biotinylated proteins. The immobilized NeutrAvidin were washed three times to remove the nonspecifically bound proteins and collect the surface proteins. The surface proteins bound to the immobilized NeutrAvidin and the total cell lysates were boiled at 100 ${}^{\circ }$C for 5 min and subjected to immunoblot analysis with anti-GluK2 antibodies.

The sample proteins were incubated with the appropriate antibodies diluted in immunoprecipitation buffer (the ratio was 1 $\mu$g antibody per 400 $\mu$g protein in 400 $\mu$l buffer) containing 50 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 10% glycerol (vol/vol), 1 mmol/L ZnCl${}_2$, 1.5 mmol/L MgCl${}_2$, 1% Triton X-100, 0.5% Nonidet P-40, and the phosphatase and protease inhibitors mentioned above overnight at 4 ${}^{\circ }$C (Zhu et al., 2012). The mixture was incubated at 4 ${}^{\circ }$C for an additional 2 h after the addition of protein A/G. The samples were washed with immunoprecipitation buffer three times and eluted by boiling for 5 min at 100 ${}^{\circ }$C in 4$\times$ Laemmli sample buffer.

After boiling at 100 ${}^{\circ }$C for 5 min, the immuno-precipitates or equal amounts of the sample proteins were separated by SDS-PAGE and then electrotransferred onto a nitrocellulose membrane (pore size: 0.22 $\mu$m). After blocking in 3% bovine serum albumin (BSA), the membrane was incubated with the indicated primary antibodies overnight at 4 ${}^{\circ }$C. Detection was performed with the appropriate HRP-conjugated IgG and developed with the SuperSignal™ West Pico Chemiluminescent Substrate assay kit (#34087) (Thermo Fisher Scientific). The bands on the membranes were scanned and analyzed with Quantity One 1-D Analysis Software (Bio-Rad).

For each type of experiment, the data were obtained from at least three independent measurements. Except for peak Ca${}^{2+}$ transient in calcium imaging, data were normalized to the corresponding control in each experiment before statistical analysis. The student’s t-test evaluated differences between the means of two groups. Multiple groups were compared by one-way ANOVA, followed by Fisher’s least significant difference test. P 0.05 was considered significant.

There are multiple pathways of internalization from the surface of eukaryotic cells that utilize different mechanisms. Based on the requirement for the coat protein clathrin, these pathways can be divided into clathrin-dependent and clathrin-independent pathways (Mayor and Pagano, 2007; Mousavi et al., 2004). Genistein (Kumar et al., 2019; Li et al., 2020; Pooja et al., 2015; Vercauteren et al., 2010) and M$\beta$CD (Guo et al., 2015; Li et al., 2020; Vercauteren et al., 2010; Zhang et al., 2018) are inhibitors of the clathrin-independent pathway; Pitstop2 (Guo et al., 2015; Kumar et al., 2019; von Kleist et al., 2011) and CPZ (Li et al., 2020; Pooja et al., 2015; Vercauteren et al., 2010; Zhang et al., 2018) are commonly used as inhibitors of clathrin-dependent endocytosis. To identify the features of GluK2 endocytosis, cultured primary cortical neurons were pretreated with the above endocytosis inhibitors before kainate stimulation. The concentration of each compound has been justified based on similar applications in the above papers. As shown in Fig. 1A, the kainate-induced reduction in surface GluK2 was prevented by genistein (200 $\mu$mol/L) or M$\beta$CD (3 mmol/L) but not by pitstop2 (25 $\mu$mol/L) or CPZ (10 $\mu$mol/L). Genistein treatment and M$\beta$CD treatment increased the surface GluK2 by 79.1% and 96.6%, respectively. Endocytosis of transferrin (a marker of the clathrin-dependent process) was used as a control to confirm the specificity and selectivity of these endocytosis inhibitors. As shown in Fig. 1B, M$\beta$CD, or genistein treatment did not impair transferrin uptake, but pitstop2 and CPZ did. These results indicate that GluK2 endocytosis is less likely to be a clathrin-mediated process.

Fig. 1.

GluK2 endocytosis is prevented by genistein or M$\beta$CD but not by CPZ or pitstop2. Cultured cortical neurons were pretreated with CPZ (10 $\mu$mol/L), pitstop2 (25 $\mu$mol/L), genistein (200 $\mu$mol/L), or M$\beta$CD (3 mmol/L) for 30 min before kainate stimulation (300 $\mu$mol/L, 10 min) or transferrin binding (20 $\mu$g/ml, 10 min). (A) The surface and total GluK2 were immunoblotted with an anti-GluK2/3 antibody. The surface/total ratios of GluK2 are normalized to that of the control (without kainate stimulation) and expressed as the mean $\pm$ SD (n = 3). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. Neuron cultures for replicates were made from different animals. (B) The endocytosis assay of Alexa 488-conjugated transferrin in cortical neurons confirms the specificity of the inhibitors. Neuron cultures for replicates were made from different animals.

Our previous research has reported that GluK2 tyrosine phosphorylation changes both intracellular Ca${}^{2+}$ and GluK2 endocytosis in HEK293T cells (Zhu et al., 2014). Here, we attempted to explore whether there is a correlation between Ca${}^{2+}$ and GluK2 endocytosis following kainate stimulation in HEK293T cells. BAPTA-AM, a permeable chelator of free Ca${}^{2+}$, was first used to restrict the intracellular Ca${}^{2+}$ concentration. The applied concentration of BAPTA-AM was decided according to similar applications (Brown et al., 2008; Tang et al., 2007; Vomaske et al., 2010). The reduction in the membrane expression of GluK2 was prevented when the cells were pretreated with BAPTA-AM (25 $\mu$mol/L) before kainate stimulation compared with non-BAPTA-AM treatment Fig. 2A. The relative surface/total GluK2 in the BAPTA-AM group is 188.7% of the non-BAPTA group. This result suggests the potential roles of intracellular Ca${}^{2+}$ in facilitating GluK2 endocytosis. Next, we investigated the mechanism of the peak Ca${}^{2+}$ transient after kainate stimulation. The GluK2-transfected HEK293T cells were pretreated with inhibitors of the Rya receptor (dantrolene [Dan] or ryanodine [Rya]). Ca${}^{2+}$-free buffer (LiCl 140 mmol/L, KCl 4 mmol/L, EGTA 10 mmol/L, MgCl${}_2$ 1 mmol/L, glucose 10 mmol/L, HEPES 5 mmol/L) was used to reduce extracellular Ca${}^{2+}$ and impair Ca${}^{2+}$ influx. As shown in Fig. 2B, the calcium image analysis showed that the peak Ca${}^{2+}$ transient after kainate stimulation was almost abolished in the Ca${}^{2+}$-free group. No significant differences were observed compared with the control group (no kainate stimulation). The peak Ca${}^{2+}$ transient was prevented to 404.43 $\pm$ 55.72 nmol/L and 280.95 $\pm$ 62.08 nmol/L by Dan treatment and Rya treatment, respectively. The peak Ca${}^{2+}$ transient in Dan treatment and Rya treatment decreased to 44.8% and 31.2% of that in the kainate stimulation group, respectively. These data demonstrate that both Ca${}^{2+}$ influxes from extracellular space and Ca${}^{2+}$ release from intracellular stores are necessary for governing the peak Ca${}^{2+}$ transient after kainate activation of GluK2. Considering the sources of intracellular Ca${}^{2+}$, the above inhibitors were used to determine the effects of Ca${}^{2+}$ on GluK2 endocytosis. As shown in Fig. 2C, relative surface localization of GluK2 in transfected HEK293T cells was 1.1867 $\pm$ 0.15567 with Rya treatment (10 $\mu$mol/L), 1.3033 $\pm$ 0.29092 with Dan treatment (10 $\mu$mol/L), and 1.3233 $\pm$ 0.19088 with Ca${}^{2+}$-free/EGTA solution treatment. The concentration of Dan or Rya was chosen, referring to several references (Diszházi et al., 2019; Shi et al., 2014; Shinohara et al., 2014; Verkhratsky, 2005). All the treatments significantly inhibited the reduction in surface GluK2 compared to 0.4967 $\pm$ 0.05132 in the kainate treatment. The relative surface localization of GluK2 in Rya, Dan, and Ca${}^{2+}$-free/EGTA increased to 238.9%, 262.4%, and 266.4% of that in the kainate stimulation group, respectively.

Fig. 2.

Ca${}^{\mathrm{2}\mathrm{+}}$ influx and intracellular Ca${}^{\mathrm{2}\mathrm{+}}$ release facilitate GluK2 endocytosis. (A) Membrane localization of GluK2 in transfected HEK293T cells (with or without BAPTA-AM pretreatment). The surface and total GluK2 were examined by immunoblotting with anti-GluK2/3 antibodies. The surface/total ratios of GluK2 are normalized to that of the control (without kainate stimulation) and expressed as the mean $\pm$ SD (n = 3). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. (B) Calcium imaging analysis in GluK2-transfected HEK293T cells. Dan (10 $\mu$mol/L) or Rya (10 $\mu$mol/L) was used to prevent ryanodine receptor opening; Ca${}^{2+}$-free/EGTA buffer was used to block Ca${}^{2+}$ influx. All the inhibitors were added 30 min before kainate stimulation (300 $\mu$mol/L, 10 min). Data are the mean $\pm$ SEM (n = 10); *P 0.05 vs. Control (no kainate). ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. (C) Surface biotinylation of GluK2-transfected HEK293T cells (with or without Ca${}^{2+}$inhibition). All the inhibitors were added 30 min before kainate stimulation (300 $\mu$mol/L, 10 min). The surface/total ratios are normalized to that of the control (untreated cells expressing wild-type GluK2) and expressed as the mean $\pm$ SD (n = 3). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation.

The Ca${}^{2+}$ sensor PICK1, a protein with both a PSD-95/Dlg/ZO1 (PDZ) domain and a C-terminal Bin/amphiphysin/Rvs (BAR) domain, has been implicated as a significant component of the machinery that regulates the internalization and trafficking of NMDAR and AMPAR (Fiuza et al., 2017; Hanley and Henley, 2005; Terashima et al., 2008). Here, PICK1-GluK2 binding significantly increased during kainate treatments (300 $\mu$mol/L for 10 min, 20 min, and 30 min) Fig. 3A. GluK2 endocytosis was decreased in the group expressing the tyrosine phosphorylation-deficient mutant Y590F (Zhu et al., 2014). Consistent with this finding, the PICK1-GluK2 interaction was reduced in the Y590F group compared with the WT group Fig. 3B. PICK1-GluK2 interaction increased to 160.3% after kainate stimulation. The PICK1-GluK2 interaction after kainate stimulation in the Y590F group decreased to was 46.8% of the WT group.

Fig. 3.

PICK1 interaction with GluK2 changes and is affected by intracellular Ca${}^{\mathrm{2}\mathrm{+}}$signaling in kainate stimulation. (A) The interaction of PICK1 with GluK2 increases during kainate stimulation (300 $\mu$mol/L for 10 min, 20 min, and 30min) in cultured cortical neurons. (B) The PICK1 interaction with GluK2 in HEK293 cells expressing the wild-type GluK2 (WT) or the tyrosine phosphorylation-deficient mutant Y590F. (C) PICK1-GluK2 binding in cortical neurons is decreased in the presence of calcium inhibitors. ${}^{*}$P 0.05 vs. no kainate.${}^{\mathrm{\#}}$P 0.05 vs. GluK2 WT without calcium inhibition. The PICK1-GluK2 interactions are normalized to that of the control (without kainate stimulation) and expressed as the mean $\pm$ SD (n = 3). ${}^{*}$P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. Neuron cultures for replicates were made from different animals.

Peak Ca${}^{2+}$ transient is a critical change during kainate stimulation of GluK2, and it is controlled by Y590 phosphorylation (Zhu et al., 2014). Therefore, we attempted to explore whether Ca${}^{2+}$ regulates PICK1-GluK2 interaction. Extracellular Ca${}^{2+}$ deletion, Rya, Dan, and BAPTA were applied to change the Ca${}^{2+}$sources or intracellular Ca${}^{2+}$concentration. The PICK1-GluK2 interaction was significantly reduced when the cells were pretreated with extracellular Ca${}^{2+}$ deletion, Rya, Dan, or BAPTA before kainate stimulation Fig. 3C. The relative levels of the PICK1-GluK2 interaction in the Rya, Dan, EGTA, and BAPTA groups were 0.9200 $\pm$ 0.1513, 1.020 $\pm$ 0.1873, 0.7200 $\pm$ 0.1300, and 0.9167 $\pm$ 0.04163, respectively, lower than the level of 1.923 $\pm$ 0.1328 in the kainate stimulation group. CaN, a calcium/calmodulin-activated phosphatase has been implicated in several Ca${}^{2+}$-sensitive pathways, including endocytosis (Cousin and Robinson, 2001; Iida et al., 2008). PICK1-GluK2 interaction was also downregulated when CaN was inhibited by CaN inhibitors (CsA and FK-506) Fig. 3D. The PICK1-GluK2 interaction decreased by 43.6% and 46.4% in CsA treatment and FK-506 treatment, respectively. This finding suggests that PICK1 serves as the pivotal effector of the elevated intracellular Ca${}^{2+}$ concentration and CaN activation to recruit GluK2.

To study the roles of CaN in GluK2 receptor endocytosis, CaN inhibitors (CsA and FK-506) were administered to inhibit the function of CaN during kainate stimulation. Their concentrations were selected based on application in several references (Koppelstaetter et al., 2018; Takadera and Ohyashiki, 2007; Xu et al., 2012). GluK2 membrane expression was increased when CaN was inhibited Fig. 4A and  4B. CsA treatment and FK-506 treatment caused 47.3% and 31.3% increase of surface GluK2 in HEK293 cells Fig. 4A, respectively. CsA treatment and FK-506 treatment caused 65.5% and 63.1% increase of surface GluK2 in cultured neurons, respectively Fig. 4B. These findings suggest a vital role for CaN in contributing to GluK2 endocytosis.

Fig. 4.

Calcineurin promotes GluK2 endocytosis. A calcineurin inhibitor (CsA or FK506) prevents the reduction in GluK2 membrane localization during kainate stimulation in GluK2-transfected HEK293T cells (A) and cultured cortical neurons (B). The surface/total ratios are normalized to that of the control and expressed as the mean $\pm$ SD (n = 3). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. Neuron cultures for replicates were made from different animals.

Clathrin-independent endocytic pathways can either use the scission GTPase dynamin or function independently of this molecule. The surface localization of GluK2 was increased when dynamin activity was inhibited by dynasore Fig. 5A, a cell-permeable inhibitor of dynamin. Dynasore treatment (50 $\mu$M and 80 $\mu$M) increased the surface localization of GluK2 to 164.8% and 154.2% of that in the kainate stimulation group. Dynamin inhibition likely prevents the process of GluK2 endocytosis. Dynamin1 is expressed in the brain and is enriched in nerve terminals (Smillie and Cousin, 2005). An obvious signal of the GluK2-dynamin1 interaction was observed in the cortical neurons, and the binding of these proteins was decreased in response to inhibition of CaN activity Fig. 5B. CaN inhibition with CsA or FK506 exhibited 65.6% or 72.3% reduction of GluK2-dynamin1 interaction. These findings suggest that GluK2 endocytosis is probably dependent on dynamin activity. Studies have shown that the acceleration of endocytosis depends on the phosphorylation status of dynamin at two specific amino acids, namely, S774 and S778 (Armbruster et al., 2015; Tan et al., 2003). Dynamin phosphorylation at S774 and S778 was also detected when a CaN inhibitor was applied to cultured primary cortical neurons. Dynamin phosphorylation at Ser774 was respectively increased to 208.6% and 188.0% of that in the kainate stimulation group when CsA and FK506 inhibited caN activity, and Ser778 was increased to 213.9% and 195.0% of that Fig. 5C. These results provide evidence that dynamin is the key substrate of CaN in regulating GluK2 endocytosis.

Fig. 5.

Dynamin serves as the substrate of CaN and participates in the regulation of GluK2 endocytosis in cultured primary cortical neurons. (A) GluK2 surface localization increases when dynamin activity is blocked by dynasore. The surface/total ratios are normalized to that of the control (without kainate stimulation) and expressed as the mean $\pm$ SD (n = 3). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no dynasore in kainate stimulation. (B) Dynamin binding with GluK2 increases and is decreased by CaN during GluK2 endocytosis. The dynamin-GluK2 interactions are normalized to that of the control (without kainate stimulation) and expressed as the mean $\pm$ SD (n = 3). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. Neuron cultures for replicates were made from different animals. (C) Dynamin1 is dephosphorylated at Ser774 and Ser778 by calcineurin. The phosphorylation levels are normalized to that of the control (without kainate stimulation) and expressed as the mean $\pm$ SD (n = 5). *P 0.05 vs. no kainate. ${}^{\mathrm{\#}}$P 0.05 vs. no inhibitor in kainate stimulation. Neuron cultures for replicates were made from different animals.