Nuclear BK Channels Regulate Gene Expression via the Control of Nuclear Calcium Signaling

Abstract

Ion channels are essential for the regulation of neuronal functions. The significance of plasma membrane, mitochondrial, endoplasmic reticulum, and lysosomal ion channels in the regulation of Ca²⁺ is well established. In contrast, surprisingly less is known about the function of ion channels on the nuclear envelope (NE). Here we demonstrate the presence of functional large-conductance, calcium-activated potassium channels (BK channels) on the NE of rodent hippocampal neurons. Functionally blockade of nuclear BK channels (nBK channels) induces NE-derived Ca²⁺ release, nucleoplasmic Ca²⁺ elevation, and cAMP response element binding protein (CREB)-dependent transcription. More importantly, blockade of nBK channels regulates nuclear Ca²⁺-sensitive gene expression and promotes dendritic arborization in a nuclear Ca²⁺-dependent manner. These results suggest that nBK channel functions as a molecular linker between neuronal activity and nuclear Ca²⁺ to convey the signals from synapse to nucleus and is a new modulator for synaptic activity-dependent neuronal functions at the NE level.

As transmembrane proteins, ion channels are present not only on the plasma membrane but also on intracellular membrane-bound organelles. Much is known regarding plasma membrane ion channels controlling action potential firing, synaptic transmission, and gene expression through the regulation of membrane potential and Ca²⁺ influx¹. In addition, the significance of mitochondrial², lysosomal^3,4, and endoplasmic reticulum (ER)⁵ ion channels in regulating ATP production, stress response and apoptosis via the control of Ca²⁺ signals has been well established. In contrast, despite the obvious structural and functional importance of the nucleus in gene expression and regulation, the functions of ion channels on the NE (nuclear ion channels) remain largely unknown, especially in neurons.

Recent studies have shown that the NE is a key source of nucleoplasmic Ca²⁺: Ca²⁺ in perinuclear space, the lumen between the outer and the inner nuclear membrane of NE, can be released into the nucleoplasm via nuclear inositol 1,4,5-trisphosphate receptors (IP₃R) and ryanodine receptors (RyR)^6–10. Therefore, the NE is a functional perinuclear Ca²⁺ store and is independent from the ER and cytoplasm in regulating the nucleoplasmic Ca²⁺ concentration ([Ca²⁺]_nu )^6–10. Meanwhile, numerous studies have demonstrated that nucleoplasmic Ca²⁺ is one of the most important signals in mediating neuronal activity-dependent gene expression^11,12, and is critical for several long lasting adaptive responses, including modulation of synapse formation and dendritic complexity¹³, regulation of synaptic plasticity and memory consolidation^13,14, and activity-dependent acquired neuroprotection^12,15.

As a key signal in neuronal functions, numerous fundamental questions remain such as how this NE-derived Ca²⁺ release regulated in neurons, whether it can be regulated by nuclear ion channels through changing the NE transmembrane potential, similar to the regulation of Ca²⁺ fluxes in the plasma membrane and mitochondria, which type of nuclear ion channel governs this transmembrane potential change, and whether this nuclear ion channel also regulate transcription factor activation and gene expression induced by neuronal activity. To address these questions, we applied a combination of biochemical and molecular biological methods, together with calcium imaging, voltage imaging, and patch clamp techniques in wild-type and gene knockout mice to study the function of the nuclear ion channel. We found that functional BK channels are present on the NE of hippocampal neurons. Pharmacological or genetic blockade of nBK channels directly regulates the NE-derived release of Ca²⁺, CREB phosphorylation, gene expression, and dendritic arborization, suggesting that nBK channels play a key role in neuronal activity-dependent functions and serve as new molecular link between neuronal activities, nuclear Ca²⁺ signaling, and gene expression at the NE level.

RESULTS

BK channels are expressed on the NE

To investigate the subcellular localization of BK channels in intact hippocampal neurons, we used a monoclonal antibody for the α subunit of the BK channel in immunocytochemistry¹⁶. The confocal optical sections demonstrated an intracellular ring-like labeling around the nucleus (Fig. 1a). Analysis of intensity profiles indicated that the labeling was co-localized with the NE marker lamin B, suggesting that the BK channels are localized to the NE (Fig. 1a). To confirm that this observation did not result from non-specific binding of the BK antibody, we conducted the same experiment in the neurons from BK channel α subunit-null mice (Kcnma1^–/–) mice. As shown in Fig. 1a, the BK channel antibody failed to detect any signal in the neurons from Kcnma1^–/– mice, validating its specificity. Furthermore, we used immunoelectron microscopy to investigate BK channel localization at high spatial resolution. As shown in Fig. 1b, gold particles signaling BK channels were localized on the NE (Fig. 1b), further confirming the NE localization of the BK channels. Consistent with previous findings^17,18, we also observed the mitochondrial localization of the BK channels (Fig. 1b). The negative labeling in Kcnma1^–/– mice once again validated the specificity of the BK antibody (Fig. 1b). To further verify the nuclear localization of the BK channels, we isolated nuclei from cultured hippocampal neurons. A ring-like labeling of BK channel immunofluorescence was detected in the isolated nuclei from wild-type mice (Fig. 1c). Analysis of intensity profiles of the isolated nuclei indicated that BK channel were co-localized with lamin B (Fig. 1c). The BK channel antibody failed to detect any signal in the nucleus from Kcnma1^–/– mice (Fig. 1c). Additionally, we transfected neurons with a green fluorescent protein (GFP)-tagged BK channel construct. As shown in Fig. 1d, green fluorescence was detected around the nucleus and co-localized with lamin B, once again suggesting the NE localization of the BK channels. This interpretation was further supported by subcellular fractionation and immunoblot analysis: BK channels were detected on the nuclear fraction, as well as the plasma membrane fraction and the whole cell lysate (Fig. 1e). The purity of the nuclear fraction was validated by the presence of lamin B, and the absence of cytosolic marker GAPDH, ER marker calnexin and mitochondrial marker COX IV (Fig. 1e). The presence of immunoreactivity in the NE fraction, but not in the denuded nuclei (NE-depleted nuclei) fraction further verified the localization of the BK channel on NE (Fig. 1f). The absence of the immunobands in Kcnma1^–/– mice confirmed that the immunoreactivity arose from authentic BK channels (Fig. 1e,f).

Figure 1. BK channels expression on the NE of hippocampal neurons.

(a) The confocal immunofluorescence of BK channel and lamin B in cultured hippocampal neurons. Analysis of intensity profiles of the vertical line across the center of the nucleus was shown on the right. Over 85% of the neurons (>100 neurons) have the positive BK channel staining in NE. (b) Immunoelectron microscopy images of BK channel localization in cultured hippocampal neurons (Top). Gold particles signaling BK channels were localized on NE (arrows) and mitochondria (M) (arrowhead) in wild-type mice. Forty-six immunoelectron microscopy images were taken totally. NE–localized gold particles were observed in all the images from the wild-type mice, and there is no signal in Kcnma1^–/– mice. Scale bar represents 1 μm. Middle, schematic diagram of the top images. Bottom, an enlargement of a square region indicated in the top images. (c) Left, immunofluorescence of BK channels and lamin B in isolated nuclei. Right, analysis of intensity profiles of the horizontal line across the center of the isolated nuclei. (d) Immunofluorescence of BK-GFP and lamin B. (e) Western blot results of the subcellular localization of the BK channel in the membrane fraction (M), the whole cell lysate (W), and the nuclear fraction (N). (f) Western blot results of the BK channel expression in intact isolated nuclei (1), denuded nuclei (2) and NE (3). (g) Single channel activities of the BK channel in isolated nuclei with 10 μM Ca²⁺ at different voltage in control group (left), with paxilline treatment (10 μM, middle), and in the nuclei from Kcnma1^–/– mice (right) (n = 6). (h) Left, single channel activities recorded in the solutions with 10 μM or 5 μM Ca²⁺ at 40 mV. Right, BK channel open probability (P_O) recorded in the solutions with 10 μM (n = 10 for each group, unpaired t-test, P = 2.30 × 10^–14, t₁₈ = 21.71 between control and paxilline) or 5 μM Ca²⁺ (n = 10 for each group, unpaired t-test, P = 1.80 × 10^–8, t₁₈ = 9.55, between control and paxilline) at 40 mV (error bars are mean ± s.d.). (i) I-V curve of the channel activity. For single channel recording, BK channel activity was recorded from 15 of the 20 successful patching of the nuclei from wild-type mice. All the data are from 3 independent cultures from at least 3 litters. All the experiments of a–f were successfully repeated for at least 3 times. WT, wild-type. Scale bars in a,c,d represent 10 μm. The full-length blots for e and f are presented in Supplementary Figure 7.

Besides the pore-forming α subunit, BK channels also contain auxiliary β subunits. The brain-specific auxiliary subunits β4 were also observed and co-localized with lamin B (Supplementary Fig. 1), suggesting that a fully assembled BK channel complex is localized on NE. To further determine whether the nuclear BK channels (nBK channels) are functional and exhibit similar electrophysiological properties to the plasma membrane BK channels (pmBK channel), we performed single channel recordings with patch clamp on isolated nuclei. We recorded a voltage-dependent and Ca²⁺ -dependent K⁺ channel activity with a conductance of approximately 220 ps, similar to that of neuronal pmBK channel^19–22 (Fig. 1g–i). The blockade of the channel activity with paxilline, a BK channel-specific blocker, and the absence of the channel activity in Kcnma1^–/– mice further confirmed that the channel activity arose from BK channels (Fig. 1g,h). Taken together, these observations demonstrate that BK channels are present and functional on the NE of hippocampal neurons.

nBK channels regulate nuclear Ca²⁺ concentrations

pmBK channels exhibit significant abilities in controlling the plasma membrane potential and accordingly regulating the voltage-dependent calcium influx^19–22. nBK channels, with similar electrophysiological properties to pmBK channels, may also regulate the nuclear transmembrane potential (ΔΨ_n), thereby regulating nuclear calcium concentration. To address this hypothesis, we first examined whether nBK channels could regulate the ΔΨ_n. Isolated nuclei were loaded with DiOC₆(3), which has been validated as a potentiometric probe in NE^23,24. Paxilline elicited a consistent increase in DiOC₆(3) fluorescence (0.28 ± 0.02 ΔF/F₀ (mean ± s.e.m.), n = 20), indicating that the perinuclear lumen became more negative^23,24 (Fig. 2a). Consistently, the paxilline-induced ΔΨ_n was completely absent in the nuclei from Kcnma1^–/– mice (0.02 ± 0.01 ΔF/F₀ (mean ± s.e.m.), n = 31, unpaired t-test, P = 2.81 × 10^–13, t₄₉ = 9.90, compared with wild-type group) (Fig. 2a). These results suggest that nBK channels can regulate ΔΨ_n.

Figure 2. nBK channels regulate the nuclear transmembrane potential and nuclear Ca²⁺ concentration in intact neurons and isolated nuclei.

(a) Fluorescence intensity changes of DiOC₆(3)-loaded isolated nuclei (insert) upon addition of paxilline. (b) Confocal nuclear calcium imaging with Fluo-4/AM of the mid-nuclear region in hippocampal neurons. (c) The changes of fluorescence intensity in nuclear calcium imaging of intact neurons after indicated treatments (left). Right, summary of the data. One-way ANOVA (n = 15 for each group, P = 4.70 × 10^–7, F_4,70 = 11.15) and post hoc test. (d) Fluorescence intensity changes in the nuclei of digitonin-permeabilized neurons after paxilline treatment. (e) Paxilline-induced fluorescence intensity changes in intact neurons from wild-type and Kcnma1^–/– mice (n = 8 for each group, unpaired t-test, P = 2.56 × 10^–5, t₁₄ = 6.14, points with error bars represent the mean ± s.e.m.). Paxilline-induced changes in Fluo-4/AM (f) or Fluo-4/dextran (g) fluorescence intensity in the isolated nuclei. (h) The changes of fluorescence intensity in isolated nuclei after indicated treatments (left). Right, summary of the data. One-way ANOVA (n = 11, 8, 12, 8, 10, 10, respectively, P = 2.50 × 10^–13, F_5,53 = 26.55) and post hoc test. (i) The changes of fluorescence intensity in isolated nuclei after knockdown of RyRs (left). Right, summary of the data. One-way ANOVA (n = 5, 6, 6, 5, respectively, P = 0.0016, F_3,18 = 7.72) and post hoc test. Fisher's least significant difference test was used for post hoc test in one-way ANOVA. Error bars represent the mean ± s.e.m.. All the data are from 3 independent cultures from at least 3 litters. All the experiments were successfully repeated for at least 3 times. **P < 0.01. WT, wild-type; Pax, paxilline; RR, ruthenium red; TG, thapsigargin; shRNA, RyRs-specific shRNAs; N.S., non-silencing shRNAs. Scale bars in a,b,f,g represent 5 μm.

Next, we investigated whether nBK channels can regulate [Ca²⁺]_nu in hippocampal neurons. Paxilline is membrane permeable and inhibits BK channels on both the plasma membrane and intracellular organelles. IbTx, a membrane-impermeant BK channel blocker, only inhibits pmBK channels on the plasma membrane. Calcium imaging of the confocal mid-nuclear section (Fig. 2b) demonstrated that paxilline induced [Ca²⁺]_nu elevation in intact neurons, whereas IbTx lacked an influence on [Ca²⁺]_nu (Fig. 2c). Moreover, paxilline elicited [Ca²⁺]_nu elevation even after complete blockade of the pmBK channels with a high concentration of IbTx (100 μM)²⁰ (Fig. 2c), which suggests that the effect of paxilline was mainly mediated by intracellular BK channels, but not pmBK channels. This idea was further supported by the results that paxilline induced [Ca²⁺]_nu elevation in digitonin-permeabilized neurons with a Ca²⁺-buffered intracellular medium (Fig. 2d). Paxilline-induced [Ca²⁺]_nu elevation was largely suppressed by prior depletion of intracellular Ca²⁺ store with thapsigargin, but not by EGTA-buffered extracellular Ca²⁺ (Fig. 2c), further suggesting that the nucleoplasmic Ca²⁺ transient is likely due to direct Ca²⁺ release from NE^{8, 24}. Importantly, neurons from Kcnma1^–/– mice were completely devoid of paxilline-induced [Ca²⁺]_nu elevation (Fig. 2e). Taken together, these results strongly supported the idea that the intracellular BK channels and perinuclear Ca²⁺ store are responsible for eliciting nuclear Ca²⁺ signals.

To further verify that the mediators of the paxilline-induced [Ca²⁺]_nu elevation are the nBK channels and perinuclear Ca²⁺ store, isolated nuclei from cultured hippocampal neurons were loaded with either the Ca²⁺ probe Fluo-4/AM or Fluo-4/dextran. Fluo-4/AM, a membrane-permeant probe, preferentially accumulated in the NE (Fig. 2f), whereas Fluo-4/dextran, a membrane-impermeant probe, was distributed uniformly in the nucleoplasm^8,10,24(Fig. 2g). Confocal microscopy revealed that paxilline produced opposite Ca²⁺ changes in the NE and the nucleoplasm. Application of paxilline led to a decrease of NE Ca²⁺ (Fig. 2f), which paralleled a transient Ca²⁺ increase in the nucleoplasm (Fig. 2g). The paxilline-induced Ca²⁺ transient was largely diminished by knockdown of BK channel with lentiviral shRNAs (Supplementary Fig. 2), and was completely abolished in neurons from Kcnma1^–/– mice (Fig. 2f,g, Fluo-4/AM intensity, 0.29 ± 0.03 ΔF/F₀ in wild-type group, 0.01 ± 0.02 ΔF/F₀ for Kcnma1^–/– group, n = 25 for each group, unpaired t-test, P = 1.16 × 10^–10, t₄₈ = 8.19; Fluo-4/dextran intensity, 0.33 ± 0.04 ΔF/F₀ in wild-type group, 0.03 ± 0.02 ΔF/F₀ for Kcnma1^–/– group, n = 22 for each group, unpaired t-test, P = 2.12 × 10^–8, t₄₂ = 6.89, data is presented as mean ± s.e.m.). Depletion of the perinuclear Ca²⁺ store with thapsigargin completely blocked paxilline-induced Ca²⁺ transient (Fig. 2h), suggesting that the perinuclear Ca²⁺ store are responsible for eliciting the paxilline-induced nuclear Ca²⁺ signals.

Numerous studies have demonstrated that IP₃R and RyR, located on the inner membrane of the NE, mediate the release of Ca²⁺ from the perinuclear space into the nucleoplasm^6,10. Therefore, we used specific IP₃R inhibitors (heparin and 2-APB) and a RyR inhibitor (ruthenium red) to test which receptor is responsible for the nBK channel-regulated NE Ca²⁺ release. The paxilline-induced Ca²⁺ release was attenuated by 10 μM ruthenium red and was completely blocked by 100 μM ruthenium red (Fig. 2h). Conversely, results with the IP₃R blockers indicated that neither heparin (50 μM) nor 2-APB (100 μM) affected paxilline-induced Ca²⁺ release (Fig. 2h). To further confirm that RyRs mediate paxilline-induced Ca²⁺ release, we designed two independent shRNAs to knock down RyRs expression in neurons²⁵. RyRs expression level was reduced to 56.4% and 51.5% of control by using the shRNA individually, and was reduced to 25.5% when both shRNAs were used together (Supplementary Fig. 3). As shown in Fig. 2i, knockdown of RyRs significantly compromised paxilline-induced [Ca²⁺]_nu elevation in intact neurons, providing direct evidence that RyRs mediate the paxilline-induced Ca²⁺ release. Collectively, these results indicate that blockade of nBK channel induces NE-derived Ca²⁺ release and [Ca²⁺]_nu elevation via RyRs.

nBK channels regulate nuclear Ca²⁺ signaling

Activity-dependent activation of CREB is heavily implicated in neuronal development and plasticity, long-term memory formation, and cell survival^12,15,26,27. Elevation of [Ca²⁺]_nu results in the phosphorylation of CREB on its activator site serine 133, which is often regarded as a marker for CREB-mediated transcription^28,29. Because blockade of the nBK channels induced [Ca²⁺]_nu elevation, we therefore further investigated whether the nBK channels could also induce CREB phosphorylation in isolated nuclei and intact neurons.

Using immunoblot in isolated functional nuclei, we observed that paxilline induced the phosphorylation of CREB, similar to the response to elevate free Ca²⁺ (Fig. 3a). The same results were also observed in immunofluorescence experiments (Fig. 3b,c). The role of the nBK channels in regulating CREB phosphorylation was further verified in isolated nuclei from wild-type and Kcnma1^–/– mice. As shown in Fig. 3d, paxilline induced CREB phosphorylation only in nuclei from wild-type mice, but not in nuclei from Kcnma1^–/– mice.

Figure 3. Blockade of nBK channels induces CREB phosphorylation in isolated nuclei and intact neurons.

(a) Immunoblots of isolated functional nuclei after stimulation as indicated. One-way ANOVA (n = 4 for each group; P = 1.75 × 10^–5, F_2,9 = 46.81 for p-CREB; P = 0.0018, F_2,9 = 13.91 for p-CaMKIV; P = 0.87, F_2,9 = 0.14 for p-ERK) and post hoc test. (b) Immunofluorescence of CREB phosphorylation. (c) Statistic results of b. One-way ANOVA (N = 5 for each group, P = 3.89 × 10^–6, F_2,12 = 41.85) and post hoc test. (d,f) Immunoblots of isolated functional nuclei (d) and intact neurons (f) after paxilline treatment. (e,g) Immunofluorescence intensity of CREB phosphorylation in isolated functional nuclei (e) and intact neurons (g) after treatments as indicated. One-way ANOVA (N = 5 for each group; P = 0.0093, F_3,16 = 5.40 for data in e; P = 0.00096, F_3,16 = 9.08 for data in g) and post hoc test. (h) CaMKIV expression after shRNA transfection. (i) Left, paxilline-induced CREB phosphorylation in the neurons transfected with GFP-tagged plasmids encoding CaMKIV-specific shRNAs or non-silencing shRNAs (N.S.), respectively. Right, summary of the data. One-way ANOVA (N = 5 for each group, P = 0.0021, F_3,16 = 7.67) and post hoc test. (j) Paxilline-induced CREB phosphorylation after treatments as indicated. One-way ANOVA (N = 5 for each group, P = 5.64 × 10^–6, F_4,20 = 15.76) and post hoc test. Fisher's least significant difference test was used for post hoc test in one-way ANOVA. Error bars represent the mean ± s.e.m.. All the data in a are from 3 independent cultures from at least 3 litters. All the other data are from 5 coverslips of cells from 3 independent cultures from at least 3 litters. All the experiments were successfully repeated for at least 3 times. **P < 0.01; n.s., no significant difference. WT, wild-type; N.S., non-silencing shRNAs. Scale bars in b,h represent 10 μm. The full-length blots for a, d, f and h are presented in Supplementary Figure 7.

CREB is a target for two Ca²⁺-regulated signaling pathways, the Ca²⁺/calmodulin (CaM)-dependent protein kinases, in particular the nuclear CaM kinase IV (CaMKIV)^28,29, and the MAP kinase/extracellular signal-regulated kinase (ERK) cascade³⁰. As shown in Fig. 3a and d, the nuclear CaMKIV were activated (phosphorylation in Thr196) by paxilline, as well as by the increase of the free Ca²⁺ concentration, suggesting the involvement of CaMKIV in paxilline-induced CREB phosphorylation (Fig. 3a). In contrast, due to the lack of cytoplasmic and plasma membrane-associated proteins in isolated nuclei, ERK1/2 were not activated by paxilline (Fig. 3a,d). In addition, paxilline-induced CREB phosphorylation in isolated nuclei was blocked by the selective CaMKIV pathway (CaM kinase kinase (CaMKK)) inhibitor STO-609, but not by the ERK pathway inhibitor U0126 (Fig. 3e). These results indicate that paxilline-induced CREB phosphorylation is likely catalyzed by nuclear CaM kinases, but not ERK, which is consistent with studies showing that CaMKIV mediates [Ca²⁺]_nu elevation-induced CREB phosphorylation³¹.

To further confirm the roles of nuclear Ca²⁺ and CaMKIV, we conducted similar experiments in intact hippocampal neurons. Similar to the results in isolated nuclei, paxilline induced CREB phosphorylation even after pmBK channel blockade by IbTx pretreatment (Fig. 3f and Supplementary Fig. 4). This event was counteracted by STO-609, but not by U0126 (Fig. 3g). Additionally, to provide the direct evidence for the role of CaMKIV, we used shRNAs to knock down the CaMKIV expression (Fig. 3h). As shown in Fig. 3i, CaMKIV-specific shRNAs significantly inhibited paxilline-induced CREB phosphorylation, further confirming the critical role of CaMKIV in paxilline-induced pCREB. Furthermore, we specifically blocked nuclear Ca²⁺ signaling by overexpressing the nuclear-targeted Ca²⁺ buffering protein parvalbumin (PV-NLS)³² and kinase-inactive form of CaMKIV (CaMKIVK75E), which functions as a dominant negative mutant of CaMKIV (dnCaMKIV)^12,13. As shown in Fig. 3j, the induction of CREB phosphorylation by paxilline was reduced in hippocampal neurons transfected with PV-NLS or dnCaMKIV, but not in U0126-treated neurons. Collectively, these results indicate that blockade of nBK channels induces CREB phosphorylation in a nuclear Ca²⁺/CaMKIV-dependent manner.

nBK channels regulate activity-evoked gene expression

Synaptic excitation on the dendritic spines of pyramidal neurons induces transcription factor activation and gene expression in the nucleus. This excitation-transcription (E-T) coupling is critically important for a broad range of processes and behaviors, including neuronal development, learning and memory, and cell survival^12,29,31. To identify the role of nBK channels in E-T coupling, we investigate the effects of the nBK channels on the regulation of synaptic activity-evoked CREB phosphorylation and gene expression.

Hippocampal neurons were preincubated with IbTx to exclude the involvement of pmBK channels (Supplementary Fig. 4) and were stimulated with the GABA_A receptor antagonist, bicuculline (50 μM). GABAergic interneurons exert a tonic inhibition onto the network. Removal of GABAergic inhibition with bicuculline leads to enhancement of neuronal activity and induces robust [Ca²⁺]_nu elevation and CREB phosphorylation^12,31 (Fig. 4a). Bicuculline-induced [Ca²⁺]_nu elevation and CREB phosphorylation are highly dependent on the filling of the perinuclear Ca²⁺ store, and a reduction of the Ca²⁺ store will significantly compromise the bicuculline-induced effects³¹. As shown in Fig. 4a,b, paxilline pretreatment attenuated both bicuculline-induced [Ca²⁺]_nu elevation and CREB phosphorylation in wild-type mice because paxilline induced the reduction of the perinuclear Ca²⁺ store content in advance (Fig. 4a,b and Fig. 2). In contrast, paxilline failed to compromise bicuculline-induced [Ca²⁺]_nu elevation and CREB phosphorylation in the neurons from Kcnma1^–/– mice (Fig. 4a,b).

Figure 4. nBK channels regulate synaptic activity-evoked [Ca²⁺]_nu elevation, CREB phosphorylation and gene expression.

(a) Bicuculline-induced changes of fluorescence intensity in nuclear calcium imaging of intact neurons after DMSO or paxilline pretreatment. (b) Left, bicuculline-induced CREB phophorylation in intact neurons with or without paxilline pretreatment. Scale bar represents 20 μm. Right, summary of the data. One-way ANOVA (N = 5 for each group; P = 4.98 × 10^–5, F_2,12 = 25.29 for wild-type group, P = 3.15 × 10^–5, F_2,12 = 27.76 for Kcnma1^–/– group) and post hoc test. (c) Left, mRNA level of c-fos after treatments as indicated. One-way ANOVA (P = 0.0015, F_2,6 = 23.34 for wild-type group; P = 0.90, F_2,6 = 0.11 for Kcnma1^–/– group) and post hoc test. Right, bicuculline-induced c-fos expression after DMSO or paxilline pretreatment (unpaired t-test, P = 0.0053, t₄ = –5.50 for control group; P = 0.0093, t₄ = 4.70 for bicuculline group). (d) Left, mRNA level of Npas4 after treatments as indicated. One-way ANOVA (P = 0.00090, F_2,6 = 28.02 for wild-type group; P = 0.99, F_2,6 = 0.00999 for Kcnma1^–/– group) and post hoc test. Right, bicuculline-induced Npas4 expression after DMSO or paxilline pretreatment (unpaired t-test, P = 0.0035, t₄ = –6.16 for control group; P = 0.0082, t₄ = 4.87 for bicuculline group). (e) Left, mRNA level of Atf3 after treatments as indicated. One-way ANOVA (P = 0.00048, F_2,6 = 35.38 for wild-type group; P = 0.90, F_2,6 = 0.11 for Kcnma1^–/– group) and post hoc test. Right, bicuculline-induced Atf3 expression after DMSO or paxilline pretreatment (unpaired t-test, P = 4.29 × 10^–5, t₄ = –19.25 for control group; P = 0.0023, t₄ = 6.88 for bicuculline group). (f) Left, mRNA level of Btg2 after treatments as indicated. One-way ANOVA (P = 0.018, F_2,6 = 8.36 for wild-type group; P = 0.16, F_2,6 = 0.16 for Kcnma1^–/– group) and post hoc test. Right, bicuculline-induced Btg2 expression after DMSO or paxilline pretreatment (unpaired t-test, P = 0.0076, t₄ = –4.98 for control group; P = 0.0057, t₄ = 5.40 for bicuculline group). (g) Left, mRNA level of Bcl6 after treatments as indicated. One-way ANOVA (P = 0.0011, F_2,6 = 25.61 for wild-type group; P = 0.93, F_2,6 = 0.074 for Kcnma1^–/– group) and post hoc test. Right, bicuculline-induced Bcl6 expression after DMSO or paxilline pretreatment (unpaired t-test, P = 0.0034, t₄ = –6.21 for control group; P = 0.0040, t₄ = 5.96 for bicuculline group). (h) Left, mRNA level of Ifi202b after treatments as indicated. One-way ANOVA (P = 0.011, F_2,6 = 10.45 for wild-type group; P = 0.93, F_2,6 = 0.075 for Kcnma1^–/– group) and post hoc test. Right, bicuculline-induced Ifi202b expression after DMSO or paxilline pretreatment (unpaired t-test, P = 0.0094, t₄ = –4.69 for control group; P = 0.0030, t₄ = 6.44 for bicuculline group). N = 3 independent cultures from at least 3 litters for each group in c–h. Fisher's least significant difference test was used for post hoc test in one-way ANOVA. Error bars represent the mean ± s.e.m.. **P < 0.01; n.s., no significant difference. WT, wild-type.

To determine whether the effects of nBK channels blockade extended to differences in activity-evoked gene transcription, we performed Real Time quantitative Reverse Transcription PCR (qRT-PCR) to analyze the expression of activity-dependent and nuclear Ca²⁺-sensitive genes^12,33,34 after IbTx preincubation. Indeed, paxilline induced increases in c-fos, Npas4, Atf3, Btg2, Bcl6 and Ifi202b expression (Fig. 4c–h), but not GADD45β and GADD45γ expression (data not shown), during basal neuronal activity. Paxilline-induced genes expression was blocked by STO-609 and was completely abolished in the neurons from Kcnma1^–/– mice (Fig. 4c–h). Similar to the effects on bicuculline-induced [Ca²⁺]_nu and CREB phosphorylation, paxilline pretreatment significantly attenuated bicuculline-induced elevation of the genes expression (Fig. 4c–h). Taken together, these results indicate that the nBK channels play an important role in regulating synaptic activity-evoked CREB phosphorylation and gene expression in E-T coupling.

nBK channels regulate dendritic arborization

Dendritic arborization is critically important for neuronal development and function, and the regulation of dendritic geometry is highly dependent on E-T coupling¹³. Given that the nBK channel is important in controlling nuclear Ca²⁺ signaling (Fig. 2 and 3) and gene expression (Fig. 4) in E-T coupling, we further explored the roles of nuclear Ca²⁺ signaling and the nBK channels in the regulation of dendritic arborization.

To first determine the role of nuclear Ca²⁺ signaling in the regulation of dendritic arborization in basal condition, we expressed CaMBP4 or dnCaMKIV in hippocampal neurons. CaMBP4 contains four repeats of the M13 CaM binding peptide and can effectively inactivate the nuclear Ca²⁺/CaM complex and block genomic responses induced by nuclear Ca²⁺ signaling^12,13,15. Consistent with the previous study¹³, morphological analyses showed that the expression of CaMBP4 or dnCaMKIV could reduce the dendritic complexity assessed by Sholl analysis and the dendritic length (Fig. 5a,b), suggesting that nuclear Ca²⁺ signaling is important for the basal dendritic arborization.

Figure 5. nBK channels regulate dendritic arborization via nuclear Ca²⁺/CaMKIV signaling.

(a) Representative micrographs of hippocampal neurons transfected with an expression vector for GFP or cotransfected with expression vectors for GFP and CaMBP4 or for GFP and CaMKIVK75E (dnCaMKIV), with DMSO or paxilline treatment. (b) Sholl analysis (left) and quantification of total dendritic length (right) in hippocampal neurons treated as indicated. Right, one-way ANOVA (P = 0.0070, F_2,57 = 5.42 for DMSO group; P = 6.22 × 10^–8, F_2,57 = 22.51 for paxilline group) and post hoc test; unpaired t-test (P = 0.010, t₃₈ = –2.70) for comparison between vector groups after DMSO or paxilline treatment. (c) Representative micrographs of hippocampal neurons transfected with GFP-tagged plasmids containing BK channel-specific shRNAs or non-silencing shRNAs, with DMSO or STO-609 treatment. (d) Sholl analysis (left) and quantification of total dendritic length (right) in hippocampal neurons treated as indicated. Right, unpaired t-test (P = 0.00011, t₃₈ = 4.32 for non-silencing shRNAs group; P = 0.0015, t₃₈ = –3.42 for BK channel-specific shRNAs group; P = 1.17 × 10^–9, t₃₈ = 7.99 for comparison between DMSO groups after non-silencing shRNAs or BK channel-specific shRNAs transfection). (e) Representative micrographs of hippocampal neurons transfected with GFP-tagged plasmids containing gene-specific shRNAs or non-silencing shRNAs, with DMSO or paxilline treatment. (f) Sholl analysis (left) and quantification of total dendritic length (right) in hippocampal neurons treated as indicated. Right, one-way ANOVA (P = 6.80 × 10^–8, F_7,152 = 7.64) and post hoc test. For the data in b,d,f, n = 20 cells from 4 independent cultures from at least 4 litters for each group. Fisher's least significant difference test was used for post hoc test in one-way ANOVA. Error bars represent the mean ± s.e.m.. Scale bar = 30 μm. *P < 0.05, **P < 0.01. N.S., non-silencing shRNAs.

To further investigate the role of nBK channels in the regulation of dendritic arborization, hippocampal neurons were pretreated with IbTx to exclude the involvement of pmBK channel (Fig. 5 and Supplementary Fig. 4). Morphometric analyses revealed that neurons treated with paxilline showed a significant increase both in the complexity of the dendrites (Fig. 5a) and in the total dendritic length (Fig. 5b). Importantly, this paxilline-induced increase was reversed by the expression of CaMBP4 or dnCaMKIV (Fig. 5a,b), suggesting that paxilline-induced increase in dendritic complexity and length is also dependent on the nuclear Ca²⁺ signaling. Similarly, neurons transfected with BK channel-specific shRNAs showed a significant increase both in the dendritic complexity and in the total dendritic length (Fig. 5c,d). STO-609 suppressed basal dendritic arborization and reversed the shRNA-induced increase, once again suggesting that both basal and paxilline-induced dendritic arborizations are dependent on nuclear Ca²⁺ signaling. To determine the specific genes mediating paxilline-induced morphological changes, we used shRNAs to knock down the expression of nBK channel-target genes (Supplementary Fig. 5). As shown in Fig. 5e,f and Supplementary Fig. 6, only shRNA against Npas4 significantly reduced dendritic arborization, and reversed paxilline-induced morphological changes, suggesting the involvement of Npas4 in both basal and paxilline-induced dendritic arborization. Taken together, these results indicate that nBK channels can regulate dendritic arborization in a nuclear Ca²⁺/CaMKIV signaling-dependent manner.

DISCUSSION

In this study, we demonstrate for the first time to our knowledge that functional BK channels are expressed on the NE of hippocampal neurons and that the activities of nBK channels regulate synaptic activity-triggered nuclear Ca²⁺ signal, CREB activation, gene expression, and dendritic arborization. These findings reveal the important role of the nBK channel as a key regulator and molecular linker between neuronal activities, nuclear Ca²⁺ signaling, and gene expression at the NE level.

The positive expression of BK channels in the nuclear envelope was based on immunofluorescence, immunoelectron microscopy and immunoblot data in both intact cells and isolated nuclei (Fig. 1a–f). The use of Kcnma1^–/– mice further confirmed the specificity of the labeling (Fig. 1). Most importantly, excised patches of nuclear membrane confirmed the presence of the nBK channels that exhibit similar voltage-dependence, Ca²⁺-dependence, single channel conductance and pharmacological properties to the BK channels found in the plasma membrane (Fig. 1g–i). Thus, several independent lines of evidence indicate the presence of functional BK channels in the NE, supporting the notion that the BK channels have multiple sites of function, including the plasma membrane^22,35-37, mitochondria^17,18,38, and, as shown here, the NE.

Nuclear Ca²⁺ is a potent signal in regulating neuronal gene expression, and represents a key player in the dialog between synapse and nucleus^11,12,31. However, the source of nuclear Ca²⁺ has been debated for many years⁶, and the mechanisms by which nuclear Ca²⁺ is regulated by synaptic activity are incompletely understood. It has been proposed that the nuclear pore complex (NPC) is large enough to provide routes for passive diffusion of ions to cross the NE, which would imply that nuclear Ca²⁺ signaling is dependent on cytosolic Ca²⁺ changes^6,7. However, numerous evidences suggested that nucleus has an autonomous Ca²⁺ signaling system, which can generate its own independent Ca²⁺ transients^6–10. (1) NPC size is regulated by the filling state of the NE lumen. Depletion of perinuclear Ca²⁺ stores has been shown to induce a conformational change of the NPC and reduce the NPC permeability⁶. (2) The ability to obtain high-resistance gigaohm seals during patch clamping in the nucleus-attached mode, when dozens of NPCs may be encompassed by the patch pipette, suggested that the NPCs can exist in a closed state that restricts ion flow³⁹. (3) The ion movement seems not freely across NPC, because studies show that there are differences in ion concentrations between cytoplasm and nucleoplasm^40–43. (4) When HepG2 cells were stimulated with extracellular ATP, nucleoplasmic Ca²⁺ signal preceded cytosolic Ca²⁺ signal by ~450 ms, suggesting the independence of nucleoplasmic Ca²⁺ signal⁹. (5) In the present study, application of paxilline to intact neurons induces nucleoplasmic Ca²⁺ elevation without raising cytosolic Ca²⁺ concentration (Fig. 2b), suggesting that the nucleoplasmic Ca²⁺ is independent from cytoplasmic Ca²⁺ and NPC-mediated Ca²⁺ transport. Nonetheless, the existence of this autonomous nuclear Ca²⁺ signaling system does not exclude the presence of the NPC-mediated Ca²⁺ transport.

Partly due to a lack of suitable experimental methods, the ionic gradient for K⁺ across the NE remains unknown in intact cells. Recent studies in isolated nuclei loaded with K⁺-sensitive fluorescent indicators showed that the K⁺ concentration in the perinuclear space is much lower than the cytoplasm and nucleoplasm⁴¹. Therefore the changes in the nuclear K⁺ channel activity would alter the K⁺ flux across the NE and ultimately lead to the alteration of the ΔΨ_n. Indeed, our results showed that the nBK channels, with the similar electrophysiological properties to pmBK channels (Fig. 1g–i), regulate the ΔΨ_n (Fig. 2a), supporting the existence of the K⁺ gradient across the NE. Nevertheless, all these results were obtained from isolated nuclei, and need further confirmation in intact cells. It is noteworthy that the open probability of RyRs is elevated upon changes of voltage^24,44–46. Thus, it is highly probable that RyRs may be affected by paxilline-induced ΔΨ_n as revealed by the voltage-sensor DiOC₆(3) (Fig. 2a). Similar voltage-induced RyRs-mediated Ca²⁺ release were observed in pancreatic beta cells²⁴. Nonetheless, we do not exclude the existence of other voltage-sensitive pathways that directly or indirectly trigger the release of nuclear calcium²⁴.

Coupling between synaptic excitation and gene expression is critical for a broad range of long-term changes in nervous system functions. Although many intracellular molecules, such as calmodulin, MAPKs and CaMKs, have been shown to participate in synapse-to-nucleus signaling, faster modulators are also required to convey fast information-laden synaptic electrical signals^6,26. Via rapid ion channel kinetics, the nBK channel may function as a fast molecular modulator between neuronal activity and nuclear Ca²⁺ to convey the signaling from synapse to nucleus. The NE is inaccessible to analysis with patch clamp pipettes and voltage-sensitive dyes in intact neurons, which makes it difficult to measure the activity of nBK channels synchronously during neuronal activity. Therefore, the effects of neuronal activity on the activity of the nBK channel require further exploration. It is noteworthy that neuronal activity can modulate the pmBK channels in a bi-directional manner: activity-induced intracellular Ca²⁺ elevation and CaMKII activation activate pmBK channels⁴⁷, whereas activity-induced calcineurin activation has the opposite effect⁴⁸. If the same regulation also applies to nBK channels, the regulation of nBK channel activity by neuronal activity may be bi-directional as well. The development of suitable experimental systems will help to elucidate this question.

The long-term maintenance of activity-induced, functional adaptations requires nuclear Ca²⁺ signaling-dependent, CREB-mediated transcription to activate gene expression^{12,15,26–28}. Using immunofluorescence, immunoblot and qRT-PCR, we show that blockade of the nBK channel increases CREB phosphorylation and neuronal activity-dependent gene expression during basal neuronal activity, but compromises those during bicuculline-induced action potential bursting (Fig. 3 and 4). Bicuculline-induced CREB phosphorylation and gene expression need sustained, Ca²⁺ store-derived nuclear Ca²⁺ elevation, and depletion of the Ca²⁺ store will significantly compromise these effects³¹. In our studies, paxilline pretreatment induced the release of nuclear Ca²⁺ and the reduction of the Ca²⁺ store content (Fig. 2), and consequently, compromise the bicuculline-induced nuclear Ca²⁺ elevation needed for CREB phosphorylation and gene expression (Fig. 4). These results are consistent with studies proposing that filling of the nuclear Ca²⁺ store is required for synaptically evoked nuclear Ca²⁺ signaling and CREB-mediated gene expression³¹. nBK channel-modulated CREB phosphorylation and gene expression is dependent on nuclear Ca²⁺ signaling because specific blockade of nuclear Ca²⁺ signaling via dnCaMKIV, PV-NLS, and STO-609, but not via the ERK inhibitor U0126, abolished the effects of nBK channel blockade (Fig. 4). In the intact cell, we cannot exclude the possibility that Ca²⁺ release from other organelles may drive gene expression in our conditions. However, as shown in Fig. 2 and 3, the NE would be responsible for an important part of these intracellular Ca²⁺ signals.

The dendritic tree determines the neuronal connectivity and has a strong influence on synaptic transmission and integration of synaptic signals. Moreover, dendritic geometry itself is regulated by neuronal activity. Our results show here that nBK channel blockade causes a significant increase both in the total dendritic length and in the complexity of the dendrites (Fig. 5), further confirming the importance of nuclear Ca²⁺ in the regulation of dendritic arborization¹³. Knockdown of one of the nBK channel-target gene, Npas4, expression completely blocked paxilline-induced morphological changes, indicating the critical role of Npas4 in this regulation. Importantly, Npas4 also functions as a transcriptional regulator, with a key role in activity-dependent regulation of inhibitory synapse development³³, contextual memory formation³⁴, and neuronal protection^12,49. Knockdown of Npas4 significantly inhibits brain-derived neurotrophic factor (BDNF) expression^11,33, implicating the potential mechanisms by which Npas4 knockdown blocks paxilline-induced morphological changes.

By controlling nuclear Ca²⁺ signaling, the nBK channel links neuronal activity to the regulation of CREB activation and gene expression, the control of total dendritic length and branching patterns. The blockade of the nBK channels represents a new strategy for the development of effective therapies for the reduction of dendrite complexity in aging, Alzheimer's disease and autism^11,33. These findings may also help explain the various phenotypes, such as Purkinje cell dysfunction³⁷, cerebellar ataxia^35,37, and abnormal circadian behavioral rhythms³⁶, found in BK channel global knockout mice.

METHODS Reagents and antibodies

Paxilline, IbTx, STO-609, thapsigargin, EGTA, ruthenium red, heparin, digitonin, 2-APB, U0126, bicuculline, DMSO, DAPI and all the other reagents to make solutions were from Sigma-Aldrich. Fluo-4/AM, Fluo-4/dextran, X-Rhod-1 were from Life Technologies. Antibodies specific for BK channel α-subunit (L6/60, Millipore MABN70, immunocytochemisty (ICC) 20 μg/ml, western blot (WB) 10 μg/ml, reference (Ref.) PMID 16566008), BK channel β4-subunit (Alomone Lab APC-061, ICC 1:100, Ref. PMID 21697543), Lamin B (Abcam ab16048, WB 1:1000, ICC 1:500, Ref. PMID 24265311) GAPDH (Cell Signalling 3683S, WB 1:1000, Ref. PMID 22895339), Calnexin (Cell Signalling 2433S, WB 1:1000, Ref. PMID 19815548), COX IV (Cell Signalling 4850S, WB 1:1000, Ref. PMID 19737931), phospho-CREB (Cell Signalling 9198S, WB 1:1000, ICC 1:500, Ref. PMID 17906627), CREB (Cell Signalling 9197S, WB 1:1000, Ref. PMID 20864036), phospho-CaMKIV (Santa Cruz Biotechnology sc-28443-R, WB 1:200, Ref. PMID 18559891), CaMKIV (Cell Signalling 4032S, WB 1:1000, Ref. PMID 21784978), phospho-ERK (Cell Signalling 9101S, WB 1:1000, Ref. PMID 11782377), and ERK (Cell Signalling 9102S, WB 1:1000, Ref. PMID 21669876), actin (Cell Signalling 4967S, WB 1:1000, Ref. PMID 17525163), RyR (Thermo MA3-925, WB 1:1000, Ref. PMID 1645737), Na/K ATPase (Cell Signalling 3010S, WB 1:1000, Ref. PMID 21724843) were used.

Cell culture and transfection and virus infection

Wild-type or Kcnma1^–/– mice (C57BL/6J strain) and Sprague-Dawley rats were used for cell culture. All of the experiments were conducted in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals (China) and were approved by the Southern Medical University Animal Ethics Committee. Three to four animals were housed per cage on a 12 h light, 12 h dark cycle, similar to those reported in previous publications. Pure primary hippocampal neurons from postnatal day 1-old wild-type or Kcnma1^–/– mice (C57BL/6J/129 strain) ³⁶ were cultured in Neurobasal medium with 2% B27 supplement, 0.5 mm glutamine, and 100 U/ml penicillin/streptomycin (Invitrogen, San Diego, CA). Cells were grown at 37°C in a humidified atmosphere containing 5% CO₂ and 95% air. The data in Fig. 1d,2b,2c and 3b were done in intact neurons or isolated nuclei of primary cultured hippocampal neurons from postnatal day 1 Sprague-Dawley rat. All experiments were performed on cells cultured for 14 d. DNA transfection was done on DIV 8 by using Lipofectamine 2000 (Invitrogen, San Diego, CA) as described⁵⁰. BK-GFP plasmid was gift from Dr. Xiangdong Tang^51,52. PV-NLS plasmid was gift from Dr. Michael Nathanson³². Kinase-inactive form of CaMKIV (dn CaMKIV, CaMKIVK75E) and CaMBP4 plasmids were gifts from Dr. Hilmar Bading¹². For shRNA, three different sequences were obtained from Shanghai GeneChem (shBK 1, shBK 2 and shBK 3), cloned and tested for silencing efficiency. shBK (1) and (2) (GTGATGCCAAAGAAGTTAA and GCACTTACGTACTGGGAATGT) was selected. The targeting sequences for all RyRs or RyR2 are ACATGGAGACCAAGTGCTT and GGAAAGAAGTCGATGGCAT, respectively²⁵. As control, a scramble version of this sequence was used. For the studies on dendrite morphology, neurons were analyzed 4–5 days after transfection. In lentivirus-delivery shRNA, virus was constructed and packed by Shanghai GeneChem Co.,Ltd, and infection was performed following the manual. 10⁷/ml of the virus was used to transfect the cultured hippocampal neurons. The shRNA target sequences of nBK channel-regulated gene are listed below:

Atf3 (1) 5’-TCCTAGCCTGTCAACATAATA-3’,

(2) 5’-TATTATGTTGACAGGCTAGGA-3’.

Npas4 (1) 5’-TCTGTGACTTAACGTCTTCAA-3’,

(2) 5’-GGGCGACAGTATCTACGATAT -3’.

Btg2 (1) 5’- GGACGCACTGACCGATCATTA-3’,

(2) 5’-GCTATCGCTGTATCCGTATCA-3’.

c-fos (1) 5’- GCAGATCTGTCCGTCTCTAGT-3’,

(2) 5’-GCTGAAGGCAGAACCCTTTGA-3’.

Bcl6 (1) 5’-GCACGCTAGTGATGTTCTTCT-3’,

(2) 5’-GGACATCTTGACGGACGTTGT-3’.

Ifi202b(1) 5’- GCTACAGAGAGCCAATATTAC-3’,

(2) 5’-GCTGCTCCTAACCAAATTATT-3’.

Western blot

Western blot were performed as described⁵⁰. Briefly, protein concentration was measured by BCA Protein Assay Kit (Thermo). Samples were combined with 2 × SDS loading buffer, boiled for 10 min, and loaded onto 10% or 4–20% gradient polyacrylamide-SDS gel. Proteins were then transferred to a PVDF membrane (Millipore, Billerica, MA) for 2 h at 350 mA, and membranes were incubated in Odyssey Blocking Buffer (LI-COR) for 2 h at room temperature. After incubation with primary antibodies for overnight at 4°C, the blots were washed three times in TBS containing 0.1% Tween 20 for 15 min and then incubated with peroxidase- or IRDye-conjugated secondary antibody for 1 h in TBS, 0.1% Tween 20 at room temperature. Immunoreactivity was detected by chemiluminescence with ECL reagent or LiCor imaging system.

Extraction of plasma membrane protein

For surface biotinylation, neurons were cooled on ice, washed twice with ice-cold PBS containing 1mM CaCl₂ and 0.5 mM MgCl₂, and then incubated with PBS containing 1mM CaCl₂, 0.5 mM MgCl₂, and 1 mg/ml Sulfo-NHS-SS-Biotin (Pierce) for 30 min at 4°C. Unreacted biotin was quenched by washing cells three times with ice-cold 100mM Glycine (pH7.4). Cultures were harvested in RIPA buffer. Homogenates were centrifuged at 132,000 rpm for 20 min at 4°C. The resulting supernatant volume was measured and 15% of it separated as the total protein. The remaining 85% of the homogenate was rotated overnight at 4°C with Streptavidin beads (Pierce). Precipitates were washed with RIPA buffer and analyzed by immunoblotting with each antibody.

Nuclei isolation

Cultured neurons were collected and resuspended in Buffer A (2.0 mM MgCl₂, 125 mM KCl, 10 mM HEPES (pH 7.5), protease inhibitors). After swelling on ice for 10 min, the cell membrane and cytoskeleton were disrupted by 15 brief homogenization. The solution were filtered with 3 layers of gauze, and then centrifuged with 1000g for 10min. The precipitate were resuspended in Buffer N (0.25 M sucrose in buffer A), and covered on the top of Buffer B (1.1 M sucrose in buffer A) in a centrifuge tube and centrifuged with 1000g for 10 min, at 4°C. We only performed experiments in intact and clean nuclei (free of ER contamination). All the data of isolated nuclei were from at least 3 independent cultures from at least 3 litters. The isolated nuclei were treated with Triton X100 buffer (1% Triton X-100, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 0.5 mM CaCl₂,1 mM DTT) for 7 min, and centrifuged with 650 g for 5 min. The resultant precipitate was saved as the denuded nuclei fraction and the supernatant as the nuclear envelope fraction for further analyses. Alternatively, nuclei isolation followed the previous report³¹. Nuclei were prepared by lysing hippocampal neurons at room temperature in PBS (pH 7.2) containing 0.5% (v/v) NP-40, 7.5 mM K⁺ 2EGTA, 2.5 mM Ca²⁺EGTA and 2.5 (free Ca²⁺ concentration is 60 nM). The free Ca²⁺ concentration was increased to 1 μM by adding 6.2 volumes of 10 mM Ca²⁺EGTA. After 60 to 90 s, 1.8 volumes of 4× sample buffer was added, nuclei were scraped off the dish, boiled for 7 min, and subjected to SDS-PAGE and immunoblot analyses.

Nuclear protein extraction

Nuclear protein was extracted using the PIERCE Nuclear Protein Extraction Kit (Thermo), following the manual. Briefly, cells were collected by centrifugation, 500g for 3 min at 4°C. Resuspend cells in 200μl precooled CER I by 15 sec of intense vortex, and incubate on ice for 10 min. Add 11μl precooled CER II to the solution, and after 5 sec of intense vortex, incubate it on ice for 1 min. After 5 sec of intense vortex, centrifuge the solution at 16000g for 5 min, at 4°C, discard the supernatant. Add 100μl NER to the precipitate and vibrate it intensively for 15 sec, and incubate it on ice for 10 min. Repeat this step 4 times, and then centrifuge the solution at 16000g for 10 min, at 4°C. The supernatant was purified nuclear protein.

Patch-clamp experiments

For whole-cell recordings, electrodes were with a final resistance of 5–10 MΩ when filled with pipette solution (in mM): 120.0 potassium gluconate, 15.0 HEPES, 2.2 CaCl2, 1.0 MgCl2, 4.0 EGTA, and 4.0 HEDTA, pH7.35, 305mOsm. Currents were recorded using an Axopatch 200B (Axon Instruments, Foster City, CA) amplifier at a sampling rate of 1–5 kHz and leak-subtracted on line. Series resistance never exceeded 25 MΩ and was compensated. Data were acquired and analyzed with pClamp9 (Axon Instruments). Single-channel currents were recorded from nuclei positively labeled with ethidium homodimer-1 by using standard patch-clamp recording procedures²⁴. Currents were recorded in symmetrical 135 mM K using a patch-clamp amplifier (Axopatch 200; Axon Instruments) at a sampling rate of 5 kHz and filtered at 1 kHz filter. Experiments were performed at RT. High resistance seals were formed (5–7 GΩ), indicating minimal contamination by the endoplasmic reticulum (ER) membrane²⁴. Under these circumstances, it has been proposed that nuclear pores would be occluded or nonconducting because of experimental conditions or lack of cytosolic factors^6,39,53. PCLAMP software was used to measure the current magnitude and duration.

Ca²⁺ imaging

Intact neurons were loading with Fluo-4/AM or X-Rhod-1 AM (Life Technology) in transfected neurons, in HEPPS buffer for 30 min, and washed 3 times before imaging. Single isolated nuclei were loaded as reported²⁴. The membrane impermeant Ca²⁺ probe Fluo-4/dextran (30 μg/ml; 30 min at 4°C; Life Technologies) loaded the nucleoplasm while the membrane permeant Ca²⁺ probe Fluo-4/AM (20 μM; 60 min at 4°C) loaded the nuclear envelope. After loading, the nuclei were washed twice with the intracellular medium (125 mM KCl/2 mM K₂HPO₄/40 mM Hepes/0.1 mM MgCl₂, pH 7.2/100 nM Ca²⁺, with 10.2 mM EGTA and 1.65 mM CaCl₂) and then were equilibrated in the same medium supplemented with 1 μM of ATP and 300 nM Ca²⁺ for a few minutes to load nuclei with Ca²⁺. After that, they were washed twice again with the intracellular buffer (without ATP and Ca²⁺). Experiments were done at RT. No probe leakage was detected during the experiment. Ca²⁺ imaging in single isolated nuclei was performed by using an Olympus FV1000 Confocal microscope. Images were collected at 1 s intervals, and fluorescence was measured by using the Olympus FV1000 software package. The difference (ΔF) between the mean fluorescence measured in a given region of interest (ROI) and the corresponding control value for each ROI (F ₀) was expressed as fraction of the control (ΔF/F ₀) and was plotted as a function of time.

Nuclear Transmembrane Potential (ΔΨ_n) Measurements

Isolated nuclei were equilibrated in intracellular medium with DiOC₆(3) (200 nM, RT, Life Technologies), which is a fluorescent cationic lipophyllic dye whose incorporation into lumen is proportional to ΔΨ²⁴. After 10 min, the nuclear envelope became stained. The nuclear were washed for 3 times before imaging with Olympus FV1000 Confocal microscope. Images were collected at 1 s intervals, and fluorescence was measured by using the Olympus FV1000 software package. The difference (ΔF) between the mean fluorescence measured in a given region of interest (ROI) and the corresponding control value for each ROI (F ₀) was expressed as fraction of the control (ΔF/F ₀) and was plotted as a function of time.

Immunocytochemistry

Cells were washed with PBS and fixed for 20 min with 4% paraformaldehyde at room temperature. To permeabilize cells, cells were treated with 0.3% Triton X-100 in PBS solutions. Nonspecific binding was blocked for 1 h at room temperature using 5% bovine serum. Primary antibodies were then added and incubated overnight at 4 °C. Cells were then washed and incubated for 1 h with the fluorescent-labeled secondary antibodies at room temperature, and washed for three times to remove the unbound antibodies.

Phospho-CREB Immunofluorescence

CREB phosphorylation was induced in isolated nuclei by increasing Ca²⁺ in an EGTA-buffered intracellular medium from 60 nM to 200 nM Ca²⁺ or by addition of paxilline for 3 min. Nuclei were fixed with 4% paraformaldehyde (wt/vol) and permeabilized with 0.1% Triton X-100 for 10 min. Nuclei were preincubated with blocking buffer; then anti-phospho-CREB antibodies were applied for 16 h at 4°C. Phospho-CREB was visualized with TRITC-conjugated secondary antibodies (1 h, RT). High Ca²⁺ gave the maximum response.

Morphometric Analyses

For three-dimensional Sholl analysis, total dendritic length and spine morphology were calculated by using Image-J freeware software. Briefly, a z-stack acquisition was imported, calibrated in Image-J, and manually traced. Total dendritic length was then computed. For Sholl analysis, the shell interval was set at 10 μm. All analyses were performed blind.

RNA Extraction and cDNA Synthesis

Total RNA was isolated from hippocampal primary neuron cultures with RNeasy Mini Kit (TAKARA) including an optional DNase I treatment at room temperature for 15 min according to manufacturer's instructions (TAKARA). 1.2 μg of extracted RNA was reverse transcribed into first strand cDNA using High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA).

Real Time Quantitative PCR

Quantitative reverse transcriptase PCR (qRT-PCR) was done on a Statagene Mx3000P thermal cycler using Universal qRT-PCR master mix for the indicated genes (TAKARA). The primers were designed and synthesized as follow:

c-Fos sense 5'- GCGGGAGTGGTGAAGACCAT-3',

antisense 5'- GCTTGGAGCGTATCTGTCAGCTC-3'.

Npas4 sense 5'- ACATCATGAGTCTTGCCTGCATC-3',

antisense 5'- TCCAGGTAGTGCTGCCACTATGTC-3'.

Atf3 sense 5'- ATGATGCTTCAACACCCAGGC-3'

antisense 5'- TTAGCTCTGCAATGTTCCTTC-3'

Btg2 sense 5'- GAGCGAGCAGAGACTCAAGGTT-3',

antisense 5'- CGATAGCCAGAACCTTTGGATGG-3'.

Bcl6 sense 5'- CAGAGATGTGCCTCCATACTGC-3',

antisense 5'- CTCCTCAGAGAAACGGCAGTCA-3'.

Ifi202b AAGAAAGGCTGGTTGATGGAGAG-3',

antisense 5'- GGTCAGTTGTCCAGATACCACC-3'.

GADD45β sense 5'- GGAGACATTGGGCACAACCGAA-3',

antisense 5'- CTGCTCTCTTCACAGTAACTGGC-3'.

GADD45γ sense 5'- TCTACGAGTCCGCCAAAGTCCT-3',

antisense 5'- CTCACAGCAGAACGCCTGAATC-3'.

GAPDH sense 5'-GGCACAGTCAAGGCTGAGAATG-3',

antisense 5'-ATGGTGGTGAAGACGCCAGTA-3'.

Expression of target genes was normalized against the expression of GAPDH as endogenous control gene. Data were derived from cells from 3 independent cultures from at least 3 litters.

Statistical Analysis

SPSS 13.0 was used for statistical analysis. The data were collected and processed randomly. No statistical methods were used to predetermine sample size, but our sample sizes are similar to those reported in previous publications.^50,54–57 Statistical significances were determined by comparing means of different groups using unpaired t-test or one-way ANOVA followed by Fisher's least significant difference (LSD) test. Two-sided test was used in the analysis. Data met the assumptions of the statistical tests used, and equal variances were formally tested. The criterion for excluding data points were established prior to data collection. An outlier was defined as a value outside the mean ± 3 s.d.. All the immunocytochemistry, immunoelectron microscopy imaging, sholl analysis, qRT-PCR, calcium imaging and electrophysiology experiments were carried out blind. The western blot was not blind when loading the samples, but data collection and analyses were performed blind. For the box and whisker plot, the bottom and top of the box are the first and third quartiles, and the band inside the box is the median. The whisker represents the minimum and maximum of all of the data. The data were expressed as mean ± s.e.m. if not otherwise specified.