PERK regulates G_q protein-coupled intracellular Ca²⁺ dynamics in primary cortical neurons

Abstract

PERK (EIF2AK3) is an ER-resident eIF2α kinase required for behavioral flexibility and metabotropic glutamate receptor-dependent long-term depression via its translational control. Motivated by the recent discoveries that PERK regulates Ca²⁺ dynamics in insulin-secreting β-cells underlying glucose-stimulated insulin secretion, and modulates Ca²⁺ signals-dependent working memory, we explored the role of PERK in regulating G_q protein-coupled Ca²⁺ dynamics in pyramidal neurons. We found that acute PERK inhibition by the use of a highly specific PERK inhibitor reduced the intracellular Ca²⁺ rise stimulated by the activation of acetylcholine, metabotropic glutamate and bradykinin-2 receptors in primary cortical neurons. More specifically, acute PERK inhibition increased IP₃ receptor mediated ER Ca²⁺ release, but decreased receptor-operated extracellular Ca²⁺ influx. Impaired G_q protein-coupled intracellular Ca²⁺ rise was also observed in genetic Perk knockout neurons. Taken together, our findings reveal a novel role of PERK in neurons, which is eIF2α-independent, and suggest that the impaired working memory in forebrain-specific Perk knockout mice may stem from altered G_q protein-coupled intracellular Ca²⁺ dynamics in cortical pyramidal neurons.

Introduction

Calcium (Ca²⁺) serves as an important second messenger in the central nervous system, as it regulates various neuronal processes including neurotransmitter release, synaptic plasticity, neuron excitability, and neuronal gene transcription [1]. Initiators of intracellular Ca²⁺ rise in neurons include the G_q-protein coupled receptors, whose activation upon agonist binding leads to the activation of G_q/phospholipase C (PLC) pathway. Activated PLC hydrolyzes phosphatidylinositol 4,5-bisphosphate (PIP₂) resulting in the generation of inositol 1,4,5-triphosphate (IP₃) and diacylglycerol (DAG). While the increased cytosol IP₃ induces internal Ca²⁺ release by binding with ER resident inositol-1,4,5-triphosphate receptor (IP₃R), the activation of G_q/PLC cascade further stimulates receptor-operated Ca²⁺ influx from external space.

The central nervous system expresses a variety of G_q protein-coupled receptors including the M1 and M3 muscarinic acetylcholine receptors (mAChR), the group 1 metabotropic glutamate receptor (mGluR1), and the bradykinin-2 receptor [2, 3]. G_q protein-coupled intracellular Ca²⁺ dynamics has been shown to play prominent roles in a number of neuronal processes. For example, in pyramidal neurons, G_q protein-coupled intracellular Ca²⁺ rise induced by mAChR or mGluR1 activation is required for the induction of Ca²⁺-activated nonselective cationic current (I_CAN) [4, 5], which is considered the ionic mechanism underlying persistent neuronal firing essential for working memory. In cerebellar Purkinje cells, mGluR1-induced postsynaptic Ca²⁺ rise in dendritic spines has been shown to be essential for long-term synaptic depression [6].

PERK, an eIF2α kinase well known for its role in eIF2α-dependent protein synthesis and translational control, has been shown to modulate Ca²⁺ dynamics-dependent working memory [7], in addition to protein synthesis-dependent cognitive functions including learning and memory flexibility [8], and long-term depression mediated by mGluR1 [9]. Moreover, it has been demonstrated that PERK acutely regulates Ca²⁺ dynamics in β-cells underlying glucose-stimulated insulin secretion, which is independent on protein synthesis [10]. Considering the critical role of Ca²⁺ signaling in working memory, we hypothesized that PERK may also regulate Ca²⁺ dynamics in pyramidal neurons. We show herein that acute PERK inhibition impairs G_q protein-coupled intracellular Ca²⁺ rise. More specifically, PERK inhibition enhances IP₃R mediated ER Ca²⁺ release, but suppresses receptor-operated external Ca²⁺ influx. G_q protein-coupled intracellular Ca²⁺ rise is also impaired in genetic Perk knockout neurons. Taken together, our findings suggest that PERK regulates G_q protein-coupled Ca²⁺ dynamics in pyramidal neurons, which may serve as the cellular mechanism underlying impaired working memory in forebrain-specific Perk knockout mice.

Methods

Reagents

PERK inhibitor GSK2606414 was a kind gift from Jeffery M. Axten and Rakesh Kumar, GlaxoSmithKline, Collegeville, PA. (S)-3,5-Dihydroxyphenylglycine (DHPG) and thapsigargin were purchased from Tocris, bradykinin acetate salt was purchased from Sigma, carbachol was purchased from EMD Millipore, and Bt₃-Ins(1, 4, 5)P₃/AM (IP₃-AM) was purchased from SiChem.

Primary neuron culture

Wild-type primary cortical neurons were prepared from wild-type mice colony in C57BL/6 J background as previously described [7]. Briefly, the cerebral cortex was isolated from day 0 pups and dissociated in 0.05 % trypsin-EDTA with DNase1 for 30 min at 37 °C (5 % CO²), followed by two washes with HBSS containing 10 % FBS and trituration in neuronal medium with DNase1. Dissociated cells were collected by centrifugation at 120Xg for 5 min and plated on glial-coated 12 mm glass coverslips in a 24-well plate at a density of approximately 150,000 cells per well. Total medium was changed on the 1^st day in vitro (DIV) and 50 % of the medium was changed on DIV 3 and DIV 5. Genetic Perk knockout neurons were prepared from Perk-floxed Nestin-Cre mice in the same way, and the genotype of each pup was determined later. The cells were maintained in MEM based neuronal medium containing 5 % FBS (GEMINI Bio-Products), 2 % B27 (Invitrogen), 1 mM L-Glutamine (Gibco), 20 mM D-Glucose, 2 μM Cytosine Arabinoside (AraC), 40 units/ml penicillin, 40 μg/ml streptomycin and 100 ng/ml Amphotericin B, and final pH was adjusted to 7.4 with NaHCO₃ (100 mg/500 ml). DIV 14–19 neurons were used for Ca²⁺ imaging experiments and immunocytochemistry.

Intracellular Ca²⁺ measurements and Ca²⁺ imaging data analysis

Intracellular Ca²⁺ levels were measured using the ratiometric Ca²⁺ probe Fura-2 AM (Molecular Probes). Briefly, coverslips seeded with neurons at DIV 14–19 were incubated in bath solution with 2 μM Fura-2-AM for 30 min at room temperature in the dark. Coverslips were then transferred to fresh bath for 15 min to allow the cleavage of AM esters by cellular esterases. After dye loading, the coverslips were put in a perfusion chamber mounted on Nikon TE-200-S inverted microscope with Xenon arc lamp as the fluorescence excitation source. Ratios of images with the fluorescent emission signal excited at 340 nm over 380 nm were obtained using an excitation filter wheel (340 nm/380 nm, Chroma Technology) and UV-2A filter cube (Nikon). Images were collected every 5 sec using a 20X objective and a cooled charge-couple device (CCD) camera. SimplePCI imaging software was used for the control of filter wheel and collection of data. Tyrode’s solution (123 mM NaCl, 30 mM Glucose, 25 mM HEPES, 5 mM KCl, 2 mM CaCl₂ and 1 mM MgCl₂) mimicking cerebrospinal fluid was used as bath solution and the pH was adjusted to 7.4 before each experiment. Cells were perfused with the bath solution at the constant rate of 1 drop/2 sec during Ca²⁺ imaging process, and this rate was also used for the application of different drugs as described in figure legends. Triangular shaped pyramidal neurons were selected for imaging and the soma was selected as the region of interest. Sister coverslips from 2 to 3 independent cultures were used for each experiment, and pooled data was analyzed.

Ca²⁺ imaging measurements were analyzed by calculating the area under the curve (AUC). Due to the inherent variation of primary neuron culture, a small percentage of neurons displayed high Ca²⁺ transients, which obscured the drug-stimulated intracellular Ca²⁺ rise. For this reason, basal Ca²⁺ transients over 100 sec were analyzed, and those with AUC >2 (<5 %) were excluded from the final analysis.

Western blot analysis

To determine PERK knockdown efficiency in genetic Perk knockout neurons, mouse cerebral cortex was isolated from day 0 Perk-floxed Nestin-Cre pups, and homogenized mechanically in ice-cold buffer (100 mM HEPES, 1 mM EDTA, 2 mM EGTA, 0.5 mM DTT, supplemented with 1X protease inhibitor and 1X phosphatase inhibitor cocktails; pH was adjusted to 7.0 before use) using a polypropylene pestle. Tissue lysates for western blot were prepared using RIPA buffer with 1X protease inhibitor and 1X phosphatase inhibitor cocktails. Samples were denatured by boiling in 2X Laemmli buffer for 5 min. The following primary antibodies were used in western blot analysis: monoclonal rabbit anti-PERK (Cell Signaling), monoclonal mouse anti-β-actin (GenScript).

Synapse formation analysis

Immunocytochemical detection of the presynaptic marker Synapsin 1 and dendritic marker MAP2 was used to visualize synapses. Neurons seeded on glial-coated coverslips were fixed with 4 % paraformaldehyde, 4 % sucrose in PBS for 15 min at room temperature, permeabilized with 0.3 % Triton X-100 in PBS for 10 min, and blocked with 5 % horse serum (Invitrogen) in PBS containing 0.1 % Triton X-100 for 1 h. The cells were then incubated with the mixture of following primary antibodies overnight at 4 °C: mouse anti-Synapsin 1 (Chemicon), rabbit anti-MAP2 (Millipore). Appropriate secondary antibodies conjugated with Alexa Fluor 546 or 488 (Molecular Probes) were applied for 1 h to visualize the signals. The coverslips were mounted using the anti-fade reagent with DAPI (Molecular Probes). The fluorescent images were captured at 40X with consistent fluorescent intensities using Nikon Eclipse E1000 and SPOT 5.1 software.

To compare synapse formation density between wild-type and Perk knockout neurons, Synapsin 1 positive puncta in contact with MAP2 labeled dendrites within middle basal dendrites were counted, and the dendrite length was measured using ImageJ.

Additional information regarding the protocols used in the supplemental figures is provided in (Additional file 1.

Results

Acute PERK inhibition impairs G_q protein-coupled intracellular Ca²⁺ rise in primary cortical neurons

To acutely inhibit PERK enzyme activity in primary cortical neurons, we took advantage of a highly specific inhibitor of PERK, GSK2606414 (PERKi), which acts by competing for the ATP binding domain in the catalytic site [11]. Previously we have shown that 500 nM PERKi pretreatment for 15 min sufficiently abolishes thapsigargin induced PERK activation and eIF2α phosphorylation in primary neurons [7]. In addition, it has been shown by others that 500 nM PERKi pretreatment fully inhibits PERK enzyme activity in a variety of cell types [11, 12]. Taken together, the above results suggest that 500 nM of the PERKi is effective in acute PERK inhibition.

At the gross subcellular level PERK has been shown to be expressed in both cell body and dendrites [9]. To determine if PERK is also present in the proximate region of the synapse, we examined isolated synaptoneurosomes from wild-type mouse prefrontal cortex and found that PERK was indeed present there (Additional file 2: Figure S1). Therefore PERK is available for participating in the molecular processes that govern synaptic plasticity.

To examine if acute PERK inhibition impairs G_q protein-coupled intracellular Ca²⁺ ([Ca²⁺]_i) rise, an intracellular response which is closely related to working memory, we used Fura-2 AM to visualize [Ca²⁺]_i dynamics in primary cortical neurons. We first tested if PERKi impacts cholinergic activation induced [Ca²⁺]_i rise by treating neurons with 250 μM carbachol, and found that 15 min 500 nM PERKi pretreatment significantly reduced carbachol-induced [Ca²⁺]_i rise (Fig. 1a). The acetylcholine receptor agonist, carbachol, activates both muscarinic acetylcholine and nicotinic acetylcholine receptors with the former being a G_q protein-coupled receptor, and the latter a Ca²⁺-permeable ionotropic receptor [13]. Thus the carbachol-induced [Ca²⁺]_i rise can be attributed to the stimulation of either receptor. To further examine how PERK inhibition impacts G_q protein-coupled intracellular Ca²⁺ signaling, we tested another G_q protein-coupled receptor, mGluR1, by the use of its specific agonist, DHPG. Again, acute PERK inhibition significantly suppressed 50 μM DHPG induced [Ca²⁺]_i rise (Fig. 1b). We also tested a third G_q protein-coupled receptor, the bradykinin-2 receptor, using its specific agonist, bradykinin, and found that 1 μM bradykinin induced [Ca²⁺]_i increase was almost abolished by PERKi treatment (Fig. 1c). That PERK inhibition is able to block cytosolic Ca²⁺ influx stimulated by three different G_q protein-coupled receptors strongly argues that PERKi impacts the G_q/PLC pathway at a point downstream of receptor binding. Since PERK is present in dendrites in addition to cell body, we examined if PERK also regulates G_q protein-coupled Ca²⁺ rise in proximal dendrites. Indeed, acute PERK inhibition impairs DHPG-stimulated Ca²⁺ rise in proximal dendrites (Additional file 3: Figure S2).

Fig. 1

G_q protein-coupled intracellular Ca²⁺ ([Ca²⁺]_i) rise is impaired by acute PERK inhibition. a [Ca²⁺]_i of primary cortical neurons in response to 250 μM carbachol treatment (DMSO n = 21, PI n = 17; ***p < 0.001, two-tailed student’s t-Test). b [Ca²⁺]_i of primary cortical neurons in response to 50 μM DHPG treatment (DMSO n = 36, PI n = 57; ***p < 0.001, two-tailed student’s t-Test). c [Ca²⁺]_i of primary cortical neurons in response to 1 μM bradykinin treatment (DMSO n = 25, PI n = 37; ***p < 0.001, two-tailed student’s t-Test). In all of the experiments above, cells were pretreated with 500 nM PERK inhibitor (PI) or DMSO for 15 min before recording. In the representative graph on the left, each Ca²⁺ trace represents the average of 6–11 neurons that were imaged from the same coverslip. Basal Ca²⁺ oscillation over 100 sec before treatment and drug-stimulated [Ca²⁺]_i rise over 200 sec were quantified by calculating the area under the curve (AUC). Final analysis is presented as AUC/100 sec and shown in the bar graph on the right

IP₃-AM induced [Ca²⁺]_i rise is impaired by acute PERK inhibition

Receptor-simulated activation of the G_q/PLC pathway produces IP₃ and increases [Ca²⁺]_i by induction of IP₃ receptor mediated ER Ca²⁺ release and receptor-operated Ca²⁺ influx. To determine if PERKi affects PLC activity or a downstream Ca²⁺ channel, IP₃-AM induced [Ca²⁺]_i rise was examined in PERK-inhibited neurons and DMSO controls. Cell permeable IP₃-AM induced a delayed but sustained [Ca²⁺]_i rise in DMSO controls (Fig. 2). The delayed response likely reflects the time required to remove the AM moiety by cellular esterases. Acute PERK inhibition substantially suppressed IP₃-AM induced [Ca²⁺]_i rise which does not require PLC activity, indicating that the likely target for PERKi is a downstream Ca²⁺ channel rather than PLC.

Fig. 2

IP₃-AM induced intracellular Ca²⁺ ([Ca²⁺]_i) rise is impaired by acute PERK inhibition. [Ca²⁺]_i. of primary cortical neurons in response to 1 μM IP₃-AM treatment (DMSO n = 48, PI n = 48; *** p < 0.001, two-tailed student’s t-Test). Cells were pretreated with 500 nM PERK inhibitor (PI) or DMSO for 15 min before recording. In the representative graph on the left, each Ca²⁺ trace represents the average of 12–14 neurons that were imaged from the same coverslip. Basal Ca²⁺ oscillation over 100 sec before treatment and IP₃-AM-stimulated [Ca²⁺]_i rise over 600 sec were quantified by calculating the area under the curve (AUC). Final analysis is presented as AUC/100 sec and shown in the bar graph on the right

Acute PERK inhibition increases IP₃ receptor mediated ER Ca²⁺ release

Two sources of Ca²⁺ influx contribute to G_q protein-coupled [Ca²⁺]_i increase: IP₃R mediated ER Ca²⁺ release and receptor-operated Ca²⁺ entry (ROCE) from extracellular medium. To study PERKi’s effect on internal Ca²⁺ release, we measured [Ca²⁺]_i rise upon carbachol treatment in the absence of extracellular Ca²⁺, to exclude any contribution from nicotinic acetylcholine receptor or receptor-operated Ca²⁺ channel (ROCC)-dependent Ca²⁺ influx. Cells were perfused with Ca²⁺-free bath for 100 sec before stimulation with 250 μM carbachol. Carbachol treatment in Ca²⁺-free bath triggered a transient and small [Ca²⁺]_i increase due to Ca²⁺ release from intracellular stores, which was significantly higher in PERK-inhibited neurons (Fig. 3a). The experiment was repeated using 50 μM DHPG to stimulate mGluR1 and similar result was obtained (Fig. 3b). Taken together, these results suggest that acute PERK inhibition increases IP₃R mediated ER Ca²⁺ release.

Fig. 3

Acute PERK inhibition increases IP₃ receptor mediated ER Ca²⁺ release. a [Ca²⁺]_i. of primary cortical neurons in response to 250 μM carbachol treatment in Ca²⁺ free bath (DMSO n = 29, PI = 26; * p < 0.05, two-tailed student’s t-Test). b [Ca²⁺]_i. of primary cortical neurons in response to 50 μM DHPG treatment in Ca²⁺- free bath (DMSO n = 33, PI = 39; * p < 0.05, two-tailed student’s t-Test). In both experiments, cells were pretreated with 500 nM PERK inhibitor (PI) or DMSO for 15 min before recording. Drug treatment started 100 sec after Ca²⁺- free bath perfusion. In the representative graph on the left, each Ca²⁺ trace represents the average of 8–12 neurons that were imaged from the same coverslip. Basal Ca²⁺ oscillation over 100 sec before treatment and drug-stimulated [Ca²⁺]_i rise over 20–30 sec were quantified by calculating the area under the curve (AUC), and shown in the middle and right bar graphs respectively

Acute PERK inhibition impairs receptor-operated Ca²⁺ entry, but not store-operated Ca²⁺ entry

Our observation that acute PERK inhibition impairs G_q protein-coupled [Ca²⁺]_i mobilization and increases IP₃R-dependent ER Ca²⁺ release suggests that ROCE is impaired as a result of PERKi treatment. To test this hypothesis, DHPG stimulated ROCE was examined in PERK-inhibited neurons and DMSO controls after ER Ca²⁺ depletion by the use of a SERCA pump inhibitor, thapsigargin [14]. The pretreatment with thapsigargin caused a rapid and irreversible depletion of ER Ca²⁺. Thus upon DHPG stimulation, the rise of [Ca²⁺]_i in ER Ca²⁺ depleted-neurons was largely contributed by ROCC-dependent extracellular Ca²⁺ influx. PERKi treatment significantly reduced DHPG induced [Ca²⁺]_i rise in ER Ca²⁺ depleted-neurons, indicating that ROCC-dependent extracellular Ca²⁺ influx is impaired upon PERK inhibition (Fig. 4a).

Fig. 4

Acute PERK inhibition impairs receptor-operated Ca²⁺ entry, but not store-operated Ca²⁺ entry. a [Ca²⁺]_i. of thapsigargin (TG) pretreated primary cortical neurons in response to 50 μM DHPG treatment. Cells were pretreated with 500 nM PERK inhibitor (PI) or DMSO for 15 min before recording, and perfused with 1 μM TG for 300 sec before 50 μM DHPG treatment. In the representative graph on the left, each Ca²⁺ trace represents the average of 8–9 neurons that were imaged from the same coverslip. Basal Ca²⁺ oscillation over 100 sec before treatment and DHPG-stimulated [Ca²⁺]_i rise over 500 sec were quantified by calculating the area under the curve (AUC). Final analysis is presented as AUC/100 sec and shown in the bar graph on the right (DMSO n = 37, PI n = 35; *** p < 0.001, two-tailed student’s t-Test). b Store-operated Ca²⁺ entry in primary cortical neurons. Cells were pretreated with 500 nM PI or DMSO for 15 min before recording, and perfused with 1 μM TG in Ca²⁺- free bath for 300 sec before reintroduction of 2 mM Ca²⁺. In the representative graph on the left, each Ca²⁺ trace represents the average of 9–12 neurons that were imaged from the same coverslip. Store-operated Ca²⁺ entry over 500 sec was quantified by calculating the area under the curve (AUC). Final analysis is presented as AUC/100 sec and shown in the bar graph on the right (DMSO n = 45, PI n = 36; n.s. not significant, two-tailed student’s t-Test)

Store-operated Ca²⁺ entry (SOCE) refers to cytosol Ca²⁺ influx mediated by cell membrane Ca²⁺ channels triggered by ER Ca²⁺ store depletion. Since ROCE and SOCE are two closely related processes, and store depletion is an integral component of ROCE, we next examined PERKi’s effect on SOCE in primary cortical neurons. As shown in Fig. 4a, in neurons perfused with Ca²⁺-containing bath, thapsigargin treatment only elicited a transient [Ca²⁺]_i rise, which is the result of the combined effect of thapsigargin-induced ER Ca²⁺ release and SOCE, suggesting that thapsigargin stimulation alone did not significantly induce SOCE in primary neurons. To maximally activate SOCE, we followed a “Ca²⁺ re-addition” protocol [15], where cells were treated with 1 μM thapsigargin in Ca²⁺-free bath for 300 sec to fully deplete ER Ca²⁺ and activate store-operated Ca²⁺ channels (SOCC). Subsequent reintroduction of 2 mM Ca²⁺ into the bath elicited a sustained [Ca²⁺]_i elevation, reflecting SOCC mediated Ca²⁺ influx. No difference was observed between PERK-inhibited neurons and DMSO controls (Fig. 4b), suggesting that acute PERK inhibition does not affect SOCE. Previous studies have shown that thapsigargin induced SOCE in pyramidal neurons is L-type voltage-gated Ca²⁺ channel (VGCC)-independent [16], thus L-type VGCC inhibitor was not included in the bath.

G_q protein-coupled [Ca²⁺]_i rise is impaired in genetic Perk knockout primary cortical neurons

To investigate if the impaired G_q protein-coupled [Ca²⁺]_i mobilization could be mimicked by genetic ablation of Perk, primary cortical neurons from brain-specific Perk KO (BrPKO) mice were examined. BrPKO mice were generated by crossing Perk-floxed mice [17] with the transgenic Nestin-Cre mice strain [18], which enables widespread deletion of the loxP-flanked Perk gene sequence in neurons and glial cells during embryonic stage [19, 20]. Western blot analysis confirmed almost complete knockdown of PERK in the cerebral cortex of BrPKO mice at postnatal day 0 (Fig. 5a). With the concern that knockdown of PERK may affect neuronal differentiation and synapse formation in vitro, synapse density was examined in BrPKO and wild-type primary cortical neurons by double immunofluorescence staining of the presynaptic marker Synapsin 1 and the dendritic marker MAP2 prior to examining their G_q protein-coupled [Ca²⁺]_i rise. No significant difference was observed in synapse density between genotypes (Fig. 5b). To determine if G_q protein-coupled [Ca²⁺]_i mobilization is impaired in BrPKO primary cortical neurons, mGluR1 agonist DHPG was applied, and significantly smaller DHPG-stimulated [Ca²⁺]_i rise was observed in BrPKO neurons (Fig. 5c), which is consistent with the pharmacological PERK inhibition results.

Fig. 5

G_q protein-coupled intracellular Ca²⁺ ([Ca²⁺]_i) mobilization is impaired in genetic Perk knockout primary cortical neurons. a Western blot analysis confirmed almost complete knockdown of PERK in the cerebral cortex of BrPKO mice at postnatal day 0 (BrPKO: Nestin-Cre Perk-floxed; *** p < 0.001, two-tailed student’s t-Test). b No difference in synapse density was observed between WT and BrPKO primary cortical neurons. Representative image on the left shows the immunofluorescent staining of Synapsin 1(red) and MAP2 (green) in primary cortical neurons. Synapse density quantification in the bar graph on the right represents pooled data from 3 mice per genotype (5 neurons were randomly picked for synapse density quantification per animal, n = 15 for each genotype; WT and BrPKO neurons were cultured from the pups in the same litter; n.s. not significant, two-tailed student’s t-Test). c DHPG stimulated [Ca²⁺]_i rise is impaired in genetic Perk KO primary cortical neurons. In the representative graph on the left, each Ca²⁺ trace represents the average of 8–10 neurons that were imaged from the same coverslip. Basal Ca²⁺ oscillation over 100 sec before treatment and DHPG-stimulated [Ca²⁺]_i rise over 200 sec were quantified by calculating the area under the curve (AUC). Final analysis is presented as AUC/100 sec and shown in the bar graph on the right (WT n = 44, BrPKO n = 34; *** p < 0.001, two-tailed student’s t-Test)

Discussion

Although earlier studies have demonstrated that PERK plays an important role in regulating cognitive functions including behavior flexibility [8] and mGluR1-dependent long-term depression [9], the underlying mechanisms remain unknown. Previously we showed that PERK regulates Ca²⁺ dynamics in electrically excitable pancreatic β cells [10], and modulates Ca²⁺ dynamics-dependent working memory [7], suggesting that PERK may regulate Ca²⁺ dynamics in neurons. Neuronal cytosolic Ca²⁺ rise is contributed by two major Ca²⁺ sources: internal Ca²⁺ release mediated by ER-resident IP₃R or Ryanodine receptor, and external Ca²⁺ influx mediated by voltage-dependent Ca²⁺ channel, ionotropic glutamate receptor, nicotinic acetylcholine receptor, or TRPCs [21]. PERK’s subcellular localization in the soma, dendrites and synaptoneurosomes suggests the possibility that it plays multiple roles in Ca²⁺ channel regulation. Moreover, its localization within ER membrane and primary spatial expression in soma and dendrites are functionally important for its regulation of ER-resident IP₃R, and potential regulation of TRPCs, which are localized mainly in soma and dendrites [22–24].

In this study, we investigated the role of PERK in G_q protein-coupled [Ca²⁺]_i mobilization in primary cortical neurons, and identified it as a negative regulator of IP₃R-dependent ER Ca²⁺ release and a positive regulator of receptor-operated Ca²⁺ entry. Our finding that inhibition of PERK alters Ca²⁺ dynamics within a few minutes after inhibitor application is inconsistent with the hypothesis that these effects are mediated by changes in protein translation. Moreover, it is unlikely that these observations are due to off-target effects because genetic ablation of Perk mimicked the impaired G_q protein-coupled [Ca²⁺]_i mobilization observed in pharmacologically PERK-inhibited neurons.

How then does PERK regulate these processes? We speculate that PERK’s regulation of IP₃R-dependent ER Ca²⁺ release is mediated by its regulation of calcineurin, a Ca²⁺/calmodulin-dependent protein phosphatase that negatively regulates IP₃R [25, 26]. PERK and calcineurin have been shown to physically interact, which impacts their individual enzymatic activities [27]. Moreover, in pancreatic insulin-secreting β-cells, PERK positively regulates calcineurin activity and calcineurin is a downstream mediator of PERK’s action on Ca²⁺-dependent insulin secretion [10]. These results led us to speculate that PERK might negatively regulate IP₃R activity through its positive regulation of calcineurin in pyramidal neurons.

For G_q/PLC coupled ROCE, the family of TRPC channels form nonselective receptor-operated Ca²⁺ channels [28]. A number of intracellular signals generated downstream of G_q/PLC pathway have been shown to activate TRPCs, which include increased PLC activity, generation of DAG and internal Ca²⁺ store depletion [28]. Among them, DAG is the only identified second messenger that directly gates TRPC activity. DAG has been shown to activate TRPC3/6/7 channels [29, 30] while inhibiting TRPC5 channel activity [31]. Since PERK has an intrinsic DAG kinase activity of converting DAG into phosphatidic acid [32], it is possible that PERK regulates TRPC activity by modulating intramembrane DAG levels. In addition, it is also possible that PERK regulates ROCE via its interaction with calcineurin. In neuronal PC12D cells, it has been shown that calcineurin is recruited to the TRPC6 centered multiprotein complex induced by M1 mAChR activation, and it is essential for TRPC6 dephosphorylation and M1 mAChR dissociation from the complex, suggesting that calcineurin might play a regulatory role in receptor-operated TRPC6 activation [33].

Receptor-operated and stored-operated Ca²⁺ entries are closely related: store depletion is an integral component of ROCE, and TRPCs have been suggested to be the Ca²⁺ channels involved in both processes. Although almost all of the TRPCs can be activated by store depletion [34–41], there is accumulating evidence suggesting that the regulation of TRPC3/6/7 [29, 30, 42] and TRPC4/5 [43, 44] activities can also be store depletion-independent. Our observation that acute PERK inhibition impairs ROCE but not SOCE suggests that PERK’s regulation of ROCE might be independent of internal Ca²⁺ release.

Does PERK’s regulation of G_q protein-coupled [Ca²⁺]_i mobilization play any physiological role in cognitive function? Previously we have observed significant working memory impairment in forebrain-specific Perk KO mice [7], and we speculate that PERK regulates working memory via its modulation of G_q protein-coupled Ca²⁺ dynamics in pyramidal neurons. Intracellular signaling pathways initiated by muscarinic acetylcholine and metabotropic glutamate receptors are critical for working memory, since blockage of either receptor impairs working memory in animals [45–48], and activation of either receptor is sufficient to induce the Ca²⁺-activated nonselective cationic current (I_CAN) [4, 5] , which is essential for working memory. G_q protein-coupled [Ca²⁺]_i mobilization regulates working memory in several ways. First, the induced [Ca²⁺]_i rise is known to activate several Ca²⁺-dependent protein enzymes, including the phosphatase calcineurin, and the kinases CaMKII and PKC, all of which have been shown to regulate working memory capacity [49]. Secondly, the I_CAN, which is identified as the ionic mechanism underlying neuronal persistent firing [4], is G_q protein and Ca²⁺-dependent [5]. Finally, G_q protein-coupled [Ca²⁺]_i rise has direct effects on intrinsic neuronal excitability. It has been demonstrated that pharmacological activation of mGluR1 in prefrontal cortex pyramidal neurons triggers a biphasic electrical response-SK channel-dependent neuronal hyperpolarization followed by TRPC-dependent neuronal depolarization, and the amplitude of both are regulated by the extent of [Ca²⁺]_i rise [50, 51]. Taken together, we speculate that PERK may regulate working memory by modulating G_q protein-coupled [Ca²⁺]_i mobilization in pyramidal neurons.

Considering PERK’s role in eIF2α-dependent protein synthesis and translational control, it has been hypothesized that PERK’s regulation over memory flexibility and mGluR1-dependent long-term depression is eIF2α-dependent [8, 9]. However, genetic reduction of eIF2α phosphorylation by single allele phosphorylation site mutation of eIF2α [52], or knockdown of other eIF2α kinases GCN2 [53] and PKR [54], lowers the threshold for late phase long-term potentiation and facilitates long-term memory storage, a phenotype that is absent in forebrain-specific Perk knockout mice [8, 9]. Thus, it is very likely that PERK imparts additional regulation on cognition that is eIF2α-independent. This study’s discovery of PERK-dependent regulation of G_q protein-coupled Ca²⁺ dynamics in primary cortical neurons, together with the earlier finding that PERK regulates Ca²⁺ dynamics-dependent working memory [7], supports the above hypothesis. Further studies are required to elucidate the specific pathways that underlie PERK’s regulation of intracellular Ca²⁺ dynamics.

As an eIF2α kinase, how did PERK evolve to be a modulator of G_q protein-coupled Ca²⁺ dynamics in pyramidal neurons? We speculate that during early vertebrate evolution, PERK first played an eIF2α-dependent role in CNS. Given its localization on the ER, which is the major organelle for intracellular Ca²⁺ storage, and its regulation by ER/cytosolic Ca²⁺[10, 55], the constant interaction with Ca²⁺ may have provided PERK the opportunity to evolve an additional function to regulate intracellular Ca²⁺ dynamics through mechanism independent of eIF2a and protein translation. The fact that PERK is activated by ER Ca²⁺ depletion [55], and the discoveries of PERK being a negative regulator of IP₃R and a positive regulator of ROCC shown herein, fit well into this hypothesis: when ER Ca²⁺ stores are depleted under physiological responses such as activation of G_q protein-coupled receptor, PERK is activated due to Ca²⁺ dissociation from its regulatory domain in the ER, and it subsequently replenishes ER Ca²⁺ by inhibiting IP₃R mediated ER Ca²⁺ release and activating ROCE (Fig. 6).

Fig. 6

Proposed model for PERK’s regulation of G_q protein-coupled Ca²⁺ dynamics in pyramidal neurons. Upon extracellular ligand binding, G_q protein-coupled receptor is activated, which subsequently activates G_q/PLC. Activated PLC hydrolyzes PIP₂ into IP₃ and DAG. Increased cytosol IP₃ induces ER Ca²⁺ depletion by binding with ER-resident IP₃R, which may activate PERK due to Ca²⁺ dissociation from its regulatory domain in the ER. Activated PERK may then restore ER Ca²⁺ level by inhibiting IP₃R mediated ER Ca²⁺ release and activating receptor-operated Ca²⁺ entry