Spontaneously active and InsP₃-activated ion channels in cell nuclei from rat cerebellar Purkinje and granule neurones

Abstract

Increases in Ca²⁺ concentration in the nucleus of neurones modulate gene transcription and may be involved in activity-dependent long-term plasticity, apoptosis, and neurotoxicity. Little is currently known about the regulation of Ca²⁺ in the nuclei of neurones. Investigation of neuronal nuclei is hampered by the cellular heterogeneity of the brain where neurones comprise no more than 10% of the cells. The situation is further complicated by large differences in properties of different neurones. Here we report a method for isolating nuclei from identified central neurones. We employed this technique to study nuclei from rat cerebellar Purkinje and granule neurones. Patch-clamp recording from the nuclear membrane of Purkinje neurones revealed numerous large-conductance channels selective for monovalent cations. The nuclear membrane of Purkinje neurones also contained multiple InsP₃- activated ion channels localized exclusively in the inner nuclear membrane with their receptor loci facing the nucleoplasm. In contrast, the nuclear membrane of granule neurones contained only a small number of mainly anion channels. Nuclear InsP₃ receptors (InsP₃Rs) were activated by InsP₃ with EC₅₀= 0.67 μm and a Hill coefficient of 2.5. Ca²⁺ exhibited a biphasic effect on the receptors elevating its activity at low concentrations and inhibiting it at micromolar concentrations. InsP₃ in saturating concentrations did not prevent the inhibitory effect of Ca²⁺, but strongly increased InsP₃R activity at resting Ca²⁺ concentrations. These data are the first evidence for the presence of intranuclear sources of Ca²⁺ in neurones. Ca²⁺ release from the nuclear envelope may amplify Ca²⁺ transients penetrating the nucleus from the cytoplasm or generate Ca²⁺ transients in the nucleus independently of the cytoplasm.

Synaptic activity can evoke a rise in Ca²⁺ concentration in the nucleus of neurones either by generating a regenerative Ca²⁺ wave, spreading from activated synapses to the cell body or by depolarizing the neurone and activating voltage-operating Ca²⁺ channels in the somatic plasma membrane (Berridge, 1998; Power & Sah, 2002). Increases in Ca²⁺ concentration in neuronal nuclei regulate gene transcription by activating nuclear Ca²⁺-sensitive kinases and phosphatases or by directly affecting Ca²⁺-binding transcription factors. The two regulatory mechanisms may be exemplified by the transcription factors cAMP responsive element binding protein (CREB) and downstream response element-antagonist modulator (DREAM), respectively (Hardingham et al. 2001; West et al. 2002). Nuclear Ca²⁺ transients have been implicated in a number of physiological phenomena including neuronal development, survival, and activity-dependent long-term neuronal plasticity (Berridge, 1998; Bading, 2000; West et al. 2002).

Little is currently known about Ca²⁺ regulation in the nuclei of neurones. In non-neuronal cells this issue has been addressed in numerous studies, which have yielded conflicting results (reviewed in Santella & Carafoli, 1997; Bootman et al. 2000). There are two potential sources of Ca²⁺ that may generate an increase inside the nucleus. Ca²⁺ ions can penetrate the nucleus from the cytoplasm through numerous nuclear pores or they can be released into the nucleoplasm from the nuclear envelope. The nuclear envelope is a flattened cistern morphologically and biogenetically related to the endoplasmic reticulum. It has been shown that the nuclear envelopes of Xenopus oocytes, hepatocytes and a number of other cells are functional Ca²⁺ stores (Nicotera et al. 1990; Gerasimenko et al. 1995; Stehno-Bittel et al. 1995). The role of the nuclear envelope in the regulation of nuclear Ca²⁺ is the main issue of the current controversy. The reason for this controversy may at least partly result from different mechanisms of Ca²⁺ regulation in the nuclei of different cells.

One of the difficulties in studying neuronal nuclei is the heterogeneous cellular composition of the nervous system. Neurones constitute no more than 10% of the total number of brain cells (Kendal et al. 2000). The standard method of nucleus isolation includes homogenization of a tissue and subsequent differential centrifugation of the homogenate (Blobel & Potter, 1966). When applied to nervous tissue this method yields a mixture where neuronal nuclei comprise a minority group. Isolation of nuclei from certain parts of the brain, such as the cerebral or cerebellar cortex, can enrich the preparation with neuronal nuclei, but does not solve the problem. Further complication arises from the extreme diversity of the neurones themselves.

We have developed a method for isolating nuclei from identified central neurones. Using this technique we have studied ion channels in the nuclei of cerebellar Purkinje and granule neurones and found that they have different properties.

Methods

Isolation of neuronal nuclei

All experimental procedures involving animals and their care were conducted in conformity with the Animals (Scientific Procedures) Act 1986 and the guidelines of the Ministry of Public Health of Ukraine. Male Wistar rats (3- to 4-weeks old) were anaesthetized with ether and decapitated. The cerebellum was rapidly removed and placed in ice-cold (1–3°C) artificial cerebrospinal fluid (ACSF), containing (mm): NaCl 120, KCl 2.5, MgSO₄ 1.3, CaCl₂ 1, NaH₂PO₄ 1.3, NaHCO₃ 26, glucose 10 and equilibrated with 95% O₂ and 5% CO₂. Coronal cerebellar slices (300 μm; see Fig. S1 of the Supplemental material available with this article online) were cut by hand or using a Vibratome (Campden Instruments, Sileby, UK). Slices were placed in a chamber on the stage of a Leica DMLFS upright microscope (Leica Microsystems, Germany) and continuously perfused with ACSF continuously oxygenated with 95% O₂ and 5% CO₂ at room temperature. Neurones were visualized using a × 40 long working distance water-immersion objective and a VX 45 Microscope CCD camera (PCO Computer Optics, Kelheim, Germany) with DIC/infrared optics.

Nuclei were collected from neurones with glass pipettes (Fig. 1A). The opening of the pipette varied depending on the size of the nucleus and was 20–30% larger than the nucleus diameter. The cytoplasm from a Purkinje or granule neurone was aspirated under visual control into the pipette until the nucleus was sucked into the pipette. This procedure was applied to several other nuclei until the collected nuclei with remains of the cytoplasm were ejected into a microtube of homogenization solution containing (mm): potassium gluconate 150, Hepes-KOH 10 (pH 7.3). Protease Inhibitor Cocktail (Roche Diagnostics UK Ltd) was added to the solution according to the manufacturer's instructions. The nuclei were cleared from remains of the cytoplasm by four to five strokes in a 2 ml Dounce tissue homogenizer (Bellco Glass, Vineland, NJ, USA). Comparison of nuclei isolated from Purkinje neurones (Fig. 1B) with nuclei isolated from the rest of the cerebellar cortex showed that because of their much larger size and distinct morphology Purkinje neurone nuclei could be easily and unequivocally identified in crude homogenates of the cerebellar cortex (Fig. 1D). This allowed us to simplify the procedure considerably. In the later stages of the work we picked out Purkinje neurone nuclei directly from homogenates of the cerebellar cortex. The homogenate was placed into a working chamber on the stage of the Leica DMIRB inverted microscope. The nuclei were allowed to attach to the glass bottom of the chamber and then washed from debris with solution (further referred to as KCl solution) containing (mm): KCl 150, Na₂ATP 0.5, K₂EGTA 0.53, CaEGTA 1.47 (calculated free Ca²⁺ concentration 250 nm) and Hepes-KOH 10 (pH 7.3). To obtain access to the inner nuclear membrane the isolated nuclei were treated with sodium citrate as follows. The homogenate was centrifuged for 15 min at 1000 g, and the resulting pellet was suspended in homogenization medium with 1% (w/v) sodium citrate added. The suspension was incubated for 30–60 min on ice while stirring gently and then used in experiments as described above.

Figure 1. Isolated nuclei from neurones of rat cerebellar cortex.

A, the method of isolating nuclei from central neurones. The cytoplasm from a neurone (here a Purkinje neurone) in a brain slice was aspirated into a glass micropipette with a relatively large opening until the nucleus was sucked into the pipette. The same pipette was used to collect nuclei from several neurones. B and C, individual nuclei isolated from a Purkinje neurone (B) and a granule neurone (C). D, a rough homogenate of the cerebellar cortex. The large nucleus from a Purkinje neurone (arrow) clearly differs from all other nuclei in the homogenate. E, a transmission electron micrograph of the periphery of the nucleus isolated from a Purkinje neurone. The nucleus had well-preserved inner (im) and outer (om) nuclear membranes.

Electrophysiology

Single ion channels were recorded from nucleus-attached and excised patches of the nuclear membrane in the voltage-clamp mode of the patch-clamp technique. Patch pipettes were prepared from borosilicate glass (Sutter Instrument, Novato, CA, USA) and filled, unless stated otherwise, with KCl solution. Pipette resistances ranged from 7 to 15 MΩ. Data acquisition was carried out using a Visual Patch VP-500 amplifier (Bio-Logic, Claix, France). Currents were filtered with a low-pass Bessel filter at 1 kHz, digitized at 5 kHz and stored on a computer disk. The indifferent electrode was an Ag–AgCl plug connected to a bath chamber via an agar bridge. In all recordings the potential of the bath solution was considered to be 0 mV. The permeability ratio P_a/P_b of ion species a and b was calculated using the following formula (see, e.g. Miedema, 2002):

where a_p, b_p and a_b, b_b refer to the activities of ion species a and b in, respectively, pipette and bath solution; z_a and z_b are valences of a and b, E_rev is the reversal potential of the single channel current, and R, T and F are standard thermodynamic parameters.

The dependence of InsP₃-activated channels on InsP₃ concentration was fitted with the Hill equation:

where P_o is the open probability of the channel at a given InsP₃ concentration x, P_max is its maximum open probability, K is half-maximum activating InsP₃ concentration (EC₅₀) and h is the Hill coefficient.

Electron microscopy

Nuclei were pelleted at 1000 g in homogenization medium. The pellet was fixed for 18 h in 2% glutaraldehyde–PBS and postfixed in 1% osmium tetraoxide–PBS, pH 7.4. Samples were dehydrated in ethanol and propylene oxide and then embedded in epon.

Results

Morphology of isolated neuronal nuclei

Nuclei isolated from Purkinje neurones had a spherical or slightly ellipsoidal shape with an average diameter of 11.8 ± 1.0 μm (n = 34; Fig. 1B). The nuclei had characteristically light nucleoplasm with a clearly visible large dark nucleolus. The nuclei from granule neurones were much smaller with an average diameter of 6.8 ± 0.7 μm (n = 78; Fig. 1C). Because of their large size and distinct morphology, nuclei from Purkinje neurones were quite different from nuclei of any other cerebellar cells and could be easily identified in rough homogenates of the cerebellar cortex (Fig. 1D). Transmission electron microscopy showed that the nuclear envelopes of isolated nuclei had a typical double-membrane structure with well-preserved outer and inner nuclear membranes (Fig. 1E). The integrity of isolated nuclear envelopes was also confirmed by their staining with fluo-3 AM (see Fig. S1 in Supplemental material). Therefore the outer nuclear membrane was readily accessible for patch-clamp recording in this preparation.

Spontaneously active ion channels in the nuclear membrane of Purkinje neurones

The majority of membrane patches (137 out of 189) from the nuclear membrane of Purkinje neurones contained large conductance ion channels. The channels were characterized by high density (usually 3–5 channels per patch), high open probability and slow kinetics (Fig. 2A). In symmetrical KCl solutions the channels had a linear current–voltage relationship (Fig. 2B, •) with a slope conductance of 198 ± 27 pS (n = 58).

Figure 2. Large conductance channel in the nuclear membrane of Purkinje neurones.

A, large conductance channels in an excised patch from the outer nuclear membrane of a Purkinje cell. The patch contained three channels; the zero current level is indicated by a dotted line. Note the slow kinetics and bursting activity of the channel. B, current–voltage relationships of the large conductance channel. In symmetrical standard KCl solution the slope conductance of the channel was 198 ± 27 pS (n = 58, •). After substitution of standard KCl bath solution with low (15 mm) KCl solution the reversal potential of the current shifted to −48.6 mV (V_rev) suggesting cationic selectivity of the channel (O).

To determine the selectivity of this channel to K⁺ and Cl⁻ we substituted the isotonic KCl solution in the bath with hypotonic solutions, which contained reduced concentrations of KCl. In these conditions the equilibrium potential for K⁺ was shifted to the left and that for Cl⁻ to the right. Consequently the current–voltage relationship of an anionic channel would shift to the right and that of a cationic channel to the left. The magnitude of the shift depends on the ratio of the channel permeability to K⁺ and Cl⁻. The reversal potential of a non-selective channel would not change.

When standard KCl bath solution was substituted with a hypotonic solution (in different experiments the concentration of KCl in the bath solution varied from 10 to 40 mm) the current–voltage relationship of the channel shifted to the left toward the reversal potential for K⁺ (Fig. 2B, O). When the pipette solution contained 155 mm K⁺ and 150 mm Cl⁻ and the bath solution contained 15 mm K⁺ and 10 mm Cl⁻ the calculated equilibrium potential for K⁺ was −59 mV. In these experimental conditions the reversal potential of the large conductance channel was −48 ± 1 mV (n = 3). The calculated ratio of the channel permeabilities for Cl⁻ and K⁺ (P_Cl/P_K) was 0.05. Therefore the large-conductance channel in the nuclear membrane of Purkinje neurones was selective to cations.

To study the permeability of the large conductance channel to other monovalent cations, K⁺ in the bath solution was replaced with an equimolar amount of the test cation (Fig. 3A). The substitution of K⁺ in the bath solution with Na⁺ shifted the reversal potential of the channels to −10.5 ± 1 mV (n = 3) indicating that the channel is permeable to Na⁺ with the permeability ratio P_Na/P_K= 0.65 (Fig. 3A, •). The slope conductance for inward current, carried by Na⁺ through the channel, was 94 pS (n = 3). The channel was also permeable to Cs⁺ (not illustrated). When K⁺ in the bath solution was replaced with the large organic monovalent cation N-methyl-d-glucamine (NMDG) no inward current through the channels was recorded (Fig. 3A, ▴). Therefore the large conductance channel was a poorly selective K⁺ channel.

Figure 3. The large conductance channel is a poorly selective K⁺ channel.

A, permeability of the large conductance channel to monovalent cations. Substitution of KCl in the bath solution with equimolar NaCl (•) shifted the reversal potential of the current through the channel to −10.4 mV suggesting that the channel is permeable to Na⁺ with P_Na/P_K= 0.65. When K⁺ in the bath solution was substituted with NMDG⁺ (▴) no inward current was observed. B, when CaCl₂ solution in the bath (▴) or in patch pipettes (•) was substituted for KCl solution no Ca²⁺ inward or outward current, respectively, was observed suggesting that the channel is impermeable to Ca²⁺.

To study the permeability of the large conductance channels to Ca²⁺ either the working chamber (n = 6) or patch pipettes (n = 4) were filled with isotonic CaCl₂ solution with standard KCl solution at the opposite side of the membrane (Fig. 3B). In both cases only K⁺ currents through the channels were recorded. In symmetric CaCl₂ solution no channel activity was observed in the range from −150 to 150 mV. These data suggest that the large conductance channel is practically impermeable to Ca²⁺. Similar experiments with BaCl₂ solutions suggest that the channels are also impermeable to Ba²⁺ (not illustrated).

The potassium channel blockers tetraethylammonium (10 mm) and 4-aminopyridine (2 mm) had no effect on either the conductance or gating of the large conductance channel. The non-selective blocker of cationic channels La³⁺ at a concentration of 10–100 μm was also ineffective. The ryanodine receptor blocker ruthenium red (10 μm), the ryanodine receptor agonist/blocker ryanodine (1–10 μm), InsP₃ (10 μm), and the InsP₃ receptor blocker heparin did not affect the large conductance cation channel. Large conductance channels were also unaffected by ATP, Ca²⁺ and Mg²⁺ at physiological concentrations (0.5–5 mm, 0.05–50 μm and 1–5 mm, respectively). At very high concentrations (100 mm) Ca²⁺ and Ba²⁺ on either side of the membrane reduced K⁺ currents through the channel (Fig. 3B).

The large conductance channels were by far the most abundant type of channel in the nuclear membrane of Purkinje cells. Nevertheless, in a small fraction of patches (11 out of 189 patches) of the nuclear membrane from Purkinje neurones a channel with smaller amplitude and very different pattern of activity was observed (Fig. 4A). The relative rarity of the channel and the abundance of the large conductance cationic channels prevented us from studying this channel in detail. The channel activity was characterized by fast fluctuations between multiple subconductance states. The slope conductances of the more frequently occurring subconductance states were 67 ± 5, 52 ± 4 and 27 ± 3 pS (n = 6). To determine the ionic selectivity of the channel, KCl in the bath solution was partly replaced with equimolar amounts of potassium gluconate. In these conditions the reversal potential of the channel shifted to positive values close to the equilibrium potential for Cl⁻ (Fig. 4B). These data suggest that the channel is selective for Cl⁻ with little if any permeability to K⁺ and gluconate.

Figure 4. Anion channels in the nuclear membrane of Purkinje neurones.

A, the channel demonstrated fast transitions between multiple subconductance states. Three of the subconductance states and the closed state (C) are shown by dashed lines. B, current–voltage relationships of the three subconductance states of the anion channel. The arrow indicates the equilibrium potential for Cl⁻ (V_Cl). Pipettes contained standard KCl solution, in bath solution 100 mm KCl was substituted with potassium gluconate.

InsP₃-activated ion channels in the nuclear membrane of Purkinje neurones

To record InsP₃-activated channels the patch pipette or the bath was filled with KCl solution containing 10 μm InsP₃. This solution also contained Na₂ATP and free Ca²⁺ in concentrations reported to be optimal for cerebellar InsP₃R stimulation (0.5 mm and 250 nm, respectively; Bezprozvanny et al. 1991). Patch-clamp recording from the outer nuclear membrane of Purkinje neurones has revealed the presence of large-conductance spontaneously active ion channels, but InsP₃-containing solution did not activate any additional ion channels in excised as well as nucleus-attached nuclear patches (n≫ 153).

To access the inner nuclear membrane isolated nuclei were treated with sodium citrate (see Methods). It has previously been reported that citrate selectively removes the outer membrane of the nuclear envelope leaving the inner nuclear membrane intact (Humbert et al. 1996).

The application of InsP₃ via bath solution and therefore to the nucleoplasmic side of the nuclear membrane of citric acid-treated nuclei activated multiple ion channels in most excised patches (136 out of 153 patches, Fig. 5). The same response was also achieved with nucleus-attached patches. The competitive blocker of InsP₃Rs, heparin (0.02–0.4 mg ml⁻¹) added to the nucleoplasmic side of the patch, suppressed the channel activity (n = 3). InsP₃ in the pipette solution (at the luminal side of the membrane) was ineffective. Therefore the internal nuclear membrane of Purkinje neurones contained InsP₃Rs with their receptor loci directed toward the nucleoplasm. In symmetric KCl solutions the InsP₃-activated channels had a linear current–voltage relationship with a slope conductance of 356 ± 4 pS (n = 7, Fig. 6A and C, •).

Figure 5. InsP₃-activated ion channels in the internal nuclear membrane of Purkinje cells.

A, in the absence of InsP₃ no channels were recorded. B, InsP₃ (10 μm) added to bath solution (the nucleoplasmic side of the membrane) activated at least 3 channels, shown as downward deflections from the zero current level indicated by arrows. C, the blocker of InsP₃Rs, heparin (0.4 mg ml⁻¹), inhibited the channel activity. The bath and pipette contained standard KCl solution with Na₂ATP (0.5 mm) and Ca²⁺ (250 nm). Holding potential was −40 mV.

Figure 6. InsP₃-activated channels are selective to Ca²⁺.

In symmetrical KCl solution (A and C, •) the current–voltage relationship of InsP₃-activated channels had a slope conductance of 356 ± 4 pS. After K⁺ in the pipette solution was substituted by NMDG⁺ (B and C, O) only inward currents were recorded indicating that the channels were permeable to K⁺ and impermeable to Cl⁻. Single InsP₃-activated currents when patch pipettes were filled with high Ca²⁺ (D and F, ▪) or Ba²⁺ (E and F, □) solution indicate that the channel was permeable to Ca²⁺ and Ba²⁺ with P_Ca/P_K= 5.2 and P_Ba/P_K= 5.7.

InsP₃ receptors in the endoplasmic reticulum are Ca²⁺-selective channels, which are also permeable to other cations such as K⁺ and Ba²⁺ (Bezprozvanny & Ehrlich, 1994; Boehning et al. 2001). An InsP₃-activated Na⁺ channel impermeable to Ca²⁺ has also been reported (Somasundaram & Mahaut-Smith, 1995). To determine the permeability of the channel to K⁺ and Cl⁻ we substituted the large organic monovalent cation NMDG⁺ mole for mole for K⁺ in the patch pipette solution. Under these conditions only inward InsP₃-activated currents were recorded, suggesting that the InsP₃-activated channels were permeable to K⁺ with little if any permeability to Cl⁻ and NMDG⁺ (n = 4, Fig. 6B and C, O). To determine the Ca²⁺ permeability of the channel, patch pipettes were filled with solution containing 50 mm CaCl₂ and 30 mm KCl. The working chamber contained standard KCl solution with 10 μm InsP₃ and 200 nm free Ca²⁺. Under these conditions the reversal potential of the InsP₃-activated channels was −20.7 ± 0.5 mV (n = 4; Fig. 6D and F, ▪). The calculated permeability ratio for Ca²⁺ and K⁺ was P_Ca/P_K= 5.2. The InsP₃-activated channels were also permeable to Ba²⁺ (Fig. 6E and F, □). When patch pipettes were filled with solution containing 100 mm BaCl₂ and 10 mm Hepes-KOH (pH 7.3) the current through InsP₃-activated channels had a reversal potential of −34 ± 0.8 mV (n = 3), corresponding to P_Ba/P_K= 5.7. The single-channel conductance of InsP₃Rs with Ba²⁺ as the current carrier in the potential range from −20 mV to 60 mV was 121 ± 2 pS (n = 3). Therefore nuclear InsP₃Rs were Ca²⁺ channels with conductance and selectivity similar to those previously reported for cerebellar InsP₃Rs incorporated into artificial lipid bilayers and recombinant type 1 InsP₃Rs expressed in mammalian cells (Bezprozvanny & Ehrlich, 1994; Boehning et al. 2001).

The open probability of the InsP₃-activated channels (P_o) depended on the ion species carrying current through the channel (Fig. S4 of Supplemental material). At a membrane potential of 60 mV in symmetrical KCl solution containing 10 μm InsP₃, 250 nm free Ca²⁺ and 0.5 mm ATP, P_o was 0.036 ± 0.04 (n = 3). With Ba²⁺ as a current carrier (patch pipettes were filled with 100 mm BaCl₂) under the same conditions P_o increased to 0.32 ± 0.03 (n = 3). The presence of Ca²⁺ in the pipette solution in concentrations ≥ 10 mm strongly decreased the P_o of outward currents through InsP₃Rs (Fig. 6D). When patch pipettes contained 50 mm CaCl₂ and 30 mm KCl, the P_o of single outward InsP₃-activated currents was reduced to 0.0023 ± 0.0003 (n = 4). The decrease in P_o apparently results from the inhibitory effect of high [Ca²⁺] in the vicinity of the receptor produced by Ca²⁺ current through the channel rather than an effect of the presence of Ca²⁺ at the luminal surface of the membrane, because the P_o of single inward K⁺ currents remained practically unchanged (0.038 ± 0.006, n = 4, V_h=−60 mV.

Most patches of the inner nuclear membrane contained both InsP₃Rs and large conductance cationic channels (see Figs S3 and S4 of Supplemental material). The superposition of multiple large conductance cationic channels complicated our study of InsP₃Rs. To separate the two channels in the following experiments Ba²⁺ was used as the current carrier. Patch pipettes were filled with solution containing (mm): BaCl₂, 100 and Hepes-KOH, 10 (pH 7.3). InsP₃-activated currents were recorded at positive membrane potentials. The bath contained standard KCl solution with InsP₃ and free Ca²⁺ as indicated. As we mention above, 198 pS channels were impermeable to Ba²⁺ and partly inhibited by large concentrations of Ba²⁺. Thus at positive membrane potentials currents through these channels were negligible.

In the absence of InsP₃, Ca²⁺ alone (20 nm–50 μm) was unable to activate the channels. In the presence of 250 nm free Ca²⁺ InsP₃ evoked a noticeable activity of the channels in concentrations > 0.1 μm and fully activated InsP₃Rs in concentrations ≥ 2–3 μm (n = 5, Fig. 7). The experimental data can be fitted by the Hill equation with EC₅₀= 0.68 μm and a Hill coefficient of 2.5.

Figure 7. Dependence of nuclear InsP₃-activated channels on InsP₃ concentration.

A, InsP₃-activated channel activity at different InsP₃ concentrations. B, dependence of the normalized open probability of InsP₃-activated channels on InsP₃ concentration. Data points represent mean ± s.e.m. of five experiments. The continuous curve is the Hill equation fit with EC₅₀= 0.68 μm and Hill coefficient = 2.5. Patch pipettes were filled with BaCl₂ solution, bath contained standard KCl solution with [Ca²⁺]_i= 250 nm. Holding potential was 40 mV.

It has previously been reported that Ca²⁺ can both activate and, at higher concentrations, inhibit InsP₃Rs (Iino, 1990; Bezprozvanny et al. 1991; Finch et al. 1991). In Xenopus oocytes the inhibitory effect of Ca²⁺ on InsP₃Rs was eliminated by treatment with saturating concentrations of InsP₃ (Mak et al. 1998). Therefore we studied the Ca²⁺ dependence of InsP₃Rs at both low (0.3 μm, about 7% of the maximum response; n = 5) and saturated (10 μm, n = 7) concentrations of InsP₃ (Fig. 8 and Fig. S5 of Supplemental material). At both InsP₃ concentrations the Ca²⁺ dependence of the InsP₃-activated channels was very similar with the maximum activity of the channel observed at [Ca²⁺]_i in the range from 200 to 400 nm and strong inhibition of InsP₃Rs by [Ca²⁺]_i≥ 1 μm.

Figure 8. Dependence of InsP₃-activated channels on Ca²⁺ concentration at the nucleoplasmic side of the membrane.

Channel activity (A) and normalized open probability of InsP₃-activated channels (B) at different Ca²⁺ concentrations in the presence of low (0.3 μm, left; n = 5) and saturated (10 μm, right; n = 7) InsP₃ concentrations. At both InsP₃ concentrations 1 μm of Ca²⁺ almost completely inhibited the channel activity. Patch pipettes were filled with BaCl₂ solution, bath contained standard KCl solution. Holding potential was 40 mV.

Apart from InsP₃Rs, ryanodine receptors have also been reported in Purkinje neurones (Pozzan et al. 1994). Therefore we tried to detect these channels in the nuclear membrane of Purkinje neurones. The agonists of ryanodine receptors, Ca²⁺ (10–50 μm) and cyclic adenosine 5′-diphosphate-ribose (cADPR, 10 μm), activated channels neither in the outer (n = 56 and 17, respectively) nor in the inner (n = 23 and 12) nuclear membrane of Purkinje neurones. The agonist of putative Ca²⁺-releasing channels, nicotinic acid adenine dinucleotide phosphate (NAADP, 20–50 nm), was also ineffective (n = 15).

Ion channels in the nuclear membrane of granule neurones

In sharp contrast to Purkinje neurones no large-conductance cationic channels were recorded in the nuclear membrane of cerebellar granule neurones. The total density of the channels was also much lower. The most common channel in the nuclear membrane of granule neurones (17 out of 78 patches) had a pattern of activity very similar to that of the anionic channel in Purkinje neurones (Fig. 9A and B). The channel quickly fluctuated between multiple subconductance states. The conductances of the more common subconductance states were 46 ± 5 and 18.8 ± 2.1 pS (n = 4). When Cl⁻ in the bath solution was replaced with equimolar amount of gluconate the current–voltage relationship of the channel shifted to the right and was close to the reversal potential for Cl⁻ (Fig. 9C). When 125 mm of the total 150 mm KCl was replaced with potassium gluconate the reversal potential of the current was 40.25 ± 1.3 mV (n = 4). These data indicate that the channel is selectively permeable to Cl⁻ with P_K/P_Cl and P_Glu/P_Cl < 0.05.

Figure 9. Anion channel in the nuclear membrane of a granule neurone.

A and B, the anion channel quickly fluctuated between multiple subconductance states. Two of the substates are shown by dashed lines (B). C, current–voltage relationships of the main conductance substates of the anion channel. The arrow indicates the equilibrium potential for Cl⁻ (V_Cl). Pipettes contained standard KCl solution, in bath solution 125 mm KCl was substituted with potassium gluconate.

The nuclear membranes of granule neurones also contained a channel with a different pattern of activity (Fig. 10A). It was found in 6 out of 78 patches. In symmetrical KCl solution the channel demonstrated small inward rectification with the slope conductance at negative membrane potentials of 53 ± 4 pS (n = 5, Fig. 10B). The substitution of isotonic KCl solution with hypotonic solutions with reduced concentration of KCl shifted the reversal potential of the channel to the left suggesting cationic selectivity of the channel (not illustrated).

Figure 10. Cation channels in the nuclear membrane of granule cells.

A, currents through a single channel at two different holding potentials. B, the current–voltage relationship of the cation channel in symmetric KCl solution demonstrated small inward rectification.

In experimental conditions identical to those used for recording InsP₃Rs in nuclear membranes of Purkinje neurones, InsP₃ (10 μm) in KCl solution containing 0.5 mm ATP and 250 nm free Ca²⁺ failed to activated InsP₃Rs in either the outer (n = 32) or the inner (n = 26) nuclear membrane of granule neurones.

Discussion

Here we have presented a method for isolating nuclei from identified neurones of the brain. We have employed this technique to isolate nuclei from cerebellar Purkinje and granule neurones, but the same approach can be readily used for isolating nuclei from any other brain cells (see Fig. S2 of Supplemental material). The significance of studying properties of nuclei from particular types of brain cells is illustrated by the findings of this work.

The major finding is that cerebellar Purkinje and granule cells express different sets of ion channels in their nuclear membranes and therefore their nuclear envelopes may play distinct functional roles. Distinct types of spontaneously active ion channels have also been reported in the outer nuclear membrane of B- and T-lymphocyte cell lines (Franco-Obregon et al. 2000). The inner, but not the outer, nuclear membrane of Purkinje neurones contained multiple InsP₃-activated channels with their receptor loci facing the nucleoplasm. No InsP₃Rs were found in the nuclear membrane of granule neurones. Although in vitro expression of InsP₃Rs in cerebellar granule neurones can be induced artificially (Choi et al. 2004), in vivo these neurones express little if any InsP₃Rs (Pozzan et al. 1994; Taylor et al. 1999). Therefore the absence of InsP₃-activated channels in the nuclear membranes of granule neurones in our experiments is hardly surprising. On the other hand, Purkinje neurones express the highest level of InsP₃Rs among mammalian cells and their absence in the outer nuclear membrane needs some discussion.

It is well-established that the endoplasmic reticulum is a functionally heterogeneous organelle (reviewed by Meldolesi & Pozzan, 1998; Papp et al. 2003). Ca²⁺-releasing channels as well as other components of Ca²⁺-signalling machinery are distributed very unevenly throughout the endoplasmic reticulum. In particular in Purkinje neurones InsP₃Rs can be packed in stacks of parallel cisternae at a density about 100 times higher than in the rest of the endoplasmic reticulum and are practically absent in some other regions (Ross et al. 1989; Satoh et al. 1990). InsP₃Rs in Purkinje neurones were found in perinuclear cisternae although the exact localization of the receptors to the inner or outer nuclear membrane was uncertain.

The localization and orientation of the Ca²⁺-release channels in the nuclear membrane that we report suggest that the nuclear envelope in Purkinje neurones is specialized to release Ca²⁺ directly into the nucleoplasm. This is the first evidence for the existence of intranuclear sources of Ca²⁺ in neurones. Ca²⁺ release from the nuclear envelope may amplify Ca²⁺ signals penetrating into the nucleus from the cytoplasm or generate Ca²⁺ transients in the nucleus independently from the cytoplasm. This hypothesis also makes it possible to explain the distribution of InsP₃Rs in the nuclear membrane – InsP₃Rs in the outer nuclear membrane would release Ca²⁺ into the cytoplasm and therefore would reduce Ca²⁺ transients inside the nucleus.

Purkinje neurones predominantly express type 1 InsP₃Rs although small amounts (∼1–3%) of type 2 and 3 receptors have also been reported (De Smedt et al. 1994; Taylor et al. 1999). Nuclear receptors comprise only a small fraction of the total pool of InsP₃Rs in Purkinje neurones and therefore their identity may differ from InsP₃Rs of the rest of the cell. Different types of InsP₃Rs have similar ion selectivity and conductance, but greatly vary in their sensitivity to InsP₃ (Thrower et al. 2001). The InsP₃Rs studied here had an EC₅₀= 0.68 μm that is close to values (0.194–0.5 μm) reported for cerebellar type 1 InsP₃Rs incorporated into artificial lipid bilayers, but differs from the EC₅₀ of type 2 and 3 receptors (58 nm and 3.2 μm, respectively; Ramos-Franco et al. 1998; Hagar & Ehrlich, 2000). These data suggest that the InsP₃Rs expressed in the inner nuclear membrane of Purkinje neurones are likely to be type 1.

Our values for EC₅₀ (0.68 μm) and the Hill coefficient (2.5) of InsP₃Rs in Purkinje neurone nuclei were somewhat higher than those reported for cerebellar InsP₃Rs incorporated into artificial lipid bilayers (0.194 μm and 0.96 reported by Ramos-Franco et al. 1998; 0.5 μm and 1.7 reported by Hagar & Ehrlich, 2000). InsP₃Rs have several putative phosphorylation sites (Thrower et al. 2001). It has been reported that phosphorylation by protein kinase A of recombinant type 1 InsP₃Rs expressed in insect cells increased the P_o of the channels incorporated into artificial lipid bilayers more than 10-fold (from < 2–3% to 30–40%) and increased their sensitivity to InsP₃ about 4-fold (Tang et al. 2003). Different level of phosphorylation may in principle explain differences in EC₅₀ reported by different authors. Tang et al. (2003) estimated the EC₅₀ of phosphorylated InsP₃Rs to be < 50 nm. If the values obtained for heterologously expressed InsP₃Rs hold for native mammalian receptors then because of the high EC₅₀ of the nuclear InsP₃Rs they must have been dephosphorylated. On the other hand, the P_o of the nuclear InsP₃Rs with Ba²⁺ as the current carrier was 0.32 which is close to the P_o of phosphorylated recombinant InsP₃Rs. Therefore most of the nuclear InsP₃Rs were apparently phosphorylated. These data suggest that the properties of heterologously expressed and native InsP₃Rs may, at least quantitatively, differ. The cause of this discrepancy is unknown and the effects of phosphorylation on native InsP₃Rs need further investigation.

Experiments on permeabilized cells and with flash photolysis of caged InsP₃ in cerebellar slices have demonstrated that InsP₃Rs in the cytoplasm of Purkinje neurones are 20–50 times less sensitive to InsP₃ than the same receptors incorporated into artificial lipid bilayers (Khodakhah & Ogden, 1993; Fujiwara et al. 2001). The reason for this difference is not known and may result from the presence in Purkinje neurones of intrinsic inhibitors of InsP₃Rs (Yang et al. 2002; Haynes et al. 2004; Kasri et al. 2004). In contrast to cytoplasmic receptors, InsP₃Rs in the native nuclear membrane of Purkinje neurones were highly sensitive to InsP₃. These data suggest that the nuclear Ca²⁺ store of Purkinje neurones has a much lower threshold for activation by InsP₃ than the rest of the endoplasmic reticulum. Similar differences in the sensitivity of nuclear and endoplasmic reticulum InsP₃Rs have been reported in a liver cell line, but in that case it was achieved by expression of different types of InsP₃Rs (Leite et al. 2003).

Several groups have reported that mammalian type 1 InsP₃Rs were activated by low (< 300 nm) and inhibited by higher Ca²⁺ concentrations (Iino, 1990; Bezprozvanny et al. 1991; Finch et al. 1991). In experiments on Xenopus oocytes Mak et al. (1998) found that an increase in InsP₃ concentration about completely eliminated the inhibitory effect of Ca²⁺. In our experiments Ca²⁺ inhibited InsP₃Rs with the same efficiency both at low (0.3 μm, ∼7% of P_max) and saturated (10 μm) InsP₃ concentrations. Therefore the inhibitory effect of Ca²⁺ on Purkinje neurone nuclear InsP₃Rs did not depend on the InsP₃ concentration. Experiments with flash photolysis of caged InsP₃ in Purkinje neurones support these data. It has been shown that Ca²⁺ entry through plasmalemmal Ca²⁺ channels strongly suppressed Ca²⁺ release from stores induced by high (25 μm) InsP₃ concentrations (Khodakhah & Ogden, 1995). Recombinant type 1 InsP₃Rs expressed in insect cells also demonstrated a bell-shaped Ca²⁺ dependence at saturated InsP₃ concentrations and it was not affected by phosphorylation (Tang et al. 2003).

There is also other evidence that Xenopus InsP₃Rs are not functionally identical to mammalian type 1 receptors. Data presented by Mak et al. (1998) suggest that Xenopus oocyte InsP₃Rs are much more sensitive to InsP₃ than mammalian type 1 receptors. The authors estimated the EC₅₀ of Xenopus InsP₃Rs to be 50 nm, less by about an order of magnitude than the EC₅₀ of mammalian type 1 InsP₃Rs. Another difference between the two receptors is the very high open probability of Xenopus InsP₃Rs (P_max∼ 0.8) with long (> 1 s) open bursts, whereas in our experiments in similar ion conditions (K⁺ as the current carrier) P_max= 0.034 and no long-lasting bursts of activity were observed. Studies on type 1 InsP₃Rs incorporated into artificial lipid bilayers also report P_max in the range of a few per cent, although these data were obtained in non-physiological ionic conditions (Ramos-Franco et al. 1998; Hagar & Ehrlich, 2000).

Xenopus oocyte InsP₃R genes are highly homologous although not identical to mammalian type 1 InsP₃Rs (∼93%; Taylor et al. 1999). Therefore the discrepancies in properties of InsP₃Rs in Xenopus oocytes and Purkinje neurones may result from structural differences between the receptors. An alternative explanation of the differences is suggested by the report that purified cerebellar InsP₃Rs incorporated into artificial lipid bilayers were not inhibited by Ca²⁺ and therefore the inhibition of InsP₃Rs by Ca²⁺ may depend on accessory proteins (Michikawa et al. 1999). The assortment of these proteins may be different in Xenopus and mammalian cells and this may be the reason for the differences in the properties of InsP₃Rs in these cells.

A decrease in the inhibitory effect of Ca²⁺ was also reported in cerebellar InsP₃Rs incorporated into artificial lipid bilayers, but this effect was observed at unusually high InsP₃ concentrations (180 μm) and the physiological relevance of this observation is unclear (Kaftan et al. 1997; Thrower et al. 2001).

An important effect of saturated InsP₃ concentrations was the high activity of InsP₃Rs at resting levels of [Ca²⁺]_i. At [Ca²⁺]_i= 100 nm InsP₃R activity reached about 46% of P_max at 10 μm InsP₃, compared with about 17% of P_max at 0.3 μm InsP₃. Taking into account that P_max at 0.3 μm InsP₃ comprised only 7% of that at 10 μm[InsP₃] the absolute activity of InsP₃Rs at resting [Ca²⁺]_i rose about 40 times with the rise in InsP₃ concentration. The bell-shaped Ca²⁺ dependence of InsP₃Rs is considered to be the cause of Ca²⁺ oscillations seen in many cells (Bezprozvanny et al. 1991; Thrower et al. 2001). Ca²⁺ oscillations are usually observed at low levels of agonist stimulation whereas stronger stimuli evoke a biphasic Ca²⁺ response (a Ca²⁺ spike followed by a smaller steady-state rise in [Ca²⁺]_i). This effect was explained by Mak et al. (1998) by the so called ‘tuning’ of the receptor, i.e. the elimination of the inhibitory effect of Ca²⁺ on InsP₃Rs at saturated InsP₃ concentrations generated by strong stimuli. Our data suggest an alternative explanation of this phenomenon – at large InsP₃ concentrations high activity of InsP₃Rs even at low [Ca²⁺]_i makes the sequestration of Ca²⁺ in the endoplasmic reticulum ineffective and produces biphasic rather than oscillatory responses to agonists.

Yang et al. (2002) reported that a family of neuronal Ca²⁺-binding proteins (CaBP) interacts with InsP₃-binding region of InsP₃Rs and showed that a member of this CaBP1 family is an agonist of InsP₃Rs of the outer nuclear membrane of Xenopus oocytes. They also demonstrated that CaBP1 is highly expressed and co-localized with InsP₃Rs in Purkinje neurones. The authors proposed that CaBP1 is a protein agonist of InsP₃Rs that enables the channel to be involved in Ca²⁺-induced Ca²⁺ release in the absence of InsP₃ in a manner similar to ryanodine receptors. In our experiments Ca²⁺ alone, in the absence of InsP₃, was unable to activate InsP₃Rs. CaBP1 is a membrane-localized protein and it is unlikely that it was washed-out during nucleus isolation. These data suggest that CaBP1 is not involved in regulation of nuclear InsP₃Rs in Purkinje neurones. There are several other reports that indicate that InsP₃-independent activation of InsP₃Rs is not a universal property of cerebellar InsP₃Rs. A local rise in InsP₃ in Purkinje neurone dendrites due to synaptic activation of metabotropic glutamate receptors or flash photolysis of caged InsP₃ evokes an increase in [Ca²⁺]_i restricted to the area of stimulation without a significant regenerative propagation of the Ca²⁺ signal, in spite of the reported presence of high levels of CaBP1 throughout the dendritic tree (Finch & Augustine, 1998; Takechi et al. 1998). In experiments on other mammalian cells CaBP1 inhibited agonist-evoked InsP₃-mediated Ca²⁺ signalling and no rise in InsP₃-independent Ca²⁺-induced Ca²⁺ release was reported by the authors (Haynes et al. 2004; Kasri et al. 2004). Therefore we, and other authors also, have not obtained any data confirming the agonist action of CaBP1, and its role in InsP₃R regulation remains contentious.

Nuclear envelopes from Purkinje neurones also contained numerous spontaneously active large-conductance channels. The channels had high open probability and were permeable to K⁺ and some other monovalent cations, but impermeable to Ca²⁺. Large-conductance cationic channels resembling the nuclear channels of Purkinje neurones have previously been reported in the sarcoplasmic reticulum of different cells (see, e.g. Picard et al. 2002). High permeability to K⁺ is generally characteristic of the membrane of endo(sarco)plasmic reticulum Ca²⁺ stores. These channels are thought to provide a route for counterflow of K⁺ to prevent changes in the membrane potential that may arise due to movements of Ca²⁺ through the endoplasmic reticulum membrane. The large-conductance cationic channels in the nuclear membrane of Purkinje neurones may have the same function. This assumption is consistent with the fact that the channels are expressed in InsP₃-sensitive nuclear envelopes of Purkinje neurones, but not in the nuclear envelopes of granule neurones.

As we have already mentioned above, InsP₃-activated Ca²⁺ channels have been recorded in the outer nuclear membrane of Xenopus oocytes (Stehno-Bittel et al. 1995; Mak et al. 1998). No native InsP₃-activated channels have until now been recorded in the nuclear membrane of mammalian cells (Mazzanti et al. 2001). On the other hand fluorescent Ca²⁺ measurements indicate that the nuclear envelope of liver and a number of other mammalian cells are InsP₃-sensitive Ca²⁺ stores (Nicotera et al. 1990; Gerasimenko et al. 1995). Incorporation of the nuclear membranes from a hepatocyte cell line into artificial lipid bilayers also revealed the presence of InsP₃Rs (Leite et al. 2003). The distribution of InsP₃Rs in mammalian nuclei has not been unequivocally established. InsP₃ binding sites were found in the internal nuclear membrane of hepatocytes (Humbert et al. 1996). It has recently been reported that InsP₃ can release Ca²⁺ from protrusions of inner nuclear membrane in the nucleoplasm of a hepatocellular carcinoma cell line, indicating the presence of functional InsP₃Rs inside the nucleus (Echevarría et al. 2003). On the other hand, careful analysis of Ca²⁺ signalling in HeLa cells showed no sources of Ca²⁺ inside the nucleus (Lipp et al. 1997).

These discrepancies may, at least partially, be explained by different mechanisms of Ca²⁺ regulation in the nuclei of different cells. In our experiments we were unable to record Ca²⁺ channels in the nuclear membrane of cerebellar granule neurones. Therefore different types of neurones may have distinct mechanisms of regulating Ca²⁺ in the nucleoplasm. The reason for these differences between neurones is unknown. One possibility is that Ca²⁺ channels in the internal nuclear membrane are expressed only in neurones where nuclear Ca²⁺ signalling plays an important role and larger Ca²⁺ transients may be required to trigger downstream events. Alternatively, the existence of a source of Ca²⁺ inside the nucleus of Purkinje neurones may result from the large size of these nuclei. The nucleus of Purkinje neurones is about twice as large in diameter and 8 times larger in volume than the nucleus of granule neurones. The inflow from the cytoplasm may by itself be insufficient to fill this volume with the necessary amount of Ca²⁺.

InsP₃ for modulating the Ca²⁺ excitability of InsP₃Rs in the inner nuclear membrane may come from two sources. It may be synthesized in the plasma membrane and enter the nucleus from the cytoplasm through nuclear pores. In neurones this possibility is limited by the fact that only a small fraction of InsP₃ synthesized in neurones can reach the nucleus, because the vast majority of the InsP₃-synthesizing metabotropic receptors are located in dendrites and produce only a local rise in InsP₃ concentration (Finch & Augustine, 1998; Takechi et al. 1998). The alternative possibility is that InsP₃ may be synthesized inside the nucleus. It has been reported that the nucleus contains a complete set of enzymes of InsP₃ metabolism (D'Santos et al. 1998; Irvine, 2000). The regulation of these enzymes is poorly understood. In Swiss 3T3 cells stimulation of insulin-like growth factor I (IGF-I) receptors activates inositide metabolism in the nucleus. IGF-I is present in climbing fibres, is released by their electrical stimulation and Purkinje neurones express IGF-I receptors (Ito, 2001). Therefore InsP₃ levels in the nucleus of Purkinje neurones may be regulated by IGF-I and/or other neuromodulators independently from the cytoplasm, switching on and off the nuclear envelope Ca²⁺ store.

The physiological role of nuclear Ca²⁺ signalling has not been unequivocally established and may be different in different cells. As we have already mentioned in the introduction, nuclei contain Ca²⁺-sensitive transcription factors and therefore a rise in nuclear Ca²⁺ can affect gene transcription. It has been hypothesized that nuclear Ca²⁺ may be involved in the formation of transcription-dependent neuronal plasticity (Bading, 2000). A rise in nuclear Ca²⁺ controls CREB-mediated gene expression triggered by synaptic activity (Hardingham et al. 2001). In Purkinje neurones associated stimulation of synaptic inputs from climbing and parallel fibres evokes long-term depression (LTD) of synaptic transmission at excitatory parallel fibre synapses (reviewed by Ito, 2001). LTD induction is Ca²⁺ dependent. Mice with a disrupted type 1 InsP₃R gene completely lack LTD suggesting an important role of InsP₃Rs in LTD formation (Inoue et al. 1998). It has been shown that the late phase of cerebellar LTD requires protein synthesis (Linden, 1996). Inhibition of CREB or CaMKIV function eliminated or strongly suppressed the late phase of LTD (Ahn et al. 1999; Ho et al. 2000). Therefore the formation of the late phase of LTD depends on Ca²⁺-sensitive transcription factors. InsP₃Rs in the internal nuclear membrane of Purkinje neurones may be involved in regulation of nuclear Ca²⁺ metabolism and induction of late phase of LTD.

Supplemental material

The online version of this paper can be accessed at: 10.1113/jphysiol.2004.081299 http://jp.physoc.org/cgi/content/full/jphysiol.2004.081299/DC1 and contains five supplemental figures.

This material can also be found as part of the full-text HTMl version available from http://www.blackwell-synergy.com